The Contribution of Mesoscale Features to the Divergence of Numerical Ocean Model Ensembles

Seasonal to decadal predictions of the climate and ocean state can be significantly improved with optimized initialization and data assimilation. Coupled climate models are now reaching resolutions capable of permitting, or in some cases resolving, a large fraction of the global ocean mesoscale field. However, little is known about the importance of the ocean mesoscale and its contribution to ensemble spread, and hence its potential contribution to forecast uncertainty. Mesoscale features are ubiquitous in the world ocean and they play a crucial role in the transport and mixing of heat and fresh water, and their momentum transfer steers larger scale current systems such as the Gulf Stream. We present an analysis of the divergence of surface fields in a set of ocean only simulations at eddy-permitting resolution (1/4°) performed with the NEMO ocean model using non-optimized initial condition perturbations. Mesoscale features contribute around 80% of the root mean square sea surface height anomaly and the ensemble divergence is fast, saturating within about 4-6 months. These results indicate that the ocean mesoscale could provide a significant source of forecast uncertainty even on monthly to seasonal timescales.