Generating soil thickness maps by means of geomorphological-empirical approach and random forest algorithm in Wanzhou County, Three Gorges Reservoir

Soil thickness, intended as depth to bedrock, is a key input parameter for many environmental models. Nevertheless, it often difficult obtain reliable spatially exhaustive soil thickness map in wide-area applications, and existing prediction models have been extensively applied only test sites with shallow depths. This study addresses this limitation by showing the results of an application section Wanzhou County (Three Gorges Reservoir Area, China), where varies from 0 ∼40 m. Two different approaches were used derive maps: modified version geomorphologically indexed (GIST) model, purposely customized better account peculiar setting site, regression performed machine learning algorithm, i.e., random forest, combined geomorphological parameters GIST (GIST-RF). Additionally, errors two quantified, validation geophysical data was carried out. The showed that model could not fully contend high spatial variability area: mean absolute error 10.68 m root-mean-square (RMSE) 12.61 m, frequency distribution residuals tendency toward underestimation. In contrast, GIST-RF returned performance 3.52 RMSE 4.56 derived be considered critical fundamental further analyses.