Climate change

Cool ozone

The influence of global warming on temperature trends at higher altitudes has been hotly debated. Stratospheric ozone depletion is another piece in the remaining tropical climate puzzle.

Climate change at the Earth's surface understandably receives the most attention from those interested in global warming. But temperature trends higher up in the atmosphere are also important, helping with the attribution of climate change and providing tests for global climate models. Debates over apparent discrepancies between observations and models in trends in the troposphere (the lower layer of the atmosphere from the surface to 12–16 km altitude) have raged for many years, and have been used to raise doubts about the validity of climate models. Forster et al.1 show that ozone depletion in the tropical lower stratosphere has contributed to recent tropospheric temperature changes in a location where discrepancies between some data and the models have stubbornly remained.

The large-scale vertical structure of temperature change in the atmosphere is an important characteristic of the forces driving climate change. Increases in greenhouse gases cause warming in the troposphere but cool the stratosphere. Greater output from the sun similarly warms the troposphere, but causes even greater warming in the stratosphere. Observations show that the troposphere has warmed in recent decades whereas the stratosphere has cooled markedly; this is clear evidence for anthropogenic warming rather than natural warming from the sun.

However, the precise structure of temperature trends with altitude is complex. Climate contrarians argued for years that models should not be trusted as they did not produce as much warming as satellite observations indicated in the troposphere. It was eventually shown that the satellite measurements were biased, and the apparent discrepancy largely vanished over most of the globe, with some problems remaining in the tropics2,3.

Therefore, our understanding of how temperature trends vary with altitude at low latitudes is still incomplete. Warming of much of the troposphere is closely coupled to surface warming in the models, with simulated temperatures up to altitudes of 10 km directly following the theoretical temperature change of air as a consequence of decreasing pressure and condensation of water vapour as the air moves upward — the so-called moist adiabatic lapse rate. In the tropics, however, trends derived from observations suggest less warming aloft than the models produce in both the lower and upper troposphere3. This discrepancy is difficult to interpret, owing to the large uncertainties in both model and observational trends results.

On the observational side, both radiosondes (balloon-borne instruments) and satellites show warming in the troposphere and cooling in the stratosphere. Whereas discrepancies between the various observational datasets are tiny near the Earth's surface, they grow steadily moving upwards to the upper troposphere and lower stratosphere3. The most likely explanation for the more negative temperature trends observed in radiosonde data (see Fig. 1) is that they contain cooling biases that have not yet been fully accounted for3 (though work continues on this topic6).

Figure 1: Tropical temperature trends.

a,b Changes in temperatures in °C per decade between 20° S and 20° N during 1979–1999 for the lower stratosphere (a) and the mid-troposphere to lower stratosphere (b). Trends8 derived from the models used in the IPCC's fourth assessment report without ozone depletion are shown as open circles, (n=5), and those with ozone depletion as solid circles, (n=12). For the observations3, diamonds are values from radiosonde observations, and satellite observations are shown as squares. The different colours are analyses of these data by different groups. Trends are calculated for the satellite lower stratospheric and mid-troposphere to lower stratospheric channels. Model error bar indicates the standard deviation among the AR4 models; uncertainties in individual models and datasets are not included.

Forster et al.1 suggest that deficiencies in the climate models' representation of ozone trends may explain the “long-standing discrepancy between modeled and measured temperature trends in the uppermost tropical troposphere”. Such a discrepancy exists in comparison with radiosonde data, but many models are in fact quite consistent with the substantially more positive trends derived from satellite measurements4,5.

Forster and colleagues make a convincing case that lower stratospheric ozone changes can have an important influence not only on local temperatures, but on the warming in the upper tropical troposphere as well. Hence models that do not take into account ozone trends will yield positively biased temperature trends not only in the stratosphere at the location of the ozone depletion, but also at the level below — the upper troposphere.

Despite the general impression that climate simulations used for assessments such as the IPCC Fourth Assessment Report (AR4) are rigorous model intercomparisons, in fact the climate models were driven by different forcings (external agents that affect the Earth's climate). Many climate models use ozone datasets that incorporate the observed depletion in lower stratospheric ozone in the tropics7 and are fairly similar to those of Forster et al. However, some climate models in the AR4 ensemble did not include ozone trends at all. Looking at the lower stratosphere, models including stratospheric ozone depletion found cooling trends that are 0.08 °C per decade greater than the average of all models3. Including ozone trends improves the agreement between models and observations both in the lower stratosphere and upper troposphere (see Fig. 1).

Thus part of the reason that models yield such a large range of temperature trends at these levels may be the lack of uniformity in their forcings, and part of the explanation for the model average continuing to show more positive temperature trends in the tropical upper troposphere than observations could indeed involve the ozone forcing.

The results presented by Forster et al. imply that projections of the expected future recovery of stratospheric ozone are important, not only for their effect on ultraviolet radiation reaching the Earth's surface, but also for their effect on the climate of the lower atmosphere.

Similar arguments could undoubtedly be made for tropospheric ozone and black carbon, which absorb solar radiation causing local heating and therefore affecting trends in the upper troposphere. Climate models need to include realistic treatments of all known climate forcing agents, not only greenhouse gases.


  1. 1

    Forster, P. M., Bodeker, G., Schofield, R., Solomon, S. & Thompson, D. Geophys. Res. Lett. 34, L23813 (2007).

    Article  Google Scholar 

  2. 2

    Fu, Q., Johanson, C. M., Warren, S. G. & Seidel, D. J. Nature 429, 55–58 (2004).

    Article  Google Scholar 

  3. 3

    Karl, T. R., Hassol, S. J., Miller, C. D. & Murray, W. L. (eds) Temperature trends in the lower atmosphere: Steps for understanding and reconciling differences (US Climate Change Science Program, Washington DC, 2006).

    Google Scholar 

  4. 4

    Thorne, P. W. et al. Geophys. Res. Lett. 34, L16702 (2007).

    Article  Google Scholar 

  5. 5

    Blog entry RealClimate <> (2007).

  6. 6

    Haimberger, L. J. Clim. 20, 1377–1403 (2007).

    Article  Google Scholar 

  7. 7

    Randel, W. J. & Wu, F. Geophys. Res. Lett. 26, 3089–3092 (1999).

    Article  Google Scholar 

  8. 8

    Santer, B. et al. Science 309, 1551–1556 (2005).

    Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Shindell, D. Cool ozone. Nature Geosci 1, 85–86 (2008).

Download citation

Further reading


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing