A 69% Chance of “Historic.” The +2.6 You Remember Is the Wrong Index.

CPC issued its monthly ENSO Diagnostic Discussion on 13 August 2026. The alert status is unchanged at El Niño Advisory. The synopsis is not unchanged: CPC now gives a greater than 90% chance of a very strong event during Northern Hemisphere fall and winter 2026-27, and adds a 69% chance of a historic event during October to December.

CPC defines that word rather than leaving it to the reader. A historic event is one that “would exceed the strength of previous El Niño events dating back to 1950,” and the bar is stated as +2.5°C or more for a 3-month RONI value. That is a specific, checkable claim about a specific published series, which is what makes it worth checking.

The Trap in the Number

+2.5 looks like a modest bar if you remember that the 2015-16 El Niño peaked at about +2.6. It reads as though CPC has set the threshold below the record. It has not, and the reason is the same index confusion this site was built to track.

The +2.6°C figure attached to 2015-16 in almost every reference is a conventional ONI value. NOAA’s ONI series puts that event at +2.59°C in November-December-January 2015. CPC’s new threshold is stated on the relative index, RONI. Per the NWS statement that implemented it on 1 February 2026, RONI subtracts the average SST anomaly of the global tropics (20°N to 20°S) from the Niño 3.4 anomaly, then rescales the result to match the amplitude of the traditional index. The two indices do not return the same number for the same ocean. When Niño-3.4 warming is partly shared across the tropical oceans, RONI generally reads lower than the conventional ONI. It is not a rule that holds in every case, and the table below is itself the counter-example: in 1982-83 the relative value came in above the conventional one.

Put both series side by side for the three biggest events on record and the picture is not a small correction. IAMElNino.com analysis, computed from the two CPC files linked below on 13 August 2026.

The ranking reverses completely. By the conventional index the order is 2015-16, then 1997-98, then 1982-83. By the relative index it is 1982-83, then 1997-98, then 2015-16. The event that holds the record on one series finishes last of the three on the other. Neither series is wrong. They are answering different questions, and only one of them is the question CPC asked this morning.

What +2.5°C Would Actually Clear

CPC’s threshold applies to October to December specifically, so the relevant comparison is the October-December value in each year, not each event’s peak. On the relative index the strongest October-December on record is +2.25°C, set in 1997, with 1982 a hundredth behind at +2.24°C and 2015 at +2.18°C.

So a +2.5°C October-December in 2026 would sit 0.25°C above the October-December record. It would also sit 0.10°C above the highest three-month RONI value ever recorded in any season, which is +2.40°C in December-January-February 1983.

That is the substance behind CPC’s wording. A reader who anchors on +2.6 and concludes the bar is easy has the sign of the comparison backwards by a quarter of a degree.

What the Discussion Actually Reports

The observational half of the discussion is straightforward. CPC gives July monthly index values of +1.4°C in Niño-3.4, +1.7°C in Niño-3 and +2.9°C in Niño-1+2, and notes sea surface temperature anomalies exceeding +2.0°C in the eastern equatorial Pacific. The equatorial subsurface temperature index, averaged from 180° to 100°W, increased during July, and CPC states that subsurface anomalies reached +10.0°C at depth.

That last number is already travelling in headlines without its qualifier. It is a subsurface value at depth, not a sea surface temperature. Nothing at the ocean surface is 10°C above normal. CPC puts eastern equatorial Pacific surface anomalies as exceeding +2.0°C in the same paragraph, so reading the subsurface figure as a surface anomaly would turn a localized measurement at depth into a several-fold exaggeration of the surface warming.

On the atmospheric side, CPC reports low-level westerly and upper-level easterly anomalies across the western to east-central equatorial Pacific, convection enhanced from the central to eastern Pacific and suppressed over Indonesia, and both the traditional and equatorial Southern Oscillation indices significantly negative at near two standard deviations during July. That is a coupled system, not a warm ocean on its own, which is the distinction the ocean versus atmosphere piece sets out.

