Cirrus anvil dissipation simulated by a mesoscale model with bulk microphysics

The dissipating phase of a cirrus anvil simulated by MM5 with a fine spatial grid (2-km) is analyzed in detail. The case of interest is the July 23, 2002 case of CRYSTAL-FACE. The numerical simulation captured the targeted convective events; however, falsely generated several convective cells downwind (upper troposphere) of the target convection. Overall, the simulation produces a much larger anvil, which dissipates more slowly than the observations. We integrate the governing equation for the ice mass over a typical GCM grid size (100-km) and examine the contribution of various resolved (by GCM) and unresolved processes, e.g., ice production by mean ascent versus production by subgrid scale motions where both are explicitly resolved in MM5. The unresolved processes are of comparable magnitude to their associated resolved terms, except for the production term. Furthermore, it is important to account for the difference between the cloud fraction and the area fraction of snow particles in order to obtain an accurate estimate of the vertical ice mass fluxes, especially for the mid-tropospheric section of a deep anvil.