Data-driven approaches for accelerating ocean spin-up in coupled climate simulations

The spin-up of the ocean component is a critical step in coupled global climate simulations, allowing the model to achieve a physically consistent equilibrium by stabilising key variables such as temperature, salinity, and ocean currents. Without an adequate spin-up, residual drifts can undermine the accuracy and reliability of long-term climate projections. This study explores data-driven strategies to accelerate the spin-up, reducing computational costs while preserving the fidelity of simulated climate states. Using a low-resolution configuration of the EC-Earth4 Earth System Model (ESM), we tested few deterministic approaches to optimise the spin-up phase. A key method relies on iterative adjustments of the oceanic state by projecting multi-decadal trends in temperature and salinity. Empirical Orthogonal Function (EOF) analysis was employed to filter internal variability and generate new initial conditions that minimise numerical instabilities. Additionally, vertical stability was ensured to reduce energy imbalances and maintain physical consistency. Overall, our approach can significantly enhance the efficiency of spin-up processes in coupled climate models by at least a factor of two. These findings pave the way for the development of more sustainable and sophisticated strategies (e.g. exploiting machine learning and AI techniques) in climate modelling. Such advancements will be particularly helpful for high-resolution simulations, where achieving computational efficiency is critical.