Evaluating the skill of daily explicit predictions of mesocyclones in multiple high–resolution WRF model forecasts during the 2005 SPC/NSSL Spring Program

The SPC/NSSL Spring Program is an annual collaborative effort designed to explore applied research topics that are of mutual interest to the research and operational forecasting communities. The program is conducted during the peak severe weather season, from mid-April through early June. This year output from three different high-resolution (grid spacing 2 - 4 km), near-CONUS domain, configurations of the WRF model was examined every day during the program.

Such high-resolution output is of particular interest to the SPC because it has the potential to provide explicit guidance regarding convective structures that can only be inferred from mesoscale model output. For example, these model configurations (especially the 2 km model) appear to have sufficient resolution to produce mesocyclones, or supercellular thunderstorms, which appear to be associated with a disproportionately large number of severe weather reports compared to other modes of convective activity. Yet, even at high resolution mesocyclonic structures are not always discernible in commonly used model output displays, so two specifically designed algorithms were developed for the Spring Program to identify mesocyclones in hourly model output.

These algorithms are both based on the concept that the primary dynamical property of the supercell updraft is a persistent, deep mesocyclone. The first algorithm computes the correlation between vertical vorticity and vertical velocity over local 3-D slab that is about 20 km on a side and 4 km deep, vertically centered on z = 3.5 km. The correlation is multiplied by the local vertical vorticity to provide a measure of magnitude for any circulation that exists. The second algorithm computes the vertical integral of vertical helicity density over the same 4 km deep layer. The vertical helicity density is defined as the local vertical velocity multiplied by the local vertical vorticity. A nine-point smoother is applied to the output from both algorithms.

Application of these algorithms during the Spring Program revealed that the high-resolution models are prodigious producers of rotating updrafts when environmental conditions are favorable. Alert thresholds in both algorithms were set quite low initially, but calibration over the course of the program allowed us to focus on more meaningful storms and quite useful predictions of supercells were generated routinely by the end of the 7-week program. This calibration was performed in a subjective manner by comparing model alerts with those obtained using the NWS WSR-88D Mesocyclone Detection Algorithm.

These results are very promising and provide us with a unique, if preliminary, assessment of the potential value of explicit numerical model predictions of supercells to severe weather forecasters. Furthermore, the diverse collection of model forecasts provides us with a unique opportunity to monitor the evolution and behavior of supercells within complex mesoscale convective structures in many different types of convective environments. The presentation will include a discussion of these topics and a comparison between explicit supercell predictions and guidance derived from algorithms used to predict supercells based on mesoscale environmental parameters.