Enhancing the Capability of a Simple, Computationally Efficient, Conceptual Flood Inundation Model in Hydrologically Complex Terrain

Abstract

The simple conceptual flood inundation model TVD (Teng-Vaze-Dutta) is more computationally efficient and cost-effective than traditional hydrodynamic models. It is especially useful for applications that do not require velocity output and have low demands on flood dynamic representation. In this study, we have addressed the main inherent limitations of the original TVD model including: the assumption that all the floodplain depressions connected to the river are instantly filled up to the in-stream water level at each time step; the lack of information sharing at the boundary of two modelling reaches; and insufficient soil moisture processes. All of these can affect the model’s applicability and accuracy, especially in very flat and hydrologically complex floodplains. A number of improvements to the model structure have been implemented to address mass conservation, reach connectivity and water balance issues. The revised model was set up to simulate a number of flood events in Australia’s lower Balonne River and Darling River to test for its enhanced capability. The modelled inundation extents before and after the improvements were assessed against remote sensing water maps. The model developments have improved the accuracy of modelled flood extent. Nevertheless, there are still remaining issues that require the model to be used with caution when simulating flood inundation in difficult-to-model topographies, largely, the demand for reliable input of overbank flow volume and the extrapolating issue with weighting schemes.