Improvements to feature resolution in the OSTIA sea surface temperature analysis using the NEMOVAR assimilation scheme

The data assimilation scheme used in the Met Office's Operational Sea Surface Temperature and Ice Analysis (OSTIA) system has been updated from an Optimal Interpolation (OI)-type scheme to a variational assimilation scheme. The updated system includes a dual length-scale background error correlation operator, and a flow-dependent component to adjust the length-scale combination in favour of the short scale in regions of high sea surface temperature (SST) variability. The variational assimilation scheme improves both the analysis performance and the representation of SST features in the OSTIA analysis compared to the OI scheme of the original system. The results of spectral analysis, assessment of horizontal SST gradients and the response of an atmospheric model to the OSTIA SST analysis as a boundary condition indicate that the flow-dependent formulation successfully contributes to improvements in the feature resolution capability of the analysis. Overall, using a short length-scale of 15 km and including a flow-dependent adjustment component produces the best results compared to using either 40 km or the first Rossby radius of deformation as the short length-scale. The new system successfully captures realistic ocean variability without introducing noise into the analysis, allowing the feature resolution capability of the new system to out-perform that of other comparable SST analysis products.