In geography and landscape ecology, machine learning techniques are applied for identifying spatio-temporal patterns and changes at different scales. This includes, for example, the interpretation of satellite images of the land surface. Furthermore, deep learning algorithms are used to identify species (e.g. bats or pollen) in large image data sets and to automatize the image segmentation for studies on wood anatomy and plant roots. Machine learning is also used to study the complex effect of climate on tree growth.

Involved Research Groups

Dr. Mario Trouillier

Landschaftsökologie und Ökosystemdynamik
Institut für Botanik und Landschaftsökologie
Universität Greifswald

Prof. Dr. Sebastian van der Linden

Institut für Geographie und Geologie
Universität Greifswald
AG Fernerkundung und Geoinformationsverarbeitung


In the project “Analysis of peatland degradation using long and dense time series of multispectral satellite data”, the focus is on a multidimensional view of damaged and rewetted peatland areas. Various satellite data with different temporal, spatial and spectral resolutions are used for the quantitative description of current and recent states of vegetation cover, land use and the hydrological situation or plant stress.