A new peer-reviewed paper links winter warmth in the tropical southern Atlantic to Pacific cooling the following summer. It is a lag relationship in past data, not a forecast for 2027, and its methods have not yet been read here.
A peer-reviewed paper published November 6 in the Journal of Geophysical Research: Atmospheres reports that sea surface temperature (SST) anomalies in the tropical southern Atlantic in winter favor the opposite sign in ENSO-related Pacific SST the following summer. A warm Atlantic favors a cooler Pacific, and a cold Atlantic a warmer one. The authors call it a potential new ENSO predictor.
That is a lag relationship in past data, not a forecast for 2027. The winter it would apply to, 2026-27, has not finished yet. We could read the abstract and the publication record. We could not read the methods, because the publisher’s site blocks automated access, so this piece covers what the abstract establishes and lists what the methods need to show.
| Study | Status | Evidence base | What we read |
|---|---|---|---|
| Zhang et al., JGR: Atmospheres, 131(21), e2026JD046812 | Peer reviewed. Received 1 March, accepted 19 October, published online 6 November 2026. | Reanalysis data and a model simulation, per the abstract. Region, months and years not read. | Abstract and publication record (Crossref, OpenAlex). Not the main text. |
Using reanalysis data and a model simulation, Zhang and colleagues report a significant lag relationship between tropical southern Atlantic (TSA) SST in winter and ENSO the following summer. The abstract says the warm (cold) TSA anomaly in the preceding winter favors cold (warm) ENSO-related SST anomalies in the following summer. The word is “favors,” a tendency, and the abstract does not describe a guaranteed sign for every year. A model simulation is said to support the result.
The abstract does not give the TSA region, the months counted as winter, the years analyzed or the strength of the correlation. Those details weren’t available in the material we read, and they decide how much weight the result can carry.
The abstract describes a chain from one ocean to the other. A warm TSA anomaly can sustain to the following summer and induces an anomalous Walker circulation spanning the tropical Atlantic and Pacific, the east-west overturning of the tropical atmosphere. That circulation triggers an anomalous low-level easterly over the tropical Pacific. The abstract does not say whether the Walker circulation strengthens or weakens. Three ocean processes then turn the wind anomaly into an SST anomaly: zonal advection, where currents carry warm or cold water east and west; Ekman pumping, where wind-driven changes in upwelling shift the depth of the thermocline, the boundary between warm surface water and cold water below; and the Bjerknes feedback, where SST, winds and thermocline depth reinforce one another.
In the usual picture, stronger easterlies cool the central and eastern Pacific, which fits a cool Pacific following a warm Atlantic. For a cold TSA anomaly, the abstract says only that it corresponds to the opposite circulation anomaly and SST anomaly.
Our reading. Every link in the chain is a process ENSO research already uses. What is new is the claim that the chain starts in the southern Atlantic and works across a lag from one winter to the following summer. The abstract describes the pathway and says modeling corroborates it. It does not say how large the effect is compared with the Pacific’s own variability.
A predictor earns its place only if it adds information that forecasters do not already have: the current ENSO state, the heat stored in the ocean, and the dynamical models. We will look for four things in the methods. First, how strong the relationship is, as a correlation and as a share of variance. Second, whether it survives once ENSO’s own winter state is accounted for, since ENSO events tend to decay by the following summer, and a lagged link can appear by that route alone. Third, whether it has been tested on years not used to find it, with data that would have been available in real time. Fourth, how many independent events the relationship rests on.
These are questions to check, not criticisms of the paper. The abstract does not say how the paper handles them.
It is one of several 2026 papers on Atlantic influence on ENSO, and they are not interchangeable. Huang et al., in Geophysical Research Letters on 26 March 2026, used a neural network and pacemaker simulations, in which Atlantic SST is imposed in a model. They report that the tropical Atlantic in summer, autumn and winter adds to ENSO predictability, while its value in spring is limited, through a summer-initiated pathway in which warm SST near the Atlantic ITCZ strengthens convection. A study by Ma et al. in The Innovation, described in a November 2 press release, reports that the link between the summer Atlantic Niño and the following winter’s ENSO switches on and off with SST gradients in the tropical Atlantic and Pacific. The release says the link was significant in 1894 to 1925 and 1984 to 2009 and not in 1928 to 1981.
The three differ in season, lead time, region and method, and we read only the abstract of one and the press release of the other. They cannot yet be said to confirm each other. The Ma et al. result is a reminder that Atlantic-Pacific links can come and go over decades.
Nothing in the abstract refers to 2027, and the relationship cannot be applied yet, because the winter it concerns has not finished. CPC’s November 12 discussion says El Niño will continue to strengthen before beginning to weaken in early 2027. Its official probabilities, issued in November, favor El Niño in March to May 2027 at 80 percent, then ENSO-neutral at 61 percent in April to June and 75 percent in May to July. The table ends with June to August 2027, at 61 percent neutral and 33 percent La Niña. Those are CPC’s calendar-season probabilities, and they are not tied to the paper’s “winter” and “summer,” which we could not confirm. If the Atlantic signal proves useful, it would enter forecasting as one more input, weighted by the numbers in the methods.
This piece makes no forecast. It reports one paper’s abstract and lists how the claim could be tested.
Checkpoints: our follow-up once the full methods are read (region, months, years, correlation, and whether ENSO’s own state is accounted for); NOAA CPC’s ENSO diagnostic discussion and updated probabilities on 10 December 2026; the end of winter 2026-27, when the paper’s TSA index can be computed once its definition is confirmed; and the following summer, when any Pacific tendency can be compared with the observed ENSO-related SST anomaly. One year cannot confirm or refute a tendency fitted over many years, and the paper’s own measures of strength, which we have not read, would define what counts as consistent with it.
No. According to its abstract, the paper reports that a warm tropical southern Atlantic SST anomaly in winter favors cold ENSO-related SST the following summer, and that a cold anomaly favors warm. The winter it would apply to, 2026-27, has not finished. CPC's November probabilities favor El Niño in March to May 2027 at 80 percent, with ENSO-neutral at 61 percent in April to June and 75 percent in May to July.
According to the abstract, a warm tropical southern Atlantic anomaly can sustain to the following summer and induces an anomalous Walker circulation spanning the Atlantic and Pacific, which triggers an anomalous low-level easterly over the tropical Pacific. Zonal advection, Ekman pumping and the Bjerknes thermocline feedback then produce the ENSO-related SST anomaly. The cold case corresponds to the opposite circulation anomaly, and a model simulation is said to support the proposal.
Not yet. The abstract calls it a potential new predictor. A predictor has to add information beyond ENSO's own state and hold up on years not used to find it. We have not read the full methods, so we cannot say how strong the relationship is or how the paper handles either test.
CPC's next ENSO discussion and updated probabilities are due on December 10, 2026. The paper's definitions of winter and summer need to be confirmed before its relationship can be checked against the 2026-27 winter and the summer that follows. We will update this piece once the paper's methods have been read.