Germany – Research and development services and related consultancy services – C3S2_384 Downscaling of CMIP6 Climate Projections using Machine Learning Tools to fill the Gaps of the Worldwide CORDEX RCM Simulations

The main objective of the contract is to create new regional climate data based on ML technology. The new data shall complement existing and planned CORDEX (RCM: Regional Climate Model) simulations and form a coherent dataset, where dynamical (RCM) and ML-based data can be used together. The work shall begin with a gap analysis checking the available and planned RCM data for the CORDEX simulations and what gaps are seen for all the 14 CORDEX domains and for all the available climate scenarios. A complete ML methodological overview is required to check whether there are ready-to-use emulators (which can provide similar set of dynamically consistent climate variables as RCMs) available to provide the ML-based data to fill the identified gaps. Ideally, the best available and tractable RCM emulator techniques need to be used to provide the missing data for the community. The data provided by the procured work needs to be published in the CDS in a similar manner to the other CORDEX climate projection datasets. User guidance and documentation will be written to fully inform the users on these new data, and particularly how they can be used alongside other available RCM datasets. ECMWF intends to award a single Framework Agreement for a period of maximum 32 months, which shall be implemented via a single Service Contract expected to commence in October 2025.