As a complement to observational meteorological data and traditional deterministic numerical model guidance, forecasters at the SPC have found much benefit from utilizing post-processed fields from the NCEP Short-Range Ensemble Forecast (SREF). The SPC's version of the SREF is constructed by post-processing all 21 members of the NCEP SREF, plus the 3-hour time lagged operational WRF-NAM, for a total of 22 members each 6 hours (03, 09, 15, and 21 UTC). Output is available at 3 hour intervals out to 87 hours. The SPC ensemble post-processing focuses on diagnostics relevant to the prediction of SPC mission-critical high-impact, mesoscale weather including severe convective storms. For more information on the SPC SREF, see: www.spc.noaa.gov/exper/sref
Previous studies (e.g. Bright et al. 2005, Bright and Wandishin 2006) have focused on calibrating probabilistic output from the SREF using an "ingredients-based approach" for predicting thunderstorms and severe convective storms. However, even with 3-hourly output and a coarser resolution compared to many operational deterministic models, the SREF can be very helpful for accessing the likelihood and timing of convective development and the potential areal coverage thereof. The diverse 22-member approach can be especially useful in more uncertain areas of convective initiation, such as when large scale forcing for ascent is limited and/or when convective inhibition or "capping" may be problematic. In particular, the model and physics diversity within the SREF that includes three different convective parameterizations [Betts-Miller-Janjic (BMJ), Kain-Fritsch (KF), and Simplified Arakawa-Schubert (SAS)] provides a wider range of performance characteristics. SPC forecasters have learned to take advantage of the differing characteristics of the parameterized convection schemes, especially between the BMJ and KF, to better understand the statistical properties of the SREF convective precipitation forecasts.
Aside from probabilistic and ingredients-based guidance, SREF output such as spaghetti diagrams and postage stamps of convective precipitation can be especially useful for deducing the prospects for severe convective development. With inherent numerical model biases and tendencies in mind, ensemble member clustering (or lack thereof) can at times imply a greater (or lesser) likelihood of convective development by proxy of convective precipitation forecasts. In addition, new experimental SREF-based point plume and probability products diagrams also provide guidance related to convective precipitation, and thus may be a benefit to the aspect of forecasting severe convective development. Case examples will ultimately be used to illustrate the utility of SPC SREF output as related to forecasting the development and coverage of severe convective storms.