Stratospheric Warming Update (Nov 2nd). We are still looking at a proper Polar Vortex flow collapse with a wind reversal and the earliest official Stratospheric Warming on record. You can monitor all this yourself in my Polar Vortex hub (including the 3D shape); link in the comments below. 👇 The models are still not in perfect agreement, which is also a bit unusual, given that the first event window is already between November 4th and 6th. There is a second reversal window between the 8th and the 11th, but that falls within the same event window (20 days). Only the AI models are currently not on board for any reversal (the AI ENS is not far off), which could be by design. But the ECM is lately pulling back the reversal, and usually the GFS follows with some delay. So definitely a race down to the finish line with this event! 🙂

Would you not be worried with the pullbacks on the ECM maybe AI knows something we don’t and With models still indecisive not sure if we can be entirely 100% just yet

The F1 NERD Max Verstappen understands the rules and regulations of the sport like it's no ones business. The rule states the yellow line is a reference for the driver to park the car in the rightly on the grid. The driver should not exceed the white line by even a centimeter and it's measured by the central alignment of wheel nut against the white line in which Max positions the car's wheel nut just centimeters off the allowed margin.. Give him a doctorate for the sport. Dr. Max Verstappen 😮‍💨

Every winter, SSW gets sold as a possible snow trigger. I checked all 41 since 1958 against 12,430 stations. First 6 weeks vs. matched winters without one: some regions got more snow days, others fewer. The overall average: +0.52%. 🧵

Also, from what I understand, CMIP6 models struggle to resolve stratosphere troposphere coupling during SSWE events. So it’s difficult at the moment to go too deep into this analysis

Love your work. It’s pretty frustrating data to work with!

Time to answer the question: What can an early-November Polar Vortex collapse (SSW) do to Winter 2026/2027? 🧵👇 These maps are not a forecast or actual temperatures. They show the DIFFERENCE between winters with an early-November collapse and winters without one, in the same El Niño+wQBO setup. Since there is no real observational data with such an early SSW and nino/QBO combo, I decided to look for data in two independent model sets: 🔹 Climate models (CMIP6, 80 runs): 32 model winters with an early collapse vs. 565 without. All of them with a strong El Niño + westerly QBO, like this year. 🔹 ECMWF SEAS5 seasonal forecast hindcast (1981–2024): 109 forecast members with a collapse vs 444 without. Each is compared only with the other members of the same forecast, so the ocean and El Niño/QBO are identical. Why 32 vs 565 is ok in the climate models: the big group is the baseline, or the normal El Niño winter without a collapse, as all are the same base state of strong Niño+wQBO, not random winters. It is supposed to look "average", and subtracting it leaves only what the collapse adds. The small group only makes the result noisier, which is why the dots mark where the difference is statistically significant. December is where both agree: 🇺🇸 Central & Eastern US 1–3 °F colder, while northern Canada around Hudson Bay runs milder 🇪🇺 Scandinavia & NE Europe 1–2 °C colder, with high pressure to the north and lower pressure over the Atlantic After December, the signal weakens. The climate models fade it out by January. SEAS5 keeps some cold over the central US and over Scandinavia/eastern Europe into January and February. Bottom line: an early collapse mainly loads the dice for December. It does not lock in a cold winter. But these are just climate simulations, not actual forecasts, as said, only meant to find an underlying signal. Use these with caution, and do not present them as an actual forecast, as they are not. These show the difference in signal, and if the mean base anomaly is +3, a -1 difference still leaves it at +2, for example. For phase 2, I am planning to use a larger CMIP6 set, and I am also thinking of making my own custom simulations, as this is a fascinating situation in development. @judah47 @RyanWeather @webberweather @TWOweather @bam_weather

I’m very curious, what kind of differences would you get for strong to super El Ninos in the CMIP6 and SEAS5 reforecast model world based on the placement of the highest mean SSTs or SSTa? I.e east-based vs “basin-wide” type events.

More great stuff! Obviously in like with my own experience and thinking but nice to see confirmation from the models. What is the latest SAI?

I worked on a study about 20 years ago that showed a cold November in the east was highly correlated to a cold winter overall. I’m not sure if those results are still robust and applicable.

Andrej, can you do the same for the 500 millibar height anomalies? Thank you and outstanding work!

Want to talk Stratospheric Warming weather impacts? Let's talk! But first of all, *this is NOT a forecast*. These are typical weather anomalies in the 0-30 days following a Stratospheric Warming event, starting with the wind reversal and the collapse of the #polarvortex. It also explains why so many people get excited when they hear about something happening in the stratosphere. What we can see in the temperatures first is a colder trend for the whole eastern half of the United States, and also for the northern and western parts of Europe. This is an average of 37 different SSW events, so individual impacts can vary from event to event. But the large majority of these events occurred during meteorological #WINTER , and we have no real analogs for an event in early November with a Super El Niño on top. In terms of snowfall, it follows the cold-spill pattern, with above-normal snowfall in the eastern and northeastern United States, and in a large part of Europe, except in the southeast. In general, SSW events are sensitive to the existing pattern in the lower levels, and can sometimes fail to couple to the surface. This means that the impacts can be deflected. Some events can also only bring cold and snow to Europe or only to the United States. So this is why there is no "one size fits all" when it comes to stratospheric impacts. Take care everyone, and thank you all for your interest in the wonders of the stratosphere! 🙏

I would be interested to see this in probabilistic form using all the SSWEs in ERA-5 since 1940. Perhaps even CPC-like where they’re probabilistic terciles of above, average, or below. Would be curious how the actual distribution shifts in these events & how confident we can be

Let's talk ECM46, the global flagship extended-range model, and the Polar Vortex. Extended range forecasting is mainly about trends on a bit larger scale and multi-day periods, not actual temperatures 30 days from now at 2:15pm. But these are often used for stratospheric forecasting too, trying to spot early signs of a Stratospheric Warming event. Below is the performance of the ECM46 for the current event, where it was long oblivious to anything happening and only started the downward disruption trend around 10 days ago. But this is actually normal and can be found in all modern forecasting solutions, especially in the new AI systems. Long-range predictions do operate on a slightly different basis than our usual daily forecasts, so I will take a look at those at some point too. It also shows how unpredictable the stratosphere really is, both in upward and downward impacts.

Ten days of lead from the 46-day model still beats nothing. Were the AI systems later than ECM46 to flip on this one, or did they all catch it around the same week?