Answers are grounded in this paper and cite their sources.
Context Soil erosion poses a major environmental challenge in semi-arid Mediterranean watersheds, where decision-support tools are needed to identify priority areas for intervention under data scarcity. Comparative evaluations of uncertainty-handling analytical hierarchy process (AHP) models remain limited in erosion susceptibility mapping, particularly for Monte Carlo simulation-aided AHP (MC-AHP). Aims This study assesses soil erosion susceptibility using AHP, fuzzy AHP (F-AHP), and MC-AHP in the Wadi Isser watershed of Algeria. We compared their predictive performance and investigating how judgement uncertainty affects model reliability. Methods Seven geomorphological and hydrological variables were normalised in GIS and weighted using AHP, F-AHP, and MC-AHP to produce susceptibility maps. The predictive performance of the three approaches was evaluated using observed eroded/non-eroded points through Receiver Operating Characteristic−Area Under Curve (ROC–AUC) analysis, DeLong significance test, and uncertainty analysis. Key results Slope and rainfall were the primary conditioning factors across all models. High to very high susceptibility covered around 26% of the watershed for MC-AHP, and 30% for AHP and F-AHP. MC-AHP achieved the highest AUC (0.803), significantly outperforming AHP (0.781) and F-AHP (0.767) (P < 0.05). Under increasing uncertainty, MC-AHP maintained its superiority (AUC ≥ 0.798), with significant score and rank divergence occurring beyond μ = 0.23 for AHP and μ = 0.17 for F-AHP. Conclusions MC-AHP offers superior discriminative ability, reduces overestimation of erosion-prone zones, and robustly handles expert weight assignment uncertainty. Implications MC-AHP provides a practical, uncertainty-aware framework for targeting soil conservation in data-scarce environments. Regarding F-AHP, further research is needed on the influence of membership function formulation and judgement transformation when analyzing uncertainty.
No takes yet. Share an insight, caveat, or question.
Guelmane et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: