Soil Spectral Library Importance and Utilization for Food

Transcription

Soil Spectral Library Importance and Utilization for Food
Soil Spectral library
Importance and Utilization for Food Security
Eyal Ben-Dor
Department of Geography and Human Environme
Tel-Aviv University
[email protected]
http://www.tau.ac.il/~geograph/bendor
OUTLINE
• What is Soil Spectral Library ?
• Reflectance spectroscopy - is a technique which measures
the ratio between irradiance and radiance light with many
number of spectral and narrow channels
• Point spectroscopy - Measuring the reflectance of a small area
at once
• Image spectroscopy – Measuring the reflectance of many points
that can merged into image
• Important of Soil Spectral Library
Definition
• Soil Spectroscopy refers to the reflectance part of the
electromagnetic radiation that interacts with the
soil (surface) matter across the VIS-NIR-SWIR spectral
region (the sun illumination) range.
Reflectance ( r )
r = Ir/ Is
Is
Ir
r
Wavelength
Definition
Hyperspectral Remote Sensing (HSR)
HSR
From A. Goetz 1994
Simultaneous acquisition of images in many registered spectrallyhigh resolution continuous bands at selected (or all) spectral
domains across the UV-VIS-NIR-SWIR–MWIR-LWIR spectral region
(0.3-12mm)
Strong Link between Point and Image
Spectroscopy
Image Spectroscopy
Point Spectroscopy
Point
fingerprint
Flower
Talc
Clay powder
Beyond visible capability
Image
fingerprint
VIS -RGB
SWIR -RGB
How point spectroscopy driven the image
spectroscopy of soils
Number of papers and Patents
Point Spectroscopy
Image Spectroscopy
Reviewing here only
key papers
1970 1980 1990 2000
Year
2010
2020
5 spectral types
in USA
1984
Stoner, E.R. and M.F., Baumgardner, 1981. Characteristic
variations in reflectance of surface soils. Soil Science
Society of American Journal 45: 1161-1165
Iron
Oxides
Main Soil
Chromophores
Particle
Size
1994
Hygroscopic
Water
Organic
Matter
Clays
Organic
Matter
Calcite
Ben-Dor E., and A. Banin 1995b, Near infrared analysis
(NIRA) as a Simultaneously method to evaluate spectral
featureless constituents in soils., Soil Science 159:259-269
Reflectance spectrum of soil is a complex of
Chemical and Physical issues
VIS
SWIR
NIR
K6
S16
C6
E1
H5
A3
0.5
1.0
1.5
Wavelength (mm)
2.0
2.5
Reflectance Spectroscopy is an Inherent property of the object which
is not related to temperature, humidity, position ect.
Thus –it is a perfect property to use for remote sensing
Figure 10
Spectral Assignments for Soil Constituents
Soil Spectral Library :
Soil samples, at storage, with wet chemistry analytical data of their attributes
plus reflectance spectra measured under a well valid protocol process
Soil Attributes
Soil Spectra Files
Sample Location
A1
34,5467.67
36,654,32
OM
2.4 %
Clay
34%
Lime….
23.4%
Various Methods, various spectrometer
Replacing Wet Analysis of Soil by Spectroscopy
A portable spectrometer
Data Mining and Soil Spetral Model : NIRA
NIRS/NIRA is a chemomtrics method where NIR-SWIR spectral region is used to predict chemical
constituents of a given matter.
b
calibration
First developed for Food Science:
Ben-Gera, I., and K.H. Norris, 1968, Determination of moisture content in soybeans by direct
spectrophotometry. Israeli Journal of Agriculture Research. 18:124-132.
Quantitative Method for spectral based SOIL properties
Dalal, R.C., and R.J. Henry. 1986. Simultaneous determination of moisture, organic carbon and total
nitrogen by near infrared reflectance spectroscopy. Soil Science Society of America Journal 50:120-12
It is possible to use spectral reflectance data of soils along with data mining
techniques to model the soil attributes and replace wet laboratory highly
labor activity
Predication
Validation
NIRS in 6 TM bands
Suggests possible utilization with Sentinel
2 Data
Ben-Dor E. and A. Banin, 1995, Quantitative analysis of convolved TM spectra of soils in the visible, near
infrared and short-wave infrared spectral regions (0.4-2.5mm). International Journal of Remote Sensing,
18:3509-3528.
• Soil Spectral Library with Data Mining Approach provides spectral
models to assess the soil attributes from laboratory, field, air and
space borne.
• Accordingly SSL can be used to map soils from both ground and
air domains in a very efficient way never before done
• With the emerging of new HSR sensors from both air and space
domains SSL will play a major role in developing new applications for
soil management in a new innovative way
The First SSL
1984
http://www.tau.ac.il/~rslweb/slis.html
91 soils with 50 soil attributes
60 soil with 7 attributes
2mm>
Soil Spectroscopy Course , Brno Czech Republic June 26-27,2013
Possible Utilization
The global vis-NIR library so
far…
Spectra from 46 countries
Soil Spectral Library
http://groups.google.com/group/soil-spectroscopy/files
Soil Spectroscopy Course , Brno Czech Republic June 26-27,2013
• http://groups.google.com.au/group/soil-spectroscopy
The LUCAS spectral library
Current status:
 23 European countries
 ~20,000 high quality spectral
readings
 Metadata: Clay, silt, sand, OC, pH,
CEC, CaCO3, Geographical
coordinates, land use, etc
Creation of four subsets: Cropland,
Grassland, Woodland, and Organic soils
Possible Solution
Data Bank Patterns :
Brazilian Spectral
Library
Principal
components
of Brazilian
Data Bank
Bellinaso (2009)
Greece Spectral Library
Clay Modeling Using PARACUDA - Preliminary
Traini Test
ng R2 R2
RMSE
P
SEP Bias SDtY RPD
0.534 0.598 6.932 6.943 0.219 10.92 1.601
7
0
6
5
5
83
6
Sand %
Silt %
Clay %
NO3 ppm
59.24
26.03
14.79
19.28
Minimum
2.00
0.00
0.00
0.00
Maximum
99.00
68.00
91.00
661.20
Mean
First Running of the PARACUDA on
the entire (~1000 samples)
dataset, without ISS correction
shows very promising results with
extremely high significance
Preprocessing used:
SNV+Final Smoothing
Important Spectral Range
Identification
VIP - Important Wavelengths
Nois
e
2.5
Nois
e
Unknown Important
Feature
Sig for Model
2
1.5
1
0.5
0
350
850
1350
Wavelength
1850
2350
Observed
Predicted
Argila, %
66
A.
