Development and application of ensemble optimal interpolation data assimilation on the Southeast Asia region and Southwestern South China Sea

Abstract

A NEMO model of the shallow south-west part of SCS and Malacca Strait (centred on the Singapore region) was tested with a compact Ensemble Optimal Interpolation (EnOI) advanced data assimilation (ADA) code. A model run is synchronously assimilated with a complete, accurate and high frequency set of observed data. Consistent reduction in Argo profile Root Mean Square Error (RMSE) with EnOI ADA applied for both temperature and salinity, with RMSE values lower as compared to a model without data assimilation, but slightly higher than a separately refined (2.5x) high-resolution model configurations. RMSEs of temperature and salinity profiles in global model compared against observations are found to be closer to the high-resolution model. Significant and consistent reduction in mean Sea Surface Temperature (SST) bias is expected when ADA is applied, with a max bias of < ± 0.5 °C for the Southeast Asia region and Southwestern South China Seas. Furthermore, improvements are less than NEMO’s built-in sea surface restoring (SSR) function seen in the high-resolution model (max bias < ±0.25 °C). Some limitations on the current implementations of EnOI synchronous daily assimilation exist where accuracy improves with the quantity of observation data. As such, greatest improvements can be seen for the SST in this study, and less so for the salinity profiles.