台灣地區臭氧濃度空間及時間分佈特性

機構典藏 > College of Engineering > Graduate Institute & Department of Water Resources and Environmental Engineering > Thesis > Item 987654321/35993

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/35993

Title:	台灣地區臭氧濃度空間及時間分佈特性
Other Titles:	A characterization of the spatial and temporal variations of ozone concentrations in Taiwan 臺灣地區臭氧濃度空間及時間分佈特性
Authors:	張勝富;Chang, Sheng-fu
Contributors:	淡江大學水資源及環境工程學系碩士班江旭程;Chiang, Hsu-cherng
Keywords:	主成份分析;經驗正交函數;群集分析;Principal Component Analysis;Empirical Orthogonal Functions;Cluster Analysis
Date:	2006
Issue Date:	2010-01-11 07:26:27 (UTC+8)
Abstract:	本研究主要利用統計分析經驗正交函數（Empirical Orthogonal Functions，EOFs）、旋轉EOF方法 (REOF) 以及群集分析（Cluster Analysis）進行台灣地區空氣品質均質區的劃分，本研究根據環保署監測站在2000 年至2004 年期間所收集的台灣地區臭氧濃度進行分析，研究中利用最大小時和最大八小時臭氧濃度進行分區，並比較所得的結果。 EOF和REOF法中均採用特徵值為最大的五個EOF負荷作為分區的依據，以EOF劃分均質區後容易將大部份測站劃為同一區而造成解釋困難，而採用REOF方法和群集分析法所得結果相似且較為合理，利用最大小時與八小時臭氧濃度進行劃分結果接近，本研究也針對EOF和REOF所求出的主成分時間序列加以分析，以了解其特性。本研究中並將REOF的主成分時間序列以群集分析的華德法分為十五個類組，以探討臭氧事件日的污染型態，除了求出各種型態的臭氧特性外，還建立轉換表，得到臭氧型態在連續兩日轉換的機率，由此表可以看出臭氧型態具有相當高的持續性。 The Empirical Orthogonal Functions (EOF), the rotated EOF (REOF) and the Cluster Analysis methods were used to carry out the delineation of air quality regions in Taiwan. This study is based on the measurements collected by the air quality monitoring stations operated by EPA (Taiwan) during years from 2000 to 2004. Both daily maximum hour and maximum 8 h ozone concentrations were used for analysis, and their results were compared. The EOF loadings corresponding to the five highest eigenvalues were used to delineate the air quality regions. The results of EOF method were difficult to interpret because they tend to cluster most stations into the first group. The results obtained by rotated EOF and cluster analysis are similar and more reasonable. It is interesting to note that the clustering results based on 1h and 8h data are similar. The principal component time series of EOF and REOF were analyzed to determine their temporal characteristics. The principal component time series of rotated EOF were used to classify the distributions of ozone concentrations into 15 patterns by cluster analysis. The characteristics of each ozone pattern were determined. The transition matrixes showing how many times each pattern made a transition either to itself or to another cluster, from day to day, were determined. High persistency of ozone pattern was noted from these tables.
Appears in Collections:	[Graduate Institute & Department of Water Resources and Environmental Engineering] Thesis

Files in This Item:

File	Size	Format
	0Kb	Unknown	352	View/Open

Loading...