| 1 |
C $Header: /u/gcmpack/models/MITgcmUV/model/src/thermodynamics.F,v 1.13 2001/09/28 03:36:16 adcroft Exp $ |
| 2 |
C $Name: release1_beta1 $ |
| 3 |
|
| 4 |
#include "CPP_OPTIONS.h" |
| 5 |
|
| 6 |
CBOP |
| 7 |
C !ROUTINE: THERMODYNAMICS |
| 8 |
C !INTERFACE: |
| 9 |
SUBROUTINE THERMODYNAMICS(myTime, myIter, myThid) |
| 10 |
C !DESCRIPTION: \bv |
| 11 |
C *==========================================================* |
| 12 |
C | SUBROUTINE THERMODYNAMICS |
| 13 |
C | o Controlling routine for the prognostic part of the |
| 14 |
C | thermo-dynamics. |
| 15 |
C *=========================================================== |
| 16 |
C | The algorithm... |
| 17 |
C | |
| 18 |
C | "Correction Step" |
| 19 |
C | ================= |
| 20 |
C | Here we update the horizontal velocities with the surface |
| 21 |
C | pressure such that the resulting flow is either consistent |
| 22 |
C | with the free-surface evolution or the rigid-lid: |
| 23 |
C | U[n] = U* + dt x d/dx P |
| 24 |
C | V[n] = V* + dt x d/dy P |
| 25 |
C | |
| 26 |
C | "Calculation of Gs" |
| 27 |
C | =================== |
| 28 |
C | This is where all the accelerations and tendencies (ie. |
| 29 |
C | physics, parameterizations etc...) are calculated |
| 30 |
C | rho = rho ( theta[n], salt[n] ) |
| 31 |
C | b = b(rho, theta) |
| 32 |
C | K31 = K31 ( rho ) |
| 33 |
C | Gu[n] = Gu( u[n], v[n], wVel, b, ... ) |
| 34 |
C | Gv[n] = Gv( u[n], v[n], wVel, b, ... ) |
| 35 |
C | Gt[n] = Gt( theta[n], u[n], v[n], wVel, K31, ... ) |
| 36 |
C | Gs[n] = Gs( salt[n], u[n], v[n], wVel, K31, ... ) |
| 37 |
C | |
| 38 |
C | "Time-stepping" or "Prediction" |
| 39 |
C | ================================ |
| 40 |
C | The models variables are stepped forward with the appropriate |
| 41 |
C | time-stepping scheme (currently we use Adams-Bashforth II) |
| 42 |
C | - For momentum, the result is always *only* a "prediction" |
| 43 |
C | in that the flow may be divergent and will be "corrected" |
| 44 |
C | later with a surface pressure gradient. |
| 45 |
C | - Normally for tracers the result is the new field at time |
| 46 |
C | level [n+1} *BUT* in the case of implicit diffusion the result |
| 47 |
C | is also *only* a prediction. |
| 48 |
C | - We denote "predictors" with an asterisk (*). |
| 49 |
C | U* = U[n] + dt x ( 3/2 Gu[n] - 1/2 Gu[n-1] ) |
| 50 |
C | V* = V[n] + dt x ( 3/2 Gv[n] - 1/2 Gv[n-1] ) |
| 51 |
C | theta[n+1] = theta[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
| 52 |
C | salt[n+1] = salt[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
| 53 |
C | With implicit diffusion: |
| 54 |
C | theta* = theta[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
| 55 |
C | salt* = salt[n] + dt x ( 3/2 Gt[n] - 1/2 atG[n-1] ) |
| 56 |
C | (1 + dt * K * d_zz) theta[n] = theta* |
| 57 |
C | (1 + dt * K * d_zz) salt[n] = salt* |
| 58 |
C | |
| 59 |
C *==========================================================* |
| 60 |
C \ev |
| 61 |
|
| 62 |
C !USES: |
| 63 |
IMPLICIT NONE |
| 64 |
C == Global variables === |
| 65 |
#include "SIZE.h" |
| 66 |
#include "EEPARAMS.h" |
| 67 |
#include "PARAMS.h" |
| 68 |
#include "DYNVARS.h" |
| 69 |
#include "GRID.h" |
| 70 |
#include "GAD.h" |
| 71 |
#ifdef ALLOW_PASSIVE_TRACER |
| 72 |
#include "TR1.h" |
| 73 |
#endif |
| 74 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 75 |
# include "tamc.h" |
| 76 |
