Improving the Effectiveness of U.S. Climate Modeling (2001)

Currently, U.S. climate modeling is characterized by a highly creative, productive, and healthy community, particularly at the intermediate and smaller scales of effort. Despite this success, high-end modeling capabilities lag behind those of other nations. The Panel on Improving the Effectiveness of U.S. Climate Modeling was convened to provide federal agencies with an assessment of the nation's technical modeling needs and to provide recommendations on how government, interacting with the scientific community, can optimize the use of modeling talents in the United States.

The two primary users of climate models are the research community, whose goals are to advance the understanding of the climate system, and the operational community, which uses models for the production of climate prediction products in response to societal demands. The two groups are connected because research is crucial for constructing and evaluating the models needed to produce useful products, and the operational infrastructure required to produce these products (such as a sustained observing system, modeling system, and resulting model assimilated data products that result) are of great utility to the research community.

There are, however, differences between these two groups. The research community's approach to modeling is exploratory and sometimes without a clear path, with success judged by peer review. The operational community operates with a more clearly defined mission and success is often judged by the utility of a given product. Operational modeling is more rigidly constrained by external controls and evaluation procedures

requiring a dependable infrastructure with a high level of organization and centralization.

During its deliberations, the panel identified the key issues influencing the effectiveness of U.S. climate modeling. The following sections identify these issues and outline the panel's recommendations. These recommendations are based on input from a survey distributed to many of the U.S. modeling centers, a workshop held in Washington, D.C., and the expertise of the panel and address some of the missing elements in U.S. climate modeling.

Information about future climate is crucial for addressing numerous societal needs. Different communities and subsets of society require distinct climate-related products. Thus, centralized modeling activities under the auspices of a single agency are needed to assemble and distribute the necessary climate information and products to diverse user groups. The United States has not as yet centralized its climate activities.

Centralized modeling activities should have close linkages to research and user groups and ought to include model building, quality control and validation of models and products, product design and regular and systematic product production, and integration of observational data. Centralized activities require computational systems adequate to address these problems. Although the operational activities would be centralized, they should take advantage of research activities external to operations, including model development and analysis, diagnostics, and interpretation.

The European Centre for Medium-Range Weather Forecasts (ECMWF) is one model of a successful dedicated modeling facility. The center was established as a European cooperative weather forecasting venture. During its lifetime it has produced some of the highest-quality, highest-resolution forecasts from any modeling group. It is not clear, however, that this model translates into leadership in modeling research or research into the long-term aspects of the climate system. Although this model may be successful in Europe, it is not clear that it can be applied to the more decentralized U.S. climate modeling community.

Many of the measurements and observations used to define climate are made in the arena of weather prediction, and many of the atmospheric processes and feedbacks that influence short-term weather contribute to climate. Therefore, the panel recognizes that strong weather forecasting capabilities are necessary preconditions for effective climate model development, and close ties should be maintained between climate and weather modeling activities.

Finding: Increased demands for operational climate products of benefit to society, such as those required for the IPCC and National Assessments and for short-range climate forecasts, have placed heavy demands on the research community, which is neither well suited by culture nor by organization to regularly produce these products. Despite this the research community is essential in providing the knowledge needed to develop effective climate products.

Finding: When comparing U.S. and European high-end modeling the panel finds that U.S. modeling is still lagging in its ability to rapidly produce accurate high-resolution model runs. This situation has worsened since the publication of the NRC report Capacity of U.S. Climate Modeling to Support Climate Change Assessment Activities (NRC, 1998a).

Recommendation 1: In order to augment and improve the effectiveness of the U.S. climate modeling effort so that it can respond to societal needs, the panel recommends that enhanced and stable resources be focused on dedicated and centralized operational activities capable of addressing each of the following societally important activities:

1. short-term climate prediction on scales of months to years;

2. study of climate variability and predictability on decadal-to-centennial time scales;

3. national and international assessments of anthropogenic climate change;

4. national and international ozone assessments; and

The most effective means to obtain superior computer performance is to utilize machines with powerful processors; small, fast networks; and rapid memory access. Currently, Japanese parallel vector supercomputers provide the fastest and most capable architecture for the sustained processing of climate model codes. Access to these computing systems for U.S. scientists is limited due to the high tariffs put on these systems as a result of a Commerce Department anti-dumping order.

