| Abstract [eng] |
In the science of geography, land cover is considered an important object of study in some respects, such as environmental protection, landscape morphology and functioning, economic and social. It is constantly changing and is influenced by economic, political, cultural and other societal factors. Long-term observations of land cover can assess the factors that affect the landscape, assess their extent, as well as predict trends in landscape change, identify patterns of development, prevent destructive processes, assess natural or man-made damage to the landscape, so long-term observations of land cover changes are relevant topic, but at the same time a problem to do so with as few resources as possible and get the most efficient process as possible. Monitoring of land cover changes is becoming easier, because of various open source satellite images for remote sensing. The aim of this master 's thesis is to develop a methodology for the identification of land cover changes using the fusion of radar and multispectral satellite images. To achieve this goal, four tasks have been set: to review the land cover classification methodologies, to apply them to the territory of Lithuania and to distinguish the land cover classes used in the research; to review the methodologies for the identification of ground cover changes and digital mapping using radar and multispectral satellite images, to prepare a common methodology that will be applied in the research of the work; to perform ground cover classification using radar and multispectral satellite image synthesis based on the prepared methodology; perform digital mapping of land cover changes based on the prepared methodology. For the research used Sentinel-1 satellite synthetic aperture radar and Sentinel-2 satellite multispectral satellite images, prepared a methodology for the fusion of these data, land classification and identification of its changes. The results of the study showed that using the fusion of the above data, it is possible to identify land cover changes, but they need to be further verified, so a qualitative accuracy check of the identified land cover changes was performed. The inspection revealed that the changes incorrectly identified (92.08% of all the inspected changes) during the inspection were false positive results and no false negative results were observed in the analysis of the images. Although changes are incorrectly identified in certain identified cases, their visual review (especially when potential locations for potential inaccuracies are known) and manual correction would still use less time than not automating the process. |
