TechniquesForHighAccuracyRelativeAndAbsoluteLocalizationOfTerraSARXTanDEMXData.ppt
- 1. Techniques for High Accuracy Relative and Absolute Localization of TerraSAR-X / TanDEM-X Data U. Balss , M. Eineder, T. Fritz, H. Breit, and C. Minet German Aerospace Center (DLR), Remote Sensing Technology Institute (IMF)
- 2. Outline Introduction Measurement Method Consideration of Continental Drift Particulars of Bistatic Acquisition Geometry Measurement Results of Absolute Localization Accuracy (TerraSAR-X) Relative Localization Accuracy (TanDEM-X) Conclusions
- 3. Introduction TSX-1 (launched June 2007) and TDX-1 (launched June 2010) constitute first bistatic SAR system in space. One satellite transmits radar pulses. Both satellites receive the echoes. High geometric accuracy requires consideration of signal propagation effects geodynamic effects in geolocation of ground targets. passive channel: Geolocation is additionally complicated by bistatic acquisition geometry
- 4. Measurement Method 6 Corner Reflectors (CR) at Oberpfaffenhofen Test Site Comparison of measured radar positions their expected values Expected values are based on precise GPS measurements of CR positions orbit interpolation of satellites’ positions The following effects have to be taken into account: additional signal delays (caused by electrons in ionosphere and water vapor in atmosphere) solid earth tides continental drift
- 5. Continental Drift and Geodetic Coordinate System s * TSX-1 / TDX-1 orbit in ITRS-2005/08 coordinates * GPS position of CR in tectonic plate fixed system (e.g. ETRS89) misinterpretation of GPS coordinates offset between expected and true CR position (e.g. approx. 60 cm if ETRS89 is taken for ITRF-2005)
- 6. Effect of Continental Drift on Radar Coordinates (e.g. corner reflector CR moved to northeast) ascending orbit (right looking): CR occurs more in late azimuth and far range than expected t 1 t 2 >t 1 expected position of CR true position of CR flight path continental drift azimuth range W S N E height
- 7. Effect of Continental Drift on Radar Coordinates (e.g. corner reflector CR moved to northeast) descending orbit (right looking): CR occurs more in early azimuth and near range than expected t 2 <t 1 t 1 expected position of CR true position of CR flight path continental drift azimuth range W S N E height
- 8. Effect of a Coordinate System Mismatch Wrong: ETRS89 coordinates are misinterpreted as ITRS-2005 Correct: ETRS89 coordinates are transformed to ITRS-2005 system
- 9. Bistatic Acquisition Geometry of TanDEM-X (Passive Imaging Channel) bistatic range depends on signal travel time which again depends on bistatic range
- 10. Computation of Bistatic Closest Approach iteration step ( n =0, 1, 2, …) initialization 2) Coordinate of hyperbola apex results by nested intervals . 1) Recursive computation of bistatic range for given slow time t :
- 11. Bistatic Acquisition Geometry of TanDEM-X (Active Imaging Channel) Satellite moves during signal travel time. Satellite position differs between instants of pulse transmission and echo reception. Thus, even this acquisition geometry is strictly speaking bistatic. Results of monostatic and bistatic computation scheme differ by some tenth of a millimeter.
- 12. Absolute Pixel Localization Accuracy of TSX-1 (Based on TSX-1 Calibration Datatakes 2007/09) L1B products created before 2011-07-15: bandwidth dependent range offset: 100 MHz : -33 cm 150 MHz : -14 cm 300 MHz : +12 cm azimuth offset: +8 cm Meanwhile bandwidth dependency is solved by code change in SAR processor and instrument is recalibrated.
- 13. Absolute Localization Accuracy of TSX-1 (Reprocessed by Actual SAR Processor Version) σ azimuth = 6.3 cm σ range = 3.8 cm The following offsets are subtracted: azimuth offset: +8 cm range offset : -29 cm
- 14. Absolute Localization Accuracy of TSX-1 and TDX-1 (Based on Calibration Datatakes 2010) TDX-1 σ azimuth = 5.5 cm σ range = 3.5 cm TSX-1 σ azimuth = 5.3 cm σ range = 3.5 cm
- 15. Relative Localization Accuracy of Bistatic TanDEM-X Acquisitions different scaling of axes azimuth: -100 … +60 mm range : -10 … +6 mm ! mean value: m azimuth = -18 mm m range = -2.1 mm standard deviation: over all acquisitions: σ azimuth = 40 mm σ range = 4.4 mm within an acquisition: σ azimuth = 16 mm σ range = 1.0 mm
- 16. Conclusions Geolocation of ground targets has to accurately consider signal propagation and geodynamic effects. Taking these effects into account, we reveal a much better absolute pixel localization accuracy of TSX-1 / TDX-1, than previous studies. The increased measurement accuracy also helped us to identify and solve a small systematic effect in SAR processing. TerraSAR-X: Absolute pixel localization accuracy in the order of magnitude of just few centimeters. TanDEM-X: Relative pixel localization accuracy in range direction even at sub-centimeter level.
- 17. Thank you for your attention!