[Help]: Conversion of uwrap phase to Elevation

[sreeram-radhakrishnan]

Hello, I want to convert unwrap phase to elevation. How to work on that in GMTSAR? Is there any scripts or command for working on it? Thank you.

Regards, Sree Ram R.

[AlexeyPechnikov]

See the function “los_displacement_mm” in PyGMTSAR Python InSAR: https://github.com/mobigroup/gmtsar/blob/pygmtsar/pygmtsar/pygmtsar/SBAS_incidence.py

[sreeram-radhakrishnan]

@mobigroup Thanks for the help. Is there any way I can derive DEM from the time series generated by SBAS method in GMTSAR? Please let me know.

As another question, should I use snaphu_interp.csh for the DEM generation which takes all pixels or normal snaphu.csh is okay for DEM generation? Thank you.

Regards, Sree Ram R.

[AlexeyPechnikov]

Sentinel-1 DEM generation is tricky while it’s still possible sometimes. See this document: https://step.esa.int/docs/tutorials/S1TBX%20DEM%20generation%20with%20Sentinel-1%20IW%20Tutorial.pdf Pay attention to the lots of cautions!

[Xiaohua-Eric-Xu]

Generally speaking Sentinel-1 is designed to do interferometry, not to generate DEM, because the baselines are too small. If you want, you may want to use bperp in GMTSAR to compute perpendicular baselines, and then stack lots of interferograms without topography correction (i.e., dem.grd filled with 0s). And then use the known formula to calculate DEM.

[AlexeyPechnikov]

@Xiaohua-Eric-Xu I suppose it would be valuable to have some example for DEM residuals estimation (like building heights). As I remember, this question raises regularly.

[Xiaohua-Eric-Xu]

That's in the current sbas program, where a dem_err.grd will be produced, but I never tried to just put in non-topographic-corrected interferograms in and try to build a DEM out of it. Also it uses a constant baseline which maynot be a good assumption, though one may correct this with a look angle grid later on. There is an example in our short course from the ERS era, but I honestly don't know who produced that. @dsandwell

[Xiaohua-Eric-Xu]

Baseline time plot here BTW.

[AlexeyPechnikov]

Hmm, if I remember it correct, GMTSAR SBAS just returns phase dispersion in percents as the DEM error estimation. There is no way to take the DEM correction in meters. Is it right?

[Xiaohua-Eric-Xu]

It is based on this paper https://www.sciencedirect.com/science/article/pii/S0034425716303510, where beta is computed as the following line in sbas.c, and when generating greens matrix it is multiplied with bperp (perpendicular baseline in meters) in sbas_utils.c.

186     scale = 4.0 * M_PI / wl / rng / sin(theta / 180.0 * M_PI);

So it should come in meters.

[AlexeyPechnikov]

That’s great, I was sure here is a simplified estimation only! Maybe do you have some results to verify?

[Xiaohua-Eric-Xu]

Not at the moment. One way is to assume 0s for dem.grd as input, and compare the output with SRTM.

[AlexeyPechnikov]

Maybe do you have an example for verification? It requires a lot of additional computations so we need to be sure that it’s valueable.

[sreeram-radhakrishnan]

Not at the moment. One way is to assume 0s for dem.grd as input, and compare the output with SRTM.

MAybe I will try to do this procedure, I can change dem.grd to 0s. could you tell me what should be done? I will do the estimation for validation. Thank you.

[Xiaohua-Eric-Xu]

I think you'll need to, produce a stack of interferograms with 0 dems, calculate corresponding baseline information with bperp, and then use the formula to estimate DEM.