Delineation of Flood Susceptibility Zone Using Analytical Hierarchy Process and Frequency Ratio Methods: A Case Study of Dakshin Dinajpur District, India

This research explored the effectiveness of two computer ensemble classifiers in assessing flood susceptibility in the Dakshin Dinajpur District, namely, the Analytical Hierarchy Method (AHP) and Frequency Ratio (FR) model. For flood susceptibility mapping, parameters such as precipitation, geomorphology, geohazard, surface, elevation, slope, curvature, flow direction, stream density, distance from the river, road distance, land use and land cover, Normalized Difference Vegetation Index, Normalized Difference Water Index, and Topographic Wetness Index are employed. The study was conducted primarily using 70% of flood points as a training dataset and 30% as model validation. In the study area, the Receiver Operating Characteristic curve results show that both the FR and the AHP models perform well for flood susceptibility mapping with an Area under curve of 0.94 and 0.92 respectively. The Flood Susceptibility Zoning map was classified into five susceptibility alternative classes viz Very Low, Low, Moderate, High, and Very High. In the model with the best results, i.e., the FR model, these zones were scattered over an area of 307.42 sq.km (13.85%) for Very Low susceptible, 547.20 sq.km (24.66%) for Low susceptible, 611.03 sq.km (27.54%) for Moderate, 504.87 sq.km (22.75%) for High and 248.48 sq.km (11.20%) for Very High flood susceptible zone. The present research demonstrated the high capacity of multi-criteria ensemble decision-making classifiers in the study area to distinguish flood susceptibility areas.