national cooperative soil survey conference 2007 madison, wi soil spectroscopy for rapid and...
TRANSCRIPT
National Cooperative Soil Survey Conference 2007 Madison, WI National Cooperative Soil Survey Conference 2007 Madison, WI
Soil Spectroscopy for Rapid and Cost-Effective Soil Mapping Across Larger Landscapes
Sabine Grunwald
NRCS-CESU (Carolyn G. Olson)“Linking Experimental and Soil Spectral Sensing for Prediction of Soil Carbon Pools and Carbon Sequestration at Landscape Scale”Investigators: Grunwald S., J.O. Sickman and N.B. ComerfordGraduate student: Gustavo M. Vasques
NASA
Soils store ~ 3 times more C than biosphere (vegetation)~ 2 times more C than atmosphere~ 1.5 time more C than surface ocean
NASA
Temp.
+
Santa Fe River Watershed
Laboratory analyses:Total, stable and labile carbon pools
Visible near-infrared diffuse reflectance spectroscopy
Regress soil analytical and soil spectral data
(chemometrics)
GIS-based quantitative soil-landscape modeling(environmental correlation)
Maps: land use,geology,topography,etc.
Predicted carbon pools derived fromsoil-landscape modeling and analytical data
Predicted carbon pools derived from soil-landscape modeling and soil spectral data
]~,[ˆ txS ichemTC
]~,[ˆ txS ichemstable
]~,[ˆ txS ichemlabile
]~,[ˆ txS isllabile
]~,[ˆ txS islstable
]~,[ˆ txS islTC
Surface Geology Coverage in the Santa Fe Watershed
Data sources:Geology: Florida DEPWatershed boundary: SRWMD
¯0 7.5 15 22.5 303.75Kilometers
Depth of Layer 1 in the Soils of the Santa Fe Watershed
SourceSoil data: NRCS, SSURGO datasetWatershed Boundary: SRWMD
¯0 5 10 15 202.5Kilometers
Surface Geology Coverage in the Santa Fe Watershed
Data sources:Geology: Florida DEPWatershed boundary: SRWMD
¯0 7.5 15 22.5 303.75Kilometers
Surface Geology Coverage in the Santa Fe Watershed
Data sources:Geology: Florida DEPWatershed boundary: SRWMD
¯0 7.5 15 22.5 303.75Kilometers
Depth of Layer 1 in the Soils of the Santa Fe Watershed
SourceSoil data: NRCS, SSURGO datasetWatershed Boundary: SRWMD
¯0 5 10 15 202.5Kilometers
Environmental landscape factors
Indicators of locally operating
ecosystem processes
C, N and Pmineralizablepools
Soil Carbon Sequestration1
Carbon (C) pools:• Total• Recalcitrant• Labile
Santa Fe River Watershed
Laboratory analyses:Total, stable and labile carbon pools
Visible near-infrared diffuse reflectance spectroscopy
Regress soil analytical and soil spectral data
(chemometrics)
GIS-based quantitative soil-landscape modeling(environmental correlation)
Maps: land use,geology,topography,etc.
Predicted carbon pools derived fromsoil-landscape modeling and analytical data
Predicted carbon pools derived from soil-landscape modeling and soil spectral data
]~,[ˆ txS ichemTC
]~,[ˆ txS ichemstable
]~,[ˆ txS ichemlabile
]~,[ˆ txS isllabile
]~,[ˆ txS islstable
]~,[ˆ txS islTC
Anthropogenic and natural forcing functions
Transfer of total, recalcitrant andlabile C imprints into landscapeselucidates onmechanisms that induce C storage/change
Assess the actualpools and potentialC sequestrationpotential
Soil Mapping 1
• Accurate • Rapid • Cost-effective
• Large soil regions• Relevant soil properties
StatisticsGIScience
Soil &Environmental
Sciences
Carto
graphy Quantitative
methods
Geo-statistics
Digital Soil Mapping1
• GIS - multi-scale data integration• Complex geospatial methods
Environmental datasets:• Field and analytical data• Soil sensing • Remote sensing
Soil-landscape models:• Functional (stochastic; deterministic)• Mechanistic (simulation)
Grunwald S. (ed) 2006. Environmental Soil-Landscape Modeling–Geographic Information Technologies & Pedometrics. CRC Press, New York.
