A National Strategy for Advancing Climate Modeling
Climate modeling isn’t simply a scientific exercise but a crucial tool that can inform planning and decisions for issues with enormous economic ramifications, potentially billions of dollars’ worth.
Insurers, for example, use climate projections to understand the risks from future hurricanes, or to figure out how sea level rise will affect flood zones. Farmers depend on climate modeling to decide which crops to grow and when to plant and harvest them. City planners, architects and builders use climate information to plan infrastructure to withstand potential snow, heat and extreme weather events, and public health officials and emergency responders use climate predictions to anticipate threats like heat waves.
In the past, records of historical weather patterns could be used to predict future climate conditions. However, as the effects of climate change become more apparent, past conditions will no longer serve as reliable predictors of future climate events. Climate change could increase the likelihood and severity of extreme weather and climate conditions, such as regional droughts and extreme flooding, events that have caused hundreds of billions of dollars in damage over the past few decades alone.
To enable society to respond and adapt to these changes, it will become increasingly important to develop climate models that can accurately simulate past and present climate, as well as project future climate. While climate models have made enormous progress in the past several decades, there is still a long way to go in order to deliver the kind of climate information that decision-makers desire. A panel of 17 leading climate experts assembled by the National Research Council (NRC) recently outlined its vision for a more unified U.S. climate modeling enterprise in ”A National Strategy for Advancing Climate Modeling.” The report offers an in-depth, independent analysis of the current state of climate modeling in the United States, and a comprehensive strategy for improvement. Panel members served as volunteers and conducted their analysis independently from the agencies that requested the study, which included the U.S. Department of Energy, National Oceanic and Atmospheric Administration (NOAA), National Science Foundation, NASA, and the U.S. Intelligence Community.
The panel recommended that building a national strategy for climate modeling — including a community-led evolution to a common software infrastructure shared by all climate researchers, and an annual climate modeling forum to facilitate communication between climate modelers and the users of climate data — would help the United States meet growing demands for accurate climate data.
Challenges Facing Climate Modeling
Over the past several decades, enormous progress has been made in developing reliable climate models. However, the NRC panel found that further advances are needed to deliver climate projections on the scale and level of detail desired by decision-makers.
The first challenge to advancing climate models is a technical one. The climate information that decision-makers desire often requires climate model projections at higher spatial resolutions, and on more specific time scales than are currently available.
There are limits to how reliable climate projections can be, but a proven strategy for gradually improving climate models has been to use finer grids (the units of climate models that contain information including physical and climate characteristics on a given location) and add in new processes of concern to users, such as interactions between climate change and ecosystems. But, to produce this sort of specific, high-resolution information, climate scientists need more powerful computing hardware and the software to use it efficiently. Rather than simply developing faster computer chips, it is likely that future climate modeling hardware will require connecting far more computer chips in parallel — a different hardware infrastructure than the one currently in use. It is expected that such hardware systems will become available over the next 10 to 20 years, but it will take significant effort to develop climate-modeling software that can take full advantage of this new hardware.
In addition to resolving these technical challenges, improving future climate modeling will require greater collaboration among scientists and decision-makers. The U.S. climate modeling community is diverse, consisting of several large global climate modeling efforts and many smaller groups running regional climate models. This diversity allows multiple research groups to tackle complex climate modeling problems in parallel, enabling more rapid progress, but it also can lead to duplication of effort and make it more difficult to prioritize limited human and computational resources. Promoting unification in some aspects of the U.S. climate modeling enterprise could enable more efficient, coordinated progress. This does not mean establishing only one U.S. center for climate modeling; instead, different climate modeling institutions could pursue their own methodologies while working within a common modeling framework in which software, data standards, and tools are shared by all major modeling groups nationwide.
Steps to a National Strategy
The report outlines a national strategy to help the United States move toward the next generation of climate models. Key steps include working toward a common software infrastructure, promoting increased collaboration through a national climate modeling forum, nurturing a unified weather-climate modeling effort, and developing a program to support those who must interpret climate modeling data to make decisions.
