The goal of this Work Package is to develop and test scientific methodologies to merge predictions from the regional and global (initialised and non-initialised) model simulations into a single prediction system, continuous across spatial and temporal scales up to 40 years in the future. A key goal of this team is determining the timescale over which an initialised model will provide more accurate (‘skilful’) predictions than a non-initialised model. Initialisation involves allowing models to simulate current conditions in order to calibrate them for accuracy, before simulating the future. This improves model accuracy on shorter timescales but takes time and computing resources to accomplish. Combining these approaches merges model outputs across different timescales. Models working on different spatial scales, such as regional and global projections, can also be merged. The resulting merged predictions present users with trustworthy, consistent climate information about the future, integrating the scientific outputs of EUCP into a single, optimised tool. This will provide useful, actionable information on a range of timescales and spatial scales to a large range of organisations, from European regional scale down to smaller, local areas, facilitating climate-related risk assessments and climate change adaptation programmes.


Work Package 5 primarily contributes to EUCP objective 1, working to develop methodologies to bring together initialised decadal climate predictions, non-initialised climate projections based on global climate models, and high-resolution regional climate projections. This provides seamless climate information for users over a period of 1 to 40 years into the future, with a focus on the European region. Its final goal is integration across the latest generation of European climate forecast systems and those that will be built on them, making them a credible, reliable, authoritative and action-oriented source of climate information to support more climate-resilient European economies and societies.

Its objectives are:

  • Comparing predictions based on global initialised versus non-initialised simulations for common prediction timescales. This will allow estimate the prediction time over which initialised models show benefits compared to non-initialised simulations and over any other benchmarks, for different large-scale and local variables.
  • Test methods of combine global initialised forecasts with non-initialised forced-only predictions with a perfect model setting, then estimate the benefits of the combined predictions for different variables and regions.
  • Develop methods to merge information from the high-resolution regional model simulations with global climate predictions using three methodologies: blending probability density functions from different models, creating discrete scenarios, and producing storylines of individual events according to the identified user needs.
  • Evaluate the extent to which observational, physical and emerging constraints on model outputs are reflected in the predictions of variables relevant to different users.

Read more about activities