Landsat imagery of Estonia.

Abstract

Landsat Thematic Mapper (TM) imagery was used to analyze the sediment and geomorphic features in southwestern Estonia. Landsat 4 and 5 image sets from June and July of 1986 and May of 1988 were used for the research. The objective of the research was to find the best way of processing and analyzing the Landsat images to use in the study of Quaternary sediments and geomorphic landforms. The sedimentary and geomorphologic features were interpreted from the scenes, using both unsupervised and supervised methods of classification. During the unsupervised classification Idrisi ISOCLUST and RECLASS modules were used. In supervised classification the author visually selected pixels or clusters, which represented a feature. Classified images were produced using the RECLASS module. The classified images were multiplied by the normalized difference vegetation index (NDVI) to recognize the vegetation type, which coincided with a sedimentary or geomorphic feature. In southwestern Estonia different types of land use coincide with the specific type of sediment or geomorphic form of terrain. Using the classified masks, different types of land use were extracted, and the attempt to link different land-use types with corresponding sediments and geomorphic features was made. During the research peat bogs appeared to be the type of sediment, which can be most easily extracted from the background and directly mapped using the Landsat TM scenes. Peat mining activity was also possible to be extracted from Tasscap moisture index scenes. Till and glaciofluvial sediments are possible to be identified and separated based on the differences of land-use types in southwestern Estonia. Areas covered by till sediment were extracted using only unsupervised methods of classification. The till sediment coincides with the specific forest-cluster, spruce, which has different spectral signatures compared to surrounding types of forests clusters, and other types of land use. Geomorphic forms including dunes, end moraines, and in many cases eskers, especially marginal eskers, gave positive results during attempts to recognize and to separate them from the background. Dunes, eskers and in some cases drumlins were possible to be extracted from the background using both supervised and unsupervised methods of classification. Drumlins, eskers and dunes are linear features, which have different types of vegetation from surrounding areas, and their spectral signatures are easy to separate from the background in many cases. End moraines are possible to be traced and located on the Landsat TM image as borders between different types of land use.