Over the past decades, the scientific community has made significant efforts to simulate flooding conditions using a variety of complex physically based models. Despite all advances, these models still fall short in accuracy and reliability and are often considered computationally intensive to be fully operational. This could be attributed to insufficient comprehension of the causative mechanisms of flood processes, assumptions in model development and inadequate consideration of uncertainties. We suggest adopting an approach that accounts for the influence of human activities, soil saturation, snow processes, topography, river morphology, and land‐use type to enhance our understanding of flood generating mechanisms. We also recommend a transition to the development of innovative earth system modeling frameworks where the interaction among all components of the earth system are simultaneously modeled. Additionally, more nonselective and rigorous studies should be conducted to provide a detailed comparison of physical models and simplified methods for flood inundation mapping. Linking process‐based models with data‐driven/statistical methods offers a variety of opportunities that are yet to be explored and conveyed to researchers and emergency managers. The main contribution of this paper is to notify scientists and practitioners of the latest developments in flood characterization and modeling, identify challenges in understanding flood processes, associated uncertainties and risks in coupled hydrologic and hydrodynamic modeling for forecasting and inundation mapping, and the potential use of state‐of‐the‐art data assimilation and machine learning to tackle the complexities involved in transitioning such developments to operation.
Every year, a large number of people are affected by flooding and suffer its costly consequences across the world. To properly manage this notorious natural disaster, the physical processes that represent riverine and coastal floods should be well understood and modeled. Over the recent decades, the scientific community has been continuously involved in characterizing the main components of floods and improving flood modeling skills using both types of physical and statistical models. Despite all these efforts, our modeling skill has major limitations which hinder an optimum performance for accurate and efficient flood forecasting. In this article, we provide a thorough review of these past efforts, highlight the main challenges, and provide potential pathways for improved flood characterization and modeling in the future. We specifically discuss the causative mechanisms of floods, physical/statistical methods used to characterize different components of flooding, coupling approaches, methods used to account for uncertainty in different layers of flood modeling, and their benefits for operational flood forecasting systems.
Causative mechanisms of floods and underlying physical processes in both riverine and coastal floods are thoroughly discussed and reviewed
The weak and selective validation of flood inundation models and the lack of sufficient validation data is a major challenge
Hybrid methods linking statistical and numerical tools are recommended for efficient and more accurate coastal flood hazard analysis