A Step Toward Rainfall Erosivity Mapping Over Türkiye Using Kriging With External Drift

Abstract

Rainfall-induced soil erosion is a significant environmental concern, leading to declining agricultural productivity, deteriorating water quality, and reduced reservoir storage capacity. The rainfall erosivity (R) factor, which quantifies the potential of rainfall to cause soil erosion, plays a pivotal role in the erosion process by directly influencing soil detachment and the transport of eroded particles through surface runoff. Its estimation requires long-term, high-frequency ground-based rainfall measurements, which are scarce in many regions worldwide. Therefore, large-scale R factor mapping typically relies on interpolating point-based station observations to unsampled locations. In this study, the R factor over T & uuml;rkiye was mapped using the kriging with external drift (KED) interpolation technique, incorporating auxiliary information from the satellite-derived Modified Fournier Index (MFI). For this purpose, initially, monthly precipitation estimates and MFI values derived from the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) were validated against station observations. The validation of MFI values yielded a correlation coefficient (CC) of 0.78 and a relative root mean square error (RRMSE) of 0.36. Subsequently, the 222 observation stations were randomly split into training (70%) and testing (30%) subsets, and R factor values from the training stations were interpolated to a grid matching the 0.05 degrees spatial resolution of CHIRPS. The resulting KED predictions outperformed the linear regression model (LM), ordinary kriging (OK), and the globally available rainfall erosivity product, Global Rainfall Erosivity Database (GloREDa), achieving a Kling-Gupta Efficiency (KGE) of 0.68 at independent testing locations. These findings are expected to support decision-making in watershed management and soil conservation planning.