Objectively Determined Model-Derived Parameters associated with Forecasts of Tropical Cyclone Formation

During the 2005 North Atlantic hurricane season, an objective tropical cyclone vortex identification and tracking technique was applied to analyzed and forecast fields of three global operational numerical models- the National Centers for Environmental Prediction Global Forecast System (GFS), the Navy Operational Global Atmospheric Prediction System (NOGAPS), and the United Kingdom Meteorological Office model (UKMET). For the purpose of evaluating each model's performance with respect to forecasting tropical cyclone formation, 14 relevant parameters are cataloged for every tropical vortex. In this study, nine of the fourteen parameters are subjected to a linear discriminant analysis applied to all forecast vortices that exceed vorticity and warm core thresholds. The goal is to determine the combination of parameters for each model, at each 12-h forecast period to 120h, that best discriminates between a vortex that is correctly forecast to intensify into a tropical cyclone (developer) and a vortex that is forecast to intensify into a tropical cyclone, but does not (false alarm). The performance of the resulting discriminant functions are then assessed using the Heidke Skill Score and Receiver Operating Characteristic curves. Overall, the methodology applied to forecasts from the UKMET model shows the most skill with regard to identifying correct forecasts of tropical cyclone formation.