25 |
C | o Controlling routine for the explicit part of the model |
C | o Controlling routine for the explicit part of the model |
26 |
C | dynamics. |
C | dynamics. |
27 |
C *==========================================================* |
C *==========================================================* |
|
C | This routine evaluates the "dynamics" terms for each |
|
|
C | block of ocean in turn. Because the blocks of ocean have |
|
|
C | overlap regions they are independent of one another. |
|
|
C | If terms involving lateral integrals are needed in this |
|
|
C | routine care will be needed. Similarly finite-difference |
|
|
C | operations with stencils wider than the overlap region |
|
|
C | require special consideration. |
|
|
C | The algorithm... |
|
|
C | |
|
|
C | "Correction Step" |
|
|
C | ================= |
|
|
C | Here we update the horizontal velocities with the surface |
|
|
C | pressure such that the resulting flow is either consistent |
|
|
C | with the free-surface evolution or the rigid-lid: |
|
|
C | U[n] = U* + dt x d/dx P |
|
|
C | V[n] = V* + dt x d/dy P |
|
|
C | W[n] = W* + dt x d/dz P (NH mode) |
|
|
C | |
|
|
C | "Calculation of Gs" |
|
|
C | =================== |
|
|
C | This is where all the accelerations and tendencies (ie. |
|
|
C | physics, parameterizations etc...) are calculated |
|
|
C | rho = rho ( theta[n], salt[n] ) |
|
|
C | b = b(rho, theta) |
|
|
C | K31 = K31 ( rho ) |
|
|
C | Gu[n] = Gu( u[n], v[n], wVel, b, ... ) |
|
|
C | Gv[n] = Gv( u[n], v[n], wVel, b, ... ) |
|
|
C | Gt[n] = Gt( theta[n], u[n], v[n], wVel, K31, ... ) |
|
|
C | Gs[n] = Gs( salt[n], u[n], v[n], wVel, K31, ... ) |
|
|
C | |
|
|
C | "Time-stepping" or "Prediction" |
|
|
C | ================================ |
|
|
C | The models variables are stepped forward with the appropriate |
|
|
C | time-stepping scheme (currently we use Adams-Bashforth II) |
|
|
C | - For momentum, the result is always *only* a "prediction" |
|
|
C | in that the flow may be divergent and will be "corrected" |
|
|
C | later with a surface pressure gradient. |
|
|
C | - Normally for tracers the result is the new field at time |
|
|
C | level [n+1} *BUT* in the case of implicit diffusion the result |
|
|
C | is also *only* a prediction. |
|
|
C | - We denote "predictors" with an asterisk (*). |
|
|
C | U* = U[n] + dt x ( 3/2 Gu[n] - 1/2 Gu[n-1] ) |
|
|
C | V* = V[n] + dt x ( 3/2 Gv[n] - 1/2 Gv[n-1] ) |
|
|
C | theta[n+1] = theta[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
|
|
C | salt[n+1] = salt[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
|
|
C | With implicit diffusion: |
|
|
C | theta* = theta[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
|
|
C | salt* = salt[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
|
|
C | (1 + dt * K * d_zz) theta[n] = theta* |
|
|
C | (1 + dt * K * d_zz) salt[n] = salt* |
|
|
C | |
|
|
C *==========================================================* |
|
28 |
C \ev |
C \ev |
29 |
C !USES: |
C !USES: |
30 |
IMPLICIT NONE |
IMPLICIT NONE |