28.02.2013 Views

A Semi-Implicit, Three-Dimensional Model for Estuarine ... - USGS

A Semi-Implicit, Three-Dimensional Model for Estuarine ... - USGS

A Semi-Implicit, Three-Dimensional Model for Estuarine ... - USGS

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

82 A <strong>Semi</strong>-<strong>Implicit</strong>, <strong>Three</strong>-<strong>Dimensional</strong> <strong>Model</strong> <strong>for</strong> <strong>Estuarine</strong> Circulation<br />

4.3.1.1 Momentum equations<br />

The solution of the momentum equations is separated into two stages in which first the explicit terms and then the implicit<br />

terms are evaluated. The separation is not a <strong>for</strong>m of time-splitting but simply a grouping of terms <strong>for</strong> convenience in making the<br />

computer program modular and facilitating the presentation of equations. The finite-difference equation <strong>for</strong> the x-direction<br />

1<br />

momentum equation is written so it is centered within layer k at the horizontal point ( i + ⁄ 2)Δx<br />

, jΔy ; the explicit stage is repre-<br />

sented by<br />

Ûi + 1⁄ 2,<br />

jk , = Ui 1⁄ 2<br />

n – 1<br />

+ , j, k 2Δt ( ADVx) n ( CORx) n<br />

[ – +<br />

( BCLINICx) n<br />

+ – + ( HDIFFx) n 1<br />

– ] i 1⁄ 2<br />

+ , j, k,<br />

(4.23)<br />

where the terms in brackets refer to the discretized <strong>for</strong>m of the corresponding groupings of terms in equation 3.59. The bracketed<br />

terms are expanded fully using the leapfrog scheme and displayed in Appendix D (eqs. D.1–D.4). The symbol ˆ<br />

( )in equation 4.23<br />

denotes a solution <strong>for</strong> the layer volumetric transport which includes only the contribution from the explicit terms. The<br />

corresponding finite-difference equation <strong>for</strong> the explicit stage of the y-direction momentum equation is centered within layer k at<br />

1<br />

the point iΔx, ( j + ⁄ 2)Δy<br />

and is represented by<br />

n – 1<br />

Ûi, j+<br />

1⁄ 2,<br />

k Uij , + 1⁄ 2,<br />

k 2Δt ( ADVy) n ( CORy) n<br />

( BCLINICy) n<br />

– ( HDIFFy) n 1 – = + [ – –<br />

+<br />

] i, j+<br />

1⁄ 2,<br />

k . (4.24)<br />

Here the bracketed terms refer to groupings of terms in equation 3.60, and they are expanded fully in Appendix D (eqs. D.5–D.8).<br />

All terms in equations 4.23 and 4.24, except horizontal diffusion (HDIFF), are centered in time at time level n to achieve secondorder<br />

numerical accuracy. The horizontal diffusion is written backward-in-time at time level n − 1 because the centering of that<br />

term can result in a weak instability; although <strong>for</strong>mally this uncentered treatment of the horizontal diffusion introduces first-order<br />

truncation error, the actual size of the error should be miniscule. Presently in the model, the horizontal momentum exchange (or<br />

eddy viscosity) coefficient AH is defined at the center of each cell with whole integers of the indices i, j, k.<br />

The shear stress terms in equations 3.59 and 3.60 can be replaced by introducing the concept of the vertical momentum<br />

exchange (or eddy viscosity) coefficient: 34<br />

τxz ∂u<br />

------ = A ----ρ<br />

V ,<br />

∂z<br />

τyz ∂v<br />

------ = A ---ρ<br />

V . (4.25)<br />

∂z<br />

Then the finite-difference equation <strong>for</strong> the implicit stage of the x-momentum equation is written as<br />

n + 1<br />

+ , j, k=<br />

Ui 1⁄ 2<br />

+<br />

⎛<br />

⎝<br />

n<br />

g<br />

n + 1<br />

n + 1 n – 1<br />

n – 1<br />

⎛( U⁄ h)<br />

i + 1⁄ 2,<br />

j, k– 1–<br />

( U⁄ h)<br />

i + 1⁄ 2,<br />

j, k ui + 1<br />

2<br />

--------------------------------------------------------------------------- ⁄ , j, k– 1–<br />

ui + 1⁄ 2,<br />

j, k⎞<br />

⎜ + ------------------------------------------------------- ⎟<br />

⎝ n + 1<br />

n – 1 ⎠<br />

⁄ , , – ⁄<br />

⁄ , , – ⁄<br />

Δt<br />

– ----- hn i + 1⁄ 2,<br />

j, k<br />

Δx<br />

ρn ⎛ i + 1⁄ 2,<br />

j, 1⎞<br />

⎜--------------------- ⎟ ⋅ ( ζn + 1<br />

⎝ρn i + 1, j – ζn + 1<br />

i, j + ζn – 1<br />

i + 1, j – ζn – 1<br />

i, j )<br />

⎠<br />

i + 1⁄ 2,<br />

j, k<br />

Ûi + 1⁄ 2,<br />

j, k<br />

Δt⎜A Vi + 1⁄ 2 jk 1 ⋅<br />

, , – ⁄ 2<br />

hi + 1<br />

2 j k 1<br />

2<br />

hi + 1<br />

2 j k 1<br />

2<br />

n + 1<br />

n + 1<br />

n – 1 n – 1<br />

n ⎛( U⁄ h)<br />

i + 1<br />

2<br />

A ⁄ , jk , – ( U⁄ h)<br />

i + 1⁄ 2,<br />

j, k+ 1 ui + 1<br />

Vi<br />

---------------------------------------------------------------------------- ⁄ 2,<br />

jk , – ui + 1⁄ 2,<br />

jk , + 1⎞⎞<br />

– ⋅<br />

+ 1⁄ 2,<br />

jk , + 1 ⎜ + ------------------------------------------------------ ⎟⎟,<br />

(4.26)<br />

⁄ 2 ⎝ n + 1<br />

n – 1 ⎠⎠<br />

⁄ , , + ⁄<br />

⁄ , , + ⁄<br />

hi + 1<br />

2 jk 1<br />

2<br />

hi + 1<br />

2 jk 1<br />

2<br />

34 The z-derivatives of u and v in equations 4.25 are mathematically undefined at the layer interfaces because the layer-averaged velocities are<br />

discontinuous there. A z-derivative at an interface is there<strong>for</strong>e understood to imply a finite-difference quotient involving the layer variables above and below the<br />

interface.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!