Toward the prediction of surface temperature in the Yellow Sea in winter

An attempt is made to predict the sea-surface temperature in the Yellow Sea in winter. A simple numerical model is adopted for the task so that the construction of the adjoint model is tractable. Direct observations of currents and sea-level fluctuations are used to establish the viability of the forward model. Published regional sea-surface temperature distributions then provide a basis for optimizing both the initial condition and the sea-surface heat flux in the integration of the heat equation. The predicted temperature using the optimized initial condition and sea-surface heat flux is then compared to the temperature time series obtained at the current-meter mooring sites to evaluate the success of the procedure.; Three elements prove necessary: (a) a good simulation of heat advection; (b) a good sea-surface heat flux estimate; and (c) an accurate initial condition.; The first part of the dissertation is devoted to the development of the numerical model and a hindcast of the wintertime Yellow Sea circulation in the time window of 13 January to 21 February 1986, when direct observations of currents and temperature are available at six moorings. Northward migration of warm water in the Yellow Sea trough is reproduced, even though the Eularian currents fluctuate with a near-zero mean. It is found that, in the Yellow Sea, sea-level fluctuations are primarily forced remotely by the Kuroshio whereas current fluctuations are primarily forced by the wind.; In the second part, a variational data assimilation procedure of determining both the initial condition and the sea-surface heat flux from heavily smoothed sea-surface temperature data is carried out. The inversely determined sea-surface heat flux is rich in spatial structure. The spatial structure is verified by uncertainty analyses to be statistically significant. The tongue-like pattern of upward sea-surface heat flux at the center of the Yellow Sea is consistent with the existence of a high sea-surface temperature anomaly there. After the data assimilation, the model predicts correctly the temperature time series at the mooring sites.