Added decadal prediction skill with the coupled regional climate model COSMO-CLM/NEMO

Regional climate models (RCMs) resolve regional climate features better than coarser-grid global climate prediction systems, which are often used to force the RCMs. This study attempted to enhance decadal climate predictions in a European modelling domain with the RCM COSMO-CLM by adding the interactive high-resolution ocean-ice component NEMO-NORDIC over the North and Baltic Seas. With this system, five decadal forecasts covering the period from 1960s to 2000s driven by global MPI-ESM predictions were prepared. The regional ocean was initialised with an ocean-alone assimilation run. The coupled RCM's results were compared with MPI-ESM and with atmosphere-only COSMO-CLM predictions. Coupling improved predictive skill for 2‑m air temperature, especially near the southern coastlines of the North and Baltic Seas. However, forecast quality and the value added by coupling decreased with lead years (as expected). The skill gain was most evident during weak phases of the North Atlantic Oscillation, with relatively weak influence of westerlies and stronger impact of regional processes.