# include "tamc_keys.h" |
| 77 |
# include "FFIELDS.h" |
| 78 |
# ifdef ALLOW_KPP |
| 79 |
# include "KPP.h" |
| 80 |
# endif |
| 81 |
# ifdef ALLOW_GMREDI |
| 82 |
# include "GMREDI.h" |
| 83 |
# endif |
| 84 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 85 |
#ifdef ALLOW_TIMEAVE |
| 86 |
#include "TIMEAVE_STATV.h" |
| 87 |
#endif |
| 88 |
|
| 89 |
C !INPUT/OUTPUT PARAMETERS: |
| 90 |
C == Routine arguments == |
| 91 |
C myTime - Current time in simulation |
| 92 |
C myIter - Current iteration number in simulation |
| 93 |
C myThid - Thread number for this instance of the routine. |
| 94 |
_RL myTime |
| 95 |
INTEGER myIter |
| 96 |
INTEGER myThid |
| 97 |
|
| 98 |
C !LOCAL VARIABLES: |
| 99 |
C == Local variables |
| 100 |
C xA, yA - Per block temporaries holding face areas |
| 101 |
C uTrans, vTrans, rTrans - Per block temporaries holding flow |
| 102 |
C transport |
| 103 |
C o uTrans: Zonal transport |
| 104 |
C o vTrans: Meridional transport |
| 105 |
C o rTrans: Vertical transport |
| 106 |
C maskUp o maskUp: land/water mask for W points |
| 107 |
C fVer[STUV] o fVer: Vertical flux term - note fVer |
| 108 |
C is "pipelined" in the vertical |
| 109 |
C so we need an fVer for each |
| 110 |
C variable. |
| 111 |
C rhoK, rhoKM1 - Density at current level, and level above |
| 112 |
C phiHyd - Hydrostatic part of the potential phiHydi. |
| 113 |
C In z coords phiHydiHyd is the hydrostatic |
| 114 |
C Potential (=pressure/rho0) anomaly |
| 115 |
C In p coords phiHydiHyd is the geopotential |
| 116 |
C surface height anomaly. |
| 117 |
C phiSurfX, - gradient of Surface potentiel (Pressure/rho, ocean) |
| 118 |
C phiSurfY or geopotentiel (atmos) in X and Y direction |
| 119 |
C KappaRT, - Total diffusion in vertical for T and S. |
| 120 |
C KappaRS (background + spatially varying, isopycnal term). |
| 121 |
C iMin, iMax - Ranges and sub-block indices on which calculations |
| 122 |
C jMin, jMax are applied. |
| 123 |
C bi, bj |
| 124 |
C k, kup, - Index for layer above and below. kup and kDown |
| 125 |
C kDown, km1 are switched with layer to be the appropriate |
| 126 |
C index into fVerTerm. |
| 127 |
_RS xA (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 128 |
_RS yA (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 129 |
_RL uTrans (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 130 |
_RL vTrans (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 131 |
_RL rTrans (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 132 |
_RS maskUp (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 133 |
_RL fVerT (1-OLx:sNx+OLx,1-OLy:sNy+OLy,2) |
| 134 |
_RL fVerS (1-OLx:sNx+OLx,1-OLy:sNy+OLy,2) |
| 135 |
_RL fVerTr1 (1-OLx:sNx+OLx,1-OLy:sNy+OLy,2) |
| 136 |
_RL phiHyd (1-OLx:sNx+OLx,1-OLy:sNy+OLy,Nr) |
| 137 |
_RL rhokm1 (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 138 |
_RL rhok (1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 139 |
_RL phiSurfX(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 140 |
_RL phiSurfY(1-OLx:sNx+OLx,1-OLy:sNy+OLy) |
| 141 |
_RL KappaRT (1-Olx:sNx+Olx,1-Oly:sNy+Oly,Nr) |
| 142 |
_RL KappaRS (1-Olx:sNx+Olx,1-Oly:sNy+Oly,Nr) |
| 143 |
_RL sigmaX (1-OLx:sNx+OLx,1-OLy:sNy+OLy,Nr) |
| 144 |
_RL sigmaY (1-OLx:sNx+OLx,1-OLy:sNy+OLy,Nr) |
| 145 |
_RL sigmaR (1-OLx:sNx+OLx,1-OLy:sNy+OLy,Nr) |
| 146 |
C This is currently used by IVDC and Diagnostics |
