High-resolution flash flood forecasting for large urban areas - Sensitivity to scale of precipitation input and model resolution

Urban flash flooding is a serious problem in large, highly populated areas such as the Dallas-Fort Worth Metroplex (DFW). Being able to monitor and predict flash flooding at a high spatiotemporal resolution is critical to mitigating its threat and cost-effective emergency management. The higher the resolution of the model and the precipitation input is, the better the spatiotemporal specificity of the model output is. Due to errors in the precipitation input, the model parameters and the model itself, however, there are practical limits to the scale of modeling. In this work, we assess this scale dependence using the National Weather Service (NWS) Hydrology Laboratory's Distributed Hydrologic Model (HL-RDHM) at different spatiotemporal resolutions ranging from ~250 m to ~4 km and from 1min to 1 hour for a large part of DFW. The high-resolution precipitation input is from the DFW Demonstration Network of CASA radars. The model simulation results are evaluated using the water level data obtained from the Cities of Fort Worth, Arlington and Grand Prairie in DFW.