IBM’s CEO Ginni Rometty unveiled a model that aims to predict the weather hyper locally based on crowdsourced data. In many regions hyperlocal predictions are already in place in the US, Japan and parts of Western Europe where there is a high density of weather stations. IBM states it can offer forecasts that are hourly updated and have a 3km resolution, while current forecasts are updated every 6 hours and have a resolution of 10-15km. Hyperlocal weather forecasts are relevant to various market segments such as farming, airlines, insurance companies and energy companies.
The model, called GRAF (Global High-Res Atmospheric Forecasting System), tracks barometric data from smartphone users that use the Weather Channel app owned by The Weather Company, a subsidiary of IBM. Besides smartphone data, the technology relies on supercomputing and other devices connected to the Internet of Things. Ultimately, sensors on buildings, automobiles and wearables could add data to the model. Implementation of the model doesn’t come without any challenges. IBM faced a lawsuit from the city of L.A. over the misuse of tracking data from its weather apps. Read more here on Time.com.
The Brussel Times reports that the Belgium national postal service Bpost is launching a pilot as well in the forecasting of hyperlocal weather by equipping 30 of its cars with sensors. As the postal cars go around the entire city, hyperlocal data can be collected and analyzed, Marc Christiaens, Business Unit Manager at Royal Meteorological Institute of Belgium states. ‘The area weather predictions are aimed at can be decreased from 16km² to 1.7km², he says. Read more in Dutch here.
