Numerical modeling of an estuary: A comprehensive skill assessment
Article first published online: 4 MAY 2005
Copyright 2005 by the American Geophysical Union.
Journal of Geophysical Research: Oceans (1978–2012)
Volume 110, Issue C5, May 2005
How to Cite
2005), Numerical modeling of an estuary: A comprehensive skill assessment, J. Geophys. Res., 110, C05001, doi:10.1029/2004JC002691., , and (
- Issue published online: 4 MAY 2005
- Article first published online: 4 MAY 2005
- Manuscript Accepted: 21 FEB 2005
- Manuscript Revised: 13 DEC 2004
- Manuscript Received: 31 AUG 2004
- model skill assessment;
- Hudson River
 Numerical simulations of the Hudson River estuary using a terrain-following, three-dimensional model (Regional Ocean Modeling System (ROMS)) are compared with an extensive set of time series and spatially resolved measurements over a 43 day period with large variations in tidal forcing and river discharge. The model is particularly effective at reproducing the observed temporal variations in both the salinity and current structure, including tidal, spring neap, and river discharge–induced variability. Large observed variations in stratification between neap and spring tides are captured qualitatively and quantitatively by the model. The observed structure and variations of the longitudinal salinity gradient are also well reproduced. The most notable discrepancy between the model and the data is in the vertical salinity structure. While the surface-to-bottom salinity difference is well reproduced, the stratification in the model tends to extend all the way to the water surface, whereas the observations indicate a distinct pycnocline and a surface mixed layer. Because the southern boundary condition is located near the mouth the estuary, the salinity within the domain is particularly sensitive to the specification of salinity at the boundary. A boundary condition for the horizontal salinity gradient, based on the local value of salinity, is developed to incorporate physical processes beyond the open boundary not resolved by the model. Model results are sensitive to the specification of the bottom roughness length and vertical stability functions, insofar as they influence the intensity of vertical mixing. The results only varied slightly between different turbulence closure methods of k-ɛ, k-ω, and k-kl.