Read our COVID-19 research and news.

Researchers use computer model outputs, such as this image from the Community Earth System Model, to study climate dynamics.

Researchers use computer model outputs, such as this image from the Community Earth System Model, to study climate dynamics.

Pacific Northwest National Laboratory

New U.S. climate model project getting cautious praise

Fifteen months ago, academic climate scientists expressed worries that a new climate model sponsored by the U.S. Department of Energy (DOE) was going to harm existing modeling efforts, siphoning off talent and funding when federally funded science is short on both. The biggest concern: that the Accelerated Climate Modeling for Energy (ACME) project, meant to forecast local impacts of climate change and to be used on DOE’s future exascale supercomputers, would dilute resources from the Community Earth System Model (CESM). That model, managed by the National Center for Atmospheric Research (NCAR) in Boulder, Colorado, is built on science and code continually developed by U.S. academics and DOE scientists and is partially funded by DOE. Besides, NASA and the National Oceanic and Atmospheric Administration, the federal space and ocean agencies, respectively, each have two other models; could creating a sixth U.S. climate model work?

The verdict so far, 6 months into the initial coding effort: mostly. DOE scientists have built ACME using CESM’s code, and the two projects are now engaged in a tandem effort that’s new and potentially good for both. Although the CESM team is still a bit uneasy, climate modeler Jean-François Lamarque of NCAR writes in an e-mail that “the potential benefits to the expansion of scope and modeling approaches offered by the two projects need to be recognized.” The “new collaborations” that the tandem effort has catalyzed have been “fruitful,” he added.

Earlier this month, DOE released an updated public document  spelling out the program’s aims and approach and hosted a public town hall at the American Geophysical Union meeting in San Francisco, California. There, scientists explained to ScienceInsider how the tandem effort to shape the two models has evolved. Think of CESM as the Swiss army knife, they said; it is built to incorporate hundreds of bits of physics and approximations from academics across the United States and is meant to serve everyone from regional modelers, bread-and-butter climate scientists, and even planetary scientists trying to model climate on other planets. In contrast, ACME is a specialized kitchen knife, focused on just three goals: modeling the water cycle, biogeochemistry, and the role of Earth’s cryosphere.

Academics that contribute to CESM and use the model in their research had several worries about DOE’s new model. First, because DOE funds some academic modelers and staff at NCAR, scientists worried that the announcement of the new model meant that DOE would cut its financial support for CESM. So far, that hasn’t happened. “Support for the development of the ACME model supports the development of CESM and vice versa,” explains David Bader, a modeler at Lawrence Livermore National Laboratory (LLNL) in California who runs the ACME project. So scientists funded to work at NCAR or in universities can develop code for both. Still, the situation seems tenuous, says Richard Neale, an NCAR modeler. “Resources are still in place from DOE. What will happen in the future, we don’t know.”

As DOE laboratories have gradually hired talented staff away from NCAR, another concern was that a new model would vacuum up valuable expertise that otherwise was needed for model development. But Neale says weekly conference calls are keeping the two teams in sync, and they’ve even created new means for sharing information on bugs and new features. “I think a couple of years ago there was some angst about the new program. Everybody’s a lot more comfortable with it these days,” said one modeler at LLNL, who requested anonymity because he was not authorized to speak to the media.

Maintaining that bonhomie may depend on whether the two modeling efforts will remain compatible into the future. “The risk of that not happening is certainly a concern to the overall CESM community,” Lamarque writes in his e-mail. And Bader warns that the models may ultimately diverge based on the computers they’re designed to run on. ACME is expected to run on DOE supercomputers, which will likely differ from those at the Wyoming Supercomputing Center, the main supercomputing tool for NCAR and CESM.