loading page

Global prediction of soil saturated hydraulic conductivity using random forest in a Covariate-based Geo Transfer Functions (CoGTF) framework
  • +3
  • Surya Gupta,
  • Tomislav Hengl,
  • Peter Lehmann,
  • Sara Bonetti,
  • Andreas Papritz,
  • Dani Or
Surya Gupta
Soil and Terrestrial Environmental Physics, Department of Environmental Systems Science, ETH Zurich, Switzerland

Corresponding Author:[email protected]

Author Profile
Tomislav Hengl
Envirometrix Ltd., Wageningen, the Netherlands
Author Profile
Peter Lehmann
Soil and Terrestrial Environmental Physics,Department of Environmental Systems Science, ETH Zurich, Switzerland
Author Profile
Sara Bonetti
Soil and Terrestrial Environmental Physics, Department of Environmental Systems Science, ETH Zurich, Switzerland
Author Profile
Andreas Papritz
Soil and Terrestrial Environmental Physics, Department of Environmental Systems Science, ETH Zurich, Switzerland
Author Profile
Dani Or
Soil and Terrestrial Environmental Physics, Department of Environmental Systems Science, ETH Zurich, Switzerland
Author Profile

Abstract

The saturated hydraulic conductivity (Ksat) is a key soil hydraulic parameter for representing infiltration and drainage in Earth system and land surface models. For large scale applications, Ksat is often estimated from pedotransfer functions (PTFs) based on easy-to-measure soil properties like soil texture and bulk density. The reliance of PTFs on data from uniform arable lands and omission of soil structure limits the applicability of texture-based predictions of Ksat in vegetated lands. A method to harness technological advances in machine learning and availability of remotely sensed surrogate information to derive a new global Ksat map at 1 km resolution using terrain, climate, vegetation, and soil covariates is proposed. For model training and testing, global compilation of 6,814 georeferenced Ksat measurements from the literature across the globe were used. The accuracy assessment results based on model cross-validations with re-fitting show a concordance correlation coefficient of 0.79 and root mean square error of 0.72 (in log10Ksat given in cm/day). The generated maps of Ksat represent spatial patterns of the vegetation-induced soil structure formation and clay mineralogy, more distinctly than previous global maps of Ksat such as computed with Rosetta 3 pedotransfer function. The validation of the model indicates that Ksat could be more accurately modeled using covariate-based Geo Transfer Functions (CoGTFs) that harness spatially distributed surface and climate attributes, compared to pedotransfer functions that rely only on soil information.