When you curse the weatherman for wrongly predicting an afternoon of clear skies, take pity. Forecasting is hard: It’s subject to tons of variables, conditions are constantly changing, and everything affects everything else. But soon the whims of the weather gods will seem a little less arbitrary: The agency that runs the National Weather Service is working on making its predictions much, much better.
In the past couple of years, the National Oceanic and Atmospheric Administration has been planning a push to overhaul the National Weather Service’s quarter-century-old system. And on Wednesday, it took its first big step: It chose a framework for a shiny new weather prediction model. Eventually, the agency hopes, its updated model will detect severe storms weeks earlier, improve forecasts eight to ten days out, and firm up predictions for hurricane tracks. And that would change everything from event planning to giving people enough time to evacuate areas before they’re blasted by the elements.
Sounds great, right? But not everyone’s happy about the model NOAA picked, an in-house solution called FV3 (for Finite-Volume on a Cubed-Sphere). The agency also considered another commonly-used model called MPAS (Model for Prediction across Scales), developed by scientists at the National Center for Atmospheric Research in Colorado. Some scientists who use MPAS in their work think it’s the stronger system, and the resulting debate is making scientists seriously consider what future weather prediction systems should prioritize.
Modeling the weather is a byzantine business, mostly because the atmosphere is so complex. So weather models start by breaking the globe up into manageable chunks and solving equations within each parcel. The shape of those parcels matter, says Colin Zarzycki, a scientist who models extreme weather at NCAR. Old models use latitude and longitude lines, which messed with their ability to predict weather at the poles where the lines converge. To solve that problem, both of the new models break up the atmosphere into roughly equal sections: MPAS uses hexagons, while FV3 is more of a warped spherical cube with an overlaid grid.
But the two geometries come with different merits. “The hexagons in MPAS let us add resolution much more gracefully,” says Cliff Mass, an atmospheric scientist at the University of Washington and a vocal critic of the FV3 model. If you zoom into one of the six-sided areas, it’s easy to break them up into smaller, individually analyzable sections. That makes it useful for detecting things like thunderstorms, which are too small to be seen by the current model.
On the flip side, NOAA’s in-house model FV3 is more efficient and ran faster in tests, says Bill Lapenta, the director of NOAA’s National Centers for Environmental Prediction–qualities that are important when you’re making weather predictions for the entire world. But Mass argues that, if you’re looking toward a future of higher and higher-res weather modeling, MPAS is still the better choice. “The atmosphere has one modeling system: nature,” he says. “And that’s the way we have to do it.”
So whither weather modeling? Both the models are essentially just skeletons, ways of setting down basic rules of the atmosphere so supercomputers can run equations on it. Now that it’s decided on a model, NOAA needs to put meat on the framework, with equations that model the physics of clouds, rainfall, and radiation. (Plus, testing and plugging millions of data points from satellites into the model itself.) That’s where most of the work lies.
That buildout process introduces another problem. Lots of atmospheric scientists in academia use the rejected MPAS model, and its developers have been great at sharing ideas and supporting researchers with documentation, says Mass. NOAA has been … less forthcoming. “The question is whether NOAA will have the wisdom not to do this alone, but to build it with the community,” Mass says. That’s a priority, NOAA’s Lapenta says.
And not just for kumbaya reasons. Meteorologists are worried that the US is falling behind in weather prediction, ever since the model at the European Center for Medium-Range Weather Forecast predicted the path of Hurricane Sandy much more accurately than the National Weather Service’s model. But Zarzycki sees the debate as a positive development. “It’s renewed dedication to the United States’ weather forecasting resources,” he says. Discussing the weather: It’s not just for small talk anymore.