LiDAR - NSGIC - National States Geographic Information Council

USGS Base
Specifications V1.2
Presented by:
Jason Caldwell, Vice President of Business
Development and Sales
Date: February 25th, 2015
LiDAR Accuracy
National Research Council of the National Academies
•
•
•
•
“LiDAR produces very-high-resolution three-dimensional point clouds in a wide
variety of land cover types at accuracies equivalent to or better than
photogrammetry…”
“LiDAR can be acquired day or night, in cloudy conditions, leaf on or leaf off and no
sun angle limitation…”
“LiDAR is able to penetrate to the bare earth in vegetated area better than either
IFSAR or photogrammetry…”
“LiDAR data processing and feature extraction can be incorporated seamlessly into
the production environment designed for photogrammetry…”
Hydro-electric dam, Puerto Rico
2
LiDAR Accuracy
USGS 30 Meter DEM
vs.
3
LiDAR DEM
LiDAR Accuracy
.
0
7
.15
Legend
EG Max WS
WS Max WS
5270
Elevation (ft)
from 7.5-minute quadrangle
map
.15
5280
Crit Max WS
Ground
5260
Bank Sta
5250
5240
-5000
-4000
-3000
-2000
-1000
0
1000
Station (ft)
.15
.
0
7
5280
.15
Legend
EG Max WS
Crit Max WS
from LIDAR data
Elevation (ft)
5270
WS Max WS
Ground
5260
Bank Sta
5250
5240
0
1000
2000
3000
4000
5000
6000
Station (ft)
4
http://pubs.usgs.gov/tm/11b4/pd
f/tm11-B4.pdf
2/25/2015
©2013, The Sanborn Map Company, Inc.
5
New Terminologies
• NPD = Nominal Pulse Density (2 ppm; 4 ppm; 8 ppm)
– NPS and NPD Evaluated on first return data of single swath
– NPS = Nominal Pulse Spacing (0.35 m; 0.7m; 1.4m; )
• ANPS & ANPD = Aggregate Nominal Pulse Density/Spacing
– The new specs have provided some provision on meeting the
required densities using swath overlaps as long as it is agreed upon
with the customer.
•
•
•
•
DPA: Defined Project Area
BPA: Buffered Project Area: DPA + 100m buffer
NVA: Non-vegetated vertical accuracy
VVA: Vegetated vertical accuracy
LiDAR Quality Levels
Quality Level (QL)
RMSEZ (nonvegetated) (cm)
NPS/ANPS (m)
NPD/ANPD
(pls/m2)
QL0
≤ 5.0
≤0.35
≥8.0
QL1
≤ 10.0
≤0.35
≥8.0
QL2
≤ 10.0
≤0.71
≥2.0
QL3
≤ 20.0
≤1.41
≥0.5
LiDAR Data Acquisition: Flight-Planning
• Requirements and guidelines for flight-line overlap and
scan angle limits have been removed.
o V1.0 used to call for Minimum 10% - 30% overlap and max 40
degrees scan angle (full -angle)
o As long as the final dataset has no gaps and meets the accuracy
requirements, there is no limitation on overlap and scan angle
• Minimum 100 m buffer around the project area
• The required project density can be achieved using
overlap as long as it is agreed upon with the customer in
the proposal
• No requirements for baseline distance:
– 30 kms for QL-2
– 40 kms for QL-3
Relative Accuracy
Quality Level
(QL)
Smooth surface
repeatability
(cm)
QL0
≤3
≤4
±8
QL1
≤6
≤8
±16
QL2
≤6
≤8
±16
QL3
≤ 12
≤ 16
±32
Swath overlap
Swath overlap
difference, RMSDZ
difference,
(cm)
maximum (cm)
• There is no guidelines on how relative accuracy is evaluated and reported
• USGS is trying to come up with some standard process
• There is no spec for the relative accuracy in X, Y
Absolute Accuracy
VVA at
NVA at 95-percent
95th
confidence level
percentile
(cm)
(cm)
≤ 9.8
≤ 14.7
Quality Level (QL)
RMSEZ (nonvegetated) (cm)
QL0
≤ 5.0
QL1
≤ 10.0
≤ 19.6
≤ 29.4
QL2
≤ 10.0
≤ 19.6
≤ 29.4
QL3
≤ 20.0
≤ 39.2
≤ 58.8
• There is no spec for the absolute accuracy in X, Y (Horizontal)
TIN
CONTOURS
LiDAR Check-Points
Accuracy Assessment
Data Courtesy 11
of State of NC
CHECK
POINT
MASSPOINTS
LiDAR Check-Points
Accuracy Assessment
• The check points should be well distributed
• All the land cover types should be well represented.
