Applying Ensemble-Based Sensitivity Analysis to WRF Convection Forecasts in the Northern Great Plains

Jeremy Berman* and Ryan Torn, Glen Romine, Morris Weisman
University at Albany/SUNY, University at Albany/SUNY, NCAR, NCAR

Numerical Weather Prediction model forecasts of severe convection are often characterized by varied degrees of predictability depending on the particular case. One hypothesis for larger errors in some cases is that the model’s initial conditions are characterized by errors in upstream sub-synoptic features before convective initiation takes place. During the Mesoscale Predictability Experiment (MPEX), one of the cases characterized by the large errors occurred on June 11-12, 2013, when some members of the NCAR Weather Research and Forecasting Model (WRF) ensemble forecasting system predicted that convection would form in South Dakota and Nebraska out ahead of a large-scale, upper-level trough lifting out of the Rockies, while other members correctly predicted less organized convection. The purpose of this study is to apply the ensemble-based sensitivity method to the aforementioned WRF ensemble forecasts to determine how uncertainty in upstream features contributed to errors in the subsequent convective forecasts.

Preliminary results suggest that convective initiation in western Nebraska and South Dakota is sensitive to the location of a lee trough that forms downstream of the Sheridan Range ahead of the upper-level trough moving out of the Rockies during the morning of June 11, 2013. Variability in the lee trough position is associated with changes in the model-forecasted thermodynamic profile of the environment prior to convection. Our findings suggest that the lee trough, and its phasing with the diurnal cycle, has an important role in pre-conditioning the environment prior to the arrival of the main dynamical forcing.

*Preference: Oral