GISTEMP & NOAA have also reported for May. They both show May’s global temperature anomalies down on April’s (GISS at +1.12ºC, down from April’s +1.17ºC, NOAA at +1.07ºC down from April’s +1.14ºC) but these May anomalies both up on last year’s May anomaly (GISS +1.08ºC, NOAA +1.06ºC).
Both GISS & NOAA are SAT/SST which would explain the April-to-May difference from wholly SAT ERA5.

SST-wise, the ERA5 re-analysis 60N-60S SST anomaly was flat through May, this in contrast to the strong warming seen Jan-Apr. Using the daily numbers at ClimatePulse, these SST60-60 anomalies Jan-May run +0.37ºC, +0.41ºC, +0.44ºC, +0.50ºC, +0.49ºC.

Projecting forward to anticipate how high the global temperature will go with the coming El Niño is perhaps best done without recourse to 12-month averages as the start-point. The 12-month aves still show a value boosted by the cooling anomalies a year ago. The present May anomalies are not greatly different to those for May/June back in 2023 in the run up to the “bananas!!!” anomalies that Autumn. The Banana Watch!!! continues.

While the coming El Niño is projected to be likely stronger than back in 2023, as the Official NOAA CPC ENSO Strength Probabilities page says “Stronger events do not always mean bigger weather and climate impacts. Stronger events can make it ‘more likely’ that certain impacts could occur.” (The page should get its June update later today.)
The NINO3.4 temperature the raw numbers used to measure ENSO strengths was stuck at +1ºC right through May but took an upward step to +1.3ºC for the first week of June. (Not sure how the NINO3.4 is now converted into the new RNINO3.4 they use.)

And here’s a thought concerning El Niño. Over the last 30 years we’ve had two “very strong” El Niños (1997 & 2015) and two managing a “strong” rating ( 2009 & 2023). The resulting temperature wobbles didn’t entirely reflect those ratings, particularly when the Southern Hemisphere is considered. Add in the stronger warming in the NH thro’ autumn months (which is not entirely an ENSO thing) and the relationship between El Niño strength and attribution of the global warm wobble to an El Niño becomes less-than-straightforward. And what’s with the arrival of that warm wobble in the SH (the El Niño is well-synchronised to the season) so early in less strong 2009 & 2023 El Niños?

Years … … … … 1997-98 … … 2009-10 … 2015-16 … 2023-24
Max RONI … … … … 2.4 … … … …1.5 … … … 2.3 … … … 1.5
Max T (Global) … +0.25ºC … … +0.19ºC … +0.32ºC … +0.27ºC
Max T (NH) … . … +0.28ºC … … +0.21ºC … +0.51ºC … +0.42ºC
Max T (SH) … . … +0.24ºC … … +0.20ºC … +0.15ºC … +0.17ºC
(NOTE – Anomaly bases for NH & SH differ from Global here)

You see the same language in dying multilateral bodies: “The OSCE is not dead.” “The INF treaty is not dead.” “The WTO’s dispute system is not dead.” Usually, that phrase appears about 2–5 years before actual brain death.

The Arctic Council may not be dead yet, but it’s still unconscious and on life-support in intensive care. The prognosis is poor.

“Can the Arctic Council, as of late 2026, issue a binding moratorium on high-seas fishing in the Central Arctic Ocean if new data shows risk?”

The answer is no — because Russia or the US can block it. That’s not a functioning body. That’s a patient in the ICU with no family consent for surgery.

Final verdict: The Arctic Council isn’t dead yet — but it’s no longer capable of governing. That’s a different, more dangerous state.

Lies, damned lies, and statistics? Depends on what you assume your “Global co2 emissions” count. While my “GHG emissions” does not equate to your “co2 emissions” no matter what the data you think you’re relying on says. Two different things.

And then there’s physics
https://tamino.wordpress.com/2026/06/10/co2-good-news-and-bad-news/

GF “The obvious next question is: why has CO2 risen so fast since 2014? The emissions data do not explain the unexpectedly high growth rate. Perhaps there’s something going on with the carbon cycle, feedbacks have kicked in, with melting permafrost releasing CO2 into the air just as the oceans are less able to absorb it. Perhaps land use change is more potent than we believe. My own answer: I don’t know.”

