here - CRSS-SCT

Transcription

here - CRSS-SCT
36th CANADIAN SYMPOSIUM ON REMOTE SENSING /
36e SYMPOSIUM CANADIEN DE TÉLÉDÉTECTION
Delta St. John’s Hotel and Conference Centre / Delta St. John’s, Hôtel et
Centre des congrès
St. John’s, Newfoundland and Labrador / St. John’s, Terre-Neuve-etLabrador
June 8 – 11, 2015 / 8 – 11 Juin 2015
ABSTRACTS / RÉSUMÉS*
*The title, abstract text and authorships are verbatim from the abstract submission. Presenting authors
have been updated with information gleaned from the online abstract registrations. / Le titre, texte du
résumé, et les noms d’auteurs sont tiré du résumé soumit au SCT. Le nom du présentateur a été mise à
jour avec l’information soumit lord de l’inscription du résumé en ligne.
List of Sessions / Liste des sessions
S1 Airborne Remote Sensing, UAVs / S1 Télédétection aéroportée, drones ............................................. 10
UAS Applications for Local Area Mapping and Volumetric Measurement ............................................. 10
Enabling Technologies for Safe and Long-Range UAV Remote Sensing Operations .............................. 11
Airborne Mapping of Sea Surface Temperature Anomalies ................................................................... 12
S2 Data/Image Processing / S2 Traitement des données et images .......................................................... 13
Scan-Line Stripping Correction in Composited Landsat 5 and 7 Image Products ................................... 13
Moderate Resolution Time Series Data Management and Analysis: Automated Large Area Synthesis
and Quality Control ................................................................................................................................. 14
Early case studies on geologic applications enabled by SWIR bands on the WorldView-3 Satellite...... 15
S3 Forestry Applications / S3 Applications en foresterie ............................................................................ 16
Object-Based Image Analysis for Operational Forest Mapping Using Aerial Imagery and Digital Surface
Model Data ............................................................................................................................................. 16
Ash Decline Assessment in Emerald Ash Borer Infested Forests using High Spatial Remote Sensing
Technologies ........................................................................................................................................... 17
Temporal disparity between leaf area index and leaf chlorophyll content in temperate deciduous
forests: Implications for GPP modelling ................................................................................................. 18
A Novel Algorithm for Reconstructing 3-D Branches from Point Clouds................................................ 19
S4 Remote Sensing for International Development / S4 Télédétection pour le développement
international................................................................................................................................... 20
The Role of Remote Sensing in Development ........................................................................................ 20
Remote Sensing Applications for Environmental and Social Due Diligence in International
Development........................................................................................................................................... 21
Remote Sensing Research and Education in Developing Countries: What Delivers and Why ............... 22
Towards the remote detection of graves: a four-year study of the ecology and spectral evolution of
experimental mass graves in the tropics ................................................................................................ 23
S5 Microwave Radar Applications / S5 Applications du radar micro-ondes............................................... 24
Wave Height Estimation from Ship-Borne X-Band Nautical Radar Images ............................................ 24
Performance Evaluation of an Interacting Multiple Model Marine-Radar Target Tracker .................... 25
Case Study: Pedestrian Bridge Static and Dynamic Monitoring Using Ground-Based Radar ................. 26
Short-Term Drift Trajectory Prediction Using X-band Marine Radar...................................................... 27
S6 Innovative Earth Observation Techniques Supporting Informed Oil & Gas Development – I / S6
Techniques d’observation de la Terre innovatrices en soutien au développement pétrolier et
gazier – I ......................................................................................................................................... 28
Earth Observation Contributions to Responsible Resource Development in the Alberta Oil Sands
Region ..................................................................................................................................................... 28
Anthropogenic Change Detection using Landsat Multispectral Datasets .............................................. 30
Remote Sensing as a Tool to Assess Reclaimed Areas: a Multi-sensor, Multi-data Approach ............... 31
Piloting a Predictive Ecosite Mapping Platform for Alberta ................................................................... 32
2
Using Satellite imagery and Automated Feature Extraction Techniques to Accurately Map Alberta’s
Pipeline Corridors.................................................................................................................................... 33
S7 Getting Ready for RADARSAT Constellation Mission / S7 Se préparer pour la Constellation RADARSAT
........................................................................................................................................................ 34
RADARSAT Constellation/ SENTINEL-1 Missions Update........................................................................ 34
RCM Data Utilization & Application Plan (DUAP) ................................................................................... 35
The Development of RCM Value-added Products and Services at Natural Resources Canada .............. 36
RADARSAT Constellation Mission: Opportunity to Expand Soil Moisture Information .......................... 37
RCM Data Utilization & Application Plan (DUAP) Activities at the Canadian Ice Service ....................... 38
S8 Marine Remote Sensing / S8 Télédétection des milieux marins............................................................ 39
Landsat-based kelp mapping in British Columbia ................................................................................... 39
Mapping of Boulders in Seismically Heterogeneous Marine Soils.......................................................... 40
Testing the Capabilities of a Landsat Time Series to Assess Changes in Marine Protected Areas in
Eastern Africa .......................................................................................................................................... 41
Results from the first topo-bathymetric lidar surveys of the Chiroptera II sensor for coastal and
freshwater sites in Maritime Canada ...................................................................................................... 42
Buried UXO Imaging Survey with a Sub-Bottom Imager™ ...................................................................... 43
S9 Airborne Imaging and Calibration / S9 Imagerie aéroportée et calibration .......................................... 44
Boreal forest and peatland classification integrating GEOBIA and Hyperspectral Imagery ................... 44
Bundle Adjustment of VNIR, SWIR and LWIR Airborne Hyperspectral Systems at Rothera Research
Station, Antarctic Peninsula .................................................................................................................... 45
Atmospheric correction of airborne hyperspectral imagery in diverse tropical ecosystems – lessons
learned from Mission Airborne Carbon 13 ............................................................................................. 46
SNR Implications of On-chip vs Off-chip Pixel Summation ..................................................................... 47
S10 Innovative Earth Observation Techniques Supporting Informed Oil & Gas Development – II / S10
Techniques d’observation de la Terre innovatrices en soutien au développement pétrolier et
gazier – II ........................................................................................................................................ 48
Utilization of Earth Observation Technologies to Assist Incident Response and Investigation ............. 48
Monitoring Pipeline Right-of-way Encroachment Using Multi-temporal High Resolution SAR ............. 50
Surface water and wetland monitoring in the Peace-Athabasca Delta .................................................. 51
Using ArcGIS as an Alternative or Complimentary Tool to ENVI for Extracting Anthropogenic Footprints
from Multispectral Imagery .................................................................................................................... 52
Systematically Derived Leaf Area Index of the Canadian Oilsands Region ............................................. 53
S11 Characterization and Change in Forest Ecosystems / S11 Caractérisation et changements dans les
écosystèmes forestiers................................................................................................................... 54
Assessment of Vegetation Conditions over Athabasca Oil Sands Region .............................................. 54
Long-term Trends in and Interrelationships between Central Boreal Plains Ecosystem Succession and
Hydro-Climatic Conditions ...................................................................................................................... 55
Using pre-fire variables in combination with Landsat difference Normalized Burn Ratio to improve the
accuracy in predicting middle burn severity classes ............................................................................... 56
3
The Multisource Vegetation Inventory: Integrating Remote Sensing and Field Data for Forest
Inventory in the Northern Boreal ........................................................................................................... 57
Systematic and Environmental Sensitivities within GLAS Estimates of Canopy Height and Crown
Closure in Northern Canada.................................................................................................................... 58
S12 Synergies Between Satellite & Marine Acoustic Remote Sensing / S12 Synergies entre la
télédétection satellitaire et la télédétection acoustique marine .................................................. 59
Multiple methods, maps, and management applications: purpose made maps in support of Ocean
Management........................................................................................................................................... 59
Where Land Meets Sea: Towards a Complete Coverage for Terrain Analysis of Coastal Environments 60
Acoustic remote sensing of Labrador fjord environments: strengths, challenges and limitations ........ 61
Using Multibeam Bathymetry and Coastal Geomorphology to Predict Shallow Marine Clam Habitat in
the Eastern Canadian Arctic .................................................................................................................... 62
Automated sonar image matching ......................................................................................................... 63
S13 River Ice Monitoring / S13 Suivi de la glace de rivière ......................................................................... 64
Augmenting River Ice Flood Forecasting Services Using Satellite Radar Imagery – A User Perspective 64
Operational Monitoring of River Ice Break-up Conditions using RADARSAT-2 Images .......................... 65
Using Multi-Source Satellite Imagery for Operational River Ice Monitoring .......................................... 66
IceFRONT (1): Monitoring River Freeze-up Processes on the Peace River from Optical Images ........... 67
IceFRONT (2): Monitoring River Freeze-up Processes on the Peace River from Radar Images ............. 68
S14 Arctic-Boreal Research and the pre-ABoVE Projects / S14 Projets de recherche sur la vulnérabilité
arctique–boréale et les projets pre-ABoVE.................................................................................... 69
The NASA Arctic-Boreal Vulnerability Experiment and the ABoVE Science Cloud ................................. 69
Quantifying change in North American Arctic lakes between 1990 and present .................................. 70
Long-Term Multi-Sensor Record of Fire Disturbances in High Northern Latitudes ................................ 71
Remotely Sensed Active Layer Thickness (ReSALT) ................................................................................ 72
Producing a High-Resolution Circumpolar Delineation of the Forest-Tundra Ecotone .......................... 73
S15 Icebergs and Extreme Ice Features / S15 Icebergs et formations glacielles extrêmes ........................ 74
Intercomparison of sea ice freeboard, thickness, and ice concentration extracted from sea ice model
with satellite data ................................................................................................................................... 74
Iceberg Detection over Northern Latitudes Using High Resolution TerraSAR-X Images ........................ 75
Detection and Discrimination of Ships and Icebergs using Satellite Altimetry: An Operational
Perspective for Estimating Iceberg Populations ..................................................................................... 76
Characterization of Hazardous Ice using Spaceborne SAR and Ice Profiling Sonar ................................ 77
Research and Development in Iceberg Monitoring at the Canadian Ice Service ................................... 78
S16 Global Vegetation Science / S16 Télédétection de la végétation à l'échelle globale .......................... 79
Satellite observed sunlight-mediated seasonality in greenness of wet equatorial Amazonian rainforest
................................................................................................................................................................ 79
Global ENSO Teleconnection Patterns derived from NDVI3g: 1981-2013 ............................................. 80
Earth Observations and Global Environmental Governance .................................................................. 81
4
A Non-Stationary 1981 to 2014 AVHRR NDVI Time Series ..................................................................... 82
Global warming and trends in seasonality of NDVI: Europe 1982-1998 versus 1999-2014 ................... 83
S17 3D Characterization of Canopy Structure using LiDAR and New Technologies – I / S17 Caractérisation
du couvert végétal avec le LiDAR et les nouvelles technologies - I ............................................... 84
GLAS Estimates of Canopy Gap Fraction ................................................................................................. 84
An individual tree based framework to support operational forest resources inventory using fullwaveform airborne LiDAR data ............................................................................................................... 85
Predicting Tree Species at the Tree-Level Using LiDAR .......................................................................... 86
Sage grouse hate trees: object-based mapping of juniper encroachment across the sage-grouse range
................................................................................................................................................................ 87
Individual tree detection from LiDAR-derived canopy height models (CHM) in longleaf pine forest .... 88
S18 Monitoring Surface Water and Wetland Ecosystems with SAR – I / S18 Suivi des milieux humides et
des surfaces d’eau avec le RSO - I .................................................................................................. 89
Mapping Ephemeral Surface Water Bodies with Radarsat-2 ................................................................. 89
Evaluation of Temporal Filters for SAR Applications in Water Resources .............................................. 90
Early results of inland water-body detection using multipolarized L-band SAR: airborne, Aquarius, and
SMAP ....................................................................................................................................................... 91
Monitoring water levels by integrating optical and synthetic aperture radar water masks with lidar
DEMs ....................................................................................................................................................... 92
Multitemporal Interferometric SAR for Accurate Water Extent Mapping ............................................. 93
S19 Coastal HF Radar Applications / S19 Applications des radars HF côtiers............................................. 94
An Ionospheric Reflection Coefficient Model for HF Ionosphere-Ocean Propagation........................... 94
Comparison of Antenna-Motion-Incorporated High Frequency Bistatic Radar Cross Sections of the
Ocean Surface with Earlier Models ......................................................................................................... 95
Design and Implementation of a High-Frequency Software-Defined Radar (HF-SDR): Initial Results ... 96
Space-Time Adaptive Processing for Coastal Surveillance Applications................................................. 98
Comparison of Spectral Estimation Methods for Rapidly-varying Currents Obtained by High Frequency
Surface Wave Radar ................................................................................................................................ 99
S20 Future Canadian Earth Observation Missions / S20 Futures missions canadiennes en observation de
la Terre ......................................................................................................................................... 100
The Canadian Hyperspectral Mission Concept ...................................................................................... 100
The Earth Observation Potential of the Polar Communication and Weather (PCW) Mission ............. 101
The WaterSat Mission Concept............................................................................................................. 102
RADARSAT Constellation Mission ......................................................................................................... 103
A Microsatellite-based Canadian Wildland Fire Monitoring System .................................................... 104
S21 3D Characterization of Canopy Structure using LiDAR and New Technologies – I / S21 Caractérisation
du couvert végétal avec le LiDAR et les nouvelles technologies – II............................................ 106
Characterizing Fine-scale Understory Plant Diversity and Fuels from Three-dimensional
Photogrammetry Points ........................................................................................................................ 106
3D Modeling of Understory Fuels Determined by Overstory Structure ............................................... 107
5
Modeling 3D understory structure for use in physics-based fire behavior simulations....................... 108
Canopy derived fuels drive patterns of in-fire energy release and plant community assembly in
longleaf pine (Pinus palustris) woodlands ............................................................................................ 109
Upscaling tree density measures from environmental monitoring plots across Eglin Air Force Base
using low density lidar .......................................................................................................................... 110
S22 Monitoring Surface Water and Wetland Ecosystems with SAR – II / S22 Suivi des milieux humides et
des surfaces d’eau avec le RSO – II .............................................................................................. 111
A Preliminary Study on the Synergistic Use of SAR and High Resolution Optical Data for Wetland
Classification in Newfoundland and Labrador ...................................................................................... 111
Water Level Changes Detection in Yellow River Delta Based on Distributed Scatterer Interferometry
.............................................................................................................................................................. 112
Temporal Coherence for Wetland Mapping and Monitoring ............................................................... 113
Integration of Multi-Frequency Data into the Periodic Remote Delineation of Wetlands .................. 114
Soil Moisture Monitoring in Mountain Areas Using Different Polarization of Radarsat-2 Data .......... 115
S23 Operational Maritime Surveillance with SAR / S23 Surveillance maritime opérationnelle avec le RSO
...................................................................................................................................................... 116
Maritime NRT Products using TerraSAR-X, Sentinel, and Radarsat data .............................................. 116
Oil Slick Discrimination using RADARSAT-2 Quad Polarized Data ........................................................ 117
Detecting and Tracking Dark Ships from Multiple Satellite Missions .................................................... 118
Evaluation of RADARSAT Constellation Mission Compact Polarimetry for ship detection .................. 119
Maritime Operational Use of SAR by Environment Canada’s Canadian Ice Service ............................. 120
S24 Arctic Ecosystem Monitoring / S24 Suivi des écosystèmes arctiques ............................................... 121
Landsat-based Mapping of Thermokarst Lake Dynamics in Continuous Permafrost of Western Canada
since 1985 ............................................................................................................................................. 121
Influence of the Environmental Conditions on Surficial Mapping Using LANDSAT-5 TM and RADARSAT2 Polarimetric SAR Images .................................................................................................................... 122
River Surface Temperature Retrieval Using Optical Remote Sensing Images ...................................... 123
Mapping discontinuous permafrost at high spatial resolution using LANDSAT and RADARSAT-2 dual
polarized images in northern Ontario, Canada..................................................................................... 124
A Multiscale Study of Tundra Vegetation Changes in the Tuktoyaktuk Coastal Plain, NWT ................ 125
S25 Hyperspectral Imaging – I / S25 Imagerie hyperspectrale – I ............................................................ 126
Validation of a Two-Step Model Inversion Approach for Regional Retrieval of Leaf Chlorophyll Content
Using Remote Sensing Data .................................................................................................................. 126
Improving hyperspectral imagery automatic segmentation by differential analysis ........................... 128
Satellite-derived bathymetry, a Canadian case study .......................................................................... 129
S26 LiDAR with other sensors / S26 LiDAR avec d’autres capteurs .......................................................... 130
Classification of LiDAR Data for Large Areas Using Kurtosis Change Curve .......................................... 130
Multispectral LiDAR Data for Land Cover Classification: Initial Results ................................................ 131
Wetland mapping using polarimetric RADARSAT-2, optical imagery and LiDAR data in Nova Scotia . 132
6
Collaborative case-studies demonstrating the effectiveness of random forest classification using
LiDAR, SAR and optical data in various landscapes .............................................................................. 133
S27 Hyperspectral Imaging – II / S27 Imagerie hyperspectrale – II .......................................................... 134
Results of longer-term airborne hyperspectral imaging over single graves in a boreal environment . 134
Multi-angular spectroscopic remote sensing of arctic plant biochemistry .......................................... 135
Direct Imaging of a Shale Gas Leak Using Passive Thermal Infrared Hyperspectral Imaging ............... 136
Thermal hyperspectral (8-12 μm) investigation of near surface gases from a clandestine mass grave
.............................................................................................................................................................. 137
S28 Land and Vegetation Classification / S28 Classification des terres et de la végétation ..................... 138
Detecting Changes in Sub-Arctic Vegetation Caused by Snow Goose Foraging on Coats Island,
Nunavut: A Multi-Temporal Analysis .................................................................................................... 138
Updating the Grassland Vegetation Inventory Using Landsat Imagery-Conversion of Native Grassland
to Cultivated Agriculture ....................................................................................................................... 140
Retrieving leaf chlorophyll in crops using Landsat-8 and RapidEye images ......................................... 141
Object-based Classification of WorldView-2 Image for Development of Stormwater Models ............ 142
Exploiting Three Decades of Continuous Satellite Data with the Canada Centre for Remote Sensing
Long Term Satellite Data Records ......................................................................................................... 143
Classification Uniformisée de la Couverture Terrestre pour une Comptabilité des Terres et des
Écosystèmes .......................................................................................................................................... 145
S29 Coastal and Ice / S29 Glace et régions côtières ................................................................................. 147
The SmartICE information system – integrating Inuit knowledge, remote sensing, and in situ
measurements for safer sea-ice travel ................................................................................................. 147
Bottom-fast Ice Delineation in the Northwest Territories using Dynamic Time Warping .................... 148
Nature and triggers of submarine mass failure in coastal waters of southeastern Baffin Island,
Nunavut................................................................................................................................................. 149
An alternative method for sea surface wind speed determination from GNSS-R delay-Doppler map 150
S30 Monitoring & Modeling Terrestrial Ecosystems (Forests) / S30 Suivi et modélisation des écosystèmes
terrestres (forêts) ......................................................................................................................... 151
National, annual, gap-free surface reflectance composites from Landsat to capture long-term forest
dynamics, land cover, and forest structure for Canada........................................................................ 151
The Newfoundland Fibre Project: Four Studies that Demonstrate the use of Remote Sensing to Predict
Wood Fibre Attributes .......................................................................................................................... 152
Airborne LiDAR enhances the competitiveness of Canada’s forest industry ....................................... 154
High-resolution global maps of 21st-century forest loss: validation and application in the Miramichi
river basin ............................................................................................................................................. 155
Operational Successes and Challenges of Deriving Forest Inventories from LiDAR in Natural and
Managed Forests................................................................................................................................... 156
S31 Classification, Modeling & Northern Applications / S31 Classification, modélisation et applications
nordiques ..................................................................................................................................... 157
Land Surface Phenology and Vegetation Productivity for Circumpolar Landmass .............................. 157
7
Remotely Sensed Caribou Habitat Indicators for Enhancing Baseline Information Preparedness for
Resource Development in Canada’s North ........................................................................................... 158
Spectral Mixture Analysis for Characterizing Tree Species Composition for a Multisource Vegetation
Inventory, Northwest Territories, Canada ............................................................................................ 159
Terrestrial Ecosystems Monitoring for the Next Decade: Measurement Needs, Technologies and
Approaches, with Perspectives on Northern Applications ................................................................... 160
1981 to 2014 Circum-Arctic Tundra Phytomass Production................................................................. 161
S32 Urban / Transportation / S32 Milieu urbain et systèmes de transport ............................................. 162
Automatic Extraction of Highway Light Poles from Mobile LiDAR Point Cloud Data ........................... 162
Early Aerial Photography in Canada after the Great War – The Case of the Halifax Air Survey Mission in
1921 ...................................................................................................................................................... 163
Multi-temporal Urban Growth Assessment in Asia Pacific Region using Time Series of Pixel-Based
Landsat Imagery Composites ................................................................................................................ 164
Infrastructure Monitoring in Urban Areas using Space Based SAR ...................................................... 165
Analysis of Aerodynamic Roughness Using SAR Data in Urban Areas .................................................. 166
S33 Synthetic Aperture Radar / S33 Radar à synthèse d'ouverture ......................................................... 167
Corner Reflectors in Canada: Challenges and Solutions. ...................................................................... 167
RADARSAT-2 Progress Report ............................................................................................................... 168
Use of RADARSAT-2 and ALOS-PALSAR images for wetland mapping in New Brunswick .................... 169
A Speckle Filtering Approach of SAR Data Based on a Coherent Decomposition ................................ 170
Mapping The C-Band Vertical Backscatter Pattern Through A Conifer Forest Canopy ........................ 171
Posters / Affiches ...................................................................................................................................... 172
Web-based applications for LiDAR data processing and visualizing trees at the plot level ................. 172
Generalizing LIDAR-based, Species-level Biomass Predictions in the Northern Rocky Mountain
Ecoregion .............................................................................................................................................. 173
Hyperspectral Non-linear Denoising with High Performance Parallel Computing ............................... 174
Working with DEMs of varying resolution and derivation.................................................................... 175
Buried Geo-Hazard Imaging and Mapping on the Grand Banks using an Acoustic CorerTM ................. 176
Using ArcGIS as an Alternative or Complimentary Tool to ENVI for Extracting Anthropogenic Footprints
from Multispectral Imagery .................................................................................................................. 177
Multispectral Lidar for 3D Land Classification and Topography/Bathymetry ...................................... 178
A Study of Microwave Sea Clutter using a Coherent X-band Radar ..................................................... 179
Delineation of within-field Spatial Variability using high resolution optical remote sensing data ...... 180
Modeling Biodiversity Response to Habitat Heterogeneity Using Multi Spatial and Temporal Remote
Sensing Data.......................................................................................................................................... 181
Using an Unmanned Aerial Vehicle and Close Range Photogrammetry to 3D Model Selkirk College's
Castlegar Campus.................................................................................................................................. 182
The Birth and Growth of Pingo in the Canadian Arctic Observed by Satellite Radar Interferometry .. 183
Spatial Distribution of Fine Particulate Matter over Southern Ontario from MODIS Data .................. 184
8
Forest Prairie Landscape Classification Based on Remote Sensing Data Fusion .................................. 185
Investigating the Effects of Input Data on the Results of Random Forest for Classification of Peatland
Landscapes ............................................................................................................................................ 186
Extracting sea ice information from the fusion of SAR and optical data .............................................. 187
Using Satellite Imagery for Visual Simulation in Aviation Applications ................................................ 188
Modelling Changes in Forest Dynamics under the Influence of Climate Change and Different Forest
Management Practice ........................................................................................................................... 189
Large-Scale Modeling of Forest Height and Biomass using the Metabolic Scaling Theory and WaterEnergy Balance Equation ...................................................................................................................... 190
A Rapid Assessment of Light Geese Damage in the Central and Eastern Arctic using Multi-temporal
Landsat Imagery Stacks. ........................................................................................................................ 191
Investigating Boreal Forest Physiology and Stand-Age Relationships Using a Landsat TM Derived
Chronosequence ................................................................................................................................... 192
Modelling and mapping soil carbon in British Columbia grasslands .................................................... 193
A Landsat-Based Study of Black Rock Coatings Proximal to Base-Metal Smelters, Sudbury, Ontario,
Canada .................................................................................................................................................. 194
Mapping northern peatland types using hyperspectral images of different spatial and spectral
resolutions ............................................................................................................................................ 195
Relating Low Arctic Tundra – Atmosphere CO2 Exchange to Satellite-Derived NDVI using Phenological
Analysis at Daring Lake, NWT ............................................................................................................... 196
Influence of Sample Distribution and Prior Probability Adjustment on Land Cover Classifier Extension
.............................................................................................................................................................. 197
Fractional Land Cover Monitoring of the Alberta Oil Sands Region using Landsat multispectral time
series and high resolution Geoeye Imagery ......................................................................................... 198
Using Airborne and Terrestrial Lidar to Estimate Biomass of Low-Stature Arctic Tundra Shrubs ........ 199
Change Detection Analysis Using Compact Polarimetry On Simulated RCM Data............................... 200
Comparing the potential of dual-pol TerraSAR-X, Sentinel, and Radarsat data for automated,
polarimetric sea ice classification ......................................................................................................... 201
Application Tests of Ground-Based Coherent Radar for Deformation and Vibration Measurements in
Canada's Atlantic Region....................................................................................................................... 202
Comparison of Different Methods of Soil Moisture Mapping in Peatlands using Synthetic Aperture
Radar Polarimetric Data ........................................................................................................................ 203
Classification of forest and wetland communities near the Victor Diamond Mine using an integrated
optical and radar satellite sensor dataset............................................................................................. 204
Helix Nebula – Supporting science with a comprehensive cloud computing infrastructure: the use case
for Earth Observation ........................................................................................................................... 205
9
S1 Airborne Remote Sensing, UAVs / S1 Télédétection aéroportée,
drones
Dirk Werle, chair / modérateur
UAS Applications for Local Area Mapping and Volumetric Measurement
Bradley J. Schmidt1*, Dawn Price
1. Business Development, Intergraph Canada, 1600 Carling Avenue, Ottawa, Ontario, Canada, 613 696-1221,
[email protected]
2. Application Consultant, Intergraph Canada, 1600 Carling Avenue, Ottawa, Ontario, Canada, 613 696-1224,
[email protected]
* Presenting Author: Bradley J. Schmidt, Business Development, 613 696-1221,
[email protected] .
ABSTRACT
With the advent of unmanned aerial systems (UAS), new applications are emerging for local area mapping
purposes. This presentation looks at using low altitude high-resolution UAS photo imagery for the creation
of highly accurate point clouds, which are then in turn used for local area mapping, volumetric
measurement, and change detection purposes.
Imagery was collected using a Nikon camera, flown on an Aibotix multicopter, over an aggregate pit
located in Phelpston, Ontario. The Aibot flight was planned and flown, the imagery and GPS data was then
combined with the camera information. The resulting data was then processed using an automated
workflow to create an orthomosaic, RGB encoded point cloud, and a digital surface model. Using this data
various analytics were performed.
Given the precision of the point cloud, highly accurate volume measurements were extracted, and
compared with other software based methodologies. Changes in volume were also calculated using a
second point cloud. The point cloud was then classified to identify ground, vegetation, buildings and other
features within the scene. Findings indicated that UAS derived imagery can be used to accurately
determine the volume of aggregate piles visible within the imagery. Practical applications include;
estimations for the quantity of truckloads required for construction or landscape purposes. The
methodology could also be used to measure the amount of debris deposited by a landslide, or snow or
salt contained in pile. In conclusion, UAV systems can be deployed quickly and efficiently for generating
accurate volumetric output and for general purpose ortho-mapping.
10
Enabling Technologies for Safe and Long-Range UAV Remote Sensing Operations
Chris D. Rouse*, Siu O’Young, Vincent M. Contarino, Robert R. MacIsaac, Dilhan J. Balage, Scott X. Fang,
Carl Thibault, Jonathan Keck, Timothy Keck
1. Seamatica Aerospace Limited, St. John’s, NL, Canada, A1B 2T4, 709-864-4488, [email protected]
* Presenting Author: Dr. Chris D. Rouse, Industrial R&D Fellow, 709-864-4488,
[email protected]
ABSTRACT
The number of commercial applications for unmanned aerial vehicles (UAVs) has been growing steadily,
especially in the area of airborne remote sensing. Examples include ice field mapping for climate change
studies, power line and oil pipeline monitoring and the mapping of coral reefs for ecosystem monitoring.
UAVs can also act as first responders during disasters for which it would be dangerous to fly manned
aircraft. The primary limitation on the commercial viability of UAVs is that their operation is restricted to
visual range, mitigating the risk of collisions with potential air, land or sea traffic. In applications where
remote sensing is desired over a relatively large area, this requires that the area be broken into smaller
regions—each involving an individual UAV operation—which increases both cost and measurement time.
The visual range limitation arises due to deficiencies in current UAV sense and avoid technology. This
motivates the development of a safety management system capable of ensuring that the desired section
of airspace is clear for safe UAV operation beyond visual range. The associated technologies include
automatic dependent surveillance-broadcast (ADS-B) and automatic identification system (AIS) for the
detection of participating aircraft and vessels, along with a coherent solid state radar system to detect
non-participating vehicles. Long-range and wideband communication with the UAV can be achieved via
high-gain antennas mounted on a pan-tilt unit which tracks the aircraft during flight. In addition to the
design details associated with each enabling technology, test results highlighting their performance and
applicability will be presented.
11
Airborne Mapping of Sea Surface Temperature Anomalies
Stephen Achal1*, Warren Shaw2 and Brandon Southgate3
1. Chief Scientist, ITRES Research Limited, Calgary, Canada
2. Research Scientist, ITRES Research Limited, Calgary, Canada
3. Technical Lead, ITRES Research Limited, Calgary, Canada
* Presenting Author: Stephen Achal, Chief Scientist, 403 250 9944, [email protected]
ABSTRACT
On May 5th 2013, a TSR-1800 (a state-of-the-art 1800-pixel real-time airborne thermal search and rescue
system) was used on an experimental trial to collect daytime search-and-rescue research data near the
coast of Tofino, British Columbia, Canada. Data were acquired over a ten square kilometer area at
25cm/pixel resolution with a temperature finesse of less than 30mK @ 12C. During the trial, several
thermal anomalies in the sea surface temperature (SST) were tagged and geo-located. These anomalies
included the control targets as well as unexpected marine animal tracks (birds and whales). The animal
tracks appeared as cooler SST features. The wakes of fishing boats were observed as cooler persistent SST
disturbances. Unusual very large-scale SST structures with intricate fine-scale features were also
observed. The SST variation within these structures ranged from 12C to 15C. These SST structures were
not likely linked with the typical convection of thermoclines, but likely due to phytoplankton or algae
activity near the sea surface.
Application of this type of state-of-the-art large-scale airborne thermal mapping of SST over photobiologically active waters may find utility in marine biology and oceanography.
Keywords: airborne mapping, ocean thermal imaging, marine animal tracks, ocean thermal anomaly,
thermoclines, sea surface temperature
12
S2 Data/Image Processing / S2 Traitement des données et images
Bahram Salehi, chair / modérateur
Scan-Line Stripping Correction in Composited Landsat 5 and 7 Image Products
Darren Pouliot1*, Rasim Latifovic2
1. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759-6341, [email protected]
2. Research Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa, Ontario
K1A 0E4, Canada, 613 759-7002, [email protected]
* Presenting Author: Darren Pouliot, Remote Sensing Scientist, 613 759-6341, [email protected]
ABSTRACT
In large area multi-scene Landsat 5 and 7 composite products, missing data due to the Scan Line Corrector
(SLC) failure of Landsat 7 can result in strong stripping artefacts. This is due to the infilling of the missing
scan lines with data from alternate dates where atmosphere and ground target conditions vary.
Atmosphere correction is often insufficient to account for this due to the difficultly of acquiring detailed
information on atmospheric properties. This combined with vegetation phenology and moisture changes
can cause significant reflectance variability resulting in errors in land cover classification and change
detection applications. In this research a Fourier Transformation correction procedure was developed that
consisted of identification of regions with scan line errors, predicting the scan line orientation angle,
generating a frequency space mask, frequency space filtering, and conversion back to the time domain.
The approach was automated and was used in the generation of a high quality national scale Landsat 5
and 7 mosaic of Canada developed for land cover applications. An overview of the methods and example
results will be presented and discussed.
13
Moderate Resolution Time Series Data Management and Analysis: Automated
Large Area Synthesis and Quality Control
Rasim Latifovic1*, Darren Pouliot2, Lixin Sun2, John Schwarz2, William Parkinson2
1. Research Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa, Ontario
K1A 0E4, Canada, 613 759-7002, [email protected]
2. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759-6341, [email protected]
3. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759-7696, [email protected]
4. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7759, [email protected]
5. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6535, [email protected]
* Presenting Author: Rasim Latifovic, . Research Scientist, 613 759-7002, [email protected]
ABSTRACT
The Canada Centre for Remote Sensing (CCRS) maintains national-scale Long Term Satellite Data Records
(LTSDR) as an essential component of Earth Observation based land surface monitoring. The CCRS LTSDR
framework provides long-term capability to generate, archive and provide access to value-added satellite
data and thematic products addressing various land surface monitoring needs of the Government of
Canada.
Until recently, cost and availability limited the usefulness of medium resolution (~30m pixel size) data for
such analyses. In 2009, the United States Geological Survey made Landsat data freely available. The
potential for moderate-resolution time series monitoring has been further strengthened by Landsat-8 and
the pending launch of Sentinel 2. For this potential to be realized, new methods and algorithms are
required to extract and analyze Landsat Time Series for monitoring aspects of land surface dynamics.
CCRS’s new framework, the Moderate Resolution Time Series Data Management and Analysis System
(TSDMAS), generates and manages value-added LTSDRs based on TM, ETM+ and OLI sensors on board the
Landsat 5, 7 and 8 missions. An overview of the TSDMAS system, and the algorithms implemented
therein, will be presented. Specific focus will be on the generation of a circa 2010 Top of Atmosphere
Reflectance Coverage of Canada, at 30 m spatial resolution. This product has been generated by TSDMAS
from data acquired by TM and ETM+ sensors. Product generation, quality control, and characteristics of
the underlying dataset will be described. By providing readily available, national-scale Landsat data
products, of “research quality”, TSDMAS harnesses the potential of moderate resolution EO data, and can
exponentially increase the downstream generation and use of medium resolution land surface
information products.
14
Early case studies on geologic applications enabled by SWIR bands on the
WorldView-3 Satellite
William Baugh1*
1. Senior Scientist/Product Development and Labs, DigitalGlobe, Longmont, Colorado, USA, 303-684-1481,
[email protected]
* Presenting Author: William Baugh, Senior Scientist/Product Development and Labs, 303-684-1481,
[email protected]
ABSTRACT
Significant new capabilities for geologic remote sensing were enabled with the successful launch of the
WorldView-3 (WV-3) commercial remote sensing satellite on 13 August 2014. Most noteworthy for
geologic remote sensing are 8 super-spectral bands in the shortwave infrared (SWIR) range of the
electromagnetic spectrum. These imaging bands are collected at a native 3.7 m resolution, and cover a
range of diverse and unique absorption features for minerals. During the calibration phase of WV-3 (and
prior to commercial availability), images have been collected over a variety of sites to demonstrate
geologic applications and capabilities. This presentation will be a survey of early results from these ongoing WV3 geologic remote sensing case studies.
15
S3 Forestry Applications / S3 Applications en foresterie
Joanne White, chair / modérateur
Object-Based Image Analysis for Operational Forest Mapping Using Aerial
Imagery and Digital Surface Model Data
Bahram Salehi1*, William Jefferies2, and Pradeep Bobby3
1. Remote Sensing Engineer, C-CORE/LOOKNorth and Cross-appointed professor, Faculty of Engineering and
Applied Science, Memorial University of Newfoundland, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL,
Canada, (709) 864 6701, [email protected]
2. Executive Director, LOOKNorth, C-CORE, 400 March Road, Suite 210, Kanata, ON, Canada,
[email protected]
3. Director, Earth Observation, C-CORE, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St. John’s, NL, Canada,
[email protected]
* Presenting Author: Bahram Salehi, Remote Sensing Engineer, 709 864 6701, [email protected]
ABSTRACT
The standard approach for forest resource inventory mapping is to interpret aerial ortho-photographs and
delineate forest stand boundaries by manual digitization of the stereo imagery, usually visualized in 3D.
However, manual digitization is costly, time consuming and results are subjective to the interpreter’s
knowledge and experience. Although there exist historical and frequently acquired aerial ortho-photos
for most Canadian forested areas, sufficient methods are still lacking for automatic delineation of such
boundaries for operational purposes in regional and provincial scales. New advances in object-based
image analysis techniques provide additional opportunity for automatic delineation of forest stand
boundaries using high resolution satellite and aerial imagery.
This paper presents an operational object-based image analysis framework for automatic forest mapping
using multispectral aerial imagery with sub-meter resolution and corresponding digital surface model
(DSM) extracted from stereo-aerial imagery. The framework comprises five steps: pre-processing of the
ortho-photos, reconstruction of digital terrain model (DTM) from DSM, image segmentation, and
classification of segments. The framework is a hierarchical-based approach which begins with image
classification to general classes of water, treed and non-treed areas. In the lower levels of the hierarchy
more detailed classes such as narrow water streams and trees with different species and densities patches
are delineated. The outputs from the analysis are shape file polygons with attributes which can be directly
imported to a GIS system for further analysis.
The method provides an automatic and relatively fast approach for operational use in mapping forest
types at regional and provincial scales using cost-effective aerial imagery.
16
Ash Decline Assessment in Emerald Ash Borer Infested Forests using High Spatial
Remote Sensing Technologies
Justin Murfitt1*, Amy Mui2, Jian Yang3 and Yuhong He 4
1. Research Assistant, Department of Geography, University of Toronto Mississauga, 3359 Mississauga Road,
Mississauga, Ontario, Canada, (905) 569-4564, [email protected]
2. PhD Student, Department of Geography, University of Toronto Mississauga, 3359 Mississauga Road,
Mississauga, Ontario, Canada, (905) 569-4564, [email protected]
3. PhD Student, Department of Geography, University of Toronto Mississauga, 3359 Mississauga Road,
Mississauga, Ontario, Canada, (905) 569-4564, [email protected]
4. Associate Professor, Department of Geography, University of Toronto Mississauga, 3359 Mississauga Road,
Mississauga, Ontario, Canada, (905) 569-4679, [email protected]
* Presenting Author: Justin Murfitt, Research Assistant, 905 569-4564, [email protected]
ABSTRACT
The Emerald ash borer (EAB) is an invasive Asian beetle that was first identified in Windsor, Ontario in
2002 and has spread throughout the Great Lakes region and into southern areas of Quebec. This pest
causes significant damage to ash trees which are a crucial natural resource within Ontario. Due to the
spatial extent and intensity of the EAB infestation, remote sensing technologies are being evaluated for
identifying and mapping early EAB infestation and subsequent stages. Using high resolution images, the
goal of this project is to establish an efficient and reliable remote sensing system for mapping EAB
infestation at different stages. EAB infestation field data were collected from three sites (heavily infested,
moderately infested, and low infested) in the Credit Valley Watershed in Southeastern Ontario in the
summer of 2014. An ash crown map was first produced from the WorldView-2 image of a moderately
infested site using a multi-band watershed segmentation approach and a supervised classification. Several
different ash health indicators including twig dieback, crown condition, general tree health, canopy
transparency, plant area index, and chlorophyll content were used to establish an ash tree health index,
which was then related to remote sensing spectral indices. The ash health index-remote sensing
regression model was applied to the ash crown image to produce an ash health map that achieved an
accuracy of 90%. Our study confirms that it is possible to use high spatial remote sensing images to
monitor the decline of ash health within emerald ash borer infested forests.
17
Temporal disparity between leaf area index and leaf chlorophyll content in
temperate deciduous forests: Implications for GPP modelling
Holly Croft1*, Jing M. Chen2, Yongqin Zhang3, Ralf M. Staebler4, Norma Froelich5 and Bin Chen6
1. 1. Holly Croft, Postdoctoral Research Fellow, University of Toronto, Department of Geography, Toronto, ON M5S
3G3, Canada, [email protected]
2. Jing M. Chen, Professor, Department of Geography, University of Toronto, Toronto, ON M5S 3G3, Canada
[email protected]
3. Yongqin Zhang, Assistant Professor, Delta State University, Division of Biological and Physical Sciences, Cleveland,
MS 38733, USA, [email protected]
4. Ralf M. Staebler, Research Scientist, Environment Canada, Atmospheric Processes Research, 4905 Dufferin Street,
Toronto, ON, M3H 5T4, Canada, [email protected]
5. Norma Froelich, Assistant Professor, Northern Michigan University, Earth, Environmental and Geographical
Sciences Department, Marquette, MI 49855, USA, [email protected]
6. Bin Chen, Postdoctoral Research Fellow, University of Toronto, Department of Geography, Toronto, ON M5S 3G3,
Canada, [email protected]
* Presenting Author: Holly Croft, Postdoctoral Research Fellow, [email protected]
ABSTRACT
Spatial and temporal variations in canopy structure and leaf biochemistry have considerable influence on
fluxes of CO2, water and energy in vegetation. The temporal relationships between canopy structure, leaf
chlorophyll (ChlLeaf) and forest carbon uptake were investigated in two temperate, deciduous forests in
Ontario, Canada. Ground data including Leaf Area Index (LAI), hyperspectral leaf
reflectance/transmittance (400–2500 nm) and ChlLeaf were measured across complete growing seasons
(Haliburton Forest in 2004, Borden Forest in 2013). ChlLeaf and LAI phenology were analysed at the
landscape scale for Haliburton Forest using ENVISAT MERIS (1200m) data, where LAI- and ChlLeaf- sensitive
vegetation indices (NDVI and Macc01) were derived for 33 dates. Phenological metrics were extracted
from the satellite time-series using a double logistic model. Results showed that ChlLeaf start of season
dates (SOS) lagged 20–35 days behind LAI SOS, and end of season (EOS) LAI dates were between 20 and
30 days later than ChlLeaf EOS. Equivalent LAI and ChlLeaf growing season dynamics were also seen at Borden
Forest in 2013, where LAI and ChlLeaf were compared with eddy-covariance flux data to assess relationships
with gross primary production (GPP); LAI: R2=0.60 (p<0.05) and Canopy Chlorophyll (ChlCanopy): R2=0.70
(p<0.001). Finally, measured GPP was compared to modelled GPP as a function of LAI*PAR (R 2=0.63,
p<0.001) and ChlCanopy*PAR (R2=0.89, p<0.001). The large temporal disparity between ChlCanopy and LAI has
important implications for biogeochemical models using NDVI or LAI to represent canopy-absorbed PAR,
by neglecting to account for delays in chlorophyll production and thus photosynthetic capacity.
18
A Novel Algorithm for Reconstructing 3-D Branches from Point Clouds
Zhouxin Xi1* and Christopher Hopkinson2
1. Student, University of Lethbridge, 4401 University Drive, Lethbridge, Canada, (403) 332-4043,
[email protected]
2. Associate Professor, University of Lethbridge, 4401 University Drive, Lethbridge, Canada, (403) 332-4586,
[email protected]
* Presenting Author: Zhouxin Xi, Student, 403 332-4043, [email protected]
ABSTRACT
The branch components play a fundamental role in tree structures, for providing the frame for the leaf
distribution and water transportation. The increasing resolution of terrestrial laser scan (TLS) makes it
possible to model the 3-D branch structure at a fine scale. Various algorithms have been experimented
on automatic branch reconstruction. However, gaps are left for understanding the algorithmic
performance on diverse branch structures and imperfect scanning conditions. The evaluation of
performance at the stage is deficient too, due to the lacking knowledge in parameter tuning complexity
and the accuracy assessment.
Our study is focused on proposing a stable and flexible algorithm for branch reconstruction with also an
emphasis on accuracy assessment. We first simulated point clouds for artificial 3D branch meshes with
varying structural parameters. For each set of the parameter values, the point clouds were reconstructed
into skeletons and meshes with the combined use of region growth and cylinder fitting. The geometrical
errors (missing error and redundant error) and the statistical errors between the reconstructed branches
and the original branches are calculated. Results using the simulation dataset show errors lower than 6cm
(missing error) and 3cm (redundant error) in best cases. The algorithm performance is affected by branch
level, bevel, taper, scan point density and the scan numbers. The effect of branch number
underestimation, instable volume estimation is constant, caused by the branch complexity and imperfect
scans. We also tested our reconstruction algorithm with the point clouds from in-situ TLS scans with
visually appealing results.
19
S4 Remote Sensing for International Development / S4 Télédétection
pour le développement international
Robert Ryerson, chair / modérateur
The Role of Remote Sensing in Development
Robert Ryerson1* and Barry Haack2
1. President, Kim Geomatics Corporation, Box 1125, Manotick, Ontario K4M 1A9 [email protected]
2. Professor of Geography, George Mason University, 4400 University Drive, Fairfax, VA 22030, USA
[email protected]
* Presenting Author: Robert Ryerson, President, Kim Geomatics Corporation, 613 692-0185
[email protected]
ABSTRACT
The importance of remote sensing in development cannot be denied. As has often been said – “everything
happens somewhere,” and remote sensing provides a very useful picture of everywhere. However, the
use of remote sensing is often at a far lower level in development activities than one would expect given
its potential importance for meeting the Millennium Development Goals, Sustainable Development Goals,
or the targets identified for the Post-2015 Agenda. Remote sensing can and should play a role in the three
major stages of development assistance: identification of the need, program development, and program
assessment and evaluation. There are a number of considerations and factors that, when taken together,
explain the relatively low use of remote sensing in development. This paper examines these
considerations and factors drawing on the work of the authors in more than forty developing countries
over the past thirty years. With better understanding of these factors it is hoped that the benefits and
that have long been promised by the remote sensing community will be delivered. The authors have
worked in the area of remote sensing and development in Africa, Asia, South America, and the Pacific.
20
Remote Sensing Applications for Environmental and Social Due Diligence in
International Development
Olivier Tsui1*, Jason Suwala 2, Andy Dean 3, Thomas Boivin 4
1. Senior Geomatics Specialist, Hatfield Consultants Partnership, Suite 200 - 850 Harbourside Drive, North
Vancouver, BC, Canada, +1 604 926 3261, [email protected]
2. Environmental Information Systems Manager, Hatfield Consultants Partnership, Suite 200 - 850 Harbourside
Drive, North Vancouver, BC, Canada, +1 604 926 3261, [email protected]
3. Senior Geomatics Specialist and Partner, Hatfield Consultants Partnership, Suite 305 - 1228 Kensington Rd. NW,
Calgary, AB, Canada, +1 403 351 0191, [email protected]
4. Director of International Operations and Partner, Hatfield Consultants Partnership, Suite 200 - 850 Harbourside
Drive, North Vancouver, BC, Canada, +1 604 926 3261, [email protected]
* Presenting Author: Olivier Tsui, Senior Geomatics Specialist, 1 604-926-3261,
[email protected]
ABSTRACT
Technologies such as remote sensing can play an important role in international development, providing
governments, industry, and non-governmental organizations with cost-effective tools to acquire
environmental baseline information and to improve monitoring and management of natural resources.
International Financial Institutions and governmental donors are key international partners for many
developing countries, providing pro-poor and climate resilient investment for economic growth in key
sectors. As part of their operating protocols, these organizations must ensure that social and
environmental analysis is part of every investment decision.
This review presents how remote sensing technologies can contribute to the evaluation of environmental
and social risks associated with a proposed investments by International Financial Institutions and their
partners. During project implementation, remote sensing provides a cost-effective means to regularly
update all partners with information on how well certain risks are being managed. Leveraging the remote
sensing data archives, remote sensing also supports project evaluation following completion. These
concepts are illustrated with example case studies, primarily from Southeast Asia.
Due diligence systems will be dramatically improved by data streams available from upcoming remote
sensing constellation missions, such as Sentinel-2, RapidEye+, UrtheCast, and nano-satellites such as
PlanetLabs. The potential impact of these systems will be evaluated with potential use cases.
21
Remote Sensing Research and Education in Developing Countries: What Delivers
and Why
Robert Ryerson1* and Barry Haack2
1. President, Kim Geomatics Corporation, Box 1125, Manotick, Ontario K4M 1A9 [email protected]
2. Professor of Geography, George Mason University, 4400 University Drive, Fairfax, VA 22030, USA
[email protected]
* Presenting Author: Robert Ryerson, President, Kim Geomatics Corporation,
[email protected]
ABSTRACT
Research in remote sensing has been undertaken around the world since the 1970s. A number of different
models have been used to develop research capacity and education in remote sensing to support
development in less developed countries. In some cases students have come to developed countries to
pursue PhDs or post-doctoral fellowships. In others, scientists from developed countries have worked for
a short time in the developing countries. In still other cases there have been research partnerships
between institutions in developed and developing countries. Scientists from developed countries have
also served on advisory and examination committees in universities in less developed countries to
strengthen them. At the same time, there have been different models to give voice to scientists from less
developed countries – different approaches to symposia, and now the Internet. There are also ongoing
issues of hardware, software, internet speed, and data availability for education. This paper examines
what seems to produce education and research useful to the developing country and why.
The authors have worked in research environments involving remote sensing and development in Africa,
Asia, South America, and the Pacific.
22
Towards the remote detection of graves: a four-year study of the ecology and
spectral evolution of experimental mass graves in the tropics
Margaret Kalacska* 1, Pablo Arroyo 1,2, Tim Moore 1, Mike Dalva 1 , George Leblanc 3
1. Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal QC, Canada
H3A 2K6
2. Geographic Information Centre, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal QC,
Canada H3A 2K6
3. National Research Council Canada, Flight Research Laboratory, 1920 Research Rd, Building U-61, Ottawa ON
Canada K1A 0R6
* Presenting Author: Margaret Kalacska, Department of Geography, McGill University, 514-398-4455,
[email protected]
ABSTRACT
Former studies have illustrated the potential of hyperspectral data for the detection of mass burials in
tropical environments and at a temperate site for large commingled burials. While single graves were
found to be detectable from airborne hyperspectral imagery both one month and four years after burial,
at present there are no longitudinal studies examining the evolution of mass graves from the ecological
perspective, the biogeochemistry or the spectral characteristics. For single graves, the spectral properties
and the feasibility of detection can vary substantially even within a time frame as short as six months.
Knowledge of how graves change over time is crucial in order to facilitate detection.
This study illustrates for the first time the vegetation successional patterns, the changes in the
biogeochemistry (production of methane and nitrous oxide) by the soil microbial community and the
changes in the spectral properties of the dominant vegetation species (Ochorma pyramidale) over the
course of four years in three experimental mass graves in Costa Rica. The study also shows that despite
being indistinguishable in the landscape from high spatial resolution aerial photography the graves are
detectable with no false positives from an airborne platform using VNIR hyperspectral imagery four years
after burial.
23
S5 Microwave Radar Applications / S5 Applications du radar microondes
Weimin Huang, chair / modérateur
Wave Height Estimation from Ship-Borne X-Band Nautical Radar Images
Xinlong Liu1*, Weimin Huang2 and Eric W Gill3
1. PhD student, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St. John’s, NL,
Canada, (709)-765-0702, [email protected]
2. Assistant Professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St.
John’s, NL, Canada, (709)-864-8937, [email protected]
3. Professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St. John’s, NL,
Canada, (709)-864-8922, [email protected]
* Presenting Author: Xinlong Liu, PhD student, 709-765-0702, [email protected].
ABSTRACT
In this paper, a signal-to-noise ratio (SNR) based method is applied to the problem of estimation of
significant wave height (HS) from ship-borne X-band nautical radar images. The method is based on the
assumption that the significant wave height is proportional to the square root of the SNR. Although this
method has been successfully applied to stationary radar systems, its application to a radar mounted on
a moving vessel is investigated here.
Low quality images acquired during rain and low wind speed are eliminated by a data quality control
algorithm. Time series of sub-images are then selected from the portion around the upwind direction. The
normalized scalar product (NSP) method is used to estimate the encounter current velocity, which is
incorporated into the dispersion relationship to separate the wave components from the noise. A linear
model is sought by curve fitting the signal-to-noise ratio (SNR) and the buoy-derived significant wave
height. Finally, based on the obtained model, significant wave height is determined using the estimated
SNR.
Validation of the method, undertaken by comparing HS obtained from the radar and buoy data, reveals a
good agreement, with a correlation coefficient of 0.78 and a root mean square (RMS) difference of 0.49
m. However, some large differences may also be observed, and these are conjectured to result from the
movement of the ship. The next phase of this work is to analyze and minimize the effect of ship motion
on wave measurements.
24
Performance Evaluation of an Interacting Multiple Model Marine-Radar Target
Tracker
Reza Shahidi1*
1. Senior Research Scientist, Rutter Technologies, Inc., 63 Thorburn Rd., St. John’s, NL, Canada, (709) 576-6666,
[email protected]
* Presenting Author: Reza Shahidi, Senior Research Scientist, 709 576-6666, [email protected].
ABSTRACT
The 8.1 release of Rutter’s sigma S6 software features a new target tracker compared to previous
releases. This new tracker is based on Interacting Multiple Models (IMM), where both constant-velocity
and maneuvering targets can be effectively tracked. In this paper, we will be comparing the performance
of the new IMM target tracker vs. Rutter’s previous single model tracker on both synthetic and real
datasets, and will show that the IMM target tracker significantly outperforms the single model tracker in
terms of both probability-of-tracking and accuracy of tracking results.
25
Case Study: Pedestrian Bridge Static and Dynamic Monitoring Using GroundBased Radar
Arpik Haikazi Hakobyan1*, Peter McGuire2, Desmond Power3, Cecilia Moloney4, Thomas Puestow5,
Guido Luzi6
1. PhD Candidate, C-CORE, MUN, 1 Morrissey Road, St. John’s, NL, Canada, 7097403137, [email protected]
2. Senior Project Engineer, C-CORE, 1 Morrissey Road, St. John’s, NL, Canada, 7098642006, [email protected]
3. Vise-President, C-CORE, 1 Morrissey Road, St. John’s, NL, Canada, 7098648353, [email protected]
4. Professor, MUN, S.J. Carew Building, 240 Prince Phillip Drive, St. John’s, NL, Canada, 709864896,
[email protected]
5. Senior Manager, C-CORE, 1 Morrissey Road, St. John’s, NL, Canada, 7098642586, [email protected]
6. Associate Professor, Senior Researcher, CTTC, Av. Carl Friedrich Gauss, Castelldefels, Spain, 34935569280,
[email protected]
* Presenting Author: Arpik Hakobyan, PhD Candidate, 709-740-3137, [email protected]
ABSTRACT
This paper demonstrates the results of applying the ground-based microwave interferometric radar
technology in monitoring a pedestrian sky-walk bridge and measuring its oscillation under a controlled
loading. The sky-walk consists of a covered bridge that connects the Aquarena and Field House buildings
located at Memorial University of Newfoundland, St. John's, NL, Canada, and it provides students, faculty,
and staff with a safe access to two major facilities at the campus, Aquarena and Field House. The sky-walk
bridge is an essential to the university because of the high volume of pedestrians using it, especially during
the winter semester. The sky-walk reliability is an important factor for the university business operations
without interruption, material damage, unanticipated costs, and possible fatalities. The possible damage
to the structure can be caused by a variety of factors, such as harsh weather conditions, high winds,
temperature fluctuations, aging of the structure. Therefore, for hazard prevention, risk mitigation and
forecasting, it is important to develop a robust and reliable approach to monitor and quantify the safety
of the structure. The radar system is able to detect the vibration response of the bridge and identify points
with high signal-to-noise ratio along the underside of the structure. Their displacements and timehistories as a response to the load are described. The ground-based radar IBIS-S(Image By Interferometric
Survey) is developed by the IDS (Ingegneria dei Sistemi SpA), Italy, and is owned by C-CORE.
26
Short-Term Drift Trajectory Prediction Using X-band Marine Radar
Armin Parsa1*, Katrin Hessner2
1. R&D Department, Rutter Inc., 63 Thorburn Road, St. John’s, Newfoundland and Labrador, Canada
2. R&D Department, Ocean Waves, Vor dem Bardowicker Tore 6b, Lüneburg, Germany
* Presenting Author: Armin Parsa, Senior Research Scientist, 709-576-6666, [email protected]
ABSTRACT
A commercial marine radar is used for predicting the drift trajectory of a light weight buoy floating on
ocean surface at Argentia, Newfoundland, Canada. The short-term drift trajectory is predicted by a Monte
Carlo simulation which takes into account the drift forces due to wind and surface current. The surface
current drifting forces are extracted from a series of radar images using WaMoS II algorithm. The radar
images were recorded from a mobile radar platform equipped with a standard marine radar transceiver,
a vertical-polarized antenna, and a sigma S6 processor. The wind field information is collected from
Argentia weather station which is approximately 1 nmi away from the drifter location, and a spatially
uniform wind field is assumed. The drifting buoy consists of a 36 inch diameter fishing buoy covered in a
radar reflective jacket. The actual location of the drifter is recorded over time using sigma S6 tracker. The
parameters used in drift prediction algorithm are optimized to reduce the error between the actual drift
trajectory and the predicted drift trajectory. The optimization parameters include drifter weight, drifter
cross sections above and below sea surface, wind direction, and wind speed.
27
S6 Innovative Earth Observation Techniques Supporting Informed Oil
& Gas Development – I / S6 Techniques d’observation de la Terre
innovatrices en soutien au développement pétrolier et gazier – I
Douglas Bancroft, chair / modérateur
Earth Observation Contributions to Responsible Resource Development in the
Alberta Oil Sands Region
Darren Pouliot 1, Richard Fernandes2, Rasim Latifovic3, Vern Singhroy4, Ridha Touzi5, Joost van der
Sanden6, Ying Zhang7, Brian Brisco 8, Todd Shipmen9 , and Darren Janzen10
1. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6341, [email protected]
2. Research Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa, Ontario
K1A 0E4, Canada, 613 759-7002, [email protected]
3. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6535, [email protected]
4. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 1047, [email protected]
5. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7181, [email protected]
6. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7208, [email protected]
7. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7987, [email protected]
8. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street,
Ottawa, Ontario K1A 0E4, Canada, 613 759-6341, [email protected]
9. Manager Landscapes and Geological Hazards, Alberta Geological Survey, Edmonton, Alberta, Canada, 780 644
5563, [email protected]
10. Section Head, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa, Ontario K1A
0E4, Canada, 613 759 6339, [email protected]
* Presenting Author: Darren Pouliot, Remote Sensing Scientist, 613 759 6341,
[email protected]
ABSTRACT
Natural Resources are fundamental to Canada’s economy and central to NRCan’s mandate. Responsible
development is important not only to Canadians but also for the social license to operate which is key to
securing export markets. For effectiveness and credibility, it is essential that science based solutions
underpin and document the responsible nature of Canada’s resource development. Earth observation can
contribute to this need through improved regulatory compliance, environmental impact assessment, and
safety/security. To this end, in collaboration with the Alberta Government and the Canadian Space
Agency, the Canada Centre for Remote Sensing launched several research and development projects in
the Alberta Oil Sands Region. The effort involved development of new methods, datasets and in depth
performance evaluations base on a variety of earth observation technologies. The main areas of research
included: change monitoring of infrastructure, surface deformation associated with in-situ oil extraction,
enhanced classification and monitoring of peatlands, measuring water level change, detailed regional land
28
surface characterization, snow cover retrieval, and river ice monitoring. This presentation will provide an
overview of these research efforts and the major outcomes to date.
29
Anthropogenic Change Detection using Landsat Multispectral Datasets
Subir Chowdhury 1*, Dennis Chao 2, Todd Shipman 3, and Ying Zhang 4
1. Remote Sensing Specialist, Alberta Geological Survey, Alberta Energy Regulator, 4999–98 Avenue NW , 4th Floor,
Edmonton, Alberta, Canada, phone: 1-780-427-4115, e-mail: [email protected]
2. GIS Specialist, Alberta Geological Survey, Alberta Energy Regulator, 4999–98 Avenue NW , 4th Floor, Edmonton,
Alberta, Canada, phone: 1-780-644-5563, e-mail: [email protected]
3. Manager, Alberta Geological Survey, Alberta Energy Regulator, 4999–98 Avenue NW , 4th Floor, Edmonton,
Alberta, Canada, phone: 1-780-644-5563, e-mail: [email protected]
4. Research Scientist, Canada Center for Mapping and Earth Observation, Natural Resources Canada, 560 Rochester
Street , 6th Floor, Ottawa, Ontario, Canada, phone: 1-613-759-7987, e-mail: [email protected]
* Presenting Author: Subir Chowdhury, Remote Sensing Specialist, 780-427-4115,
Email: [email protected].
ABSTRACT
Anthropogenic change detection associated with oil and gas activities plays an important role for the
development of effective sustainable management practices, compliance monitoring, and well site
reclamation. Satellite Earth Observation (EO) data allows us to detect reversible or irreversible changes in
infrastructures (e.g., roads, well pads, pipeline corridors, processing facilities, core hole drilling, and access
corridors to core hole sites) and transient land disturbances (e.g., cut blocks) associated with oil and gas
exploration and extraction. Quantification of these landscape changes using EO data is a solution to the
current compliance activities, which include site visits by agency personnel that is both time consuming
and costly.
The study area is located near Fox Creek, Alberta (NTS 83K / 1, 2, 6, and 7) and represents one of the major
conventional oil and gas production sites and has potential for unconventional oil and gas development
according to the AER Play-Based Regulation pilot project. Best-Available-Pixel (BAP) composite of Landsat
multispectral datasets (2005 – 2012) provided by the Canadian Forest Service, Natural Resources Canada
were used for anthropogenic change detection study. In addition, publicly available Landsat-8
multispectral datasets (2013 and 2014) provided by the United States Geological Survey were used.
The Iteratively Reweighted Multivariate Alteration Detection (IR-MAD) change detection algorithm was
applied to 2005 – 2014 multispectral datasets to extract negative (land disturbances) and positive
(vegetation recovery) changes for each year. The Support Vector Machine (SVM) classification was applied
to multispectral data of each year to classify land cover types of positive and negative changes. Ground
reference datasets were generated by matching clusters generated using the Iterative Self-Organizing
Data Analysis Technique (ISODATA) algorithm with the generalized land cover classes produced by the
Agriculture and Agri-Food Canada and the Alberta Biodiversity Monitoring Institute. The SVM was trained
using 20% pixels from each ground reference class and the rest 80% pixels of each class were used for
validation. Six land cover classes were used: developed, cut blocks, shrubland and grassland, boardleaf
forest, coniferous forest, and water. Usually, it is difficult to separate cut block class from developed land
cover class due to similar spectral characteristics. In this study, developed class was separated from cut
blocks to quantify anthropogenic changes, where a producer accuracy of 90% was achieved for this class.
30
Remote Sensing as a Tool to Assess Reclaimed Areas: a Multi-sensor, Multi-data
Approach
Kaan Ersahin 1, Leslie Brown2, Eduardo Loos3*, Gary Borstad4, Nadia Rochdi5, Karl Staenz6, David Polster7
1. Project Manager, ASL Environmental Sciences, Victoria, BC, Canada, 250-656-0177, [email protected]
2. Senior Scientist, ASL Environmental Sciences, Victoria, BC, Canada, 250-656-0177, [email protected]
3. Project Manager, ASL Environmental Sciences, Victoria, BC, Canada, 250-656-0177, [email protected]
4. Vice-President, ASL Environmental Sciences, Victoria, BC, Canada, 250-656-0177, [email protected]
5. Research Scientist, University of Lethbridge, Lethbridge, AB, Canada, 403-332-4447, [email protected]
6. Professor, University of Lethbridge, Lethbridge, AB, Canada, 403-329-2047, [email protected]
7. President, Polster Environmental Services, Duncan, BC, Canada, 250-746-8052, [email protected]
* Presenting Author: Kaan Ersahin, Project Manager, 250-656-0177, [email protected]
ABSTRACT
The objective of reclamation was usually to cover the bare ground as quickly as possible with agronomic
grasses to reduce erosion and was often done with little if any pre-disturbance information. Recent
regulation requires forestry, oil, gas, and mining companies to return disturbed land as closely as possible
to its original state upon completion of their activities in operating areas. As climate continues to change
and resource exploration and extraction increase, mapping the changing landscape becomes key for the
conservation and sustainable management of resources.
Monitoring of reclaimed sites is a complex, interdisciplinary undertaking, especially in large, disturbed
areas with difficult access. In that context, remote sensing is unique and valuable as it provides a synoptic
view of an entire reclamation program over time, extending the more detailed but sparsely-distributed in
situ monitoring. Remote sensing can provide information about a site’s vegetation history, and whether
or not it has reached and maintained biomass above the permit threshold for self-sustaining status. It also
helps decision-makers focus remediation efforts on specific locations most needing it, rather than making
unnecessary and potentially costly changes to entire sites.
The objectives of this CSA-supported project involve:

assessment of optical and RADAR remote sensing for characterization of early stage reclamation
of disturbed areas;

comparison and integration of information from optical and RADAR data analyses for different
sites; and

development of strategies for remote sensing-based reclamation monitoring, based on site
properties.
These techniques will have potential for operational, long-term, and widespread utilization in industry
and government agencies.
31
Piloting a Predictive Ecosite Mapping Platform for Alberta
Craig Aumann1*, John Simms2, Dave Springer3, Jim Schieck4, and Robert MacMillan5
1. Senior Scientist, Environment and Carbon Management, Alberta Innovates – Tech Futures, 250 Karl Clark Rd,
Edmonton AB, Canada, 780-450-5260, [email protected]
2. P.O. Box 2773, 602 7th St, Revelstoke BC, Canada, 250-837-3060, [email protected]
3. Silvacom, 3912 91st St, Edmonton, AB, Canada, 780-462-3238, [email protected]
4. Science Co-Director, Alberta Biodiversity Monitoring Institute, Bag 4000, Vegreville, AB, Canada, 780-632-8306,
[email protected]
5. LandMapper Environmental Solutions Inc., 7415 118A St NW, Edmonton AB, 780-435-4531,
[email protected]
* Presenting Author: Craig Aumann, Senior Scientist, 780-450-5260, [email protected]
ABSTRACT
Ecosite are ecologically distinct units formed from the interaction of local topography, climate, and soil
characteristics that provide information about the ecological functionality of any site. Despite the need
for consistent, comprehensive, province-wide ecosite information, the current approach to obtaining
ecosite information in Alberta is beset by information barriers (everyone collects only the minimal
information they need), is focused on site-level requirements needed for regulatory approval (not on how
to create provincial scale information), and as a result is highly fragmented, inefficient and costly.
We are exploring predictive mapping technologies through a pilot project that uses variables derived from
LIDAR DEMs and imagery (MODIS, LANDSAT), combined with existing ecosite information, to develop
predictive models using machine learning approaches. The geostatistical predictive models are being used
to produce maps over northern and southern pilot areas (each ~22,000 km2), and to design field sampling
programs to cost effectively assess and improve map accuracy.
The approach being piloted will form the core of a future predictive mapping platform for ecosites and
potentially many other terrestrial attributes (e.g. soils). The platform will inform industry on the best
locations to conduct field sampling based on the predicted spatial accuracy in the area to be sampled and
will also support use of information collected by industry and others in subsequent predictive models to
iteratively and continuously improve map accuracy across larger regions. This process of continuous
improvement ensures that map quality will continue to improve over time in a manner which is feasible,
efficient, and cost-effective.
32
Using Satellite imagery and Automated Feature Extraction Techniques to
Accurately Map Alberta’s Pipeline Corridors
Sam Lieff1*, J.V. Gill2
1. General Manager, BlackBridge Geomatics, 3528-30th St. N., Lethbridge, AB, Canada, (403.381.2800)
[email protected]
2. J.V. Gill, Sr. Manager, Operations, BlackBridge Geomatics, 3528-30th St. N., Lethbridge, AB, Canada
* Presenting Author: Sam Lieff, General Manager, 403.381.2800, [email protected]
ABSTRACT
From 2011 - 2013, Alberta Department of Energy worked with BlackBridge Geomatics to capture
anthropogenic footprint disturbance caused by Oil and Gas industrial activities. Using high-resolution
satellite imagery, BlackBridge developed methodologies to capture footprint of well pads (from oil and
gas wells drilled) and oil and gas facilities. The delivered output was spatial data (vector-based polygons)
containing relevant attributes, for the whole province of Alberta.
Currently, Alberta Energy generates the provincial pipeline footprint (the land area disturbed by pipeline
construction) by buffering the mapped pipelines to a standard width. Although this generalization of the
pipeline footprint has been adequate for approximation, actual captured data is necessary for more
accurate modelling, forecasting, and measuring cumulative effects.
BlackBridge is now building an automated workflow and intelligent quality control process to capture the
actual anthropogenic footprint disturbances caused by Oil and Gas pipeline constructions and their
associated corridors, adding to the collection of earlier footprint data. Again, high-resolution satellite
imagery is used as the base data to accurately capture the pipeline corridor locations and sizes.
33
S7 Getting Ready for RADARSAT Constellation Mission / S7 Se
préparer pour la Constellation RADARSAT
Daniel De Lisle, chair / modérateur
RADARSAT Constellation/ SENTINEL-1 Missions Update
Daniel De Lisle1 Steve Iris2 and Pierre Potin3
1. RCM Data Utilization & Applications, Canadian Space Agency, 6767 route de l’aéroport, Longueuil, QC, Canada,
J3Y 8Y9, 450-926-6611, [email protected]
2. Canadian Space Agency, 6767 route de l’aéroport, Longueuil, QC, Canada, J3Y 8Y9, [email protected]
3. European Space Agency, ESRIN, Via Galileo Galilei, 2, 00044 Frascati, Italy, [email protected]
* Presenting Author: , Daniel De Lisle, RCM Data Utilization & Applications, 450-926-6611,
[email protected]
ABSTRACT
The number of applications using radar satellite imagery has been significantly growing over the past
years. Very active in this field for more than two decades, Canada and Europe continue to maintain
leadership within C-band Synthetic Aperture Radar systems. The demand for Earth observation
intelligence to support operational monitoring applications, particularly in the maritime domain, requires
maintaining an operational satellite capability in orbit to ensure data continuity with the past/current
missions and an uninterrupted supply of key information to decision makers. Therefore, the Canadian and
European space agency’s respectively work actively on new radar missions; the RADARSAT Constellation
Mission (RCM) currently in development and Sentinel-1 (A unit already launched, B unit under
procurement). The long and successful historical collaboration between the two agencies as well as
complementary radar technologies currently in development naturally led to the establishment of an
exchange mechanism to explore and maximize the potential synergy between the two missions for the
benefit of users.
This presentation aims to provide up-to-date information on the RCM and Sentinel-1 missions’ and to
describe the main characteristic of these systems. We will also present the collaborative activities
implemented between the CSA and ESA with the aim to maximize the potential offered by their respective
systems. This cooperation and synergy could provide several benefits to the users in terms of improving
coverage of user’s areas of interest, reducing latencies, providing complementary data/modes, and
alleviating potential data acquisition conflicts. We will conclude with the activities currently ongoing to
facilitate the access to Canadian users of Sentinel data through the Canadian government satellite ground
infrastructure which will ensure the availability and continuity of imagery to support Canada’s key
operational programs.
34
RCM Data Utilization & Application Plan (DUAP)
Daniel De Lisle1*, Steve Iris2
1. RCM Data Utilization & Applications, Canadian Space Agency, 6767, route de l’aéroport, Longueuil (QC), Canada
J3Y 8Y9 450-926-6611, [email protected]
2. RCM Mission Manager, Canadian Space Agency, 6767, route de l’aéroport, Longueuil (QC), Canada J3Y 8Y9, 450926-6554, [email protected]
* Presenting Author: , Daniel De Lisle, RCM Data Utilization & Applications, 450-926-6611,
[email protected]
ABSTRACT
The RADARSAT Constellation Mission (RCM) is scheduled for a launch in 2018. A key component of the
RCM Major Crown Project (MCP) is the Data Utilization and Application Program (DUAP) intended to make
optimum use of the data when the system becomes operational. The RCM DUAP is aimed at supporting
the development of Earth observation applications and user capabilities to facilitate integration and/or
transition of those applications into their operations with a goal to start exploiting the benefits of RCM
when the system becomes operational.
Within the RCM Major Crown Project, a dedicated budget has been allocated to the Data Utilization and
Application Program in order to prepare and develop initial applications. The collaboration between the
CSA and the government users is focused on applications involving RCM data and specific activities to
exploit the unique capabilities and properties offered by the RCM:




Facilitate the integration of RCM data into existing operational applications of RADARSAT-1
and RADARSAT-2.
Support the integration of RCM data into applications that have been demonstrated to the
point where they are fully integrated into the operations of OGDs in support of their
mandates
Support science, research and development activities for potential new applications or
improve existing ones.
Facilitate the utilization and access to data, products and services produced by RCM to users.
This presentation will give an overview of the activities supported by the DUAP in collaboration with other
Government Departments.
35
The Development of RCM Value-added Products and Services at Natural
Resources Canada
Sylvia Thomas1*, Robert Landry2, Sylvain Lemay3, Sergey Samsonov4, Roger DeAbreu5, Mohamed
Habbane6, Don Raymond7
1. Canada Centre for Remote Sensing, Section Head, Canada Centre for Mapping and Earth Observation, NRCan, 560
Rochester Street, Ottawa, ON, Canada, K1A 0E4 613 694-2680, [email protected]
2. Canada Centre for Remote Sensing, Senior Physical Scientist, Canada Centre for Mapping and Earth Observation,
NRCan, 560 Rochester Street, Ottawa, ON, Canada, K1A 0E4, 613 759-7148, [email protected]
3. Canada Centre for Remote Sensing, Project Manager, Canada Centre for Mapping and Earth Observation, NRCan,
560 Rochester Street, Ottawa, ON, Canada, K1A 0E4 ,613 759-7599, [email protected]
4. Canada Centre for Remote Sensing, Research Scientist, Canada Centre for Mapping and Earth Observation, NRCan,
560 Rochester Street, Ottawa, ON, Canada, K1A 0E4613 759-1186, [email protected]
5. Canada Centre for Remote Sensing, Section Head, Canada Centre for Mapping and Earth Observation, NRCan, 560
Rochester Street, Ottawa, ON, Canada, K1A 0E4613 715-5324, [email protected]
6. Management Services and International Affairs, Deputy Director, Strategic Policy and Operations Branch, NRCan,
588 Booth Street, Ottawa, ON, Canada, K1A 0Y7 613 947-1283, [email protected]
7. Canada Centre for Remote Sensing, Physical Scientist, Canada Centre for Mapping and Earth Observation, NRCan,
560 Rochester Street, Ottawa, ON, Canada, K1A 0E4
ABSTRACT
The Canada Centre for Mapping and Earth Observation (CCMEO), Natural Resources Canada (NRCan),
is currently leading three projects that are supported by the Canadian Space Agency’s Data Utilization
and Application Program (DUAP) for the RADARSAT Constellation Mission (RCM). These projects
seek to facilitate the use of RCM data streams by providing users with new and readily accessible
value-added products (VAPs) and services. The projects are:

The development and delivery of on-demand RCM VAPs via the Earth Observation Data
Management System. Specifically, the project will identify, develop and operationalise a suite
of VAPs that range from low-level, pre-processing products with a wide user base; mid-level
products such as orthorectified imagery and polarimetric decompositions to enable broad use
of RCM’s new Compact Polarimetric beam modes; and potentially high-level downstream
applications products.

The development and delivery of on-demand RCM ground deformation products based on
advanced Interferometric Synthetic Aperture Radar (InSAR) technology. This project will
develop a framework for automatic generation of standard and advanced deformation products
based on InSAR technology from RCM data. Through automation it will be possible to extend
the mapping of surface deformation to non-SAR experts.

The transition of CCMEO’s Emergency Geomatics Service from a RADARSAT-2 based service
to the RCM, to enable provision of RCM products and services in near-real time; primarily
open water flood extent polygons, ice breakup condition maps and infrastructure damage
assessments.
This presentation will give an overview of the projects, provide preliminary results where available,
and map the way forward over the projects’ five-year duration.
36
RADARSAT Constellation Mission: Opportunity to Expand Soil Moisture
Information
Jarrett Powers1*, Heather McNairn2, Amine Merzouki3, Anna Pacheco4, Catherine Champagne5, Allan
Howard6
1. Manager, Agriculture & Agri-Food Canada – Science & Technology Branch, Manitoba Knowledge & Technology
Transfer, 200-303 Main Street, Winnipeg, Manitoba, Canada, (204) 259-4010, [email protected]
2. Research Scientist, Agriculture & Agri-Food Canada – Science & Technology Branch, Eastern Cereal & Oilseed
Research Centre, 960 Carling Avenue, Ottawa, Ontario, Canada, (613) 759-1815, [email protected]
3. Physical Scientist, Agriculture & Agri-Food Canada – Science & Technology Branch, Eastern Cereal & Oilseed
Research Centre, 960 Carling Avenue, Ottawa, Ontario, Canada, (613) 759-1815, [email protected]
4. Agroclimate Specialist, Agriculture & Agri-Food Canada – Science & Technology Branch, Eastern Cereal & Oilseed
Research Centre, 960 Carling Avenue, Ottawa, Ontario, Canada, (613) 694-2411, [email protected]
5. Earth Observation Scientist, Agriculture & Agri-Food Canada – Science & Technology Branch, National
Agroclimate Information Service, 960 Carling Avenue, Ottawa, Ontario, Canada, (613) 715-5255,
[email protected]
6. Manager, Agriculture & Agri-Food Canada – Science & Technology Branch, National Agroclimate Information
Service, 2010 12th Avenue, Regina, Saskatchewan, Canada, (306) 523-6789, [email protected]
* Presenting Author: Jarrett Powers, Manager, 204 259-4010, [email protected]
ABSTRACT
Access to timely and accurate soil moisture data is an important data variable for weather forecasting and
climate modeling, hydrology and flood forecasting and biophysical systems. The information spans the
sectors of agriculture, forestry, environment, water management and public health and safety. For
agriculture, soil moisture is vital for predicting irrigation water needs, forecasting crop production and
monitoring risks associated with floods/excess moisture, drought, pests and crop diseases.
Agriculture and Agri-Food Canada (AAFC) has worked extensively to successfully develop the processes to
derive soil moisture information from the RADARSAT-2 sensor on non-vegetated agricultural soils.
Through partnerships with the Canadian Space Agency (CSA), Environment Canada, Universities and other
organizations, AAFC is continuing to develop new approaches to derive soil moisture throughout the crop
season, evaluate and adapt these approaches using other SAR sensors such as the RADARSATConstellation Mission (RCM), and investigate the application of the information for agricultural uses.
This presentation will provide an overview of AAFC’s plans to adapt current soil moisture retrieval
processes to future RCM operations under CSA’s Data Utilization and Application Program. It will also
examine efforts undertaken to streamline data processing, allowing AAFC to more readily produce soil
moisture products; provide an overview of AAFC’s in-situ soil moisture monitoring network; and discuss
examples of agricultural applications for the information.
37
RCM Data Utilization & Application Plan (DUAP) Activities at the Canadian Ice
Service
Leah Braithwaite1*, Matthew Arkett1, Marilee Pregitzer1, Angela Cheng1, Tom Carrieres1, Lynn Pogson1,
David Bradley2, Shahid Khurshid2, and Shannon Kaya2
1. Environment Canada-Canadian Ice Service, 373 Sussex Drive, Ottawa, ON, Canada
2. Earth Observation & Geomatics, Meteorological Service of Canada, 373 Sussex Drive, Ottawa, ON, Canada, (613)
996-8637, [email protected]
* Presenting Author: Leah Braithwaite, 613 947-7514, [email protected].
ABSTRACT
Building on the success of the RADARSAT-1 & 2 missions, the RADARSAT Constellation Mission (RCM)
primary objectives are to provide greatly improved operational capability, ensure data continuity for
existing users of current RADARSAT missions, as well as adding a new series of applications enabled
through the constellation approach. With three satellites, RCM will offer more frequent observations of
the Canadian territory and its surrounding waters day and night under any weather conditions in support
of Government of Canada (GoC) priorities.
The national daily coverage and fast revisit capability offered by RCM will significantly increase Synthetic
Aperture Radar (SAR) data acquisition from current RADARSAT utilization by the Canadian Ice Service (CIS).
Recognizing that current infrastructure and processing techniques to transform data into usable
information product may not be adequate to ingest this additional data flow, the CIS will be spending the
next five years ensuring that their operational programs will be ready to make optimum use of the SAR
data when the system becomes operational.
The Data Utilization and Applications Plan (DUAP), a key component of the RCM Major Crown Project
(MCP), provides the framework of the technical and financial assistance to GoC departments to upgrade
their operational applications using SAR data and integrate new ones exploiting the new capabilities
offered by the RCM. These applications are needed to transform the data into value-added information
required by user departments to realize their mandate.
This paper will provide an overview of the DUAP activities being undertaken at CIS and will describe the
associated research and development in ice, oil and iceberg monitoring.
38
S8 Marine Remote Sensing / S8 Télédétection des milieux marins
Anders Knudby and Ian Mcdermott, co-chairs / modérateurs
Landsat-based kelp mapping in British Columbia
Anders Knudby1*, Olympia Koziatek2
1. Assistant Professor, Simon Fraser University, 8888 University Dr., Burnaby, British Columbia, Canada, +1-778782-3876, [email protected]
2. MSc student, Simon Fraser University, 8888 University Dr., Burnaby, British Columbia, Canada, +1-604-916-1513,
[email protected]
* Presenting Author: Anders Knudby, Assistant Professor, 778-782-3876, [email protected]
ABSTRACT
The coast of British Columbia is home to large stands of giant kelp (Macrocystis integrifolia) and bull kelp
(Nereocystis luetkeanaa). Both play a key role in coastal ecosystem dynamics, are sensitive to changes in
ocean climate, and support a small harvesting industry. However, the spatial extent of kelp forests in
British Columbia is unknown outside pilot sites, as is any long-term change trajectory or short-term
variability. As a first step in establishing baseline information on kelp forests in British Columbia, we report
on the effectiveness of two methods that have been used successfully to map another species of giant
kelp (Macrocystis pyrifera) in California, using Landsat data. The first method applies a land-water mask
and then uses a simple NDVI threshold to separate kelp from other water areas in a binary classification.
The second method uses multiple endmember spectral mixture analysis (MESMA) to assign per-pixel
percentage kelp cover, on which a threshold is then applied to produce a binary classification. Validation
using 2010 field data from Clayoquot Sound on the west coast of Vancouver Island suggests that the two
methods produce similar map accuracies when calibrated to a single scene, but also that MESMA
processing can be automated and thus has greater potential for time series analysis and upscaling.
39
Mapping of Boulders in Seismically Heterogeneous Marine Soils
Adam O. Gogacz1*, Shervin Azad2, Jacques Y. Guigné3, Ryan Laidley4, Ian McDermott5
1. Acoustic Zoom Inc., 685 St. Thomas Line, Paradise, NL, Canada, (709) 895-6088, [email protected]
2. Acoustic Zoom Inc., 685 St. Thomas Line, Paradise, NL, Canada, (709) 895-6088, [email protected]
3. Acoustic Zoom Inc., 685 St. Thomas Line, Paradise, NL, Canada, (709) 895-6088, [email protected]
4. PanGeo Subsea Inc., 430-434 Water St., St. John’s, NL, Canada, (709) 739-8032, [email protected]
5. PanGeo Subsea Inc., 430-434 Water St., St. John’s, NL, Canada, (709) 739-8032, [email protected]
* Presenting Author: Adam O. Gogacz, (709) 895-6088 ext 201, [email protected]
ABSTRACT
The presence of boulders, defined as rock fragments with diameter larger than 0.3 meters, poses an
engineering risk to the construction of marine structures (e.g. offshore wind-turbines or oil and gas
platforms) due to either total rejection or deflection of driven foundational piles. The glacial tills found in
the North Sea, Baltic Sea, Irish Sea, and the Grand Banks bear a substantial number of boulders; these are
also regions with increasing oil and gas as well as renewable energy sector activities. Hence, a new
technology is required to mitigate the risk in the foundation construction as posed by the boulders.
The Acoustic Corer™ is a purpose-built technology designed to acquire a scattering (acoustic) response of
the sub-seafloor on a dense grid; this acquisition and processing methodology has been thoroughly tested
in the above-mentioned marine environments. Processing of the data is based on a (acoustic)
backprojection operator obtained from the first term of the asymptotic expansion of Green’s function (ray
theory) and the Born approximation for the scattered field. This processing method, akin to and otherwise
known as beamforming or migration, requires that a background compressional-wave (P-wave) speed of
the soil be known a priori. The discussed method shows how to determine the background P-wave model
based on user-driven diffraction-focusing technique. This technique is illustrated on synthetics as well as
real data (Grand Banks) with a discussion of failure to detect boulders if the background P-wave model is
not sufficiently accurate.
40
Testing the Capabilities of a Landsat Time Series to Assess Changes in Marine
Protected Areas in Eastern Africa
Anders Knudby1*, Candace Newman2, Johan Eklöf3, Martin Gullström4
1. Assistant Professor, Simon Fraser University, 8888 University Dr., Burnaby, British Columbia, Canada, +1-778782-3876, [email protected]
2. Postdoc, Department of Ecology, Environment and Plant Sciences, Stockholm University, SE- 106 91 Stockholm,
Sweden, +1 (778) 938-8490, [email protected]
3. Associate Professor, Department of Ecology, Environment and Plant Sciences, Stockholm University, SE- 106 91
Stockholm, Sweden, +46 (0) 8 16 13 58, [email protected]
4. Associate Professor, Department of Ecology, Environment and Plant Sciences, Stockholm University, SE- 106 91
Stockholm, Sweden, +46 (0) 8 16 37 75, [email protected]
* Presenting Author: Anders Knudby, Assistant Professor, 778-782-3876, [email protected]
ABSTRACT
The implementation of Marine Protected Areas (MPAs) around the world continues to grow with recent
commitments from developed and developing states that will see several marine areas each larger than
200,000 square kilometers receive protection from fishing, industrial development and other
anthropogenic pressures. The reason for this continued expansion of marine protection is the vast
empirical evidence that demonstrates well-supported MPAs lead to higher fish biomass, species diversity,
and numbers of large organisms both inside and outside the protected area. In Eastern Africa, and many
parts of the world, the spatial extent and composition of these variables have been documented using
numerous remotely-sensed techniques; however, to-date there has been limited analysis of the long-term
change of these variables. The Landsat imagery archive, which contains images from 1972 and covers
most coastal regions of the world, could provide an invaluable data source for assessing long-term
changes. We assess the capabilities of Landsat imagery to detect changes in the spatial extent and
composition of valued substrates protected using MPAs off the coast of Eastern Africa, and report on the
extent to which change analysis can be automated, as application of a standard change analysis technique
is desired for multiple sites in subsequent project stages.
41
Results from the first topo-bathymetric lidar surveys of the Chiroptera II sensor
for coastal and freshwater sites in Maritime Canada
Nathan Crowell1*, Kevin McGuigan1, Kate Collins1, Candace MacDonald1 and Tim Webster2
1. Research Associate, Applied Geomatics Research Group, Nova Scotia Community College, Middleton, Nova
Scotia, Canada, 902 825 5478 email:[email protected], [email protected],
[email protected], [email protected], [email protected]
2. Research Scientist, Applied Geomatics Research Group, Nova Scotia Community College, Middleton, Nova Scotia,
Canada, 902 825 5475 email:[email protected]
* Presenting author: Nathan Crowell 902-825-5478 [email protected]
ABSTRACT
The Applied Geomatics Research Group (AGRG) within the Nova Scotia Community College (NSCC)
acquired a new shallow water airborne topo-bathymetric lidar sensor, the Chiroptera II, and flew the first
missions in September 2014. The survey areas consisted of several embayments along the
Northumberland Strait, an area on the Atlantic coast and two freshwater lakes. Many of the areas are
sheltered bays that host shellfish aquaculture farms. The low flow rates associated with these bays
promote high volumes of sediment cover and the presence of glacial till along the coast promotes fine
grained near shore sediments. These sediments can be re-suspended in the water column during periods
of increased wave activity. As a result, one must assess both the current environmental conditions (wind,
cloud ceiling height) as well as the water clarity conditions that may be influenced by previous conditions
such as strong winds, thus adding additional operational constraints. The reflectance of the seabed also
influences the maximum depth achieved. Having multiple study areas to choose from during a
bathymetric lidar survey campaign allows flexibility to survey areas of optimal water clarity. Results
indicate depth penetration of 5-6 m for most of the study sites and the sensor is ideal for differentiating
the water surface from the seabed at depths less than 2 m which was problematic in earlier lidars. In
addition to generating seamless DEMs of the study areas, reflectance of the green laser is used in
combination with elevation variables to map submerged aquatic vegetation.
42
Buried UXO Imaging Survey with a Sub-Bottom Imager™
Alison Brown1*, and Ryan Laidley2, Ian McDermott3
1. Geoscientist, PanGeo Subsea Inc, 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032 ext. 238,
[email protected]
2.Geoscientist, PanGeo Subsea Inc., 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032 ext. 242,
[email protected]
3. Lead Geoscientist, PanGeo Subsea Inc, 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032 ext.
229, [email protected]
* Presenting Author: Alison Brown, Geoscientist, 709-739-8032 ext. 238, [email protected].
ABSTRACT
PanGeo Subsea was contracted to carry out a survey to locate a buried unexploded ordinance (UXO) target
within a dockyard area. Previous attempts to locate the UXO using electromagnetic methods had been
unsuccessful. The UXO was known to be 5.8m long by 0.5m diameter. The seafloor in the area consisted
of 6m of soft sediment overlying a harder clay layer. The UXO was thought to have sunk through the soft
sediment, resting on the underlying clay.
PanGeo Subsea deployed their Sub-Bottom Imager™ (SBI) to carry out the survey. The SBI is an acoustic
imaging tool designed to be mounted on an ROV, drag arm, or similar mobile subsea platform. It uses
chirp sonar transmission, beamforming, and synthetic aperture processing to interrogate the sub-seabed
and form a 3D image of any buried objects or structures such as UXO, boulders, and geological formations.
The survey operations were complicated by the shallow water depth and proximity to the quayside, and
so the standard method of deployment from an ROV was unsuitable. Instead, the SBI assembly was
deployed from a mobile crane station on the quayside. In one area with particularly limited
maneuverability, the SBI was towed by a dive boat. The target was identified together with several other
non-UXO objects. SBI acoustic images and their interpretation are presented.
43
S9 Airborne Imaging and Calibration / S9 Imagerie aéroportée et
calibration
Raymond Soffer, chair / modérateur
Boreal forest and peatland classification integrating GEOBIA and Hyperspectral
Imagery
Margaret Kalacska1*, Carlomagno Soto1, J. Pablo Arroyo-Mora1,2, Ian Strachan3
1. Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal QC, Canada H3A
2K6
2. Geographic Information Centre, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal QC,
Canada H3A 2K6
3. Department of Natural Resource Sciences, McGill University, Macdonald Campus 21111 Lakeshore Rd., Ste Anne
de Bellevue, QC H9X 3V9
* Presenting Author: Margaret Kalacska, Associate Professor, McGill University, 514 398-4455,
[email protected] ,
ABSTRACT
Precise mapping of boreal ecosystems is challenging from remotely sensed data; in particular the
separation of classes such as forest and treed peatland. In order to overcome these challenges, analysis
of hyperspectral imagery has shown to be an effective alternative to multispectral satellite imagery. In
this study we combine airborne hyperspectral imagery with low altitude aerial photographs to
systematically classify boreal forest and peatland. Photographs captured with a Canon T1i DSLR (Canon
EF-S 18-55mm lens) were mosaicked and geometrically corrected to a final spatial resolution of 8cm,
which we define as Very High Resolution Aerial Photographs (VHRP).
Using Object-Oriented Image Analysis seven land cover classes were extracted from the VHRP. The VHRP
mosaic classification was carried out in eConition 8.2 by developing a series of binary classification rulesets
taking into account segment statistics such as the mean brightness, hue and saturation, among others.
Using ENVI 5.2 the classes derived from the object-oriented approach were used to train and classify
airborne hyperspectral imagery. Results show an overall accuracy of >80% suggesting that the integration
of the object oriented approach with the hyperspectral imagery improves the accuracy of landcover
classification in boreal forest and peatlands.
44
Bundle Adjustment of VNIR, SWIR and LWIR Airborne Hyperspectral Systems at
Rothera Research Station, Antarctic Peninsula
Stephen Achal1*, Hugh Corr2, John McFee3, Andrew Fleming4, Ashley Tam5 and Rita Leung6
1. Chief Scientist, ITRES Research Limited, Calgary, Canada
2. Glaciologist, British Antarctic Survey, Cambridge, UK
3. President, McFysics, Medicine Hat, Canada
4. Environmental Scientist, British Antarctic Survey, Cambridge, UK
5. Principal Geomatics Engineer, ITRES Research Limited, Calgary, Canada
6. Geomatics Research Engineer, ITRES Research Limited, Calgary, Canada
* Presenting Author: Stephen Achal, Chief Scientist, (403) 250 9955, [email protected]
ABSTRACT
In February 2011, VNIR, SWIR and LWIR hyperspectral imaging systems (CASI-1500, SASI-600, and TASI600 respectively) were installed in a Twin Otter, equipped with a large aerial survey port, at the Rothera
Research Station, Antarctic Peninsula. The installation-dependent position and angular offsets between
each system’s entrance pupil and the IMU center and GPS antennae were iteratively derived from data
acquired during a bundle adjustment (BA) flight.
Typically, BA flights are preformed over areas containing well-distributed arrays of precisely surveyed
ground control points (GCPs). The corners and edges of easily accessible anthropogenic structures
(buildings, roads, etc.) with high contrast in the VNIR, SWIR and LWIR serve as ideal GCPs.
Unfortunately, in remote and extreme environments, such as Antarctica, there are very few natural and
anthropogenic GCPs that are clearly identifiable in the VNIR, SWIR and LWIR. Predictably, the area
surrounding Rothera Research Station contains few GCP candidates.
Thus, a proof of concept scheme involving distributing sixteen field-deployable GCPs made of Aluminized
Mylar™ within the BA area was attempted. The Aluminized Mylar™ GCPs were clearly identifiable in the
data generated by all three instruments during the BA flights (consisting of fourteen north-south flight
lines and nine east-west lines crisscrossing a nine square kilometer area). This facilitated the accurate
derivation of the BA parameters for the hyperspectral systems’ installation.
The BA parameters were subsequently used with in-situ IMU data, DGPS and DEM to ortho-rectify and
mosaic the vast quantity of high-spatial resolution VNIR, SWIR and LWIR hyperspectral data collected
during the following weeks.
45
Atmospheric correction of airborne hyperspectral imagery in diverse tropical
ecosystems – lessons learned from Mission Airborne Carbon 13
Margaret Kalacska1*, J. Pablo Arroyo-Mora2, Raymond J. Soffer3
1. Associate Professor, Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705,
Montreal QC, Canada H3A 2K6, [email protected]
2. Director, Geographic Information Centre, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal
QC, Canada H3A 2K6, [email protected]
3. Research Council Officer, Flight Research Laboratory, National Research Council, Ottawa, Ontario, Canada,
(613)998-5341, [email protected]
* Presenting Author: Margaret Kalacska, Associate Professor, (514)398-4347,
[email protected]
ABSTRACT
Across the tropics, forest clearing and anthropogenic forest disturbances such as logging are an important
source of gross carbon emissions. Assessments of these land cover changes as well as the quantification
of biomass/aboveground carbon stocks are a fundamental requirement of the REDD+ mechanism. A
critical step in the derivation of biomass products from imagery is the preprocessing of the data, a key
aspect of which is atmospheric correction. In this study we examine the technical aspects of the
atmospheric correction of airborne hyperspectral imagery (CASI-1500 and SASI) using data collected
through the Mission Airborne Carbon 13 (MAC13) project. The primary aim of MAC13 is to provide
aboveground biomass/carbon accounting from airborne hyperspectral imagery in five highly diverse
ecosystems in Costa Rica.
We focus on the inclusion of the on-site aerosol optical depth data and simulations with MODTRAN, the
utility of MODIS aerosol products as well as potential illumination effects in the ground spectra and the
subsequent impact on the vicarious calibration of the imagery. We compare the results obtained from
FLAASH and ATCOR to determine the effect of ecosystem (dry vs wet) on the performance of the
atmospheric correction modules. Careful consideration must be given to the specifics of each ecosystem
in order to obtain the best results from the atmospheric correction module. Furthermore, the importance
each of these aspects varies by ecosystem type and topographic characteristics.
46
SNR Implications of On-chip vs Off-chip Pixel Summation
Raymond J. Soffer1*, Natalie Cornish2, Margaret Kalacska3
1. Research Council Officer, Flight Research Laboratory, National Research Council, Ottawa, Ontario, Canada,
(613)998-5341, [email protected]
2. Research Assistant, Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705,
Montreal QC Canada H3A 0B9
3. Associate Professor, Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705,
Montreal QC Canada H3A 0B9
* Presenting Author: Raymond Soffer, Research Council Officer, (613)998-5341, [email protected].
ABSTRACT
Certain airborne hyperspectral imaging systems allow the operator to customize the configuration for the
number of channels, their locations, and bandwidths. Summing elemental pixels into broader channels
within the imaging system is commonly applied to improve the Signal to Noise Ratio (SNR). Alternatively
it may be desirable to acquire imagery at the highest possible spectral resolution while allowing for the
possibility of summing in post-processing. In this study, the impact on the SNR of summing pixels on- vs
off-chip is evaluated.
In Charge-Coupled Device (CCD) based systems, the noise model for an individual pixel is a combination
of the read-noise, as introduced by the output amplifier, and the shot-noise, which is proportional to the
square root of the number of photo- and dark-induced electrons produced within that pixel. When charge
is summed on-chip, a single read noise impacts the signal. When pixels are read out individually and
summed in post-processing, each pixel to be summed contributes read noise to the resulting pixel.
In this presentation we will illustrate the development of a noise model describing the on- and off-chip
SNRs with the results validated against laboratory acquired CASI-1500 data of a radiometrically stable
integrating sphere. Examples of SNR spectra derived from real world input spectra will be provided
demonstrating that for situations in which the read noise is significantly small as compared to the shot
noise, summing in post-processing may be achieved with minimal negative impact to the resulting SNR.
47
S10 Innovative Earth Observation Techniques Supporting Informed Oil
& Gas Development – II / S10 Techniques d’observation de la Terre
innovatrices en soutien au développement pétrolier et gazier – II
Douglas Bancroft, chair / modérateur
Utilization of Earth Observation Technologies to Assist Incident Response and
Investigation
Subir Chowdhury 1*, Todd Shipman 2, Greg Schroter 3, and Dennis Chao 4
1. Remote Sensing Specialist, Alberta Geological Survey, Alberta Energy Regulator, 4999–98 Avenue NW , 4th Floor,
Edmonton, Alberta, Canada, phone: 1-780-427-4115, e-mail: [email protected]
2. Manager, Alberta Geological Survey, Alberta Energy Regulator, 4999–98 Avenue NW , 4th Floor, Edmonton,
Alberta, Canada, phone: 1-780-644-5563, e-mail: [email protected]
3. Emergency Response Coordinator, Alberta Energy Regulator, 4999–98 Avenue NW , 2nd Floor, Edmonton,
Alberta, Canada, phone: 1-780-427-1516, e-mail: [email protected]
4. GIS Specialist, Alberta Geological Survey, Alberta Energy Regulator, 4999–98 Avenue NW , 4th Floor, Edmonton,
Alberta, Canada, phone: 1-780-644-5563, e-mail: [email protected]
* Presenting Author: Subir Chowdhury, Remote Sensing Specialist, Phone: 1-780-427-4115, Email:
[email protected].
ABSTRACT
The Alberta Energy Regulator (AER)/Alberta Geological Survey (AGS) is investigating Earth Observation
(EO) technologies to assist incident response and investigation associated with hydrocarbon, produced
water, and hazardous gas releases from oil and gas activities. EO datasets are becoming widely available
for download and in many cases they are free. AER/AGS has an opportunity to utilize this free resource
and to develop the capacity to incorporate this technology as it is introduced.
Over past three decades, Landsat data has evolved as a valuable resource for decision makers in various
diverse fields, e.g., environmental monitoring, emergency response, agriculture, forestry, land use, water
resources, and natural resource exploration. In addition, spaceborne sensors, e.g., Ozone Monitoring
Instrument (OMI) currently provides information regarding ambient atmosphere composition. In this
study, application of Landsat multispectral datasets are demonstrated for two instances to assist
investigation of hydrocarbon and produced water spills. In addition, utilization of OMI hyperspectral
datasets for two instances of sulfur dioxide anomalies and two instances of formaldehyde anomalies are
shown as possible precursors of hazardous gas release from oil and gas activities. Biomass trends and subpixel level information were used to assess the aftermath of bitumen and produced water release as well
as to examine if there was any release before the incident was reported. Biomass trends were computed
using the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Moisture Index
(NDMI). In addition, sub-pixel level information was extracted using normalized spectral unmixing and the
Spectral Angle Mapper (SAM) target detection algorithm. For OMI sulfur dioxide and formaldehyde data,
time-averaged column amounts were used to produce the map and time-series column amounts were
used to produce area statistics.
This study demonstrated that extracted information from publicly available multispectral and
hyperspectral data can provide evidence for incident response and investigation to support good decisionmaking and to protect public and environment. In addition, it shows the potential of EO technology to
48
track performance measures, e.g., the number of incidents per kilometre of pipelines regulated by the
AER.
49
Monitoring Pipeline Right-of-way Encroachment Using Multi-temporal High
Resolution SAR
Zhaohua Chen1*, Desmond Power2, Michael Lynch3
1. Remote Sensing Research Scientist, C-CORE, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL, Canada,
A1B 3X5, Tel: 709-864-7264 Email: [email protected]
2. Vice President, C-CORE, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL, Canada, A1B 3X5, Tel: 709-8648353 Email: [email protected]
3. Remote Sensing /GIS analyst, C-CORE, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL, Canada, A1B
3X5, Tel: 709-864-3727 Email: michael.lynch @c-core.ca
* Presenting Author: Zhaohua Chen, PhD, Remote Sensing Research Scientist, 709-864-7264
[email protected]
ABSTRACT
Monitoring and identification of right of way intrusions has been a priority to oil and gas industry to
prevent pipeline failure, and to reduce the loss of life and property. Application of radar satellite imagery
may offer a cost-effective alternative to field observations and air patrol. Monitoring the oil pipeline
encroachment activities usually relies on visual interpretation or change detection based on comparison
of bi-date SAR data. Traditional mean ratio change detection and coherence change detection methods
using SAR satellite images could be affected by neighborhood backscatter variations or the loss of
interferometric coherence due to the phenological change, unstable surface, temporal decorrelation or
atmospheric interferences thus produce detection results with high false alarm rate. In this study, we
present a patch based change detection method using multi-temporal high resolution SAR. This approach
exploits information from multi-temporal acquisition to improve the detection accuracy, and provide both
short-term change and long-term change trend information. The procedure involves three steps: multitemporal SAR speckle reduction, patch based change detection and profile analysis using SAR backscatter.
An experiment has been tested on monitoring activities along pipeline in Alberta, Canada by using
COSOMO-SkyMed imagery. The results of this proposed approach are compared with traditional mean
ratio detection method and coherence change detection method. The results show that the proposed
method can provide satisfied results.
50
Surface water and wetland monitoring in the Peace-Athabasca Delta
Alexander Chichagov 1*, Brian Brisco2, Kevin Murnaghan3 and Lori White4
1. Remote Sensing Scientist, Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester St.,
Ottawa, Ontario, Canada, K1A 0E4, (613) 715-5056, [email protected]
2. Research Scientist, Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester St., Ottawa,
Ontario, Canada, K1A 0E4, (613) 759-1046, [email protected]
3. Remote Sensing Scientist, Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester St.,
Ottawa, Ontario, Canada, K1A 0E4, (613) 759-6237, [email protected]
4. Environmental Scientist, Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester St.,
Ottawa, Ontario, Canada, K1A 0E4, (613) 759-6485, [email protected]
* Presenting Author: Alexander Chichagov, Remote Sensing Scientist, 613 715-5056,
[email protected].
ABSTRACT
Water is an important resource and information about surface water conditions as well as water level can
help a number of applications including hydrology, meteorology, ecology, and agronomy. Ample
quantities of water are needed for healthy wetlands which are critical for balancing water flows and
maintaining water quality.
Synthetic Aperture Radar (SAR) has been is used operationally for flood mapping in Canada since the
launch of RADARSAT-1 in 1985 and many other countries have adopted followed our approach for flood
mapping. RADARSAT-2 (wide ultra-fine mode) data was processed and analysed to provide spatially and
temporally dynamic surface water mapping products in Alberta’s Peace Athabasca Delta (PAD).
With the increased availability of seasonal multi-temporal SAR data, new approaches can be used for noise
reduction such as temporally filtering the data to reduce speckle and increase interpretability. The multitemporal surface water products masks are generated using a thresholding technique which can be
applied to flood response situations or for monitoring seasonal changes in surface water.
Previously multi-temporal surface water extent products generated using image thresholding approaches
were delivered as well as relative water level change products. These products were generated using
interferometric techniques (InSAR), with the use of the in-situ level logger data for the phase ambiguity
resolution, and can be used for monitoring surface water and relative water level variations over space
and time. This will further enhance the use of SAR for monitoring water resources and producing flood
response products which include surface water, water level and flooded vegetation.
51
Using ArcGIS as an Alternative or Complimentary Tool to ENVI for Extracting
Anthropogenic Footprints from Multispectral Imagery
Todd Shipman1*, Subir Chowdhury2 and Dennis Chao3
1. Manager, Alberta Energy Regulator/Alberta Geological Survey, 4th floor, 4999 – 98 Ave, Edmonton, Alberta,
Canada, (780) 644-5563, [email protected]
2. Remote Sensing Specialist, Alberta Energy Regulator/Alberta Geological Survey, 4 th floor, 4999 – 98 Ave,
Edmonton, Alberta, Canada, (780) 427-4115, [email protected]
3. GIS Specialist, Alberta Energy Regulator/Alberta Geological Survey, 4 th floor, 4999 – 98 Ave, Edmonton, Alberta,
Canada, (780) 427-0107, [email protected]
* Presenting Author: Todd Shipman, Manager, (780) 644-5563, [email protected]
ABSTRACT
Alberta Geological Survey/Alberta Energy Regulator is investigating how to manage rapid growth within
the oil and gas industries though out Alberta. We are utilizing satellite earth observation data such as
multispectral imagery to extract anthropogenic footprints relating to exploration activities, such as roads,
well pads, processing facilities and access corridors to pump sites, etc. Our study area is about 200km
northwest of Edmonton, near Fox Creek where industries are producing and exploring both conventional
and unconventional oil and gas. All information is geospatial and integration of this information
contextually will help to develop better decision making.
