TimB

To me, that’s the 1991-2020 normal, which is like a degree and a half above the 1981-2010 normal due to the ridiculously warm Septembers we had in the second half of the 2010s, which is exactly what I’m grumbling about.

Beyond frustrating. Why can’t we ever get fall weather in fall anymore?

Of course it’s a little more complicated than that. It’s more like rolling a die in today’s climate, where rolling a 1 means below normal temps and rolling a 2-6 means above normal. Rolling a 1 for October doesn’t make rolling a 1 for November any less likely. That said, I would venture to guess that each roll is at least slightly dependent on the one before it, but I would guess that a below normal October would make a below normal November MORE likely rather than less.

To a certain extent, I agree, except never-ending summer temps in September and October make me angry and irritable, and getting below normal temps in September and October doesn’t make it any less likely that we’ll get them in winter. Just like flipping a coin and it landing heads doesn’t make it any more or less likely that the next flip lands heads.

Be interesting to see how it all evolves. I see a lot of heat in the eastern 2/3 of the country, but the GFS mutes it somewhat and the other models come down a bit. Getting through the first 11 days of the month without even one positive departure will be a big win.

My understanding is there are 3 categories: above normal, normal, and below normal. So it’s probably 60/30/10 or something.

This is ugly: Majority of the US under high probabilities of above normal now.

You’ll still have a hard time convincing me that we have much more than 73 years of data that isn’t at the very least somewhat suspect. September 1881 had the same average temperature as July 2020? I mean, come on.

Oops, posted this in the summer thread. This belongs in the fall thread. Very puzzling, the last sentence of this morning’s forecast discussion weighed against the last few days of the zone forecast:

Probably semi-comparable to Feb. 2015.

I’d more than willingly trade a +8 Sept/Oct for a -8 DJF.

I don’t see a -8 in anyone’s future, even if the GFS pans out.

GFS is back to wanting to shut off the heat before it gets too ugly. I see nothing of the sort on the EC or CMC.

I see the guy used Morgantown, Erie, Wheeling and Dubois, but not Pittsburgh. But I would say lows, at least at Pittsburgh. June was +1.7 for highs and +2.0 for lows, July was -1.0 for highs and -0.2 for lows, and August was +1.9 for highs and +3.1 for lows.

Interesting. Thanks for sharing!

...and that’s the bottom line. The general public isn’t looking at the GFS, the general public is looking at a forecast on an iPhone app that is a blend of models, and the vast majority of them have no idea where any of the data comes from. Beyond that, if you took 10 random people, even people who have lived in Harrisburg their whole lives, off the street and asked them “what’s the average high in Harrisburg on September 15th,” I’d say a majority of them wouldn’t be able to produce a number within 5 degrees of the actual value.

So your view on the matter is, a +5 departure on a day when a model showed a +30 departure is just as bad as a +5 departure on a day when a model showed a -20 departure. As a math/stat person, I very much agree. Just didn’t know if I was missing something. And of course, I would say such an error on D5 should be weighted a lot more heavily than such an error on D15.

So that would indicate that even in mid-September, one can expect that over 1/3 of the days would have at least slightly uncomfortable dewpoints of 60+, and that 9 out of 15 wouldn’t be that extreme.

So here’s a topic for discussion: if the average high for a particular date is 82 and the GFS models a high of 113 at 4.5 days out, then the actual high is 88, how egregious of a modeling error is that? I mean the GFS is modeling above normal temps, and above normal temps may verify.

It’s highly unlikely to happen, as there’s no support for it, but this is the GFS showing something only 105 hours out that is way out of sync with other models (and 20+ degrees above the official NWS forecast for those areas). As for the magnitude of what it’s depicting, verbatim those temps are 6-7 degrees higher at Omaha and Lincoln, NE than any September day on record.

Beyond that, is something bad running through the GFS algorithm when it’s putting out these numbers for the Central US for D5? That’s some 10-20 degrees higher than other models. Maybe our hot stretch is error carried forward from this:

I feel like I used to be able to get it from Wunderground back in the day but that was a long time ago. I don’t use that site near as much ever since TWC took it over and it went to crap. As for the programming language stuff, you might as well be speaking Russian to me.

So the question boils down to: has a 15 day stretch (if we take the GFS verbatim) where the dewpoint drops below 60 on just 7 days been common and normal enough historically in mid-September to not be irritated with the endless humidity? My gut instinct would be no, but I could be wrong. The other question, of course, is should I be concerning myself with something the GFS is showing today when it’s been flip-flopping even more than its normal high standard for flip-flopping? Might be more prudent to wait and see what happens and put off my grouchiness about endless summer heat and humidity until/unless it actually comes to fruition.

Would be interesting to compare September DPs at PIT or MDT over the past 30 years and see how much of an increase there has or hasn’t been in the number of days where the DP exceeded 60 or 65 or 70 or whatever. Maybe my mind just romanticizes Septembers past and it isn’t justified.