氣候變遷所造成的乾旱加劇了地下水的使用頻率，台灣南部地區為農業生產要地，為了更加瞭解土地利用與地下水位變化的關係，本研究以主成分分析對台南地區2004至2013年地下水月平均水位進行分析，萃取出四個含水層之主成分，再以地理加權迴歸分析土地利用與地下水之間的相關性，進而由各主成分推測，研究區域地下水位的空間變化影響力，判斷研究地區地下水位變化的原因，以利對地下水位的變化趨勢有更進一步的瞭解。主成分萃取出的各含水層前四個主成分可解釋原始資料80%以上的訊息，各含水層特徵向量高值地區都與地下水位站有高度相關，不過隨著主成分個數與含水層深度增加其相關性則逐漸減少，表示在淺層含水層，地下水位變化明顯直接受地下水位站附近人為與自然因素所影響，深層部分則會受到其他因素影響。地理加權迴歸分析方面，影響地下水位空間分佈也因為不同季節與不同土地利用而有所不同，其中以第四含水層林業與養殖用地所造成影響力特別明顯且顯著，養殖用地在冬季沿海地區(七股區)其影響力最為明顯，林業用地在夏季山區(左鎮、龍崎區)其影響力最為明顯，雖然其他季節與土地利用類型迴歸模型顯著性不高，但地理加權迴歸分析能更細緻的呈現不同時間尺度影響地下水位程度，有助於瞭解地下水與土地利用的關係，可作為預測地下水位變化趨勢之參考。
Drought caused by climate change may increase frequency of groundwater use in the southern Taiwan which is the major producer of agricultural products. In order to understand the relationship between land use and groundwater change, this study uses PCA(Principle Component Analysis) to analyze monthly average groundwater level from 2004 to 2013 at Tainan City. Components are analyzed using GWR(Geographically Weighted Regression) to compare land use and to conjecture the driving force of groundwater level change. Results show that the principle components in every aquifer can explain more than 80% of the ratio of the original data. The principle component results (high eigenvalues) of each aquifer are highly correlated with the groundwater level stations, but their correlation decreases with the increase of the number of principal components and the depth of the aquifer. This indicates that in shallow aquifers, the change of groundwater level is obviously directly affected by man-made and natural factors near groundwater stations, while the deep aquifers are affected by other factors. In GWR, the spatial distribution of groundwater level is also different because of different seasons and different land use. Among them, the influence caused by the fourth aquifer forest and aquaculture land is significant. The influence of the aquaculture land in winter coastal areas is most significant. The forest land in summer mountain area has the most obvious influence. Although regression model of seasons and land use types do not exhibit significant meanings, the GWR can still demonstrate groundwater level changes accurately under different time scales. It can also provide important information for groundwater resource management.
