Summary of Other Relevant Reports
i. The Report of the DOE/NSF National Workshop on Advanced Scientific Computing.
This workshop report (commonly called the “Langer Report”) was designed to provide a foundation for establishing a science-driven national infrastructure of terascale computing, communications, and advanced simulation. Weather and climate prediction were given as examples of scientific and engineering applications needing such expanded computing facilities. This report discussed the following obstacles to the successful use of terascale computing facilities: (a) the scarcity of human resources applied to computational and scientific research problems; (b) the difficulty of matching the financial rewards offered by private industry, and; (c) the lack of proper software available to usefully optimize performance on the new generations of massively parallel computers.
Relevant recommendations arising from this workshop were:
the U.S. should launch a vigorous effort to make high-speed computing systems accessible to the national scientific and engineering communities;
the U.S. should concurrently launch a vigorous effort to develop software, algorithms, communication infrastructure, and the visualization systems necessary for effective use of the next generation of computing facilities;
the U.S. scientific and engineering communities should prepare to use these computing facilities to solve complex problems of both basic and strategic importance;
hardware, software and communications developments should be coordinated with these scientific and engineering applications (DOE/NSF, 1998).
ii. Report of the NSF/NCEP Workshop on Global Weather and Climate Modeling
This interagency workshop report discusses the future of weather and climate modeling in the United States. An important outcome of this report was the recognition that the diversity of U.S. modeling has created a barrier to efficient collaboration between various modeling groups. Recommendations were:
a common modeling infrastructure should be established to facilitate the evaluation and exchange of technological and research advances in the broader modeling community, with exchanges envisioned not only between the operational and research communities but also between the numerical weather prediction and climate modeling communities;
the common modeling infrastructure should be advanced by establishing modeling standards and guiding principles and by focusing efforts on the development of a finite number of core models, each of which would be devoted to a major area of modeling (e.g., numerical weather prediction, seasonal to interannual prediction, decadal variability);
the National Centers for Environmental Prediction be one of the centers associated with a core model promoting the common modeling infrastructure because of its responsibility for U.S. operational forecasts and because of its critical data assimilation activities.
iii. President's Information Technology Advisory Committee
The President's Information Technology Advisory Committee on future directions in information technology (PITAC, 1999) recommended increased investment in information technology, with priorities in the areas of software, information infrastructure (including networks), and high-end computing systems, noting that “extremely fast computing systems, with both rapid calculation and rapid data movement, are essential to provide accurate weather and climate forecasting ... to conduct scientific research in a variety of different areas and to support critical national interests.”
iv. Accelerated Climate Prediction Initiative
An interagency committee charged with making recommendations on the implementation of the Department of Energy Accelerated Climate Prediction Initiative (ACPI; Gates et al., 1999) produced a report that argued for a climate modeling structure that would include a centralized computing facility; a modeling and research consortium of exclusive us-
ers of the facility for specific projects involving model development, model diagnoses and predictions on all time scales; and a group of regional climate centers (recommended to be initially three to five in number) that would interface with local user communities to examine the impacts of climate change and variability on regional scales. The committee also recommended a national software and communications infrastructure to “provide a significantly enhanced capability to store, access, transfer, diagnose and visualize the results of high-end climate model simulations, [and to] provide an effective software and communications network linking the project's components and participants ...”
A review of the ACPI program by the JASON Group noted that (JASON, 1998):
substantial increases in the computational power available to U.S. researchers are well warranted and can contribute to a better understanding of the climate system;
computational power alone will not greatly improve our abilities to predict climate, and linked observational programs and process studies are also essential for a balanced global change effort.
v. High-End Climate Science: Development of Modeling and Related Computing Capabilities (USGCRP, 2000).
This report, solicited by the Office of Science and Technology Policy, was designed to make recommendations that “responds to unmet national needs in climate prediction, climate-science research and climate-change assessment.” The basic recommendation of this report is the creation of a dedicated organization, called the “Climate Service,” designed to respond to the various requirements and demands placed on the climate community. The report distinguishes between traditional research (discovery-driven) and the type of activity required by new demands on the climate community (product-driven). This Climate Service should be product-driven, centralized, and requires a new business plan to manage the Service in an integrated manner.
The report recommends that the Climate Service should have access to computational systems with the highest level of capability and should engage in two major core simulation activities: one in weather and one in climate. It noted that the obstacles to forming the Service are organizational and computational. In particular, it noted that the U.S. policy on high performance computing, imposing restrictive duties on Japanese vector supercomputers, significantly complicates the people problem for the Climate Service in that massively parallel distributed-memory computers requires a much greater investment in software, both system and applications, and therefore requires just the type of information technologists in greatest demand in a booming software industrial environment.
In order to remedy the fragmentation of effort discussed in NRC, 1999a, the report also recommends the Climate Service seek integration of effort across the disciplines involved in weather and climate modeling, across institutions, and across modeling, data and computational systems. A software infrastructure to enable collaboration and development of software by multiple scientists at multiple institutions, to allow a transition path from discovery-driven research to product-driven activities, and to optimize across computational resources and scientific research is also recommended. Issues of machines and software form an important part of the report.