WMO: Update on 2023/4 Anomalies

The WMO released its (now) annual state of the climate report this week. As well as the (now) standard set of graphs related to increasing greenhouse gas concentrations, rising temperatures, reducing glacier mass, etc., Zeke Hausfather and I wrote up a short synthesis on the contributions to recent temperature anomalies.

Readers will recall our previous discussions on the anomalies in 2023 in particular, and the myriad of ideas that people have published to explain them. Following on from our AGU session on the topic in December, we were asked to provide a brief synthesis of the results so far. To our knowledge, this is the first quantitative attribution of the anomalies to specific processes (but it will clearly be not the last word).

First, what are we trying to explain? It is not the long term trends! The rise of temperature by ~1.5ºC since the 1850-1900 baseline is very clearly associated with the increases in greenhouse gases, slightly (and decreasingly) modulated by the changes in atmospheric pollution. Rather, we are trying to explain the residuals from that trend – why any year (or years) is much warmer or cooler than the trend. There is always something of course – the weather is variable – but for 2023 and 2024 the residuals were higher than for any other year in decades. And, as you will recall, the anomalies in 2023 specifically were not well forecast ahead of time.

We assessed the long term trend using a 20 yr loess smooth in the WMO timeseries to 2022 (which was then projected forwarded to 2024), and defined the residuals as the difference from that trend. A loess smooth has a little more structure than a linear regression, and we deliberately did not use the last two data points to define it. There is a very slight acceleration from 2000 onward in the smoothed curve, but this isn’t material for our analysis.

As you can see though, the size of the residual for 2023 was comparable to 2016 and 1998 (years that started with large El Niño events), and 2024 had the highest residual in decades despite the recent El Niño only being a moderate event. We estimated the impacts of ENSO from a statistical regression of the residuals against the Feb/Mar Nino34 index, which implicitly assumes that the El Niño in 2023/24 had similar impacts to previous events.

We estimated the other potential components from a combination of statistical modeling (for the solar cycle) and published radiative forcing estimates (for the impact of marine shipping emissions and Hunga Tonga) converted to temperature anomalies using the FaIR emulator. The impacts of East Asian SO₂ emission changes (which have dropped precipitously since ~2005) are noticeable in the global mean trends, but because they have been relatively smooth, the impact on the 2023/4 residuals is small.

A number of things stand out. There is a clear contribution to the residuals in both years from the shipping aerosol changes as has been widely expected, but given the published estimates of the radiative forcings, these contributions are only a fraction of the observed residuals. Indeed, we estimate that the contribution from the solar cycle has been comparably large. We estimate that the impact of Hunga Tonga (using the radiative forcing estimates from Schoeberl et al (2024) is actually negative (since the SO₂ contribution outweighed the added stratospheric water vapor). We find that ENSO had a negligible impact in 2023, but was a substantial contributor in 2024. And (somewhat to my surprise) the impact of changes in East Asian aerosols was negligible too.

When you sum up the contributions (assuming that they are independent, and taking into account the statistical uncertainties) the expected values fall short of the observations in 2023, but match 2024 quite well. Within the uncertainties (everything here is plotted with a 95% confidence Interval), you could say things line up, but note that the largest uncertainty comes from the modeling of the ENSO effects. More sophisticated modeling might well be able to reduce the uncertainty there.

Thus, the bottom line is, to no-one’s great surprise, that 2023 is harder to explain than 2024. As the community moves towards proper syntheses using ESMs with updated forcings, there may be some adjustments to this picture – the regionality of the aerosol effects might magnify their impact on the global mean temperature, the specifics of the El Niño effect might imply an effect beyond just what can attributed to the Nino3.4 index, and we might get a broader range of effects for the volcano and solar cycle components.

Other new studies

At few other articles on this topic have also appeared recently:

• Allen and Merchant (2025) give a “new interpretation of the drivers of Earth’s energy budget changes and their links to ocean warming”.
• Terhaar et al. (2025) look at jumps in SST in climate models and conclude that “a jump in global sea surface temperatures that breaks the previous record by at least 0.25 °C is a 1-in-512-year event”
• A new preprint from Stefan and Grant Foster is also available: Rahmstorf and Foster (2025) looks at an ENSO corrected data set and concludes that acceleration in SAT is apparent.

Stay tuned!

References

1. M.R. Schoeberl, Y. Wang, G. Taha, D.J. Zawada, R. Ueyama, and A. Dessler, "Evolution of the Climate Forcing During the Two Years After the Hunga Tonga‐Hunga Ha'apai Eruption", Journal of Geophysical Research: Atmospheres, vol. 129, 2024. http://dx.doi.org/10.1029/2024JD041296
2. R.P. Allan, and C.J. Merchant, "Reconciling Earth’s growing energy imbalance with ocean warming", Environmental Research Letters, vol. 20, pp. 044002, 2025. http://dx.doi.org/10.1088/1748-9326/adb448
3. J. Terhaar, F.A. Burger, L. Vogt, T.L. Frölicher, and T.F. Stocker, "Record sea surface temperature jump in 2023–2024 unlikely but not unexpected", Nature, 2025. http://dx.doi.org/10.1038/s41586-025-08674-z
4. S. Rahmstorf, and G. Foster, "Global Warming has Accelerated Significantly", 2025. http://dx.doi.org/10.21203/rs.3.rs-6079807/v1