A Methodology for Anomaly Coupling in Climate Simulation

We demonstrate a new methodology for anomaly coupling in general circulation models on the basis of simulations with the Norwegian Earth System Model. A correction is applied to the sea surface temperature (SST) and wind‐stress climatology of the coupled model during the integration in such a way as to match the observed climatology. The correction is found iteratively by updating the model climatology at every time step, allowing the simulation to converge toward a self‐consistent surface climatology close to that observed, with minimal impact on the simulation of variability. Our approach differs from the traditional way of anomaly coupling that employs fixed model climatologies from stand‐alone model runs to correct the model's climatological mean state. While the atmosphere and ocean components are forced by inconsistent surface fluxes in our methodology, the inconsistency becomes smaller as the simulated climatology converges toward the observations after about 30 years. Our methodology achieves a significant reduction in the systematic SST biases of the coupled model climatology, especially over the tropical and subtropical oceans. Associated with the reduction of tropical SST biases, the double‐ Inter‐tropical Convergence Zone (ITCZ) problem is mitigated, the seasonal cycle is well represented, and some aspects of coupled variability such as the El Nino Southern Oscillation are improved. The methodology can be extended in a way that ensures global energy conservation without affecting the improvements in the simulated climatology and may be thus applied for prediction purposes.