The Potential of Sentinel-2 Data for Global Building Footprint Mapping with High Temporal Resolution

Jonathan Prexl, Michael Schmitt

Abstract: The extent of human settlements is one of the most frequently captured types of information throughout many map products. Still, in many parts of the world, the information on how many buildings are present at a certain location and subsequent information such as population density or the degree of sealed surfaces is hardly known. We investigate a simple procedure to produce maps of building footprints based on the freely available optical data of the satellite mission Sentinel-2. We show that this approach is capable of mapping buildings with reasonable accuracy and discuss in detail the effects of ground sampling distance and the intrinsic georeferencing error on those maps. Given the high temporal frequency of the data acquisition, we argue that our approach is useful and capable of investigating the temporal development of fast-growing urban areas.