LARGE ALONG-TRACK BASELINE SAR-GMTI: FIRST RESULTS WITH THE TERRASAR-X/TANDEM-X SATELLITE CONSTELLATION Stefan V. Baumgartner, Gerhard Krieger Microwaves and Radar Institute, German Aerospace Center (DLR) Muenchner Strasse 20, 82234 Wessling, GERMANY, Email: stefan.baumgartner@dlr.de ABSTRACT In the paper first ground moving target indication (GMTI) and parameter estimation results obtained with the spaceborne TerraSAR-X/TanDEM-X satellite constellation are presented and discussed. For processing a dual-platform GMTI algo- rithm developed by the authors was used. This algorithm enables the estimation of the true geographical positions, the velocities and the headings of the detected targets with high accuracy. The algorithm is verified and evaluated using ground truth reference data. Index Terms— Synthetic aperture radar, pulse Doppler radar, radar signal processing, road vehicle location 1. INTRODUCTION Moving targets appear displaced from their actual positions in conventionally processed SAR images (cf. Fig. 1 top left) [1]. Therefore, the main objectives of almost any GMTI algorithm are the detection of the moving targets and the estimation of their true geographical positions. Furthermore, the velocities and headings of the targets are of interest and also should be estimated. However, parameter estimation is quite challeng- ing, especially for GMTI algorithms which have to handle SAR data acquired with spaceborne sensors. For a ’spaceborne’ GMTI algorithm one of the most criti- cal parameters to estimate is the moving target’s broadside po- sition or along-track displacement, respectively, which is di- rectly related to the target’s true geographical position. Hav- ing for instance only one single X-band satellite with two re- ceiving antennas (e.g. one TerraSAR-X satellite) and comput- ing the target’s true position by exploiting the noisy and clut- ter disturbed along-track interferometric (ATI) phase, large position errors in the order of several hundreds of meters may occur [2]. Without incorporating a priori knowledge, e.g. the knowledge about the positions of the road axes in the SAR image, reliable parameter estimation often is not possible. In [3] we have proposed a novel dual-platform SAR- GMTI algorithm which is not based on a priori knowledge. Since this algorithm does not rely on a road database, also vehicles moving on open land and on open water can be Fig. 1. Moving target displacements in the SAR images ac- quired with the first (top left) and second platform (top right). The displacement difference is shown at the bottom. monitored. Theoretical analyses and simulations have pre- dicted a high parameter estimation accuracy. For instance, the geographical position estimation error is only in the order of several meters instead of hundreds of meters using only a single SAR platform. However, since no real data were avail- able when the GMTI algorithm was proposed in 2007, up to now it was not possible to verify the performance predictions. Since June 2010 the TerraSAR-X/TanDEM-X satellite constellation is in orbit [4]. During the early commissioning phase the along-track baseline between both satellites was in the order of 20 km, corresponding to a time lag Δt b of ap- proximately 2.5 seconds. This is just the time lag promising the best performance of the proposed GMTI algorithm [3]. Therefore, during the commissioning phase several GMTI data takes over different test sites have been acquired with the satellite constellation in pursuit monostatic mode, with the