Spatial Assessment of Flood Susceptibility in Assam, India: A Comparative Study of Frequency Ratio and Shannon's Entropy Models

Flooding is a common and catastrophic natural event that has serious consequences for both human and nature. The assessment of flood susceptibility within watersheds and the implementation of measures to mitigate flood-related impacts and damages are integral aspects of environmental, water resources and disaster management. Assam is significantly impacted by recurring flood events consisting of multiple waves. Consequently, it is imperative to precisely identify the areas susceptible to flooding. Therefore, this research aims to assess the capability of the Frequency Ratio (FR) and Shannon's Entropy (SE) model for mapping flood susceptibility; and to identify extremely flood prone locations in Assam, India. The study was conducted in four main stages. Firstly, a flood inventory map was developed by using historic records (1998 to 2017) from the Assam Water Resources Department (AWRD), Assam State Disaster Management Authority (ASDMA), and extensive field surveys. A total of 1630 flood occurrence points were extracted; and of these 75% were randomly selected for model training and 25% for validation. Secondly, nine flood conditioning factors were considered, including elevation, aspect, distance from river, drainage density, Land-use and Land-cover (LULC), rainfall, geomorphology, Topographic Wetness Index (TWI), and Normalised Difference Vegetative Index (NDVI). The thematic layers were prepared in a Geographic Information System (GIS). Thirdly, flood susceptibility maps (FSMs) were generated by applying FR and SE models, using the flood conditioning factors and the occurrence locations. The FSMs were classified into five classes. Finally, the predictive capability of FR and SE models and the validation was done by using Receiver Operating Characteristic (ROC) curves and computing the Area Under Curves (AUC). The AUC for the FR model was 0.748, and the SE model, it was 0.761. The validation results suggests that the SE model is more suitable for predicting flood susceptibility in the study area. The study makes a unique contribution by identifying the flood susceptible zones in Assam, a state afflicted by recurrent floods. The derived FSMs of the region can be used for effective decision making in flood management, as well as for strategic planning and mitigation measures toward flood events in the future.