With
a joint effort of UMASSD and WHOI, we have successfully developed an
unstructured grid, finite-volume three-dimensional primitive equation
coastal ocean model (so called FVCOM) (Chen et al., 2003). In common
with POM (Princeton Ocean Model) and ECOM-si (semi-implicit code of
POM), FVCOM uses the modified Mellor and Yamada level 2.5 (MY-2.5) and
Smagorinsky turbulent closures for vertical and horizontal mixing,
respectively (Mellor and Yamada, 1982; Galperin et al., 1988;
Smagorinsky, 1963) and a sigma coordinate (vertical stretching to
follow bottom topography). Distinct difference from POM and other
finite-difference or finite-element models is that FVCOM is solved
numerically by flux calculation in the integral form of the governing
equations over non-overlapping, unstructured triangular grids.
This numerical approach combines the best features of the
finite-element method (grid flexibility) and finite-difference method
(numerical efficiency and code simplicity). Further, the flux
calculation method with an integral form of the equations provides a
better representation of momentum, mass, salt, and heat conservation.
This method avoids the occurrence of wave energy reflected from open
boundaries due to improper radiation boundary condition and also
provides exact boundary conditions of no flux of heat and salt at the
bottom of the slope. The conservative nature of FVCOM plus its grid
flexibility and the simplicity of its code also make FVCOM ideally
suited for interdisciplinary application.
FVCOM
has been developed with careful validation through comparison with
analytical solutions and finite-difference models (POM/ECOM-si and
ROMs). The model validations were conducted for idealized cases of
wind-induced surface gravity waves in a circular lake, tidal resonance
in a channel, freshwater discharges over the continental shelf and in a
sloping bottom circular lake, thermal bottom boundary layer over the
slope with steep bottom topography, equatorial Rossby Soliton, and
hydraulic jump, etc. With better fit of the curvature of the coastlines
using unstructured non-overlapped triangular cells, FVCOM demonstrates
significant improvement in numerical accuracy and capability to capture
the right physics of tide-, wind- and buoyancy-induced waves and flow
as well as the bottom boundary dynamics in the coastal ocean.
FVCOM
has been successfully applied to Lake Superior to replace our old
hydrodynamic model (a modified version of ECOM-si) used in that lake
(for the old model study, please refers to Chen et al., 2001; Zhu et
al., 2001; Chen et al., 2002a, Chen et al. 2004c, 2004d). Comparisons
between FVCOM- and ECOM-si-based Lake Superior hydrodynamic models show
that FVCOM is better in 1) resolving the high resolution thermal and
currents near the coast, 2) enabling to build a linkage with small
lakes in the adjacent coastal region, 3) capturing the right physics of
the bottom boundary layer at the steep bottom topography off the
Keweenaw coast, and 4) ensuring the mass conservation of passive
tracers. Updated FVCOM has migrated to Fortran 95/2K. This was done
primarily to facilate the use of allocatable arrays that beneficial
when one does not know a priori the dimension of the sub-domains
resulting from the an arbitrary partitioning. Four-dimensional nudging
and Kalman Filter data assimilation methods are incorporated into the
updated code to merge the model-predicted water temperature and
currents to observations taken by mooring current meters, AVHRR-derived
SST, and regional hydrographic surveys.