Development of the Ensemble Kalman Filter for the analysis and prediction of the Kuroshio variations south of Japan (Invited)

Yasumasa Miyazawa1, Toru Miyama1, Sergey M. Varlamov1, Xinyu Guo1,2, and Takuji Waseda1,3 1Research Institute for Global Change, JAMSTEC, Kanazawa-ku, Yokohama, Kanagawa 236-0001, Japan. E-mail: miyazawa@jamstec.go.jp 2Center for Marine Environmental Studies, Ehimeniversity, Matsuyama, Ehime, 790-8577, Japan. 3Graduate School of Frontier Science, University of Tokyo, Kashiwa, 277-8563, Japan We are developing the Ensemble Kalman Filter (EnKF) to allow assimilation of various types of the observational data including not only the traditional temperature and salinity data but also new types of data such as High Frequency radar ocean currents. EnKF is also useful for the detection of the highly variable phenomena associated with the western boundary currents because it represents the flow-dependent dynamic correlations among the model variables. Focusing on the Kuroshio variations south of Japan, we have developed an ocean model with horizontal resolution of 1/36 degree (3km) nested in the Northwestern Pacific OGCM with a coarser horizontal resolution of 1/12 degree. Based on 20 ensemble member models running on the parallel computers, we implemented the Local Ensemble Transformation Kalman Filter (LETKF) algorithm. Identical twin experiments using LETKF demonstrated good representation of the flow dependent correlations and effective assimilation of the ocean current observations. We further discuss results of sensitivity experiments to investigate the potential of LETKF.