The four-week plateau at +1.8°C ended in the week centred 9 September: Niño 3.4 at +2.0°C and Niño 3 at +2.8°C, while Niño 1+2 eased to +3.7°C. CPC had already moved two probabilities upward on 10 September, raising the October to December historic-event chance to 75 percent and the strength table's very strong figure to 98 percent. The threshold is a three-month RONI of +2.5°C, and that index rose 0.39°C between the last two overlapping seasons.
In its 10 September Diagnostic Discussion, NOAA's Climate Prediction Center raised its headline probability of a historic OND El Niño from 69 percent to 75 percent. At that point the weekly index most coverage quotes had not moved in a month: relative Niño 3.4 came in at +1.8°C in the weeks centred 12 August, 19 August, 26 August and 2 September.
Four days later the plateau broke. The week centred 9 September, published on 14 September, put Niño 3.4 at +2.0°C, the highest weekly relative reading of this event.
The order matters. CPC raised the probability before the weekly index moved, which means the case for the increase never rested on the weekly number. Working out what it did rest on turns on which index the threshold is stated on, and on how much of the evidence for a strengthening event sits somewhere other than a weekly surface reading.
| CPC product | 13 August 2026 | 10 September 2026 |
|---|---|---|
| Alert status | El Niño Advisory | El Niño Advisory |
| Historic event, OND (RONI ≥ +2.5°C) | 69% | 75% |
| Strength table, OND (RONI ≥ +2.0°C) | 95% | 98% |
| Synopsis, very strong event, fall and winter | greater than 90% | greater than 90% |
The September strength table is worth reading in full for that season, because the distribution has collapsed onto two bins. For October to December 2026 it puts 98 percent at a RONI of +2.0°C or above, 2 percent in the 1.5 to below 2.0°C bin, and zero in every other bin, including every negative one. The same issuance gives 93 percent at or above +2.0°C for November to January and 75 percent for December to February.
The alert status is unchanged and the synopsis wording on a very strong event is unchanged. Two probabilities moved, both upward, and they answer different questions: the strength table asks whether OND clears +2.0°C, and the discussion narrative asks whether it clears +2.5°C. This is a revision inside a forecast already pointed in the same direction, not a change of diagnosis.
CPC states what it means by historic rather than leaving it to the reader. Such an event "would exceed the strength of previous El Niño events dating back to 1950," and the bar is "+2.5°C or more for a 3-month RONI value."
That sentence rules out three comparisons that show up regularly in coverage. It is not a weekly value, since RONI is a three-month average and a single week cannot satisfy or fail it. It is not the traditional ONI, which is a different series computed from the same region. And it is not a forecast stated on some other baseline, because a peak projection computed on a different index or climatological period cannot be laid against +2.5°C without conversion, which this article does not attempt.
So when the question is whether the threshold is getting closer, the place to look is the RONI series itself.
Both published seasonal series advanced with the latest release. The RONI rose from +0.97°C in May to July 2026 to +1.36°C in June to August 2026. The traditional ONI rose from +1.39°C to +1.80°C across the same two seasons.
This is the first part of the answer. The index used for the threshold rose 0.39°C between the latest two overlapping seasons. A weekly Niño 3.4 value that holds still for four weeks is compatible with that, because a three-month mean can keep rising by dropping an earlier month while adding a warmer one.
The 0.44°C difference between the JJA RONI and the JJA ONI is an IAMElNino.com subtraction of two published CPC values, not a CPC product. It reflects the combined effect of CPC's tropical-background subtraction and subsequent variance adjustment. It should not be interpreted as a direct measurement of the tropical-mean anomaly.
Reaching CPC's historic threshold would require RONI to rise another 1.14°C between JJA and OND. Across the 24 El Niño onset years examined by IAMElNino.com, only 1982 produced a larger JJA-to-OND increase, rising 1.35°C. That makes the forecast historically aggressive, even though CPC now assigns it a 75 percent probability.
| Region | Wk centered 12 Aug | Wk centered 2 Sep | Wk centered 9 Sep |
|---|---|---|---|
| Niño 1+2 | +3.4 | +3.9 | +3.7 |
| Niño 3 | +2.5 | +2.7 | +2.8 |
| Niño 3.4 | +1.8 | +1.8 | +2.0 |
| Niño 4 | +0.1 | +0.1 | -0.1 |
Through 2 September the gain was concentrated in the far east. Niño 1+2 added 0.5°C over those four weeks, Niño 3 added 0.2°C, and Niño 3.4 and Niño 4 did not move. "El Niño stalled" only survived if Niño 3.4 was the only column you read.
In the week centred 9 September that pattern changed direction. Niño 3.4 added 0.2°C to reach +2.0°C, Niño 3 added 0.1°C to +2.8°C, Niño 1+2 gave back 0.2°C to +3.7°C, and Niño 4 slipped to -0.1°C. The centre of the basin caught up while the far east eased slightly.