Evolve to a Common Software Infrastructure
The U.S. supports several climate models, each conceptually similar but with components assembled with slightly different software and data output standards. Gradually evolving to a shared software infrastructure for building, configuring, running, and analyzing climate models could help scientists make the transition to more complex computer hardware. A common software infrastructure also would help scientists compare and interchange climate model components, such as land surface or ocean models, and would be an effective mechanism for probing uncertainties in abilities to simulate the climate system. It also might facilitate sharing of well-accepted model components across the U.S. modeling community, freeing resources to address other critical topics.
The panel concluded that the best pathway for achieving a common software infrastructure involves a community-based decision process in which individuals, institutions, and managers all see advantages, which may be scientific, computational, or resource driven, to the new system. Efforts from above to dictate transformations to specific infrastructures are likely to meet with less success than those that have come from the bottom up.
Convene a National Climate Modeling Forum
Currently, those who conduct climate modeling and those who use climate projections to make decisions are often disconnected from each other. Members of various climate-impacted fields and climate-related scientific disciplines often learn about progress through specialty conferences and scholarly journals, which can be slow, haphazard, and inefficient in communicating advances from different fields.
The NRC panel recommended holding an annual climate modeling forum to help inform the entire climate science community of current and planned activities at core modeling centers, to provide a venue for discussing priorities for the national modeling enterprise, and to bring disparate communities together to design common modeling experiments. Such a forum would provide climate model data users with an opportunity to learn more about the strengths and limitations of models and provide input to modelers on their needs.
Nurture a Unified Weather-Climate Modeling Effort
Many physical and chemical processes, from the formation of ice crystals in clouds to the circulation of ocean currents, can affect both climate and weather. For example, small cumulus clouds driven by daytime heat can trigger large thunderstorms, but can also affect the climate system over decades by changing the reflection of sunlight from Earth’s surface. Because climate varies over such long periods, it takes longer to collect observational data to test the models thoroughly. Using processes that vary on weather timescales, such as cloud cover, to test climate models can advance both weather and climate modeling. The committee concluded that the United States should nurture a unified weather-climate model as part of the national portfolio of models. Any efforts to unify weather modeling and climate modeling would be most successful if they involved collaboration among operational weather forecast centers, data assimilation centers, climate modeling centers, and the external research community.
Develop a Program for Climate Model Interpreters
The needs of climate data users are diverse and complex. Some users, such as farmers or town planners, require information about their local region, while others, such as international development organizations, require climate data on a global scale. Meeting these varied needs involves ensuring that the ever-expanding volumes of climate data are easily and freely available, and that these data are useful and easily understandable to all users. This often entails communicating the uncertainty that is inherent to all climate models.
It is a challenge for the climate modeling community to work directly with the broad array of climate model users. Although there are already organizations using climate model outputs and translating them into products to meet user needs, there are no recognized mechanisms for verifying the quality of the information provided. Developing a national education and accreditation program to train climate model interpreters to use technical findings and output from climate model in a range of applications could help ensure the accuracy and appropriateness of climate information, as well as help communicate users’ needs to climate model developers.
The panel identified areas where work is already underway to advance a more unified climate modeling enterprise. Continuing these efforts will help support the success of the national climate modeling strategy.
Sustain State-of-the-Art Computing Systems for Climate Modeling
To increase computing and data capacity, the report suggests a two-pronged approach that involves the continued use and upgrading of existing climate-dedicated computing resources at modeling centers, and research to support the transition to the more complex computer hardware systems expected over the next 10 to 20 years.
Another option is building a national climate computing facility, but this would be expensive, could divert resources away from other critical climate-modeling investments, and would only be beneficial if the current level of investment in computing capabilities at climate modeling centers also was maintained.
Continue to Contribute to a Strong International Climate Observing System
Sustained observational data on factors such as temperature, precipitation, clouds, snow and ice, and ecosystem change is critical for advancing understanding of the processes that drive the climate system. Maintaining a climate observing system is an international enterprise, but requires strong U.S. support that has come under threat due to funding cuts. For example, the number of current and planned Earth-observing satellite missions will decline by more than a factor of three by 2020. Over the next several decades, it will be important to maintain existing long-term datasets of essential climate variables, and to launch innovative new climate measurements that help characterize Earth system processes.
