PUBLIC INTERFACE ~ PUBLIC ROUTINES ~ NAMELIST

Module ocean_operators_mod

Contact:  S.M. Griffies
Reviewers:  A. Rosati ,  Zhi Liang ,  Alexander Pletzer
Change History: WebCVS Log

OVERVIEW

Operators for MOM

This module computes discrete operators used by MOM.

OTHER MODULES USED

     mpp_domains_mod
             mpp_mod
             fms_mod
   ocean_domains_mod
ocean_parameters_mod
     ocean_types_mod

PUBLIC INTERFACE

ocean_operators_init:
set_barotropic_domain:
get_use_legacy_DIV_UD:
REMAP_NT_TO_NU:
REMAP_ET_TO_EU:
REMAP_BT_TO_BU:
DIV_UD:
GRAD_BAROTROPIC_P:
S2D:
LAP_T:
FAX:
BAX:
FAY:
BAY:
BDX_EU:
BDX_ET:
FDX_U:
FDX_ZT:
FDX_T:
FDY_ZT:
FDY_T:
FDX_NT:
BDY_NU:
BDY_NT:
FDY_U:
FDY_ET:
FDZ_T:
FMX:
FMY:

PUBLIC ROUTINES

  1. ocean_operators_init

    DESCRIPTION
    Initialize the operator module


  2. set_barotropic_domain

    DESCRIPTION
    Set the barotropic domain used in barotropic time step.


    INPUT
    Domain_in    Store the barotropic domain.
       [type(ocean_domain_type)]

  3. get_use_legacy_DIV_UD

    DESCRIPTION
    Return the value of ocean_operators_nml variable use_legacy_DIV_UD


  4. REMAP_NT_TO_NU

    DESCRIPTION
    REMAP_NT_TO_NU remaps a normal flux at the north face of T-cells to the north face of U-cells


    INPUT
    a    Field to be remapped
       [real, dimension(isd:ied,jsd:jed)]

  5. REMAP_ET_TO_EU

    DESCRIPTION
    REMAP_ET_TO_EU remaps a normal flux at the east face of T-cells to the east face of U-cells


    INPUT
    a    Field to be remapped
       [real, dimension(isd:ied,jsd:jed)]

  6. REMAP_BT_TO_BU

    DESCRIPTION
    REMAP_BT_TO_BU remaps a T-cell thickness or vertical velocity on the base of T-cells to U-cells


    INPUT
    a    Field to be remapped
       [real, dimension(isd:ied,jsd:jed)]

  7. DIV_UD

    DESCRIPTION
    Compute divergence of vertically integrated velocity. For MOM with generalized vertical coordinates, ud has an extra density factor built in. The Bgrid uh and vh fields are located on the corner points, so require some backward averaging. The Cgrid uh and vh fields are located on the T-cell faces, so need no spatial averaging. Bgrid code is a speedier version of
    uhy(:,:) = BAY(ud(:,:,1)*dyu(:,:))
    vhx(:,:) = BAX(ud(:,:,2)*dxu(:,:))
    DIV_UD(ud) = BDX_ET(uhy(:,:)/dyte(:,:)) + BDY_NT(vhx(:,:)/dxtn(:,:))


  8. GRAD_BAROTROPIC_P

    DESCRIPTION
    Compute horizontal gradient of the pressure field associated with either the free surface height or the bottom pressure. Account taken here for either Bgrid or Cgrid operators. For the Bgrid, the gradient is centered onto the U-cell for use in updating barotropic velocity. For the Cgrid, the two components are centred on the T-cell faces. The Bgrid algorithm is a speedier version of
    grad_barotropic_p(:,:,1) = FDX_NT(FAY(press(:,:)))
    grad_barotropic_p(:,:,2) = FDY_ET(FAX(press(:,:)))


  9. S2D

    DESCRIPTION
    Smooth a 2D field with a 2D version of a 1D filter with weights (1/4, 1/2, 1/4)


