Global modelling of ice-nucleating particles and impacts on mixed-phase clouds

The process of cloud glaciation strongly alters the properties of mixed-phase clouds. Between 0C to about -37C, cloud liquid droplets can either exist in the liquid phase in metastable state known as supercooling, or they can be composed of solid ice crystals. For a liquid droplet to freeze at these temperatures, the action of an external agent, known as ice-nucleating particle (INP) is needed. The atmospheric distribution of ice-nucleating particles was simulated in past studies as a function of the aerosol concentration, however, new experimental information about the ice- nucleating ability of different aerosol species and several new atmospheric measurements of INP are now available to be used in models.
In this thesis, I use this new information to develop a global atmospheric model of the distribution of ice-nucleating particles to assess the relative importance of mineral dust, marine organic aerosols and black carbon for contributing to atmospheric concentrations of INPs. The model is evaluated against several datasets of INP concentrations measured in the atmosphere to test its realism and locate regions of the world where additional currently missing sources of INP could be important. The results show that feldspar aerosols dominate the atmospheric INP concentration for most parts of the globe, whereas marine organic aerosols are more relevant in the remote Southern Ocean. Black carbon particles, in contrast, seem not to play a substantial role when new estimates of its ice-nucleating ability are used.
With the information obtained by this model, I explore whether the representation of ice-nucleating particles in climate models plays a role in the Southern Ocean radiative bias. This bias is related to modelled clouds reflecting too-little solar radiation, causing large errors in sea-surface temperatures and atmospheric circulations. I combine cloud-resolving simulations over regions of 1000 km with the new estimates of the INP concentration in remote regions to show that the simulated clouds reflect much more solar radiation than predicted by a global climate model, agreeing much better with satellite observations in both magnitude and frequency.
Overall, these results will improve our understanding of the role, distribution and importance of ice-nucleating particles in the atmosphere and provide the scientific community new points of view to understand model biases.