Visible/near-infrared spectroscopy (VNIRS) is a fast, cheap and accurate alternative for the investigation of soil properties, and is now recognized as a powerful analytical tool in soil science
Soil Sensing2
Each soilhas specificreflectance signature
Modern soil surveyingModern soil surveying
Spectroradiometer2
QualitySpec® Pro (Analytical Spectral Devices Inc., Boulder, CO)
Site #38 layer 2
Site #412 layer 4
Site #923 layer 2
Reflectance
TC: 7,132 mg kg-1 (avg.)Pine plantationTypic Aquods
TC: 268,995 mg kg-1
WetlandTypic Argiaquolls
120-180 cm
30-60 cm
60-120 cm
Visible/Near-infrared Diffuse Reflectance Spectroscopy2
Total carbon (TC): 169 mg kg-1
Upland forestTypic Quartzipsamments
Soil Study – Santa Fe River Watershed, Florida3
Objectives:
• Investigate the usefulness of VNIRS for rapid and
accurate assessment of soil carbon
• Understand the linkages between labile, recalcitrant
(stable) and total organic carbon
• Assess the usefulness of VNIRS to map larger soil-
landscapes
0 10 20 30
Kilometers±
Land use categoryPine plantationUpland forestWetlandWaterImproved pastureCropUrbanOther
0 10 20 30
Kilometers±Elevation (m)
90.90 1.49
Data sources maps:DEM: National Elevation Model (US Geological Service)Land use: Florida Fish and Wildlife Conservation Commission (2003)Geology: FL Dept. of Environmental Protection Soil Orders: Soil Survey Geographic Database (SSURGO) Natural Resources Conservation Service
Santa Fe River Watershed, Florida 3
Methodology3
Laboratory soil data
VNIR spectral dataPre-treatment:Log-normalization using base-10 logarithmTesting of 30 different preprocessing transformations
Identify relationshipsIdentify relationships
Complete dataset
Methods: Stepwise Multiple Linear Regression (SMLR) Principal Components Regression (PCR) Partial Least-Squares Regression (PLSR) Regression Tree (RT) Committee Trees (CT) (bagging)
~70%of data
Modeldataset
~ 30% of dataused to test accuracyof model predictions
Validationdataset
predictions R2, RMSE
Total Carbon (TC)3
Samplingacrossland use-soil ordertrajectories
Layer 1(0-30 cm)
Layer 2(30-60 cm)
Layer 3(60-120 cm)
Layer 4(120-180 cm)
Observations (n) 143 143 141 135
Mean 14,872 8,105 3,929 1,659
Std. Error of Mean 1,828 2,127 1,111 182
Median 10,529 3,705 1,808 1,087
Mode 2,670 932 384 169
Std. Deviation 21,867 25,434 13,198 2,117
Skewness 6.361 8.503 7.998 5.021
Kurtosis 47.10 81.28 64.43 34.19
Range 199,318 268,062 112,725 18,749
Minimum 2,670 932 384 169
Maximum 201,988 268,995 113,109 18,917
Total Carbon (TC) [mg/kg]3
Mean
Min.
Max.
