The topography data there is lidar based, not from a scanned map. I do suppose there could be some data processing errors, tho I somewhat doubt it. (For example, in one of the areas I generally hike the lidar data says there are three 150 foot deep, small, steep sided pits. They are not real. I'm far from a lidar expert, but I suspect is that instead of a plane-ground-plane bounce, those areas are seeing a plane-water-canopy-water-plane bounce, and thus making it appear the ground is two tree-heights below where it really is. Oh, there are plenty of small ponds in this area, and the three pits are in fact just three of the shallow ponds.)
Clearly that's not the explanation for the anomaly in the GSM, but I suppose there could be some other reflective material in the surface there.