Lost coherence when processing large crop (TSX images)

[John L.]

Dear Batu and ADORE-DORIS members,

I am working on a set of 24 TSX images, flat urban area. 

The problem here is that I could generate quite nice interferograms with a small area (from 3000x2000 to 6000x5000). 

When I tried with the area bigger than 10000 pixels or lines, there are something wrong with the coregistration, the interferograms don't show any coherence.

M_timing and Coarsecorr look fine, the estimated translation correction are consistent for most of images. I've tried some parameters for FINE, even with the recommended values here, the same things happen.

Could you give me some suggests about the causes, or even the optimal parameters for TSX images?

Thank you!

John

[Batuhan Osmanoglu]

Hi John, 

Seems like the coarse-corr is failing when you switch to a larger scene. I processed full scene TSX scenes before so it should work. 

I would check to see the difference between coarse (orbits or) correlation offsets for small and large scenes. You should get roughly the same offsets... 

 Also I don't think I ever had to do M_Timing for TSX scenes... Especially if the area is flat, the M_Timing might be causing problems. I would say to skip it... If you have different timing errors calculated for different scenes, that could slide the slave off of master... 

Best,

batu. 

--

Batuhan Osmanoglu, Ph.D.
Partner

BOS Technologies, LLC

http://bostechnologies.com

--
You received this message because you are subscribed to the Google Groups "ADORE-DORIS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to adore-doris...@googlegroups.com.
To post to this group, send email to adore...@googlegroups.com.
Visit this group at http://groups.google.com/group/adore-doris.
For more options, visit https://groups.google.com/d/optout.

[John L.]

Hi Batu,

Thank you for your superfast reply.

Yeah I've processed without m_timing for several times, the same things happened, so I can exclude m_timing as a source of error.

Coarse_correlation_translation(lines and pixels) are roughly the same for small and large crop. I noticed that without m_timing, Coarse_orbits_translation(lines and pixels) are equal to Coarse_correlation_translation, and causes some pixels lay outside overlapped master and slave images, but forget it because it's still okay with small crop, Doris can get rid of those pixels. 

The thing that I want to mention here is that even I use the same master, copy coarse_corr information from small crop to large crop for processing, it still failed. That's why I suppose the error is from Fine coregistration.

Eventually, could you let me know your parameters for coarse-corr and Fine?

Thank you very much!

Bests,

John

[Batuhan Osmanoglu]

Hi John, 

I just checked my TSX run with Van Eq data. 

http://supersites.earthobservations.org/van.php

I used the default settings for coarsecorr and fine steps:

cc_method magfft //

cc_nwin 100 //

cc_winsize 64 64 //

cc_acc 32 8 //

cc_initoff orbit //

fc_method magfft //

fc_nwin 4000 //

fc_winsize 64 64 //

fc_acc 32 32 //

fc_initoff coarsecorr //

fc_osfactor 32 //

fc_plot 0.4 NOBG //

What are the correlation values you are getting for the coarsecorr? If they are not high, you can be sure there is something wrong. The orbits are quite good for TSX so I am puzzled you are having misregistration issues. 

Best, 

batu. 

--

Batuhan Osmanoglu, Ph.D.
Partner

BOS Technologies, LLC

http://bostechnologies.com

[John L.]

Hi Batu,

I'm in processing with your parameters, thank you for these things.

The correlations (you mean the mean coherences?) vary from 0.45 to 0.65. 

On the other hand, when I tried to use the parameters (for fine steps) below, the coherence for large crop come better (I still don't have any idea, they're the suggested parameters for oversampling scenes from StaMPS)

fc_method oversample //

fc_nwin 4000 //

fc_winsize 128 128 //

fc_acc 16 16 //

fc_initoff coarsecorr //

fc_osfactor 16 //

And when I tried to do the PSI processing by StaMPS with those succeeded small crops. The velocity map seems to be divided block by block and I'm 100% sure this is an error. (I've attached it below) Do you think there is any relationship?

Eventually, thank you very much for taking your time to help me! 

All the bests,

John