| 147 |
_RL ConvectCount (1-OLx:sNx+OLx,1-OLy:sNy+OLy,Nr) |
| 148 |
INTEGER iMin, iMax |
| 149 |
INTEGER jMin, jMax |
| 150 |
INTEGER bi, bj |
| 151 |
INTEGER i, j |
| 152 |
INTEGER k, km1, kup, kDown |
| 153 |
|
| 154 |
Cjmc : add for phiHyd output <- but not working if multi tile per CPU |
| 155 |
c CHARACTER*(MAX_LEN_MBUF) suff |
| 156 |
c LOGICAL DIFFERENT_MULTIPLE |
| 157 |
c EXTERNAL DIFFERENT_MULTIPLE |
| 158 |
Cjmc(end) |
| 159 |
CEOP |
| 160 |
|
| 161 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 162 |
C-- dummy statement to end declaration part |
| 163 |
ikey = 1 |
| 164 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 165 |
|
| 166 |
C-- Set up work arrays with valid (i.e. not NaN) values |
| 167 |
C These inital values do not alter the numerical results. They |
| 168 |
C just ensure that all memory references are to valid floating |
| 169 |
C point numbers. This prevents spurious hardware signals due to |
| 170 |
C uninitialised but inert locations. |
| 171 |
DO j=1-OLy,sNy+OLy |
| 172 |
DO i=1-OLx,sNx+OLx |
| 173 |
xA(i,j) = 0. _d 0 |
| 174 |
yA(i,j) = 0. _d 0 |
| 175 |
uTrans(i,j) = 0. _d 0 |
| 176 |
vTrans(i,j) = 0. _d 0 |
| 177 |
DO k=1,Nr |
| 178 |
phiHyd(i,j,k) = 0. _d 0 |
| 179 |
sigmaX(i,j,k) = 0. _d 0 |
| 180 |
sigmaY(i,j,k) = 0. _d 0 |
| 181 |
sigmaR(i,j,k) = 0. _d 0 |
| 182 |
ENDDO |
| 183 |
rhoKM1 (i,j) = 0. _d 0 |
| 184 |
rhok (i,j) = 0. _d 0 |
| 185 |
phiSurfX(i,j) = 0. _d 0 |
| 186 |
phiSurfY(i,j) = 0. _d 0 |
| 187 |
ENDDO |
| 188 |
ENDDO |
| 189 |
|
| 190 |
|
| 191 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 192 |
C-- HPF directive to help TAMC |
| 193 |
CHPF$ INDEPENDENT |
| 194 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 195 |
|
| 196 |
DO bj=myByLo(myThid),myByHi(myThid) |
| 197 |
|
| 198 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 199 |
C-- HPF directive to help TAMC |
| 200 |
CHPF$ INDEPENDENT, NEW (rTrans,fVerT,fVerS |
| 201 |
CHPF$& ,phiHyd,utrans,vtrans,xA,yA |
| 202 |
CHPF$& ,KappaRT,KappaRS |
| 203 |
CHPF$& ) |
| 204 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 205 |
|
| 206 |
DO bi=myBxLo(myThid),myBxHi(myThid) |
| 207 |
|
| 208 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 209 |
act1 = bi - myBxLo(myThid) |
| 210 |
max1 = myBxHi(myThid) - myBxLo(myThid) + 1 |
| 211 |
act2 = bj - myByLo(myThid) |
| 212 |
max2 = myByHi(myThid) - myByLo(myThid) + 1 |
| 213 |
act3 = myThid - 1 |
| 214 |
max3 = nTx*nTy |
| 215 |
act4 = ikey_dynamics - 1 |
| 216 |
ikey = (act1 + 1) + act2*max1 |
| 217 |
& + act3*max1*max2 |
| 218 |
& + act4*max1*max2*max3 |
| 219 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 220 |
|
| 221 |
C-- Set up work arrays that need valid initial values |
| 222 |
DO j=1-OLy,sNy+OLy |
| 223 |
DO i=1-OLx,sNx+OLx |
| 224 |
rTrans (i,j) = 0. _d 0 |
| 225 |
fVerT (i,j,1) = 0. _d 0 |
| 226 |
fVerT (i,j,2) = 0. _d 0 |
| 227 |
fVerS (i,j,1) = 0. _d 0 |
| 228 |
fVerS (i,j,2) = 0. _d 0 |
| 229 |
fVerTr1(i,j,1) = 0. _d 0 |
| 230 |
fVerTr1(i,j,2) = 0. _d 0 |
| 231 |
ENDDO |
| 232 |
ENDDO |
| 233 |
|
| 234 |
DO k=1,Nr |
| 235 |
DO j=1-OLy,sNy+OLy |
| 236 |
DO i=1-OLx,sNx+OLx |
| 237 |
C This is currently also used by IVDC and Diagnostics |
| 238 |
ConvectCount(i,j,k) = 0. |
| 239 |
KappaRT(i,j,k) = 0. _d 0 |
| 240 |
KappaRS(i,j,k) = 0. _d 0 |
| 241 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 242 |
