Co-registeration of multi-temporal PolSAR datasets

Hi everyone:
I’m confused about the co-registration when dealing with multi-temporal PolSAR datasets. Should I register each channel(VV, VH) individually and extract polarimetric parameters such as H, A, Alpha after that? I think the solution above is a little complex, do you have other better ideas?
Thanks for your reply in advance.

Sorry for asking, but why do you need them coregistered to a stack?

Sorry for my ambiguous description, I would like to do the classification by employing time-series polarimetric parameters. It seems that tutorials I find are aiming at single channel. So I want to find a solution about multi-channels.

I would then proess all products individually, apply Terrain Correction and then use the “create stack” tool.

Thanks for your reply, i’ve tried you method, it worked. But I still have 3 questtions:
1.I applied target decomposition to extract parameters after Terrain Correction, was that right?
2. Images were geo-coded after TC, did TC contain the back geocoding process?
3. Were the products after TC collocated accurately enough for creating stack? Or the errors could be ignored?

  1. No, Terrain Correction involves resampling. That means you no longer have information on the scattering mechanisms within your pixels. I would terrain correct after decomposition.
  2. I don’t think so but they have a coordinate reference now
  3. In many cases yes, but sometimes, you need to co-register the terrain corrected products to achieve good matches between your rasters.

I think terrain correction are intended to reduce the errors about topographical distortions, could the polarimetric parameters be used as the input? By the way, there are 3 kinds of methods in SNAP(range doppler TC, SAR simulation and SAR simulation TC), which one is applicable in this case?

SAR simulation TC only works when there is topographic pattern in your SAR image. This is not the case for polarimetric features so I recommend Range Doppler Terrain Correction. This should reduce the topographic distortions in your polarimetric features but keep in mind that information on layover or foreshortening areas is still considered lost.