Publicity available Landsat8 multispectral imagery (2013 and 2014) from US Geological Survey and high
resolution (6m) multispectral SPOT6 imagery for 2013 are used for this analysis. In addition, BestAvailable-Pixel (BAP) composite of Landsat multispectral image stacks (2005-2012) provided by Natural
Resources Canada.
Landsat8 and SPOT6 imagery are processed as individual scenes using Image Classification functions and
different Spatial Analyst filters in ArcGIS 10.1. Anthropogenic footprints from Landsat 8 imagery are also
extracted using NDVI and co-occurrence-based texture analysis using ENVI.
Results produced by ArcGIS for Landsat8 imagery are similar to ENVI with high degree of accuracy when
identifying footprints bigger than 100m and have difficulties detecting linear footprints such as access
corridors due to image resolution. Traditional supervised classification algorithm in ENVI, on the other
hand, is computationally expensive to process SPOT6 data because high spatial resolution. This problem
is resolved by using ArcGIS Spatial Analyst functions with high detection accuracy. Results are then
processed in ENVI using post classification techniques (i.e., clump and sieve) to further refine footprint
detection.
ArcGIS is viable alternative to dedicated remote sensing package such as ENVI for detecting
anthropogenic footprints from remote sensing imagery. In addition, it can process high resolution
imagery such as SPOT6 more efficiently and to be used for post-processing in ENVI to enhance
classification accuracy.
52
Systematically Derived Leaf Area Index of the Canadian Oilsands Region
Matthew Maloley1*, RichardFernandes1, Francis Canisius1
1. Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester, Ottawa, Canada
* Presenting Author: M. J. Maloley, Physical Scientist, Canada Centre for Remote Sensing, Natural
Resources Canada, 613-759-7993, [email protected]
KEY WORDS: Leaf Area Index, Oilsands, SPOT, Landsat, LiDAR, Hyperspectral,
ABSTRACT
The 140,200km2 oil sands region of central Canada has the 3rd largest proven reserves of oil in the world.
Currently, surface mining is concentrated in a ~600km2 region north of Fort McMurray, Alberta, Canada.
A comprehensive monitoring effort led by the Governments of Canada and Alberta is underway to track
the status of the ecosystem in the vicinity of this surface mining region. Many of these models require
systematically derived time series of gridded land surface parameters at approximately 1km
resolution together with quantified uncertainty. Leaf area index (LAI) is one such parameter that is being
systematically derived at various spatial and temporal resolutions. In order to validate coarser 1km
resolution MODIS and VGT LAI products, a 20 m resolution map of 2012-13 peak season LAI was produced
using a combination of in situ LAI estimates and vegetation indices derived from SPOT 5 satellite imagery.
A reduced simple ratio vegetation index was produced for each image to be used in the LAI processing. A
total of 204 ground plots were sampled across the study area in July 2012 and August 2013 for the
purposes of calibration and validation. Additionally, airborne hyperspectral and LiDAR data at
approximately 2m resolution were collected for portions of the study area over the summer of 2013 and
compared in LAI analysis. Using the 2m and 20m resolution LAI estimates, the performance of coarse
spatial resolution but high temporal resolution (i.e. 10 day composite) LAI products were assessed to
determine their performance as potential indicators of disturbance from emissions related to mining
activities.
53
S11 Characterization and Change in Forest Ecosystems / S11
Caractérisation et changements dans les écosystèmes forestiers
Ron Hall, chair / modérateur
Assessment of Vegetation Conditions over Athabasca Oil Sands Region
Rasim Latifovic1*, Darren Pouliot2
1. Research Scientist, NRCan/Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613-759-7002, [email protected]
2. Physical Scientist, NRCan/Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613-759-6341, [email protected]
* Presenting Author: Rasim Latifovic, . Research Scientist, 613-759-7002, [email protected]
ABSTRACT
The Canada Centre for Remote Sensing (CCRS) has a long history in the development of satellite remote
sensing applications for land surface characterization and monitoring. CCRS methods have been used to
produce datasets leading to improved understanding of complex interactions between the biosphere and
atmosphere, for modelling the effects of climate and land cover change on groundwater recharge, and
quantitative landscape studies assessing spatial and temporal land cover dynamics. Resent developments
regarding land surface characterization will be presented to inform and possibly inspire cross-disciplinary
research or the use of available methodologies/data by other scientists.
The priority of environmental research, insofar as responsible natural resources development is
concerned, is to improve our understanding of the long-term consequences that industrial activities may
have on terrestrial ecosystems. This provides for systematic progress towards effective development of
environmental standards and mitigation measures. Satellite remote sensing is an approach that is
becoming widely accepted as an observational and analytical tool for use in the scientific understanding
of issues related to the environment. Some of the topics where remote sensing observations might serve
as an additional source of information include determination of baseline conditions, quantitative impact
assessment, analysis and prediction. The methodologies presented here use vegetation indices from longterm Landsat observations to examine trends in vegetation conditions. The results presented highlight
areas in the AOSR for which more in-depth monitoring should be undertaken, offer a more quantitative
base for future cumulative impact predictions and can inform the development of mitigation measures.
In particular this study investigates cumulative long term changes in vegetation condition over the
Athabasca Oil Sands region of Alberta, Canada, between 1984 and 2012. It includes: i) analysis of the
spatial and temporal pattern of trends as a function of vegetation community, ii) trend confidence based
on statistical significance and selected time period, iii) normalization of trends for growing conditions
(temperature and precipitation) to assess the potential climate change influence, and iv) detailed change
characterization attributing occurrence, magnitude and causation.
54
Long-term Trends in and Interrelationships between Central Boreal Plains
Ecosystem Succession and Hydro-Climatic Conditions
Laura Chasmer1* Chris Hopkinson1 and Richard Petrone2
1. Department of Geography, University of Lethbridge, Lethbridge Alberta, Canada, 403-332-4661,
[email protected]
2. Dept. of Geography and Environmental Management, University of Waterloo, 200 University Ave. W. Waterloo,
Ontario, Canada, 519-888-4567, [email protected]
* Presenting Author: Laura Chasmer, Alberta Innovates Post-Doctoral Fellow, 403-332-4661,
[email protected]
ABSTRACT
In this study we examine the impacts of more than two decades of atmospheric drying on changing
vegetation trends within upland forests and lowland peatlands in the Central Boreal Plains region of northcentral Alberta. An index of annual evapotranspiration (ET) (EI) (actual ET / Precipitation) vs. dryness index
(DI) (potential ET / precipitation) per water year is compared with water runoff from six watersheds to
determine the influence of dry vs. wet periods on basin water balance. Ecosystem sensitivity (resilience
or threshold changes) associated with water balance and topography (SRTM) are examined using the
Normalised Difference Vegetation Index (NDVI) over a 23-year period from Landsat TM. Structural
vegetation changes, especially loss of foliage in upland areas and woody vegetation succession into
peatlands are further examined as a case study using multi-temporal airborne light detection and ranging
(lidar) data in one watershed.
We find that catchments with low (detrended) topographic variability containing mostly peatland land
cover types are highly resilient due to abundant water supply. Despite their resilience, these catchments
are characterised by declining water yield (greater storage) over the period of study and increased
greening (~76% of peatlands). This indicates rapid loss and conversion to riparian/forest land cover types
as peatlands dry. Catchments containing greater topographic variability with larger fractions of uplands
are typically water limited and have experienced significant browning (loss of chlorophyll, leaf cover)
within 74% of upland forests and peatlands. A possible explanation for increased catchment yield
coincident with reduced green foliage cover is a reduction in ET losses.
55
Using pre-fire variables in combination with Landsat difference Normalized Burn
Ratio to improve the accuracy in predicting middle burn severity classes
Ignacio San-Miguel1*, Nicholas C. Coops2 and Dave W. Andison3
1. MSc candidate in Forestry. Integrated Remote Sensing Studio, Department of Forest Resources Management,
2424 Main Mall, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada. Phone: 604-822-6452. Email:
[email protected].
2. Professor and Canada Research Chair in Remote Sensing. Integrated Remote Sensing Studio, Department of
Forest Resources Management, 2424 Main Mall, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada.
Phone: 604-822-6452. Email: [email protected].
3. Adjunct professor and consultant. Bandaloop Landscape-Ecosystem Services, 1011 Hendecourt Road, North
Vancouver, British Columbia, Canada V7K 2X3. Phone: 604-988-0985. Email: [email protected].
* Presenting Author: Ignacio San-Miguel, MSc candidate in Forestry at University of British Columbia,
604-822-6452, [email protected].
ABSTRACT
Ecosystem-Based Management (EBM) has emerged as a dominant paradigm for the management of the
Canadian boreal forest. One of the principles of EBM is approximating the Historical Range of Variability
(HRV) of disturbance patterns to maintain ecosystem’s function. HRV has been represented by wildfire
patterns, which are measured using vegetation severity classes derived from Landsat differenced
Normalized Burn Ratio (dNBR) images. dNBR generally achieves accurate results for low and high severity
classes when compared to ground truth data, however is poorer at accurately estimating different
gradients of moderate severity classes. One possible reason for this issue is a combination of Landsat’s
spatial and radiometric resolution. In this paper we combine a suite of pre-fire landscape variables from
a range of datasets along with dNBR to test for an improved detection of moderate middle mortality
classes.
To do so we compared 6 burn severity classes as defined from Aerial Photo-Interpretation (API) with dNBR
images over 10 fires in Alberta and Saskatchewan, Canada. We compared three sets of analysis 1. using
dNBR as alone, ; 2. combining dNBR with land cover classes; and 3. combining dNBR with forest inventory
data such as pre-fire volume and species. The results suggests that using pre-fire variables from inventory
data, as well as land cover vegetation density classes combined with dNBR information, have shown
improved results in predicting API middle severity when compared to using dNBR alone.
56
The Multisource Vegetation Inventory: Integrating Remote Sensing and Field
Data for Forest Inventory in the Northern Boreal
Ron J. Hall1*, Rob .S. Skakun2, Michael Gartrell5, Andre Beaudoin6, Philip Villemaire7, Craig Mahoney8,
Chris Hopkinson9, Lisa Smith3
1. Research Scientist, NRCan/Canadian Forest Service, 5320-122 Street, Edmonton, Alberta, T6H 3S5, Canada, 780435-7209, [email protected]
2. Physical Scientist, NRCan/Canadian Forest Service, 5320-122 Street, Edmonton, Alberta, T6H 3S5, Canada, 780435-7384, [email protected]
3. Inventory Forester, Dept. of Environment and Natural Resources/Government of Northwest Territories, P.O. Box
4354, Lot 173, Hay River, NWT, X0E 1G3, Canada, 867-874-2009, [email protected]
4. GIS Programmer, NRCan/Canadian Forest Service, 5320-122 Street, Edmonton, Alberta, T6H 3S5, Canada, 780435-7276, [email protected]
5. GIS Programmer Analyst, Natural Resources Canada, 5320 - 122 Street, Edmonton, Alberta T6H3S5, 780-4357276, [email protected]
6. Research Scientist, Natural Resources Canada,1055 Du P.E.P.S. Street, Quebec, QC, G1V 4C7,418-648-3440,
[email protected]
7. Geomatics and Remote Sensing Analyst, Natural Resources Canada,1055 Du P.E.P.S. Street, Quebec, QC G1V
4C7,418-648-4492, [email protected]
8. Post-Doctoral Fellow, University of Lethbridge,4401 University Drive, Lethbridge, Alberta T1K 3M4,403-332-4043,
[email protected]
8. Associate Professor, University of Lethbridge,4401 University Drive, Lethbridge, Alberta T1K 3M4,403-3324586,[email protected]
* Presenting Author: R.J. Hall, Research Scientist, 780-435-7209, [email protected]
ABSTRACT
The Northwest Territories (NWT) has approximately 33 million ha of forest distributed across a territory
128 million ha in size. Describing these forests is challenged by the limited availability of current inventory
data and the relative lack of access over its large geographic extent. Optical remote sensing data such as
from the Landsat Thematic Mapper has been used to generate land cover maps that depict the
composition and spatial distribution of forests in the NWT. More detailed knowledge is needed, however,
about forest structure, timber volume and above ground biomass. The vertical characterization of forests
was determined by scaling field data with samples of airborne LiDAR that were subsequently related to
full waveform satellite LiDAR provided by ICESat-GLAS. Estimates of forest structural attributes at ICESatGLAS footprint locations were subsequently computed and spatially imputed to derive continuous raster
surface maps. The challenge was how to integrate these LiDAR-derived surface maps with the satellite
land cover and limited coverages of existing forest inventory data into a format convenient for spatial
analysis?
The presentation describes the data sources, derived models and methods that form the Multisource
Vegetation Inventory (MVI). The MVI combines a raster to vector polygonization process of the LiDAR
derived attributes, satellite land cover maps and a reclassification of existing NWT inventory data. The
result is uniquely integrated forest information derived from satellite LIDAR, optical satellite and air
photo-based data sources into a single, digital inventory map product.
57
Systematic and Environmental Sensitivities within GLAS Estimates of Canopy
Height and Crown Closure in Northern Canada
Craig Mahoney1*, Chris Hopkinson2, Ron J. Hall3, and Michelle Filiatrault4
1. Post-Doctoral Fellow, Department of Geography, University of Lethbridge, Lethbridge, Alberta T1K 6T5, +1 403
332-4043, [email protected]
2. Professor, Department of Geography, University of Lethbridge, Lethbridge, Alberta T1K 6T5, +1 403 332-4586,
[email protected]
3. Research Scientist, Natural Resources Canada, Canadian Forest Service, Northern Forestry Centre, Edmonton,
Alberta T6H 3S5, +1 780 435-7209, [email protected]
4. Remote Sensing Analyst, Natural Resources Canada, Canadian Forest Service, Northern Forestry Centre,
Edmonton, Alberta T6H 3S5, +1 780 435-7291, [email protected]
* Presenting Author: Craig Mahoney, Post-Doctoral Fellow, 403 332-4043, [email protected]
ABSTRACT
The northern Canadian boreal forest is a remote, complex, fragmented landscape whose characterization
relevant to understanding its role in providing a wide range of ecosystem services. The Geoscience Laser
Altimeter System (GLAS) is ideal for characterization of such landscapes due to its unique ability to sample
large areas in three dimensions, however, data quality is sensitive to system, temporal, and spatial
influences. These sensitivities were explored within the North West Territories (NWT) of Canada using
airborne LiDAR data with the objective of identifying optimal GLAS footprints for subsequent canopy
attribute modeling.
The optimized GLAS data were utilized to obtain spatialized estimates of stand height, and crown closure
over a 200,000 km2 area in the NWT. Results were derived from random forest and k-nearest neighbour
algorithms, and empirically compared to equivalent data measures from airborne LiDAR data.
In addition to providing spatial estimates of 3D canopy attributes within remote northern boreal
ecosystems, the study will illustrate the first stages in the development of a new modeling framework.
The parallelised model implementation approach adopted makes no a priori assumptions about the most
suitable spatialization procedure, while iteratively optimising empirical measures of forest canopy
parameters using whichever model approach produces better predictions.
58
S12 Synergies Between Satellite & Marine Acoustic Remote Sensing /
S12 Synergies entre la télédétection satellitaire et la télédétection
acoustique marine
Rodolphe Devillers and Craig Brown, co-chairs / modérateurs
Multiple methods, maps, and management applications: purpose made maps in
support of Ocean Management
Craig J. Brown
Research Scientist, Applied Research, Nova Scotia Community College, 80 Mawiomi Place, Dartmouth, Nova Scotia,
Canada, 902-491-2174, [email protected]
* Presenting Author: Craig J. Brown, Research Scientist, 902-491-2174, [email protected].
ABSTRACT
The establishment of multibeam echosounders (MBES) as a mainstream tool in ocean mapping has
facilitated integrative approaches towards nautical charting, benthic habitat mapping, and seafloor
geotechnical surveys. The inherent bathymetric and backscatter information generated by MBES enables
marine scientists to present highly accurate bathymetric data with a spatial resolution closely matching
that of terrestrial mapping. A range of post-processing approaches can generate customized map
products to meet multiple ocean management needs, thus extracting maximum value from a single survey
data set.
A number of examples are presented showing how primary MBES bathymetric and backscatter data, along
with a variety of supplementary data (i.e. in situ video and stills, etc.), can be processed using a variety of
methods to generate a series of map products. These maps can provide contextual information on
seafloor geo-morphological conditions, relevant to marine engineering, resource management and other
geological related applications. In addition, methods conventionally used for classification of terrestrialbased multi-spectral data have been tested for classification of the MBES data set to produce seafloor
maps summarizing broad bio-physical characteristics of the seafloor (i.e. Benthoscape maps). These types
of map products are of value for use in many aspects of marine spatial planning. Through the process of
applying multiple methods to generate multiple maps for specific management needs, the efficient use of
survey data sets to maximize the benefit to a wide number of potential end users is demonstrated.
59
Where Land Meets Sea: Towards a Complete Coverage for Terrain Analysis of
Coastal Environments
Vincent Lecours1*, Vanessa Lucieer2, Aaron Micallef3, Margaret Dolan4, Craig Brown5, Evan Edinger6 and
Rodolphe Devillers7
1. Ph.D. Candidate, Marine Geomatics Research Lab, Department of Geography, Memorial University of
Newfoundland, 232 Elizabeth Avenue, St. John’s, Newfoundland and Labrador, Canada, 709-864-3097,
[email protected]
2. Research Fellow, Institute for Marine and Antarctic Studies, University of Tasmania, Hobart, Tasmania, Australia,
613-6227-7219, [email protected]
3. Senior Lecturer, Department of Physics, University of Malta, Msida, Malta, 356-2340-3613,
[email protected]
4. Research Scientist, Geological Survey of Norway, Trondheim, Norway, 7390-4267, [email protected]
5. Research Scientist, Department of Applied Research, Nova Scotia Community College, 80 Mawiomi Place,
Dartmouth, Nova Scotia, Canada, 902-491-2174, [email protected]
6. Associate Professor, Departments of Geography, Biology and Earth Sciences, Memorial University of
Newfoundland, 232 Elizabeth Avenue, St. John’s, Newfoundland and Labrador, Canada, 709-864-3233,
[email protected]
7. Associate Professor, Department of Geography, Memorial University of Newfoundland, 232 Elizabeth Avenue, St.
John’s, Newfoundland and Labrador, Canada, 709-864-8412, [email protected]
* Presenting Author: Vincent Lecours, Ph.D. Candidate, 709-864-3097, [email protected].
ABSTRACT
Developments in remote sensing techniques to produce digital terrain models (DTM) have provided new
methods for studying terrestrial environments, thereby improving scientific understanding of ecological
and geomorphological processes. Geomorphometry, or terrain analysis, uses DTMs to quantify physical
characteristics of the land (e.g., slope or rugosity) that are used for different applications such as habitat
mapping. The use of multibeam echosounder systems (MBES) to measure seafloor relief has changed the
way we study and understand marine environments; techniques from geomorphometry are now
commonly applied to these underwater DTMs for the investigation of marine habitats and
geomorphology.
Due to the inability of satellite remote sensing to collect data in deep waters and the limitations of MBES
data collection in shallower waters, there is often a gap in terrain data where land meets sea. A seamless
analysis of terrestrial and marine environments requires the combination of terrestrial DTMs, bathymetric
data from MBES, and bathymetric LiDAR to fill this gap. The challenges encountered with merging datasets
from different sources, such as data uncertainty or spatial resolution, makes such an approach still nascent
in the literature.
Using data from coastal Newfoundland and Labrador, this contribution reviews the potential uses of
continuous terrestrial/marine terrain models, especially for the field of geomorphometry and its
application in coastal environments. We argue that such DTMs can become essential to applications like
coastal habitat mapping, identification of hazards for navigation in shallow waters, or studying landforms
that overlap between environments. We also discuss how some techniques traditionally used in terrestrial
studies, such as Geographic object-based image analysis (Geobia), can be applied in marine studies.
60
Acoustic remote sensing of Labrador fjord environments: strengths, challenges
and limitations
Mallory Carpenter1*, Tanya Brown2, Alison Copeland3, Trevor Bell4, Evan Edinger5
1. Research Associate, Department of Geography, Memorial University of Newfoundland, St. John’s, Newfoundland
and Labrador, Canada, 709-864-7417, [email protected]
2. Post-doctoral Fellow, Department of Geography, Memorial University of Newfoundland, St. John’s,
Newfoundland and Labrador, Canada, 709-864-3038, [email protected]
3. Biodiversity Officer, Dept., of Conservation Services, Government of Bermuda, Hamilton, Bermuda, (441) 2992329 x2122, [email protected]
4. Professor, Department of Geography, Memorial University of Newfoundland, St. John’s, Newfoundland and
Labrador, Canada, 709-864-2525, [email protected]
5. Associate Professor, Departments of Geography, Biology and Earth Sciences, Memorial University of
Newfoundland, St. John’s, Newfoundland and Labrador, Canada, 709-864-3233, [email protected]
* Presenting Author: Mallory Carpenter, Research Associate, (709) 864-7417, [email protected]
ABSTRACT
Fjords represent a particularly challenging environment for remote sensing of marine habitats, due to
their widely varying depths, bottom composition, and strong gradients in water column properties.
Marine habitat mapping efforts in fjords of northern, central, and southern Labrador, Canada, have used
several multibeam sonar sensors, ground-truthed by drop-video camera and grab samples or box cores.
Sites surveyed include Nachvak and Saglek Bays (northern Labrador), Okak Bay and Lake Melville (central
Labrador), and Gilbert Bay (southern Labrador), and covered a wide range of depths and bottom types,
mostly derived from glacial geomorphic features.
Combining bathymetric data acquired at different times using different sensors was straightforward, but
integration and comparing backscatter data from various sensors using different frequencies proved much
more difficult. Statistically significant differences in species composition among habitat types mostly
reduced to hard vs. soft substrates. Acoustic detection of biogenic habitats such as kelp beds and coralline
algae (maerl) was particularly challenging; although both kelp and coralline algae were bathymetrically
limited, the backscatter strength of these habitat types was generally indistinct from that of the physical
substrates upon which the biogenic substrates occurred.
Water column properties, particularly temperature, salinity, dissolved oxygen content, and stratification
are variables likely to affect fjord flora and fauna. Habitat mapping in fjords could be improved by adding
data on the spatial variation in these and other water column properties. While satellite remote sensing
can provide sea surface temperature and salinity, oxygen availability and stratification are best assessed
using in-situ measurements such as CTD profiles.
61
Using Multibeam Bathymetry and Coastal Geomorphology to Predict Shallow
Marine Clam Habitat in the Eastern Canadian Arctic
Ben Misiuk1*, Trevor Bell2, Evan Edinger3, Alec Aitken4, Beth Cowan5, Janelle Kennedy6, John Hughes
Clarke7, Alex Flaherty8, Jonah Keyookta9, Sandrine Baillon10, Tim Siferd11
1. Master’s Student, Department of Geography, Memorial University of Newfoundland, [email protected]
2. Professor, Department of Geography, Memorial University of Newfoundland, [email protected]
3. Professor, Department of Geography, Memorial University of Newfoundland, [email protected]
4. Professor, Department of Geography and Planning, University of Saskatchewan, [email protected]
5. Master’s Student, Department of Geography, Memorial University of Newfoundland
6. Senior Advisor, Department of Environment, Fisheries and Sealing Division
7. Professor, Ocean Mapping Group, University of New Brunswick
8. Field Coordinator, Department of Environment, Fisheries and Sealing Division
9. Field Assistant, Community of Qikiqtarjuaq
10. PhD Student, Department of Biology, Memorial University of Newfoundland
11. Marine Invertebrates Program Lead, Department of Fisheries and Oceans Canada
* Presenting Author: Ben Misiuk, Master’s Student, 709 864-8998, [email protected]
ABSTRACT
A variety of natural clam resources exist in Nunavut, including Mya spp. near Qikiqtarjuaq, which have
been harvested by local gatherers for over a decade. Clam abundance and distribution appear to be
influenced by bottom composition and oceanographic conditions. Combined aerial optical and marine
acoustic remote sensing may provide a useful approach to mapping and predicting clam distributions. By
undertaking a supervised classification of multibeam bathymetry and backscatter, the benthic habitats
surrounding the community are being classified with respect to Mya spp. Multibeam data, ground-truthed
with underwater camera and benthic grab samples, are being used to differentiate areas with high
potential for clam occurrence. Coastal and paraglacial geomorphology derived from air photos is also
being assessed as a predictor of clam habitat. Multibeam collected in four regions of interest on the
eastern side of Broughton Island were ground-truthed using biological and sediment samples from 2013
and 2014 field seasons. Substrates in the Islands region were primarily cobble/boulder with some sand in
the northern portion, housing large populations of clams. Many of the Kingnelling Fjord sites were
dominated by cobbles, but the eastern portion was sandy with cobble outcrop, with moderate to dense
clam population. North Baffin was generally sandy with boulder outcrop and was relatively devoid of
benthic macro organisms. South Broughton was highly variable with cobble/boulder areas and sand with
moderate clam abundance. Sandy geomorphic features, such as raised deltas and glacial landforms, in
areas of limited wave exposure appear to be most conducive to high Mya clam abundances.
62
Automated sonar image matching
Peter King1*, Benjamin Anstey2, and Andrew Vardy3
1. MERLIN lab, Memorial University of Newfoundland, St. John’s, NL, CA
2. REALM Project, Memorial University of Newfoundland, St. John’s, NL, CA
3. Department of Computer Science – Faculty of Engineering and applied sciences, Memorial University of
Newfoundland, St. John’s, NL, CA
* Presenting Author: Peter King, Research Lab Coordinator, [email protected]
ABSTRACT
Memorial University’s Marine Environmental lab for Intelligent Vehicles (MERLIN) operates an
Autonomous Underwater Vehicle (AUV) which can perform subsea remote sensing using a variety of
sensors, including acoustic sensors such as side scan sonar. Over the past five years the MERLIN lab has
been developing tools to allow automated geo-referenced image generation, matching and of side scan
sonar data in real-time. These tools are developed for and deployed on Memorial’s Explorer AUV in
support of our work in autonomy and navigation. These tools employ image processing techniques along
with concepts from robotics.
Sonar data matching is achieved first through the transposition of raw side scan sonar data to a twodimensional image representation. Using the AUV’s on-board navigation sensors all images are
normalized in rotation and scale to improve robustness. The resulting images are then subject to the
application of feature based key point extraction and matching. Knowledge of the AUVs position and
motion are employed to filter our potential false matches through Bayesian filtering. When regions of
sonar coverage are shown to overlap the matching information is used to determine any misalignments
in the images.
Application of these tools have included autonomous path following using pre-collected sonar data and
AUV localization. Other potential applications would include data alignment and mosaicking.
63
S13 River Ice Monitoring / S13 Suivi de la glace de rivière
Sherry Warren and Alice Deschamps, co-chairs / modérateurs
Augmenting River Ice Flood Forecasting Services Using Satellite Radar Imagery –
A User Perspective
Amir A. Khan1* and Thomas M. Puestow2
1. Manager, Department of Environment & Conservation, Government of Newfoundland & Labrador, St. John’s, NL,
Canada, 709-729-2295, [email protected]
2. Senior Manager, Earth Observation, C-CORE, Captain Robert A. Bartlett Building, John’s, NL, Canada, 709-8642586, [email protected]
* Presenting Author: Amir A. Khan, Manager, Department of Environment & Conservation, 709-7292295, [email protected]
ABSTRACT
Many northern rivers developing ice covers during the winter season are prone to ice-related flooding. In
order to assess flood risk and mitigate its impact, it is imperative to monitor the development of ice covers
throughout the ice season, with particular emphasis on the freeze-up and break-up periods. This
monitoring is very hard to accomplish.
The Town of Badger, located at the confluence of three rivers, has a history of similar ice-related flooding
dating back to 1916. The Government of Newfoundland and Labrador through the Water Resources
Management Division (WRMD) provides an annual flood forecasting service for Badger using a computer
simulation model called the "Ice Progression Model" that simulates ice conditions on the Exploits River.
Following the severe flood of February 15, 2003, WRMD and C-CORE partnered to augment the flood
forecasting service with a satellite-based river ice monitoring service using RADAR imagery from the
ENVISAT and RADARSAT polar orbiting satellites. This was the first integration of RADAR imagery into
operational flood forecasting of river ice in Canada.
Since that introduction, C-CORE has, under its Polar View initiative, pioneered the provision of operational
satellite-based river ice monitoring services in Canada, the United States and Russia using the Badger River
Ice service template.
This presentation describes the River Ice flood forecasting service at Badger and the improvements that
were accomplished from WRMD’s perpsective. It describes the ice parameters derived from RADAR data
and the use of this information in model calibration and validation.
64
Operational Monitoring of River Ice Break-up Conditions using RADARSAT-2
Images
Alice Deschamps 1*, Joost van der Sanden2, George Choma3, Jean-Samuel Proulx-Bourque4, Sylvain
Lemay5
2. Canada Centre for Mapping and Earth Observation, Natural Resources Canada, 560 Rochester, Ottawa (ON),
Canada (613) 759-7949, [email protected]
3. Canada Centre for Mapping and Earth Observation, Natural Resources Canada, 560 Rochester, Ottawa (ON),
Canada, (613) 759-7197, [email protected]
4. Centre for Topographic Information, Natural Resources Canada, 2144-010 King Street West, Sherbrooke (QC),
Canada, (819) 564-5600 ext. 363, [email protected]
5. Canada Centre for Mapping and Earth Observations- Emergency Geomatics Service (CCMEO-EGS), Natural
Resources Canada, 560 Rochester, Ottawa (ON), Canada, (613) 759-7599, [email protected]
* Presenting Author: Alice Deschamps, Environmental Scientist, 613 759-7949,
[email protected]
ABSTRACT
River ice governs the winter regime of most Canadian rivers and influences the daily life of those that live
or work in riverside areas in various ways. The impact of ice covered rivers on Canadians typically peaks
at the time of breakup, in particular, when ice jams form and cause flood and/or damage to nearby
infrastructure.
Access to up-to-date information on river ice conditions facilitates decision making in the context of river
ice breakup and flood risk management. Traditional methods of river ice monitoring at the time of breakup
are expensive, provide limited spatial coverage and can be subject to weather conditions. Satellite remote
sensing systems make potentially outstanding tools for collecting current information on river ice
conditions due to their ability to repeatedly image an area of interest at a synoptic scale. Synthetic
Aperture Radar (SAR) satellites such as Canada’s RADARSAT-2 are particularly well suited for the task
because of their weather and daylight independent imaging capability, sensitivity to the presence of open
water, and sensitivity to the surface roughness of ice cover. From an emergency management perspective,
obtaining information on the development of the breaking ice cover roughness (on a scale of decimeters
to meters) can provide information on potential ice jam events that can lead to floods.
This paper will present a SAR-based approach for the classification of breaking river ice cover developed
at Canada Center for Remote Sensing. The approach was recently implemented by CCMEO’s Emergency
Geomatics Service (EGS) group to enable operational monitoring river ice breakup. The provision of
geomatics products in support of emergency management is among the core mandates of CCMEO-EGS.
The performance of the approach will be demonstrated by means of examples from a 2014 demonstration
project conducted in Northern Ontario, Manitoba and in Alberta.
65
Using Multi-Source Satellite Imagery for Operational River Ice Monitoring
Thomas M. Puestow1* and Amir A. Khan2
1. Senior Manager, Earth Observation, C-CORE, Captain Robert A. Bartlett Building, John’s, NL, Canada, 709-8642586, [email protected]
2. Manager, Department of Environment & Conservation, Government of Newfoundland & Labrador, St. John’s, NL,
Canada, 709-729-2295, [email protected]
* Presenting Author: Thomas Puestow, Senior Manager, 709-864-2586, [email protected]
ABSTRACT
Flooding due to ice jams is a major and frequent hazard for communities located near rivers in cold and
northern regions of the world. Ice jams might be formed during the development of ice covers in early
winter or during ice cover break up in early spring. A related issue is the management of river flows on
regulated rivers, especially the routing of early spring runoff during ice conditions while minimizing the
risk of initiating ice jams. Flood forecasting for such rivers is challenging due to the complexities of
modelling ice formation and breakup in real-time in an operational context. It is further complicated by
the effect of climate change, which has changed historically observed ice cover formation behavior.
This paper describes the creation and expansion of an operational river ice monitoring program based on
satellite RADAR imagery to address the challenges of flood forecasting, river flow management and
climate change adaptation. The program was developed and implemented for the Town of Badger on the
island of Newfoundland, Canada, following a severe river ice induced flood in 2003. River ice monitoring
at Badger initially relied on the use of satellite RADAR imagery from the ENVISAT, RADARSAT-1 and
RADARSAT-2. The lack of available SAR imagery due to conflict during critical periods of the monitoring
season made it necessary to integrate routinely acquired LANDSAT and MODIS data into the monitoring
process, although the utility of optical imagery is severely limited by the prevalence of cloud cover.
The availability of SENTINEL-1 imagery is expected to reduce the impact of conflict significantly and
increase the reliability of spatial and temporal coverage with SAR data. Accordingly, this investigation will
present early results of the use of SENTINEL-1 imagery for river ice monitoring, discuss issues of
operational integration and describe the initial performance of SENTINEL-1 in characterizing key river ice
parameters.
66
IceFRONT (1): Monitoring River Freeze-up Processes on the Peace River from
Optical Images
Jimmy Poulin1, Yves Gauthier2*, Antoine Padel3, Stéphane Hardy4, Monique Bernier5 and Martin Jasek6
1. Geomatics specialist, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada, G1K
9A9, Tel : (418) 654-3729, [email protected]
2. Remote sensing specialist, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada,
G1K 9A9, Tel : (418) 654-3753, [email protected]
3. Intern, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada, G1K 9A9,
[email protected]
4. Associate, Dromadaire Géo-Innovations inc., 8075 Hochelaga, bureau 201, Montréal (Québec), Canada, H1L 2K9,
Tel: (514) 493-4695 # 132 [email protected]
5. Professor, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada, G1K 9A9, Tel :
(418) 654-2585, [email protected]
6. Operations Planning Engineer, BC Hydro, 6911 Southpoint Drive, Burnaby, (BC), Canada, V3N 4X8, Tel : (604)
528-2580, [email protected]
* Presenting Author Yves, Gauthier, Remote Sensing Specialist, (418) 654-3753,
[email protected]
ABSTRACT
The IceFront project is a collaboration between Dromadaire Geo-Innovations, INRS, DLR and BC Hydro.
The initiative is funded in part by the Canadian Space Agency. The objectives of this project are 1) to help
hydropower companies to gather from satellite images, the timely information needed for operational
river freeze-up monitoring and 2) to speed up and simplify the process of analyzing the satellite images
and extracting the required information. The study is conducted on the Peace River (Alberta). The
approach integrates optical and radar images.
Here we present the approach using medium resolution optical data (MODIS and Landsat-8) to provide
quasi-daily information on the timing and location of ice lodgement and on the progression of the ice
front. An automated procedure is proposed to produce “ice/no ice” maps from 84 MODIS images of the
Peace River in 2013/2014. We first create a mask of the pixels (250m) which are totally included in the
river channel in order to exclude mixels or narrow sections. We then calculate thresholds on Band 1 and
Band 2 pertaining to open water and full ice cover conditions on these pixels. We finally apply a probability
function between these thresholds to estimate the probability of ice. A similar procedure is applied for 19
Landsat-8 images. Accuracy is over 90% in general. It is lower in incised sections (shadow effect) and it is
also affected by the quality of the cloud mask.
67
IceFRONT (2): Monitoring River Freeze-up Processes on the Peace River from
Radar Images
Yves Gauthier 1*, Stéphane Hardy 2, Achim Roth3, Helena Łoś4, Monique Bernier5, Jimmy Poulin6, and
Martin Jasek7
1. Remote sensing specialist, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada,
G1K 9A9, Tel : (418) 654-3753, [email protected]
2. Associate, Dromadaire Géo-Innovations inc., 8075 Hochelaga, bureau 201, Montréal (Québec), Canada, H1L 2K9,
Tel: (514) 493-4695 # 132 [email protected]
3. Science Coordinator, German Aerospace Center (DLR), Earth Observation Center, Oberpfaffenhofen, 82234
Wessling, Germany, Tel: +49 8153 28-2706, [email protected]
4. Intern, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada, G1K 9A9, (PhD
candidate, WUT Institute of Photogrammetry, Remote Sensing and GIS, pl. Politechniki 1, 00-661 Wasaw, Poland)
[email protected]
5. Professor, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada, G1K 9A9, Tel :
(418) 654-2585, [email protected]
6. Geomatics specialist, INRS Centre Eau Terre Environnement, 490 rue de la Couronne, Québec (Qc), Canada, G1K
9A9, Tel : (418) 654-3729, [email protected]
7. Operations Planning Engineer, BC Hydro, 6911 Southpoint Drive, Burnaby, (BC), Canada, V3N 4X8, Tel : (604)
528-2580, [email protected]
* Presenting Author Yves Gauthier, Remote Sensing Specialist, (418) 654-3753,
[email protected]
ABSTRACT
The IceFront project is a collaboration between Dromadaire Geo-Innovations, INRS, DLR and BC Hydro.
The initiative is funded in part by the Canadian Space Agency. The objectives of this project are 1) to help
hydropower companies to gather from satellite images, the timely information needed for operational
river freeze-up monitoring and 2) to speed up and simplify the process of analyzing the satellite images
and extracting the required information. The study is conducted on the Peace River (Alberta). The
approach integrates optical and radar images.
Here we present the approach using multi-platform radar data to provide specific information on the
dominant river ice types. Radarsat-2 quadpol data and TerraSAR-X dualpol are used to produce river ice
maps at 10m and 3m resolution during the freeze-up of 2013-2014, using the IceMAP-R procedure from
INRS. The analysis evaluates the complementarity of C-Band and X-Band data in order to increase image
availability for timely high resolution monitoring of river ice. We also study the potential gain of
polarimetry in the accuracy of ice characterization when using derived Kennaugh elements for both
sensors. One elusive challenge that is being specifically addressed is the improved discrimination between
open water and bubble free thermal ice. These conditions occur in calm back channels as well as in skim
ice covers. Different polarization combinations (HH/HV and HH/VV) are also compared. Validation is done
with the expertise and knowledge of BC Hydro engineers and by cross-comparison with products derived
from optical data.
68
S14 Arctic-Boreal Research and the pre-ABoVE Projects / S14 Projets
de recherche sur la vulnérabilité arctique–boréale et les projets preABoVE
Elizabeth Hoy, chair / modérateur
The NASA Arctic-Boreal Vulnerability Experiment and the ABoVE Science Cloud
Elizabeth Hoy1*, Peter Griffith1, Mark McInerney2, Dan Duffy3, Scott Sinno2, J. Hoot Thompson2, Steve
Ambrose2
1. Support Scientist, NASA Carbon Cycle and Ecosystems Office (CCEO) Global Science and Technology, Inc., NASA
Carbon Cycle and Ecosystems Office, NASA Goddard Space Flight Center, 301-614-6494, [email protected]
2. NASA Climate Model Data Services (CDS)
3. NASA Center for Climate Simulation (NCCS)
* Presenting Author: Elizabeth Hoy, 301-614-6494, [email protected]
ABSTRACT
The Arctic-Boreal Vulnerability Experiment (ABoVE), a field campaign sponsored and initiated by NASA’s
Terrestrial Ecology Program, is a large-scale study of changes to terrestrial and freshwater ecosystems in
the Arctic and boreal regions of western North America and the implications of these changes for local,
regional, and global social-ecological systems. Research for ABoVE will address questions in six thematic
areas: society, disturbance, permafrost, hydrology, flora/fauna, and carbon biogeochemistry through
integrating field-based studies, modeling, and data from airborne and satellite remote sensing. In an
effort to accelerate the pace of new Arctic science for researchers participating in the field campaign, the
NASA Center for Climate Simulation has partnered with the NASA Carbon Cycle and Ecosystems Office to
create a high performance science cloud. The ABoVE Science Cloud combines high performance
computing with emerging technologies and data management with tools for analyzing and processing
geographic information to create an environment specifically designed for large-scale modeling, analysis
of remote sensing data, copious disk storage for “big data” with integrated data management, and
integration of core variables from in-situ networks. Furthermore, by using the ABoVE Science Cloud as a
shared and centralized resource, researchers reduce costs for their proposed work, making proposed
research more competitive. Here we will provide an update on the current status of ABoVE and discuss
the ABoVE Science Cloud, which is currently in use by a number of the ABoVE-related projects.
69
Quantifying change in North American Arctic lakes between 1990 and present
Mark Carroll1*, Margaret Wooten1, Charlene DiMiceli2, Robert Sohlberg2 and John R.G. Townshend2
1. NASA Goddard Space Flight Center, Greenbelt, MD, United States
2. University of Maryland College Park, College Park, MD, United States
* Presenting Author: Mark Carroll, [email protected]
ABSTRACT
The NASA Arctic and Boreal Vulnerability Experiment (ABoVE) is a new multi-year and mulit-disciplinary
field campaign that is set to begin in 2015. One of the primary themes of this campaign is to understand
the complicated hydrology of the region. Small lakes and ponds are a prominent feature of the landscape
in the High Northern Latitudes. Previous mapping efforts have been accomplished with either single date
medium to fine resolution imagery or multiple observations with coarse resolution imagery. The maps
from single date imagery are prone to errors relate to weather phenomena such as flood and drought.
The coarse resolution products are limited to larger water bodies and miss the many smaller lakes and
ponds in the region.
We have created a time series of maps from Landsat Thematic Mapper and Enhanced Thematic Mapper
in three epochs (1990 - 1992), (2000 - 2002), and (2010 - 2012) showing the location and extent of water
bodies in the region. These maps represent the first comprehensive time series of water bodies in the
region that have been generated with a consistent input data set at 30m spatial resolution. We can use
these maps to quantify the amount of change that has occurred in the Arctic lakes over the past 20+ years.
Here we will present the first versions of the maps with the associated structure and formats and some
initial results from analysis of change in northern Canada.
70
Long-Term Multi-Sensor Record of Fire Disturbances in High Northern Latitudes
Tatiana Loboda1*, Nancy French2, Laura Bourgeau-Chavez3, Mary Ellen Miller4, Liza Jenkins5, Ivan Csiszar6
1. Associate Professor, Department of Geographical Sciences, University of Maryland, 2181 LeFrak Hall, College
Park, Maryland, USA, 1-301-405-8891, [email protected]
2. Senior Research Scientist, Michigan Technological Research Institute, Michigan Technological University, 3600
Green Ct Suite 100, Ann Arbor, Michigan, USA, 1-734-913-6844, [email protected]
3. Research Scientist, Michigan Technological Research Institute, Michigan Technological University, 3600 Green Ct
Suite 100, Ann Arbor, Michigan, USA, 1-734-913-6873, [email protected]
4. Research Engineer, Michigan Technological Research Institute, Michigan Technological University, 3600 Green Ct
Suite 100, Ann Arbor, Michigan, USA, 1-734-994-7221, [email protected]
5. Research Scientist, Michigan Technological Research Institute, Michigan Technological University, 3600 Green Ct
Suite 100, Ann Arbor, Michigan, USA, 1-734-913-6869, [email protected]
6. Chief, Environmental Monitoring Branch, Satellite Meteorology and Climatology Division, NOAA/NESDIS Center
for Satellite Applications and Research, NOAA Center for Weather and Climate Prediction (NCWCP), 5830 University
Research Court, College Park, Maryland, USA, 1-301-683-3583, [email protected]
* Presenting Author: Tatiana Loboda,301-405-8891, [email protected].
ABSTRACT
The greatest rise in global temperature is occurring in High Northern Latitudes (HNL) above 60°N and this
recent warming trend is projected to continue. Wildfire is the main disturbance agent in this ecosystem
and a major emerging issue identified by the North Slope Science Initiative. Although much research has
been done in the forested areas of HNL in the recent decade, little is still known about fire patterns in
transitional forest/tundra ecotone and tundra ecosystems. We are developing a consistent data set in
support of the NASA Arctic-Boreal Ecosystems Vulnerability Experiment (ABOVE) field campaign. We
implement several regional burned area products from coarse and moderate resolution instruments over
the full span of circumpolar extent of HNL where data are available. The data set also includes active fire
detections and burned area products from a succession of coarse resolution sensors, including Advanced
Very High Resolution Radiometer (AVHRR), Moderate Resolution Imaging Spectroradiometer (MODIS),
and Visible Infrared Imaging Radiometer Suite (VIIRS), to guide the selection and archival of clear surface
Landsat images in order to produce moderate resolution fire perimeter and fire impact characterization
maps across HNL. We will provide the first formal assessment of VIIRS and the Landsat Data Continuity
Mission (LDCM) capabilities for fire detection and mapping burned areas in HNL thus establishing the
continuation of fire mapping in HNL using a consistent algorithm into the future. Finally, we explore the
opportunities offered by Synthetic Aperture Radar (SAR) sensors to characterize burns in HNL.
71
Remotely Sensed Active Layer Thickness (ReSALT)
Kevin Schaefer1*, Lin Liu2, Andrew Parsekian3, Elchin Jafarov1, Albert Chen4, Tingjun Zhang1,5, Alessio
Gusmeroli6, Santosh Panda6, Howard A. Zebker4, Tim Schaefer7
1. National Snow and Ice Data Center, Cooperative Institute for Research in Environmental Sciences, University of
Colorado at Boulder, Boulder, CO 80309; E-Mails: [email protected]; [email protected]; [email protected]
2. Earth System Science Programme, Faculty of Science, The Chinese University of Hong Kong, Hong Kong, China; EMail: [email protected]
3. Department of Geology and Geophysics, University of Wyoming, Laramie, WY 82070; E-Mail:
[email protected]
4. Department of Geophysics, Stanford University, Stanford, CA, USA; E-Mails: [email protected];
[email protected]
5. College of Earth and Environmental Sciences, Lanzhou University, Lanzhou, Gansu Province 730000, China; EMail: [email protected]
6. International Arctic Research Center, University of Alaska Fairbanks; E-Mail: [email protected];
[email protected]
7. Galmont Consulting, Chicago, USA; E-Mail: [email protected]
* Presenting Author: Kevin Schaefer, Research Scientist, 303-492-8869, [email protected]
ABSTRACT
Active layer thickness (ALT) is a critical parameter for monitoring the status of permafrost, typically
measured at specific locations using probing, in situ temperature sensors, or other ground-based
observations. The Circumpolar Active Layer Monitoring (CALM) network has conducted ALT
measurements over the past two decades, but is under-populated, a gap which can be filled using remote
sensing. Here we describe the Remotely Sensed Active Layer Thickness (ReSALT) product developed in
support of field operations and other activities for the Arctic-Boreal Vulnerability Experiment (ABoVE) field
campaign in Alaska and northwest Canada. We use the Interferometric Synthetic Aperture Radar (InSAR)
technique to 1) measure long-term subsidence trends resulting from the melting and subsequent drainage
of excess ground ice in permafrost, and 2) measure seasonal subsidence resulting from the expansion of
soil water into ice as the active layer freezes and thaws. We estimate ReSALT from the seasonal
subsidence assuming a vertical profile of water within the soil column and identify individual thermokarst
features as spatial anomalies in the subsidence trends. We present 100x100 km2, 100 m resolution
ReSALT products for several sites on the North Slope of Alaska: Prudhoe Bay, Barrow, Toolik Lake, Happy
Valley, and the Anaktuvik fire zone. We validated ReSALT using in situ measurements of ALT from CALM
and Ground Penetrating Radar. We will develop additional ReSALT products for the Mackenzie River Delta
and other sites with concentrated field measurements in the ABoVE domain. ReSALT has proven effective
in measuring local variations in ALT and thermokarst activity on the North Slope of Alaska with broader
potential applications across the permafrost regions of North America and Eurasia.
72
Producing a High-Resolution Circumpolar Delineation of the Forest-Tundra
Ecotone
Christopher Neigh1*, Paul Montesano2, Joe Sexton3,
Min Feng4, K. Jon Ranson5, Saurabh Channan6, and Mark Chopping7
1. Research Scientist, NASA Goddard Spaceflight Center Biospheric Science Laboratory Code 618, Greenbelt, MD
20771,USA, 301-614-6681, [email protected]
2. Research Scientist, Science Systems Applications Inc., NASA Goddard Spaceflight Center Biospheric Science
Laboratory Code 618, Greenbelt, MD 20771,USA, 301-614-6695, [email protected]
3. Research Assistant Professor, University of Maryland College Park Department of Geographical
Sciences, College Park, MD 20742, USA, 301-405-8165, [email protected]
4. Research Assistant Professor, University of Maryland College Park Department of Geographical
Sciences, College Park, MD 20742, USA, 301-314-9299, [email protected]
5. Research Scientist, NASA Goddard Spaceflight Center Biospheric Science Laboratory Code 618, Greenbelt, MD
20771,USA, 301-614-6654, [email protected]
6. Director of the Global Land Cover Facility, University of Maryland College Park Department of
Geographical Sciences, College Park, MD 20742, USA, 301-314-9299, [email protected]
7. Professor, Mont Clair State University Earth and Environmental Studies, Montclair, NJ 07043, USA, 973-655-7384,
[email protected],
* Presenting Author Christopher Neigh, Research Scientist, 301-614-6681,
[email protected]
ABSTRACT
In the High Northern Latitudes climate change has altered vegetation productivity, carbon sequestration,
and many other processes. These changes have been observed within the taiga-tundra ecotone (TTE),
which is the transition zone between the two biomes. However, establishing the spatial extent of the TTE
is challenging because of variations in vegetation structure owing to site-level interactions between
microclimate, topography, winter snow depth, wind, and edaphic conditions. Such interactions have
produced TTE forest patterns that include sporadic forest cover patches to growth-stunted trees
resembling shrubs. These patterns are evident at local scales that are often not resolvable in imagery
from most earth observing satellites. Such scale problems have contributed to large differences in remote
observation of the TTE. Recent advances using the global archive of Landsat data have provided vegetation
continuous fields (VCF) at 30-m resolution, which may improve TTE delineation. Furthermore, U.S. federal
access to sub-meter DigitalGlobe data provides the means to improve estimates of the extent of the TTE
with observations of individual trees. We are using these data to optimize the Landsat VCF delineation of
the TTE ecotone between 50° and 75° north latitude. The resulting product will help to identify sites of
potential TTE forest structure change, and be used as input to ecosystem models. An overview and early
results of our current work will be presented.
73
S15 Icebergs and Extreme Ice Features / S15 Icebergs et formations
glacielles extrêmes
Pradeep Bobby, chair / modérateur
Intercomparison of sea ice freeboard, thickness, and ice concentration extracted
from sea ice model with satellite data
Siva Prasad1*, Igor Zakharov2, Pradeep Bobby3 and Peter McGuire4
1. PhD Student, Memorial University of Newfoundland, Faculty of Engineering and Applied Sciences, St. John’s, NL,
Canada, 709-763-8304, [email protected]
2. C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL Canada, A1B 3X5, 709-864-2582,
[email protected]
3. Director of Earth Observation, C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL Canada,
A1B 3X5, 709-864-8361, [email protected]
4. C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL Canada, A1B 3X5, 709-864-2006,
[email protected]
* Presenting Author: Siva Prasad, Student, 709-763-8304, [email protected]
ABSTRACT
Sea ice parameters, such as thickness, freeboard and ice concentration are important for climate studies
and marine operations. The Community Ice CodE (CICE) model was implemented in standalone mode with
a high-resolution scale in the regional domain of Baffin Bay and the Labrador Sea. The ice parameters
retrieved from the model were assimilated with OSI-SAF (Ocean and Sea Ice Satellite Application Facility)
ice concentration using a nudging method.
Model ice thickness was compared with thickness extracted from SMOS-MIRAS (Soil Moisture Ocean
Salinity – Microwave Imaging Radiometer using Aperture Synthesis). The modelled thickness was found
to be within the uncertainty limits of the observed parameters. The standard deviation of the difference
between observed freeboard and the model had its highest value when ice was forming during October.
Comparisons of the model results with freeboard data observed by satellite altimeter CryoSat-2 indicated
a maximum standard deviation of 7.5 cm, corresponding to ice formation period. Model outputs from
monthly assimilated ice concentration were compared with daily OSI-SAF products and the best results
were found for forecasts of less than four days.
The preliminary results also indicate that the accuracy of modeled ice thickness and ice concentration is
affected by including mixed layer depth parameterization and sea surface salinity from World Ocean Atlas
climatology data.
74
Iceberg Detection over Northern Latitudes Using High Resolution TerraSAR-X
Images
Anja Frost1*, Rudolf Ressel2, and Susanne Lehner3
1. Researcher, German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt, DLR), Maritime Security Lab,
Henrich-Focke-Straße 4, 28199 Bremen, Germany, +49421244201859, [email protected]
2. Researcher, German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt, DLR), Maritime Security Lab,
Henrich-Focke-Straße 4, 28199 Bremen, Germany, +49421244201858, [email protected]
3. Head of Maritime Security Lab, German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt, DLR),
Henrich-Focke-Straße 4, 28199 Bremen, Germany, +49421244201858, [email protected]
* Presenting Author: Anja Frost, PhD, 49421244201859, [email protected].
ABSTRACT
Over northern latitudes, icebergs frequently cross shipping routes and impair marine traffic. To improve
ship routing, we propose a novel algorithm that is intended to detect and chart icebergs from Synthetic
Aperture Radar (SAR) images provided by the German satellite TerraSAR-X.
TerraSAR-X is in a near-polar orbit and fully operational since January 2008. Equipped with an active Xband antenna, it is able to monitor the ocean and frozen waters regardless of cloud cover and darkness,
providing a resolution of down to 3 m. Hence, even small icebergs become visible.
