To accurately forecast the weather, you first need lots of data — not just to train your forecasting models but also to generate more precise and granular forecasts. Typically, this has been the domain of government agencies, thanks to their access to this data and the compute power to run the extremely complex models. Anybody can now buy compute power in the cloud, though, and as the Boston and Tel Aviv-based startup ClimaCell is setting out to prove, there are now also plenty of other ways to get climate data thanks to a variety of relatively non-traditional sensors that can help generate more precise local weather predictions.

Now you may say that others, like Dark Sky, for example, are already doing that with their hyperlocal

Read More At Article Source | Article Attribution