And Suddenly Everybody's a Weatherman
WHY YOU SHOULD CARE
Crowdsourcing weather forecasts could mean a heads-up on typhoons — or even just wet shoes.
By Nick Fouriezos
Last spring in Japan, it was the annual agita over cherry blossoms. Since the little pink jerks can bloom and die in the span of a week, all the weather services, trying to help a fairly hefty tourism business, went on their usual race to predict their arrival. Some were right; some were wrong.
Then there was this outfit called Weathernews. The Tokyo-based service says it accurately predicted the sakura’s arrival two weeks before most. But its edge wasn’t satellites orbiting the heavens to monitor jet streams, or a crack team of botanists checking buds. It was crowdsourcing, which came from nothing more complicated than smartphones.
Maybe it’s no surprise that a gadget we touch more than our spouse might help solve one of mankind’s oldest crises: Do I need my umbrella? Through apps, phones can collect pivotal information for meteorologists, from our most banal social posts — it’s snowing! — to key metrics like air pressure, humidity or temperature. And increasingly, a handful of groups are figuring out ways to use that data to predict phenomena large and small: the “peak” leaf-peeping season in Vermont, tsunamis in the Indian Ocean.
Sure, the idea of user-submitted weather data is hardly new. Weather Underground grew into what was likely the world’s first crowdsourced forecast company after it emerged in 1995. But until recently, the tech involved was costly, and the data coming in was crude. Enter companies like WeatherSignal and PressureNet to sift through the rubbish, which app owners agree to share. Not that any of this is easy or perfect yet. “Just being able to consolidate all that information into one meaningful story is a challenge,” Jim Menard, Weather Underground’s general manager, tells OZY.
It’s a global battle, with others also betting on the clouds.
Through its app — which has 6 million users — Japan’s Weathernews quickly gathered more than 50,000 reports of blooming cherry blossoms this spring. In the past, traditional forecasters were able to pull only a fifth of that amount. The feat “basically changed the entire face of how weather forecasting was done in Japan,” says Julia LeStage, founder of Boston’s Weathermob, an Instagram-like weather app that received $1.1 million from angel investors before being acquired by Weathernews in May.
After already shelling out to enter the U.S. market, Weathernews is now doubling down on making you and your neighbors the next stay-at-home weather forecasters. It recently gave away thousands of high-quality sensors as part of its “gamification of weather,” Weathernews director Tomohiro “Bashi” Ishibashi tells OZY. And in June, the company announced a partnership with Paris-based Netatmo to collect home sensor data from 174 countries. “We should survive this app battle,” Ishibashi says, although he admits, “It’s a world war right now.”
Indeed, plenty of others are also betting on the clouds. In May, China’s national weather service reached an information-sharing deal with global forecaster AccuWeather to beef up its local reports. Melbourne, Australia-based Ourcast, which developed weather-sharing app Minutely, has received $2 million in Australian venture capital funds. And the granddaddy of it all, Weather Underground, which now boasts 127,000 amateur meteorologists, was bought in 2012 by the industry’s biggest name: the Weather Channel.
Still, the burgeoning strategy isn’t all clear skies. Temperature and humidity readings are fickle (after all, phones spend most of their time in people’s pockets), while air pressure is affected by elevation — which complicates data analysis if, say, you work in a skyscraper. These kinds of inconsistencies have made integrating crowdsourced data with the Weather Channel’s algorithms difficult, says Allan Hui, vice president of product for the Weather Underground division. “Everyone is trying to capture data,” the former Amazon innovator acknowledges, “but no one knows how useful it’s going to be.”
What’s more, those who arguably most need accurate weather forecasts are often the least likely to get them. There are only about 11,000 traditional weather stations in the world, though poorer countries often have the biggest coverage gaps and are also the least likely to have smartphones, says Weathermob’s LeStage. And even in developed countries, the “crowd” congregates in high-density cities already covered by government-run weather reporting agencies. Rural communities are left out to dry, and crowdsourcing won’t help unless enough locals pitch in and report. “The main thing is you need to get a good sample size,” says Gerry Wiener, a senior software engineer at the National Center for Atmospheric Research in Boulder, Colorado.
These sorts of concerns may eventually be settled by another idea entirely: crowdsourcing your car. With the U.S. Department of Transportation, Wiener is exploring ways to transmit vehicle data into live-time weather updates. His Pikalert system takes information, such as whether a car’s wipers are on or the brakes are engaged, and then interprets and relays it through highway alerts. The program is already being used by state vehicles in Minnesota, Michigan and Nevada, and Wiener’s vision is to have anyone who drives become a contributor. “You could have millions of vehicle observations,” he says.