Verification, Application, and Improvements for a Statistical-Dynamical Tropical Cyclone Genesis Forecast Guidance Tool

Recently a statistical-dynamical tropical cyclone genesis tool based on global model output was developed to serve as guidance for the National Hurricane Center’s Tropical Weather Outlook (e.g., Halperin et al. 2017; http://moe.met.fsu.edu/modelgen). This presentation will discuss the recent verification statistics of the forecast tool. While the tool has produced some fairly well-calibrated forecasts, some weaknesses were identified. Specifically, it often fails to detect tropical cyclone (TC) genesis events that originate at relatively high latitudes from baroclinic pathways. Thus, the presentation also will discuss ongoing tests of the sensitivity of the forecast tool to changes in the criteria defining a TC used in the automated tracking algorithm. Finally, TC genesis forecast verification statistics comparing the performance of the 2016 and 2017 operational GFS configurations will be presented. Results indicate that the 2017 operational GFS exhibits a smaller false alarm ratio over the North Atlantic basin compared to the 2016 configuration. The 2017 operational GFS is more cyclogenetic than the 2016 configuration over the eastern North Pacific basin, exhibiting a larger probability of detection, but also a larger false alarm ratio. The presentation will close with an overview of planned improvements to the existing version of the real-time forecast tool.