Supercell Predictability: Impacts of Intra-Storm vs. Environmental Initial Condition Spread on Ensemble Forecast Uncertainty

Although several studies have evaluated the increase in storm-scale predictability gained by reducing domain-wide initial condition (IC) uncertainty, relatively little is known about the relative impacts of IC uncertainty within storms versus their environment. Therefore, building on a framework established in Flora et al. (2018), the current study evaluates the evolution of ensemble uncertainty in forecasts of supercells initialized with IC spread either everywhere or restricted to within or outside the storm of interest. In experiments with complete intra-storm (environmental) certainty, each ensemble member has their IC replaced with that of an ensemble member treated as “truth” within (outside) regions of composite reflectivity > 10 dBZ. Three cases are used from the real-time NSSL Experimental Warn-on-Forecast System for Ensembles (NEWS-e) from the 2016 NOAA Hazardous Weather Testbed Spring Forecasting Experiment. An object-oriented analysis is used that focuses on significant supercells features including the mid- and low-level mesocyclones and rainfall, similar to Flora et al. (2018). Preliminary results suggest that reducing intra-storm ensemble perturbations to zero greatly reduced forecast spread early on, while later in the simulations forecasts benefited more from eliminating uncertainty in the storm environment. Of all the supercell features, forecasts of the mid-level mesocyclone benefited the most when the perturbations inside the storm were set to zero. Substantial improvements also occurred in low-level mesocyclone forecasts, suggesting that significantly improving tornado forecast lead times does not necessarily require reducing uncertainty in the storm environment.