Forecast experiments of the Japan coastal ocean using an eddy-resolving ocean general circulation model

We have developed a high-resolution ocean forecast system. This system consists of three elements: ocean modeling, data management, and data assimilation. Synoptic variabilities in the northwestern Pacific Ocean are simulated using a high-resolution (1/12 degree and 45 levels) model based on the Princeton Ocean Model. Using an optimum interpolation method, weekly mean various data are created from sea surface height anomaly, sea surface temperature, and subsurface temperature/salinity profiles. To consistently assimilate those data into the model, the multivariate optimum interpolation method is adopted to estimate the analysis data of temperature/salinity in vertical column. The analysis data are smoothly introduced into the model using the Incremental Analysis Update. Two months forecast run driven by the monthly mean climatological surface forcing is weekly updated. The system has shown a forecasting skill of a few months in the Kuroshio region during the real-time experiment from 2002 to 2004. In particular, we discuss forecast skill of the system for the Kuroshio Large-Meander in summer 2004. It is found that eddy-Kuroshio interaction significantly affects two-months forecast of the Kuroshio path variation.