| 3 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 4 |
============================== |
============================== |
| 5 |
|
|
| 6 |
|
checkpoint47 |
| 7 |
|
differentiable version of checkpoint46n_post |
| 8 |
|
o bug fix in quasihydrostaticterms |
| 9 |
|
o * "clean" adjoint code (in terms of extensive recomputations) |
| 10 |
|
can now be obtained for all GMREDI options (i.e. for |
| 11 |
|
- GM_VISBECK_VARIABLE_K |
| 12 |
|
- GM_NON_UNITY_DIAGONAL |
| 13 |
|
- GM_EXTRA_DIAGONAL |
| 14 |
|
- GM_BOLUS_ADVEC ) |
| 15 |
|
* However, wrong gradient check problem remains unsolved. |
| 16 |
|
* New CPP options have been introduced for different |
| 17 |
|
tapering schemes |
| 18 |
|
o external_fields_load now part of differentiation list |
| 19 |
|
o pressure needs multiple storing; |
| 20 |
|
would be nice to have store_pressure at beginning or |
| 21 |
|
end of forward_step, e.g. by having phiHyd global (5-dim.) |
| 22 |
|
(NB: pressure is needed for certain cases in find_rho, |
| 23 |
|
which is also invoked through convective_adjustment). |
| 24 |
|
o recomputations in find_rho for cases |
| 25 |
|
'JMD95'/'UNESCO' or 'MDJWF' are OK. |
| 26 |
|
o #define ATMOSPHERIC_LOADING should be differentiable |
| 27 |
|
o ini_forcing shifted to begining of initialise_varia |
| 28 |
|
o Incorporating QNVS line search routines into MITgcm |
| 29 |
|
(this is separate code, and therefore not under pkg) |
| 30 |
|
lsopt/, optim/ |
| 31 |
|
o Updated verification/carbon/ |
| 32 |
|
- #define GM_VISBECK_VARIABLE_K |
| 33 |
|
- GM_taper_scheme = 'ldd97' |
| 34 |
|
|
| 35 |
|
checkpoint46n_post |
| 36 |
|
Merging from release1_p8: |
| 37 |
|
o verification/natl_box: |
| 38 |
|
updating new external_fields_load routine |
| 39 |
|
o New package: pkg/seaice |
| 40 |
|
Sea ice model by D. Menemenlis (JPL) and Jinlun Zhang (Seattle). |
| 41 |
|
The sea-ice code is based on Hibler (1979-1980). |
| 42 |
|
Two sea-ice dynamic solvers, ADI and LSR, are included. |
| 43 |
|
In addition to computing prognostic sea-ice variables and diagnosing |
| 44 |
|
the forcing/external data fields that drive the ocean model, |
| 45 |
|
SEAICE_MODEL also sets theta to the freezing point under sea-ice. |
| 46 |
|
The implied surface heat flux is then stored in variable |
| 47 |
|
surfaceTendencyTice, which is needed by KPP package (kpp_calc.F and |
| 48 |
|
kpp_transport_t.F) to diagnose surface buoyancy fluxes and for the |
| 49 |
|
non-local transport term. Because this call precedes model |
| 50 |
|
thermodynamics, temperature under sea-ice may not be "exactly" at |
| 51 |
|
the freezing point by the time theta is dumped or time-averaged. |
| 52 |
|
N.B.: fairly large differences in accuracy occur across |
| 53 |
|
different platforms/compilers; comparison between |
| 54 |
|
g77 (Linux) and F77 (SGI Origin 2000/MIPSpro) |
| 55 |
|
generated output gives: |
| 56 |
|
T S U V |
| 57 |
|
C D M c m s m s m s m s |
| 58 |
|
n p a R g m m e . m m e . m m e . m m e . |
| 59 |
|
f n k u 2 i a a d i a a d i a a d i a a d |
| 60 |
|
g d e n d n x n . n x n . n x n . n x n . |
| 61 |
|
|
| 62 |
|
Y Y Y Y 5 5 7 7 7 8 10 9 6 6 6 6 7 5 7 5 7 FAIL lab_sea |
| 63 |
|
o GAD: |
| 64 |
|
- generated new common blocks to account for call of |
| 65 |
|
same gad routines with differing traceridentities |
| 66 |
|
(needed to modify tracerIdentity indices in GAD.h) |
| 67 |
|
- generated separate common blocks for case useCubedSphereExchange |
| 68 |
|
(Department of Futurology) |
| 69 |
|
