Nobel Prize of Physics in 2021 was awarded to Klaus Hasselmann and Syukuro Manabe for identifying human fingerprints in climate warming and initiating climate model development. Their contributions pioneered the quantification of anthropogenic climate change enabling us attribute the observed warming of +1.2 C since pre-industrial primarily to increased concentration of greenhouse gases.

A key question to elucidate impacts of climate warming on our society is quantifying the threshold for emission of carbon to avoid overshooting of +1.5 or +2.0 C warming agreed in the Paris Agreement. These thresholds are determined by how sensitive the Earth system is to a single unit of radiative forcing. Our poor understanding of clouds’ response to aerosol emissions largely hampers a reliable quantification of the effective total radiative forcing and therefore the remaining carbon budget. This is because a significant fraction of CO2 warming effect is postulated to be offset by the cooling effect of aerosol-cloud interactions, but there is large uncertainty in exactly how much.  Therefore, a more robust understanding of aerosol’s fingerprint on clouds is imperative. However, previous observational evidence of aerosol-cloud interactions is either small-scale aerosol plume studies or large-scale spatiotemporal climatological correlations between aerosol and clouds. These methods are useful, but imperfect, for studying aerosol-cloud interactions, because the scale of plume studies is too small and less relevant in constraining large-scale climate models, while the climatological correlation studies are frequently contaminated by meteorological co-variability and therefore cannot confirm any causality.

Here, we pioneer a novel approach to distinguish the aerosol’s fingerprint in clouds, by combining modern machine-learning data-science technique and long-term satellite remote sensing records, to generate an emulator which tells us how clouds should look like if there is no aerosol perturbation. By applying this novel approach to an unprecedented effusive volcanic eruption in Iceland in 2014 (a good analogy of anthropogenic aerosol perturbation), we successfully distinguish aerosol’s fingerprint in cloud properties. Rather surprisingly, we find that for liquid water clouds, the cloud cover increased by about 10% when averaged over a huge region over approximate 3000 x 6000 km² and is responsible for more than 60% of the cooling effect from aerosol-cloud interactions; such a huge increase of cloud cover is not seen in most of climate models. This finding indicates a big gap in our current understanding of aerosol cooling, and urges for an improvement of aerosol-cloud interactions in our climate models. Our fingerprint study will help guide the next generation of climate models in representing clouds much better, enabling the robust prediction of climate warming and quantification of remaining carbon budgets. 

Chen, Y., Haywood, J., Wang, Y. et al. Machine learning reveals climate forcing from aerosols is dominated by increased cloud cover. Nat. Geosci. (2022). https://doi.org/10.1038/s41561-022-00991-6