EFFECTIVE METHOD SELECTION FOR FLOOD DISASTER MANAGEMENT: A DECISION SUPPORT APPROACH BASED ON RIVER TYPE

Abstract

In recent decades, the frequency and intensity of flood events have escalated, driven primarily by climate change and evolving land-use patterns. These increasingly severe hydrological hazards underscore the critical need for accurate delineation of flood-prone areas and the implementation of effective risk management strategies. This study presents a comparative analysis of flood risk mapping in two hydrologically distinct river basins: the seasonally flowing Çapakçur River and the Continuos Harşit River. A multi-method approach was employed, utilizing Fuzzy Analytic Hierarchy Process (Fuzzy AHP), Random Forest, HEC-RAS 1D, and HEC-RAS 2D modeling techniques to generate flood hazard maps for each basin. Initially, historical flood reports provided by relevant authorities were converted into spatial disaster records to define the study areas. Subsequently, model-specific data inputs and workflows were implemented, incorporating spatial parameter maps, hydrological datasets, and physical modeling inputs to simulate flood-prone zones. The performance of each method was evaluated through comparison with past flood occurrences using four validation metrics: accuracy, precision, recall, and F-score. Findings indicate that the Random Forest model consistently achieved the highest accuracy across both river types. While physically based HEC-RAS models demonstrated stable and reliable performance—particularly in continuous flow conditions—the Fuzzy AHP method showed limited predictive capability, primarily due to its reliance on subjective expert judgment. Overall, the study emphasizes the importance of aligning flood modeling approaches with the hydrological characteristics of the watershed and the nature of available data. The results provide valuable insights into method selection for flood risk assessment and contribute to more informed decision-making in disaster risk reduction and land-use planning.