I'm just making an assumption, but it looks like the temperature output is somewhat skewed over water. For example, it seems unlikely that all of Norway would be -10 to -25 C (which is below normal), while the air temperature just off the coast is 0 to 5 (which, as you note, would be well above normal). The gradient tracks the coast perfectly, which seems to me to be a red flag.
(I could be entirely wrong here. Just using logic and not any science.)