Montgomery Flora April 27

Supercell Predictability: Exploring Ensemble Sensitivity to Initial Condition Spread

The sensitivity of full-physics ensemble forecasts of supercells to initial condition (IC) uncertainty is investigated. The motivation for the study largely stems from the NOAA Warn-on-Forecast (WoF) program and the need to develop a suitable full-physics/real data numerical weather prediction (NWP) framework for studying storm-scale predictability. Three sets of initial/boundary conditions for our simulations were generated from the real-time NSSL Experimental WoF System for Ensembles (NEWS-e) during the 2016 NOAA Hazardous Weather Testbed Spring Forecasting Experiment.  The 3-km NEWS-e analyses were interpolated to a 1-km grid, then integrated for 3 hours. Our primary goal was not to replicate observed supercell evolution, but rather to isolate the effect of IC uncertainty (i.e., perfect-model assumption).

The forecast sensitivity to IC uncertainty is assessed by successively reducing the initial ensemble perturbations to 50%, 25%, and 10% of their original magnitudes.    Forecast spread was substantially reduced with decreasing initial condition uncertainty for all examined supercell features in all three cases. This was especially true for the mid-level mesocyclone and rainfall. The predictability of the features was case-dependent. However, the predictability of individual supercell features (e.g., mid- and low-level mesocyclones) were correlated for each case – an indication that features organized on larger scales can enhance the predictability of smaller features.

Experiments were also run exploring the importance of uncertainty within vs. outside of the storm. In general, 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.