本文嘗試利用環保署所提供之遍佈台灣23縣市之土壤重金屬監測資料,利用多變量統計中各種分析方法,探討八個主要重金屬變量間的相互關係,期能找出影響台灣各地區土壤污染狀況及變化趨勢的主要因子,以及探討八大重金屬變量間的相互關係,其能真正反映台灣各地重金屬之間的差異性及各等級劃分之評價模式。經由因子分析結果顯示影響台灣土壤八大重金屬污染機制的主要因子可以簡化為兩個,分別為由土壤鉻、鎳、銅、鋅所組成的「高氧化態污染因子」;此因子研究也發現四種重金屬變量其負荷程度由大至小為鉻、鎳、銅、鋅,此結果正好與週期表中之數種規則互相謀合。第二因子主要由土壤鎘、砷及鉛所組成,稱之為「半導體潛勢污染因子」,此因子不論過渡金屬之鎘、抑或典型元素中之砷及鉛,皆與半導體化合物之形成有關,且形成半導體之主要機制為金屬鍵。在「群集分析」方面,初步可以將台灣土壤重金屬污染特性劃分為五種不同的群集。最後本研究利用「判別分析」來判別確定待判別樣本的所屬類別,判識結果正確百分比高達96.87%,此結果顯示先前群集分析之分群結果是可以被接受的。本研究所得最後之結果可以提供給相關單位未來策略選擇土壤整治及應用時之參考依據。
The various analytical techniques based on multivariate statistics have been applied to evaluate the 8 soil heavy metal data collected at 23 administrative divisions provided by Environmental Protection Agency in Taiwan. The objective is to evaluate the mutual correlations among the various soil heavy metal parameters in order to reveal the primary factors that affect the soil quality, and the differences among the various soil heavy metal parameters the whole Taiwan. Results of factor analyses lead to 2 simplified principal factors that significantly influence quality of soil in Taiwan. Subsequently, “Cluster Analyses” were conducted to classify the 8 soil heavy metal quality into 5 different clusters to represent the various characteristics, and extent of potential pollution the whole Taiwan. Finally, the discriminant analyses were carried to determine the classification of soil heavy metal samples collected at 23 administrative divisions for confirming the validity of the model results. The results indicate that the accuracy of identifying the soil quality classification is as high as 97.81% and that better soil heavy metal quality determination is achieved for the soil heavy metal samples. These results that are comparable with those obtained with the aforementioned cluster analyses are acceptable. When carrying out the factor analyses, higher number of samples will favor more stable final results. In this study, using soil heavy metal samples collected over a period of three and half years will effectively raise the efficiency and reliability of the factor analysis result that will be valuable references for managing the soil pollution in Taiwan.