Unraveling the contribution of dynamical and aerosol parameters on cirrus cloud formation in the CAM 5.1 using online adjoint sensitivity analysis (Invited Presentation)

Ice clouds play a central role in the climate system impacting Earth's radiative balance, tropospheric water distribution, and stratospheric water vapor transport. Despite their climatic relevance, understanding the link between aerosol/meteorological fields and cirrus formation on a global scale remains a considerable task. Attempts to understand this link have been made through sensitivity studies where the impact of finite perturbations to aerosol/meteorological fields on formed crystal number is quantified. Although straightforward, this way of investigating crystal number dependence on model parameter variability is time consuming and forces the state of the system to be altered when the evaluations are made. Furthermore, finite-difference approaches provide limited quantification of the real time dependence crystal number on process level model parameters. Investigating sensitivities using the adjoint of the cirrus formation model itself has the potential to alleviate these issues.

Here we present the development and application of the adjoint of the Barahona and Nenes (2009) cirrus formation parameterization. This scheme allows for competition between homogeneous and heterogeneous freezing and is unique in the sense that it is able to consider any ice nuclei (IN) spectrum (which empirically or theoretically provides the number of crystals that freeze heterogeneously). The adjoint is then implemented in the version 5.1 of the Community Atmosphere Model (CAM5) to study the relative importance of model parameters to cirrus ice crystal number globally with respect to aerosol size characteristics, ice nuclei freezing characteristics, and meteorological parameters. Model sensitivities are computed at each model time step and grid point, providing an unprecedented insight on cirrus formation within the coupled model. Sensitivities were then used to determine to what extent input parameters contribute to crystal number variability, elucidating parameter importance spatially.