Geospatial assessment of soil loss in the Caatinga biome using Google Earth Engine and multi-sensor remote sensing
Description
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.
Files
Steps to reproduce
Data collection and processing workflow The datasets used in this study were obtained exclusively from publicly available geospatial and remote sensing repositories and processed using a reproducible, cloud-based workflow implemented in Google Earth Engine (GEE). Geometric and administrative reference data, including biome boundaries and state limits, were obtained from the Brazilian Institute of Geography and Statistics (IBGE) and used to spatially delimit the Caatinga biome and extract state-level statistics. Rainfall erosivity (R factor) was derived from the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) dataset. Monthly and annual precipitation time series for 2001–2020 were extracted in GEE and used to compute rainfall erosivity based on an empirically adjusted equation for the Caatinga biome, enabling estimation in regions lacking high-resolution pluviograph data. Soil erodibility (K factor) was estimated using soil texture classes from OpenLandMap, classified according to the USDA soil texture system. Literature-based erodibility coefficients associated with each texture class were assigned following established USLE methodologies and applied spatially in GEE. Topographic factors (LS) were derived from the Multi-Error-Removed Improved Terrain Digital Elevation Model (MERIT DEM). Slope and flow accumulation were computed in GEE and combined using raster-based formulations adapted from Moore and Burch (1986) and Zhang et al. (2009). Land cover and management (C factor) was derived from the MODIS MOD13A2 NDVI product. NDVI values were temporally aggregated and transformed into C-factor values using a regression-based and normalized formulation, constrained to the theoretical USLE range (0–1). Negative NDVI values were excluded to avoid non-physical results. The support practice factor (P) was indirectly estimated by integrating slope classes derived from the ALOS Digital Surface Model (ALOS DSM) with land use/land cover information from the MODIS MCD12Q1 product. Empirical P values were assigned according to slope–land use combinations, considering the predominance of non-terraced agricultural practices in the Brazilian semiarid region. All datasets were resampled to a common spatial resolution of 500 m to ensure spatial consistency and computational efficiency. A JavaScript-based workflow was developed in GEE to automate data ingestion, preprocessing, factor derivation, and computation of annual soil loss using the Universal Soil Loss Equation (USLE). Model validation used observed rainfall erosivity and soil loss data from the São João do Cariri Experimental Basin (Brazil), obtained from field-based instruments including a Ville de Paris-type rain gauge and sediment collection systems installed in an experimental micro-watershed. This integrated workflow enables full reproducibility using openly accessible datasets and cloud-based processing tools.
Institutions
- Universidade Federal da ParaibaPB, Joao Pessoa
- Universidade Federal de Campina GrandePB, Campina Grande