In order to provide high-quality climate products as well as operate an effective research program, the climate modeling community should have access to state-of-the-art, high-end computing facilities connected with centralized modeling activities as discussed previously. These computing facilities should have sufficient capabilities to comprehensively

investigate the climate system and to enhance and develop models to better understand climate change. To provide the most capable computational resources available, national boundaries should not influence where computers are purchased (NRC, 1998a).

Finding: Current U.S. high-end modeling efforts are being hindered by the forced acceptance of a computational architecture ill-suited to process the algorithms employed to model the earth's complicated climate system.

Recommendation 2: The panel recommends the adoption of a scientific computing policy ensuring open access to systems best suited to the needs of the climate modeling community.

Recommendation 3: Researchers should have improved access to modern, high-end computing facilities connected with the centralized operational activities discussed in Recommendation 1. These facilities should be sufficiently capable to enable comprehensive study of the climate system and help develop models and techniques to address relevant high-end climate modeling problems.

Effective climate model development is often hindered by an incompatibility between model components developed in different groups. A protocol to facilitate the scientific exchange of common diagnostic tools, the interchange of model components, and the exchange of data in a common data format is not yet in general use. To aid the evaluation and exchange of technological and research advances within and among the research and operational modeling communities, a set of common diagnostic and visualization tools and a set of programming and data standards is needed. The panel believes that a set of common modeling tools and standards would enhance the cooperation between high-end modelers and smaller scales of effort, would increase the efficiency of climate model development, and would reduce duplication of efforts among groups.

Finding: A common modeling infrastructure, consisting of system software and model code would reduce the inefficiencies within the climate modeling research communities and would allow the research and operational communities to interact successfully.

Recommendation 4: In order to maximize the effectiveness of different operational climate modeling efforts, these efforts should be linked to each

other and to the research community by a common modeling and data infrastructure. Furthermore, operational modeling should maintain links to the latest advances in computer science and information technology.

The climate modeling community is facing an overwhelming short-age of qualified technical and scientific staff. This difficulty is, in part, due to the inability of both research and operational modeling centers to compete with the high salaries and incentives offered by the high tech industry. Some overseas groups, (e.g. ECMWF) have overcome this difficulty by providing highly lucrative salary packages that modeling groups in the United States are unable to match. This situation is aggravated in U.S. university-based modeling groups since they are often dependent on the vagaries of short term funding for employee salaries. Furthermore, this situation is impacting university graduate programs as many students receive lucrative offers from private industry prior to the completion of their degree. This human resource problem is reflected by declining graduate enrollments in all areas of the climate sciences and in the growing disparity in the quality of life of scientists, especially young ones, living in major cities, and their private sector counterparts.

Finding: The panel finds that there is currently a strain on human resources in the climate modeling community. U.S. modeling groups are having difficulty competing with private industry and with overseas institutions for the high skilled and experienced scientists and computer technologists needed to ensure an effective modeling effort in both research and operational modeling efforts.

The shortage of highly skilled technical workers is not particular to the climate modeling community, but is part of a larger shortage of human resources affecting nearly all areas of science and engineering. The complexity of this problem, and the lack of expertise on the panel to address this issue, precludes this panel from making any specific recommendations related to human resources.

The panel has argued that the suite of designated “Climate Services” consisting of the establishment and sustenance of a climate observing system, the production of useful model products on the global and regional scale, and delivery and dissemination of these useful products would satisfy societal demands and would be of great benefit to the research community. In what form these climate services would be delivered; how much of the climate information would be developed and

delivered by public versus private sources; which agencies would take part; what roles they would play; and how the links between operational modeling, sustained observations, and research are to be established and maintained are all open questions and beyond the expertise of the panel.

Recommendation 5: Research studies on the socio-economic aspects of climate and climate modeling should be undertaken at appropriate institutions to design the institutional and governmental structures required to provide effective climate services. This assessment should include:

1. an examination of present and future societal needs for climate information;

2. a diagnosis of existing institutional capabilities for providing climate services;

3. an analysis of institutional and governmental constraints for sustaining a climate observing system, modeling the climate system, communicating with the research community, and delivering useful climate information;

4. an analysis of the human resources available and needed to accomplish the above tasks;

5. an analysis of costs and required solutions to remove the constraints in accomplishing the above tasks;

6. recommendations on the most effective form of institutional and governmental organization to produce and deliver climate information for the public and private sectors.