The proposed algorithm is based on the iterative censoring constant false alarm rate (IC-CFAR) detector,
which has proven its usefulness for target detection already. Different from the standard approach, we
not only estimate statistical properties of open water backscatter expressed by a probability density
function, but also survey in more detail recurring patterns (i.e. waves) building on former studies on
wind and sea state retrieval. This allows discriminating icebergs from most false alarms that arise from
rough sea and strong winds.
Experiments carried out with a series of HH polarized TerraSAR-X StripMap images acquired between
2012 and 2014 confirm that - due to consideration of wave pattern during image processing - the false
alarm rate is reduced by factor three. On average, we recorded 0.002 false alarms per km² and a
detection rate of 93 %. Since results are output in near real time (NRT), we provide the basis for
supporting operational ship routing in remote ice infested areas.
75
Detection and Discrimination of Ships and Icebergs using Satellite Altimetry: An
Operational Perspective for Estimating Iceberg Populations
Igor Zakharov1, Thomas M. Puestow2* and Andrew H. Fleming2
1. Research Scientist, C-CORE, Captain Robert A. Bartlett Building, John’s, NL, Canada, 709-864-2582,
[email protected]
2. Senior Manager, EO, C-CORE, Captain Robert A. Bartlett Building, John’s, NL, Canada, 709-864-2586,
[email protected]
3. Remote Sensing Manager, British Antarctic Survey, Madingley Road, Cambridge CB3 0ET, UK, +44 (0)1223
221451, [email protected]
* Presenting Author: Thomas Puestow, C-CORE, Captain Robert A. Bartlett Building, John’s, NL, Canada,
709-864-2586, [email protected]
ABSTRACT
This study investigated the utility of multi-source, multi-date satellite altmtery within the context of
iceberg monitoring and population assessment over large and generally data-poor areas. Building on the
recently documented capabilities of satellite altimneters to detect icebergs, this investigation aimed at
improving detection and incorporating the discrmination of ships and icebergs. Satellite altimeters used
include JASON-1, JASON-2 and Topex/POSEIDON, all of which operate in both Ku and C-Band.
Ship observations for training and validation data were collected in areas of known major shipping routes,
well outside any ice-affected areas. Conversely, training and validation data for icebergs were collected in
areas of known iceberg occurrence and low probability of vessel traffic.
The detection of iceberg and ship targets relied on using the phase of the Fourier-transformed altimetry
signals. Each detected target was described by different parameters describing the strength and shape of
altimetry Ku and C-Band signatures, including maximum backscatter coefficient, total backscattered
power, the number of samples per target, the number of parabolas per target and the number of specular
reflectors in parabolas.
These parameters were statistically examined to assess their ability to discriminate between ships and
icebergs and select appropriate predictor variables for use in supervised classification. In order to identify
the appropriate classification methodology, a number of parameteric and non-parameteric machine
learning algorithms were examined, including discriminant functions, k-nearest neighbor, neural
networks, support vector machines, and decision tree analysis. Several classification scenarios were
carried out using single algorithms as well as ensembles of several algorithms.
The resulting probabilities of detection and classiofiocation accuracies were assessed using the
independent validation sample as well as selected SAR imagery. The operational application for
monitoring and historial analyses was assessed.
76
Characterization of Hazardous Ice using Spaceborne SAR and Ice Profiling Sonar
Kaan Ersahin*1, Leslie Brown2, Randy Kerr3 ,Michael Henley4, Edward Ross5, Keath Borg6, and David
Fissel7
1,2,3,4,5,6,7. ASL Environmental Sciences Inc., Victoria, BC, Canada.
* Presenting Author: Kaan Ersahin, E-mail: [email protected]
ABSTRACT
Ice can pose hazard for operations (e.g., transportation, shipping, surveillance, offshore oil and gas
exploration) and for infrastructure (e.g., ports, pipelines, offshore structures). There is an increasing need
for fine scale characterization of hazardous ice conditions. This information is of interest to many
stakeholders including the industry and government departments and agencies.
Spaceborne SAR sensors are being used for near-real-time monitoring of the regional ice conditions.
Satellite derived sea ice information products typically rely on the interpretation of ice analysts,
sometimes supported by semi-automated tools. However, validation of data products is a remaining
challenge due to limited or no ground-truth.
The ASL Ice Profiling Sonar (IPS) is an upward looking sonar (ULS) device designed for high resolution ice
draft measurements. The Acoustic Doppler Current Profilers (ADCP) measure ice velocity, when used in
conjunction with IPS, horizontal extent of the ice features can be estimated. Since the mid-1990s, these
instruments provided ice draft and velocity measurements in support of research and industrial programs
in the Arctic Ocean and marginal ice zones. Analysis of ULS datasets allow us to make inferences regarding
the ice conditions that are used as input to the engineering design of structures and the development of
operational support programs. From the ULS data we characterize deformed sea ice features, e.g.,
individual keels and segments of hummocky (rubbled) ice.
This work, funded by the Canadian Space Agency (2013-2015), aimed to calibrate and validate satellitebased ice data products using continuous measurements obtained from ULS instruments. We focused on
the development of techniques to characterize hazardous ice using quad-polarized Radarsat-2 and
corresponding ULS datasets. The results include:

Paired satellite and ULS datasets to allow algorithm calibration and validation

ULS data overlays on imagery used for interpretation and analysis

Assessment of algorithms for ice draft estimation and hazardous ice characterization.

Evaluation of simulated RCM compact-pol products for characterization
77
Research and Development in Iceberg Monitoring at the Canadian Ice Service
Matthew Arkett1, Leah Braithwaite1, Angela Cheng1*, Tom Carrieres1, Adrienne Tivy1, Stephanie
Tremblay-Therrien1 , Melanie Lacelle1, Ronald Saper2, and Derek Mueller3
1. Environment Canada-Canadian Ice Service, 373 Sussex Drive, Ottawa, ON, Canada
2. Physical Scientist, 373 Sussex Drive, Ottawa, ON, Canada, [email protected]
3. Assistant Professor, Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada, (613) 520-2600 (ext. 1984),
[email protected]
* Presenting Author: Angela Cheng, 613 947-7514, [email protected]
ABSTRACT
Every year, roughly 40,000 icebergs migrate through Canada's eastern waters, while more recently, ice
islands fragments originating from the ice shelves of northern Ellesmere Island, are drifting as far south
as Newfoundland. These extreme ice features pose a significant collision risk to shipping vessels and
offshore oil and gas platforms. The Canadian Ice Service (CIS), an agency within the Meteorological Service
of Canada (MSC), and the US Coastguard’s International Ice Patrol (IIP) are the government bodies
responsible for producing and releasing iceberg information products for Canadian and US waters.
Through their North American Ice Service (NAIS) partnership agreement, both agencies share
responsibility for the development of operational iceberg products, helping to ensure the safety of life at
sea. In addition to these operational products, the CIS also produces a monthly update on the location of
ice islands and their fragments on Canada’s east coast.
Historically, both agencies have primarily used aerial reconnaissance and ship-based iceberg and ice island
sightings as the principal inputs in product development. In more recent years, ice hazard locations
derived from space based synthetic aperture radar (SAR) are incorporated into NAIS information products.
SAR constellations like Sentinel-1 and the RADARSAT Constellation Mission (RCM) will increase the
temporal frequency of SAR observations for ice hazard monitoring and should improve the level of ice
hazard information available to mariners.
This paper will discuss ongoing research at CIS in iceberg detection and discrimination from SAR. The use
of these detections in iceberg and ice island drift modelling, ice island monitoring and characterization,
and iceberg climatology are also discussed.
78
S16 Global Vegetation Science / S16 Télédétection de la végétation à
l'échelle globale
Compton (Jim) Tucker, chair / modérateur
Satellite observed sunlight-mediated seasonality in greenness of wet equatorial
Amazonian rainforest
Taejin Park1,*, Yuri Knyazikhin2, Jian Bi3, Sungho Choi4 and Ranga B Myneni5
1. PhD candidate, Boston University, Boston, MA, United States, 617-893-1988, [email protected]
2. Research Professor, Boston University, Boston, MA, United States, 617-353-8843, [email protected]
3. PhD candidate, Boston University, Boston, MA, United States, 857-472-0355, [email protected]
4. PhD candidate, Boston University, Boston, MA, United States, 617-353-8846, [email protected]
5. Professor, Boston University, Boston, MA, United States, 617-353-2525, [email protected]
* Presenting Author: Taejin Park, PhD candidate, 617-893-1988, [email protected].
ABSTRACT
Amazonian rainforests represent nearly half of the tropical terrestrial biomass and it is a major component
of regional and global carbon and hydrological cycles. Understanding their intra- or inter-annual variation
of structure and metabolic behavior is critical to projecting their response to climate change. Recent many
studies based on space-borne sensors have been conducted to understand the nature and controls of the
seasonal variation in greenness of wet equatorial Amazon forest. However, highly complicated sensing
conditions such as saturated reflectance and weak sensitivity to changes in canopy properties from this
dense forest often hinder to rightly understand the nature and controls of the seasonal variation in
greenness. In this study, we suggest a robust approach to interpret such satellite data by analyzing angular
signatures of radiation reflected from vegetation. This approach was implemented to investigate
seasonality in greenness of wet equatorial Amazonian rainforests and we utilized three independent
satellite data from Terra MODIS, MISR and Aqua MODIS sensors for this purpose. Analyzed three
independent satellite datasets consistently show higher greenness level during the dry season (June to
October) relative to the wet season (November to May). This can be interpreted as dry season increases
in leaf flushing, litterfall, photosynthesis and evapotranspiration in this forest. Additionally, this result
provides evidence for sunlight-mediated seasonality in greenness, i.e., the light is the limiting factor for
productivity of well-hydrated equatorial Amazonian rainforests.
79
Global ENSO Teleconnection Patterns derived from NDVI3g: 1981-2013
Assaf Anyamba1*, Jennifer Small2 and Compton Tucker3
1. Research Scientist, GIMMS Group, Biospheric Sciences Laboratory, Code 618.0, NASA Goddard Space Flight
Center, Greenbelt, MD 20771, USA; +1301-614-6601, [email protected].
2. Programmer Analyst, GIMMS Group, Biospheric Sciences Laboratory, Code 618.0, NASA Goddard Space Flight
Center, Greenbelt, MD 20771, USA; +1301-614-6602, [email protected].
3. Senior Scientist, GIMMS Group, Biospheric Sciences Laboratory, Code 618.0, NASA Goddard Space Flight Center,
Greenbelt, MD 20771, USA; +1301-614-6644, [email protected].
* Presenting Author: Assaf Anyamba, Research Scientist, 301-614-6601, [email protected].
ABSTRACT
Extremes phases of the El Niño Southern Oscillation (ENSO) phenomena are known to dominate
interannual climate variability of tropical and extra-tropical rainfall and temperature impacting global
vegetation dynamics. In this paper, we update the response of global vegetation patterns to ENSO events
using the new normalized difference vegetation index (NDVI; version 3g) 1981 to 2013 data set
In order the examine relationship between season NDVI and variations in the phase and amplitude of
ENSO (represented by sea surface temperature variations in the NINO 3.4 region of the equatorial Pacific
ocean: 5N-5S, 170W-120W), we compute monthly standardized NDVI anomalies and correlate them
against corresponding NINO 3.4 SST anomalies to produce a global teleconnections map.
Our findings show that, the El Niño phase of ENSO causes distinct and simultaneous patterns of anomalous
green-up and drought at various regional locations around the world. Specifically, there is a tendency for
greener-than-normal conditions associated with above-normal rainfall to occur over Eastern Africa, the
southern tier of the United States, northern US and across Canada and coastal Ecuador, and Peru.
Similarly, there is a tendency for drought to occur over India, Southeast Asia, Australia, northern and
north-eastern Brazil, and Southern Africa. These conditions are largely reversible during the La Niña phase
of ENSO. Our results confirm previous findings using precipitations data and provide a method for
mapping impacts of climate variability. Continued processing and availability of such long time series data
sets is critical to monitoring and mapping the impacts of a variable and changing climate.
80
Earth Observations and Global Environmental Governance
Lennart Olsson1* and Yangoh Tambang2
1. Director, Lund University, Centre for Sustainability Studies, Box 170, S-22100 Lund, Sweden +46 (0)46 222 0511,
<[email protected]>
2. Postdoctoral Fellow, Lund University, Centre for Sustainability Studies, Box 170, S-22100 Lund, Sweden +46 (0)46
222 0511, <[email protected]>
* Presenting Author: Lennart Olsson, professor, 46 (0)46 222 0511, <[email protected]>
ABSTRACT
Timely information on global vegetation is essential for global environmental governance. Environmental
policies are increasingly made and evaluated at a supra-national level which puts increasing demands on
the remote sensing community to develop technologies that are reliable, affordable, and accessible for
the international community. Questions about legitimacy and accountability of decision making have
recently come to the forefront in forest governance. Questions about saliency and reliability of remote
sensing based monitoring have also come under scrutiny in relation to monitoring of land degradation.
Proposed Sustainable Development Goals will give rise to new demands on monitoring global
environmental processes and phenomena. We will discuss the role of satellite remote sensing in relation
to global environmental politics. Issues of data quality, saliency, and access will be addressed as well as
legitimacy and accountability of governance based on Earth observations.
81
A Non-Stationary 1981 to 2014 AVHRR NDVI Time Series
Jorge Pinzon1 and Compton Tucker2*
1. Applied Mathematician, Earth Science Division, NASA/Goddard Space Flight Center, Greenbelt, Maryland 20771
USA, 1 301 614 6687, <[email protected]>
2. Physical Scientist, Earth Science Division, NASA/Goddard Space Flight Center, Greenbelt, Maryland 20771 USA 1
301 614-6644 <[email protected]>
* Presenting Author: Compton Tucker, Physical Scientist, 301 614 6687, <[email protected]>.
ABSTRACT
We describe an updated version of a normalized difference vegetation index data set produced from
Advanced Very High Resolution Radiometer (AVHRR) instruments that extends this record from1981 to
2014 using data only from these instruments. Minor deficiencies in a previous 1981-2011 version of our
AVHRR NDVI data set have been corrected and we describe quantification of the seasonal, inter-annual,
spatial coherence, temporal coherence, and error components of the data set’s variance.
82
Global warming and trends in seasonality of NDVI: Europe 1982-1998 versus
1999-2014
J. Ronald Eastman 1* and Elia A. MacHado2
1. Professor, Clark University, Clark University, Worcester, MA 01610-1477 USA, 1 508 849-2321,
<[email protected]>
2. Assistant Professor, Department of Earth, Environmental, and Geospatial Sciences, Lehman College, Bronx, NY
10468 USA, 1 718 960 1103, <[email protected]>
* Presenting Author: J. Ronald Eastman, professor, 508 849-2321, <[email protected]>
ABSTRACT
The last 15 years have seen a slowing in the persistent growth of tropospheric temperatures experienced
from the 1970's until the end of the century. The reasons for this slowing are still under debate, although
2014 was the warmest year on record globally. However, this change of regime represents an excellent
opportunity to evaluate the role of global warming on the seasonality of NDVI. In this exploratory study,
we focus on trends in the seasonality of NDVI from 1982-1998 and from 1999-2014 in Europe to gauge
the evidence in the NDVI record for a change regime coincident with recent global warming.
83
S17 3D Characterization of Canopy Structure using LiDAR and New
Technologies – I / S17 Caractérisation du couvert végétal avec le
LiDAR et les nouvelles technologies - I
Andrew Hudak, chair / modérateur
GLAS Estimates of Canopy Gap Fraction
Craig Mahoney1*, and Chris Hopkinson2
1. Post-Doctoral Fellow, Department of Geography, University of Lethbridge, Lethbridge, Alberta T1K 6T5, +1 403
332-4043, [email protected]
2. Professor, Department of Geography, University of Lethbridge, Lethbridge, Alberta T1K 6T5, +1 403 332-4586,
[email protected]
* Presenting Author: Craig Mahoney, Post-Doctoral Fellow, 403 332-4043, [email protected]
ABSTRACT
Canopy gap fraction is a widely used forest attribute in energy and water balance, ecological and forest
resource modeling applications. Modeling gap fraction from both discrete and waveform Light Detection
and Ranging (LiDAR) systems is an active area of research but is not always straight forwards due to
differences in vegetation and ground surface reflectance properties, as well as other localised or system
related phenomena. As a modelling solution, waveform ground returns are often calibrated to quasiarbitrary scaling factors, which can be heavily localized.
This study utilizes a normalised sampling of Geoscience Laser Altimeter System (GLAS) waveforms in
conjunction with spatially concurrent multiple return airborne laser scanning (ALS) point clouds. The
datasets are sourced from multiple sites representing a cross-section of major Australian ecosystems. The
sensitivity of waveform ground return scaling functions to localised ecosystem attributes are highlighted,
with the aim of both: i) better understanding the physics governing these sensitivities; and ii) developing
a generalised approach to GLAS waveform gap fraction assessment that can be applied equally to a wide
range of forest canopy conditions.
We acknowledge CSIRO, the Terrestrial Ecosystem Research Network and Dr. Alex Held of AusCover for
access to multiple sites of ALS data, as well as continued support throughout this study.
84
An individual tree based framework to support operational forest resources
inventory using full-waveform airborne LiDAR data
Jili Li1*, Laird Van Damme1, John A. Kershaw2 and Arnold Rudy1
1. KBM resources Group, 349 Mooney Ave., Thunder Bay, ON, Canada, 8073455445, [email protected],
[email protected], [email protected]
2. Professor, University of New Brunswick, 3 Bailey Drive, Fredericton, NB, Canada, [email protected]
* Presenting Author: Jili Li, 807-345-5445, [email protected]
ABSTRACT
In the past 20 years, laser scanning and ranging (LiDAR) technology has provided enhanced functionality
to advance the Forest Resources Inventory (FRI) and decision making. Several regression models have
been built to link LiDAR metrics with important stand parameters (e.g., basal area and volume). Those
models are generally created using fixed-area sampling plots measured by field crews, which requires
extensive fieldwork. Alternatively, stand parameters can also be estimated with less fieldwork through
the integration of individual tree parameters. To date, the tree-based approach using LiDAR has not been
operationally applied for supporting FRI in Ontario due to the issues such as: low accuracies in individual
tree detection/delineation and species identification; and, high cost in data acquisition and computation.
The objective is to design an operationally feasible framework to support FRI by applying full-waveform
airborne LiDAR data with minimum pulse density, reducing fieldwork, and improving accuracies of
tree/stand parameters. To initialize this process, high-density LiDAR data were flown over a study area
near Thunder Bay, Ontario. These LiDAR data were used to create maps of individual trees and their
attributes, as the baseline for comparison to conventional ground sampling and a progressive reduction
in LiDAR pulse density.
We used the pit-free canopy height model combined with a modified individual tree delineation algorithm
to improve tree detection; developed a simple but effective approach through the fusion of LiDAR and FRI
interpretation data to improve tree species identification; and, estimated stand basal area and piece size
distribution from derived trees parameters. We quantitatively evaluated the baseline framework accuracy
using stands reconstructed from terrestrial LiDAR data. The research also includes a novel comparison
between stand parameters obtained from our framework and those from ground point sampling. Our
results suggest that stand parameters obtained from the baseline framework are comparable with those
from terrestrial LiDAR and ground based samples.
85
Predicting Tree Species at the Tree-Level Using LiDAR
Margaret Penner1*, Murray Woods2, Benoît St-Onge3 and Doug Pitt4
1. Forest Analysis Ltd., 1188 Walker Lake Dr., RR4, Huntsville, ON, CANADA, 705-635-1314,
[email protected]
2. Senior Analyst – Forested Landscapes. Ontario Ministry of Natural Resources & Forestry, 3301 Trout Lake Road,
North Bay, ON, CANADA Phone: 705.475.5561, [email protected]
3 Professor, Département de géographie, Université du Québec à Montréal, 514- 987-3000 x 0280,
[email protected]
4. Research Scientist, Quantitative Silviculture Canadian Wood Fibre Centre, Canadian Forest Service, 1219 Queen
St. E., Sault Ste. Marie, ON, Canada, 705-541-5610, [email protected]
* Presenting Author: Margaret Penner, Forest Analysis Ltd., 705-635-1314, [email protected]
ABSTRACT
LiDAR-derived attributes were used to predict species and species groups at the tree level within a
managed Great Lakes St. Lawrence forest. A data set consisting of 1000 tree crowns was generated from
the automated segmentation of high-resolution LiDAR data followed by species identification by a
certified photo interpreter. All trees of the study area were delineated using local maxima detection and
region growing. The size and shape of trees were computed and the point cloud of each tree extracted.
Eighty-seven percent of the trees were conifer. Ten conifer species and nine hardwood species were
represented in the sample. Approximately 800 trees were selected at random to calibrate logistic
regression models, with the remaining trees reserved for validation. Initially, models were calibrated to
predict trees as hardwood or conifer with a 98% correct classification rate on the validation dataset. Once
the trees were classified as conifer, 74% of the validation trees were correctly classified at the species
level. LiDAR attributes related to tree size were avoided during the modeling in favour of tree shape (e.g.,
skewness of returns) and crown texture attributes (e.g., roughness) along with underlying terrain derived
attributes (e.g., moisture).
The greatest misclassification was associated with larch, jack pine and white spruce. Parallel nonparametric predictions were made from the same calibration dataset using randomForest without loss of
prediction accuracy.
86
Sage grouse hate trees: object-based mapping of juniper encroachment across
the sage-grouse range
Michael J. Falkowski1*, Aaron J. Poznanovic2, Nilam Kayastha3, David E. Naugle4
1. Research Associate Professor, University of Minnesota Department of Forest Resources, 1530 Cleveland Avenue
N
St. Paul, MN 55108, [email protected]
2. Research Fellow, University of Minnesota Department of Forest Resources, 1530 Cleveland Avenue N St. Paul,
MN 55108, [email protected]
3. Research Associate, University of Minnesota Department of Forest Resources, 1530 Cleveland Avenue N
St. Paul,
MN 55108, [email protected]
4. Professor, University of Montana College of Forestry and Conservation, 32 Campus Drive Missoula, MT 59812,
[email protected]
* Presenting Author: Michael J. Falkowski, Research Associate Professor, [email protected]
ABSTRACT
Conifer encroachment in sage-steppe communities has dramatically accelerated over the last 150 years.
Encroachment degrades habitat for sagebrush obligate species, including sage-grouse, by fragmenting
these historically open and expansive sagebrush communities. Research demonstrates that sage-grouse
abandon leks (breeding areas) where conifer cover is as low as 4%. Subsequently, iconic western species
such as the sage-grouse have seen declining populations. To address these threats, we are employing an
objected-oriented algorithm, called Spatial Wavelet Analysis (SWA), to map the location and size of
individual conifer trees across 425,00 km2 of sage-grouse habitat conservation areas in the western United
States. This tree location and size dataset, which is derived from multi-band high-resolution aerial
photography (acquired by the National Agricultural Imagery Program (NAIP), is then used to generate high
resolution maps of conifer canopy cover. This canopy cover product sheds light on the extent and
distribution of tree canopy cover within the sage-grouse range so that land managers can ultimately make
informed decisions on where to improve sage-grouse habitat. The accuracy of the SWA derived canopy
cover estimates was determined via a comparison with independent canopy cover measurements and
further evaluated against other object oriented approaches (image segmentation) as well as traditional
classification methods (random forests, Iterative Self-Organizing Data Analysis (ISODATA), and maximum
likelihood). Tests of statistical equivalence revealed that each of the methods varied considerably in
accuracy, with SWA preforming best when conifer cover is below 40 percent and traditional, non-object
oriented approaches performing better at high canopy covers.
87
Individual tree detection from LiDAR-derived canopy height models (CHM) in
longleaf pine forest
Carlos A. Silva1*, Andrew T. Hudak2, Lee A. Vierling3, E. Louise Loudermilk4, Joseph J. O’Brien5, Aaron
Poznanovic6, Michael Falkowski7, Carlos A. Gonzalez-Benecke8, Steve Jack9, Heezin Lee10
1. Ph.D. Student. Department of Forest, Rangeland, and Fire Sciences, University of Idaho, 875 Perimeter Drive,
Moscow, ID -83844, USA, +1 (208) 596- 4510 , [email protected]
2. Research Forester. USDA Forest Service, Rocky Mountain Research Station, 1221 South Main Street, Moscow, ID83843, USA, +1 (208) 883-2327, [email protected]
3. Associate professor. Department of Forest, Rangeland, and Fire Sciences, University of Idaho, 875 Perimeter
Drive, Moscow, ID 83844, USA, +1 (208) 885-5743, [email protected]
4. Research Ecologist. USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green St., Athens, GA 30602, USA-, +1 (208) 706 559 4309 , [email protected]
5. Research Ecologist. USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green St., Athens, GA 30602, USA-, +1 (706) 559-4336, [email protected]
6. Remote Sensing/GIS Fellow. Department of Forest Resources, University of Minnesota, St. Paul, MN 55108, USA,
+1 (218) 387-4050, [email protected]
7. Research Associate Professor. Department of Forest Resources, University of Minnesota, St.
Paul, MN 55108, USA, +1 (906) 370-7776, [email protected]
8. Program Coordinator. School of Forest Resources and Conservation, P.O. Box 110410, University of Florida,
Gainesville, FL 32611, USA, +1 (352)846-0851, [email protected]
9. Conservation Ecologist. Joseph W. Jones Ecological Research Center at Ichauway, 3988 Jones Center Drive
Newton, GA 39870, USA, +1 (229) 734-4706, ext. 247, [email protected]
10. Research Scientist. National Center for Airborne Laser Mapping, University of California, Berkeley, CA 94720,
USA, +1 (510) 642-3991, [email protected]
* Presenting Author: Carlos Alberto Silva, Ph.D. student, 208 596-4510,
[email protected].
ABSTRACT
Light Detection and Ranging (LiDAR) is currently the best remote sensing technology for characterizing 3D
canopy structure. We detected individual trees from a LiDAR-derived canopy height model (CHM) in a
longleaf pine forest (Pinus palustris Mill) at Ichauway, Georgia, USA. Tree stem locations and heights were
collected in field plots. Individual tree detection from the LiDAR-derived canopy height model (CHM) was
realized using the rLiDAR package in R. To validate the performance of the individual tree detection, the
original LiDAR point cloud was segmented into individual flight lines and treated as independent
observations. A total of 150 square plots of (10m x 10m) were used to evaluate the performance of the
individual tree detection. Within each plot, we counted the number of trees detected from two
independent CHMs generated from overlapping flight line point clouds, and from the CHM generated
from the combined point cloud data. According to preliminary results, in plots where tree density was
<100 trees/ha we had an accuracy of 96% for individual tree detection in each of the three CHMs. In plots
where the density of the trees was between 100-200 trees/ha, accuracy was ~79% , and where the tree
density was >200 trees/ha, accuracy was ~40% from all three CHMs. While we found that tree detection
accuracy suffers in denser stands as expected, we also provide evidence that tree locations derived from
LiDAR CHMs are more accurate than stem locations mapped in the field using traditional methods and
resource-grade GPS.
88
S18 Monitoring Surface Water and Wetland Ecosystems with SAR – I /
S18 Suivi des milieux humides et des surfaces d’eau avec le RSO - I
Brian Brisco, chair / modérateur
Mapping Ephemeral Surface Water Bodies with Radarsat-2
Sandra Bolanos1*, Doug Stiff1, Brian Brisco2, Al Pietroniro3, Stéphane Bélair4
1. Environment Canada, National Hydrological Services; 373 Sussex Dr., Ottawa, Ontario, K1A0H3, Canada
2. Canada Centre for Mapping and Earth Observation, 588 Booth St., Ottawa, Ontario, K1A0Y7, Canada
3. Environment Canada, National Hydrological Services; 11 Innovation Blvd., Saskatoon, Saskatchewan, S7N 3H5,
Canada
4. Environment Canada, Meteorological Research; 2121 route Transcanadienne, Dorval, Quebec, H9P1J3, Canada
* Presenting Author: Sandra Bolanos, Physical Science Officer, 613-943-9786, [email protected]
ABSTRACT
Prairie potholes are water-holding depressions of glacial origin in the prairies of the northern United
States and southern Canada. Studies have found that direct precipitation and spring runoff from snowmelt
are the major sources of water supply to prairie potholes, while evapotranspiration and overflow are the
major causes of water loss (Pomeroy, 2010). Since they generally form in low topographic terrain, their
surficial extent can vary significantly throughout the year making stage/volume relationships difficult to
establish for water quantity measurements and it is hypothesized that their surficial extent plays a
significant role in surface/atmosphere boundary models (Belair, 2014) and in flooding conditions in the
following spring (Pietroniro 2014).
Despite their importance on the water balance of prairie basins, these ephemeral water bodies are not
consistently or broadly monitored. SAR data is a well suited information source for large scale surface
water monitoring, especially in regions where hydrological information is difficult to obtain due to
inaccessibility and rapid change.
Existing threshold-based methods, used for real time flood mapping, require intensive manual input for
water surface mapping, and adjustments on a scene per scene basis. In this work, the feasibility to map
ephemeral water bodies using RADARSAT-2 imagery in Fine Wide mode (~6x6m) in an operational way is
presented, together with an analysis of the accuracy of this and alternative methods. The ability to
measure the extent of individual potholes as well as the ability to detect changes is analyzed using higher
resolution SAR data, optical imagery and lidar.
89
Evaluation of Temporal Filters for SAR Applications in Water Resources
Lori White1*, Ron Caves2, Kevin Murnaghan3, Brian Brisco4
1. Environmental Scientist, Canada Centre for Mapping and Earth Observations (EGS-CCMEO), Natural
Resources Canada, 560 Rochester, Ottawa (ON), K1A 0E4, (613) 759-6485, [email protected]
2. Senior Analyst, Research and Development, MacDonald Dettwiler and Associates Ltd, 13800 Commerce Parkway,
Richmond, British Columbia, Canada, (604) 278-3411, [email protected]
3. Environmental Scientist, Canada Centre for Mapping and Earth Observation, Natural Resources Canada, 560
Rochester, Ottawa (ON), K1A 0E4, (613) 759-6237, [email protected]
4. Research Scientist, Canada Centre for Mapping and Earth Observation, Natural Resources Canada, 560
Rochester, Ottawa (ON), K1A 0E4, (613) 759-1046, [email protected]
* Presenting Author: Lori White, Environmental Scientist, 613-759-6485, [email protected]
ABSTRACT
SAR has long been recognized as a useful sensor for mapping and monitoring surface water and floods
due to its all-weather capability and sensitivity to water in the landscape. The traditional approach of
using spatial filters to remove speckle effects in SAR data is effective for this application but may cause
the loss of spatial information due to the degradation of resolution in the output product.
The advent of SAR constellations with more rapid revisit capability allows one to generate stacks of SAR
data for various applications. This stack allows us to use temporal filters rather than spatial filter for noise
reduction while maintaining resolution. MDA recently developed a temporal filter for agricultural
applications and the GAMMA software package has two temporal filters for SAR processing. This study
will compare these three different temporal filters with the traditional spatial filtering approach using the
signal level ratio, the Equivalent Number of Looks (ENL), the resolution, and the ENL /resolution ratio.
This evaluation is conducted for detecting surface water in the Peace Athabasca Delta and for detecting
phragmites patches in the Georgian Bay Islands National Park.
The results indicate that temporal filtering is an effective way to reduce speckle noise while maintaining
resolution for the processing of stacks of SAR data. This helps maintain the edges and find small targets
such as water bodies or the small phragmites patches in the landscape.
90
Early results of inland water-body detection using multipolarized L-band SAR:
airborne, Aquarius, and SMAP
Seungbum Kim1*, Dara Entekhabi2, Simon Yueh3
1. Research Scientist, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA.
[email protected]. +1-818-354-0435
2. Professor, Massachusetts Institute of Technology, Cambridge, MA, USA
3. Senior Research Scientist, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA
* Presenting Author: Seungbum Kim, Research Scientist, 818-354-0435
[email protected].
ABSTRACT
The Soil Moisture Active Passive (SMAP) satellite will be launched in January 2015 and will provide global
measurements of normalized radar cross section (NRCS) at L-band at 8-day revisit. The dual-polarized (HH
and VV) NRCS will be used to identify inland water-bodies globally with 3-km resolution at 3-day subrepeat, to support water and carbon science, flood mapping, and soil moisture retrieval. Trade-off
between spatial and temporal resolutions would also allow 1-km detection at ~ 32-day interval. We probe
the SMAP’s new capability to resolve challenges with the existing satellites in water-body classification:
temporal resolution, spatial extent, spatial resolution, varying incidence angle, instrument noise, scene
diversity, and the capability to penetrate vegetation and clouds. The proposed approach is to combine a
threshold to the HH/VV ratio at -3 dB and a HH threshold at -25 dB. According to the prelaunch testing
with airborne radar data, the water detection error is smaller than 10 %. The false alarm rate is smaller
than 2%. The proposed approach is robust to the subpixel heterogeneity up to 75% land fraction
contributed by barren surface, 10% by shrubland, and 3% by forests. We will also investigate the utility of
the Co-/Cross-polarization ratio (HV/HH), which is nominally less than -18 dB for water surfaces and
substantially higher for land surfaces. Post-launch validations of the SMAP product include a comparison
with known static targets, and with dynamic targets such as flood events identified by spaceborne optical
or high-resolution Radarsat data.
91
Monitoring water levels by integrating optical and synthetic aperture radar
water masks with lidar DEMs
Chris Hopkinson1*, Brian Brisco2, Shane Patterson3
1. University of Lethbridge, Lethbridge AB Canada. [email protected]
2. Natural Resources Canada, Canada Centre for Remote Sensing, Ottawa ON Canada
3. Government of Alberta, Innovation and Advanced Education, Edmonton AB Canada
* Presenting Author: Chris Hopkinson, university of Lethbridge, [email protected]
ABSTRACT
The ability to map and monitor wetland and lake open water extent and levels across the landscape allows
improved estimates of watershed water balance, surface storage and flood inundation. The study
presents open water classifications over the wetland dominated Sheppard Slough watershed east of
Calgary in western Canada using parallel temporal imagery captured from the RapidEye and RadarSat
satellites throughout 2013, a year of widespread and costly flood inundation in this region. The optical
and SAR-based temporal image stacks were integrated with a high-resolution lidar DEM in order to
delineate regions of inundation on the DEM surface. GIS techniques were developed to extract lidarderived water surface elevations and track the spatio-temporal variation in pond and lake water level
across the watershed. Water bodies were assigned unique identifiers so that levels could be tracked and
linked to their associated watershed channel reach. The procedure of optical image classification through
to merging of individual water bodies into watershed channel topology and extracting reach water levels
has been automated within python scripts. The presentation will describe: i) the procedures used; ii) a
comparison of the SAR and optical classification and water level extraction results; iii) a discussion of the
spatio-temporal variations in water level across the Sheppard Slough watershed; and iv) a commentary
on how the approach could be implemented for web-based operational monitoring and as simulation
initialisation inputs for flood inundation model studies.
92
Multitemporal Interferometric SAR for Accurate Water Extent Mapping
Valentin Poncos1* and Andrew Welch2
1. SAR Scientist, Kepler Space Inc., 72 Walden drive, Ottawa, ON, Canada, 613-883-8184, [email protected]
2. President , JWRL Inc., 128 Creekside Drive, Ottawa, ON, Canada, 613-454-5807, [email protected]
* Presenting Author: Valentin Poncos, SAR Scientist, 613-883-8184, [email protected]
ABSTRACT
Flooding events are usually associated to rainy weather when the cloud coverage makes impossible the
use of the space-borne optical systems. Synthetic Aperture Radar has the capability of penetrating the
cloud coverage; this capability, associated with its own illumination that makes it usable during the night,
makes SAR desirable for operational (almost real-time) monitoring of natural hazards. Moreover, the
lower (than optical) operating frequency of SAR makes possible detecting water under vegetation, a
characteristic of wetlands and flooded areas. The water extent maps produced with the C-band Radarsat2 combine open-surface water (visible also with optical sensors) with water under vegetation (mostly
invisible to optical sensors) as well as water-saturated soil (invisible to optical frequency). These maps are
more accurate than any other water-extent map produced with optical sensors because they reveal water
masked by vegetation (such as forests, reeds), by surficial organic material covering the ground as well as
water in highly-saturated soil.
The work presented in this paper is related to three projects financed by the Canadian Space Agency:
1. R&D for Multi Satellite Data Integration, Earth Observation Applications and Utilization.
2. Earth Observation Applications Development Program (EOADP), Development of Applications for
Environmental Monitoring and Remediation.
3. Rapid Information Products and Services (RIPS), First Nation Community Monitoring (Aboriginal Affairs
and Northern Development Canada).
The goal is to produce accurate water extent maps on various ground conditions and vegetation cover.
The processing methodology exploits the SAR amplitude, interferometric phase coherence and the
temporal variation of these parameters.
93
S19 Coastal HF Radar Applications / S19 Applications des radars HF
côtiers
Weimin Huang, chair / modérateur
An Ionospheric Reflection Coefficient Model for HF Ionosphere-Ocean
Propagation
Shuyan Chen1*, Eric W. Gill2, and Weimin Huang3
1. PhD Student, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, 240 Prince
Philip Dr., St. John’s, NL, Canada, A1B 3X5, 1-(709)7659680, [email protected]
2. Professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, 240 Prince Philip
Dr., St. John’s, NL, Canada, A1B 3X5, 1-(709)8648922, [email protected]
3. Assistant professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, 240
Prince Philip Dr., St. John’s, NL, Canada, A1B 3X5, 1-(709)8648937, [email protected]
* Presenting Author: Shuyan Chen, PhD Student, 1-(709)7659680, [email protected].
ABSTRACT
High frequency surface wave radar (HFSWR) has been accepted as an important remote sensing device
for sea state monitoring. However, the performance of HFSWR may be significantly impacted by
unwanted echoes such as ionosphere clutter. During transmission, a portion of the radar radiation may
travel upwards to the ionosphere from the transmitting antenna. This may be partially reflected back to
the receiving antennas directly or via the ocean surface, the latter being referred to as ionosphere-ocean
propagation.
Based on the previously derived high frequency (HF) radar cross section for ionosphere-ocean propagation
involving a pulsed source, a modified the ionosphere reflection coefficient (IRC) is presented in this paper.
The physical influences of the ionospheric electron density on HF radar Doppler spectra are taken into
account in this new model. The relationship between the IRC and the ionospheric electron density is
derived according to a layered ionosphere model. Then, a typical power-law distribution for the electron
density is incorporated into the developed spectral density function of the IRC. Finally, the normalized
power for ionosphere-ocean propagation relative to the average first-order ocean clutter power is
simulated for different horizontal ionospheric plasma drift velocities and wind directions.
94
Comparison of Antenna-Motion-Incorporated High Frequency Bistatic Radar
Cross Sections of the Ocean Surface with Earlier Models
Yue Ma1*, Eric W Gill2 and Weimin Huang3
1. PhD student, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St. John’s, NL,
Canada, (709)-765-0702, [email protected]
2. Professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St. John’s, NL,
Canada, (709)-864-8922, [email protected]
3. Assistant Professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St.
John’s, NL, Canada, (709)-864-8937, [email protected]
* Presenting Author: Yue Ma, PhD student, 709-765-0702, [email protected].
ABSTRACT
In this paper, the first- and second-order high frequency bistatic radar cross sections of the ocean surface
with the transmitter on a floating platform are presented. When there is no antenna motion, the bistatic
radar cross section incorporating antenna motion is shown to reduce to that for the fixed antenna case.
Also, when the backscatter geometry is introduced into the bistatic radar cross section which accounts for
antenna motion, the model is seen to reduce to an earlier monostatic result. The bistatic radar cross
section with antenna motion, the monostatic radar cross section with antenna motion and the bistatic
radar cross section with fixed antennas are illustrated. Comparison of the results shows that additional
peaks caused by the swaying motion of the platform are symmetrically distributed in the spectra and
these motion-induced peaks have less energy in the bistatic cross section than those in the monostatic
case. It is also found that the effect of the antenna motion on the second-order radar cross section is
much smaller than that on the first-order case.
95
Design and Implementation of a High-Frequency Software-Defined Radar (HFSDR): Initial Results
Khalid El-Darymli1*, Noah Hansen 2, and Eric W. Gill3
1. Mr. Khalid El-Darymli is a Senior Engineer with Northern Radar Inc, 25 Anderson Avenue, St. John's,
Newfoundland, A1B 3E4, Canada. Email: [email protected]
2. Mr. Noah Hansen is a Senior Engineer with Northern Radar Inc, 25 Anderson Avenue, St. John's, Newfoundland,
A1B 3E4, Canada. Email: [email protected]
3. Dr. Eric W. Gill is a professor of electrical and computer engineering with Memorial University of Newfoundland,
St. John's, Newfoundland, Canada, A1B 3X5. Email: [email protected]
* Presenting Author: Khalid El-Darymli, Senior Engineer, [email protected]
ABSTRACT
Northern Radar Inc (NRI) is in the process of developing a high-frequency software-defined radar (HFSDR). The transceiver of the system is implemented using software-defined radio (SDR) technology, and
the radar utilizes phased-array antennas. In its current set-up, the system is designed to operate in the
surface-wave mode. Hence, its main end-market application is oceanography. The SDR is a generic radio
communication system in which components that have been typically implemented in hardware (e.g., the
mixer, modulator, demodulator, detectors, etc.) are instead implemented by means of software on a
personal computer or an embedded device. When compared to standard state-of-the-art HF radars in the
market today, the two main advantages of the SDR-based system are flexibility and cost-efficiency.
Figure 1 illustrates the overall architecture of the HF-SDR system. Primarily, SDRs connected to a host
computer function as the transceiver of the system. To provide for long-term clock stability and to
synchronize the receive and transmit channels across multiple SDRs, a GPS disciplined oscillator (GPSDO)
is used. For demonstration purposes, a representative range-Doppler map, acquired using this system
from the on-site HF antennas, installed by NRI for the HF-ROSA project [1] at the former Naval Station
Argentia, Newfoundland [2], is shown in Figure 2. In the full-paper version of this abstract, a more
thorough description will be offered along with additional results.
REFERENCES
[1] High Frequency Radar Ocean Surface Applications (HF-ROSA). (2014, 12) Memorial University of
Newfoundland. [Online]. Available: http://hfrosa.engr.mun.ca/
[2] Naval Station Argentia. (2014, 12) Wikipedia. [Online]. Available:
http://en.wikipedia.org/wiki/Naval_Station_Argentia
96
Figure 1: The overall architecture of NRI's HF-SDR radar.
z
Figure 2: Range-Doppler map due to a backscatter collected on October 29, 2014 from a single receive element. The data
collection utilized the on-site HF antennas installed by NRI for the HF-ROSA project (HFROSA). This test used a linear frequency
modulated continuous wave (LFMCW) up-chirp with the following radar parameters, center frequency = 13.3853 MHz, sweep
bandwidth = 100 kHz, sweep time = 0.784 s, and integration time = 20 minutes.
97
Space-Time Adaptive Processing for Coastal Surveillance Applications
Raviraj Adve1* and Maryam Ravan2
1. Professor, Dept. of Electrical and Computer Engineering, University of Toronto, 10 King’s College Road, Toronto,
ON M5S3G4, Canada, (416) 946 7350 [email protected]
2. Dept. of Electrical and Computer Engineering, University of Toronto, 10 King’s College Road, Toronto, ON
M5S3G4, Canada, (416) 946 7350 [email protected]
* Presenting Author: Raviraj Adve, Professor, 416 946 7350, [email protected]
ABSTRACT
This paper presents the use of space-time adaptive processing (STAP) algorithms for coastal surveillance.
STAP algorithms generally address interference, whether homogeneous or not, but are tested in
environments with a single target or with targets that do not interact with each other. However, a high
frequency surface-wave radar (HFSWR) monitoring off-shore activity likely sees many targets closely
spaced in angle, Doppler and range. Target detection in coastal surveillance applications is further
complicated by nonhomogeneous sea clutter; schemes, such as STAP, that are based on estimating
interference statistics, are particularly vulnerable to clutter heterogeneity. Clutter heterogeneity also
severely limits data available to train the adaptive processor.
In this paper we present the fast fully adaptive (FFA) approach, a STAP algorithm that has an inherent
robustness against heterogeneity. Inspired by the fast Fourier transform, the algorithm sub-divides the
overall STAP problem into a series of smaller problems; the outputs of each sub-problem are combined
adaptively. This is repeated until a single statistic is obtained. By reducing the problem size, the algorithm
benefits from significantly reduced required training; furthermore, the repeated combinations of outputs
provides an inherent robustness against heterogeneity. The proposed algorithm is tested on measured
data provided by Raytheon Canada. A comparison is made to both non-adaptive processing and to
another STAP algorithm, Joint Domain Localized processing, which also requires limited training.
98
Comparison of Spectral Estimation Methods for Rapidly-varying Currents
Obtained by High Frequency Surface Wave Radar
Wei Wang1*, Eric W. Gill2
1. Post doctor, Memorial University of Newfoundland, A1B 3X5, St John’s, NL, Canada, 763-2865, [email protected]
2. Professor, Memorial University of Newfoundland, A1B 3X5, St John’s, NL, Canada, 864-7922, [email protected]
* Presenting Author: Wei Wang, Post doctor, 763-2865, [email protected].
ABSTRACT
A comparative study of both conventional spectrum estimation methods, e.g., the periodogram method,
and modern techniques, such as the autoregressive and multiple signal classification methods, for current
mapping by a high frequency surface wave radar is undertaken. The innovation of this paper is that this
analysis is based on time series that are much shorter than the commonly adopted coherent integration
times used in such systems. This reduction in the signal length will cause two problems for the
conventional periodogram method, namely, poor Doppler resolution and reduction in signal-to-noise
ratio. Also, to calculate the radial current velocity, it is important to estimate its associated Doppler shift
from the frequency spectrum. Two Doppler shift identification methods, including the conventional
centroid method and the symmetric-peak-sum method, are examined in conjunction with each of the
spectral estimation techniques. A weighted sum of the current estimates using the two Doppler shift
identification methods is also recommended to provide a lower RMS difference. The weighting ratio is
optimized using a genetic algorithm. Field data comparison with current measurements obtained from a
current meter indicate that, in this case, the use of a combination of the high-resolution spectral
estimation methods is capable of providing the same rms difference level for short and long time series,
while the RMS of the currents obtained from the periodogram method increase dramatically for short
time series. Significant improvements in the retrieved current velocities for both stationary and rapidlyvarying current conditions have been shown.
99
S20 Future Canadian Earth Observation Missions / S20 Futures
missions canadiennes en observation de la Terre
Martin Bergeron, chair / modérateur
The Canadian Hyperspectral Mission Concept
Gary Buttner1, Jennifer Busler2*, Antonio Sanz3, Shen-En-Qian4, Martin Bergeron5, Ralph Girard6,
Karl Staenz7 and Jinkai Zhang8
1. Business Development Manager, Surveillance and Intelligence Group, MDA Systems Ltd., 13800
Commerce Parkway, Richmond, British Columbia, Canada, (604) 278-3411,
[email protected]
2. Senior Analyst, Surveillance and Intelligence Group, MDA Systems Ltd., 13800 Commerce Parkway,
Richmond, British Columbia, Canada, (604) 278-3411, [email protected]
3. Project Manager, Surveillance and Intelligence Group, MDA Systems Ltd., 13800 Commerce Parkway,
Richmond, British Columbia, Canada, (604) 278-3411, [email protected]
4. Senior Engineer (Optical), Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, Québec, Canada,
(450) 926-4618, [email protected]
5. Senior Programs Officer, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, Québec, Canada,
(450) 926-7728, [email protected]
6. Manager Radars and Antennas (Space Utilization), Canadian Space Agency, 6767, route de l'Aéroport,
Saint- Hubert, Québec, Canada, (450) 926-4654, [email protected]
7. Professor, Alberta Terrestrial Imaging Centre and Department of Geography, University of Lethbridge,
4401 University Drive, Lethbridge, Alberta, Canada, (403) 329-2047, [email protected]
8. Research Associate, Alberta Terrestrial Imaging Centre and Department of Geography, University of
Lethbridge, 4401 University Drive, Lethbridge, Alberta, Canada, (403) 329-2047, [email protected]
* Presenting Author: Jennifer Busler, Senior Analyst, 604 278-3411, [email protected].
ABSTRACT
The Government of Canada has identified wide-angle hyperspectral system remote sensing as a highly
desired capability for both government and commercial users involved in the day-to-day management
of the nation’s agricultural, forestry, marine, environmental, and other resources. In response to this,
the Canadian Space Agency has funded a variety of feasibility studies aimed at a potential Canadian
Hyperspectral Mission (CHM) which would deliver the necessary information products based upon a
constellation of microsatellites.
The CHM mission concept differs from traditional research-oriented systems in that the spectroradiometric and imaging performance are tailored to support frequent sub-weekly repeat coverage of
Canada’s enormous land and coastal regions. This combination of optical performance and large-area
coverage presents a variety of technical challenges, particularly for a low-cost microsatellite system.
However, early CHM studies have shown that it is indeed feasible to provide this capability with a 10 nm
VNIR hyperspectral system and a variety of multispectral SWIR bands using a constellation of three
microsatellites.
In this paper, we review the needs and applications which drive CHM, some of the image quality
analyses which support the proposed system performance, and the overall high-level system end-to-end
concept including the spacecraft and ground segments.
100
The Earth Observation Potential of the Polar Communication and Weather
(PCW) Mission
Martin Bergeron1*, Louis Garand2, Alexander P. Trishchenko3, Pierre Langlois4
1. Senior Programs Officer, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, Québec, Canada, (450)
926-7728, [email protected]
2. Research Scientist, Canadian Meteorological Centre, 2121, route Transcanadienne, Dorval, Québec, Canada,
(514) 421-4749, [email protected]
3. Research Scientist, Canada Centre for Remote Sensing, 560 Rochester Street, Ottawa, Ontario, Canada, (613)
759-1446, [email protected]
4. Program Lead, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, Québec, Canada, (450) 9266796, [email protected]
* Presenting Author: Martin Bergeron, Senior Programs Officer, Planning, 450 926-7728,
[email protected].
ABSTRACT
In 2007, the Canadian Space Agency in partnership with the Department of National Defense (DND),
Environment Canada (EC), and other Government Departments, initiated the Concept Development and
Requirements Identification for a Polar Communication and Weather (PCW) Mission. It proved that a
system of two satellites operating in Highly Elliptical Orbit (HEO) could provide continuous 24/7
broadband communications services and monitor arctic weather and climate change at the required
temporal and spatial resolution, throughout all of the Arctic. This was followed by a Mission Analysis and
Concept Definition study that ended in 2011 which also considered optional Science Payloads. Since,
industry has responded to a Request for Information, whose summary of Feedback and Outcomes has
been released in 2014.
PCW is conceived as a suite of payloads serving Military and Civil communications, Meteorology and Space
Weather in the arctic. This paper focuses on those capacities most relevant to Earth Observation, namely
surface parameters obtainable from the Meteorology payload and Auroral Imaging from an optional
Ultra-Violet (UV) Imager. The PCW Meteorology payload has very similar characteristics to that of GOESR (0.5-2 km resolution) and would deliver important environmental and climatic applications such as sea
or land surface temperature, land cover assessment, snow/ice cover, fire disturbances, atmospheric
motion vectors (winds), and more. With regard to Space Weather, an optional UV Imager could deliver
the first continuous real-time imagery of the entire auroras oval over the Arctic for accurate nowcasting
of the space weather situation useful to space-based and ground-based asset protection, trans-polar flight
operations and radio signals distortion.
The WaterSat Mission Concept
Martin Bergeron1*, Shen-En Qian2, Michael Ott3, Gary Buttner4
1. Senior Programs Officer, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, Québec, Canada, (450)
926-7728, [email protected]
2. Senior Engineer (Optical), Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, Québec, Canada,
(450) 926-4618, [email protected]
3. Senior Scientific Advisor, Canadian Hydrographic & Oceanographic Services, 200 Kent Street, Ottawa, Canada,
613-991-6988, [email protected]
4. Business Development Manager, Surveillance and Intelligence Group, MDA Systems Ltd, 13800 Commerce
Parkway, Richmond, British Columbia, Canada, (604) 278-3411, [email protected]
* Presenting Author: Martin Bergeron, Senior Programs Officer, Planning, 450 926-7728,
[email protected].
ABSTRACT
In 2014, the Canadian Space Agency initiated five feasibilities studies for microsatellites (≤150kg) selected
on the basis of Canadian Government needs. Of these, the Canadian Coastal and Inland Waters (WaterSat)
mission concept responds to the need to improve the monitoring and better understand the quality and
productivity of coastal and inland waters as these needs have an important impact on the health of the
population and on economic activities. WaterSat aims to: 1. Provide coastal waters and medium to large
bodies freshwater ecological information, 2. Monitor hazards, discharges, effluents and pollution events,
3. Assess productivity of marine coastal ecosystems, 4. Characterize Harmful Algal Blooms (HAB), and 5.
Monitor medium and large bodies freshwater quality. This initiative was supported by government
departments including Fisheries & Oceans Canada, Environment Canada (Water Science & Technology
Directorate and the Canadian Ice Service), Natural Resources Canada, Defence Research and Development
Canada, Public Health Agency of Canada, and Agriculture and Agri-Food Canada.
The feasibility study concluded that WaterSat could be implemented on a microsatellite platform as a
pushbroom hyperspectral sensor covering the spectral range from 400 to 1000nm with a 7.5nm spectral
sampling. A 100m Ground Sampling Distance (GSD) would be achieved over a 300km swath. Signal-tonoise of more than 400:1 (for reference albedo of 5%) would allow the retrieval of the intended products.
This paper provides an overview of the mission requirements and proposed concept. Expected benefits
are also presented.
