After more than three years of studies, Météo-France has launched a call for tenders to renew its two supercomputers, Belenos and Taranis. The objective is to improve the reliability of forecasts, better anticipate high-stakes phenomena, more precisely simulate the future climate and support actions to adapt to climate change.
Six times more computing power by 2029
The call for tenders will take place in 2025 for an initial deployment in 2027. The contract will cover the period until 2033, with the deployment of additional computing power in 2029. By that date, the objective is for Météo-France’s supercomputers to achieve a computing capacity six times greater than their current performance.
New computing architectures and artificial intelligence
The call for tenders is part of a major development plan to update Météo-France’s scientific and technical tools used for weather forecasting and climate modeling. The company’s future supercomputers must therefore integrate new computing architectures to increase their performance while improving their energy efficiency. They must also allow the coexistence of physics-based forecasting systems (notably Arome and Arpege models) with new approaches based on the use of artificial intelligence, in which Météo-France teams are already heavily involved.
Increasingly precise forecasts
Météo-France develops and operates two complementary numerical weather forecasting models – Arpège (the planetary model) and Arome (the regional model). They work by dividing the atmosphere into small units, called meshes.
New supercomputers will make it possible to reduce the resolution of forecasts, and therefore improve the accuracy of forecasts, with finer meshes, by better representing atmospheric phenomena and their interactions with continental and oceanic surfaces, and by using more numerous observations to better initialize forecasts.
Some examples of operational improvements that Météo-France expects by 2033 include an increase in the number of anticipated orange or red alerts with at least six hours’ notice and an 10% improvement in the quality of precipitation and gust forecasts. Furthermore, in Overseas France, the organization plans to create a more reliable forecast of the intensity (probable, maximum) of tropical cyclones and better forecasts of intense rainfall events and floods that regularly affect these territories. For the aviation sector, it hopes to achieve a significant improvement in the quality of forecasts at immediate forecasting timescales (up to six hours ahead), to optimize air traffic management and ‘routine’ airport management decisions. Finally, it is aiming for more precise estimation of the evolution of the frequency and severity of extreme events.
Météo-France’s current supercomputing capabilities
The organization has highlighted that its supercomputers enable it to carry out its priority missions: improving the quality and reliability of forecasts; anticipating dangerous phenomena (such as storms, tempests and cyclones); and producing relevant information for adaptation to climate change. Their computing power (up to 21 million billion operations per second), 50,000 times greater than a personal computer, makes it possible to precisely process and exploit a large amount of observation data (thanks in part to the new European meteorological observation satellites). Overall, the supercomputers are essential for initializing forecasts and verifying their reliability.
They also calculate the evolution of meteorological variables using the numerical forecast models Arpège (which covers the entire globe) and Arome (deployed model of regions of interest, which covers smaller-scale phenomena). In addition, these systems simulate future climate changes and carry out the research necessary to improve the numerical models used for weather forecasting and climate projections.
These supercomputers enable the development and execution of computer codes used to model weather phenomena on a global scale as well as in French territories, to simulate their evolution at all timescales and thus better anticipate the impacts and risks for populations. They also help to understand climate change and its consequences, and thus support adaptation actions.
In related news, Météo-France recently deployed new versions of its numerical weather prediction (NWP) models, Arpège and Arome. The organization said this upgrade will improve the quality of forecasts by 3-4% and mean the country’s weather model can be ranked second in the world for European forecasts. Click here to read the full story.
