CSL allocation requests are reviewed by the CISL HPC Assessment Panel (CHAP), a panel of experts who recommend allocation of CSL resources in the context of all the relevant criteria and considerations and available CSL resources.
The panel reports its recommendations to the CISL director, who forwards them to the National Science Foundation for final approval. Panel recommendations for CSL allocations are based on the standard CHAP review criteria, the CSL eligibility criteria, plus the following considerations:
- Priority will be given to proposals that are positioned to begin using their CSL allocation immediately. Proposed projects should not require significant “ramp-up” time.
- Any proposal requesting less than the minimum allocation size for computing will not be considered for a CSL award. In addition, any proposal for which the review panel recommends less than the minimum amount will not receive a CSL award; if resources are available, however, the panel will consider the request for a university allocation.
- Preference will be given to collaborations for the improvement of the computing methodologies for the study of climate problems—for example, collaborations with research and modeling activities at other large computing centers or programs (such as NERSC, NASA NAS, DoE SciDAC, NSF PetaApps) or collaborations between earth system modelers and mathematical or computer scientists.
- The plan for evaluating the modeling experiments is a factor. Teams should propose a balanced program of experimentation, application, analysis, and validation. With appropriate justification, CISL may provide CSL projects with continued access to analysis resources after completing their large-scale modeling runs.
- The value of and plan for sharing the resulting data sets with the broader community will be considered. While the panel is sensitive to the issue of maintaining proprietary data rights to give the principals sufficient time to analyze and publish their findings, it is also necessary for CSL projects to demonstrate concrete data management plans for providing, in a timely manner, the output data and analyzed results to the larger community for others to evaluate and to use.
- The panel considers the team’s climate modeling experience both on NCAR systems and systems at other sites, effective use of previous allocations on NCAR HPC resources (if any), and reported accomplishments from prior computational activities as factors for recommending CSL allocations.