NOAA's Climate Prediction Center puts a 69% probability on October through December 2026 producing the strongest El Nino since 1950. The threshold it named is a three-month RONI value of +2.5C. That acronym became the agency's operational definition of ENSO phase only in February 2026, when it replaced the Oceanic Nino Index. The substitution does more than relabel this event. It reverses the ranking of the three modern super El Ninos an actuary would reach for as analogs.

What the August Advisory Actually Puts at 69%

The August 13 discussion carries an El Nino Advisory and a greater than 90% chance of a very strong event through Northern Hemisphere fall and winter 2026-27. The 69% figure is the narrower claim: the probability that the October-December season registers a three-month RONI of +2.5C or higher, which would exceed every El Nino in the record back to 1950.

The event is running early. Nino3.4 crossed the strong threshold in June, and the weekly value centered on August 12 reached +2.7C, months ahead of the November-to-January window where these events normally peak. The IRI/CPC Quick Look issued August 19 carries El Nino at 100% probability for every three-month season from August-October 2026 through February-April 2027, 97% for March-May and 78% for April-June 2027. Fifteen models put Nino3.4 at or above +3.0C, which IRI describes as beyond its highest defined intensity category.

Read on its own, the +2.5C threshold looks arbitrary. It is not. On CPC's own seasonal RONI series the highest value in the record is +2.4C, set in December-February 1983. The agency placed the bar one tenth of a degree above the standing record.

El Nino seasonPeak ONIPeak RONI
1982-83+2.1 (NDJ, DJF)+2.4 (DJF), record on RONI
1997-98+2.4 (NDJ)+2.3 (OND, NDJ)
2015-16+2.6 (NDJ), record on ONI+2.3 (NDJ)
2026-27 (forecast)15 models at or above +3.069% chance of +2.5 or higher in OND

Both columns are CPC three-month seasonal series in degrees Celsius, so they are directly comparable. Monthly peaks quoted elsewhere run higher on both indexes.

Why the Index Change Reorders the Analog Set

Both indexes start in the same place: sea surface temperature anomalies in the Nino 3.4 region. ONI measures that anomaly against a centered 30-year base period updated every five years. RONI takes the same anomalies, subtracts the average anomaly across the tropical belt, and adjusts variance.

CPC's stated reason for the switch is that ONI "relies on a departure from 30-year average that struggles to keep pace with anomalous changes in tropical sea surface temperature." RONI, the announcement says, "solves this problem by comparing the ENSO region to the global tropics, thereby reducing the dependency on the climate base period."

Now look at what that does to the table. On ONI the three modern super El Ninos rank 2015-16, then 1997-98, then 1982-83, which is strictly newest to oldest. A ranking that is monotonic in time is the signature of a trend leaking into the measurement, not of three independent events happening to fall in chronological order. On RONI the order inverts and 1982-83 takes the record.

That matters because ENSO-conditioned views of risk are usually built by analog. You select the historical seasons whose index value sits nearest the forecast, then weight their landfall counts, loss experience or yield shortfalls into the current year's view. Rank that selection on ONI and the analog set skews toward recent seasons by construction, because the index credits four decades of background ocean warming to the event. Rank it on RONI and the nearest analog moves back to 1982-83: a season with a different exposure base, different building stock, and a reinsurance market that predates the modern cat bond entirely.

The forecast itself reads differently by index. Zeke Hausfather's July analysis puts the multi-model median for this event at 3.6C on the raw Nino 3.4 basis, far outside anything observed, while on RONI a record appears in 11 of 14 models: a record, but adjacent to 1982-83 rather than off the chart. "In a warming world, the raw Nino 3.4 anomaly risks conflating El Nino with the broader ocean warming trend," he writes. An actuary loading for an unprecedented event and one loading for a repeat of 1983 plus a margin are not making the same decision, and right now the two indexes support one each.

The Downstream Models Still Speak ONI

RONI became the operational standard seven months ago. Nothing that consumes it was rebuilt on that schedule.

Seasonal hurricane forecasts, crop yield models and the analog tables inside conditioned catastrophe views were fit on Nino3.4 anomalies, or on analog years chosen by ONI rank. There is now a gap between the index the agency publishes as the official definition of ENSO phase and the index the consuming models were calibrated against. For a directional diagnostic that gap is tolerable. Where an index is named as a defined term, in contract wording or in a parametric trigger, a publishing agency retiring it stops being a meteorological question and becomes a construction one.

The physics runs in RONI's favour, which is the uncomfortable part. Tropical atmospheric circulation responds to the temperature gradient between the central Pacific and the surrounding tropics, not to the absolute anomaly in one box. That is the same reasoning CPC gave for the switch. It also means the headline Nino3.4 figure overstates how anomalous the atmospheric response should be relative to past events, including the Atlantic wind shear that suppresses hurricane development.

That lands on a market which has already priced a quiet Atlantic. The Guy Carpenter US property catastrophe rate-on-line index fell 14% at the April 2026 renewals after a 12% decline at January, and June renewals produced risk-adjusted declines of 15% to 20%, with steeper cuts at more remote attachment points. The suppression signal this season is real and large. Whether it is unprecedented depends on which of two indexes you read it from, and the one that says unprecedented is the one NOAA stopped using in February.

Further Reading

Sources