U.S. markets closed
  • S&P 500

    +116.01 (+3.06%)
  • Dow 30

    +823.32 (+2.68%)
  • Nasdaq

    +375.43 (+3.34%)
  • Russell 2000

    +54.06 (+3.16%)
  • Crude Oil

    +2.79 (+2.68%)
  • Gold

    -1.70 (-0.09%)
  • Silver

    +0.09 (+0.42%)

    +0.0034 (+0.33%)
  • 10-Yr Bond

    +0.0570 (+1.86%)

    +0.0009 (+0.07%)

    +0.2770 (+0.21%)

    +526.33 (+2.51%)
  • CMC Crypto 200

    +8.22 (+1.81%)
  • FTSE 100

    +188.36 (+2.68%)
  • Nikkei 225

    +320.72 (+1.23%)

How AI Can Make Weather Forecasting Better and Cheaper

  • Oops!
    Something went wrong.
    Please try again later.
  • Oops!
    Something went wrong.
    Please try again later.
  • Oops!
    Something went wrong.
    Please try again later.
·7 min read
In this article:
  • Oops!
    Something went wrong.
    Please try again later.
  • Oops!
    Something went wrong.
    Please try again later.
  • Oops!
    Something went wrong.
    Please try again later.

(Bloomberg) -- In early February a black box crammed with computer processors took a flight from California to Uganda. The squat, 4-foot-high box resembled a giant stereo amp. Once settled into place in Kampala, its job was to predict the weather better than anything the nation had used before.

Most Read from Bloomberg

The California startup that shipped the device, Atmo AI, plans by this summer to swap it out for a grander invention: a sleek, metallic supercomputer standing 8 feet tall and packing in 20 times more power. “It’s meant to be the iPhone of global meteorology,” says Alexander Levy, Atmo’s co-founder and chief executive officer. That’s a nod to Apple Inc.’s design cred and market strategy: In many countries, consumers who’d never owned desktop computers bought smartphones in droves. Similarly, Atmo says, countries without the pricey supercomputers and data centers needed to make state-of-the-art weather forecasts—effectively, every nation that’s not a global superpower—will pay for its cheaper device instead.

For its first customer, though, the Uganda National Meteorological Authority (UNMA), Atmo is sending its beta version, the plain black box. Prizing function over form seems wise for the urgent problem at hand. In recent years, Uganda has had landslides, floods, and a Biblical plague of locusts that devastated farms. The locusts came after sporadic drought and rain, stunning officials who didn’t anticipate the swarms. “It became an eye-opener for us,” says David Elweru, UNMA’s acting executive director.

Many nations facing such ravages lack the most modern tools to plan for the changing climate. Atmo says artificial intelligence programs are the answer. “Response begins with predictions,” Levy says. “If we expect countries to react to events only after they’ve happened, we’re dooming people to disaster and suffering.” It’s a novel approach. Meteorology poses considerable challenges for AI systems, and only a few weather authorities have experimented with it. Most countries haven’t had the resources to try.

Ugandan officials signed a multi-year deal with Atmo but declined to share the terms. The UNMA picked the startup partly because its device was “way, way cheaper” than alternatives, according to Stephen Kaboyo, an investor advising Atmo in Uganda. Kaboyo spoke by phone in February, Kampala’s dry season, as rain pelted the city. “We haven’t seen this before,” he said of the weather. “Who knows what is going to happen in the next three seasons?”

Uganda currently has more than 100 weather stations and three forecasting radars across the country. Last summer, Atmo began piping data from these sensors into its California office, a home in the Berkeley hills where Levy lives. Until recently his co-founder, Johan Mathe, lived there, too. Mathe, the chief technology officer, programmed software to crunch Uganda’s data in its black box, which, before being shipped off to Kampala, sat on the ground floor emitting a loud hum from its whirring processors.

Behind the box, Atmo had placed a large map of the globe, marking countries where it had signed deals or wanted to. A placard in the room read, “You don’t have to be crazy to work here; we’ll train you.”

