Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th,...
Transcript of Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th,...
![Page 1: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/1.jpg)
Closure phases and biases in InSAR products
Francesco De Zan
Contributions by Homa Ansari, Giorgio Gomba, Alessandro Parizzi
Remote Sensing Technology Institute
German Aerospace Center (DLR)
MDIS Workshop
La Petite-Pierre, France
October 16th, 2019
![Page 2: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/2.jpg)
Performance comparison study
Data set:
Sentinel-1 A/B time series : IW mode
acquisition time span : 4 years (Oct. 2014-Sep. 2018)
size of the time series : 184 SLCs
extent of the chosen area ≈ 30000 km²
number of processed bursts : 19
Benchmark: Persistent Scatterer
Interferometry (PSI)
StBas Bandwidth 5
StBas Bandwidth 10
EMI Full Covariance
versus
Bandwidth
N37°
N36°
N38°
N39°
13° 15° 16° 17°
Mt. Etna
![Page 3: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/3.jpg)
StBAS
Bandwidth 5
905 Interferograms
Mount Etna
![Page 4: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/4.jpg)
StBAS
Bandwidth 10
1785 Interferograms
Mount Etna
![Page 5: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/5.jpg)
EMI
Full Covariance
16836 Interferograms
Deformation rate
Bias wrt PS’s [mm/year]
Dispersion wrt PS’s [mm/year]
Band 5 -6.50 2.58
Band 10 -3.05 1.55
Full Stack -0.24 0.70
Mount Etna
![Page 6: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/6.jpg)
Bias for each lag (Mt. Etna dataset)
12 mm / 365 days * 6 days = 0.2 mm = 2.6 deg
Lag−1 ≈ 6 days
Lag−2 ≈ 12 days
…
![Page 7: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/7.jpg)
Interferograms vs. closure phase
𝜙12 𝜙23 𝜙31
3 images 3 interferograms
1 closure phase
[deg]
180
-180 [deg]
𝒄 ≠ 𝒂 + 𝒃
𝒂 𝒃
Φ1,2,3 = atan exp 𝑗 𝜙12 + 𝜙23 + 𝜙31
Mis-closures are possible only with spatial averaging!
F. De Zan, M. Zonno and P. López-Dekker, "Phase Inconsistencies and Multiple Scattering in SAR Interferometry," in IEEE Transactions on Geoscience and Remote Sensing, 2015.
![Page 8: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/8.jpg)
Rain event in Japan (Kumamoto)
~12:18 11Jul-25Jul-8Aug 12:00 25Jul 11:00 25Jul
Weather radar images
(Source: tenki.jp)
Closure phase
[deg]
![Page 9: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/9.jpg)
Mts. Hakone & Fuji (Japan), ALOS-2, 2014-2015
Closure phase +/- 40 deg
![Page 10: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/10.jpg)
Mexico, Sentinel-1, Descending, Closure Phase +/- 30 deg, 2014-2016
![Page 11: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/11.jpg)
One-year deviation between 12-day and 24-day S1
interferograms
• Colorscale: +/- 360 deg => 28 mm/yr
• Far away from 1 mm / year target : necessity of log-span interferometric
measurements
forest
forest
urban
fields
+360 deg
-360 deg
0 deg
![Page 12: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/12.jpg)
Interferometric phases and velocities are biased
The presence of closure phases means that there is a path dependency in the
temporal integration
Presence of systematic closure phases means that
the interferometric phases are biased, at least some of them
velocity estimates are biased
We now know that short term interferograms are the culprit!
𝒄 ≠ 𝒂 + 𝒃
𝒂 𝒃
![Page 13: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/13.jpg)
How to estimate the bias magnitude?
