Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.


Uncertain then, irrelevant now

Uncertainty in estimates of the effects of aerosols on climate stems from poor knowledge of the past, pristine atmosphere — so getting a better understanding of these effects might not be as useful as was thought. See Article p.67

Aerosol particles in the atmosphere have long been feared to be the joker in the climate system's pack of cards. Even before the greenhouse effect became a household word, scientists had begun to consider the possibility that increasing concentrations of particles in the atmosphere's lowest layer (the troposphere) were acting to cool the planet, and thus masking what would otherwise be much larger temperature changes caused by rising concentrations of atmospheric carbon dioxide1. Reporting on page 67 of this issue, Carslaw et al.2 use an innovative approach to demonstrate that most of the uncertainty in estimates of aerosol forcing — the aerosol perturbation caused by humans that induces a global surface-temperature change — that is associated with cloud-affecting aerosols can be attributed to uncertainty in the emission rate of aerosol precursors from natural sources. Their work suggests that, if there is an aerosol joker, it was probably played a century ago, and has become irrelevant to understanding present and future changes in global climate.

The aerosol, like the Earth system itself, is fascinatingly complex. How it interacts with clouds and radiation depends not only on its composition and concentration, but also on how these properties are distributed across particles of different sizes. All of these factors depend on uncertain microphysical processes and poorly understood aerosol sources, which in turn are integrated by the vagaries of the wind. Consequently, estimates of the strength of aerosol forcing derived from models that attempt to represent these poorly understood factors are controversial3. Carslaw et al. had the brilliant insight to recognize that something could be learned from this uncertainty.

Using a comprehensive model of global aerosols to train a simple statistical model, the researchers characterized the 28-dimensional uncertainty landscape of the comprehensive model. By exploring this landscape, they find that poorly constrained natural sources determine the state of the pristine atmosphere, and thus dominate uncertainty in estimates of radiative forcing associated with aerosol–cloud interactions. Their findings aid the interpretation of earlier studies in which seemingly large differences in aerosol forcing have been found to be associated with relatively small changes in rather arbitrary parameters — such as the minimum number of cloud droplets permitted in a model of a cloud4. Carslaw and colleagues' work suggests that these parameters essentially act as a proxy for the state of the pristine atmosphere.

The disproportionate sensitivity of aerosol forcing to the background state of the pristine atmosphere suggests that the present, far-from-pristine atmosphere should be quite well buffered against further changes in sources of tropospheric aerosols. Consistent with this line of thinking, the authors used their model to show that the aerosol forcing from an important class of aerosol–cloud interaction has not changed for more than 30 years, a period that has witnessed a massive increase in the footprint of human influence on Earth and its atmosphere. Aerosol forcing is thus so well buffered that it has probably ceased to be relevant for present and future global climate changes.

Because aerosol forcing tends to plateau as aerosol concentrations rise, climate sensitivity — the change in the globally averaged temperature that accompanies a doubling of atmospheric CO2 concentration — is better constrained from observations of the more recent past (for which changes in CO2 forcing dominate; Fig. 1) than by referencing a distant, pre-industrial era whose atmospheric aerosol burden, Carslaw et al. argue, might well be unknowable. An analysis5 of observations over this more recent past, which discounts past uncertainty in aerosol forcing and exploits our better understanding of other elements of Earth's energy budget, such as average ocean heating rates, suggests that a surprisingly large climate sensitivity is not being masked by a large effect of aerosols, as was initially feared1.

Figure 1: Climate forcing by tropospheric aerosols and carbon dioxide.

The dark blue line represents the magnitude of the cooling effect from a tropospheric aerosol-forcing scenario in which a pristine natural atmosphere is strongly susceptible to perturbation by anthropogenic aerosol sources produced before the twentieth century. The light blue line depicts a scenario in which stronger natural sources of aerosols render the atmosphere less susceptible to human perturbations. The red line indicates the magnitude of the warming effect from forcing caused by changes in CO2 and other long-lived greenhouse-gas levels. Shaded bands represent the more complex pattern of global warming over the past century (green, strong warming; grey, modest warming). Carslaw et al.2 report that aerosol forcing has hardly changed since 1980, and so forcing has been dominated by CO2 changes over the past 33 years. (Curves adapted from data in refs 2 and 5.)

Because changes in tropospheric aerosols are localized near their sources, the above arguments do not rule out the possibility that anthropogenic emissions cause large regional effects. But if one follows the logic of Carslaw and co-workers' findings, such effects will also have been most evident in the more distant past — particularly for a scenario in which natural emissions of aerosol precursors are low, thus making the pristine atmosphere especially susceptible to perturbations in aerosols and their precursors (Fig. 1). In fact, there is surprisingly little evidence of changing weather patterns attributable to large alterations in regional air quality over North America and Europe in the past. But more effort should be devoted to identifying a smoking gun for regional effects of aerosols in historical records, for instance through a systematic study of temperature changes in the twentieth century. In the absence of a clear signal from the past, it seems reasonable to assume that the climatic effect of future changes in aerosol emissions, even on a regional scale, will not be large.

But the ultimate lesson to be drawn from Carslaw and colleagues' research, and from other studies6,7,8,9,10 published this year, is that a lack of understanding of basic processes — such as how patterns of cloudiness and large-scale atmospheric, or oceanic, circulation respond to warming — is limiting progress in quantifying present and future changes in Earth's climate. Only by refocusing research on these basic processes does climate science stand a chance of remaining relevant to a public that is grappling to understand the pace and pattern of climate change.


  1. 1

    Charlson, R. J. et al. Science 255, 423–430 (1992).

    CAS  Article  ADS  Google Scholar 

  2. 2

    Carslaw, K. S. et al. Nature 503, 67–71 (2013).

    CAS  Article  ADS  Google Scholar 

  3. 3

    Stevens, B. & Boucher, O. Nature 490, 40–41 (2012).

    CAS  Article  ADS  Google Scholar 

  4. 4

    Hoose, C. et al. Geophys. Res. Lett. 36, L12807 (2009).

    Article  ADS  Google Scholar 

  5. 5

    Bengtsson, L. & Schwartz, S. E. Tellus B 65, 21533 (2013).

    Article  ADS  Google Scholar 

  6. 6

    Bony, S. et al. Nature Geosci. 6, 447–451 (2013).

    CAS  Article  ADS  Google Scholar 

  7. 7

    Kosaka, Y. & Xie, S.-P. Nature 501, 403–407 (2013).

    CAS  Article  ADS  Google Scholar 

  8. 8

    Stevens, B. & Bony, S. Science 340, 1053–1054 (2013).

    CAS  Article  ADS  Google Scholar 

  9. 9

    Zhang, R. et al. J. Atmos. Sci. 70, 1135–1144 (2013).

    Article  ADS  Google Scholar 

  10. 10

    Meehl, G. A., Hu, A., Arblaster, J. M., Fasullo, J. & Trenberth, K. E. J. Clim. 26, 7298–7310 (2013).

    Article  ADS  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Bjorn Stevens.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Stevens, B. Uncertain then, irrelevant now. Nature 503, 47–48 (2013).

Download citation

Further reading


Quick links

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