- parameter lists to gmredi_?transport: added tracerIdentity |
| 70 |
|
- added new key indices to tamc.h |
| 71 |
|
o external_fields_load: |
| 72 |
|
- added this routine to TAF list |
| 73 |
|
- needed to make some common blocks global and additional storing |
| 74 |
|
along the same lines as exf package (checkpoint_lev?_directives.h) |
| 75 |
|
o exf: |
| 76 |
|
updated external forcing package |
| 77 |
|
- bug fixes carried over from ecco-branch |
| 78 |
|
(missing OBCS_OPTIONS.h in two routines) |
| 79 |
|
- enable easy to use "no forcing". |
| 80 |
|
- added exf I/O for atmospheric loading |
| 81 |
|
- added exf I/O for runoff data |
| 82 |
|
- transfered scaling between exf <-> MITgcm to exf namelist |
| 83 |
|
o Adding new verification experiment global_with_exf. |
| 84 |
|
This verification experiment is almost identical to the |
| 85 |
|
global_ocean.90x40x15 experiment, except that it uses |
| 86 |
|
the exf (and cal) package instead of the |
| 87 |
|
external_fields_load routine (cf README file). |
| 88 |
|
|
| 89 |
|
checkpoint46m_post |
| 90 |
|
o modified testscript to return -- when there are no lines of "%MON" output |
| 91 |
|
o added new flag "quasiHydrostatic" for QH mode |
| 92 |
|
and new routine quasihydrostaticterms() |
| 93 |
|
o added 2*Omega*Cos(phi)*W to u equations (non-hydrostatic Coriolis term) |
| 94 |
|
New routine: mom_u_coriolis_nh() |
| 95 |
|
o fixed sign of non-hydrostatic metric terms in u and v equations. |
| 96 |
|
Depend on correct sign of gravitySign. |
| 97 |
|
o added work.pc* to .cvsignore in aim directories using ifc |
| 98 |
|
o introduced flag for controllin non-hydrostatic metric terms |
| 99 |
|
o removed gravitySign from PARM04 |
| 100 |
|
o switched to g77/ifc from pgf77 in aim experiments |
| 101 |
|
|
| 102 |
|
checkpoint46l_post |
| 103 |
|
o made convective adjustment work with pressure coordinates: |
| 104 |
|
- changed the direction of k-loop in convective_adjustment.F for the |
| 105 |
|
case of pressure coordinates (OCEANICP,ATMOSPHERIC buoyancyRelation) |
| 106 |
|
- adjusted the reference pressure k-index in convective_adjustment.F |
| 107 |
|
- adjusted the convection condition in convect.F (in analogy to |
| 108 |
|
calc_ivdc.F) |
| 109 |
|
- convective_adjustment no longer computes anything on the halos |
| 110 |
|
- removed the warnings about negative salinity from find_rho.F and |
| 111 |
|
find_alpha.F; instead the new routine look_for_neg_salinity, called |
| 112 |
|
at the beginning of find_rho, find_alpha, and find_beta, does a |
| 113 |
|
check of the entire slice, if CPP-option |
| 114 |
|
CHECK_SALINITY_FOR_NEGATIVE_VALUES is defined |
| 115 |
|
checkpoint46l_pre |
| 116 |
|
|
| 117 |
|
checkpoint46k_post |
| 118 |
|
o fixed the verification/global_ocean.90x40x15 experiment: |
| 119 |
|
- new bathymetry (the world according to A., JMC, and M.) |
| 120 |
|
- new initial fields and forcing fields (*.bin files) |
| 121 |
|
- new POLY3.COEFFS (for the next release one should switch to a full |
| 122 |
|
equation of state: JMD95P or MDJWF) |
| 123 |
|
- fixed several errors and redundancies in the data file |
| 124 |
|
- experiment uses looped cells |
| 125 |
|
- added matlab directory with diagnostic scripts for plotting of output |
| 126 |
|
|
| 127 |
|
o S/R aim_initialise.F replace S/R aim_init from file aim_do_inphys.F: |
| 128 |
|
- read AIM physics parameters from a file (data.aimphys) |
| 129 |
|
- set defaults values = F.Molteni paper (Clim.Dyn., 2002) |
| 130 |
|
|
| 131 |
|
checkpoint46j_post |