One might wonder, like I do, why anyone would trust the emissions data compiled by national governments and the Global Carbon Project’s Global Carbon Budget anyway.

I’m with Foster: “I’ll confess that I have a hard time trusting emissions data..”

Foster 2026 found a surface warming rate of roughly 0.4°C/decade, although Foster himself added a personal opinion that the rate could drop to 0.33°C/decade, citing potential emissions reductions and renewables. That’s a policy hope, not a data conclusion.

The data shows acceleration. What humans do next might change the trajectory. But that’s mitigation, not refutation. And acceleration can only be confirmed in hindsight, after the data is in.

PS
“Lies, damned lies, and statistics” cuts the wrong way here. The statistics were always showing acceleration. The p-value just kept it from being heard.

Julian, a heads up.

MA Rodger is hyper-focusing on misleading methodological minutiae to avoid engaging with the central finding — that multiple independent lines of evidence continue to point to acceleration.

Zou 2026 is an analysis that either supports or counters the work of others. The actual final numbers are not as important as the fact it mirrors the same physical effects: the warming trend has/is accelerating.

Zou et al. provides a powerful complementary analysis to Foster 2026. Different methods, different datasets, different atmosphere level, same conclusion of a marked acceleration. When one paper (Foster 2026) finds acceleration using GMST, and another (Zou et al. 2026) finds it using satellite TLT, that’s not a weakness. That’s stronger convergent evidence.

Yes, TLT has larger ENSO wobbles, but that’s why they removed ENSO effects statistically. If anything, finding acceleration despite larger noise is stronger evidence.

Satellite calibrations have been a big issue in the past. Now resolved. Given RSS and UAH now broadly agree on trends, this is a non issue today.

The TLT range weighting function includes some upper air, but the dominant signal is lower troposphere. All trend analysis is curve-fitting. The question is whether the fit captures real physics. Calling someone “rabid” doesn’t refute their method.

The Zou paper shows the post-2015 trend is statistically distinct from the pre-2015 trend. That’s what acceleration means.

Zou estimates a short term +0.5–1.0°C total additional warming in that specific atmospheric layer over the next decade. Whereas Foster 2026 suggested an ongoing rate of possibly 0.4C / decade (0.38C?) on the surface. With a qualified personal opinion at a lower rate of only 0.33C / decade going forward.

The TLT warms faster than the surface — that’s basic atmospheric physics. So if surface warming is accelerating, TLT warming must accelerate even more. So Zou’s higher numbers aren’t a contradiction. They’re confirmation of expected behavior.

And the higher natural variability (enso/noise) in the TLT versus the surface the authors attempted to remove as shown in the paper. Precisely what Foster 2026 did for the surface temperatures.

In the earlier “non-significant” papers (eg Beaulieu, Foster etc) the rate of acceleration didn’t past the arbitrary “statistical significance” test. Meaning the data was already signaling acceleration, the threshold just kept it from being “official”. The p-value is not decisive. The data is.

Cloud/albedo research shows feedback loops amplifying recent warming. Arctic methane studies show natural emissions increasing, with microbial brakes failing. Atmospheric CO2 increase rates accelerating. Increased forcing from forest destruction, wild fires and other warming feedback loops.

All complementary research. All convergent. None contradictory. Multiple papers, different methods, different datasets, different atmospheric levels — all pointing in the same direction the acceleration is real and grounded in basic climate physics.

In summary Julian, the Zou paper doesn’t need to be the final word. It’s one more brick in a wall built from multiple papers, multiple methods, and multiple atmospheric levels — all pointing to acceleration.

Even the earlier papers that ‘failed’ the p-value test still showed accelerated rates in their point estimates. Papers that missed the threshold were not wrong. They were just early. The observation data has been signaling this for years. The only thing that’s changed is the confidence level.

