Technology Better anticipate dangerous weather phenomena: Météo France acquires new supercomputers

Better anticipate dangerous weather phenomena: Météo France acquires new supercomputers

Alain Beuraud, project manager of the new supercomputer, in front of one of the machines at Météo France in Toulouse (Haute-Garonne), June 2, 2021. (PASCALE DANYEL / FRANCE-BLEU OCCITANIE / RADIO FRANCE)

We have just experienced a particularly intense moment: 44,000 lightning strikes for Saturday, June 19 alone, whereas a very stormy day is considered to have an average of 35,000 strikes. It was unheard of since 2013. These storms are forecast by Météo France but the challenge is to be as precise as possible. This is obviously important to ensure the protection of property and people, which is one of Météo France’s main missions. This is why the service issues its famous yellow, orange or red vigilance. To establish them, we have long since abandoned the observation of frogs by replacing them with daily data, collected by satellites, ground stations, at sea and which are digested by supercomputers, to establish models of forecasts.

Météo France has just acquired new machines, “Belenos” and “Toutatis”. Two huge cabinets crammed with computers, servers and processors assembled by the Atos group in Angers and which have therefore just been installed in Toulouse. Working capacity: more than 21 petaflops, or 21 million billion operations per second. This is a computing power five times greater than the previous generation which however only dated from 2014.

With these supercomputers, Météo France is placed in the leading group of global meteorological companies with one objective: to improve forecasts, in terms of time and location. According to Météo France, the gain will be particularly significant in terms of dangerous phenomena which can now be better targeted, with a resolution around the kilometer or even less in certain specific situations, and above all, with one to two hours of additional anticipation compared to until today. This is essential when it comes to alerting populations, farmers or air transport.

These machines will also be used for longer-term calculations, in particular to model more precisely climate changes and their consequences on a local scale. Belenos and Toutatis cost 55 million euros. It may seem expensive, but according to a study cited by Météo France and taken up by the government, one euro invested in these supercomputers is 12 euros returned to society, in production gains or in avoided disasters.

Leave a Reply

Your email address will not be published. Required fields are marked *