Mea culpa: Earlier this month, I said I would tell you what happened when I started asking, out loud, how much of recent warming might be natural recovery from the Little Ice Age rather than entirely our fault – and why certain colleagues reacted as if I had defaced a cathedral. Well, I am doing that, but next month. I switched Essay 2 and 3, to convey more vulnerability and something else of concern before I tackle the cartel. On to the matter at hand …
I used to believe that the oceans were the perfect witness.
They are slow. They are vast. They hold 1,100 times more heat than the atmosphere.
Whatever signal we were looking for (whether natural and/or human) would be written clearly in their heat content; a signal stripped of the weather noise that makes surface records jump around.
That was the story I taught.
The story the models predict is a steady, almost linear climb in ocean heat from rising greenhouse gases. A fingerprint so unmistakable that even the most stubborn skeptic would have to concede.
Then I looked at the numbers with a colleague’s assistance (more to that shortly).
The Argo float network (4,000 robotic dives measuring temperature to 2,000 metres) has been running since 2004. Before its development, we had expendable bathythermographs, ship buckets, and the heroic efforts of the Hadley Centre to stitch it all together into the EN4 dataset. Together, the combined record is the longest, highest-quality ocean temperature time series that humans possess.
Here is what it shows, in one sentence:
Since 1871, the upper 700 metres of the global ocean have warmed at an average rate of 0.06°C per decade, with almost all of the variance explained by the 65–70-year stadium wave called the Atlantic Multidecadal Oscillation (AMO).
Not CO2. Not volcanoes. Not aerosols.
The AMO.
When the North Atlantic is in its warm phase (1930–1960 and 1995–present), the global oceans warm. When it flips cold (1965–1995), the warming pauses. The correlation is so tight that a simple sinusoidal fit of the observed AMO index reproduces 75% of the variance in global ocean heat content since the late 19th century.
Figure 1 – Global ocean heat content (OHC) time series from 1871–present (black line) with overlaid AMO index sinusoidal fit (red dashed line), demonstrating that natural AMO oscillations explain ~75% of OHC variance; highlighting variability over linear forcing. From Zanna et al. (2019), Figure 1a (page 1127). Global reconstruction of historical ocean heat storage and transport
I discovered this while attending a meeting on ocean heat uptake. Instead of confirming the models, they diverged sharply after 2005. The CMIP6 ensemble (our best, most expensive, most greenhouse-forced simulations) predicted twice the observed warming in the upper 700 metres and three times the warming between 700 and 2,000 m.
Figure 2 –Comparison of observed ocean heat content changes (black lines/bars) vs. CMIP6 model projections (coloured ranges) from 2005 onward, showing models overestimating warming by 2x in the upper 700m and 3x in 700–2,000m depth; evidence of divergence from reality. From Cheng et al. (2022), Figure 3 (page 780). Past and future ocean warming. Open copy available here.
The oceans were not accelerating. They were oscillating.
This was not a blog post. This was NOAA, Hadley, Argo, and every peer-reviewed reconstruction my colleague could consult.
The divergence was not subtle. It was glaring. And it was growing.
I did what any responsible academic does: I wrote it up, emailed it to other colleagues, and waited for the anticipated response: “Great catch – can we explore why those models are running hot?” (My colleague would not risk his grant by writing the email, so cheap the price of his integrity.)
The actual response:
Crickets.
Crickets morphed into concern, and concern morphed into gentle persuasions that I might want to emphasize the desired long-term trend rather than the decadal wiggles.
One colleague suggested in the hallway after a lecture, “Be careful. Administration needs the oceans to be the smoking gun. You know, for grants.”
The smoking gun having misfired, I dug.
It appears that the Pacific Decadal Oscillation (PDO) and the AMO are phase-locked in a way that suggests that the global ocean in effect breathes in and out on 50–70-year cycles. The warming periods from 1910–1945 and from 1975–2005 occurred at virtually identical rates of 0.12°C per decade in the upper ocean. This despite CO₂ forcing being three times stronger in the later period. The Argo-era warming rate (2004–present) is lower than the pre-Argo rate (1955–2003), exactly the opposite of what greenhouse physics demands.
Figure 3 – Timeline of AMO index (black line) with correlated North Pacific index (red line, lagged by ~10 years), illustrating phase-locked cycles between AMO and PDO-like patterns on 50–70-year scales, driving global ocean warming/cooling rhythms. From Zhang & Delworth (2007), Figure 1 (page 2). Impact of the Atlantic Multidecadal Oscillation on North Pacific climate variability
Together, these suggest that the evidence for recent “greenhouse physics” needs greater integrity.
They suggest that some in the scientific community playing fast and loose are, at the very least, insincere.
None of these observations disproves greenhouse warming. I still trust the science suggesting that CO2 forces the system. But they do something dangerous to the accepted narrative; they make the human fingerprint look like the natural fingerprint we have seen twice before in the instrumental record.
I crossed the rubicon, from believer to convert.
My conversion was not dramatic. There was no lightning bolt. Only the slow realization that the data trusted by those motivated to “settle the question” actually compels us to reopen the case.
The AMO is staring back at us, unblinking from the ocean heat record.
You cannot unsee it. And there is no need to avert your gaze.
You can try to go back to the old certainty. You can tell yourself the models need more time. You can tell yourself the next El Niño will fix the divergence.
I told myself at the time that pointing this out in public would only give ammunition to people who deny basic physics, and that pushing it would be a waste of time in that institution of higher learning.
It was.
But the oceans do not care about anyone’s career; they keep oscillating.
So, I am writing about it now.
I now understand why the cartel (the three institutions that I will name in my next essay) treats natural variability as a rhetorical problem rather than a scientific one. If the largest heat reservoir on Earth is still dominated by internal modes we do not fully understand, then the supremely confident attribution statements (so-called “unequivocal” declarations) sound more like marketing and less like science.
I am not asking you to believe that there is no human influence. I am asking you to believe that the influence is smaller, slower, and far more uncertain than what you have been told. The science is NOT settled.
It remains falsifiable – grants and certifications be damned.
The convert’s dilemma is simple: when you learn what the oceans speak, you cannot revert to the convenient narrative.
(Richard LeBlanc, BIG Media Ltd., 2025)
