The Washington PostDemocracy Dies in Darkness

Leading scientists say NOAA plan to improve U.S. weather forecasting is incomplete

Category 5 Hurricane Michael at landfall in Florida. (University of Wisconsin-Madison)

The latest U.S. government effort to better the nation’s weather forecast modeling may not be adequate to achieve its intended goal, according to leading scientists in the weather community. The effort, some say, lacks the necessary push for innovation.

Others say it does not provide the necessary investments in science for the United States to regain its status as the world leader in weather prediction.

On March 20, the National Oceanic and Atmospheric Administration put out a request for proposals to partners, with a $45 million reward, to help it improve its weather forecast computer modeling. But some prominent voices in the weather community worry the scope of the request is too narrow for any organization to realistically help the government reach its goal.

Over the past few decades, NOAA’s flagship prediction model, known as the Global Forecast System, or GFS, has not kept up with forecasting gains made by several other international institutions, even though it has improved. Popularly known as the American model, its accuracy ranking recently slipped to fourth place, continuing to trail models in Europe while also falling behind Canada’s main model.

The Trump administration has stated that improving U.S. weather modeling is a “top priority.” In February, NOAA announced it was spending $505 million over eight years to buy supercomputers, tripling the agency’s capacity to support its modeling needs. The upgraded computing power is expected to be on par with that of the top international modeling centers.

NOAA to triple supercomputing capacity in bid to dramatically improve forecast accuracy

To harness the power of these computers, NOAA released a March 20 request for proposals to find a partner outside the government to design and build what it calls the Earth Prediction Innovation Center.

EPIC is intended to be a “community modeling” effort, drawing in talent and expertise from universities and private companies to accelerate model improvement. A long-standing criticism of NOAA’s modeling efforts is that they have been too insular, not sufficiently engaging the outside community.

Trump administration has EPIC plan to develop the world’s smartest weather forecasting model

NOAA announced that the winning bid will receive up to $45 million over five years to work with the agency in creating “the world’s most accurate and reliable operational weather forecast model,” according to a news release.

But several voices in the weather community have expressed concern that the request for proposals is too focused on technical matters related to computing and will not allow the award recipient to advance the science underpinning the modeling.

The request “is mainly about the ‘plumbing’ of numerical weather prediction and supporting outreach, with little mention of scientific innovation,” Cliff Mass wrote in an email. Mass is a professor of atmospheric science at the University of Washington and a longtime critic of NOAA’s modeling approach.

“There is … no mention of how to bring the weather research community together, nor any requirement to demonstrate progress in providing the nation with better weather forecasts.”

“I hope that everyone remembers that the 'I' in EPIC is for innovation, and facilitating that may be its most important function,” Gary Lackmann, a professor of atmospheric science at North Carolina State University who has advised NOAA on modeling efforts, wrote in an email.

“So, while I recognize the need to discuss computing, the cloud, and software engineering, I do think greater emphasis is needed on the mechanism for reaching out to the community to draw on the potential innovations that are out there.”

NOAA’s response to this feedback is that the request is intended to identify a partner who can help it establish and sustain a process for tapping outside innovation but isn’t about conducting the innovation itself.

“The role of EPIC is to manage the process by which external innovation is integrated,” NOAA spokesman Chris Vaccaro wrote in an email.

In public forums, NOAA acting administrator Neil Jacobs, the architect of EPIC, has made this point, telling partners a computing infrastructure must first be established before innovation is brought in.

Weather community leaders say that if scientific innovation is not a focus of the request for proposals, it must come next and NOAA and Congress should strongly back it.

“NOAA must increase its support for R&D [research and development] on weather and climate prediction, in order to be able to take best advantage of EPIC’s software developments,” wrote Jim Kinter, who directs the Center for Ocean-Land-Atmosphere Studies at George Mason University.

Fred Carr, professor emeritus of meteorology at the University of Oklahoma, wrote that the computing aspect of EPIC is “absolutely a vital component.” But he noted the request for proposals provides no funding for the underlying science, so “it is also vital that Congress continues to support the NOAA programs that fund such R&D.”

In an email, Tony Busalacchi, president of the University Corporation for Atmospheric Research, called the request “just the first step” in the government’s effort to recapture U.S. supremacy in weather forecast modeling.

Weather is turning into big business. And that could be trouble for the public.

But some are hopeful the request for proposals will lay the foundation.

“I look forward to seeing EPIC rise from this process and am optimistic it will be a major milestone in improving weather and climate services for the nation,” Peter Neilley, director of weather forecasting services at IBM, wrote in an email.