One qualification belongs here immediately. A single weekly row is a short and revisable sample. CPC revises these values, and the same table shows Niño 1+2 dipping from +3.4°C to +3.2°C in mid-August before resuming its climb; reading that week as a turning point would have been wrong. What the 9 September row establishes is that the plateau in Niño 3.4 ended. It does not establish that a new direction has been set.
The 1+2 minus 3.4 difference reached 2.1°C in the week centred 2 September, the widest of this event, then narrowed to 1.7°C a week later. Those subtractions are ours, from published CPC values in the same rows. The regional averages establish an east-heavy structure that is now somewhat less east-heavy, although they cannot locate the warming precisely within the overlapping boxes.
And one piece of index discipline, because the coincidence invites the error: the +2.0°C in this weekly table is not the +2.0°C in CPC's strength table. The weekly figure is a single week of relative Niño 3.4. CPC's very strong category is a three-month RONI. Two different instruments that happen to share a number this week.
CPC reported that the equatorial subsurface temperature index, averaged from 180° to 100°W, held steady through August, reflecting a deeper than average thermocline, and that subsurface temperatures continued to be significantly above average with widespread anomalies in excess of +10.0°C at depth.
The subsurface pattern provides favorable support for additional surface warming, but it does not guarantee that the heat will reach the surface or that OND RONI will exceed +2.5°C. Whether it surfaces depends on the winds.
The September discussion lists the atmospheric evidence plainly, and the items reinforce each other rather than standing alone. Low-level westerly wind anomalies and upper-level easterly wind anomalies extended from the western to the east-central equatorial Pacific. Enhanced convection and rainfall persisted from the central to the eastern Pacific, while anomalies remained suppressed over Indonesia. Both the traditional and the equatorial Southern Oscillation indices remained negative. CPC's own summary is that "the coupled ocean-atmosphere system reflected a strengthening El Niño."
Coupling is what separates a warm patch of ocean from an El Niño event. Warm water in the central and eastern Pacific shifts convection eastward, the shifted convection weakens the trade winds, and weakened trades reduce upwelling and let warm water extend further east, warming the surface further. The negative Southern Oscillation indices are the surface pressure signature of the same rearrangement, measured independently of the SST field.
That closed loop is the second part of the answer. A forecaster raising a probability is not only extrapolating a surface curve. They are weighing whether the mechanism that intensifies these events is running, and on CPC's account it is.
It establishes that one region's surface anomaly did not increase over four particular weeks. That is all.
Weekly values are noisy and revised. Within the same stretch Niño 1+2 fell from +3.4°C to +3.2°C before resuming its climb, and reading that week as a turning point would have been wrong. Niño 3.4 is the benchmark region used for RONI, but it is not a summary statistic for the entire equatorial Pacific.
None of those observations alone establishes that the threshold will be reached. They do make four flat weekly Niño 3.4 readings weak evidence against CPC's forecast.
The forecast is falsifiable, and these are the observations that would argue against it. Declining subsurface heat content, particularly a shoaling thermocline in the 180° to 100°W band, would remove the support the current guidance leans on. A return of easterly trade wind anomalies across the central Pacific would restore upwelling and cut the mechanism described above. Weakening atmospheric coupling would appear as convection retreating westward, convection over Indonesia recovering toward normal, or the Southern Oscillation indices moving back toward zero, any of which would indicate the loop losing its grip regardless of the SST field. Cooling that begins near the dateline and spreads eastward into Niño 3 and Niño 1+2, rather than the present pattern of warmth concentrated in the east, would mark a change in structure rather than a pause. And a material downward shift in the October model guidance would be the clearest single signal, since the published probability is a forecaster judgement informed by that guidance.
Compiled 14 September 2026. Weekly Niño region values are relative anomalies from CPC's rel_wksst9120.txt, the index CPC's ENSO products have quoted since February 2026, current through the week centred 9 September 2026, published 14 September. Seasonal RONI and ONI values are as published on 14 September 2026. RONI is computed from monthly ERSST fields, following CPC's August 2026 migration from ERSSTv5 to ERSSTv6, and is not built from the weekly observations quoted above.
The May to July RONI reads +0.97°C in the current file, having previously been published as +0.98°C. That is an ordinary revision of the kind CPC notes can occur for up to two months after posting.
Probabilities come from two distinct CPC products, and this article keeps them separate. The 75 percent historic-event figure and the greater than 90 percent very strong statement are from the ENSO Diagnostic Discussion issued 10 September 2026, with the August comparison from the discussion issued 13 August 2026. The 98 percent and 2 percent OND figures are from the September 2026 issuance of CPC's ENSO strength probability table, read from the dated archive; the August issuance of the same table gives 95 percent for the same row and column.
Two differences here are IAMElNino.com subtractions of published CPC values rather than CPC products, and are labelled as such in the text: the 0.44°C RONI to ONI gap for JJA, and the Niño 1+2 minus Niño 3.4 differences of 2.1°C for the week centred 2 September and 1.7°C for the week centred 9 September. The JJA-to-OND climb comparison is also ours, computed from the published RONI series across 24 El Niño onset years, not across every year in the file.