Towards quantifying the relative information content of sea surface data relative to interior data in constraining ocean state estimates
Event: 2016 Ocean Surface Topography Science Team Meeting
Session: Science II: From large-scale oceanography to coastal and shelf processes
Presentation type: Type Poster
Owing to the relative scarcity of interior observations on synoptic temporal and spatial scales, remotely sensed sea surface data such as sea surface height (SSH) and sea surface temperature (SST) (and to a lesser extent sea surface colour) play a key role in data assimilation schemes used in operational oceanography. The relative importance and usefulness of surface data relative to interior data in constraining ocean state estimates, however, depend on a number of factors that remain poorly understood, such as the nature of the assimilation scheme or the type of surface data considered. The main purpose of this work is to develop a systematic understanding of the links between the vertical structures of the internal ocean variability and of their surface signatures by combining physical theories (based on the study of forced and free normal modes in presence of a background mean flow, topography and wave/turbulence) and empirical/statistical techniques. In this work, we will report on the preliminary results obtained from a long term and realistic simulation of the present-day ocean circulation using a high-resolution (1/12) eddy resolving model of the ocean circulation. These results demonstrate that the degrees of freedom characterising the internal variability of the ocean vary significantly from one geographical region to the other, with the equatorial regions being the more complex and the ACC and western boundary regions the simpler ones. This suggests, therefore, that a much higher relative information content of sea surface data in the simpler regions than in the complex ones. Next, we investigate the possibility to separate the vertical structure of the internal variability into a high-pass and low-pass components. Preliminary results suggest that the vertical structures of the high-pass filtered data are often quite distinct from that of the low-pass filtered data for a cut-off period of around 4-8 weeks. The possibility to exploit such an information to reconstruct the interior circulation from surface observations alone has proven difficult so far, however. Finally, we will also discuss to what extent it is possible to use SST as a proxy for density, by systematically exploiting locally defined temperature/salinity relationships, and decomposing simulated temperature and salinity vertical profiles into active (related to density) and passive (related to spiciness) components.