Sensitivity of simulated ensemble Kalman filter analyses and forecasts of low-level supercell rotation to environmental and physics parameterization errors
A set of supercell thunderstorms is simulated using the Advanced Research Weather Research and Forecasting (WRF-ARW) model with full physics and O(100 m) horizontal grid spacing. The simulation is initialized with a National Severe Storms Laboratory (NSSL) Mesoscale Ensemble (NME) analysis of the antecedent environment of a real tornado outbreak. Storms are initiated with thermal bubbles strategically positioned to achieve a range of storm environments. Multiple-Doppler velocity and reflectivity observations are synthesized from the simulation using a sophisticated radar emulator, then assimilated by the Data Assimilation Research Testbed (DART) ensemble adjustment Kalman filter (EAKF) with O(1 km) horizontal grid spacing. The resulting ensemble analyses are used to initialize ~1 h ensemble forecasts from which probabilistic low-level rotation guidance is generated. Evaluation of the simulated forecasts focuses on the sensitivity of low-level circulation track and intensity to errors in the mesoscale analysis and physics paramaterization schemes. Implications of the experiments for using real-time storm-scale ensemble forecasts to extend tornado warning lead times are discussed.