地表水與地下水間之交互作用不僅在水文生態系統中扮演著極為重要的角色,對於整個水文循環甚至資源的規劃與管理亦是不可或缺。受到外在環境變異的影響,地表水與地下水間的交互作用極為複雜,交互作用的過程不僅在時間上充滿著變異,在空間上亦呈現劇烈地變動,若採集的樣品數量很大且檢測的項目也很多時,所產生的大批數據往往容易造成混淆,且不亦判讀。本文採用中央研究院統計研究所開發廣義相關圖(generalized association plots, GAP)統計分析方法,針對濁水溪沖積扇地下水補注地質敏感區內某一工業區,以及周遭農業活動等屬複合污染源污染型態之地下水與地表水水質進行統計分析,主要歸納工業源污染性質特徵以導電度、總溶解固體等項目關聯性較高,資料差異性大者多在工業區潛在污染事業座落位置及與污水排放之匯流位置,與工業區內工廠排放廢污水有關;硝酸鹽氮偏高趨於大範圍之區域性特徵,主要由工業區外農業源流入工業區內。本文將大量潛在相關檢測項目簡化成資料矩陣視覺化方式呈現,讓研究者在對於預期的污染源有基本觀念的背景下,藉著GAP統計方法得到一些污染源解析,進一步對統計的結果作合理的解釋。
The interaction between surface water and groundwater plays an extremely important role in the ecological system, and is also indispensable for the planning and resources management of the entire hydrological cycle. Due to external environmental variability, the interaction between surface water and groundwater is extremely complex. The interaction process is not only full of variation in time, but also is drastically in space. It causes large amount of water samples and lot of analysis items, whose data generated are often confusing and difficult to interpret. In this paper, the statistical analysis method of generalized association plots (GAP) developed by the Institute of Statistics of Academia Sinica is applied for the interpretation of groundwater and surface water in the compound pollution source of an industrial zone and agricultural area in the Groundwater Recharge Geologically Sensitive Areas of the Jhuoshuei River Alluvial Fan. According to the analysis results, the characteristics of industrial source pollution is mainly concluded that the conductivity and total dissolved solids are highly correlated, whose large differences of survey locations are mostly related with the discharge of waste water from potential factories industrial areas. The high concentration of nitrate nitrogen tends to be a reginal due to agricultural sources from outside the industrial areas. The GAP statistical method can simplify a large number of potentially relevant analytical items into a data matrix and presents them in a visual format, which is allowing researchers to have a basic concept of the expected pollution sources and can explain the statistical results reasonably and furtherly.