Is it cake (or a cloud)? Using time evolution and natural experiments to uncover aerosol impacts on cloud processes

With improvements to the global observational networks and model fidelity, climate models are getting increasingly good at producing an accurate cloud climatology. However, there is still significant variation in the response of their clouds to aerosol perturbations. This variation is magnified when considering intentional perturbations to clouds (such as marine cloud brightening), where a model not only needs to get aerosol-cloud interactions right 'on average', but in specific conditions. A similar challenge exists in developing observational constraints for models, where aerosol-cloud susceptibilities (relationships determined from temporal variability over long timescales) are harder to use to assess specific conditions. We need observations that can help constrain cloud processes, ensuring that a simulated cloud is 'right for the right reasons'. 

While a simulated cloud might appear similar to an observed one, an external perturbation provides a unique opportunity to uncover the processes that have set the cloud properties. By 'poking' a cloud, they allow us to see if the cloud behaves like a real one, or it is just superficially similar (like a cake). 

Here we show how the time evolution of clouds following inadvertent perturbations (so-called 'natural experiments') can be used to identify the role of different processes in setting cloud properties. The cloud response following these experiments can be used to identify model biases, improving the accuracy of aerosol-cloud processes. We link these natural experiments to the response in large-scale temporal cloud variation, highlighting how this can be used to isolate causal aerosol impacts on clouds and providing process-level constraints on climate model behaviour.