Coregistration issues and low coherence

Hi. We’re experiencing a few issues with coregistration:

  • [standard] coregistration option “Apply Fine for SLCs”: produces blank (all zeros) slave I/Q bands
  • [standard] coregistration option “Apply Fine for SLCs” suboption “coherence based registration” causes io error
  • [standard] coregistration option “show residuals”: there’s nothing in c:\Users\login.snap\var\log\
  • DEM-assisted coregistration option “apply mask” causes interferogram formation to trigger io exception
  • DEM-assisted with xcorr coregistration won’t run and complains “The product ‘filename_Orb_Stack’ already contains a band with the name ‘Intensity_VV_mst_16Dec2005’.”

The only coregistration algorithms that ran for us were:

  • [standard] coregistration disabling option “Apply Fine for SLCs”
  • DEM-assisted coregistration disabling option “apply mask”
    But then coherence is super low and we’re not sure if its genuinely a non-coherent geographic region and spatial/temporal baseline conditions or only a coregistration artifact.

Is anyone able to reproduce the issues above?


PS: “Close all other” could keep open all selected products, not just the last one selected.
PPS: the menus could wait a few milliseconds before redrawing.

1 Like

Yesterday we were using ENVISAT ASAR and Fine Registration produced a blank slave image.
Now using ALOS-1 PALSAR and coregistration worked fine.
We’re seeing fringes! :heart_eyes:
We’re still not sure if coherence is better only because of the longer wavelength (L- vs. C-band).
Temporal baseline are similar, of the order of one month.

InSAR should work fine with ASAR. Did you apply the precise orbits first?
Maybe the threshold values aren’t ok for the data. Have you tried larger windows and many more GCPs?

Could you supply the data products to reproduce the problem?

1 Like

Hi Iveci!

Felipe and I processed the ASAR.

Yes! We applied the precise orbits before processing.
Yesterday, I tried larger windows and many more GCPs… the results are the same!
Follow the link for download of subsets:



Hi all,

Just a query.
Does image coregistration, using cross-correlation, change the pixel values in both master and slave, or in either images?


I’d say pixel values are only changed in the slave image (and only if blinear or bicubic resampling are selected)

Thanks a ton!!

How to put GCP and in what process should I put GCP? can you give me the steps?

Thank you

You don’t need to input GCPs yourself. GCPs are automatically generated points at which the images are tested for statistics that indicate correct registration. If the measure of one point is too low it will be removed from the set of GCPs. In the end, only a couple of remaining GCPs are used for the coregistration.

1 Like

if I do sentinel 1 image coregistration, do I just go through the inteferogram process or have to go through another step for make a good coherence ?

good coherence depends on the temporal and perpendicular baseline to a large degree. So a careful selection of image pairs is decesive. Have a look at this ESA document, it gives hints on how to get a good image pair. (chapter 1.3 Selecting images for InSAR DEM generation)

can you give me information to know the good perpendicular when download the image? because when I use the perpendicular results on the website and in different snap software.

sorry, I don’t understand.
The ASF baseline tool is currently the best option to determine the perpendicular baseline between two images.

Hi @ABraun,

Any clue if this kind of tool or something similar to calculate perp baseline will be implemented by ESA on a specific website, SNAP, Copernicus Hub, etc.?


I don’t know, sorry

1 Like

Yes there are plans to add a similar tool to SNAP, but not in the near future.

1 Like

Good to know, it will be a nice feature :grinning:

I generated a interferogram with sentinel 1 images, but the coherence is low and now I have to download two other images and that’s very hard.
Is there a way to get the coherence before downloading the images?
The coherence of the images was initially 0.79 but now the results are as follows

Is this acceptable !!!

no, it has to be computed based on the information in the images. The estimated coherence in the InSAR Stack Overview does not include different types of surfaces or the atmospheric conditions during the image acquisitions, so the only way to find out is to test different images’s interferograms and coherence.

There are some light patterns in your interferogram, but probably not enough to get areliable result over your image. Please have a look at the InSAR tutorials how interferograms should look before you can proceed.

The my study area in the image above is on the right side of the image that fringes are more visible.Can I subset it after debursting? And no problem to download If the images dont overlap perfectly ? Like the image below