Every fall, as the northern hemisphere approaches winter, Judah Cohen begins to put together a complex, atmospheric puzzle. Cohen, a research associate in MIT's Department of Civil and Environmental Engineering (CEE), has spent decades studying how Arctic conditions set the course of winter weather in Europe, Asia and North America. His research dates back to his postdoctoral work with Bacardi and Stockholm Water Foundations Professor Dara Entekhabi, who looked at snow cover in the Siberian region and its relationship to winter forecasting.
Cohen's forecasts for the winter of 2025-2026 highlight a season characterized by Arctic-sourced indicators using next-generation artificial intelligence tools to help provide a complete picture of the atmosphere.
Going beyond mere climatic factors
Winter forecasts rely largely on the El Niño-Southern Oscillation (ENSO) diagnostic, conditions in the tropical Pacific and atmosphere that influence weather around the world. However, Cohen notes that ENSO is relatively weak this year.
“When ENSO is weak, Arctic climate indicators become particularly important,” Cohen says.
In his subseasonal forecasts, Cohen monitors high-latitude diagnostics such as October snow cover in Siberia, early-season temperature changes, Arctic sea ice extent and the stability of the polar vortex. “These indicators can tell us about the coming winter in surprisingly detailed ways,” he says.
One of the most consistent predictors of Cohen's data is October weather in Siberia. This year, while October was unusually warm in the Northern Hemisphere, Siberia was colder than usual and snow fell early. “Low temperatures combined with early snow cover tend to promote the formation of cold air masses that can later move into Europe and North America,” Cohen says — weather patterns that have historically been linked to more frequent periods of frost later in winter.
High ocean temperatures in the Barents-Kara Sea and the “easterly” phase of the quasi-biennial oscillation also suggest a potentially weaker polar vortex in early winter. When this disturbance combines with surface conditions in December, it leads to lower than normal temperatures in parts of Eurasia and North America early in the season.
AI subseasonal forecasting
While AI-based weather models have made impressive progress in providing short-term forecasts (one to ten days), these advances are not yet applicable over longer periods. Subseasonal forecasts spanning two to six weeks remain one of the most difficult challenges in the field.
This gap means this year could be a turning point in off-season weather forecasting. Cohen's research team won first place for the fall season in the WeatherQuest 2025 AI subseasonal forecast competition organized by the European Center for Medium-Range Weather Forecasts (ECMWF). The challenge assesses how well AI models capture temperature patterns over multiple weeks, where forecasting has been limited in the past.
The winning model combined machine learning pattern recognition with the same Arctic diagnostics Cohen had spent decades refining. The system demonstrated significant improvements in multi-week forecasting, outperforming leading AI and statistical baselines.
“If this level of performance continues over multiple seasons, it could represent a real leap forward in offseason forecasting,” Cohen says
The model also detected a potential cold wave on the U.S. East Coast in mid-December much earlier than usual, weeks before such signals typically appear. The forecast was widely publicized in real-time media. If verified, Cohen explains, it would show how combining Arctic metrics with artificial intelligence could increase turnaround times for predicting severe weather.
“Reporting a potential extreme event three to four weeks in advance would be a watershed moment,” he adds. “This would give utilities, transportation systems and public agencies more time to prepare.”
What this winter may bring
Cohen's model indicates a greater likelihood of colder-than-normal conditions across parts of Eurasia and central North America later in the winter, with the strongest anomalies likely to occur mid-season.
“We are still in the early stages and patterns may change,” Cohen says. “But the ingredients for a cooler winter pattern are already there.”
As Arctic warming accelerates, its impact on winter behavior becomes increasingly apparent, making understanding these connections increasingly important for energy planning, transportation and public safety. Cohen's work shows that the Arctic has untapped off-season forecasting power, and artificial intelligence can help unlock it in time frames that have long challenged traditional models.
In November, Cohen even appeared as a tip Washington Post. crosswordwhich is a small sign of how widely his research has entered public discussions about winter weather.
“For me, the Arctic has always been a place worth seeing,” he says. “Now artificial intelligence gives us new ways to interpret signals.”
Cohen will continue to update his projections throughout the season blog.
