The Dynamical Circulation of a Partially Stratified, Frictional Estuary: Long Island Sound

Abstract

I report the results of observational and theoretical studies of the tidal and residual circulation in Long Island Sound. This estuary is near resonant at the semidiurnal frequency, has substantial variations in cross-sectional area along its length, and has spatially variable and seasonally modulated density stratification. These physical characteristic of the Sound lead to strong tidal and residual circulation that are coupled by vertical mixing and may have a major role in controlling the development of hypoxia in the western Sound in the summer. I provide the first detailed description of the vertical and horizontal structure of the circulation in the western Sound by combining both moored and ship mounted profiling current meter observations obtained over a three year period. I show that the vertical structure of the principal component (M₂) amplitude is strongly sheared and that this structure cannot be reproduced by a classic model with time-invariant eddy viscosity. Seasonal variations in stratification do not influence the vertical structure of the M₂. Strong over-tides at the M₄ and M₆ frequencies were detected with M₆ amplitudes that increase to the west. I demonstrate that this is inconsistent with a weakly non-linear vertically and horizontally averaged model driven by semidiurnal tides in a converging channel with bottom friction, likely due to neglected channel geometry and time-variant bottom friction. The observations also reveal a weaker, but persistant, non-tidal circulation. In combination with hydrography obtained during short high resolution cruises and from data archives, I examine the importance of each term in the momentum balance both along- and across- channel, and put these into context using longer-term (but more spatially sparse) velocity, wind and hydrographic data. I develop a novel application of inverse methods that utilizes data measurements to test model validity and infer unknown data parameters when the appropriate model is chosen.