CIG, which began in September 2004, has achieved most of the short-term goals outlined in the original proposal to the NSF. Significant accomplishments for CIG include (1) the adoption of modern software engineering techniques and infrastructure for our community; (2) development of several packages using common components, such as PyLith for the short-term tectonics community and Gale for the long-term tectonics community; and (3) deployment of the Seismology Science Gateway to the TeraGrid.

Before the current funding ends in August 2009, CIG will have also made substantial progress on several of the long-term goals, including a compressible mantle convection code in spherical geometry linked to seismology and mineral physics; and implementation of adaptive mesh refinement (AMR) codes within the mantle convection and magma migration subdisciplines. Yet significant computational obstacles remain an impediment to modeling geophysical processes, such as robust mesh generators and solvers that are capable of handing large variations in physical properties for anisotropic, nonlinear, and plastic materials; methodologies for multiphysics modeling that incorporate thermodynamics, multiple phases and components at different time and lengths scales; and solvers and adaptive mesh refinement techniques that are capable of scaling to machines with 10,0000 and greater computational cores.

The primary strategy employed in CIG-I has been to apply methods developed in other disciplines to geodynamics. However, CIG is rapidly approaching the limit of currently available methods and further progress in geodynamics will require development of innovative techniques in scientific computing and their application to geodynamics. The Science Steering Committee (SSC) believes CIG should continue beyond this initial funding and in a second phase, CIG-II, should seek proactive collaborations with the broader computational science community at a level far beyond that envisioned during the initial phase of CIG. This would insure that new developments in computational science target difficult obstacles in geodynamics, enlighten the computational science community of challenges in geodynamics, and bring new untapped talent to the geodynamics community.

Areas of focus for this collaboration would include developing robust mesh generation and adaption tools for geophysical applications with complex, nonplanar geometry and localized deformation, scalable solvers for problems with multiphase and multiscale physical properties and deformation, and techniques for coupled solution of physical processes operating on different spatial and temporal scales. Significant advances in these areas would permit (1) modeling of tectonic deformation across many earthquake cycles with resolution of the strain accumulation, propagating earthquake ruptures, postseismic relaxation, and evolution of fault systems; (2) integration of magmatic and fluid processes into simulations of mantle dynamics producing convection consistent with the development of plate boundaries; and (3) development of Earth Structure models that are compatible with seismic wave propagation, mantle convection, mineral physics, and the geodynamo.

These collaborative efforts would tend to be inherently more distributed than the activities in CIG-I. CIG would retain some staff members to maintain the core computational infrastructure. Most software development, however, would be done by small teams of computational scientists and geoscientists working together to overcome obstacles impeding progress of multiple subdisciplines within our community. Considerable planning will need to be directed towards developing an effective organization structure and work plan to insure that CIG maintains its vibrant, cohesive community and avoids acting as a funding agency. CIG will seek to continue in its efforts to support and promote Earth science by developing and maintaining software for computational geophysics and related fields.

Two potential sources of funding for CIG-II have been identified. The maintenance of the core computational infrastructure and community organization would potentially be funded through the NSF Division of Earth Sciences (EAR). The collaborative effort between computational scientists and geoscientists would potentially be funded through the NSF-wide Cyber-enabled Discovery and Innovation (CDI) program. In order to jump-start a detailed discussion of such a collaborative effort, NSF has tentatively agreed to fund a joint computational mathematics and geodynamics workshop to identify areas of mutual interest for collaborative research.