| 摘 要： | In this paper, a Markov-type analog model for the prediction of the tropical cyclone motion is developed for the Northwestern Pacific. The probabilities are studied by using the United States National Hurricane Center HURRAN model as a guide. In order to approach and suit to the new circumstance continuously, a Markov process has been used to describe the probability forecast for typhoon movement. The model is different from the traditional analog method both in searching probability solutions and in renewing initial data.A transplantation model (Model 1) has also been developed by using the traditional analog methods. Both Model 1 and Markov-type analog model (Model 2) have been tested for 95 simultaneous cases, which are independent data for the years 1981 and 1982. It has been shown from this examination that the Markov-type model appears to give smaller forecast errors.
|
