TerraSAR HH & VV Pol, what can I do with them?

Hello Step forum,

I have received 2 TerraSAR-X Staring Spotlight images taken in an interval of 11 days and in MGD format. One is HH, the other VV, both are the same location.

Are there any processing steps I can perform to improve image quality outside Single Product Speckle Filtering?

You could, after coregistration, apply a multi-temporal speckle filter. But it will probably not be very effective with only two images (which are even from two different polarizations).

What I like to do with my TSX ST data is calculating an average image. You will see quality increase with every image you add.

If the two scenes are acquired from the same track (as probably is the case), multitemporal speckle filtering should work great.

Coregistration is possible with ground projected data? I will attempt this now.

Do I need to take any steps before coregistration?

I believe both images are the same relative orbit, as indicated in metadata.

you can also coregister the data in slant geometry, it should actually be even exacter.

I coregistered both images with SNAP default options. The resulting stack is 2 bands, one exactly the same as the master (hh) and the other (VV) is all black (0 value). Am I doing something wrong?

you can increase the windows size of the cross correlation and the number of GCPs. Also check the box for the residual report so you can see how many GCPs were left at the end.

Please make sure that they have almost the same incidence angle and orbit/path direction before you combine them. Consult the xml header file.

They have the same incidence angle and relative orbit. Thanks.

I increased GCP’s to 10.000, still the same problem (slave band completely black).

Do the images look like there is a shift between them? Do the upper corners of the image show the same features without large shifts?

Try the DEM-based coregistration instead. It uses a downloaded DEM (preferably SRTM 1ArcSec) to simulate an image which is then matched against both radar images to find their relative shift.

There is a slight horizontal shift between images

Please let us know, which images you are working with (date and time stamp and location).

DEM assisted coregistration solved the issue, thank you.