How important are glassy SOA ice nuclei for the formation of cirrus clouds?

Extremely low ice numbers (i.e. 5 – 100 / L) have been observed in the tropical troposphere layer (TTL) in a variety of field campaigns. Various mechanisms have been proposed to explain these low numbers, including the effect of glassy secondary organic aerosol acting as heterogeneous ice nuclei (IN). In this study, we explored these effects using the CAM5.3 model. SOA fields were provided by an offline version of the University of Michigan-IMPACT model, which has a detailed process-based mechanism that describes aerosol microphysics and SOA formation through both gas phase and multiphase reactions. The transition criterion of SOA to glassy heterogeneous IN follows the parameterization developed by Wang et al. 2012. With this parameterization, glassy SOA IN form mainly when the temperature (T) is lower than 210K. In the default CAM5.3 set-up in which only the fraction of Aitken mode sulfate aerosols with diameter larger than 100nm participate in the ice nucleation (Liu and Penner 2005 parameterization), glassy SOA IN are shown to decrease the ice number (Ni) by suppressing some of the homogeneous freezing at low temperatures thereby leading to an improved representation of the relationship between Ni and T compared to the observations summarized by Kramer et al. 2009. However, when we allow the total number of the Aitken mode sulfate particles to participate in homogeneous freezing, glassy SOA IN have only a small impact on the relationship between Ni and T. If the subgrid updraft velocity is decreased to 0.1 m/s (compared to 0.2 m/s in the default set-up), there is a large decrease of Ni, since homogeneous freezing is more easily suppressed by glassy SOA IN at these updrafts. We also present the effects of glassy SOA IN using an alternative ice nucleation scheme (Barahona and Nenes, 2009).