Under SciDAC support, the PI and colleagues have developed two global cloud-resolving models (GCRM). These models are based on a new "Unified System" of equations that filters vertically propagating sound waves. They both solve the vorticity equation, and are discretized on a geodesic grid. The vorticity-divergence model, which is referred as to the "Z-grid model", predicts the vertical component of the vorticity, while the vector-vorticity model, which is referred as to the "VVM", predicts the three-dimensional vorticity.

Meanwhile, over the past ten years, the PI and collaborators have modified several successive versions of the Community Atmosphere Model (CAM) to use a "super-parameterization" in which the physical heating and drying rates are determined using a two-dimensional cloud-resolving model (CRM). The global model combined with the super-parameterization is called the multiscale modeling framework (MMF). Over the past two years, Cristiana Stan of the Center for Ocean Land Atmosphere Interactions (COLA) has coupled the SP-CAM with the ocean model component of the Community Earth System Model. Comparison with observations shows that the coupled version of SP-CAM generates very good results.

We are currently working, under other sponsorship, to make the super-parameterization a supported physics option within the CAM distribution The current MMF has inherent limits that motivate the development a second-generation version: (i) the CRM is two-dimensional; (ii) the orientation of the two- dimensional CRM must be chosen arbitrarily; (iii) the periodic boundary conditions. A prototype Quasi-3D (Q3D) MMF that removes these limitations has been developed and successfully tested by Jung and Arakawa.

We will create two new versions of the CAM that can be run with the standard CAM physics, with the Q3D MMF, or as GCRMs. One will be based on the Z-grid and the other on the VVM. It will be possible to run these models, with "conventional" horizontal grid spacings on the order of 50 to 100 km, using the CAM's standard (but evolving) physical parameterizations. The Q3D MMF will be implemented as a physics option in both models. We will also implement a cloud-resolving physics option in the new model, so that it can optionally be run as a GCRM. This will also allow us to test the Q3D MMF by comparison with the GCRM. The cloud-resolving physics will be identical to that used in the Q3D MMF. With these three alternative model configurations, the new CAM will be a very flexible modeling tool that can take advantage of increasing computer power in the years ahead.

We will also explore the possible benefits of allowing discrete CRMs to propagate, using a Semi-Lagrangian Interactive Cloud (SLIC) approach.

The model output generated by this project will be archived at the San Diego Supercomputer Center. All data generated will be made available to other interested parties on request, recognizing proprietary access to the project members according to NSF policy.