Assessing reanalysis ERA-5 LAND precipitation data for soil erosion modeling: Performance, challenges, and insights

Abstract Soil erosion due to rainfall is one of the primary soil degradation challenges and issues worldwide and significantly affects water and soil resources management, agricultural protection, and many other sectors. Among several models used for soil erosion estimation, the Revised Universal Soil Loss Equation (RUSLE) model is among the most effective. In the RUSLE model, the R-factor represents the rainfall erosivity, significantly affecting soil erosion results. However, the RUSLE model requires high spatial-resolution precipitation datasets. Still, it has some limitations, such as challenges with continuous data, particularly in developing countries, data scarcity regions, and inadequate spatial coverage, indicating the importance of using satellite and reanalysis datasets. In numerous studies, these datasets were used without validation and performance evaluation. These issues prevent researchers and local and global governments from comprehensively analyzing and estimating soil erosion and its adverse effects. This research aims to evaluate the performance of ERA 5-LAND in soil erosion estimation using the RUSLE model over the Isser basin in Algeria as an application based on the total amount of soil loss, descriptive statistics, and statistical metrics. To the best of our knowledge, this research evaluates the performance of ERA 5-LAND for soil erosion calculation for the first time. The results showed that ERA 5-LAND precipitation data underestimated soil erosion by about 24% for the second scenario by selecting the nearest sixteen grid points and 38% for the third scenario by selecting all grid points. The annual average estimated soil erosion for in-situ rain gauge stations, ERA 5-LAND grid points corresponding to the rain gauge stations, and all ERA 5-LAND grid points, 62.18-, 47.03-, and 38.40-ton ha −1 year −1 , respectively. Also, RMSE results showed slightly high errors, indicating that the ERA 5-LAND needs more correction and enhancement. Even so, ERA 5-LAND can generally be used with acceptable reliability and accuracy, especially in developing countries facing data availability problems.