ionospheric correction with stackSentinel.py

[SofiViotto]

Hello everyone,

I'm trying to create a new stack of interferograms and I would like to apply the ionospheric correction. The options are:
--param_ion ionosphere estimation parameter file. if provided, will do ionosphere estimation.
and
----num_connections_ion

What file shall I provide to perfom the ionospheric correction?

I have tried the ionospheric correction before working with topsApp, but not with stackSentinel.py. Any advice?

Best,

Sofia

[AlejandroOroDiscord]

Hello Sofía! I performed the ionospheric correction with stackSentinel.py providing the ion_param.txt file located in /home/YOURUSERNAME/tools/src/isce2/contrib/stack/topsStack.
I am actually trying to do the opposite, I want to filter the ionosphere using topsApp.py. Could you share you topsApp.xml config file? cause I don't see much improvement with the parameters I use. Thanks in advance!

[SofiViotto]

Hello! Thanks for your reply. At the end, I figured out how to run it.
A question: what is the unit of the output ion_dates? Is it radian?

With topsapp I used everything by default.

"do ionosphere correction" True
"start ionosphere step" subband
"end ionosphere step" esd
"height of ionosphere layer in km" 200.0
"apply polynomial fit before filtering ionosphere phase" True
"maximum window size for filtering ionosphere phase" 200
"minimum window size for filtering ionosphere phase" 100
"maximum window size for filtering ionosphere azimuth shift" 150
"minimum window size for filtering ionosphere azimuth shift" 75
"correct phase error caused by ionosphere azimuth shift" 1

I don't remember if the output interferogram, after the ionospheric correction, is corrected for the ionosphere or (as with the stripmap application) you have to subtract it afterwards. You may want to check that. Another reason why you don't see any improvement is that maybe you don't have a strong ionosphere impact. Normally for c-band the delay is so small that you can forget about it, but there are some cases (like my area of interest in NW Argentina and for ascending tracks) where you can really see some influence on the ionosphere.

I hope it works for you!

Best,
Sofia

[AlejandroOroDiscord]

Hi again Sofi! Thanks for your answer!!!! Then I am pretty sure I have to remove myself the contribution when using both, topsApp.py and stackSentinel.py.
So as I see it, for stackSentinel.py:
filt.ion is the ionospheric contribution.
filt_fine.unw - filt.ion is the unwrapped interferogram corrected by ionosphere.
Is this correct?
Also, is this how your filt.ion files looks like:

[fdelgadodelapuente]

I am seeing that the results are the same with and without ionospheric correction estimated.
For topsApp.py I get this:

And for stackSentinel.py I get this for one of the filt_fine files:

Maybe I need to apply the correction to the results with an additional operation?

That interferogram is not correctly processed because it includes the topographic phase. Check if your DEM is not empty. You need to remove that before being able to observe any of the ionospheric signals.

[AlejandroOroDiscord]

Thanks for your answer Francisco! I tested different DEMs in SNAP. Copernicus 30m and SRTM 1Sec improved the results but none of them remove all the topographic contribution. I guess I should create my own DEM to get a better interferogram or is there any other option? These are my results:
SRTM3Sec:

SRTM1Sec:

GETASSE30m:

COP30m:

[fdelgadodelapuente]

You should use the best available DEM, and those are SRTM 1 arcsec and Copernicus 1 arcsec. The interferograms processed with SRTM 3 arcsec and GETASSE30m have lots of high frequency signals that are topographic residuals.

There is a ramp between the SRTM 1 arcsec and Copernicus 1 arcsec corrected interferograms and I suspect it could be related to missing corrections. The Copernicus DEM is distributed with the elevation with respect to the EGM2008 geoid and it has to be referenced with respect to the WGS84 spheroid in order to be compatible with ISCE.

[AlejandroOroDiscord]

Hello Francisco! You were right, the DEM was empty, the problem was the dem.py program not downloading the data. Now, with the SRTM1Sec I am getting the same results with ISCE2 than with SNAP:

Thanks for the Copernicus DEM explanation! I think I need to find a better them to be able to completely remove the topographic phase contribution.

[fdelgadodelapuente]

Hi Alejandro,

That interferogram looks like an average pair for an area with high relief and lots of phase delays. Hence those signals are not topographic residuals.

Getting back to the point of this thread, the ionospheric signals for an ascending TOPS interferogram in areas of low geomagnetic latitude look like a weird ramp plus phase jumps between the different bursts after applying the ESD correction. This is shown clearly in Figure 5 of Liang et al., IEEE, 2019 . The interferogram that you show does not have those phase jumps.

Francisco

[AlejandroOroDiscord]

Thank you Francisco for improving my interpretation of these interferograms! I read the paper you sent and I find it very interesting. I did notice weird fringing patterns but using more than one swath, in the examples I showed you above only subswath 2 of a Sentinel-1 image was processed and you don't see those features. I wonder if it has to do with using more than 1 sub-swath.

[fdelgadodelapuente]

No, you will observe the ionospheric phase jumps in all the swaths that cover an area with a variation in the TEC.

Another example of Ionospheric corrections for TOPS
https://ieeexplore.ieee.org/document/8114326

[AlejandroOroDiscord]

I see, thanks again for the information Francisco!

[birhan22]

where can I get TEC data?