Digital Soil Mapping Basic concepts and applications in

Digital Soil Mapping: Basic concepts and applications in the Atcha project Philippe Lagacherie & Samuel Buis ATCHA project kick off meeting 14 nov 2016, Bangalore (India)

Digital Soil Mapping : A brief history n 1991 -1993: Publications of pioneer works (USA, Australia, France, …) 2003: Digital Soil Mapping as a body of soil science 2004: First International workshop on Digital Soil Mapping § 2004: IUSS Working group on Digital Soil Mapping n n Workshops : Rio (2006), Logan (2008), Rome(2010), Sydney, (2012), Nanjing, (2014), Aarhus, (2016) § 2009: Projet Global. Soil. Map Sanchez et al, Science, 2009 Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 2

Global. Soil. Map specifications Soil properties Depth Intervals Depth to Bedrock (cm) Effective rooting Depth cm) Clay, Silt and sand content (g/kg) Coarse fragment content (%vol) Organic Carbon content (g/kg) p. H (x 10) Cationic Exchange capacity (mmolc/kg) Available Water Capacity (mm) Electrical Conductivity (m. S/m) Bulk Density(Mg/m 3) Spatial Resolution Uncertainties 90% confidence interval 3

Digital Soil Mapping : The general principle S = f ( S, C , O , R , P, A, N ) + e (Mc. Bratney et al, 2003) Soil Climate Organisms Relief Parent material Age Location x, y Exhaustive Spatial data related with soil (e. g. DEM, LC images, climate model outputs, existing soil maps, …) Legacy soil data New sampling of sites with soil data Soil sensing (proximal, remore sensing) Spatial inference models (e. g. machine learning algorithms, geostatistic models, expert rules) Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 4

DSM in the Atcha project n n INPUT l Spatial exhaustive variables related with soil properties Ø DEM (at least SRTM 90 m resolution) Ø Existing soil maps : KSRSAC 1: 50 0000, NBSS 1: 10 000 (Sujala III Watershed only) ? Ø LULC and vegetation indexes images (ATCHA Task 1. 1. ) Ø Coarse resolution soil moisture maps (from AICHA project) Ø Other…. l Soil data Ø ~ 60/75 sites with measured soil properties (texture, bulk density, soil depth) Ø NBSS sites with legacy soil profiles ( from Sujala III project) (? ) Ø n sites (pixels) with AWC components (n >> 60) estimated from inversion of vegetation model and satellite images Ø p bare soil sites with surface soil property estimations from Vis-SWIR multi/hyperspectral image ? DSM OUTPUT l Predicted soil properties : AWC components , soil texture? l Spatial extent : Berambadi watershed ? , cultivated part? Subwatershed? (e. g. Sujala III) l Spatial resolution: GSM grid? Less ? More? Plots? Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 5

DSM in the Atcha project 1. “Business as usual” n INPUT l Spatial exhaustive variables related with soil properties Ø DEM (at least SRTM 90 m resolution) Ø Existing soil maps : KSRSAC 1: 50 0000, NBSS 1: 10 000 (Sujala III Watershed only) ? Ø LULC and vegetation indexes images (ATCHA Task 1. 1. ) Ø Coarse resolution soil moisture maps (from AICHA project) Ø Other…. l Soil data Ø ~ 60/75 sites with measured soil properties (texture, bulk density, soil depth) Ø NBSS sites with legacy soil profiles ( from Sujala III project) (? ) Ø n sites (pixels) with AWC components (n >> 60) estimated from inversion of vegetation model and satellite images Ø p bare soil sites with surface soil property estimations from Vis-SWIR multi/hyperspectral image ? n DSM Output l Output data : AWC components , soil texture? l Spatial extent : Berambadi catchment ? , cultivated part? Subwatershed? (e. g. Sujala III) l Spatial resolution: GSM grid? Less ? More? Plots? Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 6