MA Rodger,
Thanks for a very thorough explanation – instantly bookmarked. What immediately stood out to me in that pre-print was them making projections based on essentially fitting a polynomial to a trend, which is always a rather dangerous exercise, as you and Tamino rightfully pointed out. I just wasn’t sure about my own conclusions, so I wanted to get some sort of confirmation (or refutation) here, hence why I asked in the first place.

As for temperatures of various parts of troposphere: I tried to read about them on my own here and here with a strong focus on what that pre-print was suggesting (i.e. how does acceleration in TLT warming translate to acceleration in GMST warming), but I just couldn’t figure it out on my own in the end. I perhaps shouldn’t have spoken meanly to zebra, but I really do believe that there are just limits to how much an individual can learn and understand on their own without any mentorship or guidance or connections within broader scientific community etc. – verifying your own knowledge is hard if you don’t have anyone competent to ask, but that’s just my opinion.

in addition to my post of 9 Jun 2026 at 4:07 PM,

https://www.realclimate.org/index.php/archives/2026/06/unforced-variations-june-2026/#comment-849072

Dear Paul,

To be honest, I so far supposed that what do you have developed (or still are developing) is a theory in the usual sense of this word – a mathematical description of ENSO based on an application of your original analysis of tidal forces acting on Earth oceans in combination with some original solutions from the fluid mechanics, because you have mentioned both disciplines many times. Your posts raised my feeling that combination of these two physical disciplines forms the core of your efforts. I wondered why you keep talking about artificial intelligence (AI) in this respect, however, I supposed that you see a way how you could exploit it for formulation and/or solving the respective mathematical equations.

Now, based on your reaction to my question, I rather guess that you perhaps try to circumvent intractable fluid mechanics and hope that an AI engine could be trained with your analysis of tidal forces on one hand and with empirical data characterizing climate oscillations like ENSO or AMO on the other hand and that this way, you could replace the sought (but mathematically intractable) theory with a practical AI tool having the same capability. Is your goal, actually, the AI tool for ENSO prediction?

Maybe you describe this goal simply and understandably somewhere in your numerous posts or on your blog; I apologize if I missed it.

If you still train your AI engine or perhaps even have not started therewith yet, it could be understandable that you cannot offer any prediction that could serve as a test bed for your tidal hypothesis. Could you confirm if I indeed grasped your idea correctly? And, should I finally grasped the clue, could you explain in a plain language what progress you already achieved on your way towards the ENSO-predictive AI tool?

Thank you in advance and best regards
Tomáš

It’s not scientifically significant. El Ninos average every 3 to 7 years. No one with any training in statistics will trust a single prediction, so it will take a whole sequence of predictions to make it statistically significant. That could take at least 20 years. Well, I don’t care about any of that because I’m more interested in the physics aspects than taking on the mantle of some dime-store Nostradamus,

Take a look at this time-series of NAO where there are 80 peak excursions since 1950
https://imagizer.imageshack.com/img923/1346/Y2FBa7.gif
One does not make sense of a single next prediction without making sense of all the previous erratic cycles.

So what’s interesting is in doing as much as we can RIGHT NOW in coming up with a unifying model of geophysics. If you ask an LLM about the strengths of the lunar-based results that I have presented, it will ultimately say that the most powerful form of cross-validation is in how it may work collectively for ENSO, AMO, NAO, PDO, QBO, Chandler wobble, and regional mean sea level and atmospheric pressure measurements. That’s where the LLM starts to accept the possibility of bunking consensus, not in one-offs but in collective results. It’s also where other scientists should start to consider the possibility. And that’s the way that science should work; there really has never been a recent case of a single individual carrying the entire load of doing the calculations and experiments for some new finding. There’s always some parallel effort going on.

The main obstacle or mental stumbling block is the belief that only those with access to supercomputers should have the final say. They have created this artifice, even though it may be overkill.

in re to: “Temperature-driven cloud loss is now detectable creating an additional feedback loop of increased warming.”

