Cold water really is sitting in the eastern equatorial Atlantic, and it really would matter if it holds. But the convention for an Atlantic Niña is an ATL3 anomaly at or below -0.5°C across two consecutive overlapping three-month seasons, and no such pair has closed. In about 48 hours in mid-July, a conditional forecast turned into a flat statement of fact, and then into an "atmospheric shield." Six days after that, Bertha made landfall in Louisiana.
There is a real thing happening in the Atlantic. That should be said first, because what follows is a correction and corrections can read as dismissals.
Sea surface temperatures in parts of the eastern equatorial Atlantic have been running well below average since early July. If that persists, it will matter for West African rainfall, for northeastern South America, and plausibly for the back half of the Atlantic hurricane season. It is worth watching and it is under-covered relative to the Pacific.
What is not yet true is that the recent cooling has satisfied the duration requirement used to classify an Atlantic Niña event, that its rarity is established, or that it has combined with El Niño to build a shield over the United States. Those three claims all entered circulation in the same week, and each is a step further from the source than the last. It is worth noting that unlike ENSO there is no official alert or declaration system for the Atlantic zonal mode, so "declared" is the wrong frame entirely; what exists is a convention, and the convention has a duration clause.
It is not La Niña. This is the first thing most coverage gets right and the first thing most readers get wrong, so it is worth stating plainly. La Niña is the cold phase of ENSO in the Pacific. An Atlantic Niña is the cold phase of the Atlantic zonal mode, a separate pattern in the eastern equatorial Atlantic that swings warm and cold every few years on its own schedule.
The physics are seasonal and specific. As NOAA's explainer by Franz Philip Tuchen of CIMAS lays out, the eastern equatorial Atlantic has a counterintuitive annual cycle: its warmest water arrives in spring and its coolest, below 25°C, arrives in July and August. That summer cooling happens because the tropical rain band migrates north, pulling steady southeasterly winds across the equator, which drag surface water away and draw colder water up from below. The result is a cold tongue that forms every single summer.
So cold water in the equatorial Atlantic in July is not itself news. It is the climatology. The question is whether this year's cold tongue is anomalously cold, by enough, for long enough.
The usual answer is a threshold with a duration attached: three-month averaged anomalies at or past ±0.5°C for at least two consecutive overlapping seasons, measured over the ATL3 box, 20°W to 0° and 3°S to 3°N. Definitions are genuinely not standardised across research groups, so that number is a convention rather than a rule, but it is the common one and it is the one NOAA's own explainer uses.
Two consecutive overlapping seasons. As of today, June to August has not closed and July to September has barely started. The earliest this can be settled is autumn, and even then a JJA miss would not end it, because a later pair such as JAS and ASO could still qualify.
One more distinction that keeps getting lost. The reported cold anomalies of 1 to 3°C are local values read off an anomaly map. ATL3 is a three-month mean averaged across the whole box. A patch of water can run several degrees below normal while the box mean sits nowhere near -0.5°C. The map and the index are not the same measurement, and only one of them decides anything.
The rarity figure traces to one place: Andrej Flis, a senior meteorologist at Severe Weather Europe, on 16 July. His wording, quoted directly in subsequent coverage, was conditional twice over:
"Atlantic Niña events are not unprecedented, but strong summer events are relatively rare. If the current cooling persists and the seasonal anomaly ends below -0.5°C, the 2026 event would become only the sixth such Atlantic Niña in more than 40 years of historical data."
Read it carefully and it is a more careful sentence than what followed. It names the threshold, it names the condition, and it says "would become," not "is."
It is still not a verifiable statistic. The wording is ambiguous about what is being counted: "such" could mean strong summer events specifically or Atlantic Niñas generally, and no event list, index series, dataset, precise period or treatment of borderline seasons is supplied. That ambiguity should not be resolved charitably on the claim's behalf. The right label is unverified, not false and not established. Until someone reproduces the count from a defined ATL3 series, it is a number without a method, and this site's standing rule is that a statistic needs a source and a date before it is worth repeating.
Gizmodo's article on the same day handled the body text correctly, writing that if temperatures "remain 0.9 degrees Fahrenheit (0.5 degrees Celsius) below average for at least two overlapping seasons, this anomaly will become the sixth Atlantic Niña of the last 40 years." That is a fair rendering.
The headline deck on the same article reads: "This phenomenon has only occurred five times within the last four decades."