gT(i,j,k,bi,bj) = 0. _d 0 |
| 243 |
gS(i,j,k,bi,bj) = 0. _d 0 |
| 244 |
#ifdef ALLOW_PASSIVE_TRACER |
| 245 |
gTr1(i,j,k,bi,bj) = 0. _d 0 |
| 246 |
#endif |
| 247 |
#endif |
| 248 |
ENDDO |
| 249 |
ENDDO |
| 250 |
ENDDO |
| 251 |
|
| 252 |
iMin = 1-OLx+1 |
| 253 |
iMax = sNx+OLx |
| 254 |
jMin = 1-OLy+1 |
| 255 |
jMax = sNy+OLy |
| 256 |
|
| 257 |
|
| 258 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 259 |
CADJ STORE theta(:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 260 |
CADJ STORE salt (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 261 |
#ifdef ALLOW_KPP |
| 262 |
CADJ STORE uvel (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 263 |
CADJ STORE vvel (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 264 |
#endif |
| 265 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 266 |
|
| 267 |
C-- Start of diagnostic loop |
| 268 |
DO k=Nr,1,-1 |
| 269 |
|
| 270 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 271 |
C? Patrick, is this formula correct now that we change the loop range? |
| 272 |
C? Do we still need this? |
| 273 |
cph kkey formula corrected. |
| 274 |
cph Needed for rhok, rhokm1, in the case useGMREDI. |
| 275 |
kkey = (ikey-1)*Nr + k |
| 276 |
CADJ STORE rhokm1(:,:) = comlev1_bibj_k , key=kkey, byte=isbyte |
| 277 |
CADJ STORE rhok (:,:) = comlev1_bibj_k , key=kkey, byte=isbyte |
| 278 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 279 |
|
| 280 |
C-- Integrate continuity vertically for vertical velocity |
| 281 |
CALL INTEGRATE_FOR_W( |
| 282 |
I bi, bj, k, uVel, vVel, |
| 283 |
O wVel, |
| 284 |
I myThid ) |
| 285 |
|
| 286 |
#ifdef ALLOW_OBCS |
| 287 |
#ifdef ALLOW_NONHYDROSTATIC |
| 288 |
C-- Apply OBC to W if in N-H mode |
| 289 |
IF (useOBCS.AND.nonHydrostatic) THEN |
| 290 |
CALL OBCS_APPLY_W( bi, bj, k, wVel, myThid ) |
| 291 |
ENDIF |
| 292 |
#endif /* ALLOW_NONHYDROSTATIC */ |
| 293 |
#endif /* ALLOW_OBCS */ |
| 294 |
|
| 295 |
C-- Calculate gradients of potential density for isoneutral |
| 296 |
C slope terms (e.g. GM/Redi tensor or IVDC diffusivity) |
| 297 |
c IF ( k.GT.1 .AND. (useGMRedi.OR.ivdc_kappa.NE.0.) ) THEN |
| 298 |
IF ( useGMRedi .OR. (k.GT.1 .AND. ivdc_kappa.NE.0.) ) THEN |
| 299 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 300 |
CADJ STORE theta(:,:,k,bi,bj) = comlev1_bibj_k, key=kkey, byte=isbyte |
| 301 |
CADJ STORE salt (:,:,k,bi,bj) = comlev1_bibj_k, key=kkey, byte=isbyte |
| 302 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 303 |
CALL FIND_RHO( |
| 304 |
I bi, bj, iMin, iMax, jMin, jMax, k, k, eosType, |
| 305 |
I theta, salt, |
| 306 |
O rhoK, |
| 307 |
I myThid ) |
| 308 |
IF (k.GT.1) THEN |
| 309 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 310 |
CADJ STORE theta(:,:,k-1,bi,bj) = comlev1_bibj_k, key=kkey, byte=isbyte |
| 311 |
CADJ STORE salt (:,:,k-1,bi,bj) = comlev1_bibj_k, key=kkey, byte=isbyte |
| 312 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 313 |
CALL FIND_RHO( |
| 314 |
I bi, bj, iMin, iMax, jMin, jMax, k-1, k, eosType, |
| 315 |
I theta, salt, |
| 316 |
O rhoKm1, |
| 317 |
I myThid ) |
| 318 |
ENDIF |
| 319 |
CALL GRAD_SIGMA( |
| 320 |
I bi, bj, iMin, iMax, jMin, jMax, k, |
| 321 |
I rhoK, rhoKm1, rhoK, |
| 322 |
O sigmaX, sigmaY, sigmaR, |
| 323 |
I myThid ) |
| 324 |
ENDIF |
| 325 |
|
| 326 |
C-- Implicit Vertical Diffusion for Convection |
| 327 |
c ==> should use sigmaR !!! |
| 328 |
IF (k.GT.1 .AND. ivdc_kappa.NE.0.) THEN |
| 329 |
CALL CALC_IVDC( |
| 330 |
I bi, bj, iMin, iMax, jMin, jMax, k, |
| 331 |
I rhoKm1, rhoK, |
| 332 |
U ConvectCount, KappaRT, KappaRS, |
| 333 |
I myTime, myIter, myThid) |
| 334 |
ENDIF |
| 335 |
|
| 336 |
C-- end of diagnostic k loop (Nr:1) |
| 337 |
ENDDO |
| 338 |
|
| 339 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 340 |
cph avoids recomputation of integrate_for_w |
| 341 |
CADJ STORE wvel (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 342 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 343 |
|
| 344 |
#ifdef ALLOW_OBCS |
| 345 |
C-- Calculate future values on open boundaries |
| 346 |
IF (useOBCS) THEN |
| 347 |
CALL OBCS_CALC( bi, bj, myTime+deltaT, |
| 348 |
I uVel, vVel, wVel, theta, salt, |
| 349 |
I myThid ) |
| 350 |
ENDIF |
| 351 |
#endif /* ALLOW_OBCS */ |
| 352 |
|
| 353 |
C-- Determines forcing terms based on external fields |
| 354 |
C relaxation terms, etc. |
| 355 |
CALL EXTERNAL_FORCING_SURF( |
| 356 |
I bi, bj, iMin, iMax, jMin, jMax, |
| 357 |
I myThid ) |
| 358 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 359 |
cph needed for KPP |
| 360 |
CADJ STORE surfacetendencyU(:,:,bi,bj) |
| 361 |
CADJ & = comlev1_bibj, key=ikey, byte=isbyte |
| 362 |
CADJ STORE surfacetendencyV(:,:,bi,bj) |
| 363 |
CADJ & = comlev1_bibj, key=ikey, byte=isbyte |
| 364 |
CADJ STORE surfacetendencyS(:,:,bi,bj) |
| 365 |
CADJ & = comlev1_bibj, key=ikey, byte=isbyte |
| 366 |
CADJ STORE surfacetendencyT(:,:,bi,bj) |
| 367 |
CADJ & = comlev1_bibj, key=ikey, byte=isbyte |
| 368 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 369 |
|
| 370 |
#ifdef ALLOW_GMREDI |
| 371 |
|
| 372 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 373 |
CADJ STORE sigmaX(:,:,:) = comlev1, key=ikey, byte=isbyte |
| 374 |
CADJ STORE sigmaY(:,:,:) = comlev1, key=ikey, byte=isbyte |
| 375 |
CADJ STORE sigmaR(:,:,:) = comlev1, key=ikey, byte=isbyte |
| 376 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 377 |
C-- Calculate iso-neutral slopes for the GM/Redi parameterisation |
| 378 |
IF (useGMRedi) THEN |
| 379 |
DO k=1,Nr |
| 380 |
CALL GMREDI_CALC_TENSOR( |
| 381 |
I bi, bj, iMin, iMax, jMin, jMax, k, |
| 382 |
I sigmaX, sigmaY, sigmaR, |
| 383 |
I myThid ) |
| 384 |
ENDDO |
| 385 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 386 |
ELSE |
| 387 |
DO k=1, Nr |
| 388 |
CALL GMREDI_CALC_TENSOR_DUMMY( |
| 389 |
I bi, bj, iMin, iMax, jMin, jMax, k, |
| 390 |
I sigmaX, sigmaY, sigmaR, |
| 391 |
I myThid ) |
| 392 |
ENDDO |
| 393 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 394 |
ENDIF |
| 395 |
|
| 396 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 397 |
CADJ STORE Kwx(:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 398 |
CADJ STORE Kwy(:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 399 |
CADJ STORE Kwz(:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 400 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 401 |
|
| 402 |
#endif /* ALLOW_GMREDI */ |
| 403 |
|
| 404 |
#ifdef ALLOW_KPP |
| 405 |
C-- Compute KPP mixing coefficients |
| 406 |
IF (useKPP) THEN |
| 407 |
CALL KPP_CALC( |
| 408 |
I bi, bj, myTime, myThid ) |
| 409 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 410 |
ELSE |
| 411 |
CALL KPP_CALC_DUMMY( |
| 412 |
I bi, bj, myTime, myThid ) |
| 413 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 414 |
