Primary bio-aerosols, layer-clouds and climate simulated with deep learning AI

Motivation In a previous project, we created a new scheme for atmospheric models to represent the heterogeneous ice nucleation by five types of primary biological aerosol particle (PBAP).

These ´ice nuclei´ (IN) include bacterial and fungal particles.The mechanisms for how the  PBAP-IN can affect stratiform layer-clouds is unclear.

Yet they may be the cloud-type most susceptible to be modified by bioaerosols since their tops can be relatively warm (e.g. -20 degC), with PBAP-IN dominating overall IN activity in-cloud.If such an influence from PBAPs exists, then there may be a ´natural indirect effect´ from PBAPs on climate via their modification of the extent and properties of layer-clouds.Planned research activities and objectives of the projectThe proposed project aims to elucidate how the radiative and microphysical properties and extent of layer-cloud respond to the five types of PBAP.

The approach is to simulate with our aerosol-cloud model (AC) an observed case of a layer-cloud with a top near -20 degC in a past campaign (´ACAPEX´, funded by US Department of Energy).

After validating the simulation against aircraft data, sensitivity tests will be performed to analyse the cloud-physical reasons (e.g. altered snow production) for radiative flux changes when PBAP loadings are altered.

We will then apply a global model with our AI-accelerated super-parameterization (SP) of clouds so as to assess indirect effects from these types of PBAPs on the global scale.