Dealiasing wrongly unfolds areas of rotation

[Soudaaa]

Hi.

I've recently been testing the unravel algorithm with some radar data i got, and in general, it works way better than other algorithms, specially as the data is not of good quality (e.g., sometimes heavily contaminated by catastrophic errors with staggered PRT radars).
However, the algorithm wrongly dealiases areas where rotation is present in supercell thunderstorms. I'm including an example below with the original velocity field and the corrected one. I applied a simple gatefilter using a 10 dBZ threshold to remove some of the spurious data, but it didn't seen to improve much.
I'm not sure of the reason behind this, but I would like to know if I'm doing something wrong or is it an issue with the data, or is the algorithm failing with something.

Thanks