"If you were able to watch what the weather did in the last hour, you can get a pretty good idea of what was going to happen in the next hour," says Mark Hohmann, one of the app's creators.
Ourcast uses real-time radar data to predict, based on historical patterns, how weather at a given location will change within the next two hours. It further distinguishes between rain, snow and sunny skies by tapping into data streams from 20,000 home weather stations and asking users to check in with their local weather.
The startup's three co-founders have backgrounds in engineering and data science, but not weather, and there are no meteorological models involved. By keeping weather predictions within a two-hour window, they believe statistics are sufficient.
Meteorologists also rely on statistical models to predict the weather, but they combine that data with what Keith Seitter, the executive director of the American Meteorological Society, calls "the physics of the atmosphere."
If Ourcast is doing its statistical modeling correctly, Seitter says that the app will probably get fairly good results most of the time. But when there's a weather pattern that changes quickly rather than just moving -- such as a thunderstorm -- he says there's no replacement for a weatherman.
"When weather info is most important is exactly the situation when this app is least effective," he says.
Hohmann doesn't see the app as necessarily replacing daily weather forecasts, but rather as filling in the short-term details. He suggests runners, commuters and little league coaches can use the app to make decisions about their activities -- or even discuss them with other users.
The app's check-in feature includes an option to comment about the weather. That may sound frivolous, but it's hard to argue it isn't consistent with existing real-world behavior.
"Nobody has ever had the opportunity to try to do this," Hohmann says of his app. "We think this is something you build and people will find a number of uses for it."
