Monday, 12 January 2004: 11:30 AM
Impact of Turbulent Liquid Water Flux on Cloud Microphysics in Mesoscale Models
It has been increasingly popular in the mesoscale model community that an explicit prog-nostic approach is used to predict liquid/ice water content for cloud systems ranging from tropi-cal deep cumulus to shallow boundary layer stratocumulus clouds. In this approach, the temporal evolution of the cloud water is determined by the interaction among the major cloud dynamic-microphysical processes, two of which are the turbulence transport and condensation. Although the mixing length scheme is generally acceptable for the formulation of the turbulent transport for the quasi-conservative scalars, it does not represent the nature of that of the cloud water which is dominated by the turbulence generated condensation. Despite its severe defect, the down-gradient mixing scheme continues to be used for the cloud water prognostic equation in many mesoscale models. In this work, the impact of the liquid water flux is investigated under the stratocumulus cloud conditions with two commonly used formulations: the down-gradient and thermodynamic methods. The first uses turbulent-kinetic-energy dependent mixing scheme, the second one does not involve any severe assumption, but is only appropriate for 100 percent cloud coverage. As expected, the mixing length approach results in very large negative liquid water flux which transports the liquid water downward and leads to strong evaporation at the cloud base and condensation at the cloud top. This result is contradictory to our basic understanding of the cloud microphysics processes. The thermodynamic approach correctly predicts upward liquid water flux, and leads to a major condensation at the cloud base and evaporation at the cloud top, which is a significant improvement compared with the previous results. A new conservative mixing procedure is also introduced for the non-conservative prognostic vari-ables such as temperature and water vapor mixing ratio in a mesoscale model.