# NASA MMF

**Overview**

**Physics**

The GEOS4 has been constructed with the unique finite-volume dynamic core developed at Goddard (Lin 2004) and the physics package from the NCAR Community Climate Model CCM3 (Kiehl et al. 1998). The unique features of the finite-volume dynamical core include: an accurate conservative flux-form semi-Lagrangian transport algorithm (FFSL) with a monotonicity constraint on sub-grid distribution that is free of Gibbs oscillation (Lin and Rood 1996, 1997), a terrain-following Lagrangian control-volume vertical coordinate, a physically consistent integration of pressure gradient force for a terrain-following coordinate (Lin 1997, 1998), and a mass, momentum, and total energy conserving mapping algorithm for Lagrangian to Eulerian control-volume vertical coordinate transformation. The physical parameterizations in the GEOS have been upgraded with the gravity wave scheme from the NCAR Whole Atmosphere Community Model (WACCM) and the CLM version 2 (CLM-2). The GEOS also includes the ability to use passive water vapor tracers to diagnose the geographic source of water in precipitation and to provide quantitative diagnostics of precipitation recycling (Bosilovich and Schubert 2002). The embded 2D GCE model use the bulk single-moment microphysics (Lang et al. 2009). We are currently upgrading into the GEOS5, which features interactive aerosol transportation model and coupled ocean model.

**MMF-LIS**

**Giga-MMF**

Next generation of NASA MMF will be total grid points about Giga points, by greatly enhancing the scalability of GCE simulations, adapting to the GEOS5 cube-sphere core, and resolving a full spectrum of precipitating clouds (shallow/congestus/deep). Giga-MMF target the GCE horizontal grid points up to 1024 (dx=250m, in 256km domain) and vertical level would be up to 256 (dz=50~300m). With GEOS5 cube-sphere C48 (48x48x6 grid points) configuration, total GCE grid point will be approximately 3.6 billions. Giga-MMF development must overcome the computational limitation. First, Giga MMF requires complete replacement of the current NASA MMF MPI-domain-decomposition technique and traditional call-statement (from GEOS4 to GCE) will be replaced by asynchronized MPI approach. Namely, a set of computer processors will be independently allocated for the parent global model (GEOS) and the cloud model (GCE). By asynchronizing GEOS processes (general circulation, advection, radiation, and IO process) and GCE processes (cloud dynamics and microphysics), we can reduce computational latency as much as possible. Development will be completed in 2015.

**Applications**

Global maps of June–August 2007 mean latent heat flux in W m−2 for a) FLUXNET, b) CLM 2.0, c) MERRA, d) CLM 2.1. The color scale is the same for all maps. Grid resolution of each map is indicated.

**Citations**

*Bull*.

*Amer. Meteor. Soc.,*

**90**

*,*515-534.

*Environmental Modeling and Software*, 39, 103-115.

**Contact**