Develop a Training and Reward System for Climate Model Developers
Model development is among the most challenging tasks in climate science because it requires knowledge of climate physics, numerical analysis, and computing, as well as the ability to work effectively in a large group. Graduate fellowships in modeling centers, extended postdoctoral traineeships of three to five years, and rewards for model advancement through well-paid career tracks could help entice high caliber computer and climate scientists to become climate model developers. In addition, a method to enhance the professional recognition of climate model developers and gauge their progress would be an enhanced citation and co-authorship requirements for model computer code and the production of modeling data sets.
Enhance the National Information Technology Infrastructure that Supports the Sharing and Distribution of Climate Modeling Data
Ever larger amounts of climate model and observational data are being generated. Facilitating broad access to these data for researchers, data users, and decision makers is challenging but increasingly important. Beyond stabilizing support for current data infrastructure efforts, the committee recommended that the United States develop a national information technology infrastructure that builds on existing efforts to facilitate and accelerate data display, visualization, and analysis for experts and the wider user community.
Continue to Pursue Advances in Climate Science and Uncertainty Research
To meet national needs for improved climate information over the next several decades, U.S. climate modelers will need to address an expanding breadth of scientific problems while striving to make predictions and projections more accurate. Progress toward this goal can be made through a combination of increasing model resolution, advances in observations, improved model physics, and more complete representations of the Earth system. As a general guideline, priority should be given to climate modeling activities that focus on addressing societal needs and where progress is likely, given adequate resources.
Research on understanding and quantifying uncertainty would help climate modeling efforts support decision making. Specifically, research should help determine how to use observational records to better understand and quantify model uncertainty in projections of future climate change, how to incorporate quantifications of uncertainty more fully into the climate modeling process, and how to communicate uncertainty more effectively to users of climate models and decision makers.
Established in 1916 by the National Academy of Sciences, the National Research Council seeks to improve government decision making and public policy, increase public understanding, and promote the acquisition and dissemination of knowledge in matters involving science, engineering, technology, and health.
The report’s authoring committee includes: Chris Bretherton (Chair), University of Washington; V. Balaji, Princeton University; Thomas Delworth, Geophysical Fluid Dynamics Laboratory, Princeton; Robert E. Dickinson, University of Texas, Austin; James A. Edmonds, Pacific Northwest National Laboratory; James S. Famiglietti, University of California, Irvine; Inez Fung, University of California, Berkeley; James J. Hack, Oak Ridge National Laboratory; James W. Hurrell, National Atmospheric Research Center; Daniel J. Jacob, Harvard University; James L. Kinter III, Center for Ocean-Land-Atmosphere Studies; Lai-Yung Ruby Leung, Pacific Northwest National Laboratory; Shawn Marshall, University of Calgary, Canada; Wieslaw Maslowski, U.S. Naval Postgraduate School; Linda O. Mearns, National Center for Atmospheric Research; Richard B. Rood, University of Michigan; Larry L. Smarr, University of California, San Diego; Edward Dunlea (Study Director), Katie Thomas (Associate Program Officer), Rob Greenway (Program Associate), Rita Gaskins (Administrative Coordinator), April Melvin (Christine Mirzayan Science and Policy Fellow), Alexandra Jahn (Christine Mirzayan Science and Policy Fellow), Solmaz Spence (Communications Staff), Anne Johnson (Communications Staff), National Research Council.
The panel’s report, “A National Strategy for Advancing Climate Modeling,” can be found at dels.nas.edu/Report/National-Strategy-Climate-Modeling/13430. This site includes a recorded webinar with the committee chair, Chris Bretherton, presenting important highlights from the report.
Reports on related topics can be found at dels.nas.edu/basc.
Information from the National Academies on climate change can be found at nas-sites.org/americasclimatechoices/.
Lastly, the report’s authoring committee created a website to attempt to explain to a lay audience what climate modeling is and why it is important: nas-sites.org/climatemodeling/.
Edward Dunlea is a senior program officer at the National Research Council of the National Academy of Sciences with the Board on Atmospheric Sciences and Climate. He was the study director for the NRC report on “A National Strategy for Advancing Climate Modeling.”
Chris Bretherton is a professor in the University of Washington departments of Atmospheric Science and Applied Mathematics, and he is the former director of the University of Washington Program on Climate Change. He was the committee chair for the NRC report on “A National Strategy for Advancing Climate Modeling.”