    INPUT
    a    Field to be smoothed
       [real, dimension(isd:ied,jsd:jed)]

  10. LAP_T

    DESCRIPTION
    Compute horizontal 5-point Laplacian operator on eta_t. Result lives at T-cell center. Redundancy update for tripolar is needed to conserve total volume and tracer. It is likely unimportant when call LAP_T from within the barotropic loop. Yet it is essential when call LAP_T from ocean_surface_smooth. Mixing coefficient is assumed to be centred on the T-cell. It is averaged to compute its value on the i-face and j-face for computing fluxes.


  11. FAX

    DESCRIPTION
    Forwards average in the i-direction on the X-axis. If input is a(i,j) then output is defined at (i+1/2,j)


    INPUT
    a    Field to be averaged
       [real, dimension(isd:ied,jsd:jed)]

  12. BAX

    DESCRIPTION
    Backwards average in the i-direction along the X-axis. If input is a(i,j) then output is defined at (i-1/2,j)


    INPUT
    a    Field to be averaged
       [real, dimension(isd:ied,jsd:jed)]

  13. FAY

    DESCRIPTION
    Forwards average in the j-direction on the Y-axis If input is a(i,j) then output is defined at (i,j+1/2)


    INPUT
    a    Field to be averaged
       [real, dimension(isd:ied,jsd:jed)]

  14. BAY

    DESCRIPTION
    Backwards average in the j-direction along the Y-axis If input is a(i,j) then output is defined at (i,j-1/2)


    INPUT
    a    Field to be averaged
       [real, dimension(isd:ied,jsd:jed)]

  15. BDX_EU

    DESCRIPTION
    Backwards Derivative in X of a quantity defined on the East face of a U-cell If input is a(i,j) then output is defined at (i-1/2,j)


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  16. BDX_ET

    DESCRIPTION
    Backwards derivative in X of a quantity defined on the East face of a T-cell. If input is a(i,j) then output is defined at (i-1/2,j)


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  17. FDX_U

    DESCRIPTION
    Forward Derivative in X of a quantity defined on the grid point of a U-cell. If input is a(i,j) then output is defined at (i+1/2,j).


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  18. FDX_ZT

    DESCRIPTION
    Forward Derivative in X of a quantity defined on a tracer grid point where it is necessary to take derivative with depth held constant. When grid points live at different depths, then have an extra contribution to the derivative. Input a(i,j,1) is at the grid point of a T-cell at level k-1. Input a(i,j,2) is at the grid point of a T-cell at level k. Output is defined at (i+1/2,j) which is at the east face in a T-cell at level k.


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]
    k    Depth level
       [integer]

  19. FDX_T

    DESCRIPTION
    Forward Derivative in X of a quantity defined on the grid point of a T-cell. For lateral derivatives where vertical coordinate is held constant. Input a(i,j) is at the grid point of a T-cell Output is defined at (i+1/2,j) which is at the east face in a T-cell.


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  20. FDY_ZT

    DESCRIPTION
    Forward Derivative in Y of a quantity defined on a tracer grid point where it is necessary to take derivative with depth held constant. When grid points live at different depths, then have an extra contribution to the derivative. Input a(i,j,1) is at the grid point of a T-cell at level k-1. Input a(i,j,2) is at the grid point of a T-cell at level k. Output is defined at (i,j+1/2) which is at the north face in a T-cell at level k.


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]
    k    Depth level
       [integer]

  21. FDY_T

    DESCRIPTION
    Forward Derivative in Y of a quantity defined on the grid Point of a T-cell. For lateral derivatives where vertical coordinate is held constant. Input a(i,j) is at the grid point of a T-cell. Output is defined at (i,j+1/2) which is at the north face in a T-cell.


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  22. FDX_NT

    DESCRIPTION
    Forward Derivative in X of a quantity defined on the North face of a T-cell. If input is a(i,j) then output is defined at (i+1/2,j).


