考慮海陸風及湧浪特性的波高與風速之迴歸分析

Abstract

本文在考濾風向分區及波浪分類下，以迴歸分析建立臺北港風速與波浪的關係。採用的風速及波浪資料為港研中心於臺北港外海觀測樁的2010及2012年的觀測值。本文首先探討海陸風轉向時間發現，海陸風轉向的月份常發生在5-9月，而在2010年及2012年的陸風轉海風的平均時間分別為8時30分及8時50分，而下午海風轉陸風的平均時間分別為17時10分及17時。

本文測試三種不同風向分區後，獲得臺北港合適的分區風向方位角 25o 、70o 、205o 、250o。以二次多項式迴歸海風、陸風、西南風及東北風等四區的波高與風速後，合併計算值，再與實測值比較，二者的R2=0.66及RMSE =0.40 m。此二個檢驗指標比不考慮分區的全年樣本的結果(R2= 0.63及RMSE= 0.41m) 為佳。

本文依PM波譜法分類成2010年的波浪分成湧浪、成熟風浪、成長風浪，若再將各類波浪細分風向成4區後，共獲得12組資料。分別進行迴歸，再合併計算結果成全年波高，其與實測值比較的R2=0.76及RMSE =0.33。本文證實同時考慮波浪分類及風向分區的波高與風速而有相似統計特性時，迴歸分析的模式表現會明顯優於只考慮風向分區或波浪分類的結果。

This paper investigates the relationship between wave heights and wind speeds using quadratic polynomial regression analysis considering clusters of wind directions and kinds of wave properties. The data of wind and at the observation pole of the Taipei harbor for 2010 and 2012 wave were measured by the Harbor and Marine Technology center. The phenomenon of changing sea/land breezes during each day commonly occurs from May to September. The time of changing land breeze into sea breeze is about am 8:30 and am 8:50 and the time of changing sea breeze into land breeze is about pm 17:10 and pm 17:00, respectively, for 2010 and 2012. Three kinds of clustering wind directions were examined. Separated azimuths of wind directions being 25o, 70o, 205o and 250o are suggested. Higher R2=0.66 and RMSE =0.40 m in model performance are evaluated depending on clusters of wind dirctions than those of R2= 0.63and RMSE= 0.41m by whole the data.

Based on the Pierson & Moskowitz (1964) spectrum observed waves are separated into swells, growing waves and fully developed seas of which are divided into 12 groups of date by 4 clusters of wind directions. Calculated wave heights by the fitted quadratic equation on each group are combined to have a model performance of R2=0.76 and RMSE =0.33 m. The results show that regression analysis on data with consistent statistics has good model performance.