Philosophical question: What is the smallest unit of weather? Is it the tiny breeze that forms between the temperature differential of midday pavement and a shaded gutter puddle? Perhaps. And what about the smallest predictable unit of weather? That answer might soon be answered by a powerful new weather forecasting tool.
IBM has combined its computing power with the Weather Channel’s impressive forecasting platform to create a new tool called Deep Thunder. Yes, the name is badass, and I encourage you to say it while speaking like Xerxes from 300. Deep Thunder can, its makers claim, forecast weather events that happen on the scale of a city block–a scale known as hyperlocal. But don’t plan on getting access to it unless you are a paying client. Deep Thunder forecasts for businesses who pay top dollar to know how the slightest changes in weather will affect their bottom line.
Weather forecasting is among the most demanding work in computing. “It’s a physics problem,” says Lloyd Treinish, IBM’s chief scientist. “How is energy exchanged at the surface, how does moisture convect and form into clouds, how does a thunderstorm form.” Each of those problems can be described as a set of equations. Of course, the closer you look at the weather, the more equations you have to perform to tease out all the effects of all the little interactions.
The Weather Channel had that stuff mostly licked when IBM bought the company last December. Well, not all the company–IBM left the cable channel behind. That last bit might be surprising, but the Weather Channel has been hit hard by cord-cutting millennials. “Five years ago TV was two-thirds of the company’s revenue, and that figure has almost fully flipped around,” says Mary Glackin, head of science and forecast operations for The Weather Company.
IBM had partnered with the Weather Channel before, but decided to buy the company outright because of its global weather data collection efforts. “There was a realization that weather underpins everything, in every part of the economy,” says Glackin. Also, its platform was tight. It could move freely between different cloud computing platforms while handling incredible volumes of data. IBM’s scientists and engineers saw an opportunity to couple that with their powerful analytic tools, allowing them to determine how different phenomena affect different business sectors.
Let’s say you run a wind farm. Deep Thunder can give you 3-D wind models. Bought a farm? Deep Thunder can help you prepare for surprise storms. And if you own an airline, a weather oracle can help you fine tune your operations–from tightening delays to reconfiguring your runways. Hell, weather even matters if you work in retail. According to Trenish, Deep Thunder can tell you whether a single thunderhead is going to pass over your store–bring out the raincoats and umbrellas!–or pass you by, with a sensitivity of around one-fifth of a mile.
Deep Thunder also does retrospective forecasting. “If you’re in the insurance business, you want to understand how a past event unfolded, because it might be related to specific claims you have with your customers,” says Trenish. So maybe you’ll think twice the next time you try to blame the baseball-sized hole in your windshield on a freak hail storm.