Quality assessment of first TanDEM-X DEMs for different terrain types

The main product of the TanDEM-X mission is a global interferometric digital elevation model (DEM) that is finally calibrated due to residual systematic offsets and tilts. For the final DEM product single acquisitions (socalled data takes) are calibrated and merged to tiles of a size of 1°x1°. Finally the globe will mostly be covered with two acquisitions. The quality of the calibration of the data takes and also the accuracy of the final DEM highly depends on the terrain and the vegetation. Therefore, three different test areas where we have to deal with good, medium and difficult terrain conditions will be presented here. 1 DEM Calibration and Mosaicking Concept Within the TanDEM-X mission [1] almost the whole globe has already been measured by the two SAR satellites TerraSAR-X and TanDEM-X, so that the generation of an Intermediate global digital elevation model (IDEM), consisting of the first coverage, can start soon. In the meantime the acquisition of the second coverage, which will ensure the required absolute and relative accuracies of 10 m and 2 m respectively, is going on. Although, calibration of the SAR system and baseline errors is conducted, still smaller systematic errors in the order of few meters remain in the single acquisitions (so-called data takes). The DEM calibration ([2], [3]) estimates these residual height errors according to a functional error model which regards offset and tilts in range and azimuth. This is done by a least-squares adjustment using the elevation of tie-points in overlapping regions of neighbouring interferometric DEMs. The tie-point extraction approach is described in [4]. Prerequisite for the adjustment is the acquisition and assessment of ground control points. The height offset to WGS84 is estimated by introducing absolute height reference data like ICESat data [5]. In the subsequent step of mosaicking, the DEMs are corrected by the estimated height errors and merged to tiles of a size of 1°x1°. In this paper first examples of the calibrated and mosaicked Intermediate TanDEM-X DEM product (IDEM) which is based on one coverage will be validated. This will be carried out by comparing the IDEM to height references like SRTM, ICESat and – if available – high resolution reference DEMs and GPS tracks. Furthermore, the absolute differences between neighbouring acquisitions are computed, which are a good indicator for the relative accuracy. In this paper three test sites with different vegetation and terrain types are presented. In Section 3 a first quality assessment of the Intermediate TanDEM-X DEM is done.