Zero distance but non-zero velocity upon offset tracking calculation

For some GCPs SNAP can calculate zero distances but non- zero range/ azimuth shifts and velocities. Could please someone explain?
The last highlighted row in the sample below.

I would say that it is because of the heading which is “-1”. But then it leads to the question : why is the heading value equals to -1?

It seems that in this case there is no contradiction between the distance and the heading. No movement (distance = 0) means no direction (the heading is taken to be “-1”). But why does velocity not equal to zero? I don’t understand.

I would say that it is because of smoothing post processing. You can observe that there is a “Average Box Size” in the parameter tab. Even if the velocity is zero, nearby velocities “contaminates” the results.

Yes, but I didn’t use hole filling or averaging… a mystery…

Indeed. If you think there is some bugs inside the algorithm, please make the issue be known to the developers

The algorithm is provided “as is”. If you think there is a problem I’d encourage you to analyse the code yourself. Perhaps there is a threshold under which a movement is considered to be zero? This would not be a “bug”.