1
Introduction and
Objectives
Coastal ocean tidal models are used to define navigable depths and currents
in nearshore regions, to assess pollutant and/or sediment movement on the
continental shelf, and to evaluate coastal inundation. The hydrodynamics of
coastal tides are difficult to predict due to various complexities including irregu-
lar coastlines, intricacies of the ocean floor, and the interaction of astronomical
tides and numerous nonlinearly generated overtides and compound tides. Since
the tidal problem cannot be directly solved analytically, numerical models have
been developed to evaluate sea surface elevations and currents.
To obtain accurate tidal predictions, computer models depend on various
interrelated factors including: (a) the governing equations accurately repre-
senting the actual flow processes and phenomena, (b) the representation of the
water body and boundary forcing functions being sufficiently accurate for the
given problem, (c) the scope of the computational domain being appropriately
sized, (d) the numerical algorithms being accurate and robust, and (e) the
temporal and spatial scales being adequately and if possible optimally
discretized.
A successful strategy to enhance the accuracy of coastal ocean circulation
models has been the use of increasingly larger computational domains such as the
Western North Atlantic Tidal (WNAT) domain (Westerink, Luettich, and
Scheffner 1993; Westerink, Luettich, and Muccino 1994; Westerink, Luettich,
and Pourtaheri 2000; Blain, Westerink, and Luettich 1994, 1998). The WNAT
domain encompasses the Western North Atlantic Ocean, Gulf of Mexico, and the
Caribbean Sea. The domain has an eastern open ocean boundary along the 60W
meridian, which is placed such that an accurate set of boundary conditions can be
specified. The 60W meridian is geometrically simple and does not lie within a
resonant basin such as the Gulf of Mexico. Furthermore, the boundary is mostly
in the deep Atlantic where tides vary more gradually than on the shelf and non-
linear tidal species are minimal. However, large domains add complications to
the process of computational node placement, since they require strategic
placement of nodes in order to maintain acceptable levels of local and global
accuracy for a given computational cost.
Grids for large domains should be unstructured and nonuniform. A uni-
formly discretized grid would require a high level of resolution throughout the
domain due to resolution constraints imposed by regions with shallow depths
1
Chapter 1 Introduction and Objectives
Previous Page
