Sequential state estimation using remote-sensing measurements in the Japan/East Sea

Hirose, Naoki*; Kawamura, Hideyuki  ; Yamamoto, Masaru*

This presentation reports our assimilation modeling efforts in the Japan/East Sea. An eddy-resolving, free-surface ocean circulation model is driven by daily meteorological forcing of JMA regional forecast at the surface and monthly measurement data at the horizontal boundary. The top layer is restored to the remotely-sensed temperature. The SST assimilation reduces the average residual variance from 4.55 to 2.09 degC and presents small-scale features even finer than the original data set. An approximate Kalman filter corrects the subsurface density and velocity conditions based on the difference between the observation and the model at the sea surface height. The sequential updates of the error covariance require more computational overheads than the steady covariance but significantly corrects high-frequency barotropic changes of the northern waters. The assimilated results show realistic upper circulations such as the separation of the East Korean Warm Current or the quasi-biennial oscillation in comparison to in-situ subsurface measurements. The study also demonstrates that the assimilated estimates of the JES improve winter atmospheric conditions from meso to seasonal scales. The sharp frontal structure of the assimilated SST strongly modifies the mesoscale features over the southern JES as simulated by a non-hydrostatic atmospheric model. It is also found that the total amount of the DJF precipitation is lead by the regional NDJ SST and is further controlled by autumn transport of the Tsushima Warm Current.