The Washington PostDemocracy Dies in Darkness

Air Force’s plan to drop U.S. forecast system for U.K. model draws criticism

Senior Airman Michael Richardson takes weather readings with a TMQ-53 tactical weather sensor during training at Fort Irwin, Calif. (Val Gempis/U.S. Air Force)

The U.S. Air Force Weather Agency, which provides forecasts for Air Force and Army missions around the world, plans to replace its U.S.-based forecasting system with a model from the United Kingdom.

The U.K. model selected by the Air Force, known as the Unified Model of the United Kingdom Met Office, is widely respected. The Air Force says it will improve its forecast capabilities and lower its costs. Within a single framework, this model is able to provide both short- and longer-range forecasts over large and small areas — which is not a seamless operation within the current U.S. system.

But the decision — which was made without coordination with the National Weather Service or U.S. Navy, who partner with the Air Force to improve predictions — has drawn criticism from parts of the U.S. weather research and forecasting community. Several leading figures say they are perplexed that the Air Force has selected a foreign model when the United States is investing substantial resources to develop its own world-class models.

An Air Force memo obtained by The Washington Post (see “Memo 1” at the bottom of this post) dated March 30 states it will adopt the U.K. model in 2016 and replace the Weather Research and Forecasting (WRF) model developed by the University Center for Atmospheric Research, a federally funded laboratory in Boulder, Colo.

“We’ve been notified by the Air Force that we’re going to sustain cuts and that they are now going to go with a foreign model,” said Scott Rayder, senior adviser to the president at the University Center for Atmospheric Research. “To say we’re disappointed is an understatement. I am confused as to why they’re subsidizing the competition and not supporting the U.S. weather modeling community.”

The cuts to the WRF model, used to forecast weather systems in the United States and around the world, represent a setback for U.S. weather prediction, said Cliff Mass, professor of atmospheric sciences at the University of Washington, who has used, tested and published research on the model.

The WRF model is used and supported through a partnership involving multiple organizations, including the Air Force, National Weather Service, Navy, Federal Aviation Administration and universities.

Mass said not only is the Air Force cutting its financial support for key U.S. weather research interests but also “intellectually isolating” itself from the research community and “abandoning” a cooperative effort.

“The Air Force was the main supporter of WRF,” Mass said. “The loss of Air Force funding will seriously compromise WRF development and support, undermining the main U.S. community mesoscale prediction system.”

Mass added: “It is a national embarrassment when the U.S. Air Force feels the need to take on a foreign modeling system. This was an ill-advised decision that is harmful to U.S. interests.”

The Air Force has insisted its action supports a more seamless forecasting operation and saves money.

“U.S. involvement enhances the U.K. model, and U.K. model efforts enhance the Air Force capabilities,” Air Force spokesperson Ed Gulick said in an e-mail to The Washington Post. “Our actual cost is only for a software license, which costs $100,000 annually and is significantly less than the $2 million we currently spend annually on our Air Force model,” which is based on the WRF model.

But Rear Adm. (ret) David Titley, who previously served as oceanographer and navigator of the Navy and had oversight of its global weather operations, expressed some skepticism on the cost savings. “If the U.S. Air Force believes they can meet global weather forecasting requirements with a $100,000 per year investment and have that system responsive to their needs, they may be in for a rude surprise,” he said.

An Air Force memo dated Nov. 19, 2014, (see “Memo 2” at the bottom of this post), signed by Ralph Stoffler, director of weather, emphasized the advantages of adopting a model already used by the United Kingdom and several other U.S. allies. “We can enhance our interoperability and success ensuring ‘One Operation, One Forecast’ by using the same model as some of our coalition and international partners,” the memo said.

The U.K. Met Office Web site notes that government weather services in Australia, Korea, New Zealand and South Africa use its model.

The November Air Force memo also extolled the U.K. model’s forecasting prowess, noting it “consistently” outscored many other models “across a range of performance characteristics.”

Neil Jacobs, chief atmospheric scientist at Panasonic Avionics Corp. and chair of the American Meteorological Society’s Forecast Improvement Group, called the Air Force’s move a “logical step.” He noted the U.K. model is well-documented, has a “proven data assimilation package” (meaning it can successfully bring in data from different sources, which helps performance), and can be used on a variety of types of computers. Importantly, Jacobs said, it can produce both a global forecast, as well as detailed regional forecasts.

Though Jacobs praised the technical components of the U.K. modeling system, he expressed concern about the impact of the Air Force diverting resources overseas.  “Without funding or future revenue prospects, the incentive for model development [in the U.S.] is nonexistent,” he said.

The Air Force’s Gulick stressed the U.K. model provides “military-unique capabilities, such as dust and aerosol forecasts, vital to the safety of Air Force and Army personnel engaged in military operations around the globe.”

But Titley, who called the Air Force’s action “very disappointing,”  questioned the U.K. model’s added value given currently available Department of Defense forecasting assets. “The Navy (and therefore DoD) have spent millions of dollars over the past couple of decades developing dust and aerosol transport models that run in classified and fully IT-certified environments,” he said. “It is very difficult to believe the Air Force can not and does not take advantage of this existing Department of Defense capability.”

Both the Navy and the National Weather Service said the Air Force’s decision to adopt the U.K. model was made unilaterally, even as the Air Force is party to an agreement with them to coordinate activities and work toward a National Unified Operational Prediction Capability.  This capability fits into the  larger  national Earth System Prediction Capability (ESPC) Inter-Agency program to advance both weather and climate modeling, which also includes the Department of Energy, NASA, and the National Science Foundation.

“I want them [the Air Force] as a partner to build the best possible U.S. model,” said Rear Adm. Jonathan White, the current oceanographer and navigator for the Navy. “I don’t understand how their decision to go with the U.K. model supports that goal.”

The National Weather Service had no specific comment on the Air Force’s decision, but Chris Vaccaro, a spokesperson, said sound weather-based decisions rely on an approach in which a collection or “ensemble” of models is used, “not the exclusive dependency on just one model.”

The National Weather Service announced in January that in 2015 alone, its supercomputing power for forecasts would increase by a factor of 10 resulting from a $44.5 million investment with IBM. Vaccaro said this investment will improve its global, regional and localized forecasts “to better meet the needs of a wide range of partners and customers, including those in the Department of Defense.”

The Air Force said it will continue to “leverage” National Weather Service models. “The availability of the U.K. and the NOAA [National Weather Service] models in the Air Force inventory reduces our cost, improves our technical capability, and enhances Coalition operations,” Gulick said.

The Air Force Weather Agency, referred to as the 557th Weather Wing, provides forecasting support for Air Force and Army operations around the globe. It is made up of over 1,700 airmen.

Memo 1

Memo 2