| 摘要: | 都市人口持續上升、都市面積不斷擴張，為避免不當的開發使用造成土地資源浪費與都市問題，都市的規劃設計與永續發展逐漸受到重視，而一個完善的都市土地利用計畫應考量長期的土地利用變遷過程，預期未來可能的變遷作為參考之依據。分析不同土地利用類別之面積變化是土地利用變遷研究中最直接的量化分析方式，許多都市土地利用變遷研究應用衛星遙測影像判釋建立地表覆蓋的基礎資訊，針對不同時期都市土地利用在二維平面面積的改變分析都市發展情形，並無考慮都市發展在三維垂直結構，如樓高的變化發展。根據國土利用監測計畫的都市面積統計數據顯示，臺北市為臺灣首要的都會地區，西元1980至2010年其都市面積比從21.7%上升至41.3%，都市面積的成長有逐漸減緩的趨勢，顯示了在都市發展飽和的地區，都市二維平面面積的改變可能無法直接反應出地區的都市發展情形。
本研究選定臺北市作為研究範圍，蒐集了西元1969、1980、1991、2002、2007年共五個時期的3D建物模型，分別量化單元空間下的建物面積與建物樓高，以建物覆蓋率(Build Area Ratio，BAR)和整體建物容積(Generalized Building Capacity，GBC)，作為都市二維與三維都市發展之量化指標，利用類神經網路建立都市土地利用變遷模式，進行都市空間結構的模式化，並以類神經網路之敏感度分析法，分析都市土地利用變遷與驅動因子之關聯性。分析結果顯示，臺北市的都市發展在建物面積與建物容積上皆呈現上升趨勢，西元1991年之前都市發展主要為二維空間之建物面積的增長，西元1991年之後都市三維空間之建物容積有明顯的上升情形，顯示臺北都市面積逐漸發展飽和，都市的發展由二維平面的擴張，轉為三維垂直容積的增加，高密度、高樓層的發展取代了橫向的都市土地面積的擴張，因此本研究建議在都市發展的分析，特別是發展飽和之地區(如臺北市)，可由二維平面空間延伸探討三維空間結構的轉變。;Population growth and urban expansion lead urban sustainable development a worldwide issue in recent years. Previous studies have used satellite imagery to explore the land use change and analyzed urban growth by correlating the relationships between land uses and driving factors. Most of them investigate the change of urban development in the planar dimension (building area), while few have been focused on the development of urban structure in the third dimension (building height)—the vertical change of urban. According to the urban land monitoring data, the growth of urban area in Taipei City, the capital city of Taiwan, is becoming slow in the past two decades, and the urban area is considered fully saturated. Taipei City, the capital city of Taiwan, is selected as the study area. This study aims to develop an urban model for assessing the development process of urban development of Taipei City, in both 2D and 3D dimensions. The 3D building models of Taipei City in 1969, 1980, 1991, 2002 and 2007 are used to analyze the change of 2D and 3D urban development. In this study, the machine-learning algorithm—artificial neural networks (ANN) model is applied to assess complex and nonlinear 2D and 3D urban development processes, and the 2D and 3D urban development was quantified by two designed parameters, the build area ratio (BAR) and generalized building capacity (GBC).
Results show that the urban area (2D) represents a higher increasing rate before 1991; however, the building height (3D) represents an inverse trend against the urban area, which has a significant increase after 1991. This study considers that in an intensively developed urban, if the land available for constructing new buildings is becoming limited, the urban development could shift from 2D to 3D—the building height. |
