The Met Office, a prominent weather agency, has recently made a significant transition to a new off-site supercomputer. This latest development has the potential to enhance the accuracy of 14-day weather forecasts, bringing them on par with the current seven-day forecast standards. Utilizing Microsoft’s cloud-based Azure system, the new supercomputer is also expected to improve predictions of extreme weather events and contribute to advancements in climate research. This marks a notable shift for the Met Office, which has traditionally operated its supercomputers on-site since its inception 60 years ago.
For over a month now, the Azure-powered supercomputer has been running in tandem with the Met Office’s previous system, which has since been deactivated. This transition was carefully orchestrated to ensure a seamless transfer of operations, as Chief Executive Officer Penny Endersby likened the process to changing engines mid-flight over the Atlantic. Reflecting on this milestone, Endersby emphasized the importance of maintaining continuity and reliability in forecasting services, noting that the successful integration of the new supercomputer is imperceptible to end-users. As a result, the Met Office aims to deliver more accurate long-term forecasts while streamlining operational processes.
In a statement, Charles Ewen, the Chief Information Officer of the Met Office, shed light on the computational complexities involved in weather prediction. He explained that the agency leverages numerical weather prediction to simulate the future state of the atmosphere, a task that demands significant computational resources. With Azure’s support, the Met Office can embark on new projects without the need to build additional infrastructure, paving the way for innovative research initiatives. Moreover, Segolene Berthou, who leads a research team focused on environmental prediction capabilities, highlighted the supercomputer’s capacity to run multiple model iterations with varying parameters. This advancement is poised to bolster confidence in climate projections and enable more extensive temporal simulations, underscoring the transformative potential of the new computing infrastructure.
