Sensitivity of Homogeneous Ice Nucleation to Aerosol Perturbations and Its Implications for Aerosol Indirect Effects Through Cirrus Clouds

The magnitude and sign of anthropogenic aerosol impacts on cirrus clouds through ice nucleation are still very uncertain. In this study, aerosol sensitivity (eta(alpha)), defined as the sensitivity of the number concentration (N-i) of ice crystals formed from homogeneous ice nucleation to aerosol number concentration (N-a), is examined based on simulations from a cloud parcel model. The model represents the fundamental process of ice crystal formation that results from homogeneous nucleation. We find that the geometric dispersion (sigma) of the aerosol size distribution used in the model is a key factor for eta(alpha). For a monodisperse size distribution, eta(alpha) is close to zero in vertical updrafts (V < 50 cm s(-1)) typical of cirrus clouds. However, eta(alpha) increases to 0.1-0.3 (i.e., N-i increases by a factor of 1.3-2.0 for a tenfold increase in N-a) if aerosol particles follow lognormal size distributions with a sigma of 1.6-2.3 in the upper troposphere. By varying the input aerosol and environmental parameters, our model reproduces a large range of eta(alpha) values derived from homogeneous ice nucleation parameterizations widely used in global climate models (GCMs). The differences in eta(alpha) from these parameterizations can translate into a range of anthropogenic aerosol longwave indirect forcings through cirrus clouds from 0.05 to 0.36 W m(-2) with a GCM. Our study suggests that a larger eta(alpha) (0.1-0.3) is more plausible and the homogeneous nucleation parameterizations should include a realistic aerosol size distribution to accurately quantify anthropogenic aerosol indirect effects.