Low clouds tend to have a cooling effect, as they block the sun from reaching the ground. High clouds, however, have a warming effect, as while they let solar energy reach the ground, the energy emitted back from the Earth is different. This energy can be trapped by the clouds, enhancing the greenhouse effect. Therefore, the type and amount of cloud a warming world will produce impacts further warming potential.
Inspired by ideas from the artificial intelligence community, the researchers developed a new method to quantify relationships between state-of-the-art global satellite observations of clouds, and the associated temperature, humidity and wind conditions. From these observed relationships, they were then able to better constrain how clouds will change as the Earth warms.
They found it was very likely (more than 97.5% probability) that clouds will amplify global heating, by both reflecting less solar radiation and enhancing the greenhouse effect. These results also suggest that a doubling of CO2 concentrations will lead to around 3.2C of warming. This is the highest confidence of any study so far, and is based on data from global observations, rather than local regions or specific cloud types.
Co-author Dr Peer Nowack, from the School of Environmental Sciences and Climatic Research Unit at the University of East Anglia and Imperial’s Grantham Institute and Data Science Institute, said: “Over the last few years, there’s been a growing amount of evidence that clouds probably have an amplifying effect on global warming. However, our new approach allowed us for the first time to derive a global value for this feedback effect using only the highest quality satellite data as our preferred line of evidence.
“Our paper makes a major step towards narrowing the most important uncertainty factor in climate sensitivity projections. As such, our work also highlights a new pathway in which machine learning methods can help us constrain the key remaining uncertainty factors in climate science.”