+ 1 SD
- 1 SD
Wavelength (nm)
Spectral scans of 554 soil samples collected in the SFRW at 4 different soil depths (0-30, 30-60, 60-120 and 120-180 cm)
VNIR Scanning3
Results: Prediction Performance - logTC [mg/kg]4
Calibration ValidationR2 RMSE R2 RMSE
SMLR 0.91 0.149 0.85 0.176PCR 0.83 0.212 0.83 0.189PLSR 0.86 0.190 0.86 0.176RT 0.98 0.149 0.76 0.226CT 0.97 0.087 0.86 0.170
[30 pre-processing methods were tested]
PCR
Validation Results: Prediction Performance - logTC [mg/kg]4
SMLR
[pre-processing: Savitzky-Golay 1st-derivative using a 1st-order polynomialwith search window 9 (SGF-1-9)
[pre-processing: standard normal variate transformation (SNV)]
Laboratorymeasurements
VNIRPredictions
PLSR RT
Validation Results: Prediction Performance - logTC [mg/kg]4
[pre-processing: Savitzky-Golay 1st-derivative using a 3rd-order polynomial with search window of 9 (SGF-3-9)]
[pre-processing: Norris gap derivative with a search window of 5 (NGD-5)]
CT
Validation Results: Prediction Performance - logTC [mg/kg]4
[pre-processing: Norris gap derivative with a search window of 7 (NGD-7)]
(mg/kg)
TOC HC RC DOC07 DOC02
Observations 141 141 141 141 141
Minimum 2,670 37 1,150 214 221
Maximum 201,988 29,399 181,738 9,000 8,995
Median 10,529 2,892 7,382 644 664
Mean 14,828 3,707 11,122 799 809
Std. Deviation 21,993 3,292 19,194 827 818
Total Organic Carbon and Carbon Fractions (0-30 cm)5
TOC: Total organic carbonHC: Hydrolysable carbon (after digestion with 6N HCl) - Thermo Electron FlashEA Elemental AnalyzerRC: Recalcitrant carbon was calculated as the difference between TOC and HCDOC: Dissolved organic carbon Shimadzu TOC Analyzer after hot water extraction, then filtered into 2 classes: <0.7 µm (DOC07) and <0.2 µm (DOC02).
SOC and fractions
Best modelCalibration Validation
Rc2 RMSEC Rv
2 RMSEV
TOC LOG-PLSR 0.93 0.082 0.86 0.078
HC SAV-PLSR 0.49 0.218 0.40 0.285
RC SAV-PLSR 0.90 0.109 0.82 0.108
DOC07 SAV-PLSR 0.89 0.100 0.84 0.086
DOC02 SNV-PLSR 0.81 0.110 0.69 0.100
Statistics – Total Organic Carbon and Carbon Fractions5
LOG: Log (1/Reflectance) SAV: Savitzky-Golay smoothing, and averagingSNV: Standard normal variate transformation
VNIRS vs. Conventional Lab Analysis6
Characteristics VNIRS Conventional
Ease of sample preparation
+++ +
Ease of analysis + +
Speed +++ +
Labor +++ ++
Equipments cost + +
Use of supplies +++ +
Cost per sample +++ +
Accuracy +++ +++
VNIRS & Landscape Scale Modeling7
Organic matter (OM)lab measurements
0-30 cm 0-30 cm
OM predictions using VNIRspectral datamethod: committee trees (boosting)
0-30 cm 0-30 cm
Calibr. ValidationR2 0.94 0.83RMSE 0.49 0.77
VNIRS & Landscape Scale Modeling7
Semivariograms of OM measurements and VNIRS predictions show very similar spatial autocorrelation structure
OM lab measurements OM derived from VNIRS
VNIRS & Landscape Scale Modeling7
OM map derived using lab measurements
OM map derived from VNIR spectra
Error Sand Silt Clay
Mean -0.38 -0.50 0.40
Min -14.20 -13.24 -3.12
Max 17.21 7.30 3.05
RMSE 4.88 3.69 1.02
Lamsal S., PhD thesis
Soil Texture Geospatial Modeling - SFRW7
Soil 0-30 cmMethod: Spatial stochastic simulation
National and Global VNIRS Applications8
Locations / soil-landscape settings:• Africa (Shepherd and Walsh, 2002)• Australia (Dalal and Henry, 1980)• Australia (Viscarra Rossel et al., 2006)• Brazil (Masserschmidt et al., 1999)• Netherlands (Koistra et al., 2003)• USA, Maryland (Reeves III et al., 2001)• Global (USA & Africa) (Brown et al., 2006) …. many more
Soil properties:• Carbon; organic matter• Texture • Nutrients (N, P, Mg, Ca, …)• Metals (Fe, Al,…..)• CEC• BD …. many more
VNIRSVNIRS
Conclusions8
NASA
Soil mapping &expertise
Soil and remotesensorsincl. VNIRS
Soils – environmentalfactors(GIS; soil-landscape analysis)
Soil science- global context