search

AlexeyPechnikov / pygmtsar

PyGMTSAR (Python InSAR): Powerful and Accessible Satellite Interferometry
http://insar.dev/
BSD 3-Clause "New" or "Revised" License
421 stars 92 forks source link

[Help]: SBAS results different from GMTSAR SBAS processing results #148

Open kinreich opened 2 months ago

kinreich commented 2 months ago

Hello, I ran a time series processing once using GMTSAR manual, and once using pygmtsar, and got very different results for

OS and software version

Steps to reproduce the problem I ran the timeseries on 72 ascending scenes, and 107 interforgrams, pairs : time was 30 days between them, as well as yearly pairs. meters: up to 150 meters (default) I used the same data in both of the processing analyses: same dem file, same wavelength filter, same corr file for correlation filter.

For GMTSAR they produce a velocity grid to look at, and for pygmtsar I produced a velocity grid from the total displacement divided by the time passed (in years).

I compared between the processing stages, and found that the differences in grd files are seen only after running SBAS (unwrap and phasefilt files see no difference)

Screenshot Here is the velocity grd (mm/year) for pygmtsar analysis: pygmtsar_vell_file

Here is the velocity grd (mm/grd) for GMTSAR analysis: gmtsar_vel_ll

Could you help explain why we would see such a difference in final results with the two analyses?

AlexeyPechnikov commented 2 months ago

But what is the difference? You are only showing PyGMTSAR results without selecting a colormap or specifying a range. By the way, you are using a very old version of PyGMTSAR.

kinreich commented 2 months ago

I ran the analysis on pygmtsar a while ago, that's why its the older version. I showed the results of both pygmtsar and GMTSAR, there are two png's of grids attatched. You can see the great difference in the mm/year, although the trends are the same.

kinreich commented 2 months ago

Here I have attached another png that shows the difference in total displacement over 7 years: Cropped Grid 1 is from GMTSAR aanalysis Resampled Grid 2 is from pygmtsar Analysis And you can see the difference grid as well (in mm) final_displacement_20221204

AlexeyPechnikov commented 2 months ago

It seems you have not detrended your cloudy PyGMTSAR unwrapped interferograms before SBAS processing, but used GMTSAR atmospheric correction.

kinreich commented 2 months ago

I detrended PyGMTSAR unwrapped interferograms before SBAS processing, and didnt use atmospheric correction with GMTSAR

AlexeyPechnikov commented 2 months ago

Okay, show us your phase cleaned from stratified atmospheric phase delay and turbulence component, and show your displacement grids per date.