Numerical Modeling of Flash Floods over Complex Terrain near Black Hills, South Dakota

Large scale flow interactions with complex terrain of Black Hills in South Dakota lead to episodic events of flash floods that are often damaging to life and property. Whereas, there is potential to forecast such events through an ingredients based methodology, it does not provide information on expected spatial distribution of precipitation, which is essential for determining the populations at risk. Existing operational numerical model simulations do not adequately resolve the complex terrain of Black Hills and therefore nested grid simulations with grid spacing of ~1 km is needed. The skill of such high spatial resolution numerical simulations to capture flash flood events in the Black Hills region is not well understood. We examine the ability of two models, Weather Research and Forecast (WRF) model and Ocean Land Atmosphere Model (OLAM), to forecast the flash flood event that occurred in Black Hills region in South Dakota on 17-18 August, 2007. In WRF simulations, rainfall occurred in areas further east of where actual rainfall occurred and the rainfall accumulation was substantially underestimated. OLAM simulation was able to capture northern extent of the observed rainfall pattern, with maximum rainfall accumulations comparable to observations. A series of additional simulations using OLAM were conducted, where the sensitivity of simulated rainfall pattern to model grid configuration and parameterization options were investigated. Considerable variation occurred in simulated rainfall patterns and spatial variability of model skill in predicting accumulated rainfall. Potential causes for these differences and implications to forecasting of such events will be presented.