A physically based scheme for the treatment of stratiform clouds and precipitation in large‐scale models. I: Description and evaluation of the microphysical processes

Abstract A stratiform‐cloud and precipitation scheme, incorporating prognostic variables for cloud liquid water and cloud ice, has been developed for the CSIRO global climate model (GCM). the scheme includes physically based treatments of key microphysical processes, turbulent mixing and semi‐Lagrangian advection of cloud‐water species and interactive cloud radiative properties. Objectives in the development of the scheme were to improve upon the physical realism of parametrizations used in earlier schemes, whilst also trying to provide a scheme with moderate computational overheads. The parametrized microphysical processes are evaluated in relation to observations and theory, and are compared to treatments used in earlier schemes in a series of short GCM experiments. It is argued that the treatment of precipitation formation in warm, and mixed‐phase, stratiform clouds is more realistic in the present scheme than in earlier schemes, which used crude methods for the parametrization of autoconversion, and did not treat key ice‐processes in a consistent way. In the present scheme, accretion processes are more important, whereas autoconversion is less important than in earlier schemes. To determine whether the cloud scheme requires the use of a reduced (split) time‐step, the sensitivity of the various terms to the time‐step is evaluated in another series of short GCM experiments. It is shown that the various terms are not very sensitive to the time‐step, so the scheme can be efficiently implemented without the use of a split time‐step. Overall, analytical or time‐centred treatments perform better than implicit or explicit schemes, especially in the calculation of the precipitation of cloud ice, where only an accurate analytical treatment is found to perform satisfactorily at large time‐steps. As a preliminary validation of the scheme, zonal‐mean fields from a six‐year model‐run are presented for the month of July. the results generally agree well with observations; in particular, the modelled cloudiness and long‐wave cloud‐forcing fields are more realistic than those obtained with the standard version of the CSIRO GCM.