102
RADARSAT Constellation Mission
Daniel De Lisle1* Steve Iris
1. RCM Data Utilization & Applications, Canadian Space Agency, 6767 route de l’aéroport, Longueuil, QC, Canada,
J3Y 8Y9, 450-926-6611, [email protected]
2.RADARSAT Constellation Mission Manager, Canadian Space Agency,6767 route de l’aéroport, Saint-Hubert, Qc,
Canada 450-926-6554, [email protected]
* Presenting Author: , Daniel De Lisle, RCM Data Utilization & Applications, 450-926-6611,
[email protected]
ABSTRACT
The RADARSAT Constellation is the next step in evolution of the RADARSAT Program with the objective of
ensuring data continuity, improved operational use of synthetic aperture radar (SAR) data and improved
system reliability.
The mission consist of three identical C-band SAR satellites flying in a constellation which will provide
complete coverage of Canada's land and oceans offering an average daily revisit, as well as a potential
daily access to 95% of any location on the globe.
The main objective of the RADARSAT Constellation Mission is on meeting Government of Canada User
Department’s needs and requirements in Core Use Areas such as Maritime Surveillance, Disaster
Management, Ecosystem Monitoring and Northern Development.
The constellation is designed primarily as a wide area monitoring system, offering medium resolution
data, but it will also offers high resolution imaging capabilities, including a Spotlight Mode, as well as
multiple polarization including Compact Polarimetry and Experimental Quad-Polarization.
The greatly enhanced temporal revisit combined with accurate orbital control will enable advanced
interferometric applications in between satellites on a four-day cycle that will allow the generation of very
accurate coherent change maps. RCM frequent revisit capability, near real-time SAR data availability and
vessel identification capabilities (through an AIS payload) will provide the capability to identify and
monitor ships up before they enter Canadian waters or ports which will increase the Safety, Sovereignty
and Security of Canada.
The RADARSAT Constellation Mission is currently under construction with satellite launches planned for
2018.
This presentation will describe the RCM space and ground segments, and provide an overall project status.
It will focus on the programmatic and technical challenges and realizations, and discuss activities
surrounding application development and operational readiness by the government of Canada
departmental users.
103
A Microsatellite-based Canadian Wildland Fire Monitoring System
Timothy J. Lynham1, Joshua M. Johnston2, Marleen van Mierlo3*, Brian Lawrence4, Paul Briand5, Linh Ngo
Phong6, Jean-François Hamel7, Denis Dufour8
1. Forest Fire Research Project Leader, Natural Resources Canada, CFS, 1219 Queen St. E., Sault Ste. Marie, ON,
Canada, 705-541-5537, [email protected]
2. Forest Fire Analyst, Natural Resources Canada, CFS, 1219 Queen St. E., Sault Ste. Marie, ON, Canada, 705-5415548, [email protected]
3. Senior Engineer, Space Utilization Development, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert,
QC, Canada, 450-926-4576, [email protected]
4. Engineer, Space Utilization, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, QC, Canada, 450926-4657, [email protected]
5. Manager, Earth Observation Applications and Utilization Programs, Directorate of Space Utilization, Division of
Mission Engineering and Applications, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, QC,
Canada, 450-926-6737, [email protected]
6. Research Scientist, Canadian Space Agency, 6767, route de l'Aéroport, Saint-Hubert, QC, Canada, 450-926-4678,
[email protected]
7. Project Manager, NGC Aerospace, 1650 rue King Ouest, Bureau 202, Sherbrooke QC, Canada, 819 348-9483 ext.
225, [email protected]
8. Research Scientist, INO, 2740 Einstein Street, Québec, QC, Canada, 418-657-7006, [email protected]
* Presenting Author: Marleen van Mierlo Senior Engineer, 450-926-4576,
[email protected]
ABSTRACT
In 2011, the Canadian Space Agency (CSA) created an initiative for developing microsatellite missions that
will be part of the next generation of Canadian satellites. Five candidate missions were selected based on
their ability to respond to the needs of the Canadian government, to be low cost, and to provide a fast
turnaround for development. The plan is to launch one microsatellite every two years with the
development of the first mission starting in 2016 with a possible launch date in 2020. The selection of the
first mission will be take place in the summer of 2015.
One of the five microsatellite mission candidates is the Canadian Wildland Fire Monitoring System
(CWFMS). CWFMS has a strong user community consisting of Canadian government departments (led by
NRCAN) and academic partners. The current proposal for the CWFMS mission includes a novel, Canadian
low-cost thermal imaging technology called a micro-bolometer developed jointly by CSA and Institut
national d'optique (INO). This sensor has a space heritage after Canada provided a similar detector for
launch on a joint NASA/Argentine satellite in 2010.
The objective of CWFMS is to develop a Canadian satellite system that has the ability to monitor and track
wildland fires within Canada. CWFMS will deliver near real-time information in support of wildland fire
management to governments and commercial enterprises by providing information such as fire locations
(hotspots), fireline intensity, rate of fire spread, burned area mapping and Fire Radiative Power (FRP). FRP
is a measure of the rate of energy released from a fire and can be used to estimate carbon emissions from
wildfires. Canada has international obligations for reporting on carbon emissions and for air quality
forecasting.
Fires have important positive ecological effects but are not tolerated when they threaten human lives,
the loss of homes, and cause the evacuation of communities due to the effect of smoke on human health
and comfort. Fires also impact the Canadian economy through the loss of timber and commercial
104
infrastructure. Much of Canada is lightly populated therefore satellite monitoring can augment airborne
fire monitoring systems, especially in remote areas.
CWFMS will use a single satellite of 75-150 kg that will image in a sun-synchronous, low earth orbit of 500850 km above the earth. It will access the whole of Canada every 24 hours using Canadian ground
receiving stations. The design life will be 2 years. The CWFMS microsatellite mission is currently
undergoing a feasibility study (Phase 0) led by NGC Aerospace, that will be completed in March 2015.
CWFMS could be a stepping-stone towards an operational fire monitoring service using a constellation of
satellites.
105
S21 3D Characterization of Canopy Structure using LiDAR and New
Technologies – I / S21 Caractérisation du couvert végétal avec le
LiDAR et les nouvelles technologies – II
Andrew Hudak, chair / modérateur
Characterizing Fine-scale Understory Plant Diversity and Fuels from Threedimensional Photogrammetry Points
Benjamin C. Bright1*, E. Louise Loudermilk2, Scott Pokswinski3, Andrew T. Hudak4, Joseph J. O’Brien5
1. Geographer, USDA Forest Service, Rocky Mountain Research Station, Forestry Sciences Laboratory, 1221 South
Main Street, Moscow, ID 83843, USA, (208) 883-2311, [email protected]
2. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4309, [email protected]
3. Ecologist, University of Nevada at Reno, 1664 N. Virginia MS 0314, Reno, NV 89557, USA,
[email protected]
4. Research Forester, USDA Forest Service, Rocky Mountain Research Station, Forestry Sciences Laboratory, 1221
South Main Street, Moscow, ID 83843, USA, (208) 883-2327, [email protected]
5. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4336, [email protected]
* Presenting Author: Benjamin C. Bright, Geographer, (208) 883-2311, [email protected]
ABSTRACT
Longleaf pine ecosystems of the southeastern United States are important fire-frequented habitats for
many species of concern. Understory plant communities of longleaf pine ecosystems are often highly
diverse at sub-meter spatial scales, which is driven by low-intensity frequent fire regime (1-5 years).
Methods for quantifying understory plant diversity at fine scales will help in efforts to monitor longleaf
pine ecosystems, by relating them to surface fuel and fire behavior characteristics that are constrained by
the pine overstory. Methods for measuring fuels at fine spatial scales is an important step to
understanding plant-fire interactions across scales.
We measured understory plant diversity and fuel cell type and structure on ten 3-m2 plots at a 10 cm
resolution. For these same plots, we used photogrammetry to derive three-dimensional point clouds
representing understory plant height and color (RGB). Point clouds were summarized into various height
and density distributional metrics, similar to conventional LIDAR. Comparison of height and density
metrics with in situ measurements showed that photogrammetry points accurately measured plant and
fuel heights. We will discuss predictions of understory plant diversity and fuel cell type from
photogrammetry metrics using non-linear analyses, such as the Random Forest algorithm. This research
illustrates a new, more affordable alternative to using terrestrial LIDAR to characterize fine-scale 3D
surface vegetation and fuels.
106
3D Modeling of Understory Fuels Determined by Overstory Structure
E. Louise Loudermilk1*, Eric Rowell2, Andrew T. Hudak3, Joseph J. O’Brien4, Benjamin C. Bright5, Carl
Seielstad6, Scott Pokswinski6, Lee Dyer7
1. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4309, [email protected]
2.Remote Sensing/Programmer Analyst, National Center for Landscape Fire Analysis, College of Forestry and
Conservation, University of Montana, 32 Campus Drive, Missoula, MT 59812, USA, (406) 243-2000,
[email protected]
3. Research Forester, USDA Forest Service, Rocky Mountain Research Station, Forestry Sciences Laboratory, 1221
South Main Street, Moscow, ID 83843, USA, (208) 883-2327, [email protected]
4. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4336, [email protected]
5. Geographer, USDA Forest Service, Rocky Mountain Research Station, Forestry Sciences Laboratory, 1221 South
Main Street, Moscow, ID 83843, USA, (208) 883-2311, [email protected]
6. Associate Research Professor, National Center for Landscape Fire Analysis, College of Forestry and Conservation,
University of Montana, 32 Campus Drive, Missoula, MT 59812, USA, (406) 243-6200,
[email protected]
7. Ecologist, University of Nevada at Reno, 1664 N. Virginia St., Reno, NV 89557, USA, [email protected]
8. Professor, University of Nevada at Reno, 1664 N. Virginia St., Reno, NV 89557, USA, [email protected]
* Presenting Author: E. Louise Loudermilk, Research Ecologist, (706) 559-4309, [email protected]
ABSTRACT
Overstory structure is the main driver of surface fuels in frequent fire conifer ecosystems. Directly, the
pines provide the fine fuels (pine needles) that are an easily ignitable and continuous fuel source across
stands and landscapes. Indirectly, the pines influence the understory ground cover (shrubs, forbs, grasses)
through below or aboveground processes of plant-plant interaction and influences on fine-scale fire
behavior, creating an intricate cycle of fuel, fire, and fire effects. Understanding how surface fuels relate
to overstory structure provides the link to understanding plant diversity patterns across these fire
maintained systems.
We present a cross scale analysis and modeling study of fine-scale fuel patterns as predicted by the
overstory in a xeric longleaf pine (Pinus palustris) forest, located at Eglin Air Force Base in NW FL, USA.
Our 3D overstory structure (height, location, etc.) across a pine dominated stand was determined by low
density (1 m mean point density) aerial LIDAR. Surface fuels were measured using imagery and
photogrammetry techniques as well as point-intercept and biomass sampling in spatially located plots.
These data are used to aid in a 3D rendering process, where plot-based data informs the creation of 3D
plant models which are distributed in relation to overstory distribution. We used a cross-validation
approach (e.g., bootstrapping) to determine the accuracy of our results for predicting fuels (i.e., plant
surface area, biomass, and bulk density) across a stand. This approach will be used to further predict fire
behavior and fire effects, especially in relation to understory plant community dynamics.
107
Modeling 3D understory structure for use in physics-based fire behavior
simulations
Eric Rowell1, Carl Seielstad2, E. Louise Loudermilk3*, and Joseph J. O’Brien4
1. Remote Sensing/Programmer Analyst, National Center for Landscape Fire Analysis, College of Forestry and
Conservation, The University of Montana, Missoula, MT 59812, (406) 243-2000, [email protected]
2. Assistant Professor, National Center for Landscape Fire Analysis, College of Forestry and Conservation, The
University of Montana, Missoula, MT 59812, (406) 243-2000, [email protected]
3. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4309, [email protected]
4. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4336, [email protected]
* Presenting Author: E. Louise Loudermilk, Research Ecologist, 706 559-4309, [email protected]
ABSTRACT
As complexity of fire behavior models increase utilizing advanced physics based simulations (i.e.
HIGRAD/FIRETEC and Wildland–Urban Interface Fire Dynamics Simulator), improved estimation and
characterization of surface fuels is necessary to understand the complex relationship between fuels,
topography, and local meteorology that drive wildland fire. Recently, the employment of both airborne
and terrestrial LiDAR to estimate surface fuel availability, distribution, and mass in longleaf pine (Pinus
palustris) ecosystems of the southeastern United States suggest that even these systems struggle to
adequately characterize fuel objects in sufficient detail to extract meaningful biophysical characteristics
to drive physics-based fire behavior models.
We present an analysis and modeling research string of fine-scale fuel patterns in xeric longleaf pine
forests at Eglin Air Force Base in the Florida panhandle, USA. For this research, 1m sample plot data were
used to generate 3D plant models of dominant species of these ecosystems, using the ONYX plant library.
Surface fuels were measured using point-intercept and biomass sampling protocols for spatially located
plots. Combining high resolution nadir imagery plant objects were distributed across 3D space. A 3D
Monte Carlo Ray Tracing algorithm (librat) was used to simulate terrestrial LiDAR point clouds for these
landscapes to improve our understanding of energy/matter interactions and the ability of LiDAR to
characterize object-based fuel elements in the understory. We believe this approach will produce better
estimates of fuel characteristics in these ecosystems as well as provide a ladder between remotely sensed
and in situ field data.
108
Canopy derived fuels drive patterns of in-fire energy release and plant
community assembly in longleaf pine (Pinus palustris) woodlands
Joseph J. O’Brien1*, E. Louise Loudermilk2, Benjamin Hornsby3, Andrew T. Hudak4, Benjamin C. Bright5,
Lee Dyer6, Scott Pokswinski7
1. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4336, [email protected]
2. Research Ecologist, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green Street, Athens, GA 30602, USA, (706) 559-4309, [email protected]
3. Forester, USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320 Green
Street, Athens, GA 30602, USA, (706) 559-4307, [email protected]
4. Research Forester, USDA Forest Service, Rocky Mountain Research Station, Forestry Sciences Laboratory, 1221
South Main Street, Moscow, ID 83843, USA, (208) 883-2327, [email protected]
5. Geographer, USDA Forest Service, Rocky Mountain Research Station, Forestry Sciences Laboratory, 1221 South
Main Street, Moscow, ID 83843, USA, (208) 883-2311, [email protected]
6. Professor, Biology & Director, Ecology, Evolution and Conservation Biology Graduate Program, University of
Nevada, Reno, NV 89557, USA, [email protected]
7. Ecologist, University of Nevada at Reno, 1664 N. Virginia MS 0314, Reno, NV 89557, USA,
[email protected]
* Presenting Author: Joseph J. O’Brien, Research Ecologist, (706) 559-4336, [email protected]
ABSTRACT
Wildland fire radiant energy emission is one of the only measurements of combustion that can be made
at high temporal and spatial resolutions. Furthermore, spatially and temporally-explicit measurements
are critical for making inferences about fire effects and useful for examining patterns of fire spread. While
the correlation between fire frequency and plant biological diversity in longleaf pine (Pinus palustris) and
other frequently burned coniferous forests is well-known, the ecological mechanisms driving this
relationship are unknown. Here we describe our efforts at connecting spatial variability in fire energy
release to fuels and fire effects using longwave infrared (LWIR) thermal imagery. We compare nadir LWIR
imagery captured at fine (1 to 4 cm) scales to three dimensional fuel structure at similar scales and to
plant demography. We hypothesize that the observed variability in fire radiative energy release could be
the causal mechanism driving plant community assembly rules in these systems. We hypothesize that
since high plant diversity exists at very fine scales (30+ vascular plants per m2) in a relatively homogeneous
substrate, neutral processes are likely driving patterns of diversity seen in the frequently burned study
sites. Specifically, localized patches of high intensity fire driven by overstory derived fuels causes random
mortality in understory plants, a crucial element of ecological neutral theory. We analyzed fine-scale
spatial heterogeneity of fire radiant power and energy released in several experimental burns and found
a close connection among patterns of fire intensity and plant community responses. Our measurements
also illustrated the significance of fuel characteristics, particularly type and connectivity in driving spatial
variability of fire intensity at fine-scales. Spatially and temporally resolved data from these techniques
show promise to effectively link the combustion environment with post- fire processes, remote sensing
at larger scales, and wildland fire modeling efforts.
109
Upscaling tree density measures from environmental monitoring plots across
Eglin Air Force Base using low density lidar
Andrew T. Hudak1*, Benjamin C. Bright2, E. Louise Loudermilk3, Joseph J. O’Brien4, Carlos A. Silva5, Lee A.
Vierling6
1. Research Forester, US Forest Service, Rocky Mountain Research Station, Moscow, ID, USA, 208.883.2327,
[email protected]
2. Geographer, US Forest Service, Rocky Mountain Research Station, Moscow, ID, USA, 208.883.2311,
[email protected]
3. Research Ecologist, US Forest Service, Southern Research Station, Athens, GA, USA, 706.559.4309,
[email protected]
4. Research Ecologist, US Forest Service, Southern Research Station, Athens, GA, USA, 706.559.4336,
[email protected]
5. Ph.D. Student, University of Idaho, Department of Forest, Rangeland and Fire Sciences, Moscow, ID, USA,
208.885.5743, [email protected]
6. Associate Professor, University of Idaho, Department of Forest, Rangeland and Fire Sciences, Moscow, ID, USA,
208.596.4510, [email protected]
* Presenting Author: Andrew T. Hudak, Research Forester, 208.883.2327, [email protected]
ABSTRACT
Longleaf pine ecosystems at Eglin Air Force Base (AFB) and elsewhere in the southeastern United States
are fire dependent. Land managers maintain forest health by frequently burning surface fuels. Frequent
burns limit encroachment by competing tree species, preclude duff and fine fuel accumulations, expose
mineral soil, and facilitate longleaf pine seedling recruitment. The frequent fire regime makes longleaf
pine cones the predominant source of coarse woody debris on the forest floor. Burning cones produce
small patches with relatively high soil heating, which we hypothesize is a major mechanism that drives
exceptional fine-scale plant diversity observed in longleaf pine understories (>100 species m-2). Previous
studies show that distance to the nearest tree or tree clump is a large determining factor for understory
plant species composition and diversity. Therefore, our objective is to upscale fine-scale (3 m2) plot
measures of understory plant diversity and surface fuel structure to the stand and landscape levels, using
tree density as the forest attribute constraining understory plant diversity and surface fuel structure. We
predict tree density from low-density airborne lidar-derived canopy height and density metrics, using as
our response variable tree density measures from 201 environmental monitoring field plots (1-ha)
sampling the different forest conditions across Eglin AFB. We test the Random Forests machine learning
algorithm and some competing modeling approaches, but in all cases assume tree distributions are
random. Accuracies are assessed within five management units where high-density lidar and field data
are available and where our assumption of randomly distributed trees is tested.
110
S22 Monitoring Surface Water and Wetland Ecosystems with SAR – II
/ S22 Suivi des milieux humides et des surfaces d’eau avec le RSO – II
Brian Brisco, chair / modérateur
A Preliminary Study on the Synergistic Use of SAR and High Resolution Optical
Data for Wetland Classification in Newfoundland and Labrador
Bahram Salehi 1*, Brian Brisco 2, and Sahel Mahdavi3
1. Remote Sensing Engineer, C-CORE/LOOKNorth and Cross-appointed professor, Faculty of Engineering and
Applied Science, Memorial University of Newfoundland, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL,
Canada, (709) 864 6701, [email protected]
2. Senior Research Scientist, Earth Science Sector, Canada Centre for Mapping and Earth Observation, Ottawa,
Canada, [email protected]
3. M. Eng. Student, Memorial University of Newfoundland, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's,
NL, Canada [email protected]
* Presenting Author: Bahram Salehi, Remote Sensing Engineer, (709) 864 6701, [email protected]
ABSTRACT
Although wetlands contribute significant values to society, wetlands in boreal systems remain poorly
understood. The need for wetland inventory in Newfoundland and Labrador has been identified for over
30 years. However, despite most of the other Canadian provinces which have a wetland inventory, a
comprehensive wetland inventory identifying priority wetland area and types has not been developed for
Newfoundland and Labrador (NL). Recent advancement in both radar and optical sensor technologies
combined with new development in object-based image analysis techniques provide a unique challenge
and opportunity for wetland mapping and monitoring in the harsh environment of NL.
Because of spectral and structural similarity of land cover classes within a wetland, neither spectral
information of optical imagery nor backscatter information of SAR data alone provides sufficient
sensitivity for the separation of representative wetland land cover classes. One promising way to extend
the classification observation space is to introduce additional features inherent in the optical and SAR data
in the form of object-based image analysis. In addition to the spectral and polarimetric information of
individual pixels, this approach would also utilize spectral, polarimetric, textural, contextual, and
morphological information of objects to increase the separability of similar classes in a wetland
environment.
This paper presents a methodology for wetland classification using the combination of SAR data and high
resolution multi-temporal optical imagery and aerial ortho-photos through an object-based image
analysis framework. The method will be tested on different pilot sites across NL and it will be used later
to the creation of a province-wide wetland inventory and wetland monitoring system.
111
Water Level Changes Detection in Yellow River Delta Based on Distributed
Scatterer Interferometry
Yun Shao1*,Qianjun Zhao2, Chou Xie3, YunjunZhang4
1. Professor, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101, China,
0086-10-64878043, [email protected]
2. Professor, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101, China,
0086-10-82178013, [email protected]
3. Associate Professor, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101,
China, 0086-10-64878043, [email protected]
4. Department of Marine Geosciences, Rosenstiel School of Marine and Atmospheric School, University of Miami,
Miami, FL 33149-1098, United States,
* Presenting Author: Yun Shao, Professor, 0086-10-64878043, [email protected]
ABSTRACT
Yellow River Delta provides a productive ecosystem and favorable habitat for a wide variety of plant and
animal species, and is among the most productive ecological systems in the world. However, population
growth and increasing economic development has resulted in extremely rapid degradation and the loss
of wetlands. Many wetland-dependent species in Yellow River Delta are in decline, and the status of
coastal wetland species, especially migratory birds, is being deteriorated gradually.
A key approach for ensuring the future of wetlands and their services is to maintain the quantity and
quality of the natural water regimes on which they depend, including the frequency and timing of flows.
Stage stations can provide high temporal observations of hydrologic systems. However, the sparsely
distributed stage stations cannot supply us with high spatial resolution measurements. Space-borne
differential Interferometry SAR has proven a remarkable potential in wetland applications, including
water level monitoring with high spatial resolution, which cannot be obtained by any terrestrial-based
method.
Long term monitoring is essential to evaluate the health performance of a wetland ecosystem. A new
approach called Distributed Scatter Interferometry (DSI) has been proposed to retrieve displacement
time-series from natural targets. In this study, we adopt the DSI method to obtain long term hydrological
information. After identifying statistically homogeneous pixels (SHP) from natural target in wetlands, an
adaptive spatial phase filtering is implemented to improve the fringe quality and preserve the spatial
resolution of interferograms. An optimum target-dependent subset of interferograms is exploited to
maximize the extracted hydrological information and detect temporal and spatial evolution of water level
from InSAR data stack. The new method was applied on ALOS PALSAR data set from 2007 to 2010 in the
Yellow River Delta. The results demonstrate its effectiveness to provide temporal evolution of water level
changes in coastal wetlands.
112
Temporal Coherence for Wetland Mapping and Monitoring
Kevin Murnaghan 1*, Frank Ahern 2, Shimon Wdowinski 3, Brian Brisco 4
1. Remote Sensing Scientist, Natural Resources Canada, 560 Rochester St., Ottawa, Ontario, Canada,
1(613)759-6237, [email protected]
2. TerreVista Earth Imaging, 441 Cormac Road, Cormac, Ontario, Canada, 1(613)754-2822,
[email protected]
3. University of Miami, 4600 Rickenbacker Cswy., Miami, Florida, U.S.A., 1(305)421-4730,
[email protected]
4. Research Scientist, Natural Resources Canada, 560 Rochester St., Ottawa, Ontario, Canada, 1(613)759-6344,
[email protected]
* Presenting Author: Kevin Murnaghan, Remote Sensing Scientist, (613)759-6237,
[email protected].
ABSTRACT
Wetlands have several functions including water storage and retention which can reduce flooding and
provide continuous flow for hydroelectric generation and irrigation for agriculture. Synthetic Aperture
Radar is well suited as a tool for monitoring surface water by supplying acquisitions irrespective of cloud
cover or time of day. The wetland can be subdivided into three classes: open water, flooded vegetation
and upland. These classes vary seasonally with time and water level changes. Using repeat pass
acquisitions InSAR has been demonstrated to estimate water level of flooded vegetation. Radarsat-2 data
from the Spotlight mode were processed from 2010-2013 to provide coherence estimates. The coherence
of the wetlands is affected by vegetation type, vegetation health, and water level. This allows for the use
of coherence change detection for wetland monitoring. Examples from Lake Clear Ontario, PeaceAthabasca Delta Alberta, and Everglades National Park Florida will be shown. Guidelines for which
wetlands are suitable for monitoring using InSAR and the use of coherence as a mapping tool for large
area monitoring and mapping projects will be discussed. The findings could be used operational for
assessing wetlands using Radarsat Constellation Mission data using the rapid revisit capabilities.
113
Integration of Multi-Frequency Data into the Periodic Remote Delineation of
Wetlands
Andreas Schmitt1*, Brian Brisco2
1. Postdoc, German Aerospace Center (DLR), Oberpfaffenhofen, D-82234 Weßling, +49-8153-28-3341,
[email protected]
2. Brian Brisco, Senior Research Scientist, Canada Centre for Mapping and Earth Observation, Ottawa, Canada
(formerly CCRS)
* Presenting Author Andreas Schmitt, Postdoc, +49-8153-28-3341, [email protected]
ABSTRACT
Synthetic Aperture Radar turned out to be a suitable tool for diverse monitoring purposes because images
can be acquired independently of weather or illumination. Mainly in higher latitudes, this is the only
reliable data source available throughout the year. Regarding the increasing number of SAR satellite
sensors – each sensor with its own characteristic advantages – it is reasonable to integrate all available
image acquisitions in one uniform image processing and evaluation environment. Such a mathematical
framework was developed recently for the integration of multi-mode SAR data, i.e. multi-frequency (and
therewith multi-sensor), multi-temporal, multi-polarized, and multi-scale.
This technique bases on the description of SAR image data by the help of the Kennaugh elements. These
layers are orthorectified, further calibrated, and finally normalized in order to restrict all measured
variables to the uniform number range regardless of the data source. Thus, images of all current SAR
sensors – be it TerraSAR-X, RADARSAT-2, or ALOS-PALSAR – are processed and evaluated together in the
same geometric and radiometric frame. Multi-frequency data consequently can be used to fill temporal
gaps in-between the acquisitions of one special sensor on the one hand, but it can also be used to increase
the information content by contributing multi-spectral layers on the other hand.
The SAR image pre-processing approach additionally includes a sophisticated automated image
enhancement step: the multi-scale and multi-directional multi-looking. Instead of applying a uniform look
number to the whole image, the look number and therewith the degree of smoothing is adapted locally
to the image content both in scale and direction. This primarily results in a smooth, but detail-preserved
intensity image. Furthermore, it provides a very valuable texture information layer indicating the
appropriate shape, scale, and orientation of a set of pre-defined filter masks, known as “Schmittlets”. This
even allows the evaluation of local image patterns.
The final step of the processing chain is a very simple, but extremely effective classification technique
based on the correlation of local histograms with sample histograms of predefined classes. Thanks to the
closed and uniform number range of the normalized Kennaugh elements, the histograms can easily be
compared. The occurrences of the single Schmittlets are aggregated to sets of relative angles (e.g. parallel,
diagonal, rectangular) in order to describe local patterns regardless of their absolute orientation in the
image. Hence, radiometric information (via the Kennaugh elements) and geometric information (via the
Schmittlet indices) is utilized in conjunction.
In summary, this contribution presents an up-to-date approach to the automated monitoring of land cover
and land cover changes on a regional scale (at least) applied on TerraSAR-X and RADARSAT-2 acquisitions
of wetlands in the Peace River – Athabasca Lake – Delta in Alberta, Canada.
114
Soil Moisture Monitoring in Mountain Areas Using Different Polarization of
Radarsat-2 Data
Tingting Zhang1,Yun Shao1*, Qianjun Zhao1, Xun Chai1, and Kaixin Xie1
1. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, China
* Presenting Author: Yun Shao, Beijing, China, +86-10-6483-8047,
ABSTRACT
Soil moisture (SM) is a key parameter in the climate studies and a very important parameter in applications
such as plateau pasture. C-band RADARSAT-2 SAR (Synthetic Aperture Radar) imagery has been identified
the potential for soil moisture monitoring. Three experimental campaigns were conducted in the Qinghai
Lake watershed, northeastern Tibetan Plateau in September 2012, May 2013 and May 2014 respectively,
with simultaneously satellite overpass. Water-cloud model and Dubois model were used to eliminate the
effect of vegetation and soil surface roughness. The results demonstrated that backscattering coefficients
of HH and VV polarization can be used to build relationship with soil moisture. However, one drawback
of the developed models is the monitoring coverage is limited. HH and VV polarizations are available only
with the quad polarization data. A model with HH and HV polarizations were developed to predict soil
moisture. Backscattering coefficients of HH and HV polarization were used to build relationship with soil
moisture. The model got R2 with 0.78, 0.78 and 0.79 for 2012, 2013 and 2014 respectively. The results
indicated that the potential of new methodology to estimate soil moisture in mountain region with the
use of HH and HV polarization. We will try this method on RADARSAT-2 ScanSAR data to acquire large
scale soil moisture distribution for plateau pasture.
115
S23 Operational Maritime Surveillance with SAR / S23 Surveillance
maritime opérationnelle avec le RSO
Gordon Staples, chair / modérateur
Maritime NRT Products using TerraSAR-X, Sentinel, and Radarsat data
Anja Frost1* and Susanne Lehner
1. German Aerospace Center (DLR), EOC Bremen Airport, Henrich Focke Str. 4, Bremen, 28199, Germany,
+49421244201850, [email protected]
2. Head of Research Group, German Aerospace Center (DLR), EOC Bremen Airport, Henrich Focke Str. 4, Bremen,
28199, Germany, +49421244201850, [email protected]
* Presenting Author: Anja Frost, [email protected]
ABSTRACT
High resolution Synthetic Aperture Radar (SAR) data from TerraSAR-X/Tandem-X satellites are used to
determine and monitor the sea surface in near real time (NRT) and all weather and illumination conditions.
The radar backscatter of the sea surface is determined by the sea surface roughness caused by the wind
field and the sea state. These meteo-marine parameters are modelled by the newly developed algorithms
XMOD and XWAVE relating the wind field and sea state, depending on incidence angle and directionality
to the radar backscatter sigma0.
The TerraSAR-X Modes Stripmap, Scan SAR and Scan SAR Wide are used together with Sentinel and
RADARSAT data to detect ships, oil spills and icebergs. The detectability depending on the background
conditions is discussed. Several examples from near real time campaigns performed together with users
are given.
116
Oil Slick Discrimination using RADARSAT-2 Quad Polarized Data
Gordon Staples1* and Suzanne Brunke1
1. MDA, 13800 Commerce Parkway, Richmond, BC, Canada [email protected]
* Presenting Author: Gordon Staples, MDA, [email protected]
ABSTRACT
RADARSAT-2 quad polarized data was acquired over a controlled oil-on-water exercise in 2011 through
2013. During the exercise, three types of oil were released: plant oil, emulsion, and crude oil. Of the
fourteen images that we acquired, five were considered suitable for analysis. The Cloude-Pottier entropy
and Touzi scattering phase were calculated from the 3x3 coherency matrix after the application of 5x5
Lee polarimetric filter. For all the data, the co-polarized backscatter was nominally greater than 5 dB above
the noise floor, but decreased with increasing incidence angle. In contrast, the cross-polarized backscatter
was largely invariant with incidence and was at or very near the noise floor.
The entropy indicated discrimination between the three oil types and the ocean surface. The crude oil
entropy had the largest value, followed by emulsion, plant oil, and the ocean surface. Note that for all
data analyzed, the wind speeds were less than 6 m/s. The increase of the entropy with incidence angle
was attributed to the influence of the radar noise floor.
In contrast to the entropy, the Touzi scattering phase indicated discrimination between crude and
emulsion and plant oil plus the ocean surface, but there was no discrimination between the plant oil and
the ocean surface. The discrimination based on the standard deviation was attributed to the relationship
between the Touzi scattering phase and the co-polarized phase difference, which are essentially the same
when the target helicity approaches zero. Therefore, the Touzi S provides discrimination when there
was a difference in the scattering mechanism (i.e. ocean versus crude), but limited discrimination
otherwise. In contrast, the standard deviation of the Touzi S provided discrimination between the ocean,
plant oil, emulsion, and crude oil.
117
Detecting and Tracking Dark Ships from Multiple Satellite Missions
Jennifer Busler1*, Andrew Westwell-Roper2 and Hans Wehn3
1. Senior Analyst, MDA Systems Ltd, 13800 Commerce Pkwy, Richmond, BC, Canada, 604.278.3411,
[email protected]
2. Senior Analyst, MDA Systems Ltd, 13800 Commerce Pkwy, Richmond, BC, Canada, 604.278.3411,
[email protected]
3. Senior Analyst, MDA Systems Ltd, 13800 Commerce Pkwy, Richmond, BC, Canada, 604.278.3411,
[email protected]
* Presenting Author: Jennifer Busler, Senior Analyst, 604.278.3411, [email protected].
ABSTRACT
Maritime safety and security organizations need to detect and track ships that pose a safety or security
threat to themselves or others. Currently, the primary method for assessing these threats is through
analysis of self-reported data from services such as the Automatic Identification System (AIS) and
mandatory voyage reports. Ships that do not want to be tracked, or, as is the case for smaller vessels, are
not required by regulation to do so, do not self-report. Such vessels are referred to as “dark ships”. Earth
observation satellites can be used to detect dark ships, but it is difficult to use a single satellite source to
track a dark ship and reliably assess whether it poses a threat. Obtaining more densely sampled ship
position reports from images acquired by multiple satellite missions is a potential solution.
With the support of a grant from the Canadian Space Agency, MDA Systems Ltd is currently investigating
the feasibility of operationally detecting and tracking dark ships using radar and optical satellites from
multiple missions. Our study explores this multi-mission scenario in two stages:
1. Determine the optimum spatio-temporal sampling of vessel detections for tracking purposes
obtainable from a multiple-mission satellite constellation under finite but realistic data acquisition
scenarios and conditions.
2. Determine the performance of potential dark ship tracking algorithms, given the results of the
first stage.
This paper will present the results of our investigation, including the impact on our results of the
forthcoming RADARSAT Constellation Mission.
118
Evaluation of RADARSAT Constellation Mission Compact Polarimetry for ship
detection
Chen Liu1* and Paris W. Vachon1
1. Defence Research and Development Canada – Ottawa Research Centre 3701 Carling Avenue, Ottawa, Ontario,
Canada, K1A 0Z4, [email protected] and [email protected]
* Presenting Author: Chen Liu, [email protected]
ABSTRACT
The RADARSAT Constellation Mission (RCM) will provide compact polarimetry (CP) capability for almost
all beam modes. A CP synthetic aperture radar (SAR) system with circular polarization on transmission
and two orthogonal linear polarizations on receive offers larger swath coverage than a fully polarimetric
(quad-pol) system, while maintaining the ability of observing some of the target backscattering behaviour
of the quad-pol system.
For maritime surveillance, wide swath coverage is a key consideration. The CP SAR system provides tradeoffs among swath width, resolution and polarimetry. For the RCM Ship Detection Mode with CP, the
nominal swath coverage is 350 km, which is similar to the swaths offered by the RADARSAT-2 ScanSAR
modes.
This paper investigates the benefit of CP for ship detection and ship / iceberg discrimination by using
simulated RCM CP products derived from RADARSAT-2 quad-pol data.
The preliminary results clearly show the advantage of a CP system for ship detection over linear dualpolarization and single-polarization systems. In particular, the improved detection is superior for ships in
higher sea states.
The results also show that the dual-polarization target decomposition methods have potential to aid in
discrimination between ships and icebergs due to different target signatures from their unique structures,
sizes and shapes.
RCM is scheduled to launch in 2018. When the RCM CP system becomes operational, it is expected to be
beneficial to maritime surveillance applications including ship detection and ship / iceberg discrimination.
119
Maritime Operational Use of SAR by Environment Canada’s Canadian Ice Service
Leah Braithwaite1*, Matthew Arkett1, Marilee Pregitzer1, Yi Luo 1, Gaetan Langlois1
1. Environment Canada-Canadian Ice Service, 373 Sussex Drive, Ottawa, ON, Canada
* Presenting Author: Leah Braithwaite, Chief, Applied Science-Remote Sensing , (613) 996-4489,
[email protected]
ABSTRACT
For a significant portion of the navigation season, Canadian waters (Arctic, Hudson Bay, Great Lakes and
the St. Lawrence Seaway and Gulf), remain ice covered. This presents a challenge to efficient marine
transportation and a hazard to safety and sustainable development and an important factor in maritime
domain awareness. The mission of the Canadian Ice Service (CIS) is to provide, as much as possible, the
most timely and accurate information about ice in Canada’s navigable waters. Since the launch of
RADARSAT-1, Synthetic Aperture RADAR (SAR) data has been the primary information source for the CIS
provision of ice information products and services. The RADARSAT sensors had largely been optimized
for ice detection and CIS has maintains outstanding expertise in the interpretation of this data for the
preparation of ice charts, bulletins and warnings. Additionally, CIS uses SAR data for the detection and
monitoring of illegal marine oil releases in a partnership program with Transport Canada and the Canadian
Coast Guard.
CIS Remote Sensing and Forecast Operations has evaluated and has begun exploiting SAR data from other
sources including TerraSAR-X and Sentinel-1. Additionally, CIS has investigated new modes available from
RADARSAT-2 in partnership with Defense Research and Development Canada and, in preparation for
RADARSAT Constellation Mission, is assessing automated approaches to identifying lake and sea ice as
well as icebergs. This presentation will provide an overview of current CIS operational use of SAR data as
well as potential future applications.
120
S24 Arctic Ecosystem Monitoring / S24 Suivi des écosystèmes
arctiques
Brigitte Leblon, chair / modérateur
Landsat-based Mapping of Thermokarst Lake Dynamics in Continuous
Permafrost of Western Canada since 1985
Ian Olthof1*, Robert Fraser2 and Carla Schmitt3
1. Physical Scientist, Canada Centre for Mapping and Earth Observation, 560 Rochester, Ottawa, ON, Canada, 613759-7629, [email protected]
2. Research Scientist, Canada Centre for Mapping and Earth Observation, 560 Rochester, Ottawa, ON, Canada, 613694-2621, [email protected]
3. Physical Scientist, Canada Centre for Mapping and Earth Observation, 560 Rochester, Ottawa, ON, Canada, 613759-7216, [email protected]
* Presenting Author: Ian Olthof, Physical Scientist, 613-759-7629, [email protected].
ABSTRACT
Several remote sensing studies have documented widespread thermokarst lake expansion in continuous
permafrost regions of North America over the past few decades. Other studies have found no long-term
trends in water body extents, but large intra- and inter-annual changes driven by precipitation. These
differences could be due to geographic variability in physical conditions (geology, climate, permafrost,
hydrology) or in the data and methods used to extract water bodies. This study tested water extraction
methods over the Tuktoyaktuk Coastal Plain, Northwest Territories (NWT), Canada, based on the Landsat
shortwave infrared (SWIR) channel and validated them using water extents obtained from 0.5 m
resolution orthophoto imagery. Methods included applying optimal hard thresholds and deriving 30 m
water fractions from both linear unmixing and a new histogram breakpoint method. Results indicated that
the histogram breakpoint method outperforms other methods when evaluated against both the number
of 30 m pure water pixels and water fractions obtained from orthophotos. The breakpoint method was
then applied to a stack of 17 near peak-of-season Landsat images from 1985 to 2011 to create a water
fraction time-series for examining both trends and inter-annual variation in water extent. Unlike previous
studies that used a limited number of temporal observations, our results showed an overall expansion of
lake area along margins with isolated lakes experiencing rapid drainage likely caused by permafrost melt
and lateral breaching. Inter-annual variability and long-term trends are related to climate and geology to
explore drivers and mechanisms of thermokarst lake change.
121
Influence of the Environmental Conditions on Surficial Mapping Using LANDSAT5 TM and RADARSAT-2 Polarimetric SAR Images
Justin Byatt 1*, Armand LaRocque 1, Brigitte Leblon1, Kara Webster2, Jim McLaughlin3
(1) Faculty of Forestry and Environmental Management, University of New Brunswick, Fredericton, NB, E3B5A3,
Canada; Phone: 506-453-4924; Fax: 506-453-3538, Email: [email protected], [email protected], [email protected]
(2) Research scientist, Canadian Forest Service, Great Lake Forestry Centre, Sault-Ste-Marie (ON), Canada,
705-541-5520, [email protected]
(3) Research Scientist, Ontario Forest Research Institute, Ontario Ministry of Natural Resources,
Sault-Ste-Marie (ON), Canada, 705-946-7418, [email protected]
* Presenting Author: Justin Byatt, Student, Phone: 506-447-3430, [email protected]
ABSTRACT
The study is part of a research program on the use of RADARSAT-2 polarimetric SAR and LANDSAT-5 TM
images for mapping surficial material and land cover in a region of northern Ontario. Surficial material
maps have already been produced in previous studies by using LANDSAT and single-, dual-, and multipolairzed, or polarimetric SAR images. In this study, we further assess the influence of wetness conditions
at the time of image acquisition on mapping accuracy using a RADARSAT-2 WFQ12 descending
polarimetric SAR image and a LANDSAT-5 TM image. The Random Forests (RF) classifier was applied to
the following input features: DEM, slope, six LANDSAT-7 ETM+ optical bands, intensity images in the four
RADARSAT-2 polarizations (HH, HV, VH, VV), and several polarimetric products, such as total power, four
circular backscatters, the Freeman-Durden and Cloude-Pottier decomposition parameters, and 18
polarimetric discriminators. The best classification was obtained using both wet and dry images with
LANDSAT-7 ETM+ optical bands, RADARSAT-2 intensity images, slope, and DEM data. Further research is
needed to test image combinations that use RADARSAT-2 images acquired at the different beam modes
and orbits. This study was funded by the Canadian Space Agency Class Grant and Contribution Program
and a NSERC Post-graduate Scholarship. The Canadian Space Agency provided the RADARSAT-2 data to
the project via the Canadian Forest Service. The Canadian Forest Service and Ontario Ministry of Natural
Resources provide field observations. Support to the field work has been provided by Ontario Ministry of
Natural Resources and De Beers Canada.
122
River Surface Temperature Retrieval Using Optical Remote Sensing Images
Bing Yue1*, Joseph Chamberland2*, John Mulvie3
C-CORE, 400 March Road, Ottawa, ON, Canada K2K 3H4
1. 613-592-7700 ext 226, [email protected]
2. 613-592-7700 ext 232, [email protected]
3. 613-592-7700 ext 231, [email protected]
* Presenting Author: Joseph Chamberland, 613-592-7700 ext 232, [email protected]
ABSTRACT
Over the last several decades, resource development in the upstream portion of the Slave River watershed
has increased in various fields such as oil and gas development, paper mills, coal and uranium mining,
agriculture and forestry. Northerners have raised concerns that these activities have impacted the water
resources in the Northwest Territories [1]. Since water temperature can affect many processes of the
river's ecosystems such as evaporation, oxygen dissolving and aquatic life, the ability to monitor water
temperature change can provide knowledge and understanding on pending water quality challenges. In
this presentation, we will present our study on water surface temperature reconstruction using LANDSAT
thermal image and meteorological data which covers different atmospheric conditions from early spring
to late fall.
Water surface temperatures were obtained from the digital number (DN) of LANDSAT thermal band based
on radiative transfer equation and Plank's law inversion. To remove the atmosphere effect in DN,
atmospheric correction was necessary and applied to the data. In this work, two different atmospheric
correction approaches converting at-sensor radiance to surface-leaving radiance have been investigated.
By comparing the retrieved temperature results to the measured water surface temperatures on the study
site, both methods were seen to be able to remove the atmospheric effects efficiently and both can
reconstruct the water surface temperature correctly.
To address the water quality monitoring requirements, our work also demonstrates the feasibility of an
operational remote sensing tool for water surface temperature mapping, especially when higher spatial
and temporal resolution thermal band data available.
REFERENCE:
[1] Report Summary: Slave River Water and Suspended Sediment Quality in the Transboundary Reach of
the Slave River, Northwest Territories, Minister of Aboriginal Affairs and Northern Development, 2012
123
Mapping discontinuous permafrost at high spatial resolution using LANDSAT
and RADARSAT-2 dual polarized images in northern Ontario, Canada
Brigitte Leblon1*, Armand LaRocque2*, Chunping Ou3, Yu Zhang4, Kara Webster5, and Jim McLaughlin6
1. Professor, Faculty of Forestry and Environmental management, University of New Brunswick, Fredericton (NB),
Canada, 506-453-4924, [email protected]
2. Research associate, Faculty of Forestry and Environmental management, University of New Brunswick,
Fredericton (NB), Canada, 506-453-4932, [email protected]
3. Post-doctoral fellow, Faculty of Forestry and Environmental management, University of New Brunswick,
Fredericton (NB), Canada, 506-447-3430, [email protected]
4. Research scientist, Canadian Centre for Mapping and Earth Observation, Ottawa (ON), Canada,
613-947-1367, [email protected]
5. Research scientist, Canadian Forest Service, Great Lake Forestry Centre, Sault-Ste-Marie (ON), Canada,
705-541-5520, [email protected]
6. Research Scientist, Ontario Forest Research Institute, Ontario Ministry of Natural Resources,
Sault-Ste-Marie (ON), Canada, 705-946-7418, [email protected]
* Presenting Author: Armand LaRocque, Research Associate, 506-453-4932, [email protected]
ABSTRACT
Permafrost is an important feature that has significant biophysical and socioeconomic impacts on the
northern landscape. In order to better understand the distribution and dynamics of permafrost, there is
a need to map permafrost at high spatial resolution. This study is part of a research project that aims to
map permafrost using RADARSAT-2 and LANDSAT images and the Northern Ecosystem Soil Temperature
(NEST) model over a discontinuous permafrost region, i.e., Victor Diamond Mine area located in the
Hudson Bay Lowland (northern Ontario, Canada). The NEST model was run from 1932 to 2012 using a grid
climate datasets. The model outputs were then compared to observations acquired during 2009-2012 at
seven peat monitoring stations and two flux towers, located on three major peatland types of the study
area (bog, fen, and palsa). The simulated soil temperatures and fluxes show good agreement with the
observations, and the simulated ones. The model shows the existence of permafrost only at palsa sites,
which is in agreement with field observations. Sensitivity tests indicate that the modelled permafrost
variables are sensitive to air temperature, precipitation, and leaf area index. Permafrost maps were
derived from land cover and surficial material maps that were derived from RADARSAT-2 and LANDSAT
images. This study was funded by the Canadian Space Agency and a NSERC/NBIF Post-graduate
Scholarship. The CSA provided the RADARSAT-2 data to the Canadian Forest Service. CFS and OMNR
provide field observations. Support to the field work has been provided by OMNR and De Beers Canada.
124
A Multiscale Study of Tundra Vegetation Changes in the Tuktoyaktuk Coastal
Plain, NWT
Robert H. Fraser1, Trevor C. Lantz2, Nina Moffat3, and Ian Olthof4*
1. Research Scientist, Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester St., Ottawa,
ON, 613-694-2621, [email protected]
2. Assistant Professor, School of Environmental Studies, University of Victoria, Victoria, BC, 250-853-3566,
[email protected]
3. Graduate Student, School of Environmental Studies, University of Victoria, Victoria, BC, 250-853-3566,
[email protected]
4. Remote Sensing Scientist, Canada Centre for Remote Sensing, Natural Resources Canada, 560 Rochester St.,
Ottawa, ON, 613-694-2621, [email protected]
* Presenting Author: Ian Olthof, Remote Sensing Scientist, 613-694-2621, [email protected]
ABSTRACT
Recent field and remote sensing studies show that shrub expansion has been widespread in low-Arctic
ecosystems. However, there are still uncertainties regarding the extent of these changes, the plant
functional groups involved, and the relative importance of climate and disturbance as causes of observed
changes. One hotspot of Arctic greening identified in studies using coarse-resolution AVHRR NDVI is the
Tuktoyaktuk Coastal Plain in Northwest Territories. We investigated the detailed nature of and specific
causes for these changes using measurements collected over a range of spatial scales. These included
field plots, UAV multicopter photos, repeat ground and large-scale air photos spanning a > 30 year period,
and a dense time series (1985-2011) of 30 m resolution Landsat TM and ETM+ satellite imagery. Analysis
of data from all scales documented a consistent pattern of widespread upright and dwarf shrub expansion
and declines in lichen cover. Shrub expansion was observed in most locations, but to a lesser extent in
waterlogged soils and at sites with high shrub crown closure. The best explanation for the rapid and
widespread changes in vegetation is recent regional increases in air and ground temperatures, leading to
enhanced nutrient mineralization. These vegetation changes are likely to impact the availability of
regional lichen forage for caribou and provide additional fuels to promote increased fire activity.
125
S25 Hyperspectral Imaging – I / S25 Imagerie hyperspectrale – I
George Leblanc, chair / modérateur
Validation of a Two-Step Model Inversion Approach for Regional Retrieval of
Leaf Chlorophyll Content Using Remote Sensing Data
Jing M. Chen1*, Holly Croft2, Joyce Arabian3, Nadine Nesbitt4, Yuhong He5, Jiali Shang6, Yongqin Zhang7,
Anita Simic8, Tom Noland9, Ted Hoffman11, and Jiangui Liu12
1. Jing M. Chen, Professor, Department of Geography, University of Toronto, Toronto, Ontario, Canada, 416-9787085, [email protected]
2. Holly Croft, Postdoctoral Fellow, Department of Geography, University of Toronto, Toronto, Ontario, Canada,
416-553-8431, [email protected]
3. Joyce Arabian, M.Sc. student, Department of Geography, University of Toronto, Toronto, Ontario, Canada, 416859-4815, [email protected]
4. Nadine Nesbitt, M.Sc. student, Department of Geography, University of Toronto, Toronto, Ontario, Canada, 905669-1223, [email protected]
5. Yuhong He, Associate Professor, Department of Geography, University of Toronto, Toronto, Ontario, Canada,
905-569-4679, [email protected]
6. Jiali Shang, Research Scientist, Science and Technology Branch, Agriculture and Food Canada, 613-759-1435,
[email protected]
7. Yongqin Zhang, Assistant Professor, Division of Biological and Physical Sciences, Delta State University, USA, 662846-425, [email protected]
8. Anita Simic, Assistant Professor, College of Arts and Scence, Bowling Green State University, USA, 419-372-4035,
[email protected]
9. Tom Noland, Research Scientist, Ontario Ministry of Natural Resources and Forestry, Canada, 705-946-7421,
[email protected]
10. Ted Hoffman, Research Scientist, Science and Technology Branch, Agriculture and Food Canada, 613-759-1846,
[email protected]
11. Jiangui Liu, Remote Sensing Analyst, Science and Technology Branch, Agriculture and Food Canada, 613-7591452, [email protected]
* Presenting Author: Jing M. Chen, Professor, 416-978-7085, [email protected].
ABSTRACT
Chlorophyll content of plant leaves is an indicator of plant health and is responsible for light harvesting
during photosynthesis. For the purpose of regional and global mapping of leaf chlorophyll content using
remote sensing data, we previously developed a two-step model inversion approach: (1) inversion from
canopy to leaf reflectance, and (2) inversion from leaf reflectance to chlorophyll content. Compared with
empirical approaches based on spectral indices, our model-based approach has the advantage of the
ability to incorporate additional information such as vegetation structure, sun-target-view geometry,
background optical properties, etc. for chlorophyll estimation. This model-based approach was previously
applied to images of the Compact Airborne Spectral Imager (CASI) acquired over boreal and temperate
forests in Canada as a scaling step for evaluating its application to satellite sensors including Landsat,
CHRIS and MERIS. Ground-based chlorophyll and canopy structural data were acquired from 21 broadleaf
and conifer forest stands to validate the retrievals (R2 = 0.63, p<0.001 for CHRIS; R2 = 0.62, p<0.001 for
MERIS; and R2 = 0.53, p<0.001 for Landsat 5 TM). During the growing season of 2013, we acquired leaf
chlorophyll data from wheat and corn fields near Stratford, Ontario, and canopy-level hyperspectral data
126
were also acquired using a hand-held spectroradiometer to validate the model inversion results. An
experimental product of leaf chlorophyll has been produced over Canada’s landmass using MERIS data in
2011. Concurrent LAI data from MODIS based on the GLOBCARBON algorithm and forest background
reflectance retrieved from MISR are used as inputs in the inversion. Although we only have limited ground
chlorophyll data in 2012 for map validation, we have evaluated our model inversion approach against leaf
chlorophyll data acquired at the aforementioned ground sites as well as at sites in the Grasslands National
Park in Saskatchewan, in forests on the Vancouver Island, and wheat and corn crops near Ottawa. We
welcome researchers with chlorophyll data in various locations and ecosystems to participate in this
research to refine the algorithm and validate the product.
127
Improving hyperspectral imagery automatic segmentation by differential
analysis
Kongwen (Frank) Zhang1*, Linhai Jing2, Justin Robinson3
1. Lecturer, Selkirk College, 301 Frank Beinder Way, Castlegar, BC, Canada, 250-304-6527, [email protected]
2. Key Laboratory of Digital Earth Science, Institute of Remote Sensing and Digital Earth, Chinese Academy of
Science, Beijing, China, 010-8217-8106, [email protected]
3. Research assistant, Selkirk Geospatial Research Center, 301 Frank Beinder Way, Castlegar, BC, Canada, 250-3046527, [email protected]
* Presenting Author: Frank Zhang, Lecturer, 250-304-6527, [email protected].
