As a a few bands of rain rolled through the SF Bay Area last night and today, I took the opportunity to compare the forecast models as it went through.
I live up in the mountains, and also reasonably close to both the SF bay and the Pacific ocean, so this can be a "challenging" place for forecasts.
Almost all the models forecast continuous rain overnight and through this morning, with the exception of the HRRR, which was by far the most accurate, correctly forecasting long gaps in the rain which were obvious if you were to just look at the weather radar map.
I'm wondering two things:
If the ECMWF is really the best default model for the mainland USA?
Why the HRRR and NAM models revert to to the GFS once you look too far into the future? Wouldn't it make more sense for them to revert to the ECMWF, if that is supposed to be better?