Abstract

Eight years of daily, experimental, deterministic, convection-allowing model (CAM) forecasts, produced by the National Severe Storms Laboratory, were evaluated to assess their ability at predicting severe weather hazards over a diverse collection of seasons, regions, and environments. To do so, f... Show moreEight years of daily, experimental, deterministic, convection-allowing model (CAM) forecasts, produced by the National Severe Storms Laboratory, were evaluated to assess their ability at predicting severe weather hazards over a diverse collection of seasons, regions, and environments. To do so, forecasts of severe weather hazards were produced and verified as in previous studies using CAM output, namely by thresholding the updraft helicity (UH) field, smoothing the resulting binary field to create surrogate severe probability forecasts (SSPFs), and verifying the SSPFs against observed storm reports. SSPFs were most skillful during the spring and fall, with a relative minimum in skill observed during the summer. SSPF skill during the winter months was more variable than during other seasons, partly due to the limited sample size of events, but was often less than that during the warm season. The seasonal behavior of SSPF skill was partly driven by the relationship between the UH threshold and the likelihood of obtaining severe storm reports. Varying UH thresholds by season and region produced SSPFs that were more skillful than using a fixed UH threshold to identify severe convection. Accounting for this variability was most important during the cool season, when a lower UH threshold produced larger SSPF skill compared to warm-season events, and during the summer, when large differences in skill occurred within different parts of the continental United States (CONUS), depending on the choice of UH threshold. This relationship between UH threshold and SSPF skill is discussed within the larger scope of generating skillful CAM-based guidance for hazardous convective weather and verifying CAM predictions. Show less