‘Climate Models Can’t Explain What’s Happening to Earth: Global warming is moving faster than the best models can keep a handle on’ is the headline of an article in The Atlantic by Zoë Schlanger [1]
The content of the article does not justify the clickbait headline, which should instead read
‘Climate Models Haven’t Yet Explained an anomalous Global Mean Surface Temperature in 2023’.
Gavin Schmidt authored an earlier comment piece in Nature [2] with a similarly hyped up title (“can’t” is not the same as “haven’t yet”). He states very clearly in a discussion with Andy Revkin [3], that he fully expects the anomaly to be explained in due course through retrospective modelling using additional data. It’s worth noting that Zeke Hausfather (who also appears on Revkin’s discussion) said in an Carbon Brief article [4] that 2023 “is broadly in line with projections from the latest generation of climate models” and that there is “a risk of conflating shorter-term climate variability with longer-term changes – a pitfall that the climate science community has encountered before”.
It is not surprising there are anomalous changes in a single year. After all, climate change was historically considered by climate science as a discernible change in averaged weather over a 30 year period, precisely to eliminate inter-annual variability! Now, we have been pumping man-made carbon emissions into the atmosphere at such an unprecedented rate we don’t have to wait 30 years to see the signal.
If you look at the historical record of global mean surface temperature, it goes up and down for a lot of reasons. A lot of it has to do with the heat churning through the oceans, sometimes burping some heat out, sometimes swallowing some, but not creating additional heat. So the trend line is clearly rising and the models are excellent in modelling the trend line. The variations are superimposed on a rising trend. Nothing to see here, at this level of discussion.
The climate scientists are also, usually, pretty good at anticipating the ups and down that come from El Nino, La Nina, Volcanic eruptions, etc. (Gavin Schmidt and others do annual ‘forecasts’ of the expected variability based on this knowledge). Which triggered the concern at not seeing 2023 coming, but why expect to get it right 100% of the time?
Don’t confuse this area of investigation with extreme weather attribution, which addresses regional (ie. sub-global) and time limited (less than a year) extreme events. Weather is not climate, but climate influences weather. So it is possible using a combination of historic weather data and climate models to put a number on the probability of an extreme event and compare it with how probable it has been in the past. So, 100 year events can become 10 year events, for example. This is what the World Weather Attribution service provides. The rarer the event, the greater the uncertainties (because of less historic data to work with), but it is clear that in many cases extreme weather events are becoming more frequent in our warming world, which is no surprise at all, based purely on statistical reasoning (The Royal Statistical Society explain here.)
So back to The Atlantic piece.
The issue I feel is that journalists and lay people can’t abide uncertainty. What are the scientists not telling us! In general people want certainty and often they will choose based mostly on their own values and biases rather than expert judgment. In the case of the 2023 anomaly, the choice seems to be between “it’s certainly much worse than the modellers can model”, “it’s certainly catastrophic”, “it’s certainly ok, nothing to see here”, or something else. All without defining “it’s” or providing any margin of error on “certainty”. Whereas scientists have to navigate uncertainty every day.
The fact is that we know a lot but not everything. There is a spectrum between complete certainty and complete ignorance. On this spectrum, we know:
- a lot ‘that is established beyond any doubt’ (e.g. increasing carbon dioxide emissions will increase global mean surface temperatures);
- other things that ‘are established outcomes, but currently with uncertainties as to how much and how fast’ (e.g. sea-level rise as a result of global warming and melting of ice sheets, that will continue long after we get to net zero; before it reaches some yet to be determined new equilibrium/ level);
- and others that ‘currently, have huge uncertainties attached to them’ (e.g. the net amount of carbon in the biosphere that will be released into the atmosphere through a combination of a warming planet, agriculture and other changes – we don’t even know for sure if it’s net positive or negative by 2050 at this stage given the uncertainties in negative and positive contributions).
So we can explain a lot about what’s happening to Earth, we just have to accept that there are areas which have significant uncertainties attached to them currently, and in some cases maybe forever. Not knowing some things is not the same as knowing nothing, and not the same as not being able to refine our approaches either to reduce the levels of uncertainty, or to find ways to address those uncertainties (e.g. through adaptation) to mitigate their impacts. Don’t put it all on climate models to do all the lifting here.
