Evaluation of uncertainty in precipitation forecast products for real-time flood guidance in large metropolitan areas

Abstract:

Recurrent urban floods cause significant socioeconomic impacts in the United States, emphasizing the importance of building flood resilient communities to reduce damage and losses. An accurate water forecast system is a valuable tool to build resilience, support decision making, and launch preparedness and response actions. Precipitation is the most relevant model forcing to produce accurate flood forecasts. In this context, the objective of this research is to evaluate the impact of the High-Resolution-Rapid-Refresh (HRRR) hourly precipitation forecast model uncertainty on streamflow predictions, aiming to improve an operational real-time flood guidance system. The method includes the use of the WRF-Hydro modeling framework to simulate hydrological processes using meteorological forcings and produce short-range (18-36 Hours) forecasts. The spatial-temporal variability of the HRRR hourly precipitation was analyzed and its agreement with the National Center for Environmental Prediction (NCEP) Stage IV quantitative precipitation estimates product was evaluated. The impact of the precipitation uncertainty was estimated by comparing the performance of forecasted and hindcasted model results at different lead times. In order to assess the proposed framework, the National Capital Region was chosen due to its exposure to multiple water-related hazards such as riverine, urban, tidal and storm surge flooding, the existence of different urbanized watershed scales, as well as the fact it is one of the most rapidly-growing urban regions of the United States. This research is conducted as part of the Integrated Flood Forecast System (iFLOOD: http://iflood.vse.gmu.edu), which is a scientific experiment and educational tool to incorporate multi-scale and multi-temporal physical processes for total water prediction. The application of this study inside the iFlood will allow further investigation on the propagation of this uncertainty into estuarine environments and stormwater systems modeling for instance. It is expected that the results can lead to insights on how to improve precipitation forcing by applying methods to post-process the precipitation forecast, and consequently, reduce flood forecast uncertainty in short-range real-time guidance systems.