Evaluating the Performance of MaxEnt and Frequency Ratio Models for Flood Susceptibility Mapping in Kerala State, India

Floods are the greatest catastrophic natural disasters on a global scale. Identifying flood-prone areas is imperative for preventing floods, reducing risks, and making informed decisions. Flood susceptibility mapping plays a crucial role in flood risk assessment and management. While machine learning shows promise in flood prediction, a significant research gap remains in effectively integrating advanced algorithms with Kerala's diverse regional flood data to develop more accurate prediction models. The main objective of the proposed research is to develop, implement, and compare the performance of Maximum Entropy (MaxEnt) and Frequency Ratio (FR) models for flood susceptibility mapping of Kerala state using a spatial flood database integrated within the ArcGIS interface. The research employs two paired approaches, the MaxEnt and FR, to assess flood susceptibility levels within the study region. The results show that both models achieved strong predictive performance; AUC was 0.928 for MaxEnt and 0.946 for the FR model, indicating its enhanced capability in flood susceptibility evaluation. The MaxEnt model is a promising tool for assessing flood-prone areas and allows for proper planning and management of flood hazards.