A Silicon Valley startup that specializes in artificial intelligence (AI)-assisted weather forecasting could deeply impact the business of predicting the weather following deep budget cuts to the National Oceanic and Atmospheric Administration (NOAA), and the looming threat of hurricane season.
WindBorne, started by a group of Stanford University graduates in 2019, is increasingly replacing NOAA weather balloons with less expensive, AI-powered alternatives that provide valuable data by keeping its balloons in the sky for longer stretches of time through a combination of off-the-shelf computer parts and software. NOAA balloons operated for just a day before popping and falling to Earth.
Expect things to pick up even more for the upstart company. NOAA is reeling from proposed 28% budget cuts, or $2 billion, to its budget, necessitating growing dependence on WindBorne. The National Weather Service, for example, buys sensor readings from WindBorne each month.
The company’s current armada of 300 balloons are launched monthly from 10 sites, flying an average of 12 days, where their satellite antennas capture and record a comprehensive slate of real-time data such as temperature and humidity from the upper atmosphere.
Each balloon comes with tiny motors that control the release of sand from a small bag onboard to control the balloon’s elevation and enter different wind currents. AI determines which elevation to reach to collect the best data.
Because its balloons are airborne longer, and collect more data through AI modeling and deep learning, WindBorne claims it’s day-ahead temperature forecasts are 37% more accurate than those from the European Center for Medium-Range Weather Forecasts, the European equivalent of the National Weather Service.
WindBorne says it plans to expand its fleet of balloons, already considered the best AI-based weather prediction model, to 10,000 to cover the planet.
The company’s goal is nothing short of earthshaking for the weather business: Make long-term forecasts as useful as tomorrow’s weather report. Indeed, the five-day forecast is as accurate today as the three-day forecasts 40 years ago. Ultimately, the 10-day forecast may do the same soon.
A more robust picture of future weather, even for a few days, could be a game changer and life saver. Recent flash floods in Texas underscore the importance of time in assessing risk from extreme weather events that have become more common with climate change. Researchers have concluded that shorter forecast lead times since 2009 have prevented hundreds of millions of dollars in hurricane damage per storm.
