| 摘要: | 在快速成長的都會區，能準確與及時地描述土地資源的性質、範圍與隨時間的變遷是重要的議題。都市的擴展是導致各種都市環境問題的主因之一，像是空氣品質下降、暴雨徑流增加與導致的洪水、熱島效應與水質惡化等。因此，為了提供準確與有用的資訊處理都市擴展後續引發的種種問題，有必要了解都市擴展的程度和趨勢。
本研究中，以台北都會區作為研究案例。利用該區域多時期Landsat-5 TM (Thematic Mapper)、Landsat-7 ETM (Enhanced Thematic Mapper)與SPOT-5多光譜影像，探討自1990年至2010年，20年間的都市擴展和地表變遷。本研究採用分類後比較法 (post-classification change detection algorithm)，此演算法能使用不同時間點和不同感測器所獲得的衛星影像，提供分類成果的“From-To”差異圖 (difference maps)。
1990、2000、2010三個年度的水體、農地、林地與人造建物4項地表物分類整體精度平均為90.74％。並使用多時期的分類後比較法，比較1990年至2001年與2001年至2010年兩個區間的地表變遷。該差異圖顯示，自1990年至2010年間，人造建物增加了約總面積的16.19％，而林地覆蓋率則下降了14.45％，農業、草地與水體等地表類型也顯著地下降。由馬可夫細胞自動機 (Markov Cellular Automata, CA-Markov) 預測2020年的結果指出，都市人造建物將有進一步主要來自林地區域0.5％ (3.49平方公里) 的擴展。研究成果量化了都會區地表變遷的模式，也驗證了使用多時期Landsat與SPOT影像，可以準確與經濟地達到分析與預測各時期地表變遷的目的，提供土地管理和決策所用。;The importance of accurate and timely information describing the nature and extent of land resources and changes over time is increasing, especially in rapidly growing metropolitan areas. Urban expansion is one of the main reasons responsible for a variety of urban environmental issues like decreased air quality, increased runoff and subsequent flooding, heat island effect, deterioration of water quality, etc. Therefore, it is essential to understand its extent and trend, in order to provide accurate and valuable information for dealing with subsequent issues.
In this work, metropolitan Taipei has been taken as a case study. The urban expansion and land cover change that took place within a span of 20 years i.e. from 1990, to 2010 has been studied, using multi-temporal Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper (ETM), and SPOT 5 multispectral data of the area. The post classification change detection algorithm was adopted in this study, since it has the ability of providing difference maps, from which the “From-To” change information can be generated using satellite images acquired at different times and from different sensors.
The overall four-class classification accuracies averaged 90.74 % for the three years, and a multi-date post-classification comparison change detection algorithm was used to determine changes in land cover at two intervals, 1990-2001, and 2001-2010. The maps showed that between 1990 and 2010, the amount of urban or built-up land increased by 16.19 % of the total area, while forest cover decreased by 14.45 %. Rural cover types of agriculture, which also includes grasslands along with water bodies have also declined significantly. The results from the 2020 Cellular Automata Markov (CA_Markov) projection indicated a further increase of 0.5% (3.49 km2) in the urban built-up class category, which is mainly contributed by forested areas. The results quantify the land cover change patterns in the metropolitan area and also demonstrate the potential of multi-temporal Landsat and SPOT data to provide an accurate, and economical means to map, analyze and project changes in land cover over time that can be used as inputs to land management and policy making. |