(without doing all the processing)
Average closure phases with short and long arms
Assumption: the long arms have little bias
The asymmetric mis-closure should represents mostly the short-term bias
𝑡1 𝑡2 𝑡3
𝜙12 𝜙23
𝜙31
![Page 14: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/14.jpg)
18 April 2016 2 May 2016 16 May 2016 13 June 2016 11 July 2016
25 July 2016 8 August 2016 5 September 2016 19 September 2016 3 October 2016 31 October 2016
Moisture inversion (Kumamoto, ALOS-2)
F. De Zan and G. Gomba, Vegetation and soil moisture inversion from SAR closure phases: first experiments and results, Remote Sensing of Environment (2018)
![Page 15: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/15.jpg)
We have some validations…
Kumamoto, ALOS-2 CanEX-SM10, UAVSAR, JPL
G. Gomba and F. De Zan, Estimating soil moisture from SAR Interferometry with
Closure Phases, IGARSS 2019 (poster)
![Page 16: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/16.jpg)
• Magnitude: a few centimeters in L-band (10-20 % of wavelength, S. Zwieback)
• Corrections for InSAR: two examples over Kumamoto with our model
Moisture signal in SAR interferograms (L-band)
July 25th – August 8th 2016 July 11th – August 8th 2016
![Page 17: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/17.jpg)
Modeling the velocity bias
The moisture model seems not to describe the bias (wrong sign, more seasonal)
Some scatterer electrically moving away from the satellite at 0.1 mm / day
Biomass growth?
Observed velocity bias Modelled bias
γ = 0.2 + 0.08 ∙ exp 𝑗 ∙ 0.03 ∙ 𝑡 exp −𝑡
20
Modelled coherence
![Page 18: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/18.jpg)
• Assuming four years of Sentinel-1 with 60 acquisition / year
• To reach this performance at large scale (large distances) we need:
• Good instrument / orbits
• Tropospheric corrections (numerical weather models, e.g. ERA5)
• Ionospheric corrections (e.g. from CODE model or split spectrum)
• Accurate processing!
Current theoretical performance
Residual troposphere
Residual ionosphere
Instrument/ geometry
Total Deformation rate
Germany 1.0 cm 1.0 cm 1.5 cm 2.1 cm 1.3 mm/yr
Indonesia 3.0 cm 1.0 cm 1.5 cm 3.5 cm 2.1 mm/yr
![Page 19: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/19.jpg)
Germany
Pakistan
𝟐 𝒄𝒎𝟐
𝟏𝟎 𝒄𝒎𝟐
𝟒 𝒄𝒎𝟐
𝟐𝟎 𝒄𝒎𝟐
Performance of corrections with ECMWF ERA5
w/o corrections with corrections
w/o corrections with corrections
![Page 20: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/20.jpg)
PSI vs. GPS std: 0.86 mm/yr
![Page 21: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/21.jpg)
North and East Anatolian Faults – PSI
![Page 22: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/22.jpg)
A phase product based on the full covariance matrix
It would like to propose a phase product to be provided routinely (for instance by ESA)
Multilooked (100 m – 200 m) => much smaller than SLC’s
Based on full covariance => long-term stable
Including correction layers (troposphere, ionosphere, SET…)
Wrapped
Monti Guarnieri & Tebaldini, On the exploitation of target statistics for SAR interferometry applications, TGaRS (2008)
Ferretti et al., A New Algorithm for Processing Interferometric Data-Stacks: SqueeSAR, TGaRS (2011)
Ansari et al., Efficient phase estimation for interferogram stacks, TGaRS (2018)
Ansari et al., Sequential estimator: Toward efficient InSAR time series analysis, TGaRS (2017)
Full time covariance Sequential scheme Special interferograms from compressed images
Reduced number of regular interferograms
![Page 23: Closure phases and biases in InSAR products...MDIS Workshop La Petite-Pierre, France October 16th, 2019 Performance comparison study Data set: Sentinel-1 A/B time series : IW mode](https://reader033.fdocuments.in/reader033/viewer/2022050209/5f5c38cf626de804016152db/html5/thumbnails/23.jpg)
• Velocity biases for short lags can reach 5-10 mm/yr (or more)
• Moisture related phases
• Compensation for L-band interferograms (1-2 cm)
• Do not seem to explain the velocity biases
• The velocity biases can easily be a performance bottleneck!
• Modeling & compensation
• Use of long-term interferograms, as in Phase Linking or EMI
• Single-look interferometry
• A phase product based on the full covariance matrix
Conclusions and recommendations