Sensitivity of Physical Parameterization Schemes to Stochastic Initial Conditions of WRF Tornado Outbreak Simulations

Previous studies have analyzed the sensitivity of the WRF model to physical parameterizations and initial conditions, though none have specifically focused on severe weather outbreaks, especially tornadic outbreaks. A more complete understanding of how high-resolution models and their physical parameterizations handle different initial conditions could improve forecasting severe weather outbreaks. By identifying the performance of different parameterization schemes, the utility of the WRF model could increase, leading to an improvement of forecasts. For this study, combinations of three cumulus physics (Kain Fritsch, Bettes-Miller-Janjic, and Grell-Freitas Ensemble), three microphysics (Lin et al, Thompson, and WRF Double-Moment 6-class), and two PBL physics (Yonsei University and Mellor-Yamada-Janjic) schemes are used for WRF simulations in tandem with ten stochastic initial conditions generated by the stochastic kinetic-energy backscatter scheme (SKEBS) (Berner, 2011). Five of the top tornadic outbreaks, according to Shafer and Doswell's N15 ranking scheme (2010) are simulated at 18-km resolution. The results of these simulations will be subject to a principal component analysis to detect any outliers, bootstrapping of 95% confidence intervals on the means for each gridpoint in the simulation maps, as well an analysis on how the combinations of parameterization schemes correlate with or differ from one another. another.