An Application of clustering methods to the sea water off the eastern coast of Taiwan

• Main Authors: Shu-Chen Huang, 黃淑真
• Other Authors: 王鴻龍
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/35248731242118254796

碩士 === 國立臺北大學 === 統計學系 === 95 === An Application of clustering methods to the sea water off the eastern coast of Taiwan By Shu-Chen Huang July 2007 A clustering method is applied to study the seasonal variation of the composition of Kuroshio water east as well as northeast of Taiwan. The hydrographic data was collected by Taiwanese research vessels during 1985-2003. The data are categorized into three periods: annul、southwest monsoon season (May-October) and northeast monsoon season (November-April). 7 variables、the maximum and minimum salinity and temperature、the depths of maximum and minimum salinity and the T-S slope between the two extreme salinities estimated from T-S diagram、are used in the clustering analysis. Two clusters are defined based upon the root mean square standard deviation. Their characteristics are studied and discussed the composition of seawater. A discriminate analysis is applied to define a linear function for each cluster. The function is applied to examine the consistency of result of clustering analysis in statistically. The result of the study indicates that the water east and northeast of Taiwan is primarily Kuroshio water. The seasonal variation is vague. The difference between the two clusters is mainly limited in the upper ocean where the rainfall and river discharge could modulate the composition of Kuroshio water. Although the two clusters in the deep ocean are statistically discernable、their difference is lack of physical support according the suggestion of physical oceanographers.