New supercomputing resources available for Special Project research

ECMWF supercomputer in Bologna

A view of ECMWF’s new high-performance computing facility in Bologna, Italy.

From next year ECMWF will start to provide about five times more supercomputing resources in its new data centre in Bologna for weather and climate research as part of the Special Project framework.

The new data centre is built around an Atos high-performance computing facility specifically designed to support both time-critical forecasts and typical research workflows.

Special Project applications are open to researchers from any organisation in ECMWF Member States and should be submitted before 30 June 2022 for a start in 2023.

The annual application cycle opens at the end of May, when details about available resources will be published. It is now closed for Special Projects starting in 2022, but late requests, which might be eligible for resources from a reserve, can be submitted any time. 

What are Special Projects?

Special Projects are defined as “experiments or investigations of a scientific or technical nature, undertaken by one or more Member States, likely to be of interest to the general scientific community”.

They give researchers access not only to a top high-performance computing facility and one of the largest meteorological archives in the world, but also to full user support. Projects vary in size and can run for up to three years.

Twenty-five per cent of ECMWF’s high-performance computing resources are allocated to Member States. Of these, up to 10% are reserved for Special Projects. More than 80 Special Projects are currently running at the Centre.

Recently resources have been allocated, for example, to investigate imprecise approaches to accelerate weather forecasts, the advanced assimilation of satellite observations and the impact of improved atmospheric forcing over a limited-area Arctic region, or mining 5th generation reanalysis data for energy budget changes and forecast uncertainty growth.

Special Projects example diagram

These images from the third research project mentioned above show a machine learning mean prediction from a 2-day ensemble forecast of 500 hPa geopotential (left), the ground truth (middle), and the ensemble standard deviation (right). They show that the model correctly predicts that its forecast over the North Atlantic is uncertain, thus correctly identifying its own shortcomings. (Credit: Leopold Haimberger, Alexander Bihlo)

All Special Project applications undergo a review process by the Centre and its Scientific Advisory Committee (SAC) and are ranked primarily by their scientific quality. Allocations will be awarded based on this ranking.

How to apply

Information on how to apply and a list of current and closed projects can be viewed on the ECMWF website.