A Nested Spectral Model for Hurricane Track Forecasting

Abstract

A numerical method for analysing and forecasting a wide range of horizontal scales of motion is tested in a barotropic hurricane track forecast model. The numerical method uses cubic B-spline representations of variables on nested domains. The spline representation is used for the objective analysis of observations and the solution of the prediction equations (shallow-water equations on a Mercator projection). This analysis and forecasting system is referred to as VICBAR (Vic Ooyama barotropic model). The VICBAR model was tested in near real time during the 1989 and 1990 Atlantic hurricane seasons. For the 1989 season, VICBAR had skill comparable to, or greater than, that of the operational track forecast models. For the, 1990 season, VICBAR had skill comparable to that of the operational track-forecast models. During both 1989 and 1990, VICBAR had considerably more skill for forecasts of hurricanes than for forecasts of tropical storms. For the 1990 season, VICBAR was generalized to include time-dependent boundary conditions from a global forecast model. These boundary conditions improve the longer-range forecasts (60?72 h). The skill of VICBAR is sensitive to the choice of the background field used in the objective analysis and the fields used to apply the boundary conditions. The use of background fields and boundary-condition fields from a 12-h-old global model forecast significantly reduces the VICBAR skill (versus the use of fields from the current global forecast).