A significant upgrade to the global weather system used by the European Centre for Medium-range Weather Forecasts to produce forecasts is about to go live (from 12 May 2026) along with its AI counterpart.
A more consistent and integrated approach to forecasting the atmosphere’s interactions with the ocean through fully coupled data assimilation, and improvements to known issues with accurately forecasting sudden, heavy, localised rain, are among the changes to the Integrated Forecast System (IFS) medium-range and sub-seasonal ensemble forecasts.
The upgrade will also bring more advanced representation of waves and sea ice.
The update to the Artificial Intelligence/Integrated Forecasting System (AIFS) introduces, among other features, ECMWF’s first data-driven wave and snow cover forecasts.
Florian Pappenberger, Director-General of ECMWF, said:
“This is an important upgrade for ECMWF and for everyone who relies on our forecasts. IFS Cycle 50r1 strengthens our physics-based forecasting system, including through fully coupled atmosphere–ocean–sea-ice assimilation and new ocean and sea-ice capabilities. With AIFS v2, we are expanding its performance, bringing a new generation of AI forecasting into operation, building on ECMWF’s technical competence, operational expertise and data infrastructure. Together, these advances reflect ECMWF’s commitment to innovation and close co-development with our Member and Co-operating States in delivering better forecasts, increasing the value to society.”
Matthieu Chevallier, ECMWF’s Head of Forecasts Evaluation, added:
“The updates to these systems, being formally announced today, will together provide numerous benefits to users of our forecasts. Alongside the scientific and technical advances in coupled modelling and data assimilation, and innovation with AI techniques, we have responded to users’ feedback, for example, with new products. We look forward to hearing their reaction to the quality and accuracy of our forecasts, and on how it impacts their applications”.
One of the significant changes to IFS, introduced with Cycle 50r1, is a better representation of convective precipitation, which causes heavy rainfall and violent thunderstorms but is hard to predict because it occurs at smaller scales. The revised convection and cloud-microphysics scheme will reduce excessive stationary rainfall and, instead, represent more realistically how it moves from the ocean across the land.
IFS users will also see the introduction of the new NEMO4-SI³ ocean-sea ice model, bringing substantial improvements in forecasts and analysis over the marine areas, as well as over 40 new ocean and sea ice-related variables. Key improvements include a more accurate representation of how sea ice affects the power of waves and the interactions between waves and ocean currents, often at play in rougher seas. Other improvements for IFS include improved tropical upper-air temperature and wind forecasts, and improvements to temperature and humidity forecasts around the tropopause.
For the Copernicus Atmosphere Monitoring Service (CAMS), one of two Copernicus services operated by ECMWF on behalf of the European Commission, the IFS update will enhance the atmospheric composition and greenhouse gas forecasts. The improvements range from the assimilated observations to aerosols and reactive gases, as well as enhanced input data on human-driven emissions.
For AIFS, both the Single and Ensemble versions of the system are being upgraded to version two, with each introducing a new wave component.
11 wave-related variables are being made available to users, enabling predictions, for example, of when we will experience rougher seas. In addition, the AIFS Single v2 wave component shows substantial improvement in the medium-range skill of wave forecasts compared to the physical model, IFS Cycle 50r1.
ECMWF’s first data-driven snow cover forecasts are also being introduced with this upgrade. Again, the AIFS Single v2 forecasts show improved performance compared to IFS Cycle 50r1 forecasts, with the predicted snow cover fraction closer to observations.
Head of User Services Section, Victoria Bennett, said:
“These new cycles bring improvements to the models and introduce new products that address requests from users. In addition, with IFS, we have now removed the duplication between the old “HRES” single forecast and the Ensemble control and introduced other minor technical updates to reduce complexity. We have worked closely with users during the testing period, and with their positive feedback, we are excited to be making the switch”.
All the upgrades to our systems will be live with immediate effect from 12 May.