The current climate projections are much more precise than say the projections on stock market prices in 5 or 10 years, but we don’t use the latter as angst ridden debate about the unpredictability of the markets. We consider the risks and take action. On climate, we have enough data to make decisions in many areas (e.g. when it would be prudent to build a new, larger Thames Barrage), by using a hybrid form of decision making within which the climate models are just one input. Even at the prosaic level of our dwellings, we manage risk. I didn’t wait for certainty as to when the old gas boiler would pack up before we installed a super efficient heat pump – no, we did it prudently well beforehand – to avoid the risk of being forced into a bad decision (getting a new gas boiler). We managed the risks.
Climate models have been evolving to include more aspects of the Earth System and how these are coupled together and to enhance the granularity of the modelling (see Resources), but there is no suggestion that there is some missing process that is required to explain the 2023 uptick but probably missing data; not the same thing. Although there is a side commentary in [4] involving input from Professor Tim Palmer calling for ‘exa-scale’ computing, but Gavin Schmidt pushes back on the cost-effectiveness of such a path; there are many questions we must address and can with current models.
There are always uncertainties based on a whole range of factors (both model generated ones, and socio-economic inputs e.g. how fast will we stop burning fossil fuels in our homes and cars; that’s a model input not a model design issue). There is possibly nothing to see here (in 2023 anomaly), but it could be something significant. It certainly doesn’t quite justify the hyperbole of the The Atlantic’s headline.
If we globally are waiting for ‘certainty’ before we are prepared to act with urgency, we are completely misunderstanding how we should be managing the risks of man-made global warming.
We certainly should not, at this stage at least, be regarding what happened in 2023 as an extra spur to action. Don’t blame climate models for not having raised a red flag before or urgently enough – which is the subtext of the angst over 2023.
The climate scientists will investigate and no doubt tell us why 2023 was anomalous – merely statistical variability or something else – in due course. It is not really a topic where the public has even the slightest ability to contribute meaningfully to resolving the question. It might be better if instead The Atlantic was publishing pieces addressing the issue of what questions climate models should be addressing (e.g. constrasting the building of sea walls, managed retreat and other responses to sea level rise), where everyone can and should have a voice (as Erica Thompson discusses in her book [5]).
Climate scientists have been issuing the warning memo for decades, at least since the 1979 Charney Report, with broadly the same message. We read the memo, but then failed to act with anything like the urgency and agency required. Don’t blame them or their models for the lack of action. Ok, so the advance of models has allowed more diverse questions to be addressed (e.g. trends in flooding risks), but the core message remains essentially the same.
And please, don’t use 2023 as another pearl clutching moment for another ‘debate’ about how terrible things are, and how we need more research to enable us to take action; but then turn our heads away again. Until the next headline, of course.
(c) Richard W. Erskine, 2025
REFERENCES
- ‘Climate Models Can’t Explain What’s Happening to Earth: Global warming is moving faster than the best models can keep a handle on’, Zoë Schlanger, 6th January 2025, The Atlantic.
- ‘Climate models can’t explain 2023’s huge heat anomaly — we could be in uncharted territory’, Gavin Schmidt, 19h March 2024, Nature, https://www.nature.com/articles/d41586-024-00816-z
- ‘Factcheck: Why the recent ‘acceleration’ in global warming is what scientists expect’, Zeke Hausfather, 4th April 2024, https://www.carbonbrief.org/factcheck-why-the-recent-acceleration-in-global-warming-is-what-scientists-expect/
- ANDY REVKIN speaks with longtime NASA climate scientist GAVIN SCHMIDT about his Nature commentary on what missing factors may be behind 2023’s shocking ocean and atmosphere temperature spikes, Youtube, https://www.youtube.com/live/AYknM2qtRp4?si=fsq0y-XkYG58ITw5
- ‘Escape from Model Land: How mathematical models can lead us astray and what we can do about it’, Erica Thompson, 2022, Basic Books.
SOME RESOURCES ON CLIMATE MODEL EVOLUTION
- Timeline/ history of climate models https://www.carbonbrief.org/timeline-history-climate-modelling/
- Fast Computers and Complex Climate Models, https://scied.ucar.edu/learning-zone/how-climate-works/fast-computers-complex-climate-models
- ‘Next generation climate models: a step change for net zero and climate adaptation’, Royal Society, https://royalsociety.org/-/media/policy/projects/climate-change-science-solutions/climate-science-solutions-modelling.pdfy view.
EssaysConcerning 