頁籤選單縮合
題名 | 以群集分析法探討水庫優養化之動力研究=Cluster Analysis of Water Quality Factors of Eutrophication in the Te-Chi Reservoir |
---|---|
作者 | 陳鴻烈; 蔡大偉; | 書刊名 | 水土保持學報 |
卷期 | 41:2 2009.06[民98.06] |
頁次 | 頁197-224 |
分類號 | 445.212 |
關鍵詞 | 優養化; 群集分析; 水質因子; 德基水庫; Eutrophication; Cluster analysis; Water quality factor; TE-Chi reservoir; |
語文 | 中文(Chinese) |
中文摘要 | 本研究是以德基水庫優養化變化趨勢之相關重要參數為研究對象,運用群集分析法來進 行資料最佳分類的工作,以協助水庫優養化管理策略的制定。研究中是以先經過單迴歸分析篩 選出的優養化重要因子為對象,透過多元複迴歸分析所形成之優養全模式為基礎來進行分析。 研究中使用TSI (Trophic State Index)作為分群結果探討檢驗的指標,並採用階層與非階層群集 分析2 種方法來分群資料。研究結果顯示,階層分析結果中以馬氏法最符合優養指標TSI 的變 化趨勢,該方法建議將觀測組分為4 群,而離群值修正前後會影響分群結果,且以修正後的分 群效果較佳。另外,非階層分析結果顯示以4 群為準時,其分群效果尚稱顯著。至於模式中各 個變數根據R2 值排序結果依序為磷酸鹽、葉綠素a、COD、懸浮固體、濁度、總磷、水溫、透 明度、硝酸鹽、溶氧、氨氮及鈉,其中以磷酸鹽最為重要。若將群集分析結果與鑑別分析作一 比較,可發現分群是以磷因子最為重要,中高度優養的分群效果較為明顯,但兩者分析結果的 差異亦指出應該有更佳的分類指標存在。 |
英文摘要 | The research used the cluster analysis to classify the original water quality data of eutrophication in the Te-Chi reservoir. We wished the results could enhance and help the eutrophication management of reservoir. The full eutrophication model formed from the multiple regression analysis was utilized to analyze the important trophic factors selected by the simple regression analysis. The TSI index was adopted as classification standard, as well as two methods of hierarchical and disjoint clustering were used to categorize the data. The results of hierarchical clustering showed that McQuitty’s method mostly corresponded to the TSI variation. This method suggested the data should be separated into four clusters. And the cluster results after calibrating the outliers were better than that of before. Additionally, the results of disjoint clustering also indicated the effect of four clusters was acceptable. Furthermore, the R2 values of variables in the model was were phosphate, chlorophyl-a, COD, suspended solids, turbidity, total phosphorus, water temperature, secchi depth, nitrate, dissolved oxygen, ammonia nitrogen, and sodium in order. Among them, the most important variable was phosphate. To compare the results of cluster and discriminant analysis, we found phosphorus was the most important factor in the cluster studies, and better effects of cluster occurred in the middle and heavy eutrophication conditions. However, the differences between both methods also indicated that it is expectable for a better index of cluster. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。