| 132 |
|
o split calc_exact_eta in 2 S/R : integr_continuity & update_etaH |
| 133 |
|
o move wVel computation at the end of the time step, in S/R integr_continuity |
| 134 |
|
o create specific S/R to exchange T,S before DYNAMICS (for stagger time step) |
| 135 |
|
o update timeave pkg for wVel diagnostic. |
| 136 |
|
checkpoint46j_pre |
| 137 |
|
|
| 138 |
|
checkpoint46i_post |
| 139 |
|
o Clean up AIM package (and keep the results unchanged): |
| 140 |
|
include CPP_OPTION and use IMPLICT NONE in all routines ; |
| 141 |
|
declare all the variables _RL ; use _d 0 for all numerical constants. |
| 142 |
|
use ifdef ALLOW_AIM everywhere. And now AIM can be used with g77 ! |
| 143 |
|
|
| 144 |
|
checkpoint46h_post |
| 145 |
|
o cleaned up the use of rhoNil and rhoConst. |
| 146 |
|
- rhoNil should only appear in the LINEAR equation of state, everywhere |
| 147 |
|
else rhoNil is replaced by rhoConst, e.g. find_rho computes rho-rhoConst |
| 148 |
|
and the dynamical equations are all divided by rhoConst |
| 149 |
|
o introduced new parameter rhoConstFresh, a reference density of fresh |
| 150 |
|
water, to remove the fresh water flux's dependence on rhoNil. The default |
| 151 |
|
value is 999.8 kg/m^3 |
| 152 |
|
o cleanup up external_forcing.F and external_forcing_surf.F |
| 153 |
|
- can now be used by both OCEANIC and OCEANICP |
| 154 |
|
checkpoint46h_pre |
| 155 |
|
|
| 156 |
|
Added code to convert surface volume flux (fresh water) into |
| 157 |
|
a mass flux when using P coordinates in the ocean (OCEANICP). |
| 158 |
|
Note: It assumes you have set rho0=rhoConst=density of fresh water. |
| 159 |
|
|
| 160 |
|
checkpoint46g_post |
| 161 |
|
o Include a new diagnostic variable phiHydLow for the ocean model |
| 162 |
|
- in z-coordinates, it is the bottom pressure anomaly |
| 163 |
|
- in p-coordinates, it is the sea surface elevation |
| 164 |
|
- in both cases, these variable have global drift, reflecting the mass |
| 165 |
|
drift in z-coordinates and the volume drift in p-coordinates |
| 166 |
|
- included time averaging for phiHydLow, be aware of the drift! |
| 167 |
|
o depth-dependent computation of Bo_surf for pressure coordinates |
| 168 |
|
in the ocean (buoyancyRelation='OCEANICP') |
| 169 |
|
- requires a new routine (FIND_RHO_SCALAR) to compute density with only |
| 170 |
|
Theta, Salinity, and Pressure in the parameter list. This routine is |
| 171 |
|
presently contained in find_rho.F. This routine does not give the |
| 172 |
|
correct density for 'POLY3', which would be a z-dependent reference |
| 173 |
|
density. |
| 174 |
|
o cleaned up find_rho |
| 175 |
|
- removed obsolete 'eqn' from the parameter list. |
| 176 |
|
o added two new verification experiments: gop and goz |
| 177 |
|
(4x4 degree global ocean, 15 layers in pressure and height coordinates) |
| 178 |
|
checkpoint46g_pre |
| 179 |
|
|
| 180 |
|
checkpoint46f_post |
| 181 |
|
o Enable tangent linear (forward mode) gradient checks: |
| 182 |
|
- extended active file handling to g_... files |
| 183 |
|
- added TANGENT_SIMULATION to theSimulationMode |
| 184 |
|
- extended grdchk package accordingly |
| 185 |
|
o added tangent linear setup in verification/carbon/code_ftl |
| 186 |
|
o added adjoint setup in verification/front_relax/code_ad |
| 187 |
|
|
| 188 |
|
checkpoint46e_post |
| 189 |
|
o Added new equation of state -> MDJWF |
| 190 |
|
- EOS of McDougall et al., 2002, JAOT, submitted |
| 191 |
|
- caveat: the equation of state is only valid for a smaller (more |
