Better weather forecasts with new supercomputer
The weather forecasts from the Swedish Meteorological and Hydrological Institute (SMHI) will have a higher local precision thanks to the high performance computing system now installed at the National Supercomputer Center (NSC) at Linköping University.
A significant increase of computing capability makes it possible to accurately predict localized variation in the two-day forecasts.
The scientific advances of recent years has made it possible to meet the demand for higher resolution overviews needed by client players such as the energy sector for their production planning. The extremely comprehensive calculations that this requires are possible only through a considerable expansion of computer power.
The new NSC computer that enables highly detailed forecasts is dubbed Bore, the Scandinavian equivalent of Boreas, the god of the north wind.
"Local weather development can vary a lot even in a relatively small geographic area. By improving the resolution of the forecast images, we can map out a better, faster and more accurate picture of short-term temperatures and precipitations within a designated area," explains Lars Häggmark, a meteorologist at SMHI.
During each 24-hour period, four meteorological computations will project the land weather and two computations will project weather at sea.
The computing power of Bore, which is essentially a cluster of computers, is six times greater than its predecessor. Using a brand-new multi-core-technology, each component processor can run simultaneous tasks to achieve extremely high performance.
Bore was designed through a collaboration between NSC and SMHI. It is another example of the partnership ventures carried out by the two organizations, who collaborate on the design and operation of computer systems for weather research and development. Other areas of joint interest are the development of forecasting and climate models, and of large-scale storage systems for weather and climate data.
"Our close collaboration enabled us to customize Bore for its special applications. And we used standard core-computational components to achieve a highly cost-effective computational resource," says Torgny Faxén, project manager at NSC.
Last updated: 2009-06-03