Digital Mapping of Soil Profile Texture of the Monteregie Agricultural Area (QC, Canada) Using Analytical and Morphological Soil Legacy Data.

In order to derive digital soil maps of the sand, silt, clay and coarse fragment (CF) contents of the topsoil (0-30 cm), subsoil (30-60 cm) and substratum (60-100 cm) in the Monteregie agricultural area (1500 km²), near Montreal (QC, Canada), soil legacy data (morphological and analytical soil profile data) were used for kriging and co-kriging. The main objective of this study was to assess the usefulness of the morphological soil database (MSDB), in terms of reducing the root mean square error of prediction (RMSEP) of each soil analytical properties when implementing the global soil map of Canada (project GlobalSoilMap.net). The MSDB has been collected during soil survey works realized in this area by Agriculture and Agri-Food Canada from 1982 to 2009 (n= 44479 soil profiles). In this database, soil texture has been recorded at the subclass level (19) according to the Canadian System of Soil Classification standards. These semi-quantitative data has been converted into percent of sand, silt, and clay content using the median value of each soil textural subclass. The analytical soil database (ASDB) included 3209 soil samples for topsoil and 996 for subsoil and substratum. Anisotropic and isotropic semivariograms, block kriging (2 x 2) and co-kriging have been computed using GS⁺ (V9). No significant anisotropy has been detected. For both data sources (ASDB and MSDB), strong spatial structure were found for clay and sand contents as indicated by the relatively low nugget to sill ratio (C₀/C₀+C <0.3). Lower spatial structure (C₀/C₀+C >0.3) were obtained for silt and coarse fragment contents. Cross-validation using (Jackknife approch) showed similar prediction accuracy (8-14%) for both data sources. Co-kriging of analytical data using the MSDB as co-variables improves prediction accuracy. When available, morphological soil data recorded during conventional soil survey works should be used for deriving digital maps of soil profile texture.