台灣河川流量對河川污染性質的影響

水資源為人類社會中重要資源，攸關每個人的日常作息，而台灣的淡水水資源主要來源以水庫以及河川為主。由於台灣地小人稠，不管是在平原地區，還是高山上，均有人為行為存在。然而，在部分的水庫以及河川之集水區周遭土地正遭受土地過度開發或是土地、水源污染行為的影響，並且台灣氣候變化多端，常有強降雨的發生，在水土保持不完全以及受污染的土壤及水源的狀態下，往往會將污染物質帶進自然水體中，造成淡水資源的污染，因此，需對受污染的淡水水源加以關注，常用於表示水體水質標準的指標包含河川污染指標RPI以及水質指標WQI。目前，有關水資源的研究中，常利用相關性分析、迴歸分析以及多變量分析研究水體污染的原因及水質特性。本研究以敘述統計、相關性分析、迴歸分析以及主成分分析，探討台灣河川逕流量對河川污染性質的影響。本文以全台25條重要河川的水質監測資料以及流量資料作為研究樣本，藉由各水質數據完整度進行篩選，蒐集2016年01月至2020年10月間日平均逕流量、月平均逕流量、河川污染指數、水溫、酸鹼值、導電度、溶氧(電極法)、生化需氧量、化學需氧量、懸浮固體、大腸桿菌群以及氨氮等總計12項之水質參數，進行後續之分析研究。研究結果顯示，河川逕流量不管是在不同時間長短或是不同地區的分析，均與水中懸浮固體濃度最為相關，均呈現中度相關以上，相關係數r最高為0.577，而河川逕流量與東部河川之河川污染指數最為相關。在不同時間長短以及不同地區的單迴歸分析結果，氨氮始終為解釋能力最高之水質參數。由複迴歸分析結果顯示，當分析時間拉長，可對河川污染指數解釋的水質參數增加。然而，由於各水質參數對河川污染指數的解釋能力均不相同，因此，迴歸全模型對河川污染指數的解釋能力由71.8%減少至67.7%。在不同地區的複迴歸分析結果，對河川污染指數解釋能力最高的的為中部河川，可以對中部河川解釋78.7%的污染情形，南部河川次之，決定係數R2值為0.706，東部河川解釋能力最差，解釋能力為28.1%。由主成分分析對各地區河川之水質參數進行權重再計算之結果得知，相對於BOD來說，COD均得到較高的權重值，顯示COD對河川的污染情形具有較好的解釋。此外，當水質參數以主成分分析後，不同時間長短的分析中，KMO值最高為近五年之分析結果，達0.614，總變異解釋量最高達68.26%，可以萃取4個主成分；不同地區之PCA/FA分析以中部河川的KMO值最高，達0.679，總變異解釋量為65.63%，可以萃取3個主成分。經過因子轉軸後所得之河川污染性質影響因素中，總體萃取轉軸後的因子對河川污染性質的描述能力達到65%以上，其中，河川逕流量對中部河川以及東部河川之河川污染性質的影響能力最為顯著。

關鍵字

河川污染指數；河川逕流量；相關分析；迴歸分析；主成份分析；河川水質

並列摘要

The water resources are the essential resource in human society, which affects the daily life of everyone. The major freshwater resource of Taiwan mainly come from reservoir and river. Due to Taiwan is an island, it does have enough land for the densely populated. Thus , no matter whether it is in the plains or on the high mountains, there still have human activities exist. However, some area of reservoir or the catchment of river are suffering from excessive land development or pollution of land and water sources, and there often occurs heavy rainfall due to the variably climate and weather of Taiwan. With this conditions, in the state of incomplete water and soil conservation and contaminated soil and water sources, pollutants are often brought into natural water bodies, causing pollution of freshwater resources. Therefore, It is necessary to pay attention to contaminated fresh water sources. The indicators commonly used to indicate water quality standards include RPI and WQI. At present, in the research on water resources, correlation analysis, regression analysis and multivariate analysis are often used to study the causes of water pollution and water quality characteristics. This study uses narrative statistics, correlation analysis, regression analysis and principal component analysis to explore the impact of river runoff in Taiwan on the nature of river pollution The research results show that the river runoff is most correlated with the concentration of suspended solids in the water, whether it is analyzed in different time periods or in different regions, showing a moderate correlation or above, and the highest correlation coefficient r is 0.577. The most relevant river pollution index. In the single regression analysis results of different time periods and different regions, ammonia nitrogen is always the water quality parameter with the highest explanatory power. The results of multiple regression analysis show that when the analysis time is prolonged, the water quality parameters that can be explained by the river pollution index increase. However, since the explanatory power of each water quality parameter for the river pollution index is different, the explanatory power of the regression full model for the river pollution index is reduced from 71.8% to 67.7%. In the results of multiple regression analysis in different regions, the central river has the highest explanatory ability for river pollution index, which can explain 78.7% of the pollution situation in the central river, followed by the southern river, the coefficient of determination R2 value is 0.706, and the eastern river has the highest explanatory ability. Poor, with an explanatory power of 28.1%. The results of the recalculation of the weights of the water quality parameters of the rivers in each region from the principal component analysis showed that compared with the BOD, the COD obtained a higher weight value, indicating that the COD has a better explanation for the pollution of the river. In addition, when the water quality parameters are analyzed by principal components, the KMO value is the highest in the analysis results of the past five years, reaching 0.614, and the total variance explained is up to 68.26%, and 4 principal components can be extracted; In the PCA/FA analysis, the KMO value of the central river was the highest, reaching 0.679, the total variance explained was 65.63%, and three principal components could be extracted. Among the influencing factors of river pollution properties obtained after factor rotation, the overall factor extraction and rotation axis can describe the river pollution properties above 65%. Among them, river runoff has the most influence on the pollution properties of central and eastern rivers. Remarkably.