The current workforce of climate model developers is insufficient to meet the growing need for climate model development work (Jakob, 2010). Most modeling centers have only a small number of people directly involved in climate model development. It is difficult to quantify the number of climate model developers in the United States, because a systematic study on the climate modeling workforce has never been done. The committee estimates that the number of full-time employees who work on climate model development is on the order of a few hundred.1
CURRENT CHALLENGES IN THE CLIMATE MODEL DEVELOPMENT WORKFORCE
Climate models had their origins in both weather forecast models and very simple models describing the radiative balance of the planet that streamlined representations of the ocean and atmosphere. The earliest climate models (ca. 1980) had many simplifications, such as fixed cloudiness or oceans with no currents. Since that time the complexity of modeling has increased, including not only much greater spatial resolution and realism in the representation of the ocean-atmosphere-land-ice system, but also the inclusion of new component models, such as atmospheric chemistry and aerosols, sea ice, the terrestrial and marine carbon cycles, and ocean biogeochemical cycles. These changes have increased the demands on model development and
1 The National Research Council (NRC) report Improving the Effectiveness of U.S. Climate Modeling (NRC, 2001b) estimated that there were approximately 550 full-time employees dedicated to weather and climate modeling in the United States. The current committee requested information from several modeling centers regarding their workforce. The National Centers for Environmental Prediction has about 63 full-time employees who work on the Global Forecast System and the Climate Forecast System. The National Aeronautics and Space Administration’s (NASA’s) Global Modeling and Assimilation Office has about 7 full-time employees who work on climate modeling for the Goddard Earth Observing System Model, Version 5. NASA’s Goddard Institute for Space Studies has about 10 scientist-level people who work on the full Earth system model, with many of those people not working on the model full time. Of the 169 employees at the Geophysical Fluid Dynamics Laboratory (including federal employees, contractors, and National Oceanic and Atmospheric Administration Cooperative Institute), about 70 percent work on model development, application, analysis, and interpretations. These numbers are difficult to analyze and compare, because it is challenging to make distinct categories of people doing only model development, experiments, or analysis. In many cases the same person is doing all these activities, but at different points in time, and not necessarily full time.
analysis, but the human resources have generally not kept pace with the rapid growth in model complexity.
The development and use of comprehensive climate models in the United States requires a large number of talented individuals in the following areas:
• scientists engaged in understanding the climate system, leading to the development of new parameterizations and other model improvements (distinct cadres of scientists are often needed for various model components, such as the ocean or terrestrial ecosystem models);
• scientists engaged in using the models for well-designed numerical experiments and conducting extensive diagnostics of the models to better understand their behavior, ultimately leading both to model products and to scientific insights that provide the impetus and context for model improvements;
• scientists studying the regional details provided by the archived results from global model simulations and related downscaling efforts, and how these vary across various models;
• support scientists and programmers to conduct extensive sets of numerical simulations in support of various scientific programs and to ensure their scientific integrity;
• software engineers to create efficient and portable underlying codes, including the development and use of common software infrastructures;
• software engineers and scientists to facilitate easy and open access to model output through modern networking technologies;
• hardware engineers to maintain the high-end computing facilities that underpin the modeling enterprise; and
• climate interpreters to translate climate model output for decision makers.
The U.S. institutional and funding system has addressed some of these areas better than others; in particular, the U.S. scientific effort on model diagnostics and regionspecific analyses has kept up better than the effort devoted to model improvement. The result is that many climate modeling efforts are subcritical in some aspects. In particular, there are longstanding problems in the simulation of the atmosphere-oceanland-ice system that are not yet solved, and yet these have been somewhat neglected in the desire to add additional complexity into models. One example of a longstanding problem is the tendency for virtually all climate models to simulate an unrealistic structure of the Intertropical Convergence Zone in the eastern Tropical Pacific. Another such example is the tendency for virtually all models to simulate sea-surface temperatures in the equatorial Atlantic that increase from west to east, instead of the observed increase from east to west. These errors in the simulation of the basic state of the tropical climate can distort the overall simulation of the climate system.
One important objective is to develop a pathway that can lead to modeling efforts in the United States that have sufficient human resources to meet their challenges in all critical areas, including both persistent and longstanding problems such as those mentioned above. Efforts are needed to address the emerging scientific frontiers discussed in Chapter 4, such as the effects of aerosols on clouds (the indirect aerosol effect) or the terrestrial and oceanic carbon cycles, which are major sources of uncertainty in climate change projections. In addition, continuing model development will be required to provide the high-quality climate simulations that can provide information to decision makers at the regional and local levels, as discussed in Chapter 10.
Finding 7.1: The level of human resources available for climate modeling has not kept pace with the demands for increasing realism and comprehensiveness of the models, leading to subcritical efforts in multiple areas of core modeling efforts. This is a serious impediment to progress.
ESTABLISHING AND MAINTAINING A PIPELINE IN CLIMATE MODEL DEVELOPMENT
Workers in climate model development have primarily received postgraduate degrees. In order to maintain a pipeline of human capital to sustain the climate modeling efforts, the United States will need to ensure the current and future availability of fellowship funding for graduate students and postdoctoral researchers, including expansion of programs at national laboratories and research facilities.
Data on the numbers of students involved in climate model development do not exist.2 As a proxy for understanding trends in the training of climate model developers, the committee examines data on related fields of computer science, geosciences, mathematics, and physics. Current trends in the education pipeline in fields related to climate modeling (Figure 7.1) show that the overall number of Ph.D. degrees being awarded in some fields related to climate modeling is growing, but numbers of master’s degrees and bachelor’s degrees are not growing. The percentages of females and minorities are low and have not been growing substantially over the past decade. As stated, although none of this information is specific to the pipeline of climate model developers, the committee infers that it is indicative of a pipeline that is not growing in a robust fashion.
