May 26, 2019 - Science

America's upgraded weather forecast model still lags behind Europe

A man walking along the Tidal Basin in Washington with an umbrella

A man walks along the Tidal Basin on the National Mall in March. Photo: Brendan Smialowski/AFP/Getty Images

The National Oceanic and Atmospheric Administration is about to roll out the new version of its main weather forecasting model next monthbut it won't help the agency gain much ground against its international rivals for the title of having the world's most accurate weather model.

Why it matters: Weather forecasting today relies on numerical prediction models that simulate the current and future state of the atmosphere. If the most commonly used computer model is off target during high-impact weather events, it can affect the larger economy and possibly even cost lives.

  • NOAA, for example, has estimated that 3%–6% of U.S. GDP is sensitive to weather variability.

Driving the news: The new version of the Global Forecasting System — a model known as the GFS-FV3 — will roll out in mid-June, acting NOAA administrator Neil Jacobs tells Axios. That's after an unexpectedly lengthy trial run in which problems were discovered in how the model handles snowfall projections.

While it should result in more accurate projections, testing shows it's unlikely to make up much ground against two of NOAA's forecasting rivals: the European Centre for Medium Range Weather Forecasts and the U.K. Met Office.

  • The U.S., in other words, will still be in 3rd place even after the upgrade, Jacobs says.
  • However, Jacobs, who previously led the weather modeling group at Panasonic Avionics Corporation, says it's what is under the FV3's hood that counts. He says it will mark the beginning of a shift "that will drastically improve the skill” of U.S. forecast models from here on out.

Background: The public narrative that U.S. forecasting models are inferior have their roots in Hurricane Sandy, which struck the East Coast in 2012. The GFS failed to anticipate the storm's westerly turn into the New Jersey coast several days in advance, while the European model advertised that historically improbable outcome a week out.

  • Ever since, NOAA, working with Congress and the White House, has been trying to obtain the resources needed to catch up to other countries.

A big hang-up so far has been how the U.S. spread out its model development across agencies and research institutions without centralized oversight, including among centers within NOAA that did not collaborate despite being in the same agency.

What's next: Jacobs has big plans for changing how NOAA develops its computer models, with the goal of shifting code out of the agency and into the commercial cloud, where it can be used by university researchers and the private sector to help improve it and spin off new products.

  • “We will set ourselves up to essentially crowdsource forecast model development work,” Jacobs says, noting that before, NOAA built models that could only be run on its own machines.

NOAA also suffers from the perennial plight of every public-facing meteorologist: No one recognizes the agency when it gets the forecast right, but they are quick to point out when they screw up.

  • Jacobs says other weather agencies, including the ECMWF, are better at touting their forecasting skill: “We don’t claim the wins. We’re not actively out there saying look how good we did."

Go deeper: 5G spectrum and the potential weather forecasting apocalypse

Go deeper