Results of the soil erosion modulus in the source region of the Yellow River from 2001 to 2022 calculated by the GA-CatBoost model

This study collected 537 samples of 137Cs from the source region of the Yellow River between 2015 and 2022, forming the basis for developing a comprehensive soil erosion assessment model. The dataset highlights the estimation results of the optimal model across the entire study area.
Version 2 of this dataset contains the results of sensitivity analyses based on the original (Version 1) outputs. Sensitivity analyses were performed for the following variables: precipitation, temperature, snow, NDVI, DVI, NDVGI, NDWI, SAVI, population density, and SOC. For the three meteorological variables (precipitation, temperature, and snow), we used the average values from 1963 (the onset of ^137Cs fallout) to the sampling year. For all other variables, due to data availability constraints, we used the average values from 2000 to the sampling year in the machine learning models. This dataset provides annual results for each sensitivity analysis from 2001 to 2022. The analyses demonstrate that our research findings are robust with respect to model performance, the spatial distribution of soil erosion, and the temporal trends of total soil erosion.