Regarding coregistration of multipass TSX-TDX dataset---urgent and important

Hi, I am trying to coregister PolInSAR multipass TSX-TDX SAR dataset over the same area. Some of the passes are well coregistered whereas many are having lots of problems. Upon visualizing the coregistered dataset n ENVI, there are lots of shifts. I have tried a lot and each coregistration process takes more than one hour.

I am confused a lot. Kindly help me by providing solution for this. I am really tired doing the same thing again and again by varying window size. Kindly extend your help and support

Warm regards,

I would recommend “raster_collocate” instead of the coregistration. For processed S1-products it works better then coregistration, so I would try doing the same for non-native SAR from other sources then Sentinel.

You might be losing some meta-data though, but as for the raster values should be well coregistred by the end.

Please share your result, should it work or not.

@Suprd…Thank you for your guidance. I will try to do with that also. Actually from the original images, I am taking the subset of the study area. Does it create problem when we coregister? Also,after replacing NaN values with 0, I am calibrating the images.


Hello @Iveci sir…May I expect some guidance from you.

@suprd…sir there was still shift in the image.

Sir, there was still shift in the resampled slave image

Since you have tried apparently every option by now even outside the SNAP toolbox.
Did you check the location of the files, at least by bounding box.

Otherwise, you should be considering some geo-correction functions. Like Back-geocoding, ellipsoid correction, (or even try the Terrain Correction, although it will probably won’t output the correct results.)

As there might be some displacement for your error passes, plus your SAR processing steps could provide some insight.

[quote=“Suprd, post:7, topic:984”]
Did you check the location of the files, at least by bounding box.
[/quote] How to see this. Moreover, why for some images, the coregistration is perfect even if the processing step is same for all?

Could this possibly be related/connected to other issues posted on this forum like:

Regarding the many NaN pixels, we found a problem in the calibration of TerraSAR-X which happens only when outputting complex data and only when the DN values are below the noise vector values. This has been fixed and an update will be available soon.

I had no problem coregistering the products with the InSAR optimized coregistration.


complex calibration and Freeman-Durden decomposition


Dear sir,

Thank you for your kind concern. May I know what should be the coarse registration window width and height as when I was performing coregistration, I took different window sizes but I don’t know what is the correct one. Kindly provide necessary guidance.

I would be blessed if you can guide regarding what all values to keep in optimized coregistration window.

I request you to kindly provide the new update of SNAP ASAP as I am in urgent need of that.

Once again thank you very much.

In SNAP 6.04 I’ve used both Coregistration and DEM assisted Coregistrion, to seemingly equivalent usable results coregistering TerrasarX images.

Was, “InSAR optimized coregistration” ever included in SNAP and or is there something else you would suggest using when working with the TDX images products in SNAP?

Thank you,