Digital land cover mapping of the East Fennoscandian Arctic zone: case study of the south-eastern Kola Peninsula

The conducted research has shown that interpretation of medium-resolution multispectral images of the Landsat satellite program by supervised classification methods made it possible to form a spatial model of the land cover of the southeastern part of the Kola Peninsula with acceptable accuracy. Based on the assessment of the thematic and positional aspects of the reliability of interpretation, the researchers have concluded that such a supervised classification algorithm as the “minimum distance” method showed good reliability of the interpretation results, especially with respect to woody vegetation. Using this algorithm for an area of ​​more than 4 million hectares, a reliable spatial model of the land/vegetation cover has been developed, which is informative in a wide range of image scales. Analysis of spatial information has revealed that typical northern taiga coniferous and deciduous stands are represented to the southwest of the northern boundary of the sparse island distribution of permafrost. To the east of this boundary, woody vegetation exists in the form of separate massifs of pre-tundra sparse forests and along watercourses. In the area of ​​massive-island distribution of permafrost, there is practically no woody vegetation, or it is found only in the valleys of large rivers.