土地利用變遷是人與環境互動展現在地表的結果,土地利用的分佈狀況表現出人類在其所居住土地的活動類型與決策行為。在地區尺度下,土地利用的型態受到自然環境、社會經濟面向因子…等等的影響,這些因子作用於土地決策者,再由決策者反映在土地利用的型態上。為了評估在地區尺度下各類土地利用驅動因子的影響力,並建構土地利用變遷預測模式,本研究以宜蘭水產養殖業作為個案研究,間接剖析養殖業者的決策剖析,預測人的行為決策是否造成養殖的土地利用變遷,而非傳統研究透過土地分佈型態預測土地利用變遷。研究方法主要運用地理資訊系統與統計學的方法,GIS方法包括近鄰分析、環域分析等;統計方法包括卡方分析、t檢定、邏輯迴歸分析來尋找造成養殖土地決策者棄養的驅動因子,將上述因子納入多變量的判別分析來建構棄養行為之預測模式,期望能將個體施為決策與各類驅動因子的互動關係以模式化的方式進行分析與討論。研究結果發現卡方分析、t檢定與邏輯迴歸辨別出九個影響宜蘭地區養殖業者棄養的可能因素;判別分析則顯示棄養與否主要受到技術、自有資金、魚塭病變與距道路距離等四個因素的影響。在判別分析中,整體的預測結果可以達到70.1%的正確率,顯示透過養殖業者的屬性資料即可預測其行為決策的可能性,因此此種分析方法應用於土地利用變遷分析的效果是十分良好的。
Land-use/land-cover change (LUCC), the consequence of the interactions between humans and the environment, signifies the living pattern of human activities and the decision-making process involved in such activities. At the local level, land-use decisions are usually restricted and made by land agents. In order to evaluate the effects of various land-use driving forces and to establish a predictive model for LUCC, this project used Ilan’s aquaculture as a case study and survey data. We employed chi-square test, t-test and logistic regression analysis to find driving forces, and discriminant analysis to establish a predictive model. The results of chi-square test, t-test and logistic regression analysis showed the differences in 9 variables that were possible driving forces for the abandonment of aquaculture. These variables could distinguish abandoned farmer and unabandoned farmer efficiently. The results of discriminant analysis showed that factors determining whether land agents abandoned their aquaculture land or not could be divided into two different groups based on willing to experiment with new farming technologies, capital for the farming costs, fish disease problem and distance to road. The model achieved a high accuracy rate, demonstrating that discriminant analysis is a useful tool for analyzing land use changes and the relationships between various LUCC driving forces.