| 192 |
|
realistic?) range of values than JMD95P/Z and UNESCO |
| 193 |
|
- added masks to the calculation of pressure in store_pressure |
| 194 |
|
- added more check values for density in check_eos (ini_eos.F), some of |
| 195 |
|
the old check values are out of the range of the MDJWF-eos, so don't |
| 196 |
|
expect perfect matches for those |
| 197 |
|
checkpoint46e_pre |
| 198 |
|
|
| 199 |
|
checkpoint46d_post |
| 200 |
|
o fixed store_pressure to work with both buoyancy relation = 'OCEANIC' and |
| 201 |
|
'OCEANICP', also initialised field pressure correctly in ini_eos in the |
| 202 |
|
case of pressure coordinates. eosType='JMD95Z' in |
| 203 |
|
combination with buoyancyRelation='OCEANICP' now causes an error. |
| 204 |
|
o Changed p = pressure(i,j,k,bi,bj) to p = pressure(i,j,kRef,bi,bj) |
| 205 |
|
in find_alpha/beta. |
| 206 |
|
checkpoint46d_pre |
| 207 |
|
|
| 208 |
|
o Changed p = pressure(i,j,k,bi,bj) to p = pressure(i,j,kRef,bi,bj) |
| 209 |
|
so that JMD95Z and JMD95P give approptiate static stability in find_rho. |
| 210 |
|
|
| 211 |
|
checkpoint46c_post |
| 212 |
|
o Changes necessary for ocean in p-coordinates |
| 213 |
|
- Added new buoyancy relation = 'OCEANICP' |
| 214 |
|
- Added new parameters = gravitySign (this used to be contained inside |
| 215 |
|
the factor dRdZ which I added when we first switched to R coordinates). |
| 216 |
|
X GM/Redi is not compatible (yet) |
| 217 |
|
X bottom drag and no-slip need to be debugged. |
| 218 |
|
checkpoint46c_pre |
| 219 |
|
|
| 220 |
|
checkpoint46b_post |
| 221 |
|
o Added new equation of state -> JMD95Z and JMD95P |
| 222 |
|
- EOS of Jackett and McDougall, 1995, JAOT |
| 223 |
|
- moved all EOS parameters into EOS.h |
| 224 |
|
- new routines ini_eos.F, store_pressure.F |
| 225 |
|
o Added UNESCO EOS, but not recommended because it requires |
| 226 |
|
in-situ temperature (see JMD95) |
| 227 |
|
o Modified formatting for knudsen2.f in utils/knudsen2 and added |
| 228 |
|
unesco.f to be used with POLY3 |
| 229 |
|
checkpoint46b_pre |
| 230 |
|
|
| 231 |
|
checkpoint46a_post |
| 232 |
|
o Added (atmospheric) pressure loading |
| 233 |
|
- new field in FFIELDS.h, etc... |
| 234 |
|
- new cpp flag ATMOSPHERIC_LOADING |
| 235 |
|
o Changed hFacC to _hFacC in calc_phi_hyd.F |
| 236 |
|
o Added SHORTWAVE_HEATING to some files for consistency |
| 237 |
|
checkpoint46a_pre |
| 238 |
|
|
| 239 |
|
checkpoint46 |
| 240 |
|
further merges from release1_p5 |
| 241 |
|
o new ctrl package |
| 242 |
|
- adopted from ECCO environment to enable optimization |
| 243 |
|
- added Eliassen Palm fluxes to controls |
| 244 |
|
o added Eliassen Palm flux controls to gradient check package |
| 245 |
|
o cost package |
| 246 |
|
- Compatible with new ctrl package |
| 247 |
|
- added Eliassen Palm cost hooks |
| 248 |
|
- modif's of existing cost functions |
| 249 |
|
o Adjoint-related bug fixes in kpp: |
| 250 |
|
- kpp_calc: sore of kpphbl avoids recomputation/call to S/R kppmix |
| 251 |
|
- kpp_routines: store of Rib avoids partial recomputation bug of TAF. |
| 252 |
|
o autodiff package |
| 253 |
|
- flow directives: |
| 254 |
|
+ new for S/R do_the_model_io |
| 255 |
|
+ enhanced for S/R checkpoint |
| 256 |
|
+ new gmredi.flow |
| 257 |
|
- Introduced CPP option ALLOW_AUTODIFF_MONITOR to |
| 258 |
|
disable adcommon.h/g_common.h by default. |
| 259 |
|
- fixed problem for adjoint of global_max |
| 260 |
|
o modified verification/carbon/ |
| 261 |
|
o New setup for adjoint of global_ocean |
| 262 |
|
