Comparing MCDM methods with RUSLE and EPM models for erosion risk assessment and sub-watershed prioritization in semi-arid mountainous terrain

Abstract

Soil erosion threatens the sustainability of semi-arid watersheds. Sub-watersheds are routinely ranked for intervention, yet the reliability of those ranks is rarely quantified. This study assesses erosion risk across the 3,288 km² Al Khazer River Watershed in northern Iraq and measures the stability of the resulting priorities. Fifteen morphometric parameters were derived for 32 sub-watersheds from a 12.5 m ALOS PALSAR DEM, Sentinel-2 imagery and climate-station records. Criterion weights were obtained by the Analytical Hierarchy Process (consistency ratio = 0.0537). These weights were applied within four ranking models: VIKOR, TOPSIS, SAW and WASPAS. Two erosion models were then applied independently: the Erosion Potential Method (EPM) and the Revised Universal Soil Loss Equation (RUSLE). RUSLE agreed more closely with field observations (Observed-to-Estimated Agreement Ratio (AR) 78.1% and 86.96%) than EPM (49.16% and 53.0%). It was therefore adopted as the internal benchmark for validating the ranking models. All four models identified sub-watersheds 6, 4, 2 and 1 as the highest-priority zones. These watersheds combine steep slopes, high relief and elevated orographic precipitation in the northern mountains. Ordinary least squares regression showed the distance-based methods to correspond most closely with RUSLE (VIKOR R² = 0.763, TOPSIS R² = 0.754), with the two statistically indistinguishable (ΔAIC = 1.21). Input uncertainty was then propagated through the prioritization using 100 Monte Carlo iterations at perturbation levels of ± 10%, ± 20% and ± 30%. Classification stability fell from 96.9% to 84.4% for RUSLE and from 87.5% to 75.0% for TOPSIS. The four highest-priority sub-watersheds retained their classification at every perturbation level. The framework identifies erosion hotspots using only openly available terrain, satellite and climate data. It therefore offers a practical basis for targeting soil conservation investment, for routine environmental monitoring of erosion risk, and for guiding land management and land-use planning in ungauged semi-arid watersheds where field measurement is scarce.