The impact of ocean dynamics on predictability of North Atlantic sea surface temperatures

Abstract:

In order to determine the importance of ocean dynamics in setting predictability timescales of North Atlantic sea surface temperature (SST), we compare the predictability timescales for wintertime SST between coupled models with a dynamical ocean (henceforth fully coupled model) and models where the ocean is represented by a motionless slab (henceforth slab ocean model). For the slab ocean model, the slab thickness is set to be the annual mean mixed layer depth from the coupled model (variable in space, fixed in time). We find that in the slab ocean model, spatial variations in the predictability timescale have a strong linear relationship with spatial variations in the slab layer depth (R²>0.9), as would be expected from the null hypothesis that the ocean simply integrates stochastic atmospheric forcing. On the contrary, in the fully coupled model approximately half of the spatial variations in the predictability timescales for SST can be explained by spatial variations in the mixed layer depth. Furthermore, the western subpolar North Atlantic appears as a region where predictability timescales in the fully coupled model exceed those in the slab ocean model. These results indicate that oceanic processes are highly important in setting predictability timescales for SST in the subpolar North Atlantic. These oceanic processes include both one dimensional processes, such as interannual variations in mixed layer depths and reemergence of SST anomalies, and three-dimensional ocean dynamics, including variations in the strength of the AMOC and gyre circulations.