ABSTRACT
The automatic segmentation of hyperspectral imagery requires a reduction in the dimension of the data,
such as classification and principle component transformation. As a result, different reduction
methodologies derive varying information for segmentation, leading to distinct outcomes. In this study,
we explore the impact of multiple reduction methods and compare the differences in results. By
evaluating alternative outcomes, we propose a new differential analysis that is more effective in selecting
dimension-reducing methods, utilizing hyperspectral imagery segmentation. We use two sets of sample
hyperspectral data from SpecTIR, Beltsville, MD, USA and Reno, NV, USA, for vegetation and urban cases,
respectively. First, four different categories of processes are evaluated: unsupervised classification,
principle component transformation, vegetation index, and contribution index weighting. Then, all
outputs are stacked together to generate an accuracy matrix. Pixels with three or more concurring process
categories are kept as is, and form new segments. For every non-decided pixel, a comparison to the
neighbor segments’ average spectrum is performed using a maximum likelihood estimation. Finally, for
any remaining pixels, a prior-knowledge based decision system is used, e.g., if the surroundings are
vegetation, then the vegetation index result would be adopted. Using this approach, it is very easy to
detect any minor differences in spectral and spatial correlations, and directly provides an accuracy matrix
when generating the final results. This methodology also eliminates unnecessary iteration and allows the
minimization of computation cost by processing all spectral bands.
References
Gorretta, N., Roger, J. M., Rabatel, G., Fiorio, C., Lelong, C. 2009, “Hyperspectral image segmentation: the
buttery approach”, Proceedings of Hyperspectral Image and Signal Processing: Evolution in Remote
Sensing (WHISPERS '09).
Jia, X. and Richards, J. A., 1999, “Segmented principal components transformation for efficient
hyperspectral remote sensing image display and classification”, IEEE transactions of Geosicence and
Remote Sensing, 37:1, 538-542.
Kruse, F. A., 2005, “Multi-resolution segmentation for improved hyperspectral mapping”, Proceedings of
SPIE Symposium of Defense and Security, Orland, FL, USA.
Li, J., Bioucas-Dias, J. M., Plaza, A., 2011, “Hyperspectral image segmentation using a new Bayesian
approach with active learning”, IEEE transactions on Geoscience and Remote Sensing, 49:10 3947-3960.
Ifanrraguerri, A., and Chang, C., 2000, “Unsupervised hyperspectral image analysis with project pursuit”,
IEEE Transactions on Geoscience and Remote Sensing, 38:6, 2529 - 2538.
128
Satellite-derived bathymetry, a Canadian case study
Christopher Ilori1, Anders Knudby2*
1. PhD student, Department of Geography, Simon Fraser University, 8888 University Drive, Burnaby, BC, Canada,
+1-778-782-4567, [email protected]
2. Assistant Professor, Department of Geography, Simon Fraser University, 8888 University Drive, Burnaby, BC,
Canada, +1-778-782-3876, [email protected]
* Presenting Author: Anders knudby, Assistant Professor, 778-782-3876, [email protected]
ABSTRACT
Water depth can routinely be derived from multispectral and hyperspectral satellite data in clear tropical
waters, but no case studies from mid-latitudes have been reported in the scientific literature. We report
on preliminary results from a satellite-derived bathymetry case study in Boundary Bay, British Columbia.
In-situ spectral reflectance measurements were used with a realistic range of water optical quality and
water depth values to forward-model sea surface reflectance spectra using an open source radiative
transfer model. Landsat 8 and WorldView-2 data were atmospherically corrected, and a least squares
spectrum matching algorithm was employed to find the best match between observed and modeled sea
surface reflectance spectra, from which water depth was then derived. Based on validation using field
observations tide-corrected to match each satellite image, results show good agreement between
modeled and observed water depths, for both satellite images, in waters shallower than three meters.
Errors increase rapidly beyond three meters depth, and the imagery contains practically no information
related to water depths beyond five meters. The effectiveness of satellite-derived bathymetry depends
on sun elevation, water optical quality, and radiometric, spectral and spatial resolution of the sensor.
None of these parameters were optimized in our case study. While the results provide a good first
estimate of the depths to satellite-derived bathymetry is possible in southern Canadian waters, we thus
expect improved performance to be possible with hyperspectral satellite imagery acquired close to
summer solstice, during low tide and low turbidity.
129
S26 LiDAR with other sensors / S26 LiDAR avec d’autres capteurs
Ahmed Shaker, chair / modérateur
Classification of LiDAR Data for Large Areas Using Kurtosis Change Curve
Ahmed Shaker1*, Nagwa El-Ashmawy12
1. Civil Engineering Department, Ryerson University, (416) 979 4658,[email protected]
2. Survey Research Institute, National Water Research Center, 308 Alahram st., 12111, Giza, Giza, Egypt.
* Presenting Author: Ahmed Shaker, Civil Engineering Department, Ryerson University, 416 979 4658,
[email protected]
ABSTRACT
LiDAR has contributed to various environmental and civil engineering applications. LiDAR data has proved
its ability to contribute to the urban land cover classification. Utilizing the near Infrared signals, LiDAR got
advantage of distinguishing the land cover types. Classification techniques based on the statistical analysis
segmentation approaches have been introduced to deal with the LiDAR data. Separating terrain and nonterrain points, detecting some features such as roads, vegetation and buildings are the main use of these
approaches. Recently, a classification technique which clusters the LiDAR data based on Kurtosis Change
curve approach, and assign the produced segments to the appropriate classes, has been introduced. The
goal of this paper is to introduce several approaches to speed up the classification process of the large
areas using the new introduced classification technique. A modification of the segmentation approach to
avoid the iterative process is introduced. Apply the proposed technique on a large area where a huge
number of points exist is another aim of this paper. The proposed approach is for partitioning the large
area into small areas based on the number of points for faster handling is proposed. The partitioning
method is tested with different sizes.
130
Multispectral LiDAR Data for Land Cover Classification: Initial Results
Salem Morsy1*, and Ahmed Shaker2
1. Ph.D. Student, Department of Civil Engineering, Ryerson University, Toronto, Ontario, Canada,
[email protected]
2. Associate Professor, Department of Civil Engineering, Ryerson University, Toronto, Ontario, Canada,
[email protected]
* Presenting Author: Salem Morsy, 416-979-5000 ext. 4623, [email protected]
ABSTRACT
Multispectral LiDAR sensors are currently emerging in the industry, allowing new applications and
information extraction capabilities for LiDAR. The multispectral LiDAR sensor operates at different
wavelengths that allow a diversity of spectral reflectance collected for different land features.
Multispectral LiDAR data is now available by Optech’s new sensor “Optech Titan”. This sensor offers the
possibility of multispectral active imaging of the environment where it operates at three independent
wavelengths of 532 nm, 1064 nm, and 1550 nm. The Optech Titan is designed for a wide-range of
surveying tasks, such as land cover classification. In this context, an improvement of land cover
classification is presented using multispectral LiDAR data collected by the Optech Titan sensor. A
maximum likelihood classifier is applied to a study area covering Oshawa, Ontario, Canada. The study area
is covered by six feature classes: buildings, trees, roads, grass, soil, and water. 200 reference points are
randomly selected to assess the classification results. The results revealed 41.5%, 48.5%, and 34.0%
overall accuracy from wavelengths 532 nm, 1604 nm, and 1550 nm, respectively. The overall accuracy of
combined intensity image is improved to 65.5%. The classification results are further improved by an
additional band – Digital Surface Model (DSM) that is incorporated to the combined intensity image,
leading to an overall accuracy of 72.5%. The results are promising and showed that the use of multispectral
LiDAR data can improve the land cover classification accuracies. Further investigation is required to
explore the capability of multispectral LiDAR data.
131
Wetland mapping using polarimetric RADARSAT-2, optical imagery and LiDAR
data in Nova Scotia
Raymond Jahncke1, Peter Bush2, Brigitte Leblon3*
1. Student, Master of Environmental Studies, Dalhousie University, 6100 University Ave., Halifax, Nova Scotia,
Canada, (902) 494-6719, [email protected]
2. Protected Areas Coordinator, Nova Scotia Environment, 1903 Barrington St., Halifax, Nova Scotia, Canada, (902)
424-0202, [email protected]
3. Professor, Faculty of Forestry, University of New Brunswick, 28 Dineen Drive, Fredericton, New Brunswick,
Canada (902) 453-4924, [email protected]
* Presenting Author: Raymond Jahncke, Student, [email protected]
ABSTRACT
Nova Scotia introduced a new wetland policy in 2011 which included a goal to have no net loss of
wetlands. In order to meet this goal, the Nova Scotia government has committed to updating the
provincial wetland inventory. There are, however, wetlands that have not yet been identified in the
current wetland inventory that was based on field work and airphoto interpretation. The objective of this
study will be to identify wetlands using advanced remote sensing technology and processes based on
RADARSAT-2 polarimetric SAR images, Quickbird imagery, and lidar.
Polarimetric radar were combined in various arrangements with optical imagery and lidar terrain
derivatives. A non-parametric supervised Random Forest classifier was applied to the different date
combinations. The classified images were assessed against information collected in the field and results
show a marked improvement when radar is combined to optical imagery and lidar. The project is funded
by the Science and Operational Applications Research Education Program.
132
Collaborative case-studies demonstrating the effectiveness of random forest
classification using LiDAR, SAR and optical data in various landscapes
Millard, K1*., Banks, S2., Dingle-Robertson, L3., Richardson, M4., Duffe, J2., Pasher, J2.,
1. PhD Candidate, Carleton University, Ottawa, Ontario Canada
2. Physical Scientist, Environment Canada National Wildlife Research Centre, Ottawa, Ontario Canada
3. Visiting Fellow, AgroClimate, Geomatics and Earth Observation, Science and Technology Branch, Agri-Food and
Agriculture, Ottawa, Ontario, Canada
4. Assistant Professor, Carleton University, Ottawa, Ontario, Canada
* Presenting Author: Koreen Millard, PhD Candidate, [email protected]
ABSTRACT
Random forest (RF) is a widely used algorithm for classification of remotely sensed data. Here, we
demonstrate three innovative applications of RF using the open source program R, which has allowed us
to move beyond classifying single images to using RF for land use change over time, and allowed us to
build RF models from training data spread throughout several images covering a large area.
First, we classified a peatland in eastern Ontario (eight classes) using more than 80 LiDAR derivatives
(DEM, DSM and vegetation). RF variable importance was used to reduce input variables to only the most
important variables (OOB = 2%, independent error <10 %). Second, we conducted a Land Use Change
study in Southern Ontario and built a RF model using training data acquired from a current Landsat-7
image (six classes), then used this model to classify images acquired in the past. The independent error
ranged from 10% for 2009 to 32% for 1990 (OOB error of the original model= 2.6%). Our third
implementation focused on classifying shoreline types in the Dease Strait, Coronation Gulf, and Bathurst
Inlet, Nunavut. A single model was built using training data from 8 strips of Wide Fine Quad-Pol
RADARSAT-2 data and 12 Landsat-8 scenes. Variable importance indicates NDVI and HV intensity are the
most useful for this application, while OOB error was reported at 9.0% for 6 classes.
These applications of RF highlight its ability classify many different types of imagery (LiDAR, SAR, optical),
in different landscapes and for different applications.
133
S27 Hyperspectral Imaging – II / S27 Imagerie hyperspectrale – II
George Leblanc, chair / modérateur
Results of longer-term airborne hyperspectral imaging over single graves in a
boreal environment
George Leblanc 1*, Margaret Kalacska2, Eva Snirer2, Raymond Soffer1, Gabriela Ifimov1
1. National Research Council of Canada, Flight Research Laboratory, 1920 Research Rd, Building U-61, Ottawa ON
Canada K1A 0R6
2. Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal QC, Canada
H3A 2K6
* Presenting Author: George Leblanc, National Research Council of Canada,
ABSTRACT
The National Research Council of Canada (NRC) and McGill University partnered to develop a wellcontrolled test site for detection of clandestine graves via airborne hyperspectral imaging. The work is
based upon the use of pig carcasses (Sus scrofa) as analogues for human cadavers. The site is located on
the NRC grounds in Ottawa, Canada and consists of 18 – 1m x 2m (nominal size) grave sites that were dug
to either 60 cm or 120 cm depth (shallow and deep, respectively) and separated from each other by 8m.
Carcasses were placed in 12 of these sites (6 at shallow and 6 at deep depths) of which 6 carcasses were
in a plastic bag, and 6 carcasses, 3 of which were placed in plastic garbage bags, were surface laid. In this
study, we have closely followed the changes that have occurred over the site during a period of 1200 days
(3.3 years).
Results have shown that immediately, and up to 1 month post-burial, all the grave and surface laid sites
are easily identifiable within the imagery. Beyond the 1-month post-burial, vegetation begins to return
and over the filled sites we observe that the vegetation return is more robust than over the blank sites.
Both the quantity of biomass as well as select spectral vegetation indicies (NDVI, SGI, SIPI, etc.) are
significantly different between grave/filled grave/surface laid and surrounding natural environment. A
peak measurable contrast between filled and blank graves occurred in year 2.
134
Multi-angular spectroscopic remote sensing of arctic plant biochemistry
Blair Kennedy1*, Douglas King 2, Scott Mitchell 3, Josée Lévesque 4, H. Peter White 5, Jason Duffe 6
1. Ph.D. candidate, Department of Geography and Environmental Studies, Carleton University, Ottawa, Ontario,
Canada, 613-990-9949, [email protected]
2. Professor, Department of Geography and Environmental Studies, Carleton University, Ottawa, Ontario, Canada,
613-520-2560, [email protected]
3. Associate Professor, Department of Geography and Environmental Studies, Carleton University, Ottawa, Ontario,
Canada, 613-520-2600 x 2695, [email protected]
4. Research Scientist, Defence Research and Development Canada, Valcartier, Québec, Canada,
[email protected]
5. Research Scientist, Canada Centre for Mapping and Earth Observation, Natural Resources Canada, Ottawa,
Ontario, Canada, 613-759-6584, [email protected]
6. Geomatics Research Lab Manager - Habitat Science, National Wildlife Research Centre, Environment Canada,
Ottawa, Ontario, Canada, 613-998-9393, [email protected]
* Presenting Author: Blair Kennedy, Ph.D. candidate, 613-990-9949, [email protected]
ABSTRACT
Biogeochemical processes are driven by abiotic forces and biotic composition, where the latter is
fundamentally connected to vegetation biochemistry. Estimating the spatial distribution of foliar
biochemicals can therefore act as an important approach for monitoring plant community characteristics.
Spectroscopic instruments, which are typified by narrow spectral bandwidths, allow for an in-depth
characterization of foliar reflectance. The physical linkages between foliar chemicals and their reflectance
characteristics provide the foundation for spectroscopic remote sensing of plant biochemistry. Imaging
spectrometers mounted on satellite platforms provide the means to acquire a substantial amount of
spectral data over remote locations. Pointable multi-angular instruments (i.e. CHRIS-PROBA) help to
exploit additional reflectance information over nadir looking instruments, thus providing a greater
understanding of the independent variables contributing to reflectance. Physically based radiative
transfer models (i.e. PROSAIL) are an established technique for evaluating canopy spectral/directional
reflectance and retrieving vegetation biochemical properties from this reflectance. In this study, multiangular CHRIS-PROBA imagery, along with in situ vegetation biochemical (i.e. chlorophyll) data were
collected along a bioclimatic gradient in the Western Canadian Arctic. The PROSAIL model was
parameterized from field measurements to produce a look-up-table (LUT) of expected reflectance.
PROSAIL simulated spectra were then compared to CHRIS signatures using various cost functions.
Extracted biophysical variables from the PROSAIL inversion were compared to measured variables using
goodness-of-fit (r2). Initial results indicate that LUT-based inversion of PROSAIL from CHRIS data is suitable
for estimating chlorophyll in tundra plant canopies, where low to mid viewing angles are most suitable
(r2= ~0.6). Full results will be presented.
135
Direct Imaging of a Shale Gas Leak Using Passive Thermal Infrared Hyperspectral
Imaging
Marc-André Gagnon1*, Pierre Tremblay2, Simon Savary3 and Martin Chamberland4
1. Application Scientist, Telops, 100-2600 Saint-Jean Baptiste, Québec, Québec, Canada, +1-418-864-7808, [email protected]
2. Senior Scientist, Telops, 100-2600 Saint-Jean Baptiste, Québec, Québec, Canada, +1-418-864-7808,
[email protected]
3. System Engineer, Telops, 100-2600 Saint-Jean Baptiste, Québec, Québec, Canada, +1-418-864-7808,
[email protected]
4. Vice-President, Telops, 100-2600 Saint-Jean Baptiste, Québec, Québec, Canada, +1-418-864-7808,
[email protected]
* Presenting Author: Marc-André Gagnon, Application Scientist, Telops, +1(418)-864-7808, [email protected]
ABSTRACT
Natural gas is an energy resource in great demand worldwide. There are many types of gas fields including
shale formations which are common especially in the St-Lawrence Valley (Qc). Regardless of its origin,
methane (CH4) is the major component of natural gas. Methane gas is odorless, colorless and highly
flammable. It is also an important greenhouse gas. Therefore, dealing efficiently with methane
emanations and/or leaks is an important and challenging issue for both safety and environmental
considerations. In this regard, passive remote sensing represents an interesting approach since it allows
characterization of a large areas from a safe location. The high propensity of methane contributing to
global warming is mainly due to the fact that it is a highly infrared-active molecule. For this reason, thermal
infrared remote sensing represents one of the best approaches for methane investigations. In order to
illustrate the potential of passive thermal infrared hyperspectral imaging for research on natural gas,
imaging was carried out on a shale gas leak that unexpectedly happen during a geological survey near
Hospital Enfant-Jésus (Québec City) in December 2014. Methane was selectively identified in the scene
by its unique infrared signature. Quantitative information was also obtained. The results show how this
novel technique could be used for research work dealing with methane gas.
136
Thermal hyperspectral (8-12 μm) investigation of near surface gases from a
clandestine mass grave
Margaret Kalacska1, George Leblanc 2*, Marc-André Gagnon3, Tim Moore1, Mike Dalva1
1. Department of Geography, McGill University, 805 Sherbrooke West, Burnside Hall 705, Montreal QC, Canada
H3A 0B9
2. National Research Council Canada, Flight Research Laboratory, 1920 Research Rd., Building U61, Ottawa ON,
Canada K1A 0R6
3. Telops, 100-2600 St-Jean-Baptiste Ave, Quebec QC, Canada G2E 6J5
* Presenting Author: George Leblanc, Senior Research Officer, National Research Council Canada, Flight
Research Laboratory, 1920 Research Rd., Building U61, Ottawa ON, Canada K1A 0R6, (613) 998-3525,
[email protected],
ABSTRACT
Studies that describe the suite of volatile organic compounds (VOCs) released through the decomposition
process of human remains also reference the production of gases such as methane (CH4) nitrous oxide
(N2O) and carbon dioxide (CO2) but few have examined in depth the utility of these inorganic gases for
locating clandestine graves.
In-situ collection of gases and subsequent gas chromatography can be time consuming and requires
knowledge of the gases of interest. This study investigates the potential of thermal hyperspectral imagery
(LWIR) for the detection of the gases produced through the process of decomposition. Chemical
compounds have unique infrared signatures that can be measured outdoors by midwave or longwave
infrared hyperspectral sensors as long as certain basic requirements are met; sufficient temperature
contrast and a high enough concentration of the compounds. These methods have been successfully used
in other studies to detect a variety of chemical compounds, even if the exact gases present in a scene are
not known prior to image collection2. This study illustrates for the first time the use a Fourier Transform
infrared spectrometer (7.7-11.5 μm) for imaging gases from the surface of an experimental temperate
mass grave. Our primary gases of interest, CH4, CO2 and N2O, have identifiable features in that
wavelength range. LWIR imagery of the air column immediately above the grave was collected and
analyzed in-situ and post collection. Both the real-time in-situ analysis and the post collection analysis
illustrated the simultaneous presence of these gases in discrete ephemeral pockets above the grave.
137
S28 Land and Vegetation Classification / S28 Classification des terres
et de la végétation
Elizabeth Simms, chair / modérateur
Detecting Changes in Sub-Arctic Vegetation Caused by Snow Goose Foraging on
Coats Island, Nunavut: A Multi-Temporal Analysis
Karissa A. Reischke,1*, Mark Mallory2, Paul A. Smith3, David Colville4
1. M.Sc. Applied Geomatics Candidate, Biology Department, Acadia University, 33 Westwood Drive, Wolfville, Nova
Scotia, Canada, (905) 706-0718, [email protected]
2. Associate Professor & Canada Research Chair, Biology Department, Acadia University, 33 Westwood Drive,
Wolfville, Nova Scotia, Canada, (902) 585-1798, [email protected]
3. Research Scientist, Environment Canada, 1125 Colonel By Drive, Ottawa, Ontario, Canada, (613) 998-7362,
[email protected]
4. Research Scientist, Applied Geomatics Research Group, NSCC, 50 Elliot Road, Lawrencetown, Nova Scotia,
Canada, (902) 825-5476, [email protected]
* Presenting Author: Karissa Reischke, M.Sc. Applied Geomatics Candidate, (905) 706-0718,
[email protected]
ABSTRACT
Foraging of overabundant Snow Geese (Chen caerulescens) in North America has caused detrimental
impacts on vegetation surrounding their sub-arctic breeding grounds [1, 2]. Consequently, this increase
in barren land cover removes nesting habitat for sympatric species [3, 1, 2]. Coats Island, Nunavut is a
unique study area where changes in vegetation can be investigated before/after Snow Geese have first
begun nesting on the island [4, 5]. The objectives in this study were to: (i) create an annual land cover
classification of northern Coats Island (1991-2014) from LANDSAT imagery; and (ii) detect changes in
vegetation on Coats Island potentially attributable to foraging by Snow Geese. Using LANDSAT 5 TM and
LANDSAT 8 OLI satellite images and three ASTER GDEM tiles, a supervised classification with a random
forest classifier was used to annually classify northern Coats Island (1991-2014). LANDSAT and DEMderived input variables were also used to improve the classification accuracy. Training areas for six land
cover types were created from the ground truthing points (n=85) collected on July 1-22, 2014. Normalized
Difference Vegetation Index (NDVI) surfaces were created to assess changes in vegetation quality. Our
results show no significant change in July NDVI values (corresponding closely to goose breeding season).
Results correspond with on-site field observations which suggest that geese have had no major, negative
impact on local vegetation on Coats Island, in contrast to vegetation damage observed at other similar
coastal habitat sites with nesting Snow Goose colonies around Hudson Bay.
References:
[1] Jefferies, R.L., Jano, A.P., & Abraham, K.F. (2006). A biotic agent promotes large-scale catastrophic
change in the coastal marshes of Hudson Bay. Journal of Ecology, 94(1), 234-242.
[2] Peterson, S.L., Rockwell, R.F., Witte, C.R., & Koons, D.N. (2014). Legacy effects of habitat degradation
by Lesser Snow Geese on nesting Savannah Sparrows. The Condor, 116(4), 527-537.
138
[3] Kerbes, R.H., Kotanen, P.M., & Jefferies, R.L. (1990). Destruction of Wetland Habitats by Lesser Snow
Geese: A Keystone Species on the West Coast of Hudson Bay. Journal of Applied Ecology, 27(1), 242-258.
[4] Gaston, A.J. & Ouellet, H. (1997). Birds and Mammals of Coats Island, N.W.T. Arctic, 50(2), 101-118.
[5] Kerbes, R.H., Meeres, K.M., & Alisauskas, R.T. (2014). Surveys Nesting Snow Geese and Ross’s Geese
in Arctic Canada, 2002-2009. Arctic Goose Joint Venture Special Publication, U.S. Fish and Wildlife Service,
Washington, D.C. and Canadian Wildlife Service, Ottawa, Ontario.
139
Updating the Grassland Vegetation Inventory Using Landsat Imagery-Conversion
of Native Grassland to Cultivated Agriculture
Xiaohui Yang1, Anne Smith2*, and Michael Hill3
1. PDF, Agriculture and Agri-Food Canada, 5403-1st Ave S, Lethbridge, AB, Canada, 403-327-4561,
[email protected]
2. Research Scientist, Agriculture and Agri-Food Canada, 5403-1st Ave S, Lethbridge, AB, Canada, 403-317-2285,
[email protected]
3. Professor, Earth System Science and Policy, University of North Dakota, Grand Forks, ND, USA, 701-777-2940,
[email protected]
* Presenting Author: Anne Smith, Research Scientist, 403-317-2285, [email protected].
ABSTRACT
The Grassland Vegetation Inventory (GVI) of Alberta is the first detailed spatial, comprehensive
biophysical and anthropogenic land use inventory which has its basis in Canada’s native grasslands. The
GVI was developed using airborne imagery and manual interpretation (2006-2011) which is not
economically viable in terms of updates to determine conversion of grassland to cultivated agriculture
every 5 to 10 years. In this study the potential of Landsat TM imagery for updating the GVI was
investigated. Vegetation indices (VIs) representing photosynthetic pigments, vegetation and landscape
water content, senescent vegetation and soil, and herbaceous biomass were first evaluated for their
potential to discriminate native grassland from cultivated agriculture south east of Brooks, Alberta.
Results indicated that discrimination was superior with Landsat imagery acquired in July or August
compared to May or September. The combination of dMTVI2 (Difference Modified Triangular Vegetation
Index II) measuring vegetation photosynthesis and dSWIR32 (Difference Shortwave Infrared Reflectance
3/2 Ratio Index) sensitive to senescent vegetation and soil derived from the July image identified
conversion with an accuracy of 100% based on ground validation data. A total of four VIs, dNDII
(Difference Normalized Difference Infrared Index), dSWIR32, dSATVI (Difference Soil Adjusted Total
Vegetation Index), and dRGR (Difference Red-Green Ratio Index), were required to achieve the same
accuracy when an August image was used. A map indicating the conversion, and thus loss of native
grassland during the 5-year period 2006 to 2011 was produced though change vector analysis using the
dMTVI2 and dSWIR32 derived from July imagery.
140
Retrieving leaf chlorophyll in crops using Landsat-8 and RapidEye images
Joyce Arabian1*, Jing Chen 2, Holly Croft3, Nadine Nesbitt 4, Jiali Shang5
1. Joyce Arabian, M.Sc. student, Department of Geography, University of Toronto, Toronto, Ontario, Canada, 416859-4815, [email protected]
2. Jing M. Chen, Professor, Department of Geography, University of Toronto, Toronto, Ontario, Canada, 416-9787085, [email protected]
3. Holly Croft, Postdoctoral Fellow, Department of Geography, University of Toronto, Toronto, Ontario, Canada,
416-553-8431, [email protected]
4. Nadine Nesbitt, M.Sc. student, Department of Geography, University of Toronto, Toronto, Ontario, Canada, 905669-1223, [email protected]
5. Jiali Shang, Research Scientist, Science and Technology Branch, Agriculture and Food Canada, 613-759-1435,
[email protected]
* Presenting Author: Joyce Arabian, MSc Candidate, 416-859-4815, [email protected].
ABSTRACT
Leaf chlorophyll content (Chlab) acts as a main indicator of vegetation health and is vital for cropland
management as well as understanding plant productivity. The purpose of this study is to develop an
accurate method of modelling crop Chlab using remote sensing data within a physically-based modelling
approach. During the 2013 growing season, ground data was collected at 15 wheat sites and 13 corn sites
over 11 days between May-September near Stratford, Ontario. Of the total 28 sites, 11 were within
controlled areas of zero nitrogen fertilizer applications. Effective leaf area index was measured using the
Li-Cor LAI-2000 and hyperspectral leaf reflectance and transmittance data was sampled using an ASD
FieldSpecPro spectroradiometer (400-2500 nm). Chlab was chemically extracted and measured using a
Cary-1 spectrometer. A two-step inversion process is being developed to model crop Chlab using Landsat8 and RapidEye satellite imagery. In this process, a look-up-table (LUT) is developed using 4-Scale, a
geometrical-optical radiative-transfer model, to simulate canopy reflectance values using both canopy
structural parameters and viewing/illumination geometry. This LUT will be utilized to calculate leaf level
reflectance which is used as an input to PROSPECT, a leaf-level radiative-transfer model in order to
estimate leaf-level Chlab. Validation of PROSPECT with ground-based Chlab from Stratford using simulated
Landsat-8 bands shows an R²= 0.78 and RMSE=9.73 for corn and R²= 0.85 and RMSE=8.75 for wheat. This
research provides an operational basis for modelling within-field variations in Chlab across a growing
season, which is critical for precision agriculture applications and managing nutrient optimization
programs that increase crop yield.
141
Object-based Classification of WorldView-2 Image for Development of
Stormwater Models
Moh Moh Lin Khin1, Ahmed Shaker2*, Darko Joksimovic3 and Wai Yeung Yan4
1. MASc Graduate, Ryerson University, 350 Victoria Street, Toronto, ON, Canada, [email protected]
2. Associate Professor, Ryerson University, 350 Victoria Street, Toronto, ON, Canada, [email protected]
3. Assistant Professor, Ryerson University, 350 Victoria Street, Toronto, ON, Canada, [email protected]
4. Postdoctoral Fellow, Ryerson University, 350 Victoria Street, Toronto, ON, Canada, [email protected]
* Presenting Author: Ahmed Shaker, Associate Professor, 4169795000x6458,
[email protected].
ABSTRACT
Surface runoff source controls such as Low Impact Development (LID) techniques are being used as retrofit
options for older developed areas that lack available land to implement conventional practices,
stormwater management ponds. The complexity of the catchment area with LID requires detailed
distributed urban drainage models, which need high resolution land cover information to accurately
estimate the benefit that LIDs may provide. Most of the urbanized areas lack such detailed information;
therefore, this study explores the use of high resolution WorldView-2 satellite image in urban drainage
modeling. The objectives of this study are: 1) to assess the accuracy of a proposed two-stage image
classification method on WorldView-2 image and 2) to analyze the feasibility of using extracted land cover
information in-lieu of GIS data in subsequent urban drainage modelling. A proposed two-stage image
classification method is composed of rule-based decision tree in first stage and topological rules in second
stage. A proposed method produced 83% accuracy in classifying an urban residential area into tree, grass,
road and building. Then, the U.S Environmental Protection Agency (EPA) Storm Water Management
Models (SWMM) were developed using the extracted land cover information from WorldView-2. These
urban drainage models gave comparable but mixed results for different drainage systems: curb and
gutter, roadside ditches and LIDs against the simulated results of models developed using GIS data and
observed runoff data. The comparative analyses signified the guidance data to be used for the purpose of
planning and managing stormwater systems.
142
Exploiting Three Decades of Continuous Satellite Data with the Canada Centre
for Remote Sensing Long Term Satellite Data Records
Darren Janzen1*, Alexander Trichtchenko2, Rasim Latifovic3, Shusen Wang4, Richard Fernandes5, Darren
Pouliot6, Fuqun Zhou7, Junhua Li8, John Schwarz9, Lixin Sun10
1. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6339, [email protected]
2. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 1446, [email protected]
3. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7002, [email protected]
4. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6462, [email protected]
5. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6193, [email protected]
6. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6341, [email protected]
7. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7963, [email protected]
8. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7205, Li, [email protected]
9. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7759, [email protected]
10. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 7696, Lixin.sun @nrcan.gc.ca
* Presenting Author: Darren Janzen, Remote sensing Scientist, 613 759 6339,
[email protected]
ABSTRACT
How can decades of satellite data help us understand our planet and manage our activities? Canada is
vast and has extensive sparsely populated areas. Thus, spatially explicit and detailed long-term records of
the environment do not exist for the majority of the country. To assess and address change consistently
across Canada, the best available data is satellite imagery spanning decades. However, extracting
systematic and meaningful information from raw satellite data requires methods to address a wide range
of issues including clouds, cloud shadow, atmospheric haze, viewing geometry, and geolocational
accuracy among others.
The Canada Centre for Remote Sensing has addressed these challenges, generating the Long Term
Satellite Data Records. The foundation of these records are national scale 10-day cloud free composites
spanning decades. These data products and their associated methods are used by a wide range of
governmental agencies, academia, and commercial groups for monitoring, reporting, decision-making,
and facilitating further analysis and research.
There are currently over 1,000 composites spanning 30 years of environmental change across Canada. For
the first time, these datasets are available to the public through a public data access portal. An overview
will be provided on the implementation and exploitation of these datasets, including methods and
development for LTSDR successor sensors, application of LTSDR methods to moderate resolution sensors,
and efforts to produce comprehensive value-added datasets of national interest, such as land cover, snow
143
cover, albedo, and vegetation indices. Examples will illustrate what these data are telling us about the
state of Canada.
144
Classification Uniformisée de la Couverture Terrestre pour une Comptabilité des
Terres et des Écosystèmes
Standardized Land Cover Classification for Land and Ecosystem Accounting
Stéphanie Uhde1, Richard Fournier2*, et Marcel Darveau3
1. Économiste, Direction des statistiques sectorielles et du développement durable, Institut de la statistique du
Québec, 200, chemin Sainte-Foy, 3e étage, Québec, Québec, G1R 5T4, Canada, 418-691-2411, p. 3002,
[email protected]
2. Professeur titulaire, Département de géomatique appliquée, Centre d’Applications et de Recherches en
Télédétection, Université de Sherbrooke, 2500 boulevard de l’Université, Sherbrooke, Québec, J1K 2R1, Canada,
819-821-8000 p. 63209, [email protected]
3. Chef, recherche et conservation boréales pour le Québec, Canards Illimités Canada, 710 rue Bouvier, bureau 260,
Québec, Québec, G2J 1C2, Canada, 418-623-1650 p. 26, [email protected]
* Conférencier/Speaker : Richard Fournier, Professeur titulaire, 819-821-8000 p. 63209,
[email protected].
RÉSUMÉ
Les effets de plus en plus grands de l’activité humaine sur l’environnement rendent nécessaire une
meilleure gestion du capital naturel et des services produits par les écosystèmes. Plusieurs pays ou
gouvernements sous-nationaux élaborent actuellement des systèmes de comptabilité des terres et des
écosystèmes, qui font appel à des données recueillies par télédétection ou photos aériennes.
La comptabilité des terres et des écosystèmes demande un grand niveau de rigueur et de comparabilité,
à l’instar du Système de comptabilité économique, dont sont dérivés les grands indicateurs économiques.
Nous avons développé une classification uniformisée de la couverture terrestre pour le Québec. Cette
classification hiérarchique comprend douze classes de premier niveau et une trentaine de classes de
deuxième et troisième niveau. Bien qu’elle ait été élaborée pour être appliquée au Québec, cette
classification est conforme à la classification standard du Système de comptabilité environnementale et
économique (SCEE), laquelle respecte les concepts du Land Cover Classification System (LCCS) de la FAO.
L’ensemble des classes de couverture terrestre, qui sont mutuellement exclusives, permet de décrire la
totalité du territoire québécois. La structure hiérarchique de la classification la rend flexible et par le fait
même applicable à différents exercices de cartographie et de télédétection.
Dans le cadre de la production des comptes des terres pour le sud du Québec, la classification proposée
a déjà été utilisée pour l’interprétation des cartes écoforestières du Québec. Elle constitue un aspect
important de l’utilisation des produits de la télédétection en appui à la création d’un système rigoureux
de comptabilité des écosystèmes pour le Québec.
SUMMARY
Increasing pressure of human activities on the environment has led to a need to improve the management
of natural capital and services produced by ecosystems. Many countries or sub-national governments are
presently putting together land and ecosystem accounts, which draw on data acquired by remote sensing
or aerial photography.
Land and ecosystem accounting requires a high level of rigour and comparability similar to the System of
National Accounts used for economical reporting. We developed a standardized land cover classification
for Quebec. This hierarchical classification comprises twelve first-level classes and about thirty second145
and third-level classes. Although it was developed for application in Quebec, this classification is
compatible with the System of Environmental-Economic Accounting (SEEA)’s classification, which respects
FAO’s Land Cover Classification System (LCCS). The land cover classes are mutually exclusive and, taken
together, describe all land in Quebec. The classification’s hierarchical structure makes it flexible to apply
to mapping and remote sensing exercises at various levels of detail.
The proposed classification has already been used in southern Quebec, with forest maps of different
dates, for the accounting of land cover change. The proposed classification is an important element
guiding the use of remote sensing products in support of the implementation of a rigorous ecosystem
accounting system for Quebec.
146
S29 Coastal and Ice / S29 Glace et régions côtières
Joseph Chamberland, chair / modérateur
The SmartICE information system – integrating Inuit knowledge, remote
sensing, and in situ measurements for safer sea-ice travel
Robert Briggs1*, Trevor Bell2
1. Research Scientist, C-CORE, Captain Robert A. Bartlett Building, Morrissey Road, St. John's, NL, Canada A1B 3X5,
(709) 895-6197, [email protected]
2. University Research Professor, Geography Department, Memorial University of Newfoundland, St. John's,
Newfoundland and Labrador, A1B 3X9, (709) 864-2525, [email protected]
* Presenting Author: Robert Briggs, Research Scientist, (709) 895-6197, [email protected]
ABSTRACT
Sea ice is a key component of the Arctic coastal environment. For northern communities, it is a critical
part of culture, wellbeing, and livelihood. Recent changes in Arctic climate have led to greater
unpredictability in sea-ice conditions, making travel and hunting more hazardous, particularly during the
dynamic freeze-up and break-up periods. For northern resource industries and maritime shipping, the
presence of sea ice represents additional operational risk. In regions where shipping routes overlap with
community on-ice travel routes, the risk is greater; route planning in these coastal areas is less predictable
and the potential for delay and additional costs is increased. SmartICE represents a communitygovernment-university-industry collaboration that aims to deliver timely information of sea ice
conditions, at the relevant spatial scales, presented in a format that is appropriate for each user group. It
will support planning and decision-making, leading to decreased usage conflicts and help ensure safe and
efficient winter travel for both local communities and industry. The key components of SmartICE are: (i)
A network of automated in situ sensors that measure sea-ice thickness and other characteristics; (ii)
Repeat satellite imagery from which sea-ice surface conditions (e.g., concentration, roughness, water
content) are mapped following user-defined classification systems; and (iii) GIS Information technology
that integrates the in situ and remotely sensed sea-ice data to generate digital products that match the
needs of user groups. We present RADARSAT2 derived results for sea-ice season 2013/14 and, for season
2014/15, preliminary results from Sentinel 1 and the automated in situ sensors.
147
Bottom-fast Ice Delineation in the Northwest Territories using Dynamic Time
Warping
Tsui, Olivier W.1*, Chiang, Michael2, Dean, Andy3
1. Senior Geomatics Specialist, Hatfield Consultants Partnership, Suite 200 - 850 Harbourside Drive, North
Vancouver, BC, Canada, +1 604 926 3261, [email protected]
2. Data Scientist, Isturary innovation Labs LLC, Vancouver, BC, Canada, +1 604 299 0388 [email protected]
3. Senior Geomatics Specialist and Partner, Hatfield Consultants Partnership, Suite 305 - 1228 Kensington Rd. NW,
Calgary, AB, Canada, +1 403 351 0191, [email protected]
* Presenting Author: Olivier W. Tsui, Senior Geomatics Specialist, +1 604 926 3261,
[email protected]
ABSTRACT
Given its size and remoteness, harsh climate, and limited infrastructure, environmental baseline data
describing most Canadian northern regions are limited. This is particularly evident for water resources,
which can impede effective environmental planning, assessment, and monitoring. In the Northwest
Territories, the vast number of freshwater lakes in the North Slave and Central Mackenzie Valley require
effective and cost-efficient methods to monitor changes in ice cover and status across large areas.
Multi-date RADARSAT-2 Wide Fine dual-polarized data were acquired and processed to provide indirect
information of ice thicknesses and water depth through the detection and delineation of bottom-fast ice.
A dynamic time warping (DTW) methodology was used, which exploits the temporal profile of radar
backscatter as ice growth progresses and then becomes grounded. Backscatter time series for each lake
pixel was compared to a set of reference time series, collected using ground penetrating radar,
representing different backscatter profiles of bottom-fast ice. The methodology accounted for speckle
noise, non-synchronicity (e.g., grounded-ice occurring at different times and rates on different lakes), and
scaling (e.g., varying backscatter levels for different lakes due to incidence angle).
Individual lakes were used as the fundamental processing unit, with the proportion of bottom-fast or
floating ice mapped and visualized. Large volumes of data were processed with little manual intervention.
Overall accuracy was obtained using a leave one out cross-validation approach and overall classification
accuracy was 89.1%. This study suggests the potential of DTW as an efficient method to map the
spatial distribution of bottom-fast ice.
148
Nature and triggers of submarine mass failure in coastal waters of southeastern
Baffin Island, Nunavut
Robert Deering1*, Trevor Bell2, Donald L. Forbes3
1. MSc. Student, Memorial University of Newfoundland, Department of Geography, St. John’s, NL, Canada, A1B 3X9,
(709) 697-3667, [email protected]
2. Professor, Memorial University of Newfoundland, Department of Geography, St. John’s, NL, Canada, A1B 3X9,
(709) 864-2525, [email protected]
3. Adjunct Professor, Memorial University of Newfoundland, Department of Geography, St. John’s, NL, Canada, A1B
3X9, (902) 426-7737, [email protected]
* Presenting Author: Robert Deering, MSc. Student, (709) 697-3667, [email protected]
ABSTRACT
With advances in seabed mapping technology and its application in the Canadian Arctic, new records of
submarine mass failures (SMF) are being revealed in coastal and nearshore environments. SMFs are
typically features of continental and island margins, ranging from small-scale features (<100 m3) to
enormous failures (3000 km3). The largest SMFs in shelf and slope environments have received more
attention, understandably, than smaller inshore features. Recent mapping, however, has shown that small
failures are prevalent in the fjords of eastern Baffin Island, whiles studies in Norway and Scotland have
demonstrated that small failures may be potentially destructive hazards; for example, causing tsunamis
in fjords and damaging seabed infrastructure. SMFs can be triggered in a number of ways, such as
overloading by sediments, undercutting of slopes, and seismic events. Of particular interest is the
potential for triggering SMFs by sea-level adjustments. The easternmost fringe of Baffin Island has
experienced substantial postglacial relative sea-level rise (30-50 m). Through coring of SMF deposits, it
may be possible to determine the ages of events in relation to episodes of rapid relative sea-level change.
Our research methods include mapping and characterizing the morphology of mass movements to
develop a typology of forms and associated processes, multibeam bathymetric surveying of features to
examine the seabed settings in which these events occur, and acoustic subbottom profiling to examine
the stratigraphy underlying past events. These data will permit a fuller understanding of SMF events and
their seabed settings, as well as their potential triggers in Baffin Island fjords.
149
An alternative method for sea surface wind speed determination from GNSS-R
delay-Doppler map
Qingyun Yan1* and Weimin Huang2
1. Master student, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St. John’s,
NL, Canada, (709)-330-3273, [email protected]
2. Assistant Professor, Faculty of Engineering and Applied Science, Memorial University of Newfoundland, St.
John’s, NL, Canada, (709)-864-8937, [email protected]
* Presenting Author: Qingyun Yan, Master student, (709)-330-3273, [email protected].
ABSTRACT
In this paper, a new method is proposed to retrieve sea surface wind speed from Global Navigation
Satellite System-Reflectometry (GNSS-R) Delay-Doppler Map (DDM). In this work, the maximum-power
curve which contains points of maximum power along axes is extracted from DDM for the wind speed
retrieval, and this process can reduce the computation cost. The wind speed associated with an initial
DDM is known, the maximum-power curve in the initial DDM is deducted from those of new DDMs,
generating the maximum-power-difference curve for each. For the first consideration, an assumption is
made that the measuring system is stationary so that the power difference between the afterwards
maximum-power curve and the initial one is caused by the variation in sea surface wind field. Next, a
relationship between power difference and wind speed is fitted based on simulated DDMs. In application,
with an initial DDM and associated wind speed, the sea surface wind speed corresponding to a new DDM
can be retrieved from the fitted relationship. Based on the discussion of the simplified case, a more
general scenario in which the transmitter and receiver are moving is also investigated. The method is
tested using simulated data. The retrieval result is compared with the input wind speed values and good
consistency is observed. Root mean square errors of 0.12 m/s and 0.53 m/s for the simplified and general
scenario are obtained, respectively. Conclusion can be drawn that this method is accurate when the DDM
detection interval is less than 0.5 s.
150
S30 Monitoring & Modeling Terrestrial Ecosystems (Forests) / S30
Suivi et modélisation des écosystèmes terrestres (forêts)
Richard Fournier and Nicholas Coops, co-chairs / modérateurs
National, annual, gap-free surface reflectance composites from Landsat to
capture long-term forest dynamics, land cover, and forest structure for Canada
Joanne C. White1*, Michael A. Wulder2, Geordie Hobart3, Txomin Hermosilla4, Nicholas C. Coops5
1. Research Scientist, Canadian Forest Service, Natural Resources Canada, Pacific Forestry Centre, Victoria, British
Columbia, Canada, 250-298-2402, [email protected]
2. Senior Research Scientist, Canadian Forest Service, Natural Resources Canada, Pacific Forestry Centre, Victoria,
British Columbia, Canada, 250-298-2401, [email protected]
3. Physical Scientist, Canadian Forest Service, Natural Resources Canada, Pacific Forestry Centre, Victoria, British
Columbia, Canada, 250-298-2403, [email protected]
4. Post-Doctoral Fellow, Department of Forest Resource Management, University of British Columbia, Vancouver,
British Columbia, Canada, 604-822-6452, [email protected]
5. Canada Research Chair in Remote Sensing, Department of Forest Resource Management, University of British
Columbia, Vancouver, British Columbia, Canada, 604-822-6452, [email protected]
* Presenting Author: Joanne White, Research Scientist, 250-298-2402, [email protected]
ABSTRACT
It is important for Canada to produce and be able to stand behind data products used for science,
reporting, and monitoring of forest ecosystems. To support these aims, we have been using bestavailable-pixel image composites generated from tens of thousands of Landsat images to characterize
conditions and changes over Canada's forests from 1984 to 2012. We have developed approaches to
produce annual gap-free, surface reflectance image composites (hereafter referred to as proxy
composites), as well as a suite of informative change metrics derived from the time series of annual proxy
composites. Our ongoing work is focused on investigating the spatial and temporal characteristics of the
detected changes, automating change attribution, mapping land cover, and estimating forest structural
attributes of interest. Given Landsat’s 30 m spatial resolution, the information products developed are
informative and relevant across a range of spatial scales, effectively capturing human and natural
processes over large areas in a systematic and repeatable fashion. A consistent, national data set
characterizing forests and forest change since 1984 provides valuable baseline information for science,
monitoring, and reporting activities. This presentation will describe key methods developed to produce
NTEMS products (gap-free annual reflectance imagery, times series change metrics, land cover, and forest
structural attributes), provide an update on project status, and communicate key findings to date.
151
The Newfoundland Fibre Project: Four Studies that Demonstrate the use of
Remote Sensing to Predict Wood Fibre Attributes
Richard A. Fournier1* Joan E. Luther2, Wade W. Bowers3, Tim Moulton4, Olivier R. van Lier5, JeanFrançois Côté6, Danny Blanchette1 and Émilie Lessard1
1. Prof. Titulaire, Département de géomatique appliquée, Université de Sherbrooke, 2500 boul. de l’Université,
Sherbrooke, QC, J1K 2R1, CANADA, 819-821-8000 ext. 63209, [email protected],
[email protected], [email protected]
2. Research Scientist, Natural Resources Canada, Canadian Forest Service - Atlantic Forestry Centre, 26 University
Drive, Corner Brook, A2H 5G5, Canada, 709-637-4917, [email protected]
3. Professor, Grenfell Campus, Memorial University of Newfoundland, 20 University Drive, Corner Brook, NL, A2H
5G4, Canada, 709-637-6200, [email protected]
4. General Operations Superintendent, Corner Brook Pulp and Paper Limited, 1 Mill Rd., Corner Brook, NL, A2H 6J4,
Canada, 709-637-3393, [email protected]
5. Geospatial Research Assistant, Natural Resources Canada, Canadian Forest Service – Canadian Wood Fibre
Centre, 26 University Drive, Corner Brook, NL, A2H 5G5, Canada, 7-9-637-4944, [email protected]
6. Research Scientist, Natural Resources Canada, Canadian Forest Service – Canadian Wood Fibre Centre, University
Drive, QC, G1V 4C7, Canada, 418-648-3982, [email protected]
* Presenting Author: Richard Fournier, Prof. Titulaire, 819-821-8000 ext. 63209,
[email protected].
ABSTRACT
Forest industry and forest managers require information regarding variation in wood fibre attributes over
large areas in order to expand the economic benefits from wood raw materials and to optimize forest
value. The Newfoundland Fibre Inventory Project (NFIP) was initiated in 2009 with a goal to develop new
methods and tools to enhance current forest inventories with information on wood fibre attributes. The
proposed project had three main objectives aimed towards progressive improvements in the ability to
characterize fibre attributes in the forest: (i) quantify relationships among forest and environmental
variables and fibre attributes at tree, plot and landscape scales (ii) develop new inventory tools to enhance
the measurement of forest characteristics linked to fibre attributes, and (iii) improve mapping of fibre
attributes at the landscape scale. These innovative developments were possible because of an extensive
university, government and industry partnership and the availability of an exceptional database on wood
fibre attributes from about 2,000 trees distributed over about 200 plots in Newfoundland. Terrestrial and
airborne lidar data supplemented ground measurements and supplied detailed structural measurements
of trees and stands.
The NFIP completed four interrelated studies. The first study assessed the capability to predict wood fibre
attributes of balsam fir and black spruce forests using environmental variables and forest variables
measured at inventory plots or available from stand-level maps. The study provided models to predict
fiber attributes at plot and landscape levels and demonstrated potential to map fibre attributes over the
boreal forest of Newfoundland. The second study explored the use of terrestrial lidar data to enhance the
measurement of forest characteristics linked to fibre attributes and established enhanced predictability
of wood fibre attributes using new structural metrics derived from terrestrial lidar data. The third study
explored in fine detail which structural characteristics of trees and stands are best related to wood fiber
attributes. Lastly, the fourth study quantified statistical relationships between airborne lidar data and
wood attributes and demonstrated the predictive capacity of airborne lidar data to map wood attributes.
The results demonstrate enhanced inventory tools for forest characterization. New projects are on-going
152
to improve the prediction power of existing models and to improve the capability to predict wood
attributes across measurement scales.
153
Airborne LiDAR enhances the competitiveness of Canada’s forest industry
Doug Pitt1* Murray Woods2, Greg Adams3, and Chad St. Amand4
1. Research Scientist, Canadian Wood Fibre Centre, Canadian Forest Service, 1219 Queen St. E. Sault Ste. Marie, ON,
CANADA, P6A 2E5, 705-541-5610, [email protected]
2. Senior Analyst – Forested Landscapes, Ontario Ministry of Natural Resources and Forestry, 3301 Trout Lake Road,
North Bay, ON, CANADA, P1A 4L7, 705-475-5561, murray.woods@ontario
3. Manager, R&D, Nurseries and Tree Improvement, J.D. Irving Ltd., 181 Aiton Rd., Sussex East, NB, CANADA, E4G
2V5, 506-432-2844, [email protected]
4. Forest Information Services Coordinator, Tembec, 5310 Hwy 101 West , PO Box 1100, Timmins, ON CANADA, P4N
7H9, 705-360-1283, [email protected]
* Presenting Author: Doug Pitt, Research Scientist, 705-541-5610, [email protected].
ABSTRACT
Since its introduction to Canadian forest industry in 2005, airborne LiDAR flown in wall-to-wall fashion
over large forested tracts has precipitated a sea change in the precision and economy with which forests
are managed. Tembec, situated in Northeastern Ontario, and J.D. Irving, in the Maritime Provinces, have
been among a small group of forest companies that have helped pioneer the implementation of LiDARderived digital elevation models (DEMs) and area-based predictions of stand metrics for forest
management. For example, DEMs approaching 1-m accuracy have resulted in significant cost savings and
mitigation of environmental impacts associated with road construction and harvesting operations. These
DEMs are enabling forest managers to better understand how water flows across and through forest
landscapes, to better size culverts and bridges, and make spatial predictions of the productive capacity of
forest soils. Areas of operation and safe machine limits can now be efficiently mapped and displayed onscreen, with GPS providing true spatial awareness to operators, without the cost of physically identifying
features on the ground. Accurate predictions of tree size distribution and volume derived from LiDAR
point clouds are enabling managers to plan with much more spatial precision than previously possible.
Such predictions are enabling the monitoring of silvicultural success and the development of enhanced
prescriptions associated with activities such as commercial thinning and the timing of final harvest, with
significant cost-savings in the amount of field time needed to accomplish these activities. Airborne LiDAR
has truly been a “game changer” for forest industry during a period of serious economic constraint.
