Ever since humans first harnessed fire, we have emitted microscopic particles called aerosols into the atmosphere. These particles remain suspended in the air and alter the amount of sunlight that reaches Earth’s surface. Low-level clouds efficiently cool the planet by reflecting sunlight back into space and are readily exposed to human-made aerosols. Such particles can change the reflectance of clouds, and the associated cooling, by modifying the size of droplets1 or the amount of water2 in the clouds. Writing in Nature, Toll et al.3 provide compelling evidence that human-made aerosols cause a weak average decrease in cloud water content compared with unpolluted clouds. This result provides an important constraint on the overall cooling effect of aerosol emissions and reduces one of the key uncertainties in climate science.
Earth’s global mean temperature is governed by the interplay of competing warming and cooling processes. Since 1850, there has been a net warming owing to greenhouse-gas emissions from human activity4. Over this time, the global mean temperature has increased5 by 0.9 °C, although the warming caused by greenhouse-gas emissions has been partially compensated for by the cooling effect of aerosol emissions. Therefore, a precise quantification of this cooling could have profound implications for projections of future climate.
Clouds that form in the presence of high aerosol concentrations contain droplets that are smaller and more numerous than usual (Fig. 1). These droplets therefore have a large total surface area for sunlight to bounce off. Consequently, the reflectance of polluted clouds is greater than that of unpolluted ones1. As a result of the aerosol emissions of the early twenty-first century, as opposed to pre-industrial conditions, this enhanced reflectance generates a substantial cooling effect on Earth’s climate6.
Whether cloud reflectance is increased or decreased by changes in water content, and to what extent, has been highly uncertain. A greater water content in polluted clouds than in unpolluted ones could enhance the net cooling effect of aerosol emissions2. This possibility is suggested by many global climate models7. Some scientists have argued that the increase in cloud water content, and the associated cooling effect, caused by aerosols might be even larger than these models indicate8. By contrast, other evidence suggests that there could be considerably less water in polluted clouds than in unpolluted ones, which would reduce the net cooling effect substantially9.
To address this uncertainty, Toll and colleagues looked at features of polluted clouds called pollution tracks (see Fig. 1 of the paper3). These features were produced downwind of sources of human-made aerosols such as coal-fired power plants, oil refineries, smelters, cities, ships and wildfires. Like the cloud trails that form behind aircraft at high altitudes, these pollution tracks in low-level clouds are visible from space. As a result, polluted and less polluted cloudy regions can be clearly distinguished. Observed changes in droplet size or cloud water content can, therefore, be unequivocally attributed to variations in aerosol concentrations.
Using 15 years of high-resolution satellite data of near-global coverage, the authors built an unprecedented database of thousands of such tracks across Earth’s climate zones. Overall, they found that the average droplet size was at least 30% lower in polluted clouds than in unpolluted ones. Although differences in cloud water content varied, the mean water content was slightly lower in polluted clouds than in unpolluted ones (Fig. 1). This finding suggests that the effect of aerosols on cloud water content slightly reduces the overall aerosol-induced increase in cloud reflectance.
Toll et al. then extrapolated their findings to all low-level clouds across Earth, considering global changes in aerosol emissions from human activity. They estimate that the identified decrease in cloud water content offsets only 23% of the net cooling effect caused by the reduction in droplet size. However, the precise estimate remains uncertain. Although the authors sampled thousands of pollution tracks, these features are scarce. For example, it is extremely rare for a ship to leave a pollution track in its wake10, and probabilities of track generation for the other sources of human-made aerosols are likely to be similarly low. This rarity raises the question of whether observations made using pollution tracks can be generalized to all other conditions in which pollution tracks are not seen.
The most common hypothetical situations in which pollution tracks are not identified are: when clouds are already bright, so that added aerosols have no impact on reflectance; and when cloud properties are rapidly varying because of changes in humidity, stability or horizontal winds. The aerosol-induced decrease in cloud water content might therefore be smaller or larger than is estimated from pollution tracks. However, there is no a priori reason for the clouds to respond in a fundamentally different manner in conditions in which pollution tracks are not observed. Toll and colleagues’ work therefore strongly suggests that the sensitivity of cloud water content to changes in the concentration of human-made aerosols might not be accurate in many current global climate models, and that large cooling effects caused by variations in cloud water content are unlikely.
Nature 572, 35-36 (2019)