Tuesday, June 7, 2011

Who Saw This Heat Coming? One Forecaster ... Sort Of

Last Wednesday night, we noted the various forecasts for the extended outlook (which took us to Monday thru Wednesday of the current week). We were gobsmacked to see the forecast of one station and one forecaster stand out. Per the Monday through Wednesday period detailed below, KSTP was predicting temperatures considerably above the others, particularly for the middle day (Tuesday).

WCCO: 85/85/80
KSTP: 88/92/94
FOX: 84/85/84
KARE: 84/--/--
NWS: 83/81/83
Weather.com: 87/85/77
Accuweather.com: 80/75/74
Strib: 79/76/75

We raised the seemingly odd prediction with a member of the KSTP meteorology staff, who confirmed the prediction with us last Thursday morning (below):

As everyone knows by now, the Twin Cities shattered a record, and KSTP's seemingly overly enthusiastic prediction of 92 for today (as of last Wednesday) turned out to off by double digits from the actual temperature.

Still, being off by 11 degrees passes as a sterling forecast in comparison to the prognostications from other forecasts, one of which (the Star Tribune forecast) was off by 27 degrees. Here's how far off the Twin Cities (and national) forecasters were six days in advance of today's record heat:

WCCO: 18
KSTP: 11
FOX: 18
KARE: no prediction made beyond five days (a wise choice?)
NWS: 22
Weather.com: 18
Accuweather.com: 28
Strib: 27

While KSTP deserves credit for predicting a heat wave that nobody else was able to see in the extended outlook, KSTP also detailed in its blog a day or two ago that temperatures would remain above normal for at least 10 days. That, it certainly appears based on a now-universally predicted chill down, will not come to fruition. Ah, the weather, ever hard to predict.

High Dewpoints Exaggerated by Some Forecasters
In several cases, a few weathercasters overpredicted dew points. The Strib called for dew point temperatures to be in the 70s for both Monday and Tuesday. Neither happened, and in fact dew points actually spent some significant time in the 50s. Fox also suggested that dew points would get into the 70s on at least one occasion. From our perspective, this was really not an especially steamy scenario. Warm and uncomfortable to be sure, but not as sauna-like as it could have been or as some forecasters would have you believe.

A Picture For Your Grandchildren
As of late Tuesday afternoon, here's a graphic you'll hardly ever see. The Twin Cities boasted the hottest temperature in the country.

The smoky sunset that concluded the 103-degree day:


  1. Does everyone realize that today was a mere 5 degrees off of the hottest recorded temperature at MSP(1936-108 degrees) I hope we got our 90's and 100's out of our systems,I for one is counting the minutes till the cold front arrives this weather can stay in Texas where it belongs(and the 9 people who answered the poll question "the hotter the better" can hop on the next plane to Texas and follow it down there)bring on the 60's and 70's.

  2. This is a great post Bill.

    To think that a Met could be 20 to 30 degrees off with their temp forecast, even 5 days in advanced, is alarming. The real question is: Why did so many other outlets disregard the Euro models when Dave Dahl bought into the solutions? The Euro models predicted this heat wave well in advance.

    Were the other Mets lazy and did they not even consider looking at the Euro solutions? Do these Mets specifically live and die by the U.S. models including the GFSX & GFS?

    These are questions that need to be answered internally within our industry. What shocks me is that news directors and/or station administrators rarely question their broadcast Mets on their forecasts. Maybe they don't care?

  3. Just as amazing: three days out (the Saturday night forecast for Tuesday) no forecasters were within 14 degrees of the actual high temp.

  4. It's from 2005, but this blog post from ForecastAdvisor.com shows that in general, forecasters do a much poorer job forecasting extremes (of which this was one) than forecasting "normal" tempeatures:


  5. When you forecast an extreme, whether it be temp or precipitation you are taking a much bigger risk that it won't happen. For example in 1991 Paul Douglas was the only forecaster who even remotely thought that accumulating snow could happen on halloween. Its a gamble that the forecaster is taking. Play it safe and go for the mid 80s or double down and go for the 90's (which in the end wasn't even warm enough).

  6. Novak...a fantastic post, I think that comment is spot on. I think most mets relay on Moss guidance, when it is notoriously bad when it comes to strong warm and cold fronts.

    This was a classic case that should be studied at places like PSU,SCS and the U of OK. Looking back, the Upper Mississippi Valley was under the influence of a high pressure ridge at the 500mb level, and with it's clockwise winds it was pulling in warm air aloft into our region. At the same time at the surface there was a low pressure system (with it counterclockwise winds) that was forecast to undercut the ridge, bringing in more humid air the first day of the heat wave, and on the second day the winds at the LLJ level would turn around to the sw as it moved into the Dakota's bringing dryer(but still very warm) air from the SW US allowing the temps to warm even more. The surface maps were showing the warm front stalling out near the international border...this whole set up was aching for a mid 90's forecast and possibly higher a good 5 days out. For my own personal reasons did I see 100° that far out?...no way, but I thought mid 90's were likely..I think the industry relays to much on the crutches than the shovels.

    Any way that's my take on it

    BTW....I would grade most of the outlets on this one a D (out of fear of over sensationalizing it) but give KSTP a B, they went far above MOSS guidance, just not far enough

  7. Interesting post, Randy! You make a lot of sense.

  8. most mets are so afraid of forecasting mid 90's out 5-7 days...as there are so few up here...even though they see it, they are afraid to take the head first dive..They need to stop that...tell what they see, don't sugar coat guys and gals

  9. sorry to take up so much space. Next question is this? If the local stations have their own in house models and subscribe to them are they obligated to show that model on their broadcast,based on contractual agreements? even though they say they disagree with it? Seems to me that's the case. I think it time to end those contracts, so that the local can show what they want.....that brings back another question??? weather producers at the stations? or are local mets responsible for that? For instance can I show the HRRR model or do I have to stick with the contract? even though I believe the HRRR is the correct solution? Hmmm some heavy question's there.

  10. 68 on thursday much better,frost north of duluth,got to love minnesota,dont like the weather wait 24 hours!

  11. Nice blog post, Bill. Temperature forecasts can be just as interesting as snow storms and possibly as difficult in certain situations. Kudos to KSTP for making the gutsy call. If it walks like a duck, quacks like a duck then ......

  12. next Tuesday into Wednesday is starting to look interesting based on the Euro model which has been very consistent in the last few runs...the GFS is starting to lean that way.

  13. Plymouth Weather LoverJune 13, 2011 at 1:02 PM

    I think next weekend (June 17--late through June 19)look even more interesting based on the recent model runs.