Linear multiple regression and discriminant analyses provide estimates
of the errors of track forecasts from a nested barotropic hurricane t
rack forecast model(VICBAR), which was run in the North Atlantic Basin
during the 1989-94 hurricane seasons. Predictors are determined from
the synoptic situation, the magnitude of atmospheric changes in the en
vironment of the tropical cyclone, the consistency between current and
past predictions, and the past performance of the model for each part
icular storm. This technique distinguishes cases in which VICBAR perfo
rms well from those for which it performs poorly and can provide skill
ful operational predictions of model performance to forecasters.