The Weather Company, an IBM Business initiative, has announced that it proposes to use the latest IBM Global High-Resolution Atmospheric Forecasting System (GRAF) to serve the needs of “an under-served population” across the globe.
As per a report in September 2019, GRAF is said to be the first hourly-updating weather system that can predict something as small as a thunderstorm that occurs anywhere on the Earth. When compared to the weather models that exist, it claims a nearly 200 per cent improvement in forecasting resolution for much of the globe (from 12 sq km resolution to 3 sq km)!
The Weather Company has been leveraging models that are in some cases developed by national meteorological agencies, and uses technologies such as Artificial Intelligence (AI) to extract additional information to make high-quality forecasts. GRAF is the latest of its own prediction models.
Over the years, weather modelling has become global, including European Centre for Medium-Range Weather Forecasts, US National Weather Service, the UK Met Office, and the Japanese Met Agency. The models are typically run at a scale of 10 sq km or 15 sq km and are updated four times a day at the most. But, the Weather Company is looking to run global models at 3 sq km over the globe with high population concentrations it order to come up with high-quality forecasts that are updated hourly.
GRAF is to be used to better define and capture thunderstorms. Three key areas are to be in focus: observations, computing power, and modelling. Observations are sourced from millions of sensors worldwide and in-flight data information is used for the purpose. Data simulations are based on them and these are streamed into models to assess current conditions. Then comes the computing power. With graphical processing units that accelerate workloads for faster performance, there is more power at hand to run the models with high resolution globally.