ABSTRACT

A global eddy-resolving forecast system LFS is developed based on a primitive ocean general circulation model. The system's configuration, forecast experiments, and a preliminary evaluation of the preoperational system are shown. In the preoperational stage, since the full data assimilation has not yet been set up for LFS, the initial state is obtained by nudging the ocean temperature and salinity from an ocean analysis dataset. Despite this, the LFS demonstrates a generally good performance in short-term oceanography forecasting, except for sea level anomaly (SLA). The median values of the 1-day forecast leading root mean square error (RMSE) for the sea surface temperature (SST), SLA, upper 2000 m temperature, and salinity are approximately 0.52°C, 0.10 m, 0.57°C and 0.13 psu, respectively. Although there are slight warm biases in the forecasted SST, the forecasts of temperature and salinity in the thermocline by the LFS are comparable with the results of operational oceanography systems under the framework of the Intercomparison and Validation Task Team. However, the forecast SLA has a relatively large RMSE related to the absence of direct observational constraints in the initial state. Further investigations are needed to improve the performance of LFS.