Book Description
Despite its spatial confines, the east Pacific basin is one of the most active basins in the world for tropical cyclone (TC) genesis. While the TCs that form in the basin have important implications for Central America and the southwestern U.S., relatively little research (compared to other tropical basins) has been done on eastern Pacific tropical cyclogenesis. The present study uses two statistical techniques - linear discriminant analysis (LDA) and a Bayesian probabilistic model to identify those variables that are associated with the development of nascent vortices in the east Pacific and uses them to predict tropical storm formation for lead times out to 48 hours. All nascent vortices that last for a minimum of 48 hours and form during the 2001-2009 "peak" hurricane seasons (July-September) are considered in the study. An initial set of 27 spatially averaged variables is considered as potential predictors for the statistical models. Results from both the LDA algorithm and Bayes probabilistic model show that a number of predictors improve the forecast skill of both models. These predictors include the 900hPa relative vorticity, latitude of the vortex, 900hPa deformation fields, 900hPa-500hPa relative humidity, 900hPa zonal wind, and the 900hPa-200hPa equatorward vertical shear of the meridional wind. Using the aforementioned predictors as a basis, a composite-based conceptual model for environmental elements that favor tropical storm formation is constructed to explain the physical mechanisms of the process. In conjunction with the statistical and composite-based models, Tropical Storm Gil (a 2007 storm that is accurately forecast by the statistical models) and Tropical Storm Enrique (a 2009 case that is poorly forecast) are used to highlight the strengths and weaknesses of each model.