Many tech startups find it crazy to wade through the red tape necessary to serve governments. But Atmo is intent on pitching public agencies. “The decisive organizations—the ones with the most to gain and lose—are governments themselves,” Levy says.

Computers were born to predict the weather. In 1950 the ENIAC, developed by the U.S. Army, the earliest digital computer, spent an entire day issuing the world’s first machine forecast. Over time, computers got faster. They used the traditional forecasting model of numerical weather prediction, which divvies up the Earth’s surface into a series of grids—cells for calculating temperatures, winds, humidity—then spits out a prediction.

But in these models, each zoom in on the grids requires an ever greater increase in computing horsepower. As satellite images added more snapshots of the planet in recent years, computers struggled to keep up, says Martin Schultz, a senior scientist at Germany’s Jülich Supercomputing Centre. Some of the world’s largest supercomputers are devoted to weather predictions; in 2020 the National Oceanic and Atmospheric Administration added two machines that brought its overall capacity to more than 40 petaflops, giving it more than 15,000 times the performance power of the latest Mac computer.

Artificial intelligence, in theory, can deliver on-par forecasts with less computing. Early research has shown progress in forecasting rainfall and “nowcasting,” predicting the weather over the next hour or two. About 20 “really serious” meteorological programs have started incorporating AI in the past five years, Schultz says. But machines have a much more difficult time predicting the weather than recognizing a photo or finishing a sentence. The atmosphere has a mess of volatile variables (turbulence, high-pressure systems, etc.) that must be forecast for different times over the next hour, for the day, and a week ahead. Experts say a fully accurate forecast beyond two weeks is impossible, even with the aid of AI.

Atmo reports that its early tests have doubled the accuracy scores of baseline forecasts in Southeast Asia, where the startup is pursuing contracts. Initial tests on the ground in Uganda correctly predicted rainfall when other systems didn’t, according to UNMA officials.

The flashiest part of Atmo’s pitch is its hardware. The startup shrank a weather data center into its new device, an L-shaped machine slightly longer than a Ford F-150 that packs 50 kilowatts of power. Atmo borrowed a liquid cooling technique developed for Bitcoin mining for its ventilation system. To blueprint the device, it hired Frank Stephenson, an acclaimed designer who worked on the Mini Cooper. He strove to make an object that would look as if it was from the natural world and some imagined world to come. “Did it come to Earth a million years ago?” he asks. “Or is it from a million years in the future?” Atmo says it hopes customers will display it outdoors like a sculpture.

Its computer is designed to work as a standalone device or in interlocking sets, depending on forecasting needs. UNMA has agreed to lease one Atmo device. The company also has a contract with 51 Degrees, a nonprofit in Kenya, as well as a regional bloc in East Africa. It’s also in talks with Kenyan officials.

Atmo’s success there may depend on its ability to diverge from Silicon Valley’s checkered history of parachuting into Africa with dreams that were soon aborted. The startup knows this well: Mathe and Anna Prouse, Atmo’s vice president, both worked at Loon, a Google project that provided rural internet service with balloons floating in the stratosphere. Loon cut deals with telecommunications operators in a few countries, including Kenya, but ran into issues settling on a business model and securing flyover rights from neighboring nations. It shut down in 2021.

Atmo is taking a more collaborative approach, says Prouse, who negotiated many of the Loon deals: “Officials talk to each other all over the world. So you do not want to mess up your first engagement.”

Better diplomacy and AI may only go so far. Compared with Western countries, most African nations collect relatively little atmospheric data and have limited bandwidth to process it. “You just don’t have enough observations to make good forecasts,” says Gregory Jenkins, a professor of meteorology and African studies at Penn State University. He questions the spending on fancy equipment over increased weather sensors: “Is this the best use of resources?”

Apuuli Bwango, the UNMA chair, says the agency is investing both in better data collection and in forecasting systems like Atmo’s. Climate change, he adds, isn’t giving the government much of a choice. “This is an imperative.”

Most Read from Bloomberg Businessweek

©2022 Bloomberg L.P.