Complementary Use of Short-Range Ensemble and 4.5 km WRF-NMM Model Guidance for Severe Weather Forecasting at the Storm Prediction Center

Large increases in computer power and communications capabilities in recent years have facilitated the development and operational testing of two key Numerical Weather Prediction initiatives: 1) short-range ensemble forecast (SREF) systems, and 2) high resolution deterministic models. The application of ensemble concepts to short-range prediction provides forecasters with information about the possible range of solutions and measures of forecast uncertainty, which can then be used to better convey appropriate levels of forecaster confidence to the user community. Typically, ensemble members are run at coarser resolution (~30-40 km grid length) compared to current operational deterministic model runs. SREF output can be created to provide basic synoptic and mesoscale guidance for a variety of products ranging from synoptic pattern evolution and the likelihood of precipitation to more specialized fields such as thermodynamic and kinematic parameters related to convective storm potential. Additional research efforts have been focused on high resolution models that use explicit cloud and precipitation microphysics to generate precipitation (no convective parameterization is used in these models). The convection-allowing models are typically run with grid lengths of ~5 km or less, and have the capability to generate explicit convective systems such as MCSs, as well as near-storm scale convective elements including model generated storms containing rotating updrafts. Thus, high resolution models have potential to provide unique guidance to severe weather forecasters regarding critical topics of convective initiation, evolution, mode, and intensity.

Since 2003, SPC has played a leading role in testing various configurations of SREF systems and high resolution Weather Research and Forecasting (WRF) models for their operational utility. This testing has occurred both within SPC operations and as part of organized activities in the NOAA Hazardous Weather Testbed in Norman. The HWT is designed to bring research scientists, model developers, and forecaster together to work on issues of mutual interest, facilitating the rapid transfer of research to operations. Currently, the NCEP Environmental Modeling Center (EMC) is running a 21 member multi-model, multi-analysis SREF system with enhanced physics diversity twice daily at 09 and 21 UTC, with output through 87 hours. SPC processes the grids from all SREF members and produces a large variety of products for severe weather forecasting, including standard spaghetti, mean and spread, probability, and max/min charts, as well as specialized multi-parameter convective fields and post-processed calibrated probabilities for the occurrence of thunderstorms, dry thunderstorms, and severe thunderstorms.

The EMC has also been running an experimental 4.5 km WRF-Non-hydrostatic Mesoscale Model (WRF-NMM4) for the SPC since April 2004. This model is run with 4.5 km grid length and 35 vertical levels over a domain covering approximately the eastern three-fourths of the United States. It is currently initialized from a cold start once daily at 0000 UTC using initial and lateral boundary conditions from the operational North American Mesoscale model, and provides forecasts through a 36 hour period. Over the last two years, the WRF-NMM4 has been periodically upgraded and currently runs using the community WRF version 2 framework. Several unique WRF-NMM4 products have been developed for use by severe weather forecasters, including simulated reflectivity and a measure of updraft rotation in model-generated storms.

The incorporation of SREF and high resolution WRF-NMM4 guidance into an operational severe weather forecasting environment already dealing with high volumes of observational and model data requires careful assessment of the unique strengths of each modeling system, and knowledge of the specific needs of SPC forecasters. Since the SPC severe weather forecast mission focuses on phenomena smaller than that predicted by mesoscale models, such as tornadoes and severe thunderstorms, the traditional forecast methodology has focused on first predicting the evolution of the mesoscale environment and then determining the spectrum of convective storms a particular environment may support. SREF output has been found to be particularly useful in quantifying the likelihood that the environment will occupy specific parts of convective parameter space, as well as the likelihood and timing for thunderstorms and severe thunderstorms to develop over Outlook-scale regions. While this can be extremely helpful to SPC forecasters, more detailed information about the intensity and mode of storms is also needed, since the type of severe weather (e.g., tornadoes, damaging wind) is often strongly related to convective mode. The value of the high resolution WRF-NMM is most evident here, as it has capability to resolve near storm-scale convective characteristics, such as the development of discrete cells ahead of a line of storms, and the development of model storms with rotating updrafts.

A number of major tornado and severe thunderstorm occurred during the period from late Fall 2005 into the Spring of 2006. We examine the complementary role of SREF and high resolution WRF-NMM output during several strongly-forced and weakly-forced severe weather days during this period, and illustrate the operational application of these model datasets in the SPC decision-making process for both Convective Outlooks and Watches.