Description:

Convection-allowing modeling systems that are designed to predict storm-scale severe and hazardous convective weather phenomena, such as supercells and flash flooding, on short time scales of 0 to 6 hours usually require unique verification techniques to assess forecast performance. In general, verification techniques should be designed to elicit relevant performance information from forecasts that are deemed useful to not only model developers but also to end users, such as operational forecasters. For short-term CAM systems, such as WoFS or HRRR, verification methods might focus on assessing the performance of forecasts for individual thunderstorms and their associated hazards rather than forecast probabilities of hazards over a wider area, as is typically done for later-day and next-day CAM forecasts. This session welcomes contributions from studies with a focus on the use of verification techniques designed for short-term predictions of severe and hazardous convective weather from deterministic and ensemble CAM systems. These verification techniques could include spatial, object-based, or other novel methods. In addition, studies with topics related to the verification of short-term predictions of thunderstorm hazards produced by machine learning models and the sensitivities of those models to different verification datasets are also encouraged to submit to this session.