appropriate local selection of the parameter. Therefore, regardless of flow depth,
a locally consistent τ0 is chosen.
The details of ADCIRC, our implementation of the GWCE based shallow
water equations, accuracy tests, and basic algorithm analysis are provided in a
series of reports and papers (Luettich and Westerink 1991; Luettich, Hu, and
Westerink 1994; Kolar et al. 1994a, 1994b; Kolar, Gray, and Westerink 1996;
Westerink et al. 1992, 1994). Additional information available on the World
Wide Web at http://www.marine.unc.edu/C_CATS/adcirc/adcirc.htm). As with
most GWCE based solutions, ADCIRC applies three node triangles for surface
elevation, velocity, and depth. Time-stepping for all linear terms is based on a
three-level implicit scheme for the GWCE and a two-level Crank-Nicholson
scheme for the momentum equations. Nonlinear terms are treated explicitly,
which imposes a Courant-based stability constraint. The decoupling of the time
and space discrete form of the GWCE and momentum equations, time inde-
pendent and/or tridiagonal system matrices, elimination of spatial integration
procedures during time-stepping, and full vectorization of all major loops result
in a highly efficient code.
ADCIRC has also been implemented in parallel using domain decomposi-
tion, a conjugate gradient solver and MPI (Message Passing Interface) based
message passing. When a low ratio of interface to processor partition nodes is
maintained, linear or even superlinear speedups are achieved. Thus, the wall
clock time is reduced by a factor equal to or greater than the number of pro-
cessors that the code is being run on. Superlinear speedups are possible since the
problem sizes are reduced such that the portion of the simulation being run on
each processor can take advantage of the on chip caching available on Random
Instruction Set Computer (RISC) RISC-based chips used in parallel computers.
Benchmarks have been run on a variety of platforms with up to 128 processors.
ADCIRC also includes a wide range of additional hydrodynamic features
including wetting/drying of elements based on water-surface elevations and
gradients (Luettich and Westerink 1995a, 1995b, 1999).
Model Input Parameters
The Eastcoast 2001 tidal database was derived from a 90-day simulation run
with the O1, K1 and Q1 diurnal and the M2, N2, S2, and K2 semidiurnal astro-
nomical tidal constituents forced on the open ocean boundary and within the
interior domain. A smooth hyperbolic tangent time ramp function, which acts
over 20 days, is applied to both the boundary forcing functions and the tidal
potential forcing functions. Computed time-histories were calculated and then
harmonically analyzed at all of the nodes in the domain, as well as at 101 tidal
elevation stations where measured tidal constituent data are available. The
harmonic analysis was based on the last 45 days of record using time-history
values recorded every 5 min. Since harmonic constituents are allowed to fully
interact through various nonlinear terms in the shallow-water equations, non-
linear overtides and compound tides are generated as well. Therefore, the
6
Chapter 2 Governing Equations and 2-D Modeling
Previous Page
