Analysis on urban land cover classification in Taipei City using FORMOSAT-2 MS images

Abstract

Impervious surface ratio (ISR), green cover ratio (GCR) and normalized difference vegetation index (NDVI) are recognized as keys in assessing urban environments, and respectively apply to assess urban flood phenomenon, vegetation quantity and quality. This paper classified the surfaces in Taipei City into impervious road and building categories, and pervious bare land, vegetation and water body ones using Taiwan FORMOSAT-2 MS data by remote sensing methods. The overall classification accuracy of 90.56% for the five urban land cover types was obtained. In qualitative analysis, four digital index maps were drawn, the land cover map, the ISR map, the GCR map and the NDVI map, to present the distribution characteristics of the indexes above in Taipei City. Researches showed that there is a medium high relationship between the GCR map and the NDVI map (R 2=0.8678). In quantitative analysis, all the component of the five land cover categories, ISR and GCR of the 12 districts in Taipei City were calculated basing on the classification data. The results present that both the household density (r=0.96) and the population density (r=0.95) are highly related to ISR, and the ratio of reserved area (r=0.97) is related to GCR. In addition, the ISR is highly related to the GCR (r= -0.98). According to the results above, there are several policies recommended to reduce ISR in this paper: to planting trees, to establish open water spaces, and to improve the pervious performance of roads and pavements especially in high population area. Finally, appropriate reserved area in urban is also suggested to enhance GCR.