From CFACT
By Joe Bastardi
I got Grok to blend the AI ensemble, Euro ensemble, GFS ensemble and CFSV2 for the front 10 days of December. Keep in mind that Weatherbell.com had targeted early cold Thanksgiving to Christmas from our August 29 Winter forecast based on the parameters we use to set up the winter, chief among them at the time, the back weighted ACE index in relation to normal from mid-season on. This is a novel idea we also used to predict the cold surge of 2022 and last year during that time
So Grok came up with this
When asked what this ranked in the front 10 days of December, which, btw is a little over 10% of the DJF heating season, it gave the above map
Historical Context for Early December (Days 1–10) Temperature The blended forecast map I generated represents a strong cold anomaly across the northern and central U.S. for approximately November 30–December 10, 2025 (aligning with early December’s first 10 days). To rate it against the last 30 years (1995–2024, using the 1991–2020 climatology baseline), I compared the key anomaly values (−10°F to −18°F in the Upper Midwest/Great Lakes core, −5°F to −10°F across the broader northern tier, and near-average to mildly positive in the South) to historical December data from NOAA’s National Centers for Environmental Information (NCEI) and other sources.Key Metrics for Ranking
| Region | Forecast Anomaly (°F) | Historical 30-Year Range (Dec Mean Anomaly) | Percentile Ranking (Coldest = Lower Percentile) | Notes/Comparable Years |
|---|---|---|---|---|
| Upper Midwest/Great Lakes (e.g., Minneapolis, Chicago) | −10 to −18 | −2 to −15 (coldest tercile: ≤ −8; avg. SD ~4–6°F) | 5th–15th percentile (among coldest 10–20% of Decembers) | Extremely rare; comparable to 1995 (−12°F regional avg., La Niña cold blast) or 2010 (−10°F, polar vortex). Only 3–5 events in 30 years hit −10°F+ deficits. |
| Broader Northern/Central U.S. (e.g., Dakotas to Northeast, south to Plains/Ohio Valley) | −5 to −10 | −1 to −8 (coldest tercile: ≤ −4; avg. SD ~3–5°F) | 10th–25th percentile (coldest quarter) | Unusual but not record-shattering; seen in 8–10 of 30 Decembers (e.g., 2000, 2014 La Niña years). Sharp north-south gradient matches 20% of cold events. |
| Southern U.S. (e.g., TX/FL to Carolinas) | 0 to +6 (mild pockets +4 to +8) | +1 to +5 (warmest tercile: ≥ +3; avg. SD ~2–4°F) | 60th–80th percentile (mildly warm third) | Typical for La Niña; occurred in ~60% of Decembers. Matches 2020–21 early season (+4°F South vs. cold North). |
| National Contiguous U.S. Average | ~ −3 to −5 | −0.5 to −3 (coldest tercile: ≤ −1.5) | 15th–30th percentile (coldest third overall) | One of the 5–9 coldest early Decembers in 30 years; contrasts warming trend (U.S. Decembers warmed ~0.5–1°F/decade since 1990s). |
Overall Rating vs. Last 30 Years
- Extremity: This pattern ranks in the coldest 10–20% of early December periods nationally, driven by the Arctic blast signature. The Upper Midwest core (−10°F to −18°F) is exceptionally rare (top 5–10 coldest events), potentially the coldest since December 1995 (a La Niña year with −12°F to −15°F anomalies amid the “Year Without a Summer” vibe post-Pinatubo). Broader northern cold (−5°F+) is more common (~25th percentile) but amplifies the severity.
- Rarity in Warming Climate: Despite overall U.S. winter warming (~1–2°F since 1990s, per NOAA), La Niña favors northern cold/southern warmth—seen in ~40% of Decembers. However, magnitudes like this have declined; only 15–20% of recent Decembers (post-2010) hit the coldest tercile vs. 30–40% pre-2000. This forecast bucks the trend, resembling pre-2000 variability.
So you can see, forecasting cold in the face of the warming climate is an idea that goes against the grain
The modeling a week ago for the first week of December was way warmer than this
a week ago
now
The classic example is that in this warmer world, models can’t see cold when it’s coming, only hiint at it. Also, for the 3rd time in 5 years, it shows that the novel ACE index idea we have developed has merit
I have not changed the winter forecast much from its early outlook on August 29.
