th1.l09 - investigations on tops interferometry with terrasar-x
TRANSCRIPT
[email protected] and Radar Institute
Investigations on TOPS Interferometry Investigations on TOPS Interferometry with TerraSAR-Xwith TerraSAR-XPau Prats, Luca Marotti, Steffen Wollstadt, Rolf Scheiber
Microwaves and Radar Institute (HR)
German Aerospace Center
July 29th, 2010Slide 2Microwaves and Radar Institute
TOPS is a wide-swath mode that achieves an azimuth-invariant DTAR and SNR, i.e., no scalloping.
The antenna is steered from backwards to forwards, so that every target on ground is observed under the same antenna pattern.
The TOPS mode will be the default mode for the IWS (250 km) and EWS (400 km) for the Sentinel-1 satellites.
The TOPS mode was first demonstrated in-orbit by TerraSAR-X.
Terrain Observation by Progressive Scans (TOPS)
F. De Zan and A. Monti Guarnieri , “TOPSAR: Terrain Observation by Progressive Scans,” IEEE Trans. On Geoscience and Remote Sensing, vol. 44, no. 9, Sept. 2006.
July 29th, 2010Slide 3Microwaves and Radar Institute
TOPS Time-Frequency Diagram
P. Prats, R. Scheiber, J. Mittermayer, A. Meta and A. Moreira, “Processing of Sliding Spotlight and TOPS SAR Data Using Baseband Azimuth Scaling,” IEEE Trans. On Geoscience and Remote Sensing, vol. 48, no. 2, Feb. 2010.
July 29th, 2010Slide 4Microwaves and Radar Institute
*M. Bara, R. Scheiber, A. Broquetas and A. Moreira, “Interferometric SAR Signal Analysis in the Presence of Squint,” IEEE Trans. On Geoscience and Remote Sensing, vol. 38, no. 5, Sep. 2000.
SAR Impulse Response in the Presence of Squint*
July 29th, 2010Slide 5Microwaves and Radar Institute
Due to the large Doppler centroid variations within a burst in the TOPS modes, small coregistration errors can introduce severe azimuth phase ramps, e.g. 0.1 samples with TerraSAR-X introduces a ramp of 1.6 within a burst (5.4 kHz Doppler variation). Therefore, an error smaller than 0.001 samples is necessary in the TerraSAR-X case to have an error smaller than 3º.
TOPS Interferometry: Problem Statement
July 29th, 2010Slide 6Microwaves and Radar Institute
When using the orbit and an external DEM, very good relative accuracy can be obtained in the estimation of the range and azimuth coregistration errors. However, constant azimuth and range offsets still need to be estimated.
This offset comes principally from the orbit accuracy of the satellite. In TerraSAR-X the orbit position is measured with an accuracy of 5 cm (1 ) ~ 0.025 azimuth samples in stripmap, which scales to only ~0.005 azimuth samples in TOPS (which introduces a phase ramp along the burst of ~15º).
Spectral diversity* (SD) has been shown to achieve the Cramér-Rao bound in the estimation of the coregistration offsets. Two approaches have been investigated based on SD.
*R. Scheiber and A. Moreira, “Coregistration of Interferometric SAR Images Using Spectral Diversity,” IEEE Trans. On Geoscience and Remote Sensing, vol. 38, no. 5, Sep. 2000.
Error: 0.05 pixelsError: 0.05 pixelsTOPS Interferometry: Problem Statement
July 29th, 2010Slide 7Microwaves and Radar Institute
*R. Scheiber and A. Moreira, “Coregistration of Interferometric SAR Images Using Spectral Diversity,” IEEE Trans. On Geoscience and Remote Sensing, vol. 38, no. 5, Sep. 2000.
Look separation
Coregistration error
Spectral Diversity
July 29th, 2010Slide 8Microwaves and Radar Institute
Two approaches are suggested:
The use of SD within one burst (conventional SD). The spectral separation between looks is given by the processed bandwidth.
The use of SD on the overlap region. The spectral separation is in this case given by the steering of the antenna.
ta
fa
Spectral Diversity: Approach with TOPS
July 29th, 2010Slide 9Microwaves and Radar Institute
Spectral Diversity: Performance
July 29th, 2010Slide 10Microwaves and Radar Institute
R. Bamler, M. Eineder, “Accuracy of Differential Shift Estimation by Correlation and Split-Bandwidth Interferometry for Wideband and Delta-k SAR Systems,” IEEE Geoscience and Remote Sensing Letters, vol. 2, no. 2, April 2005.
