Spatial phenomena like the expansion of artificial land and the decrease of agricultural land lead to significant change rates over time for the main land cover types. For European policies, comparable information on land cover change in all European countries is required. The Statistical Office of the European Union (Eurostat) uses the nomenclature of the Land Use/Cover Area Frame Survey (LUCAS) as a basis for compiling areal statistics across the entire EUs territory. As there is presently no dataset in Germany which can be used as a stand-alone source to fulfill Eurostat´s requirements, the project Cop4Stat_2015plus was initiated. The aim is to assess the feasibility of providing the needed land cover information by using remote sensing techniques and satellite data from the Copernicus program and contributing missions. In this study, a method for classification of high-resolution RapidEye time series images for the year 2015 in a study area in Germany is presented. Machine-learning algorithms in combination with topographic reference data as training and validation datasets are used for an object-based classification of artificial land, cropland, woodland, shrubland, grassland and water bodies. For a better separation between shrubland and woodland a normalized digital surface model is used. Classification results show an overall accuracy of above 95 %. The accuracies for the classes range between 76 % for shrubland and 98 % for water bodies. The information of the near infrared band, the red edge band and the height show the highest relevance for good classification results. At the edge of forests, shrublands are partly misclassified instead of forest or cropland because the height of trees and the height of the neighboring objects are averaged and the outcome then corresponds to the height of shrubland. The results will subsequently be compared with official areal statistics and will be tested to support the provision of areal statistics for national and European purposes.