Global hydrodynamic modelling of flood inundation in continental rivers: How can we achieve it?

Global-scale modelling of river hydrodynamics is essential for understanding global hydrological cycle, and is also required in interdisciplinary research fields (e.g. earth/climate system modelling and flood risk forecasting/assessment). Global river models have been developed continuously for more than two decades, but modelling river flow at a global scale is still a challenging topic because surface water movement in continental rivers is a multi-spatial-scale phenomena. We have to consider the basin-wide water balance (>1000km scale), while hydrodynamics in river channels and floodplains is regulated by much smaller-scale topography (<100m scale). For example, heavy precipitation in upstream regions may later cause flooding in farthest downstream reaches. In order to realistically simulate the timing and amplitude of flood wave propagation for a long distance, consideration of detailed local topography is unavoidable.
Regional flood inundation models (e.g. LISFLOOD-FP developed by University of Bristol) realistically simulate 2D hydrodynamics by explicitly representing detailed floodplain topography (<100m scale), and it is now virtually possible to apply these regional models at a global scale. However, because of their large computational cost, regional flood inundation models are not suitable for long-term ensemble simulations or real-time applications at a global scale. A global river hydrodynamic model with both high computational efficiency and high simulation accuracy has to be developed to satisfy the requirements from interdisciplinary research communities.
I have developed the global hydrodynamic model CaMa-Flood (Catchment-based Macro-scale Floodplain model) in order to overcome this scale-discrepancy of continental river flow. The CaMa-Flood divides river basins into multiple “unit-catchments”, and assumes the water level is uniform within each unit-catchment. One unit-catchment is assigned to each grid-box defined at the typical spatial resolution of global climate models (10~100 km scale). Adopting a uniform water level in a >10km river segment seems to be a big assumption, but it is actually a good approximation for hydrodynamic modelling of continental rivers. The number of grid points required for global hydrodynamic simulations is largely reduced by this “unit-catchment assumption”. Alternative to calculating 2-dimensional floodplain flows as in regional flood models, the CaMa-Flood treats floodplain inundation in a unit-catchment as a sub-grid physics. The water level and inundated area in each unit-catchment are diagnosed from water volume using topography parameters derived from high-resolution digital elevation models. Thus, the CaMa-Flood is at least 1000 times computationally more efficient compared to regional flood inundation models while the reality of simulated flood dynamics is kept. In the seminar, I will explain in detail how the CaMa-Flood model has been constructed from high-resolution topography datasets, and how the model can be used for various interdisciplinary applications.

Details available on the Oxford Water Network: