| 1 |
C $Header: /u/gcmpack/MITgcm/pkg/mom_vecinv/mom_vi_hdissip.F,v 1.5 2004/02/07 23:15:47 dimitri Exp $ |
| 2 |
C $Name: $ |
| 3 |
|
| 4 |
#include "CPP_OPTIONS.h" |
| 5 |
|
| 6 |
SUBROUTINE MOM_VI_HDISSIP( |
| 7 |
I bi,bj,k, |
| 8 |
I hDiv,vort3,hFacZ,dStar,zStar, |
| 9 |
c I viscAh_Z,viscAh_D,viscA4_Z,viscA4_D, |
| 10 |
O uDissip,vDissip, |
| 11 |
I myThid) |
| 12 |
IMPLICIT NONE |
| 13 |
C |
| 14 |
C Calculate horizontal dissipation terms |
| 15 |
C [del^2 - del^4] (u,v) |
| 16 |
C |
| 17 |
|
| 18 |
C == Global variables == |
| 19 |
#include "SIZE.h" |
| 20 |
#include "GRID.h" |
| 21 |
#include "EEPARAMS.h" |
| 22 |
#include "PARAMS.h" |
| 23 |
|
| 24 |
C == Routine arguments == |
| 25 |
INTEGER bi,bj,k |
| 26 |
_RL hDiv(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 27 |
_RL vort3(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 28 |
_RS hFacZ(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 29 |
_RL dStar(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 30 |
_RL zStar(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 31 |
_RL viscAh_Z(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 32 |
_RL viscAh_D(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 33 |
_RL viscA4_Z(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 34 |
_RL viscA4_D(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 35 |
_RL uDissip(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 36 |
_RL vDissip(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 37 |
INTEGER myThid |
| 38 |
|
| 39 |
C == Local variables == |
| 40 |
INTEGER I,J |
| 41 |
_RL Zip,Zij,Zpj,Dim,Dij,Dmj,uD2,vD2,uD4,vD4 |
| 42 |
_RL Alin,Alth,grdVrt,vg2,vg4 |
| 43 |
LOGICAL useVariableViscosity |
| 44 |
|
| 45 |
useVariableViscosity= |
| 46 |
& (viscAhGrid*deltaTmom.NE.0.) |
| 47 |
& .OR.(viscA4Grid*deltaTmom.NE.0.) |
| 48 |
& .OR.(viscC2leith.NE.0.) |
| 49 |
& .OR.(viscC4leith.NE.0.) |
| 50 |
|
| 51 |
IF (deltaTmom.NE.0.) THEN |
| 52 |
vg2=viscAhGrid/deltaTmom |
| 53 |
vg4=viscA4Grid/deltaTmom |
| 54 |
ELSE |
| 55 |
vg2=0. |
| 56 |
vg4=0. |
| 57 |
ENDIF |
| 58 |
|
| 59 |
C - Viscosity |
| 60 |
IF (useVariableViscosity) THEN |
| 61 |
DO j=2-Oly,sNy+Oly-1 |
| 62 |
DO i=2-Olx,sNx+Olx-1 |
| 63 |
C This is the vector magnitude of the vorticity gradient |
| 64 |
c grdVrt=sqrt(0.25*( |
| 65 |
c & ((vort3(i+1,j)-vort3(i,j))*recip_DXG(i,j,bi,bj))**2 |
| 66 |
c & +((vort3(i,j+1)-vort3(i,j))*recip_DYG(i,j,bi,bj))**2 |
| 67 |
c & +((vort3(i+1,j+1)-vort3(i,j+1))*recip_DXG(i,j+1,bi,bj))**2 |
| 68 |
c & +((vort3(i+1,j+1)-vort3(i+1,j))*recip_DYG(i+1,j,bi,bj))**2 )) |
| 69 |
C but this approximation will work on cube (and differs by as much as 4X) |
| 70 |
grdVrt=abs((vort3(i+1,j)-vort3(i,j))*recip_DXG(i,j,bi,bj)) |
| 71 |
grdVrt=max(grdVrt, |
| 72 |
& abs((vort3(i,j+1)-vort3(i,j))*recip_DYG(i,j,bi,bj))) |
| 73 |
grdVrt=max(grdVrt, |
| 74 |
& abs((vort3(i+1,j+1)-vort3(i,j+1))*recip_DXG(i,j+1,bi,bj))) |
| 75 |
grdVrt=max(grdVrt, |
| 76 |
& abs((vort3(i+1,j+1)-vort3(i+1,j))*recip_DYG(i+1,j,bi,bj))) |
| 77 |
Alth=viscC2leith*grdVrt*(rA(i,j,bi,bj)**1.5) |
| 78 |
Alin=viscAh+vg2*rA ( i , j ,bi,bj) |
| 79 |
viscAh_D(i,j)=min(viscAhMax,Alin+Alth) |
| 80 |
|
| 81 |
Alth=viscC4leith*grdVrt*0.125*(rA(i,j,bi,bj)**2.5) |
| 82 |
Alin=viscA4+vg4*(rA ( i , j ,bi,bj)**2) |
| 83 |
viscA4_D(i,j)=min(viscA4Max,Alin+Alth) |
| 84 |
|
| 85 |
C This is the vector magnitude of the vorticity gradient |
| 86 |
c grdVrt=sqrt(0.25*( |
| 87 |
c & ((vort3(i+1,j)-vort3(i,j))*recip_DXG(i,j,bi,bj))**2 |
| 88 |
c & +((vort3(i,j+1)-vort3(i,j))*recip_DYG(i,j,bi,bj))**2 |
| 89 |
c & +((vort3(i-1,j)-vort3(i,j))*recip_DXG(i-1,j,bi,bj))**2 |
| 90 |
c & +((vort3(i,j-1)-vort3(i,j))*recip_DYG(i,j-1,bi,bj))**2 )) |
| 91 |
C but this approximation will work on cube (and differs by as much as 4X) |
| 92 |
grdVrt=abs((vort3(i+1,j)-vort3(i,j))*recip_DXG(i,j,bi,bj)) |
| 93 |
grdVrt=max(grdVrt, |
| 94 |
& abs((vort3(i,j+1)-vort3(i,j))*recip_DYG(i,j,bi,bj))) |
| 95 |
grdVrt=max(grdVrt, |
| 96 |
& abs((vort3(i-1,j)-vort3(i,j))*recip_DXG(i-1,j,bi,bj))) |
| 97 |
grdVrt=max(grdVrt, |
| 98 |
& abs((vort3(i,j-1)-vort3(i,j))*recip_DYG(i,j-1,bi,bj))) |
| 99 |
Alth=viscC2leith*grdVrt*(rAz(i,j,bi,bj)**1.5) |
| 100 |
Alin=viscAh+vg2*rAz( i , j ,bi,bj) |
| 101 |
viscAh_Z(i,j)=min(viscAhMax,Alin+Alth) |
| 102 |
|
| 103 |
Alth=viscC4leith*grdVrt*0.125*(rAz(i,j,bi,bj)**2.5) |
| 104 |
Alin=viscA4+vg4*(rAz( i , j ,bi,bj)**2) |
| 105 |
viscA4_Z(i,j)=min(viscA4Max,Alin+Alth) |
| 106 |
ENDDO |
| 107 |
ENDDO |
| 108 |
ELSE |
| 109 |
DO j=1-Oly,sNy+Oly |
| 110 |
DO i=1-Olx,sNx+Olx |
| 111 |
viscAh_D(i,j)=viscAh |
| 112 |
viscAh_Z(i,j)=viscAh |
| 113 |
viscA4_D(i,j)=viscA4 |
| 114 |
viscA4_Z(i,j)=viscA4 |
| 115 |
ENDDO |
| 116 |
ENDDO |
| 117 |
ENDIF |
| 118 |
|
| 119 |
C - Laplacian and bi-harmonic terms |
| 120 |
DO j=2-Oly,sNy+Oly-1 |
| 121 |
DO i=2-Olx,sNx+Olx-1 |
| 122 |
|
| 123 |
Dim=hDiv( i ,j-1) |
| 124 |
Dij=hDiv( i , j ) |
| 125 |
Dmj=hDiv(i-1, j ) |
| 126 |
Zip=hFacZ( i ,j+1)*vort3( i ,j+1) |
| 127 |
Zij=hFacZ( i , j )*vort3( i , j ) |
| 128 |
Zpj=hFacZ(i+1, j )*vort3(i+1, j ) |
| 129 |
|
| 130 |
C This bit scales the harmonic dissipation operator to be proportional |
| 131 |
C to the grid-cell area over the time-step. viscAh is then non-dimensional |
