Sub‐Watershed Prioritization Using Advanced Multi‐Criteria Decision‐Making Approaches for Proactive Flood Risk Management

ABSTRACT The study prioritizes sub‐watersheds (SWs) from the Taleqan Watershed in Iran based on Flood Generation Potential (FGP) using Multi‐Criteria Decision Making (MCDM) approaches, including Condorcet, Borda scoring, Fallback bargaining algorithms (Game Theory [GT]), and Best‐Worst Method (BWM), and compares the results with observed discharge data. Based on a review of reliable resources, 30 geo‐environmental criteria were identified to prioritize FGP. Quantitative values for each criterion were then calculated for all SWs. The Principal Component Analysis (PCA) was employed to reduce the FGP conditioning criteria. Next, using the decision matrix based on the selected criteria, three GT algorithms and BWM were applied, and the SWs were prioritized accordingly. FGP maps were then created using all four approaches, with results categorized into five classes, namely, very low, low, moderate, high, and very high. Finally, the prioritization patterns from the MCDM approaches were compared against observed discharge data from all 18 SWs to select the optimal method. The results from all four MCDM approaches indicated that SW5 had the highest potential for flood generation. The BWM approach classified about 56% of the study watershed as having low or very low FGP. In contrast, GT identified only 45% of the area in this category, with approximately 39% of SWs classified with high or very high FGP. The comparison of means using the One‐Way ANOVA test revealed insignificant differences between the MCDM approaches and the observed discharge data. However, the similarities between the results of the Condorcet, Borda scoring, Fallback bargaining, and BWM approaches, compared with the observed discharge data, were 27.78%, 27.78%, 33.33%, and 44.44%, respectively. Accordingly, the BWM approach with the highest similarity was chosen as the optimal model. The summary indicated that the BWM approach outperformed the GT algorithms in mapping flood generation compared to the observed flood discharge data.