Co-registration of Sentinel-1

dear all, I am trying to co-register the time series of sentinel-1 data. I get an error that not all images are overlapping with master image. Have you received this error? What is the way to select a new image as master and not SNAP selects the master image?

Are you working with GRD or SLC products?
In case of GRD, the image at the top of the list in the coregistration module is the master.

Are the products acquired in different tracks?

Thanks for your reply. Yes the producst are from 2 different tracks. But I want to coregister all of them to 1 master image. Maybe a mosaic of the whole country as th master image and then co-register all other images from the time series to that one. Do you think it is possible? or there is another way to cover a whole country with coregistered sentinel-1 time series?

If they are from different tracks, you might get problems with the traditional co-registration because the surfaces and objects are illuminated from different angles and directions.

To still bring them all into one large stack with optimized accuracy, you can terrain correct all rasters fist, then apply the coregistration module, but select “product geolocation” as initial offset method instead of “orbit”. This worked for me in most of the cases.

A stack of SLCs of different tracks (for interferometric processing) is not possible.

Dear ABraun,
I use this topic to show my problem with coregistration and I am sure you can help me with that.
I am having a problem with coregistration of a time series of GRD products.
My work flow is:

  1. Subset data
  2. Apply orbit file
  3. Remove Thermal Noise
  4. Calibration
  5. Create stack
  6. Multispeckle
  7. Terrain Correction
  8. Linear to dB

Now, I would like to apply the coregistration step, but the tool asks for two input data and I have only one as a stack.
How can I do that? For the time series analysis, the only “create stack” is enough for it or it is necessary to coregister all the datasets?
Thank you in advance for your help.


If all images were acquired under same conditions (same relative orbit), you can probably use the “Create Stack” operator. However, this might still lead to some smaller misregistrations in your data. Especially when you combine ascending and descending data or images of different orbits, you should use the Coregistration

If you already created a stack, you can no longer apply Coregistration. To get precise image registration, you should apply this instead of “Create Stack” at step 5 of your analysis.

Thank you for your quick response. I knew I could have your help.
They are all “descending”, but I thought that the coregistration before the multispeckle could give some alterations. Am I wrong? Otherwise I will substitute the step 5 with the coregistration.
Thank you again,


what do you mean by “multispeckle”?

Instead to apply single speckle filtering, I would like to use the multi-temporal speckle filtering. I had read on one of your post, the advantage to use that, when you have available more datasets as a time series. Is this approach correct?


sorry, I thought you confused it with multi-looking.

I agree, this form of filtering is more effective whenever you have larger stacks.

You can create an RGB of your stack to see if the quality of registration is sufficient. If you notice some offsets between the different dates, I would recommend going back to step 5 and using Coregistration instead.

OK great, thank you again!
I will update you about my results.
I appreciate a lot your help.