摘要：Supervised classification of digital Landsat satellite images was used to locate seabird nesting habitats in the Russian High Arctic archipelago of Franz Josef Land, a region where the avifauna is poorly known and ecologically vulnerable. Major seabird nesting colonies are readily identifiable in Landsat Thematic Mapper (TM) imagery of the region due primarily to the distinctive spectral signature of vegetation on ornithogenically altered soils below bird cliffs. Supervised image classification was used to pinpoint areas displaying spectral characteristics typical of documented seabird nesting habitats. A total of 101 seabird nesting colony locations identified in Russian and Western literature from 1898 to 1996 was used as training sites to develop spectral signatures from a summer TM image mosaic for use in a supervised maximum likelihood classification. The classified image was thresholded and compared to a map of documented nesting locations. Of the 101 field-documented nesting sites, 96 were clearly identified in the classified image. An inventory was produced of all undocumented seabird habitats suggested by the classification, totalling over 300 sites. The methodology used may be applicable to other arctic regions and is intended as a first step when planning ecological protection zones in remote and inaccessible arctic regions.