First example : Digital Soil Mapping from legacy data (e. g. Vaysse et al, 2015) S = f ( S, C , O , R , P, A, N ) + e (Mc. Bratney et al, 2003) Soil: Soil Climate Organisms The set of GSM soil properties Relief Parent material Age Location x, y Spatial data related with soil (soil covariates) DEM, geology map, land use map, climate mod outputs, small scale soil maps Legacy measured profiles (1/13. 5 km 2) Spatial inference models : Quantile Random Forest (bagged regression tree) Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 7

Digital soil maps of soil properties in Languedoc-Roussillon (27 000 km 2) Estimated p. H (5 -15 cm depth) Digital Soil Mapping. P. Lagacherie, S Buis Uncertainty map (confidence interval width) ATCHA KO meeting 14 nov 2016 Bangalore (India) 8

DSM in the Atcha project 2. Methodological advances n n INPUT l Spatial exhaustive variables related with soil properties Ø DEM (at least SRTM 90 m resolution) Ø Existing soil maps : KSRSAC 1: 50 0000, NBSS 1: 10 000 (Sujala III Watershed only) ? Ø LULC and vegetation indexes images (ATCHA Task 1. 1. ) Ø Coarse resolution soil moisture maps (from AICHA project) Ø Other…. l Soil data Ø ~ 60/75 sites with measured soil properties (texture, bulk density, soil depth) Ø NBSS sites with legacy soil profiles ( from Sujala III project) (? ) Ø n sites (pixels) with AWC components (n >> 60) estimated from inversion of vegetation model and satellite images Ø p bare soil sites with surface soil property estimations from Vis-SWIR multi/hyperspectral image ? DSM Output l Output data : AWC components , soil texture? l Spatial extent : Berambadi catchment ? , cultivated part? Subwatershed? (e. g. Sujala III) l Spatial resolution: GSM grid? Less ? More? Plots? Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 9

Second example : Digital Soil Mapping with adding of imaging spectroscopy estimations of soil properties (e. g. Ciampalini et al, 2012) S = f ( S, C , O , R , P, A, N ) + e (Mc. Bratney et al, 2003) Soil Climate Organisms Relief Parent material Age Location x, y Spatial sampling of topsoil clay content measurements hyperspectral imagery estimations of topsoil clay content Spatial inference models : Co-kriging Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 10

Co-kriging of soil properties with Vis-NIR hyperspectral covariates in the Cap Bon region (Tunisia). Ciampalini et al, 2012 Confidence interval map Coregionalization model Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India) 11

DSM in Atcha project 3. Link between the approaches DSM “Business as usual” (example 1) Rough estimations of soil properties with confidence intervals Improved DSM (example 2) Digital Soil Mapping. P. Lagacherie, S Buis Drive additional sampling Constraint for inversion Inversion of STICS model n sites with estimations of AWC components ATCHA KO meeting 14 nov 2016 Bangalore (India) 12

Data available / needed for inversion n STICS calibration: l l n 2010 -2013 monitored plots LAI, SM, farming technics for maize for the monitored plots with soil measurements for STICS recalibration on currently used varieties (20142017)? STICS inversion: l l l Past survey, new survey (fert. date and qty, sowing date, variety)? Remotely-sensed LAI / SSM 2011 -2013, 2014 -2017 (2018 ? ) Land Use (identification of maize plots): 2014 -2017 (2018 ? ) : WP 1. 1 + human identification? Ø Possible for 2010 -2013 ? Ø • Finer estimation weather ATCHA data. KO (meeting rain 14, nov ET) 2011 Digital Soil Mapping. P. Lagacherie, of S Buis 2016 for Bangalore (India)

Partnership / Link with other tasks n Partnership: LISAH/EMMAH/IISc/NBSS n Link with other tasks l l WP 1. 1 Land Use, Remotely sensed LAI-SSM WP ? Digital Soil Mapping. P. Lagacherie, S Buis ATCHA KO meeting 14 nov 2016 Bangalore (India)