I have a question regarding incidence angle effect.
I have TanDEM-X data that I did not apply incidence angle correction because actually I did not see any gradient over the backscatter intensity (incidence angle near range to far range were from 19 to 21 degree).
I made an interferogram and unwrapped it and again I did not see any incidence angle effect but when I converted unwrapped result to elevation then I can see an effect like gradient.
Ramps in TanDEM-X data are quite common, at least in my experience. This can occur because of slight coregistration errors or bad orbit quality. Also the phase to height conversion is somehow prone to errors.
Maybe this discussion here helps you a bit:
You can try a higher polynomial for the flat earth phase and the number of estimation points:
One way to get a better unwrapping is to remove the topographic phase, perform the unwrapping and add the topographic phase again after the unwrapping.