Probabilistic Forecast Guidance for Severe Thunderstorms Based on the Identification of Extreme Phenomena in Convection-Allowing Model Forecasts

Abstract

Abstract With the advent of convection-allowing NWP models (CAMs) comes the potential for new forms of forecast guidance. While CAMs lack the required resolution to simulate many severe phenomena associated with convection (e.g., large hail, downburst winds, and tornadoes), they can still provide unique guidance for the occurrence of these phenomena if “extreme” patterns of behavior in simulated storms are strongly correlated with observed severe phenomena. This concept is explored using output from a series of CAM forecasts generated on a daily basis during the spring of 2008. This output is mined for the presence of extreme values of updraft helicity (UH), a diagnostic field used to identify supercellular storms. Extreme values of the UH field are flagged as simulated “surrogate” severe weather reports and the spatial correspondence between these surrogate reports and actual observed severe reports is determined. In addition, probabilistic forecasts [surrogate severe probabilistic forecasts (SSPFs)] are created from each field’s simulated surrogate severe reports using a Gaussian smoother. The simulated surrogate reports are capable of reproducing the seasonal climatology observed within the field of actual reports. The SSPFs created from the surrogates are verified using ROC curves and reliability diagrams and the sensitivity of the verification metrics to the smoothing parameter in the Gaussian distribution is tested. The SSPFs produce reliable forecast probabilities with minimal calibration. These results demonstrate that a relatively straightforward postprocessing procedure, which focuses on the characteristics of explicitly predicted convective entities, can provide reliable severe weather forecast guidance. It is anticipated that this technique will be even more valuable when implemented within a convection-allowing ensemble forecasting system.