STOCHASTIC SIMULATION OF TROPICAL CYCLONE TRACKS IN THE NORTHWEST PACIFIC REGION WITH CLASSIFICATION MODEL

Abstract

Accurate simulation of tropical cyclone tracks is a prerequisite for tropical cyclone risk assessment. Against the spatial characteristics of tropical cyclone tracks in the Northwest Pacific region, stochastic simulation method based on classification model is used to simulate tropical cyclone tracks in this region. Such simulation includes the classification method, the genesis model, the traveling model, and the lysis model. Tropical cyclone tracks in the Northwest Pacific region are classified into five categories on the basis of its movement characteristics and steering positions. In the genesis model, Gaussian kernel probability density functions with the biased cross validation method are used to simulate the annual occurrence number and genesis positions. The traveling model is established on the basis of the mean and mean square error of the historical 6h latitude and longitude displacements. The termination probability is used as the discrimination standard in the lysis model. Then, this stochastic simulation method of tropical cyclone tracks is applied and qualitatively evaluated with different diagnostics. Results show that the tropical cyclone tracks in Northwest Pacific can be satisfactorily simulated with this classification model.