In this talk, I will provide a brief introduction to two main research topics that I am currently working on, which involve integrating observation, machine learning, and numerical models to enhance our understanding of aerosol-cloud interactions.

The first topic explores how semi-volatile compounds affect aerosol activation into cloud droplets on a global scale. In my PhD, we found that these semi-volatile compounds could halve the critical supersaturation for cloud formation, as calculated from Köhler theory, in Delhi, India - an effect not yet incorporated into current models. Currently, there is no direct evidence of co-condensation effect under supersaturated conditions. Our team at ETH designed smog chamber experiments to quantify this co-condensation effect under subsaturated conditions and developed a cloud parcel model to simulate it, with the aim of parameterising it in global models. I would like to discuss the feasibility of conducting co-condensation experiments in a cloud chamber setting.

A second focus of my research is understanding the impacts of aerosols impact on clouds using satellite data and machine learning. In a recent collaboration, we used volcanic degassing events as natural experiments, combing machine learning with satellite data to disentangle aerosols’ impact on clouds from confounding meteorology on a large scale. This provides a robust observational benchmark for climate models. I will introduce the intercomparison of this constraint with ECHAM6-HAM2 and CESM2 model simulations, to explore how we can further improve these models.