Sounds like a nightmare to get properly labeled data… those NWS “trained spotter” reports would only go so far. Taken at different times, at inconsistent locations, tricky to account for drift and aspect.

If you know anything about the shenanigans Killington pulls with their snow stake, the flaws here would be very obvious. They report from a spot on the mountain that is not representative at all — the snow piles up there twice as deep as elsewhere. Their snow stake is a dirty liar.

Also, the best model this winter will probably be very different from the best model in an El Niño year.