桃園地區土地利用變遷與影響因素之空間分析

Abstract

工業區開發與交通建設吸引大量人口移入，使桃園縣近年發展迅速，並帶動土地利用變遷。受歷史、政治脈絡影響，呈現北桃園、南中壢的雙元發展特性。然而過往土地利用變遷研究多忽略地理分布現象所具有的空間特性，使模式推導產生偏誤，對此空間統計方法的創新有助於處理此類空間效應問題。本研究目的在於瞭解桃園地區建成地變遷的分布型態與影響因素，並檢視空間效應的影響。先以空間自相關指標偵測建成地分布與變遷的空間型態；進一步利用空間延遲模型與地理加權迴歸探討建成地變遷之影響因素，以及空間延遲相依、空間異質性的作用效力。

研究結果顯示，桃園地區於1995-2006年間建成地大量增加，相對的農業用地流失迅速。建成地分布以桃園市、中壢市為核心，近年在中壢平鎮市區外圍、桃園市區外圍、龜山公西地區發展最快。整體上鄰近村里建成地發展較快、人口與工商業員工成長、位於市區外圍、近交流道、位於都市計畫區或工業區內，工業用地比例較少，空置地及農業用地比例較多的村里，有較高的建成地開發可能性。在空間效應方面，多元線性迴歸中存在空間延遲相依與空間異質性，分別能夠過空間延遲模型與半參數地理加權迴歸有效地修正二者衍生的模型偏誤，並提升模型適配度。由半參數地理加權迴歸的局部迴歸係數得知，人口密度變遷與三級產業員工成長在桃園市北區、蘆竹鄉、龜山鄉、大園鄉等地對建成地變遷有較高的影響性；二級產業員工成長的效應侷限於沿海地區；原工業用地比例於桃園市、八德市負向效應最大；與交流道距離僅在林口、中壢、內壢、大湳等特定交流道周邊有較高影響性；與火車站距離、原農業用地比例、原空置地比例三者在桃園市區、八德市、中壢平鎮市區等核心區域佔關鍵。總體而言，桃園地區建成地變遷因素具有明顯的南北、城鄉差異性。A large amount of immigration attracted by the development of industrial areas and transportation construction causes Taoyuan County to develop rapidly, and leads landuse changes. Influenced by historical and political context, it shows dual developmental traits of Taoyuan City in the north and Chungli City in the south. However, most research on the landuse changes in the past neglects spatial traits of phenomena of geographical distribution, which causes bias against derivation of patterns. The innovation in ways of the spatial statistics helps to handle such problem of spatial effects. The purpose of this research is to help everyone understand the distribution patterns and the influencing factors of the changes of the build-up areas in Taoyuan district, and examine the influence the spatial effects have. This study used spatial autocorrelation index to detect the distribution of the build-up areas and the changes of the spatial patterns; furthermore, we used Spatial Lag Model and Geographically Weighted Regression to probe into the influencing factors of the changes of the build-up areas, and the effectiveness of spatial lag dependence and spatial heterogeneity.

The result of the research shows that the build-up areas increased dramatically in Taoyuan area from 1995 to 2006, and the agricultural land of the counterparts eroded fast. Taoyuan City and Chungli City are the cores of the distribution of the build-up areas, and in recent years the districts have developed fastest in peri-urban areas of Chulgli-Pingzhen as well as in peri-urban areas of Taoyuan City, and Gongsi, Guishan. Overall, the villages listed below have stronger possibility of developing the build-up areas: villages of the build-up areas developing faster neighbour villages, of population and employee growth of business, in peri-urban areas, near interchanges, and in urban planning areas or industrial areas; villages of smaller proportion of industrial land; villages of more vacant land and agricultural land. In the spatial effects, the spatial lag dependence and spatial heterogeneity exist in multiple linear regressions, which can effectively correct the derivative bias in pattern respectively with the Spatial Lag Models and Semiparametric Geographically Weighted Regression, and elevate goodness of fit of models. We can know from local regression coefficient of Semiparametric Geographically Weighted Regression that, the changes of population density and growth of employees of tertiary sector in such areas as the northern part of Taoyuan City, Luzhu Township, Guishan Township, Dayuan Township, have higher impact to the changes of built-up area; effects of growth of employees of secondary sector are confined to coastal districts; the proportion of negative effects is the largest in original industrial land of Taoyuan City and Bade City; the changes of build-up areas have higher impact on the distance from interchanges—only around specific interchanges, such as Linkou, Chungli, Neili, Danan—; the distance form train stations, the proportion of original agricultural land, and the proportion of original vacant land, these three in such core areas as Taoyuan City, Bade City, Chulgli-Pingzhen City, are key to the changes of the build-up areas. In totality, the factors of the changes of build-up areas in Taoyuan area show great differences between the north and south, as well as urban and rural.

Keywords: Taoyuan, landuse change, spatial effects, Spatila Lag Model, Geographically Weighted Regression