The goal of the SEA-CISM project is to develop a scalable, efficient, and accurate Community Ice Sheet Model (CISM). A state-of-the-art algorithms and tools framework will be developed, within which climate model developers can create a predictive ice sheet modeling capability. A fully integrated parallel hierarchical blocking structure will allow maximized
use of the latest high-performance computing architectures. An implicit solution algorithm with scalable preconditioning will be implemented using the Trilinos package of solvers to maintain efficiency with resolution and physical enhancements of ice representation.
CISM has short-term requirements to provide accurate quantitative estimates of ice sheet behavior, so our project will provide initial throughput gains for practical use. The model will be developed in extensible form, such that ongoing code improvements will transfer to the climate model development effort with minimal structural modification. CISM is already being developed as a component of the Community Earth System Model (CESM); the computationally modern CISM could provide a template for the next generation of other climate model components in the CCSM. At the completion of the project, CISM will contain demonstrably scalable solution algorithms and high performance computing tools and allow for the incorporation of future improvements.