Range offsets issue

[vjpbd]

We are working with X-band SAR images of glaciers acquired 1-4 days apart with the aim to derive velocity fields from them. Most of the images 1 day apart allow extraction of coherent LOS InSAR images. But we are applying the denseoffset tracking function in ISCE2 to get the along track component of the ice motion and to fill in the gaps of the LOS motion where the InSAR fails (particularly for scenes >1 day apart) . We expected to have the LOS InSAR displacements and the slant-range pixel offsets agree with each other but this is not the case.
To check if the problem was with the InSAR or the offset tracking we compared with the offset tracking made by Enveo (www.enveo.at). The LOS InSAR results and the Enveo slant-range offset tracking agrees quite well (see attached figures including maps with results projected into map coordinates and graphs). This mean that ISCE2 offset tracking is the odd one out.

LOS InSAR vs ISCE2 range offsets:

LOS InSAR vs Enveo range offsets:

ISCE2 and Enveo range offset comparison:

This does not seem to be a scaling issue (comparison does not show a linear trend) nor does this behave in the same manner for different tracks. Comparison for a descending track with small incidence angle (9-14° relative to vertical) indicates overestimate for ISCE2 range offsets while ascending track with relatively large incidence angle (36-38° relative to vertical) indicates underestimate for ISCE2 range offsets.

Note that both ISCE2 and Enveo azimuth offset agree very well with each other, hence our problems seems to be isolated to the range offsets from ISCE2:

Any ideas what could have gone wrong with the ISCE2 range offset tracking? Is possible that we are not running the function correctly to get the actual slant-range displacements out of it or is the output something else than the slant-range displacements?