[an error occurred while processing this directive]

Implementation of a two-moment microphysics scheme into the MMF

Marat Khairoutdinov and Hugh Morrison

The multiscale modeling framework (MMF) is a new type of global model that uses a cloud resolving model (CRM) within each grid column of a low-resolution general circulation model (GCM). This approach was first introduced about a decade ago (Grabowski and Smolarkiewicz 1999; Khairoutdinov and Randall 2001), and the development and testing of MMF is a significant part of the work done at CMMAP. In the MMF, the CRM replaces the conventional GCM parameterizations of convection and moist processes, so that deep convective-scale and mesoscale motions are explicitly resolved. A key point is that in the MMF, small-scale processes like shallow convection, turbulence, and microphysics are not resolved and must still be parameterized. The representation of these processes can have a large impact on simulations.

In this work, a new microphysics scheme is implemented and tested in the superparameterized Community Atmosphere Model (SpCAM), which is the MMF originally developed by Marat Khairoutdinov at CSU. Microphysics schemes treat cloud and precipitation particles and the various physical processes that act upon them. The representation of microphysics is especially important in high-resolution models that resolve deep convective-scale motions, since the microphysics helps to determine buoyancy and hence the convective dynamics through latent heat release and condensate loading. Microphysics is also important for simulating the impact of aerosols on clouds, which is key since anthropogenic aerosol effects represent one of the largest uncertainties in estimates of climate change (IPCC 2007).

The goals of the new microphysics scheme are to improve the representation of microphysical processes and allow a physically-based treatment of aerosol effects on clouds in SpCAM. The new scheme predicts number concentrations and mass mixing ratios of five different species: cloud droplets, cloud ice, rain, snow, and graupel (Morrison et al. 2009). The prediction of both mass and number allows for a more flexible representation of mean particle size and droplet activation on cloud condensation nuclei (CCN), which is the key to simulating aerosol effects on precipitation formation and radiative transfer. CCN concentrations are diagnosed from monthly climatology derived from fully-coupled runs of the Community Climate System Model (CCSM3), and consist of sulfate, organic aerosols, and sea salt. The original scheme in SpCAM predicts only mass mixing ratios of cloud water and total precipitation, and uses a temperature-dependent partitioning to diagnose liquid and ice. Because of the extra prognostic variables in the new scheme and added complexity of the microphysical processes, there is an increase in model run time of about a factor of 2.5 using the new scheme.

Results of 3-year simulations (plus 4 months spin up) using the new and old schemes are compared. With some tuning related to representation of vertical velocity for droplet activation and autoconversion, the new scheme produces a comparable climate simulation to the old scheme. An example is illustrated by plots of shortwave cloud forcing (Fig. 1). While results using the new and old schemes are similar, there are some notable improvements, such as the reduced magnitude of the cloud forcing over the tropical western Pacific. The new scheme produces reasonable geographic distributions of droplet effective radius (overall values are smaller than MODIS retrievals but the retrieved values are believed to be too large) (Fig. 2). In particular, values are much smaller over land than ocean, corresponding with the differences in aerosol loading and CCN concentrations (Fig. 2). Additional simulations have been run comparing results for preindustrial and present-day aerosols to estimate anthropogenic aerosol effects due to sulfate.

Future work will focus on more detailed analyses, testing of sensitivity to various parameters in the microphysics scheme, and development of approaches to reduce the computational cost.




References

[an error occurred while processing this directive]