Development of a Data Sampling Strategy for Semienclosed Seas Using a Shallow-Water Model

The interdependence of numerical coastal ocean models and field observations forms a basis for development of a data sampling strategy. The sampling strategy is centered on the identification of regions for field measurement that have the greatest impact on model-simulated dynamics. Comparisons, both qualitative and quantitative, between model-computed tides from a data-assimilative, barotropic tidal model in the Yellow and East China seas that assimilates 114 tidal height observations, and model simulations that assimilate a single station group, defined by bathymetry or location within a dynamically active region, indicate that spatial location of the data has the greatest impact on computed model response. Tidal height information from deep water is shown to minimally influence predictions in shallower shelf waters. Coastally located data capture finescale, highly nonlinear tidal interactions spawned by complex bathymetry and irregular shorelines in localized areas. Tidal stations at shallow intermediate depths on the continental shelf make the largest contributions to model computed tides, particularly in the representation of dominant amplitude features and phase structure. Observations along the Korean coast, in the Bohai Sea near river inflows, or at the Taiwan Strait enhance model computations of the nonlinear tidal dynamics present in these regions. Procedures implemented to derive a data sampling strategy are readily transferred to other marginal seas and are not limited to tidal prediction but can be applied to other physical phenomena for which measurements are available or desired.