| 132 |
C and should be less than 1/8, for example viscAh=0.01 |
| 133 |
if (useVariableViscosity) then |
| 134 |
Dij=Dij*viscAh_D(i,j) |
| 135 |
Dim=Dim*viscAh_D(i,j-1) |
| 136 |
Dmj=Dmj*viscAh_D(i-1,j) |
| 137 |
Zij=Zij*viscAh_Z(i,j) |
| 138 |
Zip=Zip*viscAh_Z(i,j+1) |
| 139 |
Zpj=Zpj*viscAh_Z(i+1,j) |
| 140 |
uD2 = ( |
| 141 |
& cosFacU(j,bi,bj)*( Dij-Dmj )*recip_DXC(i,j,bi,bj) |
| 142 |
& -recip_hFacW(i,j,k,bi,bj)*( Zip-Zij )*recip_DYG(i,j,bi,bj) ) |
| 143 |
vD2 = ( |
| 144 |
& recip_hFacS(i,j,k,bi,bj)*( Zpj-Zij )*recip_DXG(i,j,bi,bj) |
| 145 |
& *cosFacV(j,bi,bj) |
| 146 |
& +( Dij-Dim )*recip_DYC(i,j,bi,bj) ) |
| 147 |
else |
| 148 |
uD2 = viscAh*( |
| 149 |
& cosFacU(j,bi,bj)*( Dij-Dmj )*recip_DXC(i,j,bi,bj) |
| 150 |
& -recip_hFacW(i,j,k,bi,bj)*( Zip-Zij )*recip_DYG(i,j,bi,bj) ) |
| 151 |
vD2 = viscAh*( |
| 152 |
& recip_hFacS(i,j,k,bi,bj)*( Zpj-Zij )*recip_DXG(i,j,bi,bj) |
| 153 |
& *cosFacV(j,bi,bj) |
| 154 |
& +( Dij-Dim )*recip_DYC(i,j,bi,bj) ) |
| 155 |
endif |
| 156 |
|
| 157 |
Dim=dyF( i ,j-1,bi,bj)*dStar( i ,j-1) |
| 158 |
Dij=dyF( i , j ,bi,bj)*dStar( i , j ) |
| 159 |
Dmj=dyF(i-1, j ,bi,bj)*dStar(i-1, j ) |
| 160 |
|
| 161 |
Zip=dxV( i ,j+1,bi,bj)*hFacZ( i ,j+1)*zStar( i ,j+1) |
| 162 |
Zij=dxV( i , j ,bi,bj)*hFacZ( i , j )*zStar( i , j ) |
| 163 |
Zpj=dxV(i+1, j ,bi,bj)*hFacZ(i+1, j )*zStar(i+1, j ) |
| 164 |
|
| 165 |
C This bit scales the harmonic dissipation operator to be proportional |
| 166 |
C to the grid-cell area over the time-step. viscAh is then non-dimensional |
| 167 |
C and should be less than 1/8, for example viscAh=0.01 |
| 168 |
if (useVariableViscosity) then |
| 169 |
Dij=Dij*viscA4_D(i,j) |
| 170 |
Dim=Dim*viscA4_D(i,j-1) |
| 171 |
Dmj=Dmj*viscA4_D(i-1,j) |
| 172 |
Zij=Zij*viscA4_Z(i,j) |
| 173 |
Zip=Zip*viscA4_Z(i,j+1) |
| 174 |
Zpj=Zpj*viscA4_Z(i+1,j) |
| 175 |
uD4 = recip_rAw(i,j,bi,bj)*( |
| 176 |
& ( (Dij-Dmj)*cosFacU(j,bi,bj) ) |
| 177 |
& -recip_hFacW(i,j,k,bi,bj)*( Zip-Zij ) ) |
| 178 |
vD4 = recip_rAs(i,j,bi,bj)*( |
| 179 |
& recip_hFacS(i,j,k,bi,bj)*( (Zpj-Zij)*cosFacV(j,bi,bj) ) |
| 180 |
& + ( Dij-Dim ) ) |
| 181 |
else |
| 182 |
uD4 = recip_rAw(i,j,bi,bj)*( |
| 183 |
& viscA4*( (Dij-Dmj)*cosFacU(j,bi,bj) ) |
| 184 |
& -recip_hFacW(i,j,k,bi,bj)*viscA4*( Zip-Zij ) ) |
| 185 |
vD4 = recip_rAs(i,j,bi,bj)*( |
| 186 |
& recip_hFacS(i,j,k,bi,bj)*viscA4*( (Zpj-Zij)*cosFacV(j,bi,bj) ) |
| 187 |
& + viscA4*( Dij-Dim ) ) |
| 188 |
endif |
| 189 |
|
| 190 |
uDissip(i,j) = uD2 - uD4 |
| 191 |
vDissip(i,j) = vD2 - vD4 |
| 192 |
|
| 193 |
ENDDO |
| 194 |
ENDDO |
| 195 |
|
| 196 |
RETURN |
| 197 |
END |
Parent Directory
| 
Revision Log
| 
Revision Graph