Digital Soil Texture Maps of Argentina and Their Relationship to Soil-Forming Factors and Processes

Abstract

Soil texture is determined by the parent material of the soil and the resulting pedogenetic processes. Because the spatial distribution of soil texture governs several soil physical and chemical processes, our goals were to map clay, silt, and sand at different depths using the SISINTA soil profile database; generate textural soil class maps for the same soil depths; and describe the distribution of soil texture in relation to different landscape units. We used 4663 soil texture observations and 64 environmental covariates to represent the soil-forming factors (e.g., remote sensing data, climate data or geomorphology maps). We modelled clay, silt and sand at 0-15, 15-30, 30-60 and 60-100 cm, establishing an empirical relationship with environmental covariates using Random Forest to predict their spatial distribution. Finally, we performed an analysis of uncertainty through repeated cross-validation. We observed model efficiency coefficient (MEC) values between 0.452 and 0.557, with an RMSE between 8.77% and 11.21% for the clay fraction. The MEC for the silt fraction ranged from 0.561 to 0.638 with an RMSE of 10.50% to 12.01%. For the sand fraction the MEC ranged from 0.587 to 0.640 with RMSE values between 16.19% and 16.76%. The general patterns of uncertainty are consistent with areas of limited data. Our results increased the quality, quantity and accessibility of information on soil texture in Argentina by providing new insights into both the distribution of parent materials and the intensity of pedogenetic processes in each region.