New supercomputer could be a forecasting game-changer


A challenger is entering the ring of weather models.
Recently, IBM launched the GRAF (Global High-Resolution Atmospheric Forecasting) model, and this newcomer will offer forecasters an incredible new tool.
Twelve trillion pieces of weather data will be issued daily, helping to produce super hi-res weather forecasts as small as 2 miles wide, every single hour, worldwide. Compare that with the detail and timing for current global weather models, which is 6 to 9 miles and every six to 12 hours, respectively.
Basically, this new model is offering the same resolution and rapidness that we currently only get from localized models - the HRRR and NAM to name a few - for the entire planet.
This is big for everyone, but especially less advanced civilizations. Billions of people around the world lack basic weather forecasts, while bigger countries receive alerts on their phone or mobile device for storm warnings, and can access highly accurate forecasts with a few clicks. 
GRAF will also have implications beyond just your personal forecasting needs. For example, airlines can use the better international weather products to route planes around storms and/or turbulence, making for smoother flights. This is especially helpful when flying over desolate areas with a lack of weather data.
Lack of data is actually one of the models few weaknesses, but GRAF is able to pull in data from airplane sensors to help offset this limitation. This is furthered on the ground, where it can receive temperature and pressure readings from smartphones.
The GRAF model will be running off of the new IBM supercomputer DYEUS, named after an ancient daytime sky God. It was built solely to run the model. IBM is no stranger to the forecasting game - in 2016, it purchased the Weather Company conglomerate (what began as the Weather Channel) and became a leader in the meteorology business.