154
High-resolution global maps of 21st-century forest loss: validation and
application in the Miramichi river basin
Julia Linke1*, Marie-Josée Fortin2 and Simon Courtenay3
1. NSERC PDF, Department of Ecology and Evolutionary Biology, University of Toronto, 25 Harbord Street, Toronto,
ON, Canada, 506 3670987, [email protected]
2. Professor, Department of Ecology and Evolutionary Biology University of Toronto, 25 Harbord Street, Toronto,
ON, Canada, 416 9467886, [email protected]
3. Scientific Director, Canadian Water Network, University of Waterloo, 200 University Ave. W., Waterloo, ON,
Canada, 519 8884567, [email protected]
* Presenting Author: Julia Linke, NSERC PDF, 506 3670987, [email protected]
ABSTRACT
Monitoring forest resources requires accurate, up-to-date information on the type and extent of forest
depletions. While these data may be available for public lands, its attributes or quality vary across
jurisdictions, and they may be completely inaccessible for private forests land where the owners may be
reluctant to publicly share information on their harvesting operations. An opportunity to tackle this
situation could be provided by a new Landsat-based global dataset of forest change developed by Hansen
et al. (2013). The dataset contains year-by-year forest changes for the period 2000 to 2012, with annual
updates planned. Here we report on the methodology and preliminary findings of our accuracy
assessment of this dataset in the basin of the Miramichi river (NB), which was carried out as part of the
National Science Engineering Research Council Canadian Network for Aquatic Ecosystem Services (NSERC
CNAES) project on the combined cumulative effects of climate change and forestry and agriculture on
aquatic systems. We vectorised the portion of the dataset corresponding to public lands in the Miramichi
basin (ca. 11,000 polygons, MMU 0.5 ha), for which we had a complete provincial inventory of harvest
cutblocks, and compared both datasets through GIS overlays and visual checks in satellite imagery. We
present 1) thematic accuracy (harvested or not, year); 2) detectability of various harvest types (standreplacing versus selective); and 3) annual trends of forest loss. Our results confirm the synergies that this
global map of forest loss can provide to large-area, multi-disciplinary research efforts.
155
Operational Successes and Challenges of Deriving Forest Inventories from LiDAR
in Natural and Managed Forests
Murray Woods1*, Doug Pitt2 and Margaret Penner3
1. Senior Analyst – Forested Landscapes, Ontario Ministry of Natural Resources and Forestry, 3301 Trout Lake Road,
North Bay, ON, CANADA, P1A 4L7, 705-475-5561, murray.woods@ontario
2. Research Scientist, Canadian Wood Fibre Centre, Canadian Forest Service, 1219 Queen St. E. Sault Ste. Marie, ON,
CANADA, P6A 2E5, 705-541-5610, [email protected]
3. Forest Analysis Ltd., 1188 Walker Lake Dr., RR4, Huntsville, ON, CANADA, 705-635-1314,
[email protected]
* Presenting Author: Murray Woods, Senior Analyst – Forested Landscapes, 705-475-5561,
murray.woods@ontario.
ABSTRACT
LiDAR-derived (400m2 raster) forest inventories for large boreal forest management units have been
developed and operationally implemented in Ontario and elsewhere. Predicted attributes include height
(top, dominant & co-dominant and mean), basal area (merchantable and total), volumes (merchantable
and total), quadratic mean diameter and size-class distributions (basal area and volume). The accuracy
and precision of these predictions in the largely even-aged stands of the boreal have been suitably high
to garner rapid and widespread adoption and implementation by the forest sector.
We recently applied similar area-based prediction approaches to a Great-Lakes St. Lawrence forest and
encountered larger than anticipated prediction errors. In these species diverse forests, commonly
practiced silvicultural systems like uniform shelterwood, single-tree selection and row thinning regimes
alter the horizontal and vertical height structure of stands creating unique challenges that are not
encountered in the boreal forest .
We will present situations where traditional LiDAR area-based modeling approaches performed
acceptably in these non-boreal conditions and where approaches need to be modified to adapt to the
more challenging structural conditions encountered.
156
S31 Classification, Modeling & Northern Applications / S31
Classification, modélisation et applications nordiques
Derek R. Peddle, chair / modérateur
Land Surface Phenology and Vegetation Productivity for Circumpolar Landmass
Alemu Gonsamo1*and Jing M. Chen1
1. Department of Geography and Program in Planning, University of Toronto, Rm. 5047, Sidney Smith Hall, 100 St.
George St., Toronto, Ontario, Canada
* Presenting Author: Alemu Gonsamo, Postdoctoral fellow/Dr., +1 4169467715,
[email protected].
ABSTRACT
The effects of climate change in the circumpolar north are no longer debatable; radical changes are
happening in sea ice, vegetation, and animals. Land surface phenology (LSP) defined as the study of the
timing of recurring seasonal pattern of variation in vegetated land surfaces observed from synoptic
sensors has become a key research topic in the understanding of the effects of climate change on
vegetated lands surface. Despite the fact that LSP has emerged as an important focus for ecological and
global change researches, there are still large lingering uncertainties related to the determination of the
key phenological phases such as: start of growing season (SOS) and end of growing season (EOS).
Uncertainties in interpreting of the most commonly used normalized difference vegetation index (NDVI)
for LSP study arise from limitations such as contaminations by background reflectances from soil, leaf
litter, dead branches, snow, soil thaw, snow thaw, and shadows, which also have distinct seasonal
dynamics often misinterpreted as vegetation LSP. Here we present an LSP product for circumpolar
landmass (>45o N) using a newly developed remote-sensing-based phenology index (PI) which combines
the merits of NDVI and normalized difference infrared index (NDII) by taking the difference of squared
greenness and wetness to remove the soil and snow cover dynamics from key vegetation LSP cycles. We
have also produced vegetation productivity maps using the growing season integrated NDVI. Both the LSP
and vegetation productivity maps are for 1999-2013 using the SPOT VGT satellite sensor.
157
Remotely Sensed Caribou Habitat Indicators for Enhancing Baseline Information
Preparedness for Resource Development in Canada’s North
Wenjun Chen1*, Jan Z. Adamczewski2, Bruno Croft2, Lori White1, Sylvain Leblanc1, Kerri Garner3, Adeline
Football3, and Boyan Tracz4
1. Canada Centre for Remote Sensing, Natural Resources Canada, Ottawa, ON, Canada
2. Environment and Natural Resources, GNWT, Yellowknife, NT, Canada
3. Tlicho Government, NT, Canada
4. Wek'èezhìi Renewable Resources Board, Yellowknife, NWT, Canada
* Presenting Author: Wenjun Chen, Research Scientist, (613) 759-7895, [email protected]
ABSTRACT
On the one hand, resource development is a cornerstone of economy in Canada’s North. On the other
hand, ccaribou are important in economy, culture, health, and way of life of northern aboriginal peoples.
Balancing resource development and caribou conservation is thus a top priority in northern governance
and decision making.
Caribou are affected by many factors such as habitat, harvest, predators, diseases and parasites, extreme
weather, climate change, industrial development, and pollution. Their impacts on caribou are
compounded in a complex manner spatially and temporally. Assessing the impacts of manmade activities
(e.g., resource development) on caribou is thus a very complex task, as illustrated by the “Science and
Technology Plan for 2014 to 2019” of the Canadian High Arctic Research Station (CHARS), in which
companies have identified the capacity to distinguish project-specific environmental, social, economic,
and health impacts from those that result from broader-scale changes such as global warming or from the
cumulative effects of multiple projects in a region as a critical area for science and technological
development to support resource development.
As a part of NWT CIMP (Cumulative Impact Monitoring Program), we have been working towards a
solution to this challenge. By using historical time series of satellite data from the NOAA AVHRR (Advance
Very High Resolution Radiometer) acquired and pre-processed by Canada Centre for Remote Sensing, we
have estimated the forage availability and quality over the Bathurst caribou summer range since 1985.
Community-based vegetation monitoring results from sites near Wekweeti in 2013-14 were used for
calibrating these estimates. We calculated a summer range cumulative indicator (SRCI) to measure overall
forage capacity by combining forage availability in early summer and late fall with forage quality at the
peak of leaf biomass. By determining the upper envelope curve between SRCI and caribou productivity,
we were able to quantify the impact of habitat changes on caribou, for the first time at the population
level in the world. With SRCI, we explained 54% variation in the late winter calf:cow ratio of Bathurst
caribou since 1985. Using the SRCI-caribou productivity relationship as an analogue, we could potentially
estimate impacts of resource developments on caribou’s productivity and population changes. However,
it should be pointed out that to convert this potential into a scientifically defendable tool, many challenges
needed to be addressed.
158
Spectral Mixture Analysis for Characterizing Tree Species Composition for a
Multisource Vegetation Inventory, Northwest Territories, Canada
Jurjen van der Sluijs 1,3, Ronald J. Hall 2,1, and Derek R. Peddle 1*
1. Department of Geography, and Alberta Terrestrial Imaging Centre (ATIC),
University of Lethbridge, Lethbridge, AB., Canada
2. Natural Resources Canada, Canadian Forest Service, Edmonton AB., Canada
3. Government of the Northwest Territories, NWT Centre for Geomatics, Yellowknife NT., Canada
[email protected], [email protected], [email protected]
* Presenting Author: Derek R. Peddle, [email protected]
ABSTRACT
Forest resource management and conservation in the Northwest Territories (NWT) requires land-cover
and tree species identification and to support initiatives such as the NWT Biomass Energy Strategy and
the Action Plan for Boreal Woodland Caribou conservation. Satellite remote sensing provides a costeffective and time-efficient way to obtain this information for the large, remote, and inaccessible northern
boreal forests. However, satellite signals are highly mixed due to the presence of tree shadows and
understory vegetation in open canopy forests that are typical in the NWT. Therefore, spectral mixture
analysis (SMA) was used to separate sub-pixel scale components and provide a better focus on deriving
forest information. SMA of Landsat TM imagery was used to classify four conifer and deciduous forest
species, which is an important information need yet a considerable challenge in low-density northern
forests. Our first classification product derived leading (or dominant) tree species information for each
pixel with 72% accuracy validated over 48 ground plots. Stakeholders indicated a specific need for tree
species composition estimates (relative abundance of all tree species) at the scale of the forest stand, and
this will be the focus of this presentation in terms of how species composition was determined, as well as
its integration into an existing geospatial forest inventory system. The validation indicated that the imagederived species composition was within 20% and 30% of field measured species proportions for 56% and
73% of the 48 field plots, respectively. Work is continuing to improve the species composition estimates
through updated image classifications and filtering of erroneous pixels.
159
Terrestrial Ecosystems Monitoring for the Next Decade: Measurement Needs,
Technologies and Approaches, with Perspectives on Northern Applications
Forrest G. Hall 1*, Scott J. Goetz 2
1. Senior Research Scientists, NASA GSFC/UMd/Joint Center For Earth System Technology
UMBC JCET, c/o NASA GSFC (618.0), Greenbelt, MD, USA. (301) 614-6659. [email protected]
2. Woods Hole Research Center, 149 Woods Hole Road, Falmouth, MA 02540, USA (508) 444-1530
[email protected]
* Presenting Author: Forrest G. Hall, Senior Research Scientist, (301) 614-6659.
[email protected]
ABSTRACT
This paper summarizes the outcomes of deliberations resulting from a two and one-half day workshop, at
the Goddard Space Flight Center in October 2014. The workshop engaged the relevant terrestrial
ecosystem, carbon cycle, landcover/landuse change and biodiversity science communities to describe and
prioritize measurement needs. Participants included 50 scientists from a range of Earth science research
institutions as well as government agencies including NASA, USGS, NSF, NOAA and the DOE.
NASA relies on the Earth science community to identify and prioritize societally relevant scientific
questions and the observations required to address them. The workshop addressed key objectives
articulated in the 2010 NASA Earth science plan: “How is the Earth changing and what are the
consequences for life on Earth?” NASA seeks to characterize, understand and predict the current and
future state of global Earth systems by addressing three critical science questions; (1) How is the global
Earth system changing? (2) What are the sources of change in the Earth system and their magnitudes and
trends? (3) How will the Earth system change in the future?
Measurement requirements for the next decade were addressed in the context of existing data records
and measurement technology (satellite, aircraft and ground) and feasible future technology and
measurement approaches to acquire and manage the data collections. A white paper articulating the
societal issues, science questions, analysis framework and prioritized measurement needs was published
and placed online. Major workshop conclusions will be presented, including perspectives related to
northern applications.
160
1981 to 2014 Circum-Arctic Tundra Phytomass Production
Compton Tucker1*, Jorge Pinzon2, Donald Walker3, Howard Epstein4, Martha Reynolds5, Umma Bhatt6,
and Gerald Frost7
1. Physical Scientist, Earth Science Division, NASA/Goddard Space Flight Center, Greenbelt, Maryland 20771 USA 1
301 614-6644 <[email protected]>
2. Applied Mathematician, Earth Science Division, NASA/Goddard Space Flight Center, Greenbelt, Maryland 20771
USA 1 301 614-6687, <[email protected]>
3. Environmental Biologist, Institute of Arctic Biology, University of Alaska, 311 Irving St., P.O. Box 757000,
Fairbanks, AK 99775 USA, 1 907 474-2459,<[email protected]>
4. Ecologist, Department of Environmental Sciences, University of Virginia, Charlottesville, VA 22904-4123 USA, 1
434-924-4308,<[email protected]>
5. Arctic Ecologist, Institute of Arctic Biology, University of Alaska, 311 Irving St., P.O. Box 757000, Fairbanks, AK
99775 USA, 1 907-474-6720, <[email protected]>
6. Atmospheric Scientist, Geophysical Institute, University of Alaska, 903 Koyukuk Drive, Fairbanks, AK 99775 USA, 1
907 474-2459, 1 907 474-7558, <[email protected]>
7. Environmental Ecologist, ABR Inc., P.O. Box 80410, Fairbanks AK, 99708, 907 455-6777,<[email protected]>
* Presenting Author: Compton Tucker, Senior Earth Scientist, 1 301 614-6644,
<[email protected]>.
ABSTRACT
We describe 34 years of non-stationary photosynthetic capacity 8-km data from the AVHRR instruments
from 1981 to 2014 and 15 years of MODIS 250-m data from 2000 to 2014, both covering the entire
circum-polar Arctic Tundra. We couple these data with ground-sampled total dry photomass from sites
in Eurasia and North America to express our spatially continuous satellite data in units of plant mass. We
observed peak tundra phytomass production was still relatively high in 2013 and 2014 for North America
and the Arctic as a whole, indicating a continued trend in increasing vegetation productivity, consistent
with a longer growing season and warmer summers. We also observe some interesting trends in both
Eurasia and the Arctic as a whole, with time-integrated photosynthetic capacity values that suggest
increased snow extent in these areas, perhaps delaying the start of the tundra growing season. Since
1999 we also observed increased areas of Arctic photosynthetic capacity decreases, suggesting tundra
water limitations as the growing season progresses. We will describe these results and how we use
MODIS 250 m data from both Aqua and Terra to confirm our 8 km AVHRR data results from the last
fifteen years.
161
S32 Urban / Transportation / S32 Milieu urbain et systèmes de
transport
Elizabeth Simms, chair / modérateur
Automatic Extraction of Highway Light Poles from Mobile LiDAR Point Cloud
Data
Wai Yeung Yan1, Salem Morsy2*, and Ahmed Shaker3
1. Postdoctoral Fellow, Department of Civil Engineering, Ryerson University, Toronto, Ontario, Canada,
[email protected]
2. Ph.D. Student, Department of Civil Engineering, Ryerson University, Toronto, Ontario, Canada,
[email protected]
3. Associate Professor, Department of Civil Engineering, Ryerson University, Toronto, Ontario, Canada,
[email protected]
* Presenting Author: Salem Morsy, Ph.D. Student , 416-979-5000 [email protected]
ABSTRACT
Maintaining a comprehensive 3D road database is one of the tasks required by the local/municipal
transportation authority for inventory purpose. The resulted database can also serve as an input for
building a virtual 3D scene to facilitate driving simulation and intelligent vehicle navigation. This study
presents a workflow for automatic extraction of light poles and towers from mobile LiDAR data point
clouds collected for a section of highway 401 located in Toronto, Ontario, Canada. A ground filtering
algorithm is first implemented to separate the above-ground features from the ground based on a
statistical analysis – skewness and slope-based change mechanism. The process results in a layer of aboveground features with normalized height, and this layer is subsequently being used to perform
unsupervised clustering where the density-based spatial clustering of applications with noise (DBSCAN)
algorithm is implemented. The clustered point clouds of the above-ground features are then used to find
out the light poles and towers based on certain decision rules. With reference to the specification of
highway/street lighting, a set of decision rules is built based on the dimension in order to delineate
different types of lighting equipment situated along the highway. The results show that the proposed
method can achieve over 95% of detection rate for different types of light poles and towers along the
highway 401.
162
Early Aerial Photography in Canada after the Great War – The Case of the
Halifax Air Survey Mission in 1921
Dirk Werle1*
1. Partner / Geoscientist, Ærde Environmental Research, 19 Forward Avenue, Halifax, NS, Canada
Telephone: (902) 423-2211, E-mail: [email protected]
* Presenting Author: Dirk Werle, Partner / Geoscientist, Tel. (902) 423-2211, [email protected]
ABSTRACT
The paper presents research into the military and civilian history, technological development, and
practical outcomes of aerial photography in Canada immediately after the First World War. The collections
of early Canadian aerial photography, as well as the institutional and practical circumstances and
arrangements of their creation, represent not only an important part of Canada's remote sensing heritage
but it also contains valuable information for thematic inquiries and mapping exercises today.
An episode of one of the first urban surveys, carried out over Halifax, Nova Scotia, in 1921, is highlighted
and an air photo mosaic and interpretation is presented. Using the almost one hundred year old air photos
and a digitally re-assembled mosaic of a substantial portion of that collection as a guide, a variety of
features unique to the post-war urban landscape of the Halifax peninsula are analyzed, illustrated and
compared with current land use.
The pan-chromatic air photo ensemble at a nominal scale of 1:5,000 is placed into the historical context
with contemporary thematic maps, ground-based photography, modern satellite imagery, as well artists’
representations, which – similar to the 1921 air photos – have preserved urban streetscapes that no
longer exist. Further research opportunities concerning outcome and present day applications of early
Canadian aerial photography are discussed.
163
Multi-temporal Urban Growth Assessment in Asia Pacific Region using Time
Series of Pixel-Based Landsat Imagery Composites
Yuhao Lu1*, Nicholas C Coops2, and Txomin Hermosilla3
1. Msc Candidate, Integrated Remote Sensing Studio, Department of Forest Resources Management, University of
British Columbia, 2424 Main Mall, Vancouver, B.C., Canada, phone: +1 778-320-3326, e-mail:
[email protected]
2. Professor, Integrated Remote Sensing Studio, Department of Forest Resources Management, University of British
Columbia, 2424 Main Mall, Vancouver, B.C., Canada, phone: +1 604-822-6452, e-mail: [email protected]
3. Postdoctoral Fellows, Integrated Remote Sensing Studio, Department of Forest Resources Management,
University of British Columbia, 2424 Main Mall, Vancouver, B.C., Canada, phone: +1 604-822-6456, e-mail:
[email protected]
* Presenting Author: Yuhao Lu, Msc Candidate, +1 778-320-3326, [email protected]
ABSTRACT
Human modifications in the terrestrial biosphere have intensified over the past several decades. Covering
3% of the Earth’s land surface yet consuming nearly 75% of global primary energy, the urban environment
comprises some of the world’s most profound anthropogenic impacts. Satellite imagery and other forms
of remote sensing have been utilized to monitor and understand spatial and temporal changes of the
urban environment in terms of its physical attributes such as built-up areas, and green space. Yet
challenges remain due to the rapidity of city development especially in areas in highly populous regions,
such as the Asia-Pacific, spurring need for improved monitoring methods and seamless image data
archives. In this paper, we demonstrate an approach to produce a near-annual and gap-free Landsat time
series from 1984 to 2012 at 30m spatial resolution. These annual image composites are then used to
examine changes in urban density and greenspace in 3 cities in the Asia-Pacific region. To do so we
quantify both the trends and interactions between urban expansion, as represented by the Normalized
Difference Built-up Index (NDBI), and urban green spaces dynamics, as detected using the Normalized
Difference Vegetation Index (NDVI). Results indicate there is significant potential of gap-filled Landsat
imagery to observe changing in urban dynamics, even in areas with high cloud cover, and urban expansion
in the focus cities follows known models of geographic urban development. Future work will expand the
number of cities and utilize other remote sensing data to provide additional socio-economic context.
164
Infrastructure Monitoring in Urban Areas using Space Based SAR
Joseph Chamberland1*, Pierre-Jean Alasset2 and John Mulvie3
400 March Road, Suite 210, Ottawa, Ontario, Canada K2K 3H4
1*. Director of Ottawa Operations, 1 (613)-592-7700 ext. 232, [email protected].
2. Senior EO Scientist, 1 (613)-592-7700 ext. 227, pierre-jean.alasset@c-core.
3. Geomatics Analyst, 1 (613)-592-7700 ext. 231, [email protected].
* Presenting Author: Joseph Chamberland, Director of Ottawa Operations, 1 (613) 853-9419,
[email protected] address.
ABSTRACT
The Ottawa City Council approved and awarded the contract for construction of the Confederation Line
through downtown Ottawa in December 2012. This $2 billion investment into transportation
infrastructure features a 12.5km light rail span, including a 2.5km tunnel through the downtown area. To
monitor construction effects, the Ottawa Light Rail Constructors (OLRT) have included extensive GPS and
leveling and total stations along the tunnel corridor. They have also worked with C-CORE to define a space
based monitoring program using Synthetic Aperture Radar and interferometric processing techniques.
The information extracted from the space borne SAR data is used to ensure infrastructure stability of key
regions of quick clay (known as leda clay) surrounding the excavation corridor.
By using the Small Baseline Subset interferometric SAR (InSAR) technique and frequent SAR data
acquisitions using the Italian COSMO-SkyMed constellation, a displacement map of the downtown area
was available to the end-users within 4 months of project start. During the continued excavation phase,
this map is updated monthly with new acquisitions. The accuracy of the displacement has been evaluated
to be within the 3-5mm range and displacement information is measured on city infrastructure, such as
fences and buildings, with sufficient coverage that a risk assessment can be made for key buildings in the
leda clay zone.
This talk will focus on the OLRT monitoring requirements and show how an InSAR monitoring program is
meeting these requirements and is a suitable tool for monitoring of city infrastructure.
165
Analysis of Aerodynamic Roughness Using SAR Data in Urban Areas
Fengli Zhang1, Qianjun Zhao2, Minmin Sha3, Yun Shao4*
1. Associate Professor, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101,
China, 0086-10-64838047, [email protected]
2. Professor, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101, China,
0086-10- 82178013, [email protected]
3. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101, China, 0086-1064838047, [email protected]
4. Professor, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, 100101, China,
8610-648763138047, [email protected]
* Presenting Author: Tun Shao, Professor, 8610-648763138047, [email protected]
ABSTRACT
Urban areas are roughest and most complex among all aerodynamic boundaries. Buildings and trees in
urban areas dramatically increase drag and reduce wind speed, and there are great variations for the
distribution of these roughness elements within urban areas. Thus estimation of aerodynamic roughness
length z0 within urban areas is very important for urban climate studies, air pollution control and urban
planning.
Backscattering received by Synthetic Aperture Radar (SAR) sensor is very sensitive to the composition and
structure of the roughness elements, thus it maybe helpful to aerodynamic roughness estimation in urban
areas. This paper analyzed the correlation of backscattering coefficient σ0 and the aerodynamic roughness
taking Beijing city as an example. Land use and land cover of Beijing have changed greatly since 1980s,
with great spatial variations. Beijing 325-meter high meteorological tower records 15-layer wind velocity
and wind direction using anemometers mounted in this tower. It is a member of Urban FluxNet and plays
an important role in aerodynamic process studies in Beijing.
Totally 11 scenes of ALOS PALSAR images obtained during 2010 was used for experiments. These SAR
images were first processed with radiometric correction, geometry correction and co-registration. Field
aerodynamic roughness z0 was calculated using the fitting iteration method according to wind data
obtained by the meteorological tower. The fetch area was defined as a sector area with the central line
same to the wind direction. Then backscattering coefficients σ0 within the fetch area were extracted.
Experiments show that there is great consistency between backscattering coefficient extracted from SAR
data and aerodynamic roughness measured in field.
166
S33 Synthetic Aperture Radar / S33 Radar à synthèse d'ouverture
Tom Lukowski, chair / modérateur
Corner Reflectors in Canada: Challenges and Solutions.
Kevin Murnaghan1* and Vern Singhroy2
1. Remote Sensing Scientist, Natural Resources Canada, 560 Rochester St., Ottawa, Ontario, Canada,
1(613)759-6237, [email protected]
2. Research Scientist, Natural Resources Canada, 560 Rochester St., Ottawa, Ontario, Canada, 1(613)759-1047,
[email protected]
* Presenting Author: Kevin Murnaghan, Remote Sensing Scientist, (613)759-6237,
[email protected].
ABSTRACT
Purpose built artificial targets have been used in Synthetic Aperture RADAR for calibration, geolocation,
and InSAR since its inception. The trihedral corner reflector is commonly used due to the wide symmetric
beam pattern of reflection, ease of construction, and reasonable size.
There are many challenges between a theoretical corner reflector and using them in the field. The four
main areas of concern are mechanical design, support of the reflector, orientation and the
longevity/stability of the deployment. The design of the reflector must ensure flatness and orthogonality
of the surfaces and be sized correctly. The support of the reflector is determined by the ground
conditions, accessibility, and tools/machinery to be used. The orientation can be estimated or accurately
calculated using Canada Centre for Remote Sensing (CCRS) software Target to Satellite Pointing (TSP). The
longevity of the deployment of the radar reflector is affected by the types of damage that it may be
subjected to such as snow and ice buildup, rock impacts, vegetation, animals, or human interference.
Examples of deployments will be shown from several past and current projects across Canada including:
airborne SAR 580 calibration, geohazard InSAR monitoring of pipeline and transportation corridors,
wetland monitoring, and carbon sequestration testing. Issues and solutions will provide guidance to help
new users in preparation for RCM where InSAR and CCD techniques will be more operational.
167
RADARSAT-2 Progress Report
David Belton1* and Warren Cartwright 2
1. Vice President, MDA Geospatial Services Inc., 13800 Commerce Parkway, Richmond, BC, Canada, (604) 231-2736
2. Product Manager, MDA Geospatial Services Inc., 13800 Commerce Parkway, Richmond, BC, Canada, (604) 2312032, [email protected]
* Presenting Author: David Belton, Vice President, MDA Geospatial Services Inc., 13800 Commerce
Parkway, Richmond, BC, Canada, (604) 231-2736, [email protected]
ABSTRACT
RADARSAT is one of the most successful and important commercial Synthetic Aperture Radar (SAR) Earth
observation programs, having contributed data, products and services to the global Earth Observation
community since 1995. MDA has been deeply involved in all aspects of the program, having designed,
built and operated all of the missions to date.
RADARSAT-1 was launched in November 1995 with a five-year design life, which it exceeded by over 13
years, when it reached its end of life in April 2013. RADARSAT-2, the current mission, was the successful
follow-on, and was launched in December 2007. It provides continuity into the RADARSAT Constellation
Mission (RCM). RCM is a constellation of three SAR Earth observation satellites, and is scheduled to launch
in 2018.
At the 7-year anniversary of the start of operations of RADARSAT-2, we provide the Earth observation
community with an update on the status of the mission. This presentation provides a status summary of
the satellite, system, and operations performance, describes some highlights of recent challenges,
changes, and achievements, and outlines some enhancement work in progress and in future plans for
RADARSAT-2.
In addition, the presentation will provide an overview of the plans for the launch and operations of the
RADARSAT Constellation Mission, and how RCM will provide data continuity for the users of RADARSAT-1
and RADARSAT-2. The RADARSAT Program has a bright future ahead of it, and we encourage you to join
us as we explore the past, present and future of the mission.
168
Use of RADARSAT-2 and ALOS-PALSAR images for wetland mapping in New
Brunswick
Armand LaRocque1*, Brigitte Leblon2, Renata Woodward3, Nancy French4 and Laura Bourgeau-Chavez5
1. Research associate, Faculty of Forestry and Environmental management, University of New Brunswick,
Fredericton (NB), Canada, 506-453-4932, [email protected]
2. Professor, Faculty of Forestry and Environmental management, University of New Brunswick, Fredericton (NB),
Canada, 506-453-4924, [email protected]
3. Executive director, Nature Trust of New Brunswick, Fredericton (NB), Canada, 506-457-2398,
[email protected]
4. Senior research scientist, Michigan Tech Research Institute, Ann Arbor, Michigan, U.S.A., 734-913-6844,
[email protected]
5. Research scientist, Michigan Tech Research Institute, Ann Arbor, Michigan, U.S.A., 734-913-6873,
[email protected]
* Presenting Author: Armand LaRocque, Research Associate, 506-453-4932, [email protected]
ABSTRACT
This study used different combinations of LANDSAT-5 TM images with RADARSAT-2 C-band and ALOSPALSAR L-band SAR images for mapping wetland areas in New Brunswick. The resulting maps were
compared to GPS field data as well as the two wetland maps currently in use by the Province of New
Brunswick, namely the Department of Natural Resources (DNR) wetland and forested wetland maps.
Whatever the image combination we used (LANDSAT + RADARSAT-2 and/or ALOS-PALSAR), the overall
accuracy for the image classification is always higher than 90%, with a maximum of 93.9%. In addition,
the number of correctly identified sites is higher with the ALOS-PALSAR-based classified image (91.1%) or
the RADARSAT-2-based classified image (88.4%) than with the DNR maps (44.5%). From the classified
images, the few misclassifications are due to wetland sites classified in another wetland class. For the DNR
maps, about half of them are associated to wetland sites that not being mapped, the remaining half being
wetland sites that are not classified in the right wetland class. The classified images show that the
RADARSAT-2 C band is more sensitive to detect marshes than the ALOS-PALSAR L band, but this is the
opposite for the detection of the shrub wetlands. Therefore, all the SAR images are complementary and
their combination gives the best ground-truth identification accuracy (98.6%). The study was funded by
two NB Environmental Trust Fund grants and supported by a NASA Interdisciplinary Science Program
grant. The RADARSAT-2 images were provided under a Canadian Space Agency SOAR grant awarded to
the Province of New Brunswick.
169
A Speckle Filtering Approach of SAR Data Based on a Coherent Decomposition
Sahel Mahdavi1*, Yasser Maghsoudi2, Bahram Salehi3
1. Master of Electrical Engineering, Faculty of Engineering and Applied Science, Memorial University of
Newfoundland, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL, Canada, (709) 864 6701,
[email protected]
2. Assistant Professor, Department of Remote Sensing, K. N. Toosi University of Technology, ValiAsr Street,
Mirdamad Cross, Tehran, Iran, +982188770218, [email protected]
3. Remote Sensing Engineer, C-CORE/LOOKNorth and Cross-appointed professor, Faculty of Engineering and
Applied Science, Memorial University of Newfoundland, Capt. Robert A. Bartlett Bldg, Morrissey Rd, St John's, NL,
Canada, (709) 864 6701, [email protected]
* Presenting Author: Sahel Mahdavi, Master of Electrical Engineering, (709) 864 6701,
[email protected]
ABSTRACT
Speckle degrades the radiometric resolution of the image and complicates the visual interpretation of
RADAR data. Speckle reduction is an important topic within radar remote sensing community and
significant amount of research has been conducted on it. One strategy for speckle reduction is to apply
the decomposition to the image followed by filtering of the decomposed image based on the dominant
scattering mechanism. This strategy has been previously used for incoherent decomposition. Incoherent
decompositions, however, are effective when applied to the filtered image (i.e. after speckle reduction).
In this paper, we propose a speckle reduction approach based on the Pauli decomposition which is a
coherent decomposition method. This way, pixels with same dominant scattering mechanism within a
window are averaged out and speckle noise is filtered in an effective way. The proposed method is, in
particular, effective when applied to synthetic aperture radar (SAR) images of coherent targets such as
dominant man-made structure of urban areas. We tested the method on a RADARSAT-2 image of an
urban area over San Francisco and results showed that as the size of the filtering window increases, the
proposed approach preserves more details compared to the standard average filter.
170
Mapping The C-Band Vertical Backscatter Pattern Through A Conifer Forest
Canopy
Keith Morrison1*, Svein Solberg2, and John C. Bennett3
1. Radar Group, Cranfield University, Shrivenham, Swindon SN6 8LA, UK, +44 1793 785050,
[email protected]
2. Norwegian Forest and Landscape Institute, Pb 115, NO-1431, Ås, Norway, [email protected]
3. Consultant, retired, Sheffield, UK
* Presenting Author: Dr. Keith Morrison, Head of GB-SAR Laboratory, +44 1793 785050,
[email protected]
ABSTRACT
Many issues remain to be understood on the interaction of a radar wave with a forest canopy and retrieval
of forest parameters. Forest biomass is one of the most significant and some techniques seek to determine
it from stand height. Synthetic aperture radar (SAR) interferometric techniques use a pair of images
slightly displaced from each other in order to determine an apparent ‘effective’ scattering height of a
canopy, but it provides no detail on the vertical backscatter distribution within the target’s volume. More
recently, SAR polarization coherence tomography is an attempt to provide a 2D vertical profile of
backscatter through a canopy using as little as a single interferometric image pair. However, the retrieval
algorithm is model-based, relying on a priori assumptions about the backscattering pattern.
Tomographic profiling (TP) is a recent innovation, and the presented result has many similarities to the
final image product from SAR tomographic schemes, but requiring only a single-pass collection. Cranfield
University’s portable Ground-Based SAR (GB-SAR) system acquired TP C-band imagery of the vertical
polarimetric backscattering patterns through a conifer forest canopy. Each TP scan simultaneously
captured along-track image transects over the incidence angle range 0-60 degrees. HH, VV backscatter
profiles were very similar through the canopy and 6dB above the cross-polarised return. This likely
indicates the radar likely saw the canopy volume as a random collection of scatterers. Backscattering
profiles showed close agreement with branch biomass distribution through the canopy. Equivalent
interferometric tree heights were estimated from the centre-of-mass of the backscatter-height
distributions.
171
Posters / Affiches
Web-based applications for LiDAR data processing and visualizing trees at the
plot level
Carlos A. Silva1*, Andrew T. Hudak2, Lee A. Vierling3, E. Louise Loudermilk4, Joseph J. O’Brien5
1. Ph.D. Student. Department of Forest, Rangeland, and Fire Sciences, University of Idaho, 875 Perimeter Drive,
Moscow, ID -83843, USA, +1 (208) 596- 4510 , [email protected]
2. Research Forester. USDA Forest Service, Rocky Mountain Research Station, RMRS, 1221 South Main Street,
Moscow, Idaho- 83843, USA, +1 (208)-883-2327, [email protected]
3. Associate professor. Department of Forest, Rangeland, and Fire Sciences, University of Idaho, 875 Perimeter
Drive, Moscow, ID 83843, USA, +1 (208) 885-5743, [email protected]
4. Research Ecologist. USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green St., Athens, GA 30602, USA-, +1 (706) 559 4309 , [email protected]
5. Research Ecologist. USDA Forest Service, Southern Research Station, Center for Forest Disturbance Science, 320
Green St., Athens, GA 30602, USA-, +1 (706) 559-4336, [email protected]
* Presenting Author: Carlos Alberto Silva, Ph.D. student, (208) 596-4510,
[email protected].
ABSTRACT
LiDAR remotely sensed data have been a focus for individual tree attribute extraction for forest inventory
and modeling. However, there are few specific software packages available to process LiDAR data
efficiently in terms of individual tree detection and extraction. Programming skills are required to create
algorithms to perform these tasks. Yet, it is hard to find freely available routines and/or codes. Therefore,
we present some simple, free, and publicly available online tools to visualize and process small (plot-scale)
LiDAR datasets. We present the Web-LiDAR suite, currently comprised of eight web-based applications
for LiDAR data processing and visualization at the individual tree level: LiDARTreeTop for individual tree
detection from the LiDAR-derived canopy height model (CHM); LiDARTreeExtractor, LiDAR3DclusterTree,
and LiDAR3DClipTree for individual tree extraction from the 3D LiDAR point cloud; LiDARAlphaShape3D
for individual canopy volume computation; and LiDARstandModel3D, LiDARstand3D, and LiDARpine3D for
virtual tree visualization. The Web-LiDAR applications have three major tab panels. The first one is a short
presentation of the tools (i.e. “Welcome panel”); the second is the main page of the application (i.e.
“Application panel”), where the users can visualize and process their own LiDAR datasets; and the last one
(“About panel”) is the tab panel that describes the web-LiDAR application and presents a tutorial in both
pdf and a youtube link to help users better understand how to use the tool. The links to access the WebLiDAR applications suite are presented below:
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDARTreeTop/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDARTreeExtractor/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDAR3DclusterTree/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDAR3DClipTree/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDARAlphaShape3D/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDARstandModel3D/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDARstand3D/
http://forest.moscowfsl.wsu.edu:3838/csilva/LiDARpine3D/
172
Generalizing LIDAR-based, Species-level Biomass Predictions in the Northern
Rocky Mountain Ecoregion
Patrick Fekety1*, Michael Falkowski2, Andrew Hudak3
1. Research Fellow, University of Minnesota, Department of Forest Resources, St. Paul, MN, USA, 208.883.2326,
[email protected]
2. Research Associate Professor, University of Minnesota, Department of Forest Resources, St. Paul, MN, USA,
[email protected]
3. Research Forester, US Forest Service, Rocky Mountain Research Station, Moscow, ID, USA, 208.882.3557,
[email protected]
* Presenting Author: Patrick Fekety, Research Fellow, 208.883.2326, [email protected]
ABSTRACT
Airborne scanning LiDAR data are becoming a valuable tool for forest management. Numerous studies
have demonstrated that project-level prediction models using LiDAR data can be developed to create wallto-wall estimates of forest attributes such as biomass, basal area, and volume. Combining spatially
disparate LiDAR datasets is a promising approach for developing a generalized model to predict biomass
across large areas such as ecoregions. In this study, we explored the potential of combining four spatially
disparate LiDAR collections across the Northern Rocky Mountain Ecoregion, USA. We leveraged field plot
measurements, LiDAR-derived structural metrics, topographic metrics, and climate data to develop a
random forest-based imputation model to predict aboveground biomass, dominant species, and biomass
of the dominant species across multiple spatial extents. A PCA-convex hull analysis demonstrated that
structural conditions among the project areas were similar. Additionally, a “leave-one-study-area-out”
analysis provided evidence that the imputation model is spatially transferable within the ecoregion.
Preliminary results show the general model had R2 values of 61% and 47% for aboveground biomass and
biomass of the dominant species, respectively. The model had a 41% success rate at identifying the
dominant species, but had difficulty distinguishing among plots dominated by Douglas-fir, grand fir, and
western redcedar. We expect the results of this study will be useful for large-scale planning of forest
resources. The results can be used to parameterize climate sensitive forest dynamics models such as
LANDIS-II and Climate-FVS to simulate forest responses to a changing climate.
173
Hyperspectral Non-linear Denoising with High Performance Parallel Computing
Aimin Guan1* and David Goodenough2
1. Masters Student, Department of Computer Science, University of Victoria, 3800 Finnerty Road, Victoria, BC,
Canada, 250-885-5333, [email protected]
2. Adjunct Professor, Department of Computer Science, University of Victoria, 3800 Finnerty Road, Victoria, BC,
Canada, 250-516-5157, [email protected]
* Presenting Author: Aimin Guan, Masters Student, 250-885-5333, [email protected]
ABSTRACT
Hyperspectral remote sensing imagery provides high spectral resolution, which enables forest species
recognition and forest chemistry mapping for large area forest monitoring. Hyperspectral data noise
reduction is an important image pre-processing step and can improve classification and chemistry
accuracy. Hyperspectral image data are nonlinear, but in hyperspectral imagery processing, they are often
modeled using linear stochastic processes based on the assumption of linearity. It is also assumed the
noise and signal are independent. Therefore, commonly used de-noise algorithms deal with the noise
based on linearity. Han and Goodenough [1] demonstrated the existence of nonlinearity in a 4-m
hyperspectral Airborne Visible/Infrared Imaging Spectrometer image acquired over an area of coastal
forests on Vancouver Island. A nonlinear noise reduction algorithm was proposed and implemented. The
algorithm efficiently reduces noise and boosts image signal to noise ratio. The algorithm is
computationally very expensive and demands fast processing solutions. In this paper, we offer a
framework which we have implemented and evaluated in high performance parallel computing
environment. This frame work can be adapted to other remote sensing applications.
Reference:
1. Han, Tian and David G. Goodenough 2008, "Investigation of Nonlinearity in Hyperspectral Imagery
Using Surrogate Data Methods," IEEE Transactions on Geoscience and Remote Sensing, Vol. 46,
No. 10, pp. 2840-2847.
174
Working with DEMs of varying resolution and derivation
Shane Furze1*, Jae Ogilvie2, Paul Arp3
1. PhD Candidate, Faculty of Forestry and Environmental Management, University of New Brunswick, 28 Dineen
Drive, Fredericton, New Brunswick, Canada, [email protected]
2. Research Associate, Faculty of Forestry and Environmental Management, University of New Brunswick, 28
Dineen Drive, Fredericton, New Brunswick, Canada, [email protected]
3. Professor, Faculty of Forestry and Environmental Management, University of New Brunswick, 28 Dineen Drive,
Fredericton, New Brunswick, Canada, [email protected]
* Presenting Author: Shane Furze, PhD Candidate, 506-451-6823, [email protected],
ABSTRACT
This presentation focuses on improving already existing digital elevation layers such as, e.g., satellite
generated data (SRTM, ASTER), photo-grammetrically derived data, and traditional contour elevation data
through calibration with LiDAR-derived bare-earth elevation data sets. Doing so provides at least three
advantages: (i) a systematic reduction in DEM artifacts such as “ridging” and errors in elevation
registration, (ii) the improvements apply for areas that do not as of yet have LiDAR coverage; (iii) for areas
with LiDAR coverage, the improvements show whether there have been significant changes in elevation
in the time interval between obtaining the original elevation data and the LiDAR data. Preliminary results
indicate a 30% reduction of elevation error to within ± 1m. Since many of the non-LiDAR DEM derivations
registered canopy top elevations rather than bare-earth elevations, there is also the added opportunity
to obtain changes in forest height that are specific to the interim time lapse. This presentation illustrates
some of the results obtained, and how the results enable and improve research and operations planning
in wetland delineations, flood risk mapping, soil-remapping, and soil trafficability and forest productivity
mapping.
175
Buried Geo-Hazard Imaging and Mapping on the Grand Banks using an Acoustic
CorerTM
Ryan Laidley1*, Ian McDermott2, Alison Brown3, Jacques Guigne4
1. Geoscientist, PanGeo Subsea Inc., 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032,
[email protected]
2. Lead Geoscientist, PanGeo Subsea Inc., 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032,
[email protected]
3. Geoscientist, PanGeo Subsea Inc., 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032,
[email protected]
4. Chief Scientist, PanGeo Subsea Inc., 430-434 Water St., St. John’s, Newfoundland, Canada, 709-739-8032,
[email protected]
* Presenting Author: Ryan Laidley, Geoscientist, 709-739-8032, [email protected]
ABSTRACT
The main objective of an Acoustic Corer™ survey is to interrogate the sub-seabed to identify any buried
geo-hazards and to delineate any stratigraphy. The Acoustic Corer™ is an acoustic acquisition tool that
transmits and receives with dual subsurface scanning sonar packages on a 12 meter boom that rotates
180o. The received signals are then processed using synthetic aperture sonar (SAS) algorithms, a technique
that generates the effect of a large transmit-receive aperture by signal processing means rather than by
actual use of a large array. By coherently summing together the returns, the Acoustic Corer™ is able to
clearly and accurately identify geohazards.
PanGeo Subsea was commissioned to undertake an Acoustic Corer™ geo-hazard survey at a site off the
east coast of Newfoundland, on the Grand Banks. Four over-lapping Acoustic Corer™ surveys were
completed at the site and the results were interpreted. The survey revealed a sub-seabed with an acoustic
character consistent with till. There were numerous acoustic anomalies identified consistent with geohazards exceeding 0.5 m in diameter. Thirteen anomalies were suggestive of isolated boulders and 8
anomalies were identified as clusters of boulders and cobbles. The over-lapping surveys were correlated
together to increase confidence in the results. The conclusion of this exercise was that there was good
agreement in both acoustic character and positioning of anomalies identified in the overlapping Acoustic
Corer™ surveys. This reaffirms the initial independent interpretation of individual Acoustic Corer™ surveys
and increases the confidence in these data sets.
176
Using ArcGIS as an Alternative or Complimentary Tool to ENVI for Extracting
Anthropogenic Footprints from Multispectral Imagery
Dennis Chao1*, Subir Chowdhury2 and Todd Shipman3
1. GIS Specialist, Alberta Energy Regulator/Alberta Geological Survey, 4th floor, 4999 – 98 Ave, Edmonton, Alberta,
Canada, (780) 427-0107, [email protected]
2. Remote Sensing Specialist, Alberta Energy Regulator/Alberta Geological Survey, 4 th floor, 4999 – 98 Ave,
Edmonton, Alberta, Canada, (780) 427-4115, [email protected]
3. Manager, Alberta Energy Regulator/Alberta Geological Survey, 4 th floor, 4999 – 98 Ave, Edmonton, Alberta,
Canada, (780) 644-5563, [email protected]
* Presenting Author: Dennis Chao, GIS Specialist, (780) 427-0107, [email protected]
ABSTRACT
Alberta Geological Survey/Alberta Energy Regulator is investigating how to manage rapid growth within
the oil and gas industries though out Alberta. We are utilizing satellite earth observation data such as
multispectral imagery to extract anthropogenic footprints relating to exploration activities, such as roads,
well pads, processing facilities and access corridors to pump sites, etc. Our study area is about 200km
northwest of Edmonton, near Fox Creek where industries are producing and exploring both conventional
and unconventional oil and gas. All information is geospatial and integration of this information
contextually will help to develop better decision making.
Publicity available Landsat8 multispectral imagery (2013 and 2014) from US Geological Survey and high
resolution (6m) multispectral SPOT6 imagery for 2013 are used for this analysis. In addition, BestAvailable-Pixel (BAP) composite of Landsat multispectral image stacks (2005-2012) provided by Natural
Resources Canada.
Landsat8 and SPOT6 imagery are processed as individual scenes using Image Classification functions and
different Spatial Analyst filters in ArcGIS 10.1. Anthropogenic footprints from Landsat 8 imagery are also
extracted using NDVI and co-occurrence-based texture analysis using ENVI.
Results produced by ArcGIS for Landsat8 imagery are similar to ENVI with high degree of accuracy when
identifying footprints bigger than 100m and have difficulties detecting linear footprints such as access
corridors due to image resolution. Traditional supervised classification algorithm in ENVI, on the other
hand, is computationally expensive to process SPOT6 data because high spatial resolution. This problem
is resolved by using ArcGIS Spatial Analyst functions with high detection accuracy. Results are then
processed in ENVI using post classification techniques (i.e., clump and sieve) to further refine footprint
detection.
ArcGIS is viable alternative to dedicated remote sensing package such as ENVI for detecting
anthropogenic footprints from remote sensing imagery. In addition, it can process high resolution
imagery such as SPOT6 more efficiently and to be used for post-processing in ENVI to enhance
classification accuracy.
177
Multispectral Lidar for 3D Land Classification and Topography/Bathymetry
Paul LaRocque1* and Ahmed Shaker AbdElrahman2
1. VP Advanced Technology, Optech Incorporated, 300 Interchange Way, Vaughan, ON, Canada, +1 905 660 0808
(ext. 3308), [email protected]
2. Associate Professor, Ryerson University, 350 Victoria Street, Toronto, ON, Canada, +1 416 979 5000 (ext. 6458),
[email protected]
* Presenting Author: Paul LaRocque, VP Advanced Technology, +1 905 660 0808,
[email protected].
ABSTRACT
Traditional lidar systems employ one or more channels operating at a single wavelength, typically in the
infrared for topographic mapping or green for bathymetry. This design has been very successful at creating
high-accuracy 3D maps, but it has proven difficult to properly classify survey targets with singlewavelength lidar data without support from passive imagers. Furthermore, no one wavelength can
efficiently map both topographic and bathymetric targets, forcing surveyors to either specialize in one
domain or acquire multiple lidar systems.
Both of these issues can be remedied with a multispectral lidar, which employs multiple laser channels
operating at separate wavelengths. By converting the radiometrically corrected intensities from each
channel into a raster format, the user can determine how each target reflects light at different
wavelengths, enabling automated classification of targets with lidar data alone. Furthermore, a
multispectral lidar can employ wavelengths optimal for both topographic and bathymetric mapping,
meaning surveyors can transition seamlessly between both applications as the project requirements
demand.
To realize the potential of this technology, Optech has designed and built the Titan multispectral airborne
lidar. Initial testing of Titan’s classification abilities used responses from all three channels plus the DEM
to create an unsupervised maximum likelihood classifier, which achieved 78% accuracy. Bathymetric
testing shows penetration down to 8 m in Lake Ontario, Canada, with a maximum depth of 15 m expected
in clear water.
178
A Study of Microwave Sea Clutter using a Coherent X-band Radar
Joseph Ryan1*
1.President, Deltaradar Research Incorporated, PO Box 5414, St John’s, NL, Canada, +1(709)800-0786.
[email protected]
* Presenting Author: Joseph Ryan, President, (709)800-0786. [email protected]
ABSTRACT
Microwave marine radar has been used for many years for general purpose navigation and collision
avoidance. Most standard marine radars utilize a non-coherent signal source(magnetron)for generation
of a high power transmit pulse. These radars cannot take advantage of performance improvements that
may be achieved from Doppler processing available when using a coherent signal source and processing.
New radar technology is starting to appear in the commercial market that offers full coherence between
transmitted and received radar signals. These coherent radars utilize solid state transmitters of relatively
low power. They achieve high effective power by using transmit waveform encoding techniques such as
pulse compression and FMCW(Frequency Modulated Continuous Wave). While use of coherent radar for
civil marine applications has begun the understanding of microwave scattering from the ocean is still not
complete and so optimum processing algorithms for target detection and ocean parameter extraction are
not available for these radars.
This paper provides some preliminary results from a baseline study of microwave sea clutter data
collected off the coast of Newfoundland over the past year(2014/2015). The data was collected with a
vehicle mounted coherent X-band radar with horizontal and vertical polarization and radar range
resolutions from 4 to 16 m. Corresponding to the radar data there is in-situ environmental data that
includes sea surface data from a directional wave rider buoy, a current meter and full meteorological
information.
Results of analysis to date indicate the data to be of very good quality and of sufficient variety to permit
the investigation of microwave scattering phenomena for marine applications. Doppler spectra for
horizontal and vertical polarizations have been shown to agree with published literature and indicate a
marked dependence upon wind speed and direction.
179
Delineation of within-field Spatial Variability using high resolution optical
remote sensing data
Sow, A.A.1*, Shang, J.1, Liu, J1, Ma, B1., Jiao, X2, Geng, X.1, Huffman, T.1, Kovacs, J.2, Waters, D.2
1. Research Scientists, Science and Technology Branch, Agriculture and Agri-Food Canada, Ottawa, ON, Canada 960
Carling Avenue, K.W. Neatby Building, Ottawa, Ontario K1A-0C6, Canada. [email protected]
2. Department of Geography, Nipissing University, Canada.
*Presenting Author: A.A. Sow, Research Scientist, 613-759-6930 [email protected]
ABSTRACT
Within-field spatial variability of soil and crop growth condition may indicate variability in crop yield
limiting factors. The delineation of the variability is useful for implementing site-specific management
practices in precision agriculture. Remote sensing data has been used for crop and soil biophysical and
biochemical parameter estimation, therefore are useful for this purpose. This study used multi-temporal
high resolution optical data acquired by the RapidEye 5-satellite constellation for within-field
homogeneous zone delineation using a fuzzy k-means unsupervised classification method. The study site
is a small farm-based community located in the West Nipissing agricultural district, Verner, Ontario. The
soils comprised primarily of azilda clay loam. Production potential is limited by poor drainage, hence much
of the land has been systematically tile drained. Remote sensing and field data were acquired in 2012.
Crop and field parameters extracted from the remote sensing data are used for zone delineation, and the
results of the delineated zones were interpreted according to the estimated crop and field parameters
and the measured field variables for two crop types, spring wheat and canola. The results of the study
show that the delineated zones relate well to plant and soil parameters. This suggests that high-resolution
optical data can provide valuable information in delineating within-field spatial variability and hence
determine the optimum N fertilizer rate to maximize yield of canola and wheat.
180
Modeling Biodiversity Response to Habitat Heterogeneity Using Multi Spatial
and Temporal Remote Sensing Data
Niloofar Alavi1*, Doug King2, Scott Mitchell3, Dennis Duro4, Lenore Fahrig5, Kathryn Linsy6
1. PhD Candidate, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Canada, 613-262-6604,
[email protected]
2. Professor and Chair and Co-director of Geomatics and Landscape Ecology Lab (GLEL), Carleton University, 1125
Colonel By Drive, Ottawa, Ontario, Canada, 613-520-2560, [email protected]
3. Professor and Co-director of Geomatics and Landscape Ecology Lab (GLEL), Carleton University, 1125 Colonel By
Drive, Ottawa, Ontario, Canada, 613 520 2600 x2695, [email protected]
4. Independent Geomatics Consultant, 3144 rue Marcel-Proust, Saint-Augustin-de-Desmaures, Quebec City,
Quebec, Canada, 418-558-5019, [email protected]
5. Professor and Co-director of Geomatics and Landscape Ecology Lab (GLEL) Carleton University, 1125 Colonel By
Drive, Ottawa, Ontario, Canada, 613-520-2600 x 3856 [email protected]
6. Landscape Science Sr. Advisor and Co-director of Geomatics and Landscape Ecology Lab (GLEL), Environment
Canada, National Wildlife Research Centre, 1125 Colonel By Drive, Ottawa, Ontario, Canada, 613-998-7385,
[email protected]
* Presenting Author: Niloofar Alavi, PhD. Candidate, 613-262-6604, [email protected]
ABSTRACT
Habitat heterogeneity is defined as the variation in spatial composition and configuration of different land
cover patches in a given landscape. The habitat heterogeneity hypothesis assumes that heterogeneous
landscapes promote biodiversity by providing available resources and niches for species requirements.
Remote sensing data and techniques have been used to model biodiversity response to habitat
heterogeneity but most studies have relied on single spatial and temporal resolution data. In this study,
Landsat data with moderate spatial resolution and MODIS data with coarse spatial resolution but high
temporal resolution are used to model biodiversity response to habitat heterogeneity in agricultural lands.
The study sites are comprised of 96 agricultural landscapes located within Eastern Ontario. The dominant
crops are corn, soybean and hay, followed by cereals. In a large field study, diversity data for seven taxa
(birds, butterflies, bees, syrphids, plants, carabids and spiders) were collected within 1 km × 1 km extents
at each landscape in 2011 and 2012. Non-crop area was kept relatively constant and only the cropped
areas were sampled. Thus far, biodiversity modelling using metrics such as field size and Shannon Diversity
as predictors have shown significant negative effects of field size on biodiversity. A subsequent result also
showed that a continuous metric derived directly from the image data (e.g., NDVI texture) was better
related to biodiversity than the above two metrics derived from land cover maps. This paper will present
the overall study design and preliminary results of biodiversity modelling using multi spatial and temporal
metrics as predictors.
181
Using an Unmanned Aerial Vehicle and Close Range Photogrammetry to 3D
Model Selkirk College's Castlegar Campus
Peter LeCouffe1*, Kongwen (Frank) Zhang2
1. Student, Selkirk College, 301 Frank Beinder Way, Castlegar, British Columbia, Canada, 250-353-3300,
[email protected]
2. Research Scientist, Selkirk Geospatial Research Center, 301 Frank Beinder Way, Castlegar, British Columbia,
Canada,250-304-6527, [email protected]
* Presenting Author: Peter LeCouffe, Student, 250-353-3300, [email protected]
ABSTRACT
3D ground object reconstruction is crucial for understanding the geospatial properties of the Earth's
surface (Zhu and Kanade, 2008). Photogrammetry, a technique previously used in aerial photography on
traditional film cameras has been revived by the abundance of consumer grade cameras and the
availability of powerful processing hardware (Singh Et al, 2013). The availability of UAV sensed data fills
the spatial scale gap in traditional remote sensing. The goal of this project was to combine the use of an
Unmanned Aerial Vehicle (UAV) and terrestrial close range photogrammetry to create a complete 3D
model of Selkirk College's Castlegar campus. Incorporating both surveying and Geographic Information
Systems (GIS) methods, the data collected is now properly scaled and georeferenced. Ground control
points were established using real time kinetic GPS. The aerial survey has been conducted by Harrier Aerial
Surveys using a fixed wing UAV capturing low altitude nadir imagery, georeferenced using the ground
control points. A terrestrial survey was conducted using both a digital single lens reflex camera and a
consumer grade action camera to capture oblique imagery. Building corners within the aerial data, which
were visible in the terrestrial data, then were used as terrestrial control. Using this technique, both data
were properly scaled within the same coordinate system. Merging the two datasets then resulted in one
complete and georeferenced 3D model of the Selkirk College Castlegar Campus that can now be used for
better understanding the geospatial properties of the campus.
Keywords: photogrammetry, unmanned aerial vehicle, terrestrial, 3D model
Reference
Singh, S. P., Jain, K., and Mandla, V. R., 2013. “Virtual 3D Campus Modeling by Using Close Range
Photogrammetry”, American Journal of Civil Engineering and Architecture, 1(6), pp.200–2005.
Zhu, Z. and Kanade, T., 2008, “Modeling and Representations of Large-Scale 3D Scenes”, International
Journal of Computer Vision, 78(2–3), pp. 119–120.
182
The Birth and Growth of Pingo in the Canadian Arctic Observed by Satellite
Radar Interferometry
Sergey Samsonov1, Trevor Lantz2, Steve Kokelj3, Yu Zhang1, Robert Fraser1*and Ian Olthof1
1. Canada Centre for Mapping and Earth Observation, Natural Resources Canada, 560 Rochester Street, Ottawa,
ON Canada, Phone: 613-759-1186, Email: [email protected]
2. Environmental Studies, University of Victoria, Victoria, BC, Canada
3. Northwest Territories Geoscience Office, Government of the Northwest Territories, Yellowknife, NWT, Canada
* Presenting Author: Robert Fraser, 613-759-1186, [email protected]
ABSTRACT
In the framework of the Northwest Territories Cumulative Impact Monitoring Program Differential
Synthetic Aperture Radar (DInSAR) was used to monitor ground deformation related to permafrost thaw
in the Tuktoyaktuk Coastlands. Across this region vertical deformation generally did not exceed 2
cm/year, but we observed one large area of localized heave. In this paper we describe the evidence that
this feature is a pingo. Pingos are ice-cored hills that can grow only in the permafrost environment. There
are about 5000 or more pingos in the world, of these about 1350 are located in the Tuktoyaktuk
Coastlands. Thirty four high resolution Ultra-Fine RADARSAT-2 images acquired during 2011-2014 were
processed with the advanced Multidimensional Small Baseline Subset (MSBAS) DInSAR to produce vertical
and horizontal deformation time series and mean deformation rate. Uplift was detected with the
maximum deformation rate of 2.7 cm/year in a drained lake bottom. The detected uplifting rate is much
larger than the ground surface heave induced by uniform permafrost aggradation after draining of the
lake, which suggests that the detected feature is a pingo. Historical stereophotos, field observations and
borehole analysis also suggest that the observed feature of the elliptical shape with dimensions of about
600x600 m as a young pingo growing in a recently drained area. Prior to lake drainage the thermal mass
of the lake maintains unfrozen talik beneath the lake bottom. Following lake drainage, the bed is exposed
to subfreezing air temperature and permafrost aggrades causing pore water to freeze. In the freezing of
saturated sandy sediments, volumetric expansion forces about 9% of the pore water and excluded solutes
away from the freezing front. This process causes pressure to increase in the unfrozen talik creating
conditions favorable for the growth of segregated ice and the generation of heave necessary to deform
the frozen lake bottom sediments. Pingo growth can continue until the talik has refrozen. We modeled
observed deformation as uniformly loaded elliptical plate with clamped edges, with semi-major and semiminor axis of 348 m and 290 m tilted 27 degrees clockwise and centered at 69.3728N133.1688W. Such
model describes observed ground deformation in the spatial domain very well. Acceleration of vertical
deformation rate observed during summer, when mean daily temperature reaches maximum. Long term
measurement of heave at this site may contribute to better understanding processes governing temporal
evolution of pingos and growth of permafrost.
183
Spatial Distribution of Fine Particulate Matter over Southern Ontario from
MODIS Data
Jane Liu1,*, Jenny Cui1
1. Department of Geography and Planning, University of Toronto,100 St. George Street, Toronto, Ontario, Canada
M5S 3G3, 416-978 1672, [email protected]
* Presenting Author: Jane Liu, Assistant Professor/Dr., 416-978 1672, [email protected].
ABSTRACT
Particulate matter (PM) is a combination of fine solids and liquid droplets suspended in the air. PM 2.5 is
fine particulate matter with aerodynamic diameter less than 2.5 𝜇m. Epidemiological studies suggested
that both acute exposures and chronic exposures of PM impose a significant adverse health impact on
people.
PM2.5 concentrations were derived from the satellite instruments MODIS (The Moderate-Resolution
Imaging Spectroradiometer) and MISR (The Multi-Angle Imaging Spectroradiometer) (van Donkelaar,
2010). The data are used to examine the spatial distribution of PM2.5 concentrations over Ontario, in
combination with ground measurements from the Ontario Ministry of the Environment and Climate
Change.
The PM2.5 values derived from satellite data are generally consistent with the PM2.5 observations from the
surface, with low PM2.5 at northern cities (Thunder Bay, Sudbury, North Bay, and Sault Ste. Marie), high
PM2.5 in southern metropolitan and cities with heavy industrial activities (Toronto, Hamilton, and
Windsor), and intermediate PM2.5 for the middle-sized cities (Barrie, London, and Oakville). However, the
PM2.5 annual mean for Sarnia from the ground measurement was the highest (11 𝜇g/m3) among the 12
cities and this is not reflected in the satellite image, perhaps due to the coarse horizontal resolution (0.1°)
of the satellite data. The strong influence of long-range transport of PM2.5 over large areas in Ontario is
indicated by the high correlations among Ontario cities.
184
Forest Prairie Landscape Classification Based on Remote Sensing Data Fusion
Junna Yuan1*, Kun Li2, Yun Shao3, Chou Xie4, Fengli Zhang5
1. Assistant Researcher, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences , Datun Road,
Beijing, 100101,China,010-64838047,[email protected]
2. Assistant Researcher, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences , Datun Road,
Beijing, 100101,China,010-64838047,[email protected]
3. Researcher, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences , Datun Road, Beijing,
100101,China,010-64876313,[email protected]
4. Research Associate, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences , Datun Road,
Beijing, 100101,China,010-64838047,[email protected]
5. Research Associate, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences , Datun Road,
Beijing, 100101,China,010-64838047,[email protected]
* Presenting Author: Junna Yuan, Assistant Researcher, 010-64838047, [email protected].
ABSTRACT
The shortcoming of traditional use of hyperspectral image classification is that due to a variety of
vegetation types mutual mixed and with the impact of foreign body spectrum and synonyms spectrum,
the results of the vegetation type recognition accuracy is not high. Radar images have the ability to
penetrate through the clouds and fog, and could better display the backscattering characteristics and
the space texture feature. However, due to the complexity of the feature distribution of different
objects often results in similar backscatter signal characteristics, thus increasing the difficulty of the
extracted feature information. At the same time due to the polarization SAR images have a high
correlation, it decreases the image classification accuracy.
In this study, spectral radiation and spatial texture information of image were used to classify the forest
prairie landscape of SaiHanBa area. Both spectral and spatial texture information of the GF-1 optical
data and RADARSAT-2 SAR data were used .The first step is data preprocessing which include geometric
and radiometric correction、image filtering、orthorectification、geometric registration,etc.Then
supervised classification - the method of maximum likelihood classification was conducted.
There were 5 landscape units,which were deciduous Coniferous forests, evergreen coniferous forests,
grasslands, water, and residential areas. The overall classification accuracy was 85.8175%,the Kappa
coefficient was 0.8055.It showed that the classification method can effectively classify forest landscape
unit with high accuracy.The hyperspectral data and radar data acquisition are more convenient.So the
research has a certain universality.
185
Investigating the Effects of Input Data on the Results of Random Forest for
Classification of Peatland Landscapes
Koreen Millard 1 and Murray Richardson2
1. PhD Candidate, Carleton University, Ottawa, Ontario, Canada
2. Assistant Professor, Carleton University, Ottawa, Ontario, Canada
* Presenting Author: Koreen Millard, PhD Candidate, Carleton University, Ottawa, Ontario, Canada,
[email protected]
ABSTRACT
Random Forest is a widely used algorithm for classification of remotely sensed data. Most research on
image classification with Random Forest has shown it to be extremely useful for producing highly accurate
classifications of non-parametric and high-dimensional data, which is generally not possible with
traditional, parametric classifiers. In the evaluation of classification effectiveness with Random Forest,
most studies focused on performance against traditional parametric classifiers (such as Maximum
Likelihood, e.g. Akar and Gungor, 2012) or other machine learning techniques (such as Support Vector
Machines, e.g. Adam et al, 2014; Sonobe et al, 2014). A number of studies have examined the two key
parameters used in random forest (mtry: the number of variables sampled at the split, and ntree: the
number of trees 'grown' in the forest, e.g. Lawrence et al, 2006). We present results of an analysis of the
effects of input training data characteristics on Random Forest classifications in a peatland landscape
using LiDAR-derived DEM, DSM and vegetation derivatives. Several characteristics of input data are
shown to have significant impacts on the results of both out-of-bag and independent classification
accuracy. The number of training data samples used in building the model, the proportions of data in
each class within the training dataset, correlation of input variables, and the autocorrelation of training
data samples were found to lead to changes in the resulting classifications and reported error. These
factors have been quantitatively assessed and recommendations will be made on the optimum input data
configuration for classification with Random Forest.
References:
Adam et al (2014) IJRS, 35(10).
Sonobe, et al (2014), IJRS, 35(23).
Lawrence et al (2006), Remote Sensing of Environment, 100(3).
Akar and Gungor (2012), Journal of Geodesy and Geoinformation, 1(2).
186
Extracting sea ice information from the fusion of SAR and optical data
Igor Zakharov 1*, Pradeep Bobby 2, Desmond Power 3, Matt Arkett 4, Yi Luo 5, Matthew Smith 6,
Sherry Warren 7, Mark Howell 8
1. Dr. C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL Canada, A1B 3X5, 709.864.2582,
[email protected]
2. Director of Earth Observation, C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL Canada,
A1B 3X5, 709-864-8361, [email protected]
3. Vice President Remote Sensing, C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL
Canada, A1B 3X5, 709.864.8535, [email protected]
4. Remote Sensing Manager, Canadian Ice Service, 373 Sussex Drive, Ottawa, ON K1N 7B1, 613.947.7514,
[email protected]
5. Yi Luo, Physical Scientist, Canadian Ice Service, 373 Sussex Drive, Ottawa, ON K1N 7B1, 613.943.5755,
[email protected]
6. Student, Ocean Mapping program, Marine Institute, 155 Ridge Road, St. John's, NL, 709.687.2162,
[email protected]
7. Senior Remote Sensing/GIS Specialist, C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL
Canada, A1B 3X5, 709.864.7696, [email protected]
8. GIS Analyst, C-CORE, Captain Robert A. Bartlett Building, Morrissey Road St. John's, NL Canada, A1B 3X5,
709.864.8374, [email protected]
* Presenting Author: Igor Zakharov, Senior Research Engineer/Scientist, 709.864.2582,
[email protected]
ABSTRACT
Data fusion can be defined as integrating or combining data from multiple sources to provide improved
accuracy and information content that cannot be achieved from a single sensor alone. Several data fusion
techniques have been analyzed and implemented for combining SAR and optical data of different modes
and resolutions in order to extract additional sea ice characteristics. The objectives were focused on
understanding possible improvements in identifying and characterizing extreme ice features (features
which can cause extreme loads on offshore infrastructures) using medium resolution data and ice types
from low resolution data.
After preprocessing, MODIS bands were fused with Radarsat-2 (RS-2) ScanSAR imagery using the SARMultispectral Fusion Algorithm (ASMF). Despite limitations of coarse resolution, the fused product was
useful for discriminating between ice and open water, especially in high winds which resulted in high SAR
ocean clutter. In most cases there is a time difference between MODIS and RS-2 data acquisition which
causes difficulties in data interpretation, however this time difference allows to determine the speed and
direction of ice floes in the fused image.
Several dual polarized Radarsat-2 and TerraSAR-X data sets, their combinations and color composites
were useful for discriminating glacier ice in sea ice. Fusing medium resolution SAR and optical data
improves identification of extreme ice features (glacier ice and multiyear hummock fields). Fusion product
helps identify multi-year hummock fields surrounded by rubble fields, which is important since they are
stronger than first year features and are more hazardous to marine operations.
187
Using Satellite Imagery for Visual Simulation in Aviation Applications
Sam Lieff1*
1. General Manager, BlackBridge Geomatics, 3528 30th Street North, Lethbridge, Alberta, Canada, +014033326011,
[email protected]
* Presenting Author: Sam Lieff, General Manager, +14033326011, [email protected].
ABSTRACT
The best training is real world experience in the safety of a controlled environment. By using satellite
imagery, we can create visual solutions to bring reality into the virtual world of simulation. Leveraging the
unique capabilities of imagery from the latest satellites, we can produce an ultra-realistic visual model for
commercial aviation training applications.
The information in a satellite image gives an objective, reliable picture of the Earth’s surface. The imagery
is integrated into the simulation environment by processing it to meet the specific and rigorous needs of
the simulator. By using the latest software technology and techniques, the imagery is aligned to match
the existing spatial database of the airport, such as runway vectors, and building footprints. Colours are
then calibrated and matched to blend with the environment, and visual artifacts such as terrain
distortions, haze, clouds and sensor errors are removed. Using a multistep process, different images are
blended for a seamless transition from one image type (different resolution, color etc.) to another. By
creating a multi-resolution environment, we create a solution with high resolution imagery in areas that
require extra detail, and lower resolution imagery when the simulation is representing high altitude flight.
The end result is a clean, realistic, digital representation of the airport and the surrounding area.
188
Modelling Changes in Forest Dynamics under the Influence of Climate Change
and Different Forest Management Practice
Nilam Kayastha1*, Michael J. Falkowski2, Andrew T. Hudak3, Patrick A. Fekety4, Linda M. Nagel5, Asim
Banskota6
1. Research Associate, Department of Forest Resources, University of Minnesota, St. Paul, MN USA,
[email protected]
2. Research Associate Professor, Department of Forest Resources, University of Minnesota, St. Paul, MN USA,
[email protected]
3. Research Forester, USDA Forest Service Forestry Sciences Laboratory, Moscow, ID USA, [email protected]
4. Research Fellow, Department of Forest Resources, University of Minnesota, St. Paul, MN USA,
[email protected]
5. Professor, Department of Forest Resources, University of Minnesota, St. Paul, MN USA, [email protected]
6. Research Associate, Department of Forest Resources, University of Minnesota, St. Paul, MN USA,
[email protected]
* Presenting Author: Nilam Kayastha. Research Associate, , [email protected]
ABSTRACT
Over the next century, changes in temperature, precipitation and atmospheric CO2 concentration are
expected to impact forest dynamics by altering forest productivity, disturbance rates, and the distribution
of tree species. Understanding climate change impacts on forest composition, structure, and function is
essential for mitigating the effects of climate change via forest management adaptation strategies. In this
study, we used a spatially explicit forest succession and disturbance model, LANDIS-II, to investigate how
the interaction of climate and forest management practices alters the composition and structure of a
forest in northern Idaho, USA. LiDAR data was used to parametrize the LANDIS-II model across the study
area. Statistical imputation was used to upscale LiDAR-derived forest inventory attributes across the study
area. Forest change was simulated over one-hundred years (2000-2100) using three different climate
emission scenarios. A current climate scenario was also simulated using the historical climate data from
the PRISM climate dataset. Two downscaled climate projections PCM B1 (low emission scenario) and GFDL
A1FI (fossil fuel intensive scenario) were used to evaluate the effect of projected climate change in future
forest conditions. The three climate scenarios were then combined with different forest management
scenarios specifically crafted to enhance forest adaption to a changing climate. Climate change had a
significant effect on both aboveground biomass and species diversity across the study area. However, our
results also demonstrate that climate related impacts on forests can be somewhat tempered via forest
management activities specifically designed to increase forest resilience to a changing climate.
189
Large-Scale Modeling of Forest Height and Biomass using the Metabolic Scaling
Theory and Water-Energy Balance Equation
Sungho Choi1*, Taejin Park2, Yuri Knyazikhin3, Ranga B. Myneni4
1. PhD candidate, Boston University, Boston, MA, United States, 617-353-8846, [email protected]
2. PhD candidate, Boston University, Boston, MA, United States, 617-353-8846, [email protected]
3. Research Professor, Boston University, Boston, MA, United States, 617-353-8843, [email protected]
4. Professor, Boston University, Boston, MA, United States, 617-353-2525, [email protected]
* Presenting Author: Sungho Choi, PhD candidate, 617-353-8846, [email protected].
ABSTRACT
Forest height and biomass are major biophysical properties to understand carbon cycles across scales.
Because in-situ measurement is labor-intensive and thus impractical for the large-scale monitoring,
previous studies have used modeling approaches (e.g., spatial statistics or machine learning algorithms)
to alleviate the data discontinuity in space and time. In this study, we explored the application of the
metabolic scaling theory and water-energy balance equation in the extensive prediction of forest height
and aboveground biomass (AGB). Our model, called Allometric Scaling and Resource Limitations (ASRL),
accounts for the size-dependent metabolism of trees and local availability of light and water. Geospatial
predictors incorporated in the model are altitude and long-term monthly precipitation, air temperature,
solar radiation, vapor pressure and wind speed. Disturbance history is further implemented to estimate
contemporary forest height and biomass. This study provides baseline maps (circa 2005; 1 km2 grids) of
maximum forest canopy height and AGB over the continental USA. Uncertainties at the pixel-level are also
delivered. The modeled canopy height and biomass are validated with field measured data and
airborne/spaceborne lidar data. Our approach is promising for prognostic applications because the model
grants a simple and clear mechanistic understanding in the large-scale modeling compared to the
conventional black-box type approaches.
190
A Rapid Assessment of Light Geese Damage in the Central and Eastern Arctic
using Multi-temporal Landsat Imagery Stacks.
Jon Pasher1, Rasim Latifovic2, Darren Pouliot3, Valerie Wynja4*, Victoria Putinkski5 and Ryan Zimmerling6
1. Remote Sensing Scientist, Environment Canada, 1125 Colonel By Drive, Ottawa, Ontario, Canada, 613-990-9941,
[email protected]
2. Remote Sensing Scientist, Natural Resources Canada, 560 Rochester Street, Ottawa, Ontario, Canada,
[email protected]
3. Remote Sensing Scientist, Natural Resources Canada, , 560 Rochester Street, Ottawa, Ontario, Canada,
[email protected]
4. Geomatics Analyst, Environment Canada, 1125 Colonel By Drive, Ottawa, Ontario, Canada, 613-990-9947,
[email protected]
5. Geomatics Technician, Environment Canada, 1125 Colonel By Drive, Ottawa, Ontario, Canada, 613-998-8261,
[email protected]
6. Management Unit Head, Environment Canada, 351 St. Joseph Blvd., Gatineau, Québec, Canada, 613-404-3900,
[email protected]
*Presenting Author: Valerie Wynja Geomatics Analyst, 613-9909947, [email protected].
ABSTRACT
Light geese (Ross’s and Lesser Snow Geese) populations have been steadily increasing across the Canadian
Arctic since the 1970s, likely the result of a trophic cascade, beginning with improved agricultural practices
in over-wintering grounds. The avian herbivores cover large areas of marsh, intensively grazing the
sedges, sometimes pulling roots and shoots out beyond repair. A positive feedback loop is thus created
between grazing, the land’s exposure to sediment and the resulting change in vegetated land. As a result,
great swaths of vegetation have been irreversibly degraded due to excessive foraging in the central and
eastern arctic. Monitoring these changes in vegetation over large areas with remote sensing is critical for
ecosystem and wildlife management.
This paper presents the ongoing efforts to perform a rapid assessment of the extent of vegetation damage
in three study sites which have experienced varying degrees of goose damage over the past several
decades: Queen Maud Gulf Migratory Bird Sanctuary, East Bay Migratory Bird Sanctuary, and La Pérouse
Bay. The use of Landsat allows frequent assessment and monitoring, which is necessary to determine the
magnitude and extent of vegetation degradation and/or vegetation regeneration. For each study site,
Landsat 4/5 TM images were relatively calibrated and assembled into dense time series of images ranging
from 1984 to 2011. Several vegetation indices, soil indices and a Tassled Cap Transformation were applied
to the image time series. Following this, a trend analysis is being performed to detect subtle vegetation
changes caused by geese damage.
191
Investigating Boreal Forest Physiology and Stand-Age Relationships Using a
Landsat TM Derived Chronosequence
Holly Croft1*, Jing M. Chen2, Thomas L. Noland3
1. Holly Croft, Postdoctoral Fellow, Department of Geography, University of Toronto, Toronto, ON M5S 3G3, Canada,
[email protected]
2. Jing M. Chen, Professor, Department of Geography, University of Toronto, Toronto, ON M5S 3G3, Canada,
[email protected]
3. Thomas L. Noland, Research Scientist, Ontario Ministry of Natural Resources and Forestry, Ontario Forest Research
Institute, Sault Ste. Marie, ON P6A 2E5, Canada, [email protected]
* Presenting Author: Holly Croft, Postdoctoral Research Fellow, [email protected]
ABSTRACT
Forest stand age affects a number of ecological variables and processes, including biodiversity, leaf area
index (LAI), nutrient and water cycling, and biomass production. However, investigations into stand age
dependency are limited by the availability of measurement sites and have typically focused on stand
structure. This study uses a measured chronosequence of 9 managed Pinus banksiana stands, ranging
from 15 to 90 years old. Ground measurements at the sites in Northern Ontario, Canada were collected
for a range of canopy structural parameters (LAI, stand density, crown radius, tree height) and leaf
chlorophyll (Chlab). A time-series of satellite-derived data (Landsat 5 TM; 30 m) from 1989 to 2011 was
used to extend the chronosequence, with Chlab- and LAI- sensitive spectral vegetation indices calculated
for each image date. Stand age showed strong relationships with tree height (R2 = 0.95, p<0.001), canopy
radius (R2 = 0.68; p<0.01) and stand density (R2 = 0.49; p<0.01). A spherical model was fitted to the LAI
and Chlab time-series, where an asymptote was reached at Chlab = 44 years and LAI = 22 years, after which
both variables showed no further increase with stand age. Temporal lag variations indicate differences in
the maturation period for Chlab or LAI, with LAI likely related to stand density reduction. The demonstrated
stand age-dependency of Chlab is crucial for understanding stand age effects on photosynthetic processes
and carbon assimilation; for quantifying net primary production within carbon budgets and guiding forest
management in light of a changing climate.
192
Modelling and mapping soil carbon in British Columbia grasslands
Heather Richardson1*, Lauchlan Fraser2, and David Hill3
1. MSc Candidate, Thompson Rivers University, Environmental Sciences, 900 McGill Road V2C 0C8, Kamloops, BC,
Canada, (250) 571-3402, [email protected]
2. Professor, Thompson Rivers University Department of Biological Sciences, 900 McGill Road V2C 0C8, Kamloops,
BC, Canada, (250) 377-6135, [email protected]
3. Assistant Professor, Thompson Rivers University Department of Geography and Environmental Studies, 900
McGill Road V2C 0C8, Kamloops, BC, Canada, (250) 828-5099, [email protected]
* Presenting Author: Heather Richardson, MSc Candidate, 250 571 3402,
[email protected]
ABSTRACT
Industrialization, production and consumption of fossil fuels, and land use changes have resulted in
increased concentrations of CO2 and other greenhouse gases in the atmosphere causing changes in
ecosystem structure and properties. Soil C sequestration, the process of storing carbon dioxide in the soil
through crop residues and other organic solids, has been an area under much investigation as it relates to
reducing atmospheric carbon and mitigating climate change. Since grasslands predominately sequester
carbon below ground through root growth and consequent soil-building processes, they have a high
potential for long term C storage and therefore are of major importance for maintaining earth’s carbon
cycle. Despite advances in soil C determination in recent years, it remains a challenge to model and
monitor soil C across large regions. There are several factors, both anthropogenic and environmental, that
influence carbon sequestration. Given this complex system, the possibility of using remote sensing (RS)
applications in conjunction with accurate field measurements is a topic of much interest. Here, we
examine the mechanisms that affect soil C storage and subsequently model and map soil C in the southern
interior grasslands of British Columbia (BC). Soil C prediction will be based on the Normalized Difference
Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI), which have demonstrated high
correspondence with soil C distribution in past studies. The relationship of soil C and NDVI/EVI will be
evaluated on two scales using: i) MODIS (MOD13Q1, 250m/ 16day resolution) imagery, and ii) a handheld
Multispectral Radiometer (MSR16R, Cropscan Inc., 1m resolution ) device. Other factors included in the
model are: i) grazing, ii) climate data, iii) vegetation community zones, iv) soil classification and soil
drainage, and v) topography. A traditional stepwise linear modelling approach was compared with
random forest modelling, which is a recursive partitioning technique that employs randomized bagging
and bootstrapping of samples. Based on comparisons of mean squared error, random forest models
created better predictions than stepwise linear models. When MSR-derived NDVI data was inputted in
models, the percentage of explained variance was greater than for models which used NDVI derived from
MODIS data, showing the potential of increased model accuracy with higher resolution RS data. This
project will help create the groundwork for effective monitoring techniques of soil C levels using RS
techniques and help direct land management efforts to increase C sequestration in BC.
193
A Landsat-Based Study of Black Rock Coatings Proximal to Base-Metal Smelters,
Sudbury, Ontario, Canada
Kelly J. Malcolm1, David W. Leverington2*, and Michael Schindler3
1. Student, Earth Sciences, Laurentian University, Sudbury, ON, Canada, [email protected]
2. Associate Professor, Geosciences, Texas Tech University, Lubbock, TX, USA, [email protected]
3. Associate Professor, Earth Sciences, Laurentian University, Sudbury, ON, Canada, [email protected]
* Presenting Author: David Leverington, Assoc. Professor, (806) 834-5310, [email protected]
ABSTRACT
Past emission of metal-bearing particulate matter, sulfur dioxide, and sulfuric acid by base-metal smelters
in the Sudbury region led to widespread loss of vegetation, contamination of soils, and formation of black
coatings on rock surfaces. These black coatings formed through the incorporation of smelter-borne
particulate matter into the partly-dissolved uppermost layers of siliceous minerals on exposed rock.
This study involved assessment of the reflectance properties of black coatings in the Sudbury region, and
determination of the spatial distribution of coatings through supervised classification of reflectance data
derived from a Landsat Enhanced Thematic Mapper Plus (ETM+) image. Classifications involved the use
of the spectral angle mapper (SAM) and maximum likelihood algorithms.
The reflectance spectra of black coatings in the Sudbury region are generally flat and relatively featureless,
and are characterized by low reflectance values of under 12% across the visible, near-infrared, and shortwave infrared. Weak absorption features variously exist in some spectra at ~950-1000, 1400, 1915, 2200,
2210, 2260, 2310, 2350, and 2385 nm. Though SAM classification results are characterized by the
widespread mislabeling of uncoated urban and open-pit sites as mantled by black coatings, results
generated by the maximum likelihood algorithm properly depict the general distribution of exposed black
coatings in the Sudbury region. The mapping of black coatings using remote sensing methods can provide
useful information on the spatial character of environmental degradation in the vicinity of smelters, and
should be helpful in the monitoring of environmental recovery where emissions have been reduced or
eliminated.
194
Mapping northern peatland types using hyperspectral images of different
spatial and spectral resolutions
Asim Banskota1*, Mary Ellen Miller2, Michael J. Falkowski3, and Laura Bourgeau-Chavez4
1. Research Associate, Department of Forest Resources, University of Minnesota, St. Paul, MN USA,
[email protected]
2. Research Engineer, Michigan Technological University, Michigan Tech Research Institute, 3600 Green Ct Suite
100, Ann Arbor, Michigan, USA, [email protected]
3. Research Associate Professor, Department of Forest Resources, University of Minnesota, St. Paul, MN USA,
[email protected]
4. Research Scientist, Michigan Technological University, Michigan Tech Research Institute, 3600 Green Ct Suite
100, Ann Arbor, Michigan, USA, [email protected]
* Presenting Author: Asim Banskota. Research Associate, [email protected]
ABSTRACT
Northern peatlands reserve a large amount of carbon that is vulnerable to climate-induced changes in
hydrological dynamics and fire regimes. Understanding the role of peatlands in the global carbon cycle
and their future responses under a changing climate requires accurate mapping of peatland features and
types. Identifying peatlands, which requires estimating the depth of peat, is challenging with remote
sensing data. Alternatively, vegetation maps together with hydrological and geomorphological
characteristics can serve as proxy variables for identifying peatland types. In this study, we compare
hyperspectral images of different spectral (AVIRIS classic vs AVIRIS Next Generation (NG)) and spatial
resolutions (1 m vs 4 m) for their ability to discriminate characteristic peatland types, such as bogs and
patterned fens, among others, in the Seney National Wildlife Refuge in Michigan, USA. AVIRIS classic data,
with a spectral resolution of 224 bands (10 nm ± 0.1 nm) and a spatial resolution of approximately 3.8 m,
were acquired in August 2013. AVIRIS NG data with a spectral resolution of 426 bands (5 nm ±- 0.5 nm)
were acquired from two different altitudes in September 2014, yielding spatial resolutions of
approximately 3.8 m and 1 m. Ground data characterizing peatland types along several transects were
collected coincidently with the AVIRIS over-flights. In order to classify peatlands type, we employ an
approach that selects a useful subset of spectral bands and their transformed components (using a
continuous wavelet transform) followed by image classification using the Random Forests classifier. The
classified map is evaluated using an independent validation data set as well as via a comparison with a
peatland type map previously derived from classification of Radarsat-2 data.
195
Relating Low Arctic Tundra – Atmosphere CO2 Exchange to Satellite-Derived
NDVI using Phenological Analysis at Daring Lake, NWT
Claire Elliott 1*, Elyn Humphreys 2
1. Student, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Canada, K1S 5B6, 613-520-2600 x 4125,
[email protected]
2. Associate Professor, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Canada, K1S 5B6, 613-5202600 x 4125, [email protected]
* Presenting Author: Claire Elliott, Student, 613-520-2600 x 4125, [email protected]
ABSTRACT
Tundra ecosystems facing a rapidly warming climate are predicted to see alterations in vegetation
community phenology and composition resulting in changes to carbon cycling. At the Daring Lake
Ecosystem Research Station, NWT (64°52 N, 111°35 W) carbon exchange between the biosphere and
atmosphere has been monitored continuously since 2004 using eddy covariance systems. Eddy covariance
towers placed in three distinct tundra vegetation communities measure net ecosystem exchange of
carbon dioxide (CO2) over footprints of ~200 m. Spatial and temporal trends of vegetation productivity
measured via the normalized difference vegetation index (NDVI) can be used to infer CO2 exchange should
the relation between the two variables be simply generalized over a full growing season. In the summer
of 2014 (June – August) a series of nested sampling plots were established in the tower footprints to
provide weekly field-based measurements of NDVI at 0.5 m, 5 m, 30 m, and 200 m. Concurrent
measurements of leaf area index and percent cover were also collected. For each measurement
resolution, the temporal variations in field-based NDVI will be compared with related IKONOS, Landsat 8,
and MODIS NDVI data. The satellite derived NDVI data will then be compared with CO2 exchange derived
from the eddy covariance towers. This analysis will be executed using TIMESAT software to extract
phenological event information from the NDVI and CO2 flux datasets. The fundamental goal of this study
is to determine the potential for satellite NDVI-derived phenological indices for estimating growing season
CO2 exchange in Low Arctic tundra communities.
196
Influence of Sample Distribution and Prior Probability Adjustment on Land
Cover Classifier Extension
Darren Pouliot1*, Rasim Latifovic2, William Parkinson3
1. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759-6341, [email protected]
2. Research Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa, Ontario
K1A 0E4, Canada, 613 759-7002, [email protected]
3. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6535, [email protected]
* Presenting Author: Darren Pouliot, Remote Sensing Scientist, , 613 759-6341,
[email protected]
ABSTRACT
Machine learning algorithms are widely used for remote sensing based land surface characterization.
Successful implementation requires a representative training sample for the domain to which it is applied.
However, accessibility and cost strongly limit the acquisition of suitable training samples for large regional
applications. Further, it is often desirable to use previously developed datasets where significant
resources have already been invested, such as data developed from extensive field survey or high
resolution remotely sensed imagery. However, these data often only partially represent the domain of
interest and can lead to sample bias. Classifier spatial extension is an extreme case, where a sample is
trained from one region (or domain) and applied in another. This approach is desirable from a cost
perspective, but achieving acceptable accuracy is often difficult. In this research we investigate two
approaches to account for possible differences between the sample and domain of interest land cover
distributions. The first is an iterative resampling approach to predict the domain distribution and adjust
the sample to match. The second is the use of prior probabilities to adjust class memberships. Results
reveal that sample adjustment is superior if the domain distribution is well known. However, if it is not
then the use of prior probabilities preforms similarly. Examination of factors influencing the performance
of either approach revealed that for a given land cover problem the absolute difference between the
sample and application distributions was indicative of the performance improvement.
197
Fractional Land Cover Monitoring of the Alberta Oil Sands Region using Landsat
multispectral time series and high resolution Geoeye Imagery
William Parkinson1*, Darren Pouliot2, Rasim Latifovic3
1. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759 6535, [email protected]
2. Remote Sensing Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa,
Ontario K1A 0E4, Canada, 613 759-6341, [email protected]
3. Research Scientist, Canada Center for Mapping and Earth Observation, 560 Rochester Street, Ottawa, Ontario
K1A 0E4, Canada, 613 759-7002, [email protected]
*Presenting Author: William Parkinson, Remote Sensing Scientist, 613 759 6535, [email protected]
ABSTRACT
For change monitoring in the Alberta Oil Sand Region (AOSR) high resolution (<5 m) multispectral time
series would be preferred to capture the varying size and rates of change that occur. However, limited
spatial-temporal coverage and cost of current high resolution sensors make such an approach impractical
to derive historical information and provide for continued monitoring. Using moderate resolution (~30m)
time series such as that available from Landsat together with an appropriate fine resolution sample may
provide an alternative. In this research the potential to derive sub-pixel information on land cover was
evaluated. Sub-pixel land cover fractions were trained for Landsat using high resolution (2 m) Geoeye data
classified into basic land cover types. The point spread function of Landsat was modeled to ensure that
the reflectance properties measured by Landsat were coincident with the training footprint in the Geoeye
scenes. Random Forest and SEE5 decision tree classifiers were tested for prediction. Results showed that
land cover fractions could be extracted over the region with an average error ranging from 7-17%.
Sampling exerted a significant effect where validation using a holdout Geoeye scene preformed inferior
to sampling from all available scenes by on average 3%. Water and bare covers had limited sampling for
fractions between 25-75% and therefore the results for these covers are not indicative of performance.
Better controlling for spectral variability, training and Landsat data quality in site specific analysis suggests
significant improvement in accuracy of predicting fractions compared to the regional analysis. The
improvement for the site specific analysis ranged from 5-10%. Examination of forest fraction sensitivity to
change revealed good agreement with forest harvesting and fire. For areas known to not have change the
average inter-annual difference was <10%. Additional research will seek to improve regional application
for achieving site specific accuracy.
198
Using Airborne and Terrestrial Lidar to Estimate Biomass of Low-Stature Arctic
Tundra Shrubs
Heather E Greaves1*, Lee A Vierling2, Jan UH Eitel3, Troy S Magney4, Natalie T Boelman5, Kevin L Griffin6
1. PhD Student, Geospatial Laboratory for Environmental Dynamics, Department of Forest, Rangeland, and Fire
Sciences, University of Idaho, 875 Perimeter Dr MS 1133, Moscow, ID, 83844 USA. [email protected].
2. Associate Professor, Geospatial Laboratory for Environmental Dynamics, Department of Forest, Rangeland, and
Fire Sciences, University of Idaho, 875 Perimeter Dr MS 1133, Moscow, ID, 83844 USA. [email protected].
3. Research Assistant Professor, Geospatial Laboratory for Environmental Dynamics, Department of Forest,
Rangeland, and Fire Sciences, University of Idaho, 875 Perimeter Dr MS 1133, Moscow, ID, 83844 USA.
[email protected].
4. PhD Candidate, Geospatial Laboratory for Environmental Dynamics, Department of Forest, Rangeland, and Fire
Sciences, University of Idaho, 875 Perimeter Dr MS 1133, Moscow, ID, 83844 USA. [email protected].
5. Assistant Research Professor, Lamont-Doherty Earth Observatory, Columbia University, 61 Rte 9W, Palisades, NY,
10964 USA; Department of Earth and Environmental Sciences, Columbia University, Mail Code 5505, New York, NY,
10027 USA. [email protected].
6. Professor, Lamont-Doherty Earth Observatory, Columbia University, 61 Rte 9W, Palisades, NY, 10964 USA;
Department of Earth and Environmental Sciences, Columbia University, Mail Code 5505, New York, NY, 10027 USA;
Department of Ecology, Evolution, and Environmental Biology, Columbia University, 1200 Amsterdam Avenue, New
York, NY, 10027 USA. [email protected].
*Presenting Author: Heather E Greaves, PhD Student, 83844 USA. [email protected].
ABSTRACT
Rapid climatic warming in Arctic tundra ecosystems may be causing shifts in tundra vegetation
composition, including increasing aboveground shrub biomass. Understanding these shifts is important
for predicting ecosystem trajectories, but establishing methods for quantifying and scaling biomass in lowstature biomes is challenging. We seek to improve landscape-level estimates of aboveground tundra
woody biomass by linking fine-scale destructive biomass measurements to terrestrial and airborne lidar
metrics.
Using a volumetric surface differencing approach, we found strong relationships between harvested
biomass of 2 dominant low-stature (<1.5m) Arctic shrub species and terrestrial lidar metrics in 24 small
(1m diameter) plots in north central Alaska, USA. Relationships established using data from 5 or 6 coregistered scans taken at varying distances (2-50m) from the target plots were only slightly less strong (R2
= 0.90; RMSE = 128g) than relationships based on data from 2 close-range (2m) co-registered scans (R2 =
0.92; RMSE = 117g). In an additional 60 small plots located in the overlap of the terrestrial and airborne
lidar acquisitions, we similarly found strong relationships between harvested aboveground biomass of all
shrubs >5cm tall and terrestrial lidar metrics (R2 = 0.83; RMSE = 96g) and airborne lidar metrics (R2 = 0.77;
RMSE = 113g).
We also found that the strength of the airborne lidar-shrub biomass relationship depended strongly on
the choice of ground-finding algorithm used with the airborne lidar data. Continuing comparison of
terrestrial and airborne lidar data will allow us to optimize ground-finding algorithms to permit accurate
estimates of tundra shrub biomass across large spatial scales.
199
Change Detection Analysis Using Compact Polarimetry On Simulated RCM Data
Vincent Decker1* and François Charbonneau1
1. Canada Center for Mapping and Earth Observation, Natural Resources Canada, 560 Rochester, Ottawa, Ontario,
Canada
* Presenting Author: Vincent Decker, 613-694-2688, [email protected]
ABSTRACT
The future RADARSAT Constellation Mission (RCM) is designed to respond efficiently to operational user
requirements. Consequently, new capability features will be added to the RADARSAT family: Rapid Revisit
(RR) and Compact Polarimetry (CP).The RCM CP configuration, circular transmit–linear receive (horizontal
and vertical polarizations), will be available on all beam modes and it will compensate partially for the lack
of operational fully polarimetric mode. Simulated RCM CP data has shown promising potential over a wide
range of valuable applications for the Government of Canada such as ecosystem monitoring (forestry,
agriculture, wetlands and coastal change monitoring). Change detection techniques based on the
Bhattacharya distance have been performed on fully polarimetric and simulated RCM CP data that was
acquired over two sites: Kelowna British Columbia and Elk Island National Park (EINP) Alberta. Change
detection products generated from seasonal and multi-year stacks show that CP can perform as well as
fully polarimetric data. In EINP, vegetation around lacustrine environments showed the most sensitivity
to change with regards to seasonal water levels and ice forming/thawing. Grasses such as cattail also
showed high change intensity from spring to summer. In Kelowna, abrupt land cover changes from
orchard to urbanized settings showed the highest dispersion in change intensity. Lastly, the classification
of types of change between pairs of SAR acquisitions was assessed according to basic scattering
mechanisms derived from the m-chi decomposition approach. As development work continues, RCM CP
combined with fast revisit may offer an additional source of information to geoscientists studying change
detection across a wide range of applications.
200
Comparing the potential of dual-pol TerraSAR-X, Sentinel, and Radarsat data for
automated, polarimetric sea ice classification
Rudolf Ressel1*, Anja Frost2, and Susanne Lehner3
1. Researcher, DLR EOC Bremen Airport, Henrich Focke Str. 4, Bremen, 28199, Germany, +49421244201858,
[email protected]
2. Researcher, DLR EOC Bremen Airport, Henrich Focke Str. 4, Bremen, 28199, Germany, +49421244201859,
[email protected]
3. Head of Research Group, DLR EOC Bremen Airport, Henrich Focke Str. 4, Bremen, 28199, Germany,
+49421244201850, [email protected]
* Presenting Author: Rudolf Ressel, Researcher, +49421244201858, [email protected] .
ABSTRACT
In contrast to SAR single-pol data, which allow only classical image analysis, SAR dual-pol imagery can be
analyzed by means of complex polarimetry. Our work investigates the potential of different dual-pol
configurations (co-pol, compact polarimetry) in different satellite SAR sensors (TerraSAR-X, Sentinel;
Radarsat) for automatic sea ice classification. The first step of our analysis comprises the extraction of
polarimetric features. To enrich the information content of image segments, second order statistics on
these polarimetric features are additionally computed. The discriminative power and relevance of the
different features are ranked by utilizing the concept of mutual information. Different selections of the
most relevant features are then fed into a neural network classifier. We explore different network
configurations for optimal classification results. Performance is compared for different selections of
relevant features. In order to evaluate the generalizability of trained classifiers, data for classification is
taken from various geographical regions (Svalbard, Kara Sea, Baffin Island Coast, Antarctic). The outcome
for the different sensors is then also discussed in terms of reliability and applicability.
The implemented dual-pol processing chain exhibits improved performance over classical single-pol
texture based ice classification approaches and is well-suited for fully automated ice charting purposes in
near real-time situations. The promising results we achieved for our single-pol based classification
algorithm during field campaigns (Akademik Shokalskyi, Polarstern, Lance) can therefore also be expected
for dual-pol data, complementing our portfolio of navigation assistance products.
201
Application Tests of Ground-Based Coherent Radar for Deformation and
Vibration Measurements in Canada's Atlantic Region
Arpik Haikazi Hakobyan1*, Peter McGuire2, Desmond Power3, Cecilia Moloney4, Thomas Puestow5,
Guido Luzi6
1. PhD Candidate, C-CORE, MUN, 1 Morrissey Road, St. John’s, NL, Canada, 7097403137, [email protected]
2. Senior Project Engineer, C-CORE, 1 Morrissey Road, St. John’s, NL, Canada, 7098642006, [email protected]
3. Vise-President, C-CORE, 1 Morrissey Road, St. John’s, NL, Canada, 7098648353, [email protected]
4. Professor, MUN, S.J. Carew Building, 240 Prince Phillip Drive, St. John’s, NL, Canada, 709864896,
[email protected]
5. Senior Manager, C-CORE, 1 Morrissey Road, St. John’s, NL, Canada, 7098642586, [email protected]
6. Associate Professor, Senior Researcher, CTTC, Av. Carl Friedrich Gauss, Castelldefels, Spain, 34935569280,
[email protected]
* Presenting Author: Arpik Hakobyan, PhD Candidate, 7097403137, [email protected]
ABSTRACT
To reduce and prevent disasters and avoid human and economic loses, our knowledge about ground
movements, deformations of critical infrastructures, and their triggering factors must be improved to
make the disasters predictable, and thus preventable. One approach to derive deformation
measurements and to enhance short- and long-term forecasting methods and techniques is development
of advanced remote sensing methods. This research examines the ground-based radar technology in
measuring the displacements and deformations of various origin; demonstrates the operational and
performance characteristics of ground-based radar through a series of field tests carried out in Atlantic
Canada in 2012. The primary goal of case studies presented here is to apply ground-based radar
technology to estimate structural instabilities and ground motions, validate the capabilities of the
instrument, and to evaluate the instrument in detecting displacements in various application settings. The
field tests include data collection cases for the ground-based Synthetic Aperture Radar(IBIS-L), and for
ground-based Real Aperture Radar(IBIS-S). Some examples of applications include Big Falls Hydro-Electric
Generating Plant, a gravel pit, a building on MUN campus, high rise structure such as chimney. The
outcomes of data analysis that are presented here underline the feasibility and the importance of such
studies. The IBIS(Image By Interferometric Survey) ground-based radar sensor is developed by the IDS
(Ingegneria dei Sistemi SpA), Italy, and is owned by C-CORE.
202
Comparison of Different Methods of Soil Moisture Mapping in Peatlands using
Synthetic Aperture Radar Polarimetric Data
Koreen Millard1 and Murray Richardson2
1. PhD Candidate, Carleton University, Ottawa, Ontario, Canada
2. Assistant Professor, Carleton University, Ottawa, Ontario, Canada
* Presenting Author: Koreen Millard, PhD Candidate, Carleton University, Ottawa, Ontario, Canada,
[email protected]
ABSTRACT
Peatland surface soil moisture is important in many biophysical models but access to peatlands is often
limited or difficult. Therefore many models currently rely on single point-scale measurements which may
not accurately reflect the true spatial variability of soil moisture. Remote sensing provides a synoptic
view of inaccessible areas and Synthetic Aperture Radar (SAR) is generally thought to be the most
promising sensor for soil moisture retrieval due to its sensitivity to the dielectric constant of its targets.
Throughout the literature several different modelling approaches have been examined for relating in-situ
measurements with remotely sensed SAR data. However, the high degree of variability in both soil and
vegetation parameters is widely cited as the fundamental barrier to the development of reliable predictive
models.
Throughout the summer of 2014, in-situ soil moisture measurements were acquired during the same day
as RADARSAT-2 Fine Quad acquisitions in a peatland in Eastern Ontario. Here, we present results of
several different modeling techniques for estimating soil moisture in peatlands using polarimetrc SAR
data, as well as LiDAR DEM, DSM and canopy derivatives and optical data. Preliminary results show that
C-Band SAR intensity data alone were not sufficient for retrieving soil moisture in peatlands (best MLR R2
< 0.4) but the addition of elevation improved the model (R2 > 0.75). Results of the Water Cloud Model
and temporal differencing analysis were highly variable depending on date and specific model parameters
employed.
203
Classification of forest and wetland communities near the Victor Diamond Mine
using an integrated optical and radar satellite sensor dataset
Michael Stefanuk1*, Dr. Steven Franklin2, Dr. Oumer Ahmed3
1. Michael Stefanuk*, Undergraduate Student, Trent University, 257 Fern Cr., Waterloo ON, CA, N2V 2P9, 705-9270911, [email protected]
2. Dr. Steven Franklin, Professor of Geography and Environmental Science, Trent University, 1600 West Bank Dr.,
Peterborough ON, CA, K9J 7B8, 705-748-1011, [email protected]
3. Dr. Oumer Ahmed - Post-Doctoral Fellow in Geography, Trent University, 1600 West Bank Dr., Peterborough ON,
CA, K9J 7B8, 705-748-1011 ext.7965, [email protected]
* Presenting Author: Michael Stefanuk, Student, 705-927-0911, [email protected].
ABSTRACT
Remote sensing techniques have become a powerful tool for landscape-scale examinations of ecosystems
with an ever-increasing degree of accuracy. This analysis can be used to characterize ecosystems for a
variety of purposes including mapping for land inventory, resource management and informed
development. Optical and radar datasets from satellite sensors have been shown to be independently
effective for classifying forest and wetland land cover types. Classification accuracy of optical datasets
such as Landsat-7 ETM+ has been shown to be approximately 75%. This accuracy has been shown to be
increased to around 80% when analysis is augmented by radar data such as RADARSAT-2 because of
radar’s increased ability to sense water and better differentiate between wooded wetlands and upland
forests. This research will determine the classification accuracy of forest and wetland communities in the
area around the Victor Diamond Mine through analysis of a combination of Landsat-7 ETM+ and Radarsat2 quadpolarization synthetic aperture radar (SAR) dataset using maximum likelihood classification
methods. If optical satellite data (e.g., Landsat) and SAR (e.g., Radarsat-2) data are combined into a single
dataset for analysis, then the classification accuracy will be higher than if these datasets were classified
independently. It is predicted that analysis of the combined Landsat and Radarsat dataset will show
increased classification accuracy relative to classifications of either dataset independently.
204
Helix Nebula – Supporting science with a comprehensive cloud computing
infrastructure: the use case for Earth Observation
Marc-Elian Begin1, Cedric Seynat2*
1. Marc-Elian Begin, Founder and CEO, SixSq, 8 Rue du Bois-du-Lan, 1217 Geneva, Switzerland, [email protected]
2. Cedric Seynat*, Manager, Space Engineering, ADGA-RHEA Group, 6700 Côte-de-Liesse, Suite 105, Montreal,
Québec H2T 2B5, Canada, [email protected]
*Presenting Author: Cedric Seynat, Manager, [email protected]
ABSTRACT
‘Helix Nebula – The Science Cloud’ is a European initiative established in 2011 by 3 large
intergovernmental agencies (the European Space Agency, the European Molecular Biology Laboratory and
the Centre Européen de Recherche Nucléaire). Helix Nebula provides a multidisciplinary cloud platform
for data intensive science. It brings together leading IT providers of cloud infrastructure and services with
the science community to provide computing capacity and services that elastically meet big science’s
growing demand for computing power. The Helix Nebula marketplace (HNX) delivers easy and large-scale
access to a range of commercial Cloud Services through innovative broker technology.
The European Earth Observation program Copernicus provides a timely opportunity to start an open
ecosystem building on the Helix Nebula platform. As a pilot project, the European Space Agency is creating
a platform focusing on earthquake and volcano research. The project, named Geohazard Supersites,
focuses on monitoring the dynamic and complex solid-Earth system and the assessment of geohazards.
Its mission is to support the Group of Earth Observation in the effort to reduce the impact of natural
disasters. The project aims to provide an open source, unified e-infrastructure for solid Earth data, and
improved data products for solid Earth monitoring. In the long term, the project will provide a platform
for the optimal exploitation of EO data, and first choice workspace for scientists and academics, service
providers and other value adding organisations.
The Helix Nebula initiative is not restricted to Europe, and Canadian scientists and industry can potentially
benefit from joining the Helix Nebula initiative and from accessing or contributing to the Helix Nebula
marketplace.
205