A Review of River Flood Models: Methods and Applications for Forecasting and Simulation

Abstract

Flood modeling plays a pivotal role in mitigating flood impacts and ensuring public safety, water management, and environmental preservation. This paper presents a comprehensive review of sixty-two research papers available on “Google Scholar”, analyzing and categorizing various flood models into five types: Time series models, Physical-based models, ANN models, Fuzzy logic models, and Other models. Time series and fuzzy logic models can be applied to forecast floods by predicting key flood-causing factors such as water levels, discharges, and rainfall. Detailed flood extent information can then be simulated using physically based models. Notably, the 1D-2D coupled model stands out as the most effective physical-based model, providing detailed flood information for both rivers and floodplains. The paper identifies user-friendly software packages like HEC-HMS, HEC-RAS, MIKE FLOOD, and LISFLOOD-FP that make physical-based flood models popular despite requiring more input data. Existing flood models exhibit an accuracy level of over 80%, and further improvements can be achieved by incorporating additional flood-dependent factors such as rainfall, slope, topography, and land use. Moreover, combining ANN hydrological model with 2D hydrodynamic model shows promise in obtaining detailed results for future flood events. Commonly used statistics such as RMSE, NSE, R2, and flood area-related metrics are employed for evaluating flood model accuracy.