The previous edition of the Newsletter provided an insight into the years of research that went into the development and implementation of one of the most substantial model upgrades our Integrated Forecasting System (IFS) has undergone in many years. IFS Cycle 41r2 is set to offer improved range, reliability and accuracy to help national meteorological services provide earlier warnings of adverse conditions and extreme weather to better protect property and vital infrastructure, and to aid long-term planning for weather-dependent industries.
Several weeks into the new model cycle becoming operational, we can report a smooth transition with extremely limited disruption to users, which in itself is an achievement. Most importantly, the evaluation results presented in this Newsletter give us great confidence in the quality of 41r2 analyses and forecasts. How did we get here?
The first ingredient is of course the research that ECMWF conducts into all aspects of numerical weather prediction (NWP), from the modelling of physical processes to the development of new numerical schemes and data assimilation methods. The second ingredient is the exchange of ideas and solutions with our Member State partners and through academic collaborations, for example the recent adaptation of the input/output server. Last but certainly not least, the upgrading and expansion of our supercomputers from Cray XC30s to Cray XC40s will deliver the computing capability needed to make it all work.
The model upgrade illustrates how intimately linked progress in NWP is with supercomputing and the scalability of NWP codes. This is why we were delighted to be invited to host the Cray User Group meeting 2016 in London, which is dedicated to the overarching theme of scalability. This gathering of some 200 supercomputing experts is held once a year to review and discuss developments at Cray, and sometimes to challenge the American vendor. But the conference, which takes place from 8 to 12 May, is also about ensuring that Cray and its users have a platform for sharing ideas and prospects for development.
Our Scalability Programme is taking the same approach, teaming up forecasting experts with university research and high-performance computing centres as well as world-leading hardware companies to bring entirely new knowledge and technology to the field. The close partnership we have enjoyed with computing vendors over the years has played a critical role in ensuring ECMWF’s success and is one of our key priorities, as our Workshop on High-Performance Computing in Meteorology to be held in November will once again illustrate this year.
Maintaining a strong and constant dialogue with the computing industry has proven to be in all our best interests. It ensures that advances in computing match what our science requires – which is a win for them and for us – and it also ensures that we can take research in NWP to the frontiers of knowledge, producing the best possible predictions for the benefit of our Member and Co-operating States.
Florence Rabier
Director-General