Investigation of Atmospheric Modelling Framework for Better Reconstruction on Historical Extreme Precipitation Event in PMP Estimation

Abstract:

During May 1-2, 2010, a record-breaking storm hit Nashville, and caused huge humanity and societal loss. It raises the importance of forecasting/reconstructing these types of extreme weather systems once again, in the meanwhile providing an excellent case for such atmospheric modelling studies. However, earlier studies suggest that successful reconstruction of this event depends on and is sensitive to a number of model options, making it difficult to establish a better model framework with more confidence. In this study we employed the Weather Research and Forecast (WRF) model to investigate how this extreme precipitation event is sensitive to the model configuration, and identified options that would produce better results. We tested several combinations of modelling grid sizes together with initial/boundary conditions (IC/BC). At different grid sizes, we conducted a set of tests on various combinations of microphysics (Morrison, new Thompson and WSM5) and cumulus process (Kain-Fristch, Grell-Devenyi and Grell-Freitas) parameterization schemes. The model results were intensively evaluated under bias analysis as well as other metrics (probability of detection, bias, false alerts, HSS, ETS). The evaluation suggests that in general, simulation results benefit from finer model grids (5km). At 5km level, NCEP2 or NAM IC/BCs are more representative for the 2010 Nashville storm. There are no universally good parameterization schemes, but the WSM5 microphysics scheme, Kain-Fristch and Grell-Freitas cumulus schemes are recommended over other tested schemes. These better schemes would help to make better estimation of PMP in the region.