| Summary: | Approved for public release; distribution unlimited. === An empirical orthogonal function (EOF) representation of relative vorticity is used to forecast recurvature (change in storm heading from west to east of 000 deg N) of western North Pacific tropical cyclones. The time-dependent coefficients of the first and second EOF eigenvectors vary in a systematic manner as the tropical cyclone recurves around the subtropical ridge and tend to cluster in a different region in EOF space. Exploiting this Euclidean distance approach, additional EOF coefficients are identified that best represent the vorticity fields of recurving and straight-moving storms. Classification of an individual case is then into the closest time-to-recurvature in 12-h intervals or straight-moving storm category as measured in multidimensional EOF space. Although rather subjective, the Euclidean method demonstrates skill relative to climatological forecasts. A more objective discriminant analysis technique is also tested. A final version that involves the first six EOF coefficients of the 250 mb vorticity field is useful (72% correct) in identifying recurvers or straight-movers during the 72-h forecast period. Skill in classifying situations within 12-h time-to-recurvature groups is low, but might be improved using other analysis techniques or in combination with other predictors.
|
|---|