Conventional SDConventional SD SD with Overlap RegionSD with Overlap Region
max
288
[pixels] 1.11
f Hz
t
max
5386
[pixels] 0.059ovlf Hz
t
CRB
Spectral Diversity: Performance
July 29th, 2010Slide 11Microwaves and Radar Institute
Not CorrectedNot Corrected CorrectedCorrected
Results with Real Data: Mexico City (Descending)
July 29th, 2010Slide 12Microwaves and Radar Instituteaz
imut
h
range
22 days repeat-pass (20.09.2009 – 12.10.2009)
Results with Real Data: Mexico City (Descending)
July 29th, 2010Slide 13Microwaves and Radar Institute
5 months repeat-pass (20.09.2009 – 21.02.2010)az
imut
h
range
Results with Real Data: Mexico City (Descending)
July 29th, 2010Slide 14Microwaves and Radar Institute
ScanSAR – TOPS Interferogram Comparison
11-day repeat-pass11-day repeat-pass
TOPSTOPS
azim
uth
range
SNR degradation at burst edges
ScanSARScanSAR
azim
uth
range
July 29th, 2010Slide 15Microwaves and Radar Institute
Stripmap - TOPS DEM Comparison
11-day repeat-pass11-day repeat-pass
DEM = 15m
For atm = 5mm DEM = 27m
July 29th, 2010Slide 16Microwaves and Radar Institute
ta
fa
ta
fa
TOPS – Stripmap Interferograms: Rationale
July 29th, 2010Slide 17Microwaves and Radar Institute
11-day repeat-pass11-day repeat-pass
TOPSTOPS StripmapStripmap
TOPS – Stripmap Interferograms
July 29th, 2010Slide 18Microwaves and Radar Institute
22 days 44 days 66 days
88 days 110 days 132 days 154 days
July 29th, 2010Slide 19Microwaves and Radar Institute
20.09.2009 – 30.01.2010
Measured subsidence
July 29th, 2010Slide 20Microwaves and Radar Institute
Interferometric TOPS is very sensitive to small azimuth coregistration errors, which introduce severe azimuth phase ramps.
A geometrical coregistration approach (DEM + orbits) is recommended very good relative accuracy only a constant offset in azimuth needs to be estimated, which comes from the orbit accuracy (~5cm in TSX ~0.005 pixels with TOPS).
Two approaches based on spectral diversity have been investigated in order to estimate this offset, namely
the use of spectral bands within bursts.
the use of the overlapping area between bursts.
Experimental results with TSX data have been used to validate the proposed approaches, as well as a comparison with ScanSAR and stripmap.
TOPS - stripmap interferograms seem feasible under a PS approach.
The presence of an azimuth displacement might require the use of more elaborated approaches.
Conclusion
July 29th, 2010Slide 21Microwaves and Radar Institute
Thank you for your attention!Thank you for your attention!
July 29th, 2010Slide 22Microwaves and Radar Institute
88 days 110 days 132 days 154 days
22 days 44 days 66 days
July 29th, 2010Slide 23Microwaves and Radar Institute
[1] R. Bamler, M. Eineder, “Accuracy of Differential Shift Estimation by Correlation and Split-Bandwidth Interferometry for Wideband and Delta-k SAR Systems,” IEEE Geoscience and Remote Sensing Letters, vol. 2, no. 2, April 2005.
In [1] the performance of the spectral diversity approach is retrieved. It can be shown that the accuracy in the estimation of the differential shift in samples with non-overlapping spectra is
where dt is the image sampling and
where N is the number of averaged samples, is the oversampling factor, B is the processed bandwidth, b is the look bandwidth, and is the coherence.
The Cramér-Rao bound (CRB) in the estimation of the differential shift is met when
Spectral Diversity: Performance
July 29th, 2010Slide 24Microwaves and Radar Institute
For the second approach, it can be shown that the spectral separation in the overlap region is given by
where Tcycle is the cycle time, is the wavelength, veff is the effective velocity, rrot is the rotation range, and r is the nominal range of the target.
Spectral Diversity: Performance
July 29th, 2010Slide 25Microwaves and Radar Institute
ScanSAR – TOPS Interferogram Comparison
11-day repeat-pass11-day repeat-pass
ScanSARScanSAR TOPSTOPS