B.
56
Spatial
Spectral
information
46
Clay
36
26
16
6
N
Areia, %
90
C.
D.
70
Sand
50
30
10
0
2000
3000 m
100
A.
R2Ajust.= 0.96
RMSE = 3.26
ME = -0.22
m = 0.87
Intercesção = 3.03
60
Areia % - Mapa espectro-digital
Argila % - Mapa espectro-digital
80
1000
40
20
R2Ajust.= 0.97
B.
RMSE = 4.38
ME = -0.04
m = 0.86
Intercesção = 10.03
80
60
40
20
0
0
0
20
40
60
Argila % - Mapa digital
80
0
20
40
60
80
Areia % - Mapa digital
100
Ramirez-Lopes
e Demattê (in press 2009)
Compaction
Detection
0-20 cm, RP soil, Dry
a.
0,25
Higher intensity for Compacted samples
Reflectance Factor
0,20
Cd
Compacted: less pores
0,15
NCd
0,10
0,05
Non compacted: more pores
0,00
450
650
850
1050 1250 1450 1650 1850 2050 2250 2450
Wavelength, nm
Demattê et al.
(in press, IJRS 2009)
Ideas: Real Time Profile information
Pixel
Portable Hyperspectral
systems
Fe2O , TiO2 , Al2O3, Cor, Conhecimento
experto
etc
+
≈
In situ soil
class
determination
Precision Agriculture: On-the-go vis-NIR proximal sensors
VERIS Tech.
Mouazen et al.
Erosion Detection
0,35
0,3
a: LVA
0,25
0,2
0,15
0,1
0,05
0
400
700
p1a (nula)
p1b (ligeira)
p1c (moderada)
p1d (severa)
1000
1300
1600
1900
2200
2500
2800
Demattê & Fotch (1998)
Ideias: Pedologia Espectral em tempo real
0.80
0.75
LPP16B
LPP16C
LPP16D
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
500
750
1000
1250
1500
1750
2000
Probabilidade
Softwares “inteligentes”
Quantitavivo
Soil Spectral library
0.80
0.80
0.80
0.75
0.75
LPP05A
LPP05B
LPP05C
0.70
0.65
0.75
LPP06A
LPP06B
LPP06C
LPP06D
0.70
0.65
0.65
0.60
0.60
0.55
0.55
0.55
0.50
0.50
0.50
0.45
0.45
0.45
0.40
0.40
0.40
0.35
0.35
0.35
0.30
0.30
0.30
0.25
0.25
0.25
0.20
0.20
0.15
0.15
0.10
0.10
0.05
0.05
0.00
0.00
500
750
1000
1250
1500
1750
2000
2250
2500
LPP12A
LPP12B
LPP12C
0.70
0.60
0.20
0.15
0.10
0.05
0.00
500
750
1000
1250
1500
1750
2000
2250
2500
500
750
1000
1250
1500
1750
2000
2250
2500
2250
2500
Soil Profile not Visible
3D view of the SSA property in the study
field
Combining field
and
image spectroscopy
Soil mapping: spatial variation determination
Detailed Soil Map (Field method)
Soil Map by aerial photographs+punctual spectral
information
Demattê et al. (2001)
Several Soil Spectral Library are existed
world wide
• Each were measured with a different protocol and spectrometers
• Still problematic to use as is for robust utilization
• The primary goal is to help and exploit HSR satellite data for
monitoring food security and increasing food production.
Soil Spectroscopy Course , Brno Czech Republic June 26-27,2013
SHALOM Products for Costumers
Product Name
Crop, Rangeland and Invasive Species Map
Burnt Area Map
Vegetation Status Indicators
Vegetation Damage and Stress Indicators
Fire Fuel Map
Mineral Map
Coastal Bathymetry Map
Urban And industrial Functional Area Map
Lithological Map
Lava Flow Parameters
Soil Surface Pollutants Map
Volcanic Gas And Aerosol Emission Map
Forest Species Map
Forest Biomass Map
Ice Cover Map
Soil Characterization Map
Land Cover Map
Land Cover Change Detection Map
Snow Cover Map
Forest Nitrogen and Chlorophyll Map
Wetlands Classification Map
Marine And Aquatic Quality And
Productivity Indicators
Lava and ash distribution map
Snow And Ice Cover Characterization
Soil Characterization Map
In each of us, a little man is hidden. This man is our imagination:
use it to further exploit the spectral information of soil for the
benefit of mankind
Hyperspectral Remote Sensing
IMAGING SPECTROSCOPY -- Imagination before Technology
Width (km)
think cube
Length km)