蘭陽平原自1991年國道五號開工，2000年農業發展條例修正通過後，因預期心理大量興建農舍，此一農地轉用不僅破壞了農地的生產功能，其對於社會經濟系統以及環境生態系統所扮演的重要文化、調節與支持功能也一併消失。為能進一步探討其農地轉用之影響因子，本研究引用系統生態學（system ecology）的「系統」方法為基礎，透過專家學者與農民之訪談，以農地尺度（farm scale）與過程基礎（process-based）觀點建立蘭陽平原農地轉用系統與假說，並以1977年、1995年至2006年的蘭陽平原的農地轉用為例，整合GIS與逐步迴歸分析方法進行多時期（multi-periods）的分析，其中主要將蘭陽平原劃分為245個網格空間（1.5×1.5公里），以進一步指認出系統中，農地本身特性與鄰域條件於蘭陽平原農地轉用的影響關係。經蘭陽平原農地轉用假說驗證結果顯示，一、農地破碎化會影響農地轉用率，以耕地（含旱作地、稻作地）具有解釋力；二、農地的單位面積土地價格確實會影響農地轉用；三、未轉用廢耕地、稻作地面積比會影響農地轉用率；四、生活機能影響對農地轉用具重要影響力；五、緊鄰住宅、商業、四米（含以上）道路、大專院校的農地，及離河流較近的農地，因周邊投資效益佳，對於農地轉用影響較為顯著；六、農地轉用受周圍的住宅數量影響，並以後期的都市地區較為明顯；七、農地轉用具有次要火車站衛星現象。最後，本研究在農地轉用因子之時間與空間交叉分析後建議，農地保育政策必須更貼近農民決策之農地轉用因子，並且能夠將非農業的潛在轉用影響因子納入整合性的考量，以針對具有空間差異性的轉用因子提出農地保育的空間策略。

Since the construction of National Freeway No. 5 in 1991 and the amendment of the Agricultural Development Act in 2000, real estate prospects have driven a rapid bloom of housing developments in agricultural lands in the Lan-Yang Plains. Such landuse transformations have not only diminished the provisioning services of agricultural land, but also degraded their cultural, regulating, and supporting services. In this study, a systems approach is used to examine various hypotheses regarding the transformation of farms to housing developments in Lan-Yang Plains. Information gathered from interviews with farmers and experts in agricultural land-use planning are used to form a farm-scale, process-based systems model of agricultural land-use change. Land use maps from 1977, 1995, and 2006 are divided into 245 square grids (1.5 km × 1.5 km) and analyzed with GIS to derive farm, neighborhood (within same grid), and external factors. Stepwise regression models with these factors are then used to parameterize the probability of agricultural land-use change in the systems model. Empirical results suggest that the rate of agricultural land-use change is affected by: 1) fragmentation of agricultural lands; 2) unit land price; 3) the area of abandoned fields still developable under current regulations; 4) availability of shopping and services in the same grid; 5) the density of housing in the same grid; 6) proximity of grid to other housings, business districts, roads wider than 4m, university campuses, riverfronts, and other factors that enhance investment prospects; and 7) proximity of grid to secondary train stations. We suggest that agricultural land use policies should be more aligned to the factors driving the land-use decisions of individual farmers, especially because such decisions have spatial implications much larger than the space allotted to housings.