Catchment-Scale Hydrodynamic Modelling of Seasonal Fluvial Processes

Modelling has provided an essential tool for predicting and understanding the complex surface and subsurface hydrological processes induced by rainfall, and also for assessing the impact of flooding of on environment, people and their properties. Traditionally modelling of seasonal or longer-term fluvial processes is achieved by hydrological models of different levels of complexity. However, due to their inherent limitations, hydrological models are not able to accurately simulate those highly transient overland flow and flooding processes induced by intense rainfall and therefore cannot reliably assess their impact on the longer-term fluvial processes. On the other hand, sophisticated hydraulic/hydrodynamic models have been widely applied to predict both slowly-varying and rapidly-varying flash flood events lasting for hours or days, but they are rarely used for the simulation of long-term fluvial processes due to their incapability in representing certain hydrological processes (e.g. infiltration and evaporation) and prohibitively long simulation time. Potentially, hydrodynamic modelling may provide more detailed information for better understanding of the dynamics of fluvial processes and the impact of extreme events on the natural and engineering environments if they can directly support long-term simulations.

In this work, a finite volume shock-capturing hydrodynamic model that solves the full 2D nonlinear shallow water equations is developed and applied to reproduce the seasonal fluvial processes at high spatial resolution in the 12398km² Chaohu River basin for the entire calendar year of 2016. New model components are developed and fully coupled to the hydrodynamic model to take into account the infiltration, subsurface flow, interception and evaporation effects. The use of high-resolution topographic data (< 10m) will enable the model to automatically capture surface storage and reliably represent river networks and flows. The overall modelling system is implemented on high-performance graphics processing units (GPUs) to substantially improve the computational efficiency for long-term simulations.

The simulation results are verified by systematically comparing with historical flood records and gauge measurements of water level and discharge. The model performance will be further demonstrated by comparing with main stream hydrological models, including e.g. HSPF. Further simulations will be also carried out to investigate model sensitivity to key model parameters, e.g. grid resolution, and friction, infiltration and evaporation parameters.