Coastal-ocean hindcast/forecast model

Flows in the coastal oceans are produced by interactions of different components: tides, winds, buoyancy discharge from estuaries, topography and remote forcing of deeper-ocean origin. We present here a general methodology for efficient coastal-ocean hindcast and forecast, taking advantage of the disparate time scales which exist between the different components. Tides are repetitive and can be calculated. River discharge and remote forcing are assumed to have long time scales O(months or seasons), so that they can be considered to be quasi-steady when one computes wind and density-induced flows, which have time scales of the O(1-10 days). Thus, for a given topographic region, one can first perform a long-term simulation using a general circulation model (GCM), and including as much physics and different forcing as possible. One can then decompose the density field using, for example, the method of empirical orthogonal functions (EOFs). The first few EOF modes can be correlated with the winds. Once this wind/density correlation is done, one can then use it to drive a simplified model which excludes the density calculation. The method is applied to simulate the spring and summer circulation in the New York Bight. The hindcast fields compare favorably with the true-state results obtained from the GCM.