It's been that way for literally ever. You telling me weather models 10 years ago had better long-range accuracy than one's today? lol. There's so much that weather models cant see, I mean they are literally attempting to predict the interactions between a million frickin variables when there are at least 10 million more they just can't account for. Think of all the small details they don't see that add up into big changes, small topographical changes compounded over long distances, maybe even the interaction between built environments and the atmosphere, or even something like chaos theories butterfly affect.
I mean check this out, this is what goes into weather models. literal chaos theory and disorder. Look at this lorentz attractor. funny I learned about this from an article about dimensions of consciousness. This is weather turbulence modeled, a set of chaotic solutions. I also wonder how small temp changes work. I mean, our weather sensor might read something that rounds too 70.4 degrees, when in reality if we could break up a 1 cubic meter area of air and break that even further into smaller areas and be able to measure the temps in each of those smaller areas we would find so many variations! from 70.4, to 70.3893 to 70.2 to 70.7 etc etc. I mean weather models are just not able to see that stuff.
www.ncbi.nlm.nih.gov