ENDIF |
| 415 |
|
| 416 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 417 |
CADJ STORE KPPghat (:,:,:,bi,bj) |
| 418 |
CADJ & , KPPdiffKzT(:,:,:,bi,bj) |
| 419 |
CADJ & , KPPdiffKzS(:,:,:,bi,bj) |
| 420 |
CADJ & , KPPfrac (:,: ,bi,bj) |
| 421 |
CADJ & = comlev1_bibj, key=ikey, byte=isbyte |
| 422 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 423 |
|
| 424 |
#endif /* ALLOW_KPP */ |
| 425 |
|
| 426 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 427 |
CADJ STORE KappaRT(:,:,:) = comlev1_bibj, key=ikey, byte=isbyte |
| 428 |
CADJ STORE KappaRS(:,:,:) = comlev1_bibj, key=ikey, byte=isbyte |
| 429 |
CADJ STORE theta(:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 430 |
CADJ STORE salt (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 431 |
CADJ STORE uvel (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 432 |
CADJ STORE vvel (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 433 |
#ifdef ALLOW_PASSIVE_TRACER |
| 434 |
CADJ STORE tr1 (:,:,:,bi,bj) = comlev1_bibj, key=ikey, byte=isbyte |
| 435 |
#endif |
| 436 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 437 |
|
| 438 |
#ifdef ALLOW_AIM |
| 439 |
C AIM - atmospheric intermediate model, physics package code. |
| 440 |
C note(jmc) : phiHyd=0 at this point but is not really used in Molteni Physics |
| 441 |
IF ( useAIM ) THEN |
| 442 |
CALL TIMER_START('AIM_DO_ATMOS_PHYS [DYNAMICS]', myThid) |
| 443 |
CALL AIM_DO_ATMOS_PHYSICS( phiHyd, bi, bj, myTime, myThid ) |
| 444 |
CALL TIMER_STOP ('AIM_DO_ATMOS_PHYS [DYNAMICS]', myThid) |
| 445 |
ENDIF |
| 446 |
#endif /* ALLOW_AIM */ |
| 447 |
|
| 448 |
#ifdef ALLOW_MITPHYS |
| 449 |
C MITPHYS - atmospheric model, MIT physics package code. (sb) |
| 450 |
IF ( useMITPHYS ) THEN |
| 451 |
CALL TIMER_START('MITPHYS_DO_ATMOS_PHYS [DYNAMICS]', myThid) |
| 452 |
CALL MITPHYS_DO_ATMOS_PHYSICS( phiHyd, myTime, myThid ) |
| 453 |
CALL TIMER_STOP ('MITPHYS_DO_ATMOS_PHYS [DYNAMICS]', myThid) |
| 454 |
ENDIF |
| 455 |
#endif /* ALLOW_MITPHYS */ |
| 456 |
|
| 457 |
#ifndef DISABLE_MULTIDIM_ADVECTION |
| 458 |
C-- Some advection schemes are better calculated using a multi-dimensional |
| 459 |
C method in the absence of any other terms and, if used, is done here. |
| 460 |
C |
| 461 |
C The CPP flag DISABLE_MULTIDIM_ADVECTION is currently unset in GAD_OPTIONS.h |
| 462 |
C The default is to use multi-dimensinal advection for non-linear advection |
| 463 |
C schemes. However, for the sake of efficiency of the adjoint it is necessary |
| 464 |
C to be able to exclude this scheme to avoid excessive storage and |
| 465 |
C recomputation. It *is* differentiable, if you need it. |
| 466 |
C Edit GAD_OPTIONS.h and #define DISABLE_MULTIDIM_ADVECTION to |
| 467 |
C disable this section of code. |
| 468 |
IF (multiDimAdvection) THEN |
| 469 |
IF (tempStepping .AND. |
| 470 |
& tempAdvScheme.NE.ENUM_CENTERED_2ND .AND. |
| 471 |
& tempAdvScheme.NE.ENUM_UPWIND_3RD .AND. |
| 472 |
& tempAdvScheme.NE.ENUM_CENTERED_4TH ) THEN |
| 473 |
CALL GAD_ADVECTION(bi,bj,tempAdvScheme,GAD_TEMPERATURE, |
| 474 |
U theta,gT, |
| 475 |
I myTime,myIter,myThid) |
| 476 |
ENDIF |
| 477 |
IF (saltStepping .AND. |
| 478 |
& saltAdvScheme.NE.ENUM_CENTERED_2ND .AND. |
| 479 |
& saltAdvScheme.NE.ENUM_UPWIND_3RD .AND. |
| 480 |
& saltAdvScheme.NE.ENUM_CENTERED_4TH ) THEN |
| 481 |