That is the version in the social card. It is the version that travels. The conditional is gone, and a forecast about a threshold has become a fact about a phenomenon. Nobody had to be careless for this to happen; the body was fine. The summary field is simply where nuance goes to die, and the summary field is what gets shared.
One number is doing a lot of load-bearing work across this coverage: during an Atlantic Niña, tropical cyclone counts are about 50 percent lower.
Lower than what, exactly, is the part that keeps falling off. The comparison is to an Atlantic Niño, the opposite warm phase, not to a normal year. Measuring one extreme against the other extreme produces a bigger gap than measuring either against climatology, which is the number a reader assumes they are being given. It is also attributed to Severe Weather Europe rather than to a study.
The composite also appears to come without a stated sample size, uncertainty range, or any control for ENSO state, which matters a great deal when the years being compared may differ in Pacific conditions too.
None of which means the direction is wrong. There is real peer-reviewed support that this axis matters: Kim and colleagues showed in Nature Communications in 2023 that Cape Verde hurricanes increase during Atlantic Niño. What that literature supports is modulation of the ITCZ, African easterly waves and Cape Verde-region genesis. The honest summary of the cold phase is therefore a possible statistical headwind for parts of Atlantic hurricane development, not a uniform suppression of the whole basin and certainly not protection of the United States. The warm-phase evidence is also stronger than the mirror-image assumption that the cold phase does the same thing in reverse with equal force. "Roughly half compared with the opposite phase, per a forecaster's composite" and "half as many hurricanes" are different statements, and only one of them is supported.
For how a similar substitution works in the Pacific, see our note on why the same box in the same week yields different numbers in different datasets.
Worth flagging because it will propagate. Gizmodo's piece places the Atlantic Niña "off the southeastern coast of Africa," twice, once in the body and once in a chart caption. Southeastern Africa faces the Indian Ocean. The eastern equatorial Atlantic sits off Africa's west coast, in the Gulf of Guinea sector, which is what the article's own first image caption correctly says.
A small thing, and easy to do. But readers building a mental map from that sentence will look at the wrong ocean.
The "atmospheric shield" framing spread between 16 and 18 July. On 22 July, Tropical Storm Bertha made landfall in St Bernard Parish, Louisiana, then made a second landfall near the Louisiana-Texas state line.
This does not refute anything. A suppressed season is a shifted distribution, not a closed basin, and one storm no more disproves suppression than one hot day proves a trend. We made exactly that argument ourselves in our piece on Bertha: the forecast is down to nine named storms, and the second one still hit the Gulf Coast twice, and both things are true at once.
What the six-day gap does illustrate is the cost of the word "shield." Suppression language that describes odds survives a landfall. Language that describes a barrier does not. NWS Director Ken Graham put the operational version of this more bluntly when NOAA issued its outlook: "It only takes one storm to make for a very bad season."
The underlying seasonal numbers, for reference, are genuinely quiet. NOAA's May outlook gives a 55 percent chance of a below-normal season and forecasts 8 to 14 named storms, 3 to 6 hurricanes and 1 to 3 major hurricanes, against an average season of 14, 7 and 3. Our earlier coverage of the CSU forecast and the early-season reality check goes through what that does and does not promise.
| Claim | Status |
|---|---|
| Local cold anomalies in eastern equatorial Atlantic | Observed, reported mid-July. Not the same as ATL3 |
| An Atlantic Niña has developed | Duration requirement not yet satisfied. No pair has closed |
| Would be the sixth in 40+ years | Unverified. No event list, dataset or period published |
| Cyclone counts about half | Versus Atlantic Niño, not versus normal. No stated sample or ENSO control |
| Basins perfectly aligned | Hypothesis. Cold-phase evidence weaker than warm-phase |
| El Niño suppressing the Atlantic | Well established, and separately supported by CPC |
The last row is the one worth holding on to. The Pacific case does not need the Atlantic case to stand up. CPC's July 9 discussion put an 81 percent chance on a very strong El Niño during October to December and a 97 percent chance of persistence through early spring 2027, and the relative weekly Niño 3.4 index reached +1.4°C in the week centred 22 July. El Niño-driven wind shear is a long-established suppressant with decades of literature behind it. Stapling a second, weaker, unconfirmed mechanism to it does not make the forecast stronger. It makes it easier to falsify.
Cold water off West Africa is a real signal and a good story. It does not need to be rare yet, and it does not need to be a shield, to be worth watching. For the wider pattern picture see our teleconnections dashboard, and for the Pacific side of the same season, how El Niño is already suppressing the Atlantic.