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  23. BDY_NU

    DESCRIPTION
    Backward Derivative in Y of a quantity defined on the North face of a U-cell. If input is a(i,j) then output is defined at (i,j-1/2).


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  24. BDY_NT

    DESCRIPTION
    Backward Derivative in Y of a quantity defined on the North face of a T-cell. If input is a(i,j) then output is defined at (i,j-1/2).


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  25. FDY_U

    DESCRIPTION
    Forward Derivative in Y of a quantity defined on the grid Point of a U-cell. If input is a(i,j) then output is defined at (i,j+1/2).


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  26. FDY_ET

    DESCRIPTION
    Forward Derivative in Y of a quantity defined on the East face of a T-cell. If input is a(i,j) then output is defined at (i,j+1/2).


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]

  27. FDZ_T

    DESCRIPTION
    Forward Derivative in Z of a field on the T-cell point at level k. input a(i,j,1) is at the grid point of a T-cell at level k. input a(i,j,2) is at the grid point of a T-cell at level k+1. output is at (i,j,k+3/2) which is bottom face of T-cell at level k. minus sign due to convention that z-increases upwards, whereas k increases downward.


    INPUT
    a    Field to be finite differenced
       [real, dimension(isd:ied,jsd:jed)]
    k    Depth level index
       [integer]

  28. FMX

    DESCRIPTION
    Forwards Minimum in the X direction. If input is a(i,j) then output is defined at (i+1/2,j).


    INPUT
    a    Field to find minimum
       [real, dimension(isd:ied,jsd:jed)]

  29. FMY

    DESCRIPTION
    Forwards Minimum in the Y direction. If input is a(i,j) then output is defined at (i,j+1/2).


    INPUT
    a    Field to find minimum
       [real, dimension(isd:ied,jsd:jed)]

NAMELIST

&ocean_operators_nml

use_legacy_DIV_UD
Set use_legacy_DIV_UD=.true. to reproduce Riga results for DIV_UD on Bgrid. For the case that the model grid is tripolar grid, when barotropic_halo > 1 in ocean_barotropic.F90, then we must set use_legacy_DIV_UD=.false., since will not reproduce between different number of processors if set use_legacy_DIV_UD=.true. Tests indicate that with wider barotropic halos, there are some performance enhancements for use_legacy_DIV_UD=.false. Hence, the default is use_legacy_DIV_UD=.false. For the case that the model grid is regular lat-lon grid, use_legacy_DIV_UD could be set to .true. or .false. for any positive value of barotropic_halo. Note that the only difference between the new and old DIV_UD is order of operations induced by parentheses, which occurs in the tripolar fold region in the Arctic: old: DIV_UD(i,j) = (uh_bay - uhim_bay + vh_bax - vhjm_bax)*datr_bt(i,j) new: DIV_UD(i,j) = ((uh_bay - uhim_bay) + (vh_bax - vhjm_bax))*datr_bt(i,j)
[logical]

NOTES

All operators will be replaced by generic forms when Fortran can properly support functions of allocatable arrays. The problems presently with this replacement are are as follows: Allocatable arrays cannot be inside of derived types. Only pointers to allocatable arrays can be inside derived types. Supposedly the former will be allowed in Fortran 95 Also, functions cannot be typed as a derived type without conflicts which preclude using function as general operators operating on derived types.

Mnemonics for simple operators 1st letter (direction of operation)
F => Forward direction with respect to the index.
B => Backward direction with respect to the index. 2nd letter (operation)
D => Derivative
A => Average
M => Minimum 3rd letter (axis) X => along the X axis
Y => along the Y axis
Z => along the Z axis 4th letter (placement of quantity being operated on) E => East face
N => North face
B => Bottom face
P => Point (grid point within cell) 5th letter (type of grid cell) U => U-cell
T => T-cell

top