IBM and its subsidiary the Weather Company are offering GRAF (Global High-Resolution Atmospheric Forecasting System) as the first global weather model that runs operationally on a GPU-based high-performance server.
If the people of Mozambique, Malawi and Zimbabwe had prior information about Cyclone Idai, or residents of japan knew about Typhoon Hagibis beforehand, they would have surely moved to safer locations and would have tried to limit the extent of destruction. Thanks to IBM and its subsidiary, The Weather Company, getting such information timely will now become possible. The two are now offering IBM GRAF (Global High-Resolution Atmospheric Forecasting System) as the first global weather model that runs operationally on a GPU-based high-performance server.
High resolution forecast
The new weather model will offer high resolution forecasts globally with detail as small as 2 miles wide and up to 12 hours in advance. It will have many applications globally for governments and industries, including airlines, agriculture and retail, and will issue 12 trillion pieces of weather data every day and process forecasts every hour.
One of the limitations of global weather forecasts is lack of data. Currently, there are many computer models available which cover 10-15 square kilometers (6.2-9.3 miles) and are updated every 6-12 hours. The GRAF will take it down to 3 km. There are models like High-Resolution Rapid Refresh (HRRR) model of the United States that give short-term weather forecasts. However, these are limited to the boundaries of the country and work only in the time of a weather event. But, with GRAF, information will be available to every country.
In an IBM press release, Kevin Petty, Head of Science and Forecasting at The Weather Company, says, “We realize that there’s a gap globally in that only a small portion of the world really has access to it. We’re getting down to the cellular level of the thunderstorms now, where we weren’t able to do that before. And with that information, we can provide better support to critical decision-makers. ″
GRAF will forecast every landmass on Earth, can provide local forecast at scales of 3km that can extend up to 2,000 km beyond coastlines. It can also predict atmospheric conditions, including humidity, air pressure, temperatures and wind speed.
Power of Cloud and supercomputer
IBM GRAF taps into data from a broad range of sensors around the world, including satellites, aircraft, ground radar, weather balloons and private weather stations. Data sources in future might include readings from the barometers built into smartphones.
It is a numerical modelling system that simulates the global atmosphere through the use of honeycomb mesh reaching 40 kilometers, or about 25 miles, into the atmosphere. The model generates data along and within millions of sectors or cells, almost all of them hexagonal.
IBM GRAF’s speed, accuracy and resolution depend on massive computing power, a new weather model and the use of a wide variety of data from traditional and new sources. Its welter of incoming data is processed by the Dyeus Supercomputer in Raleigh, N.C. which is developed by the National Centre for Atmospheric Research (NCAR) in concert with the Climate Modelling Group at Los Alamos National Laboratory. It was later refined in collaboration with IBM and The Weather Company.
Dyeus further transmits its result to the weather company’s cloud computing system around the world. Further, smartphones which have embedded barometers can be an important local data source in aggregate, people choose to share.