Representing Uncertainty in Extreme Precipitation Using a WRF Ensemble

As extreme precipitation and urban flooding become more common, reliable precipitation forecasts are necessary. Ensemble forecasts of extreme precipitation provide valuable probabilistic information for decision makers. We develop an ensemble forecast system to produce convection-permitting forecasts of extreme precipitation over the southeastern Michigan region, including the US Army Garrison Detroit Arsenal, using the Weather Research and Forecasting (WRF) model. The ensemble is produced by varying the initial and boundary conditions, planetary boundary layer, surface layer, and microphysics parameterizations. The WRF ensemble is compared against operational model forecasts of precipitation accumulation for twenty historical extreme precipitation events. The WRF ensemble demonstrates higher forecast skill for the majority of events than the operational models, which generally exhibit an underforecast of precipitation. Using a search area to find the maximum precipitation from the ensemble that could have potentially fallen at the the Detroit Arsenal further improves the extreme precipitation forecast for that specific location. Future work includes incorporating the precipitation forecasts from this ensemble into a hydrological model to produce probabilistic inundation forecasts at the building scale.