That's an odd bug. I did notice that if I entered -0.1 for the minimum value, -100 for the maximum value, clicked preview, changed maximum back to 100, and then clicked apply, the mound was showing as getting increasingly lower. It was also going to -100, which is not as desired. However, using Filter>>Mathematical>>Scale with a value of 0.1 did properly scale the below-sea part to from -0.01 to -100.

Further uses of the mound look like there is an exponential operation being applied somewhere. Lazy software developers producing shoddy products is the only excuse. Doesn't anyone ever test these things or do they just get pushed out and wait for end-user reports?