CALL GAD_ADVECTION(bi,bj,saltAdvScheme,GAD_SALINITY, |
| 482 |
U salt,gS, |
| 483 |
I myTime,myIter,myThid) |
| 484 |
ENDIF |
| 485 |
ENDIF |
| 486 |
#endif /* DISABLE_MULTIDIM_ADVECTION */ |
| 487 |
|
| 488 |
C-- Start of thermodynamics loop |
| 489 |
DO k=Nr,1,-1 |
| 490 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 491 |
C? Patrick Is this formula correct? |
| 492 |
cph Yes, but I rewrote it. |
| 493 |
cph Also, the KappaR? need the index and subscript k! |
| 494 |
kkey = (ikey-1)*Nr + k |
| 495 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 496 |
|
| 497 |
C-- km1 Points to level above k (=k-1) |
| 498 |
C-- kup Cycles through 1,2 to point to layer above |
| 499 |
C-- kDown Cycles through 2,1 to point to current layer |
| 500 |
|
| 501 |
km1 = MAX(1,k-1) |
| 502 |
kup = 1+MOD(k+1,2) |
| 503 |
kDown= 1+MOD(k,2) |
| 504 |
|
| 505 |
iMin = 1-OLx |
| 506 |
iMax = sNx+OLx |
| 507 |
jMin = 1-OLy |
| 508 |
jMax = sNy+OLy |
| 509 |
|
| 510 |
C-- Get temporary terms used by tendency routines |
| 511 |
CALL CALC_COMMON_FACTORS ( |
| 512 |
I bi,bj,iMin,iMax,jMin,jMax,k, |
| 513 |
O xA,yA,uTrans,vTrans,rTrans,maskUp, |
| 514 |
I myThid) |
| 515 |
|
| 516 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 517 |
CADJ STORE KappaRT(:,:,k) = comlev1_bibj_k, key=kkey, byte=isbyte |
| 518 |
CADJ STORE KappaRS(:,:,k) = comlev1_bibj_k, key=kkey, byte=isbyte |
| 519 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 520 |
|
| 521 |
#ifdef INCLUDE_CALC_DIFFUSIVITY_CALL |
| 522 |
C-- Calculate the total vertical diffusivity |
| 523 |
CALL CALC_DIFFUSIVITY( |
| 524 |
I bi,bj,iMin,iMax,jMin,jMax,k, |
| 525 |
I maskUp, |
| 526 |
O KappaRT,KappaRS, |
| 527 |
I myThid) |
| 528 |
#endif |
| 529 |
|
| 530 |
iMin = 1-OLx+2 |
| 531 |
iMax = sNx+OLx-1 |
| 532 |
jMin = 1-OLy+2 |
| 533 |
jMax = sNy+OLy-1 |
| 534 |
|
| 535 |
C-- Calculate active tracer tendencies (gT,gS,...) |
| 536 |
C and step forward storing result in gTnm1, gSnm1, etc. |
| 537 |
IF ( tempStepping ) THEN |
| 538 |
CALL CALC_GT( |
| 539 |
I bi,bj,iMin,iMax,jMin,jMax, k,km1,kup,kDown, |
| 540 |
I xA,yA,uTrans,vTrans,rTrans,maskUp, |
| 541 |
I KappaRT, |
| 542 |
U fVerT, |
| 543 |
I myTime,myIter,myThid) |
| 544 |
CALL TIMESTEP_TRACER( |
| 545 |
I bi,bj,iMin,iMax,jMin,jMax,k,tempAdvScheme, |
| 546 |
I theta, gT, |
| 547 |
I myIter, myThid) |
| 548 |
ENDIF |
| 549 |
IF ( saltStepping ) THEN |
| 550 |
CALL CALC_GS( |
| 551 |
I bi,bj,iMin,iMax,jMin,jMax, k,km1,kup,kDown, |
| 552 |
I xA,yA,uTrans,vTrans,rTrans,maskUp, |
| 553 |
I KappaRS, |
| 554 |
U fVerS, |
| 555 |
I myTime,myIter,myThid) |
| 556 |
CALL TIMESTEP_TRACER( |
| 557 |
I bi,bj,iMin,iMax,jMin,jMax,k,saltAdvScheme, |
| 558 |
I salt, gS, |
| 559 |
I myIter, myThid) |
| 560 |
ENDIF |
| 561 |
#ifdef ALLOW_PASSIVE_TRACER |
| 562 |
IF ( tr1Stepping ) THEN |
| 563 |
CALL CALC_GTR1( |
| 564 |
I bi,bj,iMin,iMax,jMin,jMax, k,km1,kup,kDown, |
| 565 |
I xA,yA,uTrans,vTrans,rTrans,maskUp, |
| 566 |
I KappaRT, |
| 567 |
U fVerTr1, |
| 568 |
I myTime,myIter,myThid) |
| 569 |
CALL TIMESTEP_TRACER( |
| 570 |
I bi,bj,iMin,iMax,jMin,jMax,k,tracerAdvScheme, |
| 571 |
I Tr1, gTr1, |
| 572 |
I myIter,myThid) |
| 573 |
ENDIF |
| 574 |
#endif |
| 575 |
|
| 576 |
#ifdef ALLOW_OBCS |
| 577 |
C-- Apply open boundary conditions |
| 578 |
IF (useOBCS) THEN |
| 579 |
CALL OBCS_APPLY_TS( bi, bj, k, gT, gS, myThid ) |
| 580 |
END IF |
