Wind Field Estimation discontinuities - 0.1 values


I am trying to retrieve wind field from GRD products in 2018 and I am comparing the results of the wind field estimation (in SNAP) with the respective OCN products of the same date and time. I have applied preprocessing steps (workflow: apply orbit file, thermal noise removal, border noise removal, calibration, speckle filtering, wind field estimation) and I am experimenting with different speckle filtering methods and parameters of these methods but I always get the same result, with slight differences. Lots of 0.1 values that don’t fit to the surrounding pixels, plus some areas that show significantly larger values of wind speed comparing them with the respective OCN images. The image in general fits the patterns of OCN images but there seem to be a persistent noise that I do not know what to do to remove(Fig.1 is the generated wind filed from GRD while Fig.2 is the OCN product). Does anyone have any idea what steps am i missing?