Global Prediction of Soil Saturated Hydraulic Conductivity Using Random Forest in a Covariate?Based GeoTransfer Function (CoGTF) Framework

Saturated hydraulic conductivity (Ksat) is a key soil parameter for representing infiltration and drainage in land surface models. For large scale applications, Ksat often estimated from pedotransfer functions (PTFs) based on easy-to-measure properties like texture bulk density. The reliance of PTFs data uniform arable lands the omission structure limits applicability texture-based predictions vegetated lands. To include effects terrain, climate, vegetation derivation new global map at 1 km resolution, we harness technological advances machine learning availability remotely sensed surrogate information. model training testing, compilation 6,814 geo-referenced measurements literature was used. accuracy assessment spatial cross-validation shows concordance correlation coefficient (CCC) 0.16 root mean square error (RMSE) 1.18 log10 values cm/day (CCC = 0.79 RMSE 0.72 non-spatial cross-validation). generated maps represent patterns formation processes more distinctly than previous properties. validation indicates that could be modeled without bias using Covariate-based GeoTransfer Functions (CoGTFs) spatially distributed climate attributes, compared to information PTFs. relatively poor performance all models (low CCC high RMSE) highlights need collection additional train regions with sparse data.