During the last decades, the Earth has been changing rapidly. For instance, with the change of its temperature, the glaciers are melting. The number of humans as well as natural disasters have increased considerably. The impact of these events causes short-time, but also long-time damages to the regions, which are affected. In order to show the impact of these events to their surrounding area, multispectral as well as Synthetic Aperture Radar satellite images have been classified by their content. The classification was done in a supervised approach, using a Support Vector Machine with relevance feedback, as well as an alternative unsupervised parameter-free and featurefree approach based on Kolmogorov Complexity. With the supervised approach it is possible to show the effects of certain events on their area, while the unsupervised approach can create a change map in a binary form derived from a satellite image time series. With this change map one can recognize the impact of the event.
Publications: The work from the master thesis has been published in five papers.