High-Resolution, Fully Distributed Hydrologic Event-Based Simulations Over a Large Watershed in Texas

The spatial variability of watershed characteristics has to be considered in detail when developing a flood prediction model because it has great influence on the generation of runoff and other hydrologic processes. Physically based, distributed-parameter models can use detailed watershed data that are becoming more readily available to improve hydrologic simulations. However, the computational penalty can be very high for simulations over large catchments. In this study, a physically based, distributed-parameter hydrologic model was used to simulate different flooding events over an 11,285 basin at 150 m 150 m grid resolution. To overcome the computational demand, the model was run on nine sub-catchments separately, without losing hydrologic connectivity among the sub-catchments. The outflow from an upstream sub-basin was used as the inflow into the next sub-basin, allowing this inflow to join the flow generated in the downstream sub-basin in a dynamic mode. Rainfall data from weather radar were used as input. This approach also allowed excluding certain sub-basins from the model simulations without affecting the hydrologic connectivity. A calibration was performed on one sub-basin only, and then, the model was validated over the entire watershed. The simulation results were reasonable and similar to results from single catchment simulations. By using this approach, it was possible to employ observed stream discharge at interior points of a catchment as part of the model inputs. It will also make it possible to run fully distributed, physically based models over very large catchments at very high resolution.