• No objective guidelines are provided for ‘well-distributed’ and ‘well-represented’ terms
Fundamental Vertical Accuracy (FVA)
Supplemental Vertical Accuracy (SVA)
Class
number
1
2
3
4
5
Land cover class or Description
Clear or Open, bare earth, low grass; e.g. sand,
rock, dirt, plowed fields, lawns, golf courses
Urban areas; for e.g. tall, dense man-made
structures
Tall grass, tall weeds, and crops; e.g. hay, corn,
and wheat fields
Brush lands and short trees; e.g. chaparrals,
mesquite
Forested areas, fully covered by trees; e.g.
hardwoods, conifers, mixed forests
Previous
Reporting
group
FVA
SVA
Current
reporting
group
NVA
SVA
SVA
SVA
VVA
LiDAR Check-Points
Accuracy Assessment
•
•
•
•
The no. of check point calculation comes from the new ASPRS specs
Instead of the FVA, SVA and CVA there are new terms now called the NVA and VVA
No of check points required is based on the square miles of the project
There is no clear guidance of the horizontal accuracy testing of LiDAR data.
Vertical and Horizontal Accuracy Testing of Elevation Data sets
Project Area (square
Number of Static 3D
Number of Static 3D
Total Number of Static
kilometers)
Check Points in NVA
Check Points in VVA
3D Check Points
<500
201-750
751-1000
1001-1250
1251-1500
1501-1750
1751-2000
2001-2250
2251-2500
20
20
25
30
35
40
45
50
55
5
10
15
20
25
30
35
40
45
25
30
40
50
60
70
80
90
100
Classification Levels
Code
Description
1
Processed, but unclassified
2
Bare earth
7
Low noise
9
Water
10
Ignored ground (near a breakline)
17
Bridge decks
18
High noise
• Classes 17 and 18 are new mandatory classes introduced
• All deliveries in LAS 1.4 format
• Tolerance for Misclassification
• QL-0 and QL-1: 0.5%
• QL-2: 1%
• QL-3: 2%
DEM Spacing
Quality Level (QL)
QL0
QL1
QL2
QL3
Minimum cell Minimum cell
size (m)
size (ft)
0.5
0.5
1
2
1
1
2
5
Hydro Flattening/Enforcement
• No changes in the
tolerances of double-line
streams and water-bodies
– 100 ft. wide rivers/streams
– 2 acre water-bodies
• Provision for single line
streams added
– No guidelines provided.
Collection of single line
streams depends on the
special requirements from
the customer
Hydro Flattening Vs. Enforcement
•
Hydro-flattened describes the specific type
of DEM required by the USGS National
Geospatial Program (NGP) for integration
into the NED.
•
A traditional topographic DEM such as the
NED represents the actual ground surface,
and hydrologic features are handled in
established ways.
•
In HF surface Roadways crossing drainages
passing through culverts remain in the
surface model because they are part of the
landscape (the culvert beneath the road is
the manmade feature). Bridges, manmade
structures above the landscape, are
removed.
– In HF: Terrain topography gets
preference
– In HE: Hydro-features get preference
over terrain features
Hydro-Flattening Examples
Lidar only
Hydro-flattened Lidar
Hydro Flattened Breaklines
Base Earth LiDAR
No Hydro Breaklines
• DEM created only using
Bare-Earth LiDAR points.
• Surface contains extensive
triangulation artifacts
(“tinning”).
• Cause by the absence of
LiDAR returns from water.
• Breakline constraints that
would define buildings,
water, and other features.
• Aesthetically and
cartographically
unacceptable to most users.
Triangulation in Water
Hydro Flattened
Topographic Surface
• Removes the most offensive pure
LiDAR artifacts: those in the water.
Constant elevation for water
bodies.
• Wide streams and rivers are
flattened bank-to-bank and forced
to flow downhill (monotonic).
• Carries ZERO implicit or explicit
accuracy with regards to the
represented water surface
elevations – Typically used for
Cartographic/Aesthetic
enhancement.
• Most often achieved via the
development and inclusion of hard
breaklines.
Water Body
Stream
Hydro Flattened
Monotonicity (downhill order)
Flat Water Bodies
Hydro Enforced
Hydrologic Surface
• Surface used by engineers in
Hydraulic and Hydrologic (H&H)
modeling.
• NOT to be used for traditional
mapping (contours, etc.)
• Similar to Hydro Flattened with
the addition of Single Line
Breaklines: Pipelines, Culverts,
Underground Streams, etc…
• Terrain is then cut away at bridges
and culverts to model drain
connectivity.
• Water Surface Elevations are
often set to known values
typically from surveying methods.
Culverts Cut Through Roads
Automated Building Extraction: Example 1
Automated 3D Building Extraction
25
Thanks