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.本研究基于谷歌地球引擎（Google Earth Engine, GEE）平台上的通用土壤流失方程（Universal Soil Loss Equation, USLE）自动化实现方案，对2001-2020年巴西卡廷加生物群区（Caatinga biome）的年度土壤流失情况开展生物群区尺度评估。研究整合了公开可用的遥感与地理空间数据集，以推导降雨侵蚀力（CHIRPS）、土壤可蚀性（OpenLandMap）、地形因子（MERIT DEM与ALOS DSM）以及土地覆盖与管理指标（MODIS NDVI及土地覆盖产品）。所有数据集均统一至500米空间分辨率，并通过可复现的云端工作流完成处理。研究采用巴西半干旱区域某实验微流域的实测降雨侵蚀力与土壤流失数据，对模型输出结果进行了验证。所得土壤流失图及衍生统计数据，可为数据匮乏的半干旱环境开展环境评估、土地利用规划及土壤保护策略制定提供支撑。