Spatial evaluation of flood susceptibility on a national scale across Ghana using a fuzzy analytic hierarchy process and evidential belief function: an ensemble approach

Abstract

Insights conveyed by flood susceptibility maps are critical to lessen the devastating impacts of floods. This research seeks to create an ensemble modeling method using a fuzzy analytic hierarchy process (FAHP) and evidential belief function (EBF) for flood susceptibility classification. The proposed FAHP-EBF method has scarcely been utilized in flood susceptibility mapping, particularly for national-scale assessments. Therefore, through multicollinearity diagnostics, a host of conditioning parameters including elevation, slope, flow accumulation, curvature (plan), distance to rivers, aspect, topographic wetness index, rainfall, and land use/land cover were employed for investigation across Ghana. Furthermore, flood inventory data, essential for building and validating models, were generated using a modified normalized difference water index. Results indicate that a significant share (43.3%) of Ghana’s land is within a low flood susceptibility zone. Additionally, approximately 5%, 10%, 16%, and 26% of the country fall within a very high, high, moderate, and very low susceptibility category, respectively. Areas that are extremely prone to floods include the southwestern part, sections along the coast, and lands along major rivers. In terms of geographical distribution, the Western, Central, Western North, and Savannah regions had greater percentages of land exposed in extremely high areas. The map’s accuracy value of 0.848 based on the area under the curve demonstrated a very good performance. The finalized susceptibility map provides flood management authorities with a blueprint for forecasting and planning interventions in highly susceptible zones and regions.