台灣港灣地區颱風波浪推算之應用研究

Abstract

本計畫的主要目的係利用港研中心所收集到之歷年氣象資料，包括花蓮、蘇澳、台北港等海域測站的波浪資料進行分類分析，同時應用類神經網路及模糊理論之方法分析颱風波浪、颱風波高、颱風週期、氣象及地理環境的關係特性，建立一個颱風與波浪的關係模式。 另外，由建立之颱風與波浪特性關係後，建立輸入颱風預報資料即可推算颱風波浪之模式。

A hybrid neural and fuzzy network model for forecasting typhoon waves was developed in this study. The present consists of two parts. The first part is to set up a database for typhoon data and the corresponding wave conditions. The database is used to promote the predictive capability of the neuro network model. The other part is to connect the typhoon waves to typhoon condition by using fuzzy-neuro network. The wave data were collected by Hwa-Lien,Su-Ao and Taipei harbor. Comparisons with observed wave height shows that the present model has high capability to predict wave height from typhoon data.