Summary: | 博士 === 國立臺灣大學 === 生物環境系統工程學研究所 === 99 === Understanding the landcover pattern in a region is essential for landuse planning and resources management. In this study ALOS multispectral images were used to compare landcover patterns in three study areas, namely Tokyo, Kyoto, and Taipei, of different de-grees of urbanization. From the results of landuse/landcover classification, Shannon diver-sity index at cell level was used for landcover pattern analysis. Existing landcover pattern of the three study areas were also compared by investigating cell distribution in a landcover coverage-ratio space. Both the landcover type richness and evenness are low in the Tokyo study area and built-up is the single dominant landcover type in almost all cells. In compar-ison, landcover patterns of the Kyoto and Taipei study areas are more diversified, with sig-nificant amount of cells having mixed and non-dominant landcover types. Kyoto is least urbanized and enjoys a good mixture of different landcover types. It was found that cell-average NDVI alone can be used for delineating areas of certain dominant landcover types. Implementation of such method does not require an a priori LULC classification, and thus is particularly useful when good training data for LULC classification are not available. An urbanization index which integrates the coverage-ratio of built-up landcover type and the cell-average NDVI was proposed and used to explore the spatial variation of degree of ur-banization. Area-average urbanization indices of the Tokyo, Kyoto, and Taipei study areas were calculated to be 0.91, 0.55, and 0.72, respectively. Such results are consistent with the results of qualitative evaluation using different landscape metrics.
|