| 581 |
#endif /* ALLOW_OBCS */ |
| 582 |
|
| 583 |
C-- Freeze water |
| 584 |
IF (allowFreezing) THEN |
| 585 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 586 |
CADJ STORE gT(:,:,k,bi,bj) = comlev1_bibj_k |
| 587 |
CADJ & , key = kkey, byte = isbyte |
| 588 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 589 |
CALL FREEZE( bi, bj, iMin, iMax, jMin, jMax, k, myThid ) |
| 590 |
END IF |
| 591 |
|
| 592 |
C-- end of thermodynamic k loop (Nr:1) |
| 593 |
ENDDO |
| 594 |
|
| 595 |
|
| 596 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 597 |
C? Patrick? What about this one? |
| 598 |
cph Keys iikey and idkey dont seem to be needed |
| 599 |
cph since storing occurs on different tape for each |
| 600 |
cph impldiff call anyways. |
| 601 |
cph Thus, common block comlev1_impl isnt needed either. |
| 602 |
cph Storing below needed in the case useGMREDI. |
| 603 |
iikey = (ikey-1)*maximpl |
| 604 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 605 |
|
| 606 |
C-- Implicit diffusion |
| 607 |
IF (implicitDiffusion) THEN |
| 608 |
|
| 609 |
IF (tempStepping) THEN |
| 610 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 611 |
idkey = iikey + 1 |
| 612 |
CADJ STORE gT(:,:,:,bi,bj) = comlev1_bibj , key=ikey, byte=isbyte |
| 613 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 614 |
CALL IMPLDIFF( |
| 615 |
I bi, bj, iMin, iMax, jMin, jMax, |
| 616 |
I deltaTtracer, KappaRT, recip_HFacC, |
| 617 |
U gT, |
| 618 |
I myThid ) |
| 619 |
ENDIF |
| 620 |
|
| 621 |
IF (saltStepping) THEN |
| 622 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 623 |
idkey = iikey + 2 |
| 624 |
CADJ STORE gS(:,:,:,bi,bj) = comlev1_bibj , key=ikey, byte=isbyte |
| 625 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 626 |
CALL IMPLDIFF( |
| 627 |
I bi, bj, iMin, iMax, jMin, jMax, |
| 628 |
I deltaTtracer, KappaRS, recip_HFacC, |
| 629 |
U gS, |
| 630 |
I myThid ) |
| 631 |
ENDIF |
| 632 |
|
| 633 |
#ifdef ALLOW_PASSIVE_TRACER |
| 634 |
IF (tr1Stepping) THEN |
| 635 |
#ifdef ALLOW_AUTODIFF_TAMC |
| 636 |
CADJ STORE gTr1(:,:,:,bi,bj) = comlev1_bibj , key=ikey, byte=isbyte |
| 637 |
#endif /* ALLOW_AUTODIFF_TAMC */ |
| 638 |
CALL IMPLDIFF( |
| 639 |
I bi, bj, iMin, iMax, jMin, jMax, |
| 640 |
I deltaTtracer, KappaRT, recip_HFacC, |
| 641 |
U gTr1, |
| 642 |
I myThid ) |
| 643 |
ENDIF |
| 644 |
#endif |
| 645 |
|
| 646 |
#ifdef ALLOW_OBCS |
| 647 |
C-- Apply open boundary conditions |
| 648 |
IF (useOBCS) THEN |
| 649 |
DO K=1,Nr |
| 650 |
CALL OBCS_APPLY_TS( bi, bj, k, gT, gS, myThid ) |
| 651 |
ENDDO |
| 652 |
END IF |
| 653 |
#endif /* ALLOW_OBCS */ |
| 654 |
|
| 655 |
C-- End If implicitDiffusion |
| 656 |
ENDIF |
| 657 |
|
| 658 |
Ccs- |
| 659 |
ENDDO |
| 660 |
ENDDO |
| 661 |
|
| 662 |
#ifdef ALLOW_AIM |
| 663 |
IF ( useAIM ) THEN |
| 664 |
CALL AIM_AIM2DYN_EXCHANGES( myTime, myThid ) |
| 665 |
ENDIF |
| 666 |
_EXCH_XYZ_R8(gT,myThid) |
| 667 |
_EXCH_XYZ_R8(gS,myThid) |
| 668 |
#else |
| 669 |
C modif omp |
| 670 |
C previous version |
| 671 |
C IF (staggerTimeStep.AND.useCubedSphereExchange) THEN |
| 672 |
C changed version --> exchange after the timestep |
| 673 |
IF (staggerTimeStep) THEN |
| 674 |
_EXCH_XYZ_R8(gT,myThid) |
| 675 |
_EXCH_XYZ_R8(gS,myThid) |
| 676 |
ENDIF |
| 677 |
#endif /* ALLOW_AIM */ |
| 678 |
|
| 679 |
RETURN |
| 680 |
END |
Parent Directory
| 
Revision Log
| 
Revision Graph