Page 4

supercomputing capabilities, need to be provided to the climate modeling community.

4. The reliance of the United States upon other countries for high-end climate modeling must be redressed.

In order to avoid the aforementioned problem regarding priority setting, the USGCRP could assume increased responsibility for identifying, from an interagency perspective, any gaps or imbalances in the research priorities established by the individual agencies. At present, however, this is made difficult because some agencies have excluded from their USGCRP budgets the computational and human resources to support comprehensive, coupled atmosphere-ocean climate modeling efforts on a par with those in several foreign countries. Although an entirely top-down management approach for climate modeling is viewed as undesirable, national economic and security interests nevertheless require a more comprehensive national strategy for setting priorities, and improving and applying climate models. An effective national approach to climate modeling should ensure that available resources are allocated appropriately according to agreed upon science research and societal priorities and are efficiently utilized by the modeling community. We acknowledge that justification for and design of such a strategy would require a more complete evaluation of the current status of climate modeling in the United States than was possible in developing this report. Development of such a strategy should take place with full involvement of climate modelers within academia and the national climate research centers, along with users of climate modeling results and agency program managers.

Climate modeling in the United States promotes a healthy competition among various groups, but without better coordination of research among national laboratories and between them and the academic community, it may be difficult to optimally utilize available human and high-end computer resources. In particular, standardization of model output, model evaluation tools, and modular programming structures can facilitate model development and minimize duplication of effort, with the possibility that prudent standardization may yield some cost savings. High-end modeling coordination could also be