FORUM FORUM
Categories » SAR Discussions » Interferometry » Interferogram for IEEE Data Fusion Contest TSX samples?
Combination View Flat View Tree View
Threads [ Previous | Next ]
Interferogram for IEEE Data Fusion Contest TSX samples?
toggle
Interferogram for IEEE Data Fusion Contest TSX samples?
coherence insar tsx error messages
1/24/12 9:10 AM
Dear NEST,

I have installed NEST 4B-1.1 (64 bit LINUX) to try out some InSAR processing
on the TSX samples provided for the IEEE Data Fusion Contest 2012.

See: http://www.digitalglobe.com/data-fusion-contest for details.

I managed to create a co-registered stack for the Oct13 and Oct24 2011 images,
but when I try to take the next step, either interferogram generation or coherence
image calculation results in the non-nonsensical error message "2" (with a "STOP"
traffic sign, in which "STOP" is replaced by "!"). There is no related output to the
console, so I have no idea what this means.

I saw from earlier posts that InSAR on TSX (SPOTLIGHT) is not available/stable
until 4C. Still, something more informative than "2" would help,

With greetings

GGL
RE: Interferogram for IEEE Data Fusion Contest TSX samples?
Answer Answer (Unmark)
1/24/12 9:12 AM as a reply to Guido Lemoine.
so, do we share the prize? ;) no spot-interferometry is currently not supported, and it will also not be supported in 4C. (if you want it, make a post & argument in a wishlist part of the forum). \p
RE: Interferogram for IEEE Data Fusion Contest TSX samples?
1/24/12 9:23 AM as a reply to Petar Marinkovic.
Ok, thx, even though that is sad news.

As a consolation, can we expect the "2" message to become self-explanatory?

GGL
RE: Interferogram for IEEE Data Fusion Contest TSX samples?
1/24/12 10:19 AM as a reply to Guido Lemoine.
NEST is an open source project: feel free to contribute. ;)

UPDATE#1:

Regarding the very self-explanatory "2" error message is an internal error code of operators. In principle if the stack you use for input for interferometry is properly coregistered no user will be exposed this error - in an optimal situation.

I think in this case the bigger issue is that you were allowed to do coreigstration on TSX spot. NEST coregistration modules are not optimized for "time varying doppler" and you were computing spectral rubbish. Still, you might get lucky with existing modules, if the variability of doppler between master/slave crops is similar, and manage to get some coherence. But, results would anyways sub-optimal for any serious InSAR application.
RE: Interferogram for IEEE Data Fusion Contest TSX samples?
1/24/12 10:48 AM as a reply to Petar Marinkovic.
Ok, I understand. In fact, the coregistering for stack production only worked
well for the 2 images I mentioned, but not the one of 2 Oct in the same series.
I was impressed by the (purely visual) quality of the co-registration, but did
not do any spectral analysis (so might well be rubbish).

Just for info, are other (non-interferometric) routines in NEST supposed to
work OK for TSX SPOT data (time varying doppler) or would, e.g. geocoding
not work?

Thanks for the fast replies, btw.

GGL
RE: Interferogram for IEEE Data Fusion Contest TSX samples?
1/24/12 11:28 AM as a reply to Guido Lemoine.
(I will let non-InSAR guys give a more detail answer, and/or check if its needed, but just quickly now):

For non-InSAR applications you can always convert Level1 product from Complex -> Detected and work from there. There is an operator that does all the work for you: Utilities/Dataset Conversion/Complex to Detected GR. This operator will calibrate, satellite-to-ground range, and multilook your product in one go - refer to the manual for specific info. Btw, I haven't tested this for this specific stack, but it should work.

On this _new_ product you should be able to execute any non-InSAR operator in a conventional way. Eg coregistration will then work on amplitude signal, and problems with de-ramping image spectra for warping/resampling needed in complex-coregistration are then avoided.

Note: make sure that for visualization your units are ok (linear vs db), since you can end up with "black band" as output ;)

UPDATE: while typing this, in background I did a test of the procedure I listed - complex -> detected. Input is IEEE Contest data, and master/slave are ones you listed as "good" ones. Indeed spectacular signal! I hope I am not breaking any EULA clause with attaching this screenshot emoticon
Attachment

Attachments: SPOT_TSX_Detected_Stack_SanFran.png (733.1k)
NEST Forum Participation NEST Forum Participation

We encourage you to  join the NEST community and participate in the forum. The forum is maintained by the NEST project team who will most likely answer your technical questions related to NEST if not done so by other community members. Before posting on the Forum please have a look at the User manual, FAQ and previous posts.

Chat
Chat is temporarily unavailable.