Assessing aerosol indirect effect through ice clouds in CAM5

Abstract

Ice clouds play an important role in regulating the Earth’s radiative budget and influencing the hydrological cycle. Aerosols can act as solution droplets or ice nuclei for ice crystal formation, thus affecting the physical properties of ice clouds. Because the related dynamical and microphysical processes happen at very small spatial and temporal scales, it is a great challenge to accurately represent them in global climate models. Consequently, the aerosol indirect effect through ice clouds (ice AIE) estimated by global climate models is associated with large uncertainties. In order to better understand these processes and improve ice cloud parameterization in the Community Atmospheric Model, version 5 (CAM5), we analyze in-situ measurements from various research campaigns, and use the derived statistical information to evaluate and constrain the model [1]. We also make use of new model capabilities (prescribed aerosols and nudging) to estimate the aerosol indirect effect through ice clouds, and quantify the uncertainties associated with ice nucleation processes. In this study, a new approach is applied to separate the impact of aerosols on warm and cold clouds by using the prescribed-aerosol capability in CAM5 [2]. This capability allows a single simulation to simultaneously include up to three aerosol fields: online calculated, as well as prescribed pre-industrial (PI) and present-day conditions (PD). In a set of sensitivity simulations, we use the same aerosol fields to drive droplet activation in warm clouds, and different (PD and PI) conditions for different components of the ice nucleation parameterization in pure ice clouds, so as to investigate various ice nucleation mechanisms in an isolated manner. We also applied nudging in our simulations, which helps to increase the signal-to-noise ratio in much shorter simulation period [3] and isolate the impact of aerosols on ice clouds from other factors, such as temperature and relative humidity change. The results show that homogeneous ice nucleation is the main contributor that leads to strong longwave ice AIE in this model. The estimated PD-PI longwave cloud forcing (LWCF) change is strongly sensitive to the simulated sub-grid updraft velocity. Considering the effect of pre-existing ice crystals on ice nucleation can help to significantly reduce the LWCF change. In comparison, the effect of heterogeneous ice nuclei spectra is relatively small, although the perturbations in the LWCF and shortwave cloud forcing are still non-negligible.