One does not “detect” temperature-driven effects per se. One observes an unexpected change in the real world and then attempts to attribute its cause. The ongoing interpretation/rationalization/conjecture/normalization that anomalous SW cloud radiative effect change is attributable mostly to temperature mediated feedback is not very well supported by existing TCR screened CMIP-class models, and is not really reconciled with a classical hypothetical framework such as those originating from Manabe. From an energy budget standpoint in terms of TOA net radiation, it resembles something more like an effective radiative forcing. We should hope the re-parameterized CMIP7 members are able to offer new insights into missing inputs, potential issues in observation, or bona fide missing physics. The detection of an unexpected trend in reflected shortwave radiation is distinct from, and does not by itself demonstrate, that the trend is attributable to temperature-mediated feedback.

Australia’s Snowy 2.0 pumped-hydro project, initially touted in 2017 as a $2 billion scheme, has spiraled into a massive infrastructure controversy. While direct construction costs are officially estimated around $12 billion, energy economists and analysts warn that once interest charges and mandatory transmission upgrades are factored in, the true price tag is approaching $42 billion

Originally slated to be finished by 2021, the expected completion date has been pushed back to late 2028 at the earliest.Current leadership at Snowy Hydro has conceded that even this revised 2028 deadline is in doubt, with some analysts predicting completion could stretch into 2030 or beyond.

The astronomical $42 billion figure often cited by critics includes the cost of associated transmission infrastructure like HumeLink, as well as over a decade of compound interest during the prolonged construction phase.

Despite these hurdles, the government-owned company maintains that the project is roughly two-thirds complete and is critical for firming up the National Electricity Market as coal plants retire

And some more info on the acceleration matter on Tamino I addressed above

Fire accounted for one-third of global land cover change in 2023, with forest loss due to fire increasing dramatically in boreal and tropical regions. Emissions from forest fires have risen 60% since 2001 . The positive feedback is explicit.
https://www.wri.org/insights/land-use-climate-change-feedback-loop

Methane emissions from fossil fuel operations are still not falling despite proven mitigation pathways, according to the IEA’s Global Methane Tracker 2026
https://van.nongnghiepmoitruong.vn/tackling-methane-emissions-key-for-climate-change-and-energy-security-d809515.html

new Nature Climate Change study (June 2026) found that under warming, natural methane emissions from lakes, ponds and wetlands will increase inexorably — the methane-consuming microbes cannot keep up with the extra methane being produced . That’s another direct positive feedback
https://www.eurekalert.org/news-releases/1130715

The data from 2025 and 2026 is painting a picture of an Arctic where the natural environment is becoming a major, and increasingly uncontrollable, driver of permafrost melt methane emissions though difficult to quantify into the future.

The Ice Loss & Albedo Feedback
Fresh research (May 2026) in Nature Geoscience confirms Antarctic ice melt creates a self-reinforcing feedback loop that most climate models don’t include. The study warns this means the out of date IPCC scenario projections are “too conservative”. Will AR7 improve reporting?
https://www.eurekalert.org/news-releases/1128447

Fossil Fuel Emissions Are Still Rising
Wood Mackenzie’s Energy Transition Outlook 2025-26 confirms: peak emissions now projected for 2028 been (pushed back), with the global warming pathway at 2.6°C, above their base case from a year ago. The report states bluntly: “No major country is on track to meet their 2030 emissions targets”
https://storage.pardot.com/131501/1761584185BajrOANV/Wood_Mackenzie_ETO_2025_26_executive_summary.pdf

Of course it all adds up alongside current and future data centre rollouts.

Alberta is actively pitching cheap natural gas to tech companies for AI data centers — directly contradicting Canada’s “clean power” AI strategy. 90% of Canada’s 100 planned hyperscale data centers are in Alberta, where the grid is 60% natural gas and emissions intensity is 5x the national average

The 2026 Shell “Surge” scenario projects data centres consuming an additional 5,000 terawatt-hours by 2050 — about 8% of global electricity supply. In that scenario, AI infrastructure becomes a primary driver of new generation, but fossil fuels remain in the mix for decades still.

Renewables are expanding, but they’re barely keeping up with demand growth, not displacing fossil fuels. While the developing nations seek to rapidly increase their energy supply. As Wood Mackenzie puts it: “Renewables largely add to supply rather than displacing fossil fuels”.

Not surprising if warming is accelerating, with feedbacks kicking in, and fossil fuel emissions still rising into the future.