Reporting out today says this El Niño could push the global monthly average past 2°C of warming for the first time on record. That is a real possibility, it is worth taking seriously, and it is not a breach of the Paris Agreement. Here is the arithmetic, the lag that puts the test in early 2027, and the two very different numbers this coverage keeps setting side by side.
Bloomberg reported this morning that the developing El Niño may push the global monthly average temperature past 2°C above pre-industrial levels for the first time in the instrumental record. The figure comes from projections by Ben Kirtman, dean of the University of Miami's Rosenstiel School of Marine, Atmospheric, and Earth Science, who put the odds of a single month crossing that line in early 2027 at roughly 35 to 40 percent. The framing that travels from a story like that is "world passes 2°C." The framing that survives contact with the data is narrower, more interesting, and considerably more useful.
One qualifier before anything else, because it gets dropped fast: this is about a monthly mean. Individual days have already exceeded 2°C above pre-industrial, four of them in a row during February 2024 in the Copernicus record. No monthly average has.
Start here, because it is the most common error in this week's coverage. Two figures are circulating in the same stories: a forecast El Niño peak of 3.6°C, and a possible global temperature month of 2°C. Readers understandably assume the first is a bigger version of the second. It is not the same measurement at all.
The 3.6°C figure is a detrended monthly Niño 3.4 sea surface temperature anomaly. It comes from Zeke Hausfather's synthesis of 667 ensemble members across 14 seasonal forecast models, and it is the multi-model median forecast peak for that index, which covers a box in the tropical Pacific. It measures how strong El Niño is, and nothing else. For scale, the previous record in those terms is 2.75°C from 2015-16, with the 1877-78 event effectively tied at 2.73°C, so a 3.6°C peak would clear the record by about 0.8°C. Hausfather's models put the odds of a record event at about 91 percent in Niño 3.4 terms and 77 percent using RONI, which strips out the broader ocean warming trend.
The 2°C figure is global mean surface temperature across the entire planet, measured against the 1850-1900 pre-industrial baseline. Different area, different variable, different baseline. One is a regional index of a Pacific oscillation; the other is the headline number of climate change. Putting them in adjacent paragraphs without saying so is how a reader ends up thinking the planet is about to warm 3.6°C.
The Paris goals refer to the long-term increase in global average temperature above pre-industrial levels. The Agreement itself does not specify an averaging period or a dataset. Scientific assessments estimate that underlying warming over multidecadal periods, separating it from temporary variability such as ENSO: IPCC assessments generally operationalize this as a 20-year or 30-year average depending on the framework, with AR6 commonly using a 20-year mean and the earlier 1.5°C special report using a 30-year one. Either way, the target was never defined as a monthly or annual value. That is not a loophole invented to soften bad news; single months and single years swing on natural variability including ENSO and volcanic eruptions, so a short-window reading would measure noise as much as trend.
We have already run this experiment at the lower threshold. Individual months crossed 1.5°C during the 2023-24 El Niño, and calendar year 2024 finished above it, the first year on record to do so. Published attribution work put observed 2024 warming near 1.52°C and estimated the underlying human-caused warming level near 1.36°C. El Niño and other short-term climate variability helped lift the observed annual temperature above that underlying level. Those records were real and they mattered. They also did not constitute failure of the 1.5°C goal, and the researchers who reported them said so at the time.
So the honest sentence about a 2°C month is: it would be the highest monthly value ever recorded, it would be driven substantially by a natural oscillation riding on top of a warming trend, and it would not mean the world has passed 2°C in the sense the Paris Agreement means.
Here is what the jump requires. June 2026 came in at 1.39°C above pre-industrial in the Copernicus ERA5 record, the second-warmest June in that dataset, with a 12-month running average of 1.43°C. The warmest single month in ERA5 is December 2023 at 1.78°C, with February 2024 just behind at 1.77°C. Reaching 2.0°C therefore means clearing the ERA5 monthly record by roughly 0.2°C, and clearing the current running average by about 0.6°C.
That is a large jump, which is why the stated odds are 35 to 40 percent rather than a near certainty. It is also not an absurd one. Strong El Niño events can add several tenths of a degree to short-term global temperature variability when combined with other oceanic and atmospheric conditions, though the response is not proportional to the Niño 3.4 index alone: NOAA's rough historical relationship is on the order of 0.07°C of global surface warming per 1°C of Niño 3.4 warming, at a lag. Carbon Brief projects a central 2027 anomaly near 1.71°C, with a 90 percent range of roughly 1.49°C to 1.93°C, and considers a new annual record likely. If 2027 averages near that central estimate, a monthly value approaching 2°C is plausible, because monthly anomalies fluctuate around the annual mean and El Niño's global influence is expected to concentrate in part of the year.
Global temperature does not respond to El Niño instantly. El Niño increases the transfer of heat from the tropical Pacific to the atmosphere and reorganizes tropical convection and circulation, and the global surface response follows with a lag of roughly two to five months, with about three months a common estimate. This is not a fixed physical timer: different datasets, events and response variables yield different lags. Because strong El Niño events commonly peak near the end of the calendar year, their largest global temperature influence often appears during the opening months of the following year. Kevin Trenberth, commenting on Hausfather's analysis, put the biggest influence on teleconnections and global conditions about three months after a December peak.
That is why the test is early 2027, not this autumn. This event started from near-La Niña conditions in January, and by Hausfather's detrended daily Niño 3.4 measure, 2026 was running warmer in mid-July than either 1997 or 2015 was at the same point in the calendar, already near 2°C above its era-adjusted average. The 2026 calendar year has only a modest chance, around 28 percent by Hausfather's dashboard, of edging past 2024 as the warmest on record. The main event is next year.
None of this is a reason to shrug. The reason a 2°C month is even on the table is that the background trend has climbed high enough that a strong natural oscillation can carry a single month there. In 2016, at the peak of what was then the strongest El Niño on record, the same oscillation could not do that, because the baseline was lower. The threshold moved into reach because the trend rose, and the trend is the part that does not reverse when El Niño decays.
That is the useful thing to take from the headline: not that a line has been crossed, but that the distance to the line has shrunk. The monthly figure is a thermometer reading on a noisy day. The multi-decade average is the climate. Watching the first and calling it the second is how this coverage goes wrong, in both directions, and we covered how quickly that translation happens in our trace of the CPC bulletin through its syndication chain.
We wrote in June that El Niño did not cause that heat wave, and the same discipline applies here in the other direction: a record month would be genuinely significant and it would still not be the threshold it gets called. Track the indices behind the forecast on our teleconnections dashboard, and see how the models producing these numbers actually work in our guide to ENSO forecast systems.