CCSM is a climate model that has grown significantly over the past several years. Early on, there were a handful of scientists and software engineers working on the project and the project was reasonable to manage. In the past few years, the number of developers and collaborators has increased significantly as has the importance of this work.
Over the past year, concerns have been initiated by NCAR scientists, software engineers, the community, and external collaborators about how CCSM can continue to develop effectively as the number of people working on the project increases. Some of these concerns are summarized in Appendix A which came out of a discussion at the CCSM programmers meeting on September 19, 2001. Currently, there is a loosely defined set of rules that governs model development. Independent research and loose coordination is the status quo. This no longer works well for CCSM.
CCSM is currently coordinated as follows. There are individual working groups that are responsible for particular model components and/or analysis. These groups are relatively free to develop and test their components any way they choose. The one requirement is that they check the model into a CVS repository at NCAR at some point. The overall CCSM model is periodically tagged and production runs are then typically started. Very limited testing is conducted, there is no formal review of changes and change requests, and there are few policies with regard to process or coordination. Separate scientist and software engineering meetings are held nearly weekly and much coordination is carried out through personal communication.