2 Jill Karsten, National Science Foundation (NSF), personal communication, 2011.
FIGURE 7.1 Education data for scientific disciplines related to climate model development indicate that the pipeline for climate model developers is not flowing robustly. The upper two panels show the trend in Ph.D. degrees (first panel) and bachelor’s degrees (second panel) awarded over the past decade, showing no increase for the category of “Earth, Atmospheric, and Ocean Sciences.” The third panel shows the percentage of females awarded doctorates over the past decade, showing relatively low percentages (less than 50 percent) for several fields related to climate modeling. The fourth panel shows the percent-
age of bachelor’s degrees awarded in geosciences by ethnicity, showing the relative lack of diversity; the percentages are similar at higher education levels. For these figures, even though only one set of data is shown, the trends among doctorates, master’s degrees, and bachelor’s degrees are all similar. SOURCE: NSF, Division of Science Resources Statistics, special tabulations of U.S. Department of Education, National Center for Education Statistics, Integrated Postsecondary Education Data System, Completions Survey, 2000-2008.
Finding 7.2: From the limited data available, it is surmised that the pipeline of climate modelers being trained is not growing robustly in overall numbers or in diversity within the United States.
Climate model development is a challenging job. It involves synthesizing deep and broad knowledge, working across the interface between science and computational algorithms, and working well in a team. Thus, it is important to hire the most talented available people into this field. One obstacle to getting more students who are interested in climate science and other related fields to go into climate model development comes from the current incentive system for U.S. early-career scientists, which heavily favors those who produce more first-authored peer-reviewed journal articles. Students are disinclined from undertaking model development projects because of the fear of a “black-hole syndrome” wherein climate model development projects take long time periods (longer than a typical Ph.D. length) and do not result in many journal publications.3 This is a systemic issue within the field—the credibility of climate change science is heavily dependent on the fidelity of the climate models used, and yet the process of improving such models is often not particularly rewarding to a young scientist’s career. For example, a scientist who spends 2 years analyzing the output of existing simulations and writing papers on the findings may be more likely to advance in his or her career than a scientist who spends 2 years working on the details of a physical parameterization in a model. From the perspective of the entire field the efforts of the second scientist may well be more important in the long run, and yet the personal rewards will likely accrue more to the former scientist. One mechanism to combat this bias would be an enhanced recognition and reward system for climate model computer code writing and for the production of modeling data sets, including the recognition of such effort through stronger requirements for citation and coauthorship, both within modeling institutions and by academic users and collaborators.
A significant challenge is the entraining of top students interested in software engineering or computational science to work on developing climate models in comparison to other career tracks. Promising young computer programmers may have other more lucrative career opportunities at large software companies or startups. Climate modeling groups must compete by marketing relatively stable career tracks and the opportunity for stimulating cross-disciplinary interactions with a variety of scientists. A
3 This is a conclusion largely drawn from anecdotal evidence; further quantification is needed to determine how pervasive student bias is against tackling model development projects.
positive step is the development of computational (as opposed to computer) science programs at a number of U.S. universities that provide applied training well suited to a career in computational aspects of climate model development.
Finding 7.3: The current professional recognition system that heavily weights journal publications is a barrier to entraining more young scientists into climate model development.
The European Centre for Medium-Range Weather Forecasts as an Example
One potential example of how to entrain more people in climate model development work is from the European Centre for Medium-Range Weather Forecasts (ECMWF). ECMWF is an intergovernmental organization supported by 34 countries dedicated to operational medium- and extended-range forecasts combined with an extensive scientific research program. In these roles of operations and research, the Center employs about 150 staff members and 80 consultants coming from member and cooperating states. Rather than strictly a technician staff, members and consultants are generally highly reputed scientists who not only serve to deliver the end-use product to the member institutions, but also employ the supercomputing facilities and myriad data services of cooperating institutions to provide cutting-edge science with regard to the Center’s many research projects. The utility of the modeling is more directly coupled with research and provides justification for the latter. ECMWF appoints model developers for 5-year terms, which is longer than typical research grant cycles in the United States (3 years). ECMWF offers strong incentives to attract top scientists such as access to excellent facilities, excellent tools (e.g., the best numerical weather prediction [NWP] model in the world), and high, tax-free salaries.
THE WAY FORWARD
In order to ensure a capable and robust workforce in climate modeling in the future, it is crucial that young scientists and software engineers entering this field be appropriately trained and have highly attractive career paths. This involves a partnership between funding agencies, universities, and national laboratories. Universities can offer innovative coursework, degree pathways, and research opportunities for students and postdoctoral researchers combining climate and computational science. National laboratories can also host postdoctoral researchers and partner with universities in graduate student training. They can create stable career paths and change the current system of professional recognition and incentives to favor important contributions to
team projects with long development cycles and more risk. In order for these efforts to succeed, funding agencies will need to nurture training activities and provide adequate opportunities for stable funding to those who choose climate model development careers both in national laboratories and in academia, so that the best scientists and engineers do not seek greener pastures.
As described above, there are limited data on the existing climate model development workforce or future needs. More information on gaps in the workforce pipeline and future workforce needs could help inform better planning by universities, national laboratories, and funding agencies.
Recommendation 7.1: The United States should attempt to entrain top students into choosing climate model development as a career by providing more graduate and postgraduate training opportunities, enhanced professional recognition and career advancement for participation in climate model development projects, and adequate incentives to attract software engineers who could also choose private-sector careers.
Recommendation 7.2: In order to assess future needs on the climate model development workforce, the United States should obtain quantitative information about the workforce needs and required expertise base to support climate modeling.