Data and code for: Effects of Terrain Derivation and Vegetation Data Filtering on Satellite-Based RUSLE Erosion Potential Estimates in Northeast China, 2001-2023

Data tables and analysis code for the study “Effects of Terrain Derivation and Vegetation Data Filtering on Satellite-Based RUSLE Erosion Potential Estimates in Northeast China, 2001-2023”. The package contains the annual model tables for 2001-2023, the experiment tables behind the terrain, coverage and rainfall comparisons, the run manifests recording Earth Engine asset identifiers and export parameters, and the analysis code. Running code/reproduce/reproduce.py recomputes every statistic reported in the manuscript from the tables and compares each value with the printed one (206 checks in the full run, 201 with the --quick option that skips the 50 000-replicate Monte Carlo); no Earth Engine account or network access is required. No raw input data are included: all input layers are public Google Earth Engine assets (ERA5-Land monthly and daily, MOD13Q1, MCD12Q1, SRTM, SoilGrids) and are loaded directly by the scripts in code/gee_model/. Derived raster layers are not included because of their size, and the provincial boundary geometry is not redistributed because the source administrative dataset does not permit it; the script that rebuilds the boundary from the source is included.