The Comparison the Synopsis Hides

CPC’s 9 July discussion put an 81% chance of a very strong El Niño during October to December. Today’s synopsis reads “greater than 90% chance of a very strong event during the Northern Hemisphere fall and winter 2026-27.” Those two sentences are not the cleanest comparison available, because the second one covers several seasons at once.

The season-by-season numbers behind both sit in CPC’s archived strength tables, and they are directly comparable. In that product, very strong means a RONI value at or above +2.0°C, using the 1991-2020 base period over the Niño-3.4 region. IAMElNino.com analysis, read from CPC's July 2026 and August 2026 archived strength tables on 13 August 2026.

So the October-December very-strong probability moved from 81% to 95% in one cycle, and the outlook strengthened across every season through JFM. That is the like-for-like comparison, and it is a clearer statement than the synopsis.

It also shows why the synopsis is shorthand. “Greater than 90% during fall and winter” holds for SON, OND and NDJ, but the table puts December-February at 70% and January-March at 35%. The seasonal table is the more precise document, and where the two differ, the table is what CPC actually forecast.

One further point about the 69%. The strength table’s top category is index at or above +2.0°C, and it is open-ended, so a +2.5°C outcome would still fall inside it and still be classified very strong. The table publishes no finer split above +2.0°C, which is why the 69% cannot be read off it and appears only in the discussion text. The 95% is the probability of landing anywhere in that open-ended band; the 69% is the probability of landing in a narrower region within it. They are answering different questions, and neither substitutes for the other. Corrected 15 August 2026: an earlier version described the +2.5°C threshold as sitting above the table’s highest bin. It sits inside it. See the corrections log.

How Far There Is Still to Go

The most recent published seasonal RONI is +0.98°C for May to July 2026, the first season to clear the +0.5°C numerical threshold, having read +0.49°C in April to June. Clearing that number is not the same as classifying an episode. Under the NWS statement that implemented RONI, historical events are categorised only where there are five or more consecutive overlapping three-month seasons above the threshold. Reaching +2.5°C in October to December therefore requires a rise of about 1.52°C across roughly five overlapping seasons.

It is worth asking how that compares with what past events actually did over the same stretch. IAMElNino.com analysis, computed from RONI.ascii.txt on 13 August 2026.

The rise required is larger than any of these five events managed over the same span. The steepest was 1982 at +1.40°C, and 2026 would need +1.52°C, about a tenth of a degree more, from a May-July starting point 0.22°C below where 1997 stood at the same season.

Two cautions on reading that. Five events is a small sample and says nothing about what is physically possible, only what has been observed since 1950. And the required figure moves if the May-July value is revised, which it still can be. This is a statement about historical rates of change, not a forecast, and it does not contradict CPC: a 69% probability already carries a 31% chance the threshold is not reached. It does show that the outcome CPC is describing would be unusual in its rate as well as its magnitude.

What to Watch

• Which index a headline is quoting. If a figure near +2.5 or +2.6 appears without the word relative or conventional attached, it cannot be placed against CPC’s threshold.
• The weekly relative Niño 3.4 value, published each Monday in rel_wksst9120.txt. It moves first and is the earliest sign of whether the seasonal index keeps climbing.
• Niño-4. CPC did not quote it this month. The eastern regions are carrying this event, and if the dateline region stays near neutral the structure remains sharply east-based, which changes the teleconnections rather than just the headline number.
• The 10 September discussion, for whether the 69% holds, rises or falls.
• Revisions. Recent RONI seasons are provisional for up to two months. A record claim resting on a first-posted value is resting on a number that can still move.

The Short Version

CPC raised its outlook and defined its terms: historic means +2.5°C on the three-month relative index in October to December, and it gives that 69% odds. Its season-by-season table separately lifts the very-strong probability for October-December from 81% to 95%. On the relative index the October-December record is +2.25°C from 1997 and the highest value in any season since 1950 is +2.40°C from early 1983, so the historic threshold clears both. The +2.6°C that most readers associate with 2015-16 belongs to a different index, on which that event ranks first, while on the relative index it ranks third of three. The bar is a quarter of a degree higher than it looks, and the value that settles it does not publish until 2027.