Weather forecasting supercomputer gets a $36m upgrade

One of the world's biggest weather prediction systems is set to get a major upgrade.


The finished Cray panel design for a supercomputer at ECMWF.

Image: ECMWF

Cray has signed a $36m deal to upgrade and expand the supercomputers used by the European Centre for Medium-Range Weather Forecasts (ECMWF).

The centre provides medium-range forecasts of global weather to 15 days ahead as well as with monthly and seasonal forecasts up to a year ahead.

Read this

Raspberry Pi meets Lego in supercomputer-like cluster: Photos

Six-year-old James Cox has helped his dad and other engineers at the University of Southampton to build a £2,500 supercomputer-style cluster out of 64 Linux-based Raspberry Pi PCs and Lego.

Read More

The data-crunching supercomputer system based at its headquarters in Reading, England, is one of the largest in the world of meteorology and contains the world's largest archive of numerical weather prediction data. It also runs a sophisticated global prediction model of the atmosphere and oceans.

The upgrade will allow the research center to further develop its complex models to provide more accurate weather forecasts.

Back in June 2013 Cray was awarded a contract to provide ECMWF with two Cray XC30 supercomputers and a Cray Sonexion storage system. Under the terms of the new deal, Cray will expand and upgrade the supercomputers at ECMWF to Cray XC40 systems. ECMWF will also get additional Cray Sonexion 2000 scale-out Lustre storage, and a 32-node Cray XC40-AC system with the next-generation of the Intel Xeon Phi processor code-named Knights Landing.

The new systems are due to be delivered this year.

ECMWF Director of Research Erland Källén said the upgrade will enable the centre to develop high-resolution forecasts that improve the prediction of severe weather events up to about two weeks ahead.

It will also allow for the introduction of improved data assimilation methods, allowing the centre to use more Earth system observations, and to produce more detailed and better-quality atmospheric composition forecasts.

While supercomputing has traditionally been limited to use for scientific research, increasingly it is being viewed with interest by more mainstream enterprise customers too as the search for ways to analyse their big data stores.

More on supercomputing