Prediction of Flood Using Adaptive Neuro-Fuzzy Inference Systems: A Case Study

Abstract

Among natural hazards, flood is the one which occurs in all parts of the world and is essential to control it by appropriate administration. Floods in Basantpur, India cause destruction to life and property each year, and hence for recognition of vulnerable parts in watersheds flood model development is significant for decision makers. BPNN and ANFIS techniques can be valuable in the study of hydrology even though these techniques are capable of fulfilling all necessities for comprehensive, hurried, and precise analysis. The purpose of present research is the comparison of prediction performances of two altered approaches for flood susceptibility mapping at proposed watershed. The performance of standardized ANN-based model has been accessed by taking peak of observed and simulated floods and calculation of root mean squared error (RMSE) for intermediate gauging stations on the projected basin. ANFIS gives better performance with coefficient of determination 0.9676 and 0.9347 for both testing and training phase while in case of BPNN it gives 0.9536 and 0.9227.