Assimilating remote sensing and in situ observations into a coastal model of northern South China Sea using ensemble Kalman filter

Abstract

Major forecast errors on the background error covariance from initial conditions, atmospheric forcing, model open boundary conditions, and the river discharges are examined in a coastal model of northern South China Sea. The analysis of background error covariance matrix produced by model ensemble shows that the perturbations of the initial conditions and atmospheric forcing play major roles in producing and maintaining the amplitude of ensemble spread except for the sea surface height (SSH) field. The perturbation of model open boundary conditions can influence ensemble spread of all variables and covariance between temperature and velocity or between temperature and SSH. The perturbation of river discharge mainly affects the covariance of salinity in river estuary. A data assimilation experiment of northern South China Sea is conducted using ensemble Kalman filter (EnKF) in the Princeton Ocean Model. In the experiment the ensemble model forecasts are made by perturbing the above mentioned four major model inputs. The assimilated data include sea-surface temperature (SST) and conductive–temperature–depth (CTD) observations. The assimilation experiment suggests that assimilating SST and CTD data can effectively improve the model simulation that has a shallower thermocline and weaker plume comparing to the observations. Moreover, consistent with these improvements of temperature and salinity, the along-shore velocity, cross-shore velocity, and characters of water mass are also corrected, respectively.