IBM will be able to bring the full force of big data to predicting the weather.
IBM’s Watson might soon be your new weatherman.
The company’s deal, said to be worth $2 billion plus, to buy the digital and technology pieces of the Weather Company’s business—the company that runs the Weather Channel—marks IBM’s first consumer-facing purchase in years. And its big bets on the cloud and Watson, its cognitive computing system, seem to be driving its decision. It means IBM will be able to bring the full force of big data to weather, and potentially predict the weather—and the chances of natural disasters—far more accurately than ever before.
Weather will essentially be another real world application of Watson, which CEO Ginni Rometty told Charlie Rose back in April is IBM’s “moonshot.” Watson’s already being used by doctors to more quickly and accurately make diagnoses and by the city of Austin to help tourists find better tacos.
Weather Company CEO David Kenney said it has “billions of data points from sensors” — in fact, three billion weather forecast reference points, as well as weather data from more than 40 million smartphones and 50,000 daily airplane flights. All this will be a perfect data set to run through Watson and IBM’s cloud computing prowess to mine for new insights on weather patterns that impact our daily lives.
“We’ve known about clouds for a long time,” Kenney added.
IBM has actually been working on weather for years. Since the 1990s, IBM has collaborated with the US National Oceanic and Atmospheric Administration (NOAA) to help the government with short-term forecasting models. Meanwhile, in the Physical Analytics Lab at the company’s Yorktown Heights, NY, research facility, a team has been working with the Department of Energy on technology that could one day (fairly soon) accurately predict when clouds will form (and therefore, rain). IBM also already partnered with the Weather Company earlier this year to provide cities with a disaster management tool as the Atlantic hurricane season began.
Now with direct access to the Weather Company’s multitudinous sensors and digital channels, IBM could revolutionize how well we’re able to predict—and plan for—weather events using data already available today.
Tennis offers a handy analogy. Since IBM got involved in tennis—it powers the point tracking and analytics at the US Open and the other Grand Slam tournaments—it’s shaped the way we think about the sport. Commentators no longer simply talk about how many second serves a player has had in a match.
Now, thanks to IBM’s data analysis, we know in an instant the likely velocity of the second serves the player is going to make, and the likelihood that the serve will be in, given the player’s history in the tournament, against that particular player, and on that surface. IBM brought big data to the hardcourt. The Weather Company acquisition has set IBM up to do the same thing for weather.
Watson is designed to literally crunch all of the data it can get its hands on. Even before the deal, IBM already had lots of weather data to play with. The Physical Analytics team is using information from essentially every weather report ever disseminated in history to try to build a system to more accurately predict the weather. The team, lead by Hendrik Hamann, is trying to build this database to figure out when it’s going to get cloudy so that solar energy generating companies can more accurately plan what resources will be available when.
IBM announced in July that its research has already led the team to create models that are 30% more accurate in predicting cloud cover than any other models out there. And their system—which is a machine learning system—will only get better with time, as more data is fed into the system. Rather like Watson.
The team has been working on this project for three years, Hamann told Quartz in May. They have a solar array set up on the roof of Yorktown Heights, which they use as one of the myriad data points they’re collecting weather information from. Their Department of Energy partnership gives them access to real-time weather information from sensors across the country. Now, theoretically, they could have access to the Weather Channel’s sensors and historical data, which will only add to the machine’s ability to predict future cloud coverage, and potentially, down the road, other forms of weather.
IBM didn’t explicitly confirm that there was a correlation between the Weather Company purchases and its weather research. But it did tell Quartz that the company has “has long focused on pursuing the best forecast skill and decision tools” in its partnerships with local and national government bodies, whether that’s helping predict rain or the next Hurricane Katrina.
Predicting the weather more effectively could have a plethora of positive effects on the economy, enabling businesses to plan around inclement weather, from farmers and fishermen to builders and truckers. You’d even have a decent chance of knowing if a ball game would be rained out, or if there would be rain on your wedding day.
“And where this acquisition comes in is how weather science will coalesce around cloud, Internet of Things, and Watson,” IBM added.
Meanwhile, with the Weather Channel’s apps, IBM can get its brand in front of more consumers. The Weather Company says its Weather Channel mobile app is the fourth-most used app in the US. (According to app analysis company App Annie, it’s consistently been near the topof the iOS download chart for the last five years.) Every time you check the weather forecast, it will be IBM telling you what to expect. Even on the cable TV channel, which IBM isn’t buying, the data will be provided by IBM—most likely along with its logo.
It’s entirely conceivable that IBM will one day have a system that will tell you with 95% certainty that it will only be cloudy for 10 minutes today and that you can leave your jacket at home, or that it’s sure the next hurricane won’t come through your area. Now that’s something that will get consumers—and perhaps investors—excited about IBM’s growing artificial intelligence prowess.