Land-Cover Mapping of Boreal Regions through Landsat and SPOT MultispectralImages. A Case Study: The Skedsmo Area in Southeast Norway

摘要

This report discusses mapping of boreal vegetation by the use of Landsat and SPOTmultispectral images. Such images are based on spectral signatures of vegetation, and consist of a series of spectral bands. The reflectance values within the bands may be used to infer the vegetation canopy. First, the authors discuss visual analysis through false-color composite images. They are based on selecting three spectral bands and assigning them to the three primary colors of the display device. The authors derive a color scheme for recognizing major types of boreal vegetation across various band combinations. Second, the authors perform a digital analysis through unsupervised classification using the Isodata clustering algorithm. This method will always partition the data into groups, but these may not necessarily correspond to distinctly different vegetation types.