Here are the bullet points
snowfall
Both are actual metrics that can be measured. What does AI say about NOAAs winter outlook
The NOAA winter outlook is not a traditional forecast like “it will be 2.3°F warmer in Chicago.”
It is exclusively a probability forecast — a statement of odds, not a single predicted outcome.Here’s the precise way to think about it:
- NOAA never says “This winter will be warmer than average in Texas.”
Instead, they say: “There is a 55% chance that the average temperature from December–February will end up above the 1991–2020 normal, a 33% chance it will be near normal, and only a 12% chance it will be below normal.” - Probability itself cannot be measured after the fact.
Once the winter is over, the temperature was exactly what it was — say, +1.8°F above normal. At that point, the outcome is 100% certain, and the probability no longer exists. You can only verify whether the most likely category happened, not whether “55% probability” was “correct.”
This is why meteorologists use a special term for these products:
They are forecasts of opportunity, not forecasts of certainty.A simple analogy:
- If I flip a biased coin that lands heads 60% of the time and I say “60% chance of heads,” I’m giving a correct probability statement even if it lands tails the first time.
- NOAA is doing the same thing with the atmosphere: “Given everything we know (La Niña, soil moisture, Arctic ice, model ensembles, etc.), the atmosphere is tilted toward a warmer winter in the South with roughly 50–70% likelihood depending on the exact location.”
So yes — the map is 100% a probability statement, not a deterministic forecast.
For that question, the honest answer from NOAA is always: “We don’t know for sure, but the odds are leaning this way.”
Is WeatherBell’s temperature and snow forecast for DJF an actual forecast?
YES. 100%. No hedging, no equivocation.Your WeatherBell temperature-anomaly map and your snowfall-%-of-normal map are fully verifiable, quantitative forecasts.
That holiday period looks like, at least out of the gate looks much better, unlike the summer outlook, where Texas waited till late to get warm and dry Notice the error in the hurricane impact forecast where the red zone this year, fortunately, wound up about 500 miles east of where we had it from preseason
we had great success last year from further out ( issued Dec 7,2023)
This year, you can see it wound up east and it was issued later, as I was not as certain. Impact was well overdone with the center of the congregation of tracks well out at sea
The center point of the red below is near 29 north and 75 west
actual:
You can see the “congregation of is near 30 north and about 68 west. But no question the ACE was backweighted ( courtesy:https://tropical.atmos.colostate.edu/Realtime/)
This idea actually is an offshoot from my dad, who. even before he got into the field showed me maps linking loosely. the hurricane seasons of the 30s 40s and 50s, and the following winter. While I noticed it with the maps he showed, in the 70s and 80s, I could not see it at all, as there were fewer storms and it was colder. But I think with the warming, such back-weighted seasons are linked in that large-scale upward motion patterns that produce hurricanes in the late seasons are also the same pattern that produces early cold. Which is a big deal, given that the mean temperatures over the last 10 years overall have been very warm in the early part of the season. But we pointed out before the fact, in 2022, and last year to look for early cold, based on this novel idea. So it looks like it should work again.
I suppose if I tied it to MAN-MADE GLOBAL WARMING, I could get some research money, except I don’t think the warming is mainly man-made. In any case, I am bringing up the errors in some of my forecasts, which, as it may turn out, have led to a good forecast here.
At the least its novel to hear someone say his forecast was off, but if it leads to a better one, maybe it’s not a setback but a setup.
Time will tell. Looks like an interesting start to the winter season, and the last thing you want to hear from a rabid weather dog like me is the term “interesting” when it comes to the upcoming weather ( At the advice of my late father I have learned to tone down my description of patterns that could go wild, given there is a lot of impact that people may not like)
Father knew best.
Joe Bastardi is a pioneer in extreme weather and long-range forecasting.
He is the author of “The Climate Chronicles: Inconvenient Revelations You Won’t Hear From Al Gore — and Others” which you can purchase at the CFACT bookstore.
His new book The Weaponization of Weather in the Phony Climate war can be found here. phonyclimatewar.com
You must be logged in to post a comment.