How much of the ocean interannual variability (with periods over 18 months) is driven by the atmosphere and how much is generated by the ocean itself? To help answer this question, the Drakkar project ran its global eddying ocean model for several decades twice. The first run was driven by the full range of atmospheric forcing, including direct interannual forcing; the second one was driven only by a repeated seasonal atmospheric forcing. The first simulation produces the "total" interannual variability, which manifestation in sea level anomalies proves to be very close to altimetric measurements; the second simulation also produces an interannual ocean variability, but which comes from nonlinear, purely oceanic, processes.

Depending on the latitude, up to 40% or even 80% of the variability is thus intrisincally generated by the ocean. This proportion is also substantial on sea-surface temperature. The regions were mesoscale eddies are strong (Southern ocean, Western boundary currents) are mostly concerned with intrinsic interannual variability. In contrast, in the tropical Pacific and Indian Ocean, the ocean interannual variability is mostly due to the atmospheric forcing.
This study also shows that when oceanic mesoscale eddies are not simulated by the model (as in present IPCC coupled climate models), the intrinsic variability is almost absent. These results suggest that the eddying ocean models used in future climate prediction systems may generate a low-frequency variability component which is presently ignored.

Altimetry, especially with now about 20 years of data available, enable to validate such studies by providing comparison with measurements.