App could help apply big data principles to weather forecasting

Posted by Justin Hesser on October 15, 2013

You might soon be able to help meteorologists as easily as you check your email. That’s the goal of the developers at OpenSignal, who have launched a new app called WeatherSignal that uses existing smartphone sensors to build a live map of atmospheric readings. The developers hope to use the application to eventually build custom database software that will be able to apply big data principles to weather forecasting.

There are some inherent drawbacks to using smartphones as weather sensors, however. People often keep their phones in their pockets, which obscures readings, and even the most advanced smartphones aren’t deliberately designed for meteorology. However, the first issue can be solved by using multiple sensor readings: The amount of light recorded can determine whether a phone is actually outside. The second issue can be alleviated through the power of big data.

One of the values of big data is that it allows imperfect data to be collected in large enough quantities to glean valuable insight. Smartphones are everywhere, and using them as weather sensors would mean millions of granular data points that could be used to derive correlative insight. Not only could such information be used to predict small shifts in daily weather, it could also help determine the probability of “outlier” weather events, such as hurricanes – knowledge that could save lives.

The next step is determining what role this massive amount of information can play when it comes to forecasting weather. The Birmingham University Climate Lab is working with the developers of WeatherSignal to determine the exact utility of this crowdsourced information. If successful, this would be a huge breakthrough for meteorology, and another example of the utility of custom database software.