A new approach to flood susceptibility assessment in data-scarce and ungauged regions based on GIS-based hybrid multi criteria decision-making method

Identifying the flood susceptible areas is a vital and substantial element of disaster management for every country to control and mitigate injuries of the natural hazards. The current research presents a framework for the preparation of flood prone areas' maps by the integration of Geospatial Information System (GIS), fuzzy logic, and Multi-Criteria Decision Making (MCDM). To achieve this goal, a spectrum of geophysical, geomorphological, meteorological, hydrological, and geographical criteria have been addressed. Considering the linkage and the interdependencies of the criteria, DEcision-MAking Trial and Evaluation Laboratory (DEMATEL) approach are used to form the network of relations among the criteria. Moreover, considering inevitable uncertainty, ambiguity/vagueness of the experts' opinions, fuzzy theory is integrated with DEMATEL to handle the issue. Finally, Analytic Network Process (ANP) are implemented to calculate the final weight of every single criterion. The Kurdistan province, located in the North-West of Iran, is considered as the case study in which numerous flood events has had recently happened. The efficiency of the model is assessed through the area under the curve (AUC) and statistical measures such as the Kappa index. In order to evaluate the produced classified flood susceptibility map, the map of historical flood events in the province is also used. About 85% of validation area is classified as "Very High flood Susceptible" which implies the efficiency of proposed framework for flood susceptibility mapping. Furthermore, for evaluating the functionality of the framework in comparison with traditional approaches, the well-known AHP methodology is implemented too. The validation results demonstrate that the Fuzzy-DEMATEL ANP model (AUC-ROC = 0.938, Kappa = 0.88) has a higher performance accuracy compare to the AHP model (AUC-ROC = 0.918, Kappa = 0.79). Moreover, comparing the proposed and validation model shows that the proposed framework can effectively improve decision makers to provide flood susceptibility maps, and recognize flood susceptible areas in data-scarce and ungauged regions.