Geospatial assessment of soil loss in the Caatinga biome using Google Earth Engine and multi-sensor remote sensing

This study provides a biome-scale assessment of annual soil loss in the Caatinga biome (Brazil) for the period 2001–2020 using an automated implementation of the Universal Soil Loss Equation (USLE) within the Google Earth Engine (GEE) platform. Publicly available remote sensing and geospatial datasets were integrated to derive rainfall erosivity (CHIRPS), soil erodibility (OpenLandMap), topographic factors (MERIT DEM and ALOS DSM), and land cover and management indicators (MODIS NDVI and land cover products). All datasets were harmonized to a 500 m spatial resolution and processed through a reproducible, cloud-based workflow. Model outputs were validated using observed rainfall erosivity and soil loss data from an experimental micro-watershed in the Brazilian semiarid region. The resulting soil loss maps and derived statistics support environmental assessment, land-use planning, and soil conservation strategies in data-scarce semi-arid environments.