GIS-driven hybrid multicriteria model for flood susceptibility assessment in a coastal metropolis of Ghana

• A novel hybrid method is proposed for flood susceptibility mapping in STM. • The proposed method combines DEMATEL, CRITIC, and SAW. • A network structure highlights complex interconnections among variables. • STM’s coast and mid sections showed moderate to very high flood susceptibilities. • An AUC of 0.947 demonstrates the hybrid model's excellent performance. Mapping regions prone to flooding constitutes a crucial step toward developing localized solutions for flood preparedness and mitigation. This study presents a geographic information system (GIS)-driven approach that combines the decision-making trial and evaluation laboratory (DEMATEL), criteria importance through intercriteria correlation (CRITIC), and simple additive weighting (SAW) methodologies to identify flood-prone areas in the Sekondi-Takoradi Metropolis (STM), Ghana. The study's originality lies in using the hybrid DEMATEL-CRITIC-SAW model for flood susceptibility assessment, a novel integration of decision-making methods and analytical techniques not previously applied together for this purpose. This novel framework provides a comprehensive approach to analyze relationships among 11 flood-inducing variables, determine variable importance, and integrate these findings to produce a more accurate and robust flood susceptibility map. The results reveal a constructed network structure that highlights the complex relationships and dependencies among the variables. Among the assessed criteria, stream power index was identified as the most significant factor due to its high total interaction with other criteria. The flood susceptibility zones within STM are classified into five levels: very low (15%), low (27%), moderate (21%), high (22%), and very high (14%). Notably, the coastal and central sections of STM were marked as areas with moderate to very high flood susceptibility. The results, measured using the area under the curve, indicate that the proposed approach achieved a score of 0.947, demonstrating its superior performance over other existing hybrid models in the literature. The method provides actionable recommendations to authorities in STM for developing flood prevention/mitigation measures.