Geomorphometric soil erosion modelling in Sub-Saharan Africa: A systematic review of model applicability, validation gaps, and policy integration

Soil erosion poses a fundamental threat to sustainable development in Sub-Saharan Africa (SSA), where ∼490 million hectares of agricultural land are already degraded. Here we present a PRISMA-guided systematic review of 399 peer-reviewed studies (2000–2024) that evaluates the applicability, performance, and policy relevance of geomorphometric soil erosion models in SSA’s data-scarce, topographically complex landscapes. Empirical models (chiefly RUSLE and its derivatives) dominate the literature (∼68 %), followed by physically based (WEPP, KINEROS2, EUROSEM) and emerging hybrid/machine learning frameworks. Although high resolution DEMs and multi-temporal satellite imagery (Landsat and Sentinel) have markedly improved spatial prediction of erosion hotspots, critical limitations persist: (i) widespread absence of field based validation, with most studies relying on indirect proxies or qualitative checks; (ii) underrepresentation of Central and Southern African semi-arid catchments, producing a strong East African bias; (iii) negligible integration of model outputs into national or sub-national land use policy, spatial development frameworks, or climate adaptation strategies. Hybrid and data driven models consistently outperform traditional empirical and process-based approaches (mean AUC 0.83–0.86 vs. R² ∼0.55 for RUSLE type models), yet their adoption remains marginal owing to calibration and computational barriers. We identify an urgent need for regionally calibrated ensemble models, continent wide validation networks using UAV/LiDAR ground truthing, and institutional mechanisms to translate geomorphometric risk maps into actionable policy. Closing these gaps is essential to deliver reliable, policy-ready erosion assessments capable of guiding sustainable land management and safeguarding agricultural productivity in one of the world’s most erosion vulnerable and food insecure regions.