I can’t exactly tell which one is more promising for InSAR applications of surface deformation, but there are studies which argue in favor of certain polarizations.
My main point was that if Sentinel-1 data is being used, the co-polarization channel (mostly VV, but HH in some specific regions, see the last figure on this page) is better than the cross-pol channel, mainly because of two reasons:
- better fringes and higher coherence, see here: Effect of polarization on interferometric products
- using both channels requires more processing time and disk space and can cause errors in SNAP related to file naming during the the unwrapping steps.
This was why I advised to reduce the data to the VV or HH information (whatever is available, mostly not both at a time) already at the interferogram generation step.