Just to play devil's advocate:
It looks like the original images were taken in 8-bit per channel format instead of the preferable 16-bpc. Since the dark material's color varriation seems to occur in a narrow range it may be overwhelming the available pallet causing banding along intersecting color gradients - possibly giving rise to the angular shaped regions.

This bump mapped image can be missleading - it takes the original image and asigns "high" to white and "low" to black so even a cloud or fog bank can appear to be a mountain complete with a shadow. On the other hand a dark hill may come out looking like a depression.

I tend to agree with the artifact interpretation over the crater view because they look like they happen in roughly the same places in the images.

... or maybe bergs are dislodged and partially submerged by impacting ring material kicked out by a wayward moonlet ;)