I guess I should have been paying a little bit more attention to the vertically integrated smoke forecast and not just the near-surface smoke forecast. rapidrefresh.noaa.gov/hrrr/H… says the Bay Area is going to be stuck with dark orange skies for at least another 40 hours...
2
6
4
20
Note that the NWS publicly stated that their models aren't working for this, so I'd book this in "who knows" territory.
1
1
Replying to @groby @slightlylate
Based particularly on nitter.vloup.ch/NWSBayArea/statu… I think what they were saying is that forecasts for things like temperature were not accounting for smoke... not that the smoke model itself (which is separate from other models) is broken.
NWS Bay Area 🌉
 @NWSBayArea
9 Sep 2020
Replying to @disco_lemonade1
The models do not seem to be keeping up with the thick layer of smoke as far as temperature forecast data, visibility, etc. We just have so much smoke over us right now. Yesterday the models were way off and we expect the same today (for afternoon temperatures).

Sep 9, 2020 · 10:26 PM UTC

2
1
Replying to @davidbaron @groby @slightlylate
That said, I wouldn't expect models to do particularly well in edge cases that they weren't well tested on. But I still think they're likely better than nothing. (Where the standard "nothing" baseline is the best linear combination of climatology and persistence.)
Replying to @davidbaron @slightlylate
I'd be surprised if you can correctly work out convection when you don't know temperatures :)
1