This dataset features a map of building fractions (as opposed to built-up fractions including other impervious surfaces such as roads) for Germany on a 10m grid based on Sentinel-1A/B and Sentinel-2A/B time series. The data were created by using machine learning regression and spectral unmixing, using synthetically mixed training data. The dataset is completely based on freely accessible satellite imagery, and was validated with freely available building footprint reference data for three federal states. We recommend to use data at an aggregated resolution of 20m, 50m, or 100m, and to clip data at about 20% building fraction when using 10m resolution maps (or roughly the corresponding RMSE at any other resolution). Temporal extent Used Sentinel-2 data were acquired in 2018, and Sentinel-1 data were acquired in 2017 (see publication). The map is, thus, representative for 2017/2018. Validation results can be affected by building footprint reference data from different years. Data format The data come in tiles of 30x30km (see shapefile). The projection is EPSG:3035. The images are compressed GeoTiff files (*.tif). There is a mosaic in GDAL Virtual format (*.vrt), which can readily be opened in most Geographic Information Systems. Building fraction values are in percent, from 0 to 100. In the original dataset with 10m spatial resolution, fraction values are equivalent to area in m². In the aggregated dataset with 100m spatial resolution, the values must be multiplied with 100 in order to see area in m². Further information For further information, please see the publication or contact Franz Schug (franz.schug@geo.hu-berlin.de). A web-visualization of this dataset is available here. Publication Schug, F.; Frantz, D.; Okujeni, A.; Hostert, P. (2022). Sub-pixel building area mapping based on synthetic training data and regression-based unmixing using Sentinel-1 and -2 data. Remote Sensing Letters. DOI: 10.1080/2150704X.2022.2088253 Acknowledgements The dataset was generated by FORCE v. 3.6.1 (paper, code), which is freely available software under the terms of the GNU General Public License v. >= 3. Sentinel imagery were obtained from the European Space Agency and the European Commission. Sentinel-1 data were provided by EODC. We thank the providers of the building footprint reference data (see publication). Funding This dataset was produced with funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (MAT_STOCKS, grant agreement No 741950).