Digital mapping of the soil thickness of loess deposits over a calcareous bedrock in central France

Soil thickness (ST) plays an important role in regulating soil processes, vegetation growth and land suitability. Therefore, it has been listed as one of twelve basic properties to be delivered GlobalSoilMap project. However, ST prediction reported with poor performance previous studies. Our case study is located the intensive agriculture Beauce area, central France. In this region, mainly depends on loess (TOL) deposits over a calcareous bedrock. We attempted test TOL by coupling large dataset (10978 sampling sites) 117 environmental covariates. After variable selection recursive feature elimination, quantile regression forests (QRF) was employed for spatial modelling, able directly provide 90% intervals (PIs). Averaging total 50 models, generated repeated stratified random sampling, showed substantial model mean R2 0.33, RMSE 30.48 cm bias ?1.20 cm. The interval coverage percentage that 86.70% validation samples fall within predefined PIs, which also indicated uncertainty produced QRF reasonable. relative importance airborne gamma-ray radiometric data Sentinel 2 products prediction. map PIs makes sense from science physiographic point view. final product can guide evidence-based decision making agricultural management, especially irrigation our study.