Will La Niña follow the record El Niño?
A strong El Niño spends the heat stored along the equator, and the deficit it leaves has often tipped the Pacific into La Niña the next year. For the southern U.S. that would mean the drought of the La Niña years coming back. No forecast model reaches the 2027–28 winter yet. This page watches the ocean's heat reservoir month by month and sets beside it what the models can see and what history says, each with its sample size and its track record, without blending them into one number.
The heat reservoir, live
Warm water volume and Niño-3.4, month by month
Top: the heat stored above the 20 °C layer along the equator (anomaly). It builds before an El Niño (recharge) and drains as the event peaks and decays (discharge). Bottom: the surface temperature of the Niño-3.4 box, on its own axis.
How fast the reservoir drained after every strong El Niño since 1980
| El Niño | Reservoir fullest | Peak WWV | August WWV | Niño-3.4 peak | WWV below zero |
|---|
What the reservoir has told us before
Spring warm water volume and the winter that followed
La Niña by spring reservoir level
| Feb–Apr WWV | La Niña | 95% interval |
|---|
How far ahead does the reservoir know?
What SEAS5 has known, start by start
Skill by lead, for forecasts started around an El Niño peak
Every start month: correlation by lead (hindcasts 1981–2016)
Did SEAS5 see La Niña coming after each strong El Niño?
La Niña probabilities, scored
When SEAS5 said X%, how often did La Niña come?
The independent check: the real forecasts, 2017 onward
The newest SEAS5 forecast, with its own record
Four lines of evidence, side by side
What a 2027–28 La Niña would mean
How many of these winters fell in each division's warm or cold third (trend removed)
How many of these winters fell in each division's wet or dry third (trend removed)
Why no single probability
Each line of evidence answers a slightly different question, on a different index, with a different and usually small sample. The models stop in spring 2027, and SEAS5's own record (line 4) can say only how much its late-winter reading has told about the winter after. The reservoir's clearest signal, its level after the peak, does not exist until February–April 2027. History has 5 to 14 cases depending on how the question is cut, and a leave-one-out test shows the plain counts would not have beaten climatology. A weighted average of the four would look precise, but there is no record of past blends to score it against, so this page does not make one. The page rebuilds with every new month of warm water volume, so the spring 2027 numbers will show here as they arrive.
Phases use the archetype library's rule: a winter's Dec–Feb MEI-ERA5 (MEI.v2 recipe with OLR) at +0.5σ or more is El Niño, at −0.5σ or less La Niña. Winters are named by their Dec–Jan years.