China reported +2.09°C for July while NOAA reported +1.4°C for Niño 3.4. China’s preliminary first-half August reading then reached +2.64°C. The numbers differ because the agencies are not publishing the same index, and their two +2.5°C thresholds turn out to be doing entirely different jobs.
On 18 August, China’s National Climate Centre told state media that sea surface temperatures in the equatorial central-eastern Pacific are expected to peak around November or December, forming a super-strong El Niño event. China’s National Climate Centre says the event will probably become the strongest in its observational record.
That is a genuine forecast from a major national climate agency, not a rewrite of NOAA. It also arrives with numbers that look nothing like NOAA’s. Reading them side by side without checking the definitions would leave you thinking one agency is badly wrong. Neither is.
Per the China News Service report of 18 August, citing the National Climate Centre:
One thing the announcement does not do is name the monitoring region. It says “监测区海温指数”, the monitoring-region sea temperature index. For that we have to go to the standard.
The +2.64°C is a first-half-of-August value. The threshold it is being measured against is a three-month running mean. Those are not the same quantity, and a half-month reading cannot cross a seasonal threshold any more than one warm afternoon can raise a monthly average on its own.
This is the same discipline that applies to NOAA’s weekly figures. A weekly or fortnightly value moves first and moves furthest; the seasonal index it feeds lags it by design. China’s own announcement is careful about this, attaching the +2.5°C threshold explicitly to the three-month running mean at the expected November-December peak, not to the August reading. Coverage that reports “China says 2.64, threshold is 2.5, therefore threshold crossed” would be inventing a conclusion the agency did not draw.
China’s national standard is GB/T 33666, Identification Method for El Niño/La Niña events, drafted by the National Climate Centre and the Chinese Academy of Meteorological Sciences. It is public. Three things in it matter here.
First, the region. The standard identifies events on the NINO3.4 index, defined as the mean SST anomaly over 170°W-120°W, 5°S-5°N. That is the same box NOAA uses.
Second, the event rule. An event is declared when the absolute value of the three-month running mean of NINO3.4 reaches or exceeds 0.5°C and persists for at least five months. For retrospective classification of a full episode, NOAA likewise uses at least five consecutive overlapping three-month seasons beyond the threshold. Its real-time advisory procedure does not wait for all five seasons to be observed.
Third, and this is where NOAA’s current operational index parts company with China’s conventional index, the construction. RONI subtracts the tropical-mean SST anomaly and then rescales the result. Both China’s index and NOAA’s legacy ONI use changing 30-year climatologies, although their exact implementation and operational data products may differ. The Chinese standard sets out its own schedule explicitly: anomalies for 1950-1990 computed against 1951-1980, those for 1991-2000 against 1961-1990, and so on, with the document naming 1981-2010 for the period from 2011.
Worth noting what they share, and where that similarity stops. From 1982 onward the Chinese standard specifies NOAA’s own OISST product as its input data, with HadISST before that. NOAA has since moved its historical ONI and RONI tables to ERSSTv6, while some of its weekly and monthly monitoring products still run on OISST. So the systems draw from closely related observations of the same region, including NOAA-produced SST datasets, but they do not necessarily use the same dataset, climatology, averaging period or index construction. The difference between relative and conventional indices is set out in full elsewhere on this site.
Here is the part that matters most, and it is not visible from either agency’s announcement alone.
One wording note before the point, because it matters for a boundary. The version of GB/T 33666 available on the National Climate Centre’s server is the approval draft, and it phrases every grade as “reaching or exceeding” its floor, which would leave strong and super-strong overlapping at exactly 2.5. Published descriptions of the standard resolve this as strong running 2.0 through 2.5 and super-strong above 2.5. This article uses the latter. Since the index is reported to one decimal place, the first value that would display as super-strong is +2.6°C.
Both headline thresholds are +2.5°C. In China’s system that number is the floor of the top category. GB/T 33666 grades events on the peak three-month running mean of NINO3.4: weak from 0.5, moderate from 1.3, strong from 2.0 through 2.5, and super-strong above 2.5. Exceeding +2.5°C makes an event super-strong. It does not make it a record.
In NOAA’s system, +2.5°C on RONI is not a category floor at all. CPC introduced it in its 13 August discussion as the level that would exceed every El Niño back to 1950, and gave that outcome 69% odds for October-December. NOAA’s own top published category, very strong, starts at +2.0°C and is open-ended.
So the same figure means “qualifies for the top tier” in one system and “beats everything ever measured” in the other.
This becomes concrete when you look at how each system scores the three past super-strong events. The Chinese figures are from Appendix A of the standard; the NOAA figures are from CPC’s published index files, read on 13 August 2026.
| Event | China, NINO3.4 peak changing 30-yr climatology | NOAA ONI peak conventional | NOAA RONI peak relative |
|---|---|---|---|
| 1982-83 | +2.7 | +2.14 | +2.40 |
| 1997-98 | +2.7 | +2.37 | +2.28 |
| 2015-16 | +2.8 | +2.59 | +2.25 |
IAMElNino.com analysis, compiled from GB/T 33666 Appendix A and NOAA CPC’s ONI and RONI seasonal files as read on 13 August 2026. Bold marks the record holder on each index. The three columns use different datasets, climatologies and index constructions, and are set side by side to show how far those choices move the answer, not to be read as competing estimates of one quantity.
Three consequences follow.
The record to beat is not the same record. On China’s index the mark is +2.8°C, set in December 2015. On NOAA’s RONI it is +2.40°C, set in the winter of 1982-83. A 2026 peak of, say, +2.6°C on China’s index would be super-strong but not a record; the same event might or might not set a RONI record depending on how the two indices diverge this year.
The ranking is not the same either. China’s index and NOAA’s conventional ONI both put 2015-16 on top. NOAA’s relative index puts 1982-83 on top and 2015-16 last of the three. We covered why that reversal happens in the piece on what a historic event would have to beat.
The spread between systems is larger than the margin being forecast. For 2015-16 alone, China’s +2.8 and NOAA’s RONI +2.25 differ by 0.55°C. NOAA’s historic threshold sits only 0.10°C above its own record. That is a comparison of scales, not a statement about forecast uncertainty: it simply means the choice of index moves the number by more than the margin that decides whether a record falls. The index label is not a footnote.
It would be wrong to leave the impression that this is two countries talking past each other. The agreement is substantial and worth stating plainly.
The disagreement is not about the ocean. It is about which yardstick is laid against it.
China’s claim is precise within its own framework: strongest in its observational record, on its index, which means beating +2.8°C. NOAA’s 69% is precise within its framework: exceeding every event in its RONI record, which means beating +2.40°C.
Neither statement is a claim about the other. A headline that says only “strongest El Niño ever” has dropped the one piece of information that makes it checkable, and it is entirely possible for this event to end up setting a record on one index and not the other. That is not a paradox. It is what happens when two indices with different baselines and different treatments of background warming are asked to rank the same ocean.
China’s National Climate Centre expects a super-strong El Niño peaking around November to December and probably the strongest in its record. NOAA gives 69% odds that October-December exceeds every event since 1950 on its relative index. Both headline a +2.5°C threshold, but in China’s system that is the boundary of its top category, which all three past super-strong events already exceeded at +2.7, +2.7 and +2.8, while in NOAA’s it would clear a record that currently stands at +2.40°C. The two agencies agree the event is strengthening, agree roughly on timing, use a similar retrospective classification rule and draw on closely related observations. What they do not share is a yardstick, and the difference between their yardsticks is bigger than the margin either record turns on.