Flood hazard susceptibility areas mapping using Analytical Hierarchical Process (AHP), Frequency Ratio (FR) and AHP-FR ensemble based on Geographic Information Systems (GIS): a case study for Kastamonu, Turkiye

Global climate change brings with it various natural disasters. In particular, natural disasters such as floods destroy nature and human resources. The flood disaster in Kastamonu province, primarily striking Bozkurt district and many other districts in Turkiye on August 11, 2021, causing both life and material losses, has been one of the most devastating disasters in the Black Sea region. In this study, various geospatial and statistical methods were used to produce flood hazard susceptibility maps of Kastamonu province. In order to evaluate the flood risk in Kastamonu, eleven different variables, i.e. rainfall, slope, elevation, distance from stream, land-use-land cover, lithology, curvature plan, curvature profile, Topographic Wetness Index, Stream Power Index and Normalised Differences Vegetation Index were used. Flooded areas were determined by the Modified Normalised Water Index (MNDWI) on the Google Earth Engine platform using Remote Sensing techniques. Flood points determined on the calculated MNDWI image are divided into 70% training and 30% testing dataset. Geographical Information Systems-based Analytical Hierarchy Process (AHP), Frequency Ratio (FR), and ensemble AHP-FR were used in the creation of flood hazard susceptibility maps. The maps were divided into five classes: very low, low, moderate, high, and very high. On the map classified using AHP-FR, areas in high and very high sensitivity classes were calculated as 128.72 km(2) and 6.89 km(2), respectively. These calculated areas constitute 0.99% and 0.05% of the entire region. On the other hand, part of Kastamonu province with an area of 484.07 km(2) was determined as a moderate-risk area. This area covers 3.71% of the entire province. The remaining part of the province, with an area of 8729.39 km(2) and 3697.30 km(2), is classified as very low and low, respectively. These areas cover 66.91% and 28.34% of the entire province, respectively. The study's accuracy was tested using the receiver operating characteristic curves method. Area under curve values for AHP, FR, and AHP-FR were calculated as 0.965, 0.989, and 0.992, respectively. According to these values, using the AHP-FR ensemble gave more successful results than the other two methods.