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

During the 2005 tropical cyclone season over the western North Pacific, eastern North Pacific, and North Atlantic, an objective tropical cyclone vortex identification and tracking technique was applied to analyzed and foreacast fields of three global operational numerical models for the purpose of evaluating each model's performance with respect to forecasting tropical cyclone formation. Analyzed and forecast vortices, defined by 850 hPa relative vorticity above a threshold value are automatically identified, tracked, and cataloged in a database. For each tropical vortex in the database, fourteen parameters in model analyses and forecasts that are relevant to tropical cyclone formation (i.e., vertical wind shear, midlevel moisture, midlevel warm core) are also cataloged in the tropical vortex database. Therefore, the database structure allows for detailed analysis of forecast performance with respect to a variety of parameters relevant to tropical cyclone formation.

In this study, forecasts of tropical vortices made by the National Centers for Environmental Prediction's Global Forecast System, the United States Navy Operational Global Atmospheric Prediction System, and the United Kingdom Meteorological Office Global Model are analyzed with respect to physical quantities that are relevant to tropical cyclone formation. While summarizing statistics such as false alarm rates and probability of detection are readily identifiable from the database, this study seeks to examine the potential for correct forecasts of tropical cyclone formation in each model. The set of fourteen parameters are subjected to a discriminant analysis to examine the most relevant parameters that correctly identify a potential for intensification of a tropical vortex to a tropical cyclone. To place the analysis in a probabilistic framework, the discriminant analysis is applied in a fuzzy framework such that each physical parameter is assigned a weight that identifies a weighted linear combination of parameters that best identify the potential for tropical cyclone development given the forecast parameters. This type of analysis is made possible by the objective identification and catalog of model parameters relevant to each tropical vortex. This procedure allows for continuous assessment of model performance with respect to a variety of parameters and vortices. Based on the comprehensive database, the discriminant analysis is able to consolidate a comprehensive set of parameters for the purpose of assessing the development potential given the forecast associated with each tropical vortex. A comparison of each model's ability to correctly identify tropical vortices that develop into a tropical cyclone will be presented.