Volume XLI-B8
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLI-B8, 1305-1308, 2016
https://doi.org/10.5194/isprs-archives-XLI-B8-1305-2016
© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLI-B8, 1305-1308, 2016
https://doi.org/10.5194/isprs-archives-XLI-B8-1305-2016
© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.

  24 Jun 2016

24 Jun 2016

IMPROVING GLOBALlAND30 ARTIFICIAL TYPE EXTRACTION ACCURACY IN LOW-DENSITY RESIDENTS

Lili Hou1, Ling Zhu1, Shu Peng2, Zhenlei Xie1, and Xu Chen1 Lili Hou et al.
  • 1School of Geometrics and Urban Information, Beijing University of Civil Engineering and Architecture,100044 Beijing China
  • 2National Geometrics Center of China,100036 Beijing China

Keywords: GlobalLand 30, TM, Low-density residents, nighttime light remote sensing image, TM6, TR-NDBI

Abstract. GlobalLand 30 is the first 30m resolution land cover product in the world. It covers the area within 80°N and 80°S. There are ten classes including artificial cover, water bodies, woodland, lawn, bare land, cultivated land, wetland, sea area, shrub and snow,. The TM imagery from Landsat is the main data source of GlobalLand 30. In the artificial surface type, one of the omission error happened on low-density residents’ part. In TM images, hash distribution is one of the typical characteristics of the low-density residents, and another one is there are a lot of cultivated lands surrounded the low-density residents. Thus made the low-density residents part being blurred with cultivated land. In order to solve this problem, nighttime light remote sensing image is used as a referenced data, and on the basis of NDBI, we add TM6 to calculate the amount of surface thermal radiation index TR-NDBI (Thermal Radiation Normalized Difference Building Index) to achieve the purpose of extracting low-density residents. The result shows that using TR-NDBI and the nighttime light remote sensing image are a feasible and effective method for extracting low-density residents’ areas.