A Process-Based Cloud Cover Parameterization for the ECHAM-HAM Global Climate Model

Clouds have a major impact on the climate system and are a dominant source of uncertainty in future climate projections. However, modelling clouds in a global climate model (GCM) is challenging as many processes work together and one has to find a good balance between simplicity and still representing all important processes. Cloud cover and the cloud droplet and ice crystal size distributions are essential to simulate the radiative effect of clouds. Currently, cloud cover is just a function of relative humidity in the ECHAM-HAM GCM. Because this simple approach has its limitations, we developed a new cloud model that includes and is oriented around a prognostic cloud cover. With our new model we are able to simulate the different cloud processes and cloud types more consistently and realistically:

• Liquid clouds form when the air is supersaturated. Supersaturation can be reached either by large-scale motions in the whole gridbox or through smaller-scale motions in only part of the gridbox as shown in Figure 1. In the second case, we calculate the cloud cover based on the likelihood that an air parcel reaches supersaturation via subgrid-scale motions.
• Mixed-phase clouds develop from liquid clouds that have formed in or are advected to temperatures below the melting point when cloud droplets start to freeze. We are able to simulate this freezing process quite well with simulated ice nucleating particle concentrations in agreement with observations. Cloud cover is conserved during freezing. We explicitly calculate ice crystal growth by cloud droplet riming and the Wegener-Bergeron-Findeisen process and thereby ice crystals can glaciate a whole liquid cloud. Our improved ice crystal sedimentation replaces the diagnostic treatment of snow and ensures consistency between the presence of ice crystals and cloud cover.
• Cirrus clouds form in ice supersaturated regions. As for liquid clouds, we calculate the cover at cirrus cloud formation based on large-scale and small-scale motions. The in-situ formation of cirrus clouds is a competition for water vapor between ice nucleation on dust particles and homogeneous freezing of solution droplets which determines the concentration and size of the newly formed ice crystals. Due to our improved sedimentation treatment, the ice crystals can sublimate during their decent or trigger a glaciation of an underlying supercooled liquid cloud.
• Convective clouds occur in a smaller area of the gridbox. They efficiently transport energy, mass, specific humidity and other trace substances to higher altitudes. Many of the cloud processes proceed in a short time. The link between the convective and stratiform cloud scheme is convective detrainment. We simulate cloud droplet formation and freezing in convective clouds to get an estimate of the detrained cloud droplet and ice crystal concentrations. Together with a calculation of the convective cloud cover, we consider convective detrainment as a subgrid cloud formation process for stratiform clouds.

Preliminary results show a better agreement of our simulated cloud cover with the CALIPSO-GOCCP satellite product. Our new cloud model simulates the geographic distribution of low-level and high-level clouds better than the previous model version. In particular, we can reduce the overestimation of cirrus clouds in the tropics and of low-level clouds in the extra-tropics and polar regions. Also other cloud properties, like cloud droplet and ice crystal number concentrations, compare well with observations. Further, our changes influence the partitioning of homogeneous vs. heterogeneous ice crystal nucleation in cirrus clouds, and the supercooled liquid fraction in mixed-phase clouds.

Figure 1 shows the different mechanisms of cloud formation in our model. Stratiform clouds form by large-scale ascent and smaller-scale motions. While large-scale motions lead to cloud formation in the whole gridbox, small-scale motions form a cloud in part of the gridbox. Besides stratiform cloud formation, convective detrainment can form a stratiform anvil or marine stratocumulus in part of the gridbox.