The central puzzle the team confronted is one that has long frustrated the subseasonal-to-seasonal, or S2S, forecasting community. The skill of forecasts targeting East Asian winter cold surges does not vary smoothly or randomly; instead, it clusters. Some initialization dates yield forecasts that capture the timing and intensity of a surge with impressive accuracy, while others—initialized only days apart—miss the event entirely. Historically, the only way to know whether a forecast fell into a high-skill window was to wait several days after issuance, when verifying observations began to arrive. By then, of course, the forecast had already been delivered to energy planners, agricultural agencies, and disaster-preparedness officials, who had no way of knowing whether to lean on it or discount it. The new study asks a deceptively simple question: can those windows be recognized in advance, before the forecast is made, using precursor signals in the ocean?

To answer it, the researchers mined ECMWF hindcast archives covering the winters of 1997 through 2021, a quarter-century record that spans numerous El Niño and La Niña events and a wide range of Arctic sea-ice conditions. From this archive they systematically identified fourteen high-skill windows, encompassing fifty-six individual forecast cases in which week-three cold surge predictions performed exceptionally well. The first, and in some ways most sobering, finding was that the pre-forecast atmospheric circulation offered almost no help. When the team examined the state of the atmosphere in the days leading up to initialization, they could not reliably distinguish upcoming high-skill windows from ordinary periods, leaving a false alarm ratio of roughly nineteen percent. The atmosphere, in other words, does not announce when it is about to become predictable. The answer, it turned out, lay beneath the surface—in the sea.

Regardless of whether the tropical Pacific was in an El Niño or La Niña phase, every one of the fourteen high-skill windows was preceded by the same sea surface temperature pattern in the mid-latitude North Pacific: a warm anomaly over the western basin paired with a cold anomaly over the eastern basin, a configuration the authors describe as a warm-west/cold-east dipole. This dipole, averaged over the seven days before forecast initialization, emerged as a consistent precursor signal across all ENSO backgrounds. Its physical significance is considerable, because mid-latitude SST anomalies of this kind can reshape the baroclinic environment of the North Pacific storm track and excite Rossby wave trains—vast, undulating disturbances in the atmospheric flow that propagate downstream and can lock the wintertime circulation into persistent, high-impact patterns such as the Siberian high and Ural blocking.

Yet the study’s most important conclusion is that the North Pacific dipole, on its own, is not sufficient. A high-skill window materializes only when the dipole is joined by a matching sea surface temperature configuration in the other key ocean basins—and, remarkably, the state of those ancillary regions also determines which of two distinct dynamical pathways the atmosphere will follow. When warm SST anomalies occupy the Barents-Kara Seas, a condition widely regarded as an indicator of regional sea-ice loss, the Rossby wave train excited by the North Pacific dipole is favored to maintain itself and then to sustain the Ural blocking high, the anticyclonic anomaly over western Russia that acts as a gateway for Arctic air to spill into East Asia. In this pathway, the oceanic signal reinforces the blocking ridge, and the cold surge follows a well-teleconnected route from the polar reservoir southward.

The alternative pathway unfolds when cold anomalies instead dominate the Indian Ocean. In that case, the wave energy excited by the North Pacific dipole is confined to the North Pacific–polar sector rather than propagating through the Ural sector, and the circulation response takes the form of a meridional dipole that steers polar air directly southward into East Asia. Two different oceanic backgrounds, two different wave-guide behaviors, two different routes to the same destructive outcome. The practical implication is that forecasters cannot simply check one index or one basin; they must verify that the sea surface temperature anomalies in all the key regions—the mid-latitude North Pacific, the Barents-Kara Seas, the Indian Ocean, and the tropical Pacific—meet their respective thresholds at the same time. Only this simultaneous, basin-wide alignment marks a genuine high-skill window.

When the team applied this multi-region screening criterion to the hindcast archive, the results were dramatic. Forecasts selected by requiring all key-region SST thresholds to be satisfied simultaneously achieved hit rates of one hundred percent under La Niña-A conditions, ninety-five percent under La Niña-B conditions, and ninety percent under El Niño conditions. Those figures stand in stark contrast to the roughly nineteen percent false alarm ratio that prevailed when no such screening was applied. In effect, the oceanic precursor pattern functions as a reliability certificate: if the prescribed SST configuration is present at initialization, the week-three cold surge forecast can be issued with a degree of confidence that S2S prediction has rarely been able to claim, and if it is absent, users know to treat the forecast with caution.

Beyond its operational value, the study carries a deeper scientific message: forecast skill windows are not random accidents of chaos but physically traceable events, anchored in slow, predictable components of the climate system. The ocean evolves on timescales of weeks to months, far slower than the atmosphere, and its anomalies act as a kind of memory that conditions how atmospheric disturbances will amplify, propagate, and persist. By mapping which oceanic configurations render the atmosphere more predictable, the researchers have effectively converted an abstract question about ensemble spread and verification statistics into a concrete, observable checklist. The work also underscores the growing recognition that mid-latitude predictability is jointly governed by the tropics, the Arctic, and the mid-latitudes themselves, with the Barents-Kara Seas linking the story of Arctic sea-ice decline directly to the practical skill of winter forecasts thousands of kilometers away.

For the agencies that must act on winter forecasts—grid operators deciding when to pre-position fuel reserves, transportation authorities planning for ice and snow, and farmers protecting overwintering crops—the ability to know in advance that the coming two to three weeks are likely to be a high-skill period is of immediate practical value. It allows limited confidence to be spent where it is justified and withheld where it is not, sharpening decisions in energy dispatch and agricultural disaster prevention alike. As subseasonal prediction matures from a research frontier into an operational mainstay, studies of this kind suggest that the future of reliable week-three forecasting may depend less on faster supercomputers than on learning to read the ocean’s slow, patient signals before the atmosphere ever makes its move.

Subject of Research: Identifying sea surface temperature precursors that reveal high-skill windows in subseasonal forecasts of East Asian winter cold surges

Article Title: A "warm-west/cold-east" North Pacific sea surface temperature dipole, working in concert with anomalies in other key ocean regions, reveals when subseasonal forecasts of East Asian winter cold surges can be trusted

Article References: A "warm-west/cold-east" North Pacific sea surface temperature dipole, working in concert with anomalies in other key ocean regions, reveals when subseasonal forecasts of East Asian winter cold surges can be trusted. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: subseasonal forecasting, East Asian winter monsoon, cold surges, sea surface temperature, North Pacific dipole, El Niño, La Niña, Barents-Kara Seas, Ural blocking, Rossby wave train, ECMWF hindcasts, forecast skill windows