The GUF has been generated by means of the dedicated Urban Footprint Processor (UFP) implemented at the German Remote Sensing Data Center (DFD) of the German Aerospace Center (DLR).
Input data are single look slant range complex (SSC) TerraSAR-X and TanDEM-X amplitude images acquired in Stripmap mode at 3m ground resolution. The processing consists of three basic steps. First, a texture feature (speckle divergence) is extracted from alloriginal amplitude images in order to highlight areas characterized by highly diverse and heterogeneous backscattering – a typical characteristic of built-up areas in radar imagery that results from the double bounce effects from buildings and other vertical structures in combination with extensive shadow regions. Secondly, a fully-automated classification procedure derives a binary settlement layer for each scene based on both the corresponding amplitude and texture images. Thereby, pixels exhibiting high values in the texture and the amplitude images are defined as built-up areas, whereas all remaining regions are assigned to the class non-built-up area. The third and final step is the mosaicking and post-processing of the data, supported by a semi-automated quality assessment. More details on the GUF methodology and UFP technique are provided by the publications listed in the References section of the GUF webpage.