以多變量分析法探討水庫優養化之動力研究

Multivariate Analysis on Water Quality Factors of Eutrophication in the Te-Chi Reservoir

陳鴻烈(Paris Hong-Lay Chen) ； 蔡大偉(David D-W. Tsai)

《坡地防災學報》 8卷3期 (2009/12) Pp. 1-22

https://doi.org/10.29995/JSHR.200912.0001

摘要

本研究是利用多變量分析統計方法來尋求造成德基水庫優養化的動力，了解形成的動力可協助該區優養化問題的管理規劃。研究以多變量統計原理為基礎，尋求優養形成之動力，研究中可分為(1)尋求可能優養動力因子、(2)複迴歸模式、(3)離群值的探討、(4)共線性的討論、(5)共線性的修正、(6)優養主要動力因子等6部分，第1部分藉由敘述統計的資訊得知該區水質狀況，並用單迴歸分析尋找與優養相關的動力因子。第2部分則是應用複迴歸分析來建立基本優養模式。而第3部分則是使用統計方法尋找可能離群值，在經過離群值修正後，模式R^2值由0.9116提升至0.9664。第4部分則是運用共線性分析來了解模式共線性情形，結果找出懸浮固體、COD、葉綠素a三變數共線最嚴重。第5部分則是使用複迴歸模式選擇法與主成分分析法來消除共線的問題，並更進一步優化模式。第6部分則是綜合討論所有多變量分析結果來深入尋求動力。最後研究結果顯示，此區優養最為重要的動力因子為營養鹽磷與藻類濃度指標。

關鍵字

優養動力 ； 多變量分析 ； 優養化

並列摘要

This study was to search for major factors which causing Te-Chi reservoir eutrophication dynamics by the multivariate statistics analysis. We wish the results could enhance and help the management of reservoir's eutrophication. The investigations were divided to six parts: (1) to look for the factors of eutrophication dynamics, (2) to study multiple regression model, (3) to explore the outliers, (4) to discuss colinearity, (5) to correct colinearity problems, and (6) to find out major factors of eutrophication dynamics. The first part was to understand the water quality conditions by the descriptive statistics, and to search for important factors of eutrophication dynamics by simple regression analysis. The second part was to establish the basic eutrophication model by multiple regression analysis. The third was to look for the outliers by statistic methods. The results showed that R^2 values of the model could be raised to 0.9664 from 0.9116 after correcting the outliers. The fourth was to analyze the colinearity problems in the model. We found the most serious colinearity was occurred among suspended solid, COD, and chlorophyll-a. The fifth was to solve the coliearity problems by using multiple regression model selection and principal component analysis to improve the model. The sixth part was to discuss the results from all multivariate statistics analyses to find out the most important factors of eutrophication dynamics. Final results presented that the most important factors were phosphorus nutrition and algae concentration.

並列關鍵字

Eutrophication dynamics ； Multivariate statistic analysis ； Eutrophication