Landsat data for Samoylov Island (Lena Delta, Siberia) was classified using a k-means unsupervised algorithm in the ENVI 4.7 software. Unsupervised classifications are based solely on the natural groupings within the image, i.e. the spectral properties of the surface, and as such return classes containing spectrally similar pixels. The resulting satellite spectral classification was compared with the aerial land cover classification in order to assess the fine-scale land cover variability within each satellite pixel and k-means class. Classification of was performed with 15 clusters and 15 iterations. The 15 clusters were reduced to 9 groups by merging clusters with little pixel count into spectrally adjacent classes. k-means classification of Landsat pixels is determined by the proportions of open water and dry tundra within each pixel. Class 1 is a water class. Classes 2-9 are characterized by a gradual decrease in open water and an increase in dry tundra. […]
