Evaluation of WRF model output for severe-weather forecasting from the 2008 NOAA Hazardous Weather Testbed Spring Experiment

The NOAA Hazardous Weather Testbed (HWT) conducted the 2008 Spring Experiment over a seven-week period during the peak severe convective season, from mid April through early June. As in recent Spring Experiments, the primary focus was on the examination of convection allowing (Δ = 2-4 km) configurations of the WRF model covering approximately the eastern three-fourths of the U. S. in a simulated severe-weather-forecasting environment. Output from various configurations of the WRF model were provided to the HWT during the experiment by The Center for Analysis and Prediction of Storms (CAPS), the NCEP Environmental Modeling Center (EMC), the National Center for Atmospheric Research (NCAR), and the National Severe Storms Laboratory (NSSL). As in previous experiments, these simulations were evaluated on their ability to predict the location and timing of thunderstorm initiation and evolution, and offer useful information on thunderstorm morphology. In addition, the experiment continued testing and refining a real-time, large domain 10-member convection-allowing storm scale ensemble forecast (SSEF) system provided by CAPS to gauge technical issues related to high performance computing, networking, data transfer and processing, product creation, and workstation display requirements for future high impact weather forecasting initiatives, and the potential benefits of uncertainty information provided by the SSEF.

New endeavors this year included 1) an exploration of the impact of assimilating radar reflectivity and velocity data into SSEF members on short-term forecasts of hazardous convective weather; and 2) a more detailed examination of the relationship between model forecasts of convective storms and model predictions of the environment, focusing on boundary-layer thermodynamic structure, airmass boundaries, and sub-synoptic scale features in the free atmosphere. The goal of both endeavors is to provide specific information to model developers that can guide their efforts to improve various components of the WRF model. A finding from the latter effort is that the perceived value and accuracy of the convection-allowing forecasts is often tied to the characteristics of the lower resolution operational NAM model analysis and forecasts that provide the initial and boundary conditions to most of the WRF models. Mesoscale and even synoptic scale errors in the environment are often large enough in the convection-allowing model domains to greatly diminish the predictive value of the explicit 18 to 30 h thunderstorm forecasts. This highlights the importance of providing an accurate mesoscale environment to the convection allowing models and underscores the need to develop model perturbation strategies that are appropriate for convection-allowing ensembles. In addition, it was also found that incorrect prediction of nocturnal and morning convection would at times erroneously modify the model mesoscale environment during the next afternoon and negatively impact WRF model forecasts of thunderstorms. Finally, these observations emphasize the need to develop tools to diagnose effectively and efficiently areas of high or low forecast confidence, and maybe even recognize errant model solutions or ensemble members to aid forecasters in deciphering the voluminous amount of output available.