o added hooks for Eliassen Palm fluxes (dynamics.F, ini_fields.F) |
| 263 |
|
o cleaned TAF keys iikey, idkey (dynamics.F, thermodynamics.F) |
| 264 |
|
o incorporated hooks for sponge layer code (external_forcing.F) |
| 265 |
|
|
| 266 |
|
checkpoint45d_post |
| 267 |
|
o import 2 fixes from release1_p5 (CG2D.h & packages_readparms.F) |
| 268 |
|
o use recip_dx*,recip_dy* instead of /dx*,/dy* in orlanski_E,W,N,S |
| 269 |
|
(affects plume_on_slope "cg2d" output) |
| 270 |
|
o add new parameter deltaTfreesurf for free-surface time stepping |
| 271 |
|
o exchange gT,gS when using staggerTimeStep & GM & Oly<4 |
| 272 |
|
o NONLIN_FRSURF: rescale gNm1 to get a better conservation with A-B |
| 273 |
|
o change phiMin in exp1 (to agree with documentation) |
| 274 |
|
o add new flags: |
| 275 |
|
- T,S forcing inside/outside Adams-Bashforth |
| 276 |
|
- turn off temp,salt advection and/or forcing |
| 277 |
|
- for each tracer, define internal flag for multiDimAdvection & A-B |
| 278 |
|
o monitor: change definition of KE to check conservation of Energy |
| 279 |
|
|
| 280 |
|
checkpoint45c_post |
| 281 |
|
o Removed f90 comments ("!") |
| 282 |
|
o Modified initialisations to enable exf package for MITgcm |
| 283 |
|
(without invoking ECCO_PACKAGE). |
| 284 |
|
o modifications to gradient check package (Martin Losch) |
| 285 |
|
- enable centered differences |
| 286 |
|
- modified format of standard output |
| 287 |
|
|
| 288 |
|
checkpoint45b_post |
| 289 |
|
o cleaned exchange calls in cg2d |
| 290 |
|
o Included CPP option ALLOW_AUTODIFF_MONITOR in forward_step |
| 291 |
|
o included CPP option SINGLE_LAYER_MODE in thermodynamics |
| 292 |
|
to configure barotropic setup (Martin Losch) |
| 293 |
|
o moved some initialisations within bi,bj loop in dynamics |
| 294 |
|
(TAF flow dependency) |
| 295 |
|
o in initialise_varia, moved call packages_init_variables |
| 296 |
|
before convective_adjustment_init |
| 297 |
|
(corresponding adjustment of TAF tape initialisation). |
| 298 |
|
o in calc_diffusivity/calc_viscosity extended domain to full overlap. |
| 299 |
|
|
| 300 |
|
checkpoint45a_post |
| 301 |
|
o compute Rho over the whole domain (extend iMin,jMin indices) |
| 302 |
|
o specific exchange for X-slice domain (case Ny=1) |
| 303 |
|
o argument futureIter of S/R obcs_calc missing in exp4 & internal_wave /code |
| 304 |
|
o solve_for_pressure : set cg2d_x to zero where OBCS are applied |
| 305 |
|
|
| 306 |
|
checkpoint44h_post = checkpoint45 |
| 307 |
|
o Fix to restore differentiability: |
| 308 |
|
place at which packages_init_variables is called in |
| 309 |
|
initialise_varia is crucial for reverse control flow. |
| 310 |
|
o modifications in GMRedi to enable stable adjoint |
| 311 |
|
(corresponding modif's in thermodynamics, dynamics) |
| 312 |
|
o added missing hook for sponge layer code in external_forcing |
| 313 |
|
o modified test cost functions |
| 314 |
|
o storing in gad_advection (plus removal of write(0,*)! |
| 315 |
|
o missing headers TR1.h added in convective_adjustment routines |
| 316 |
|
o errorMessageUnit set non-equal zero in eeboot |
| 317 |
|
(conflict on some platforms, e.g. IBM SP3) |
| 318 |
|
o modified carbon verif. |
| 319 |
|
|
| 320 |
|
checkpoint44h_pre |
| 321 |
|
o vertical grid option: allow to put the Interface at the middle between |
| 322 |
|
2 cell-centers ; replace delZ in pkg/kpp by drF. |
| 323 |
|
o GM Advective form: Tracers are advected using the residual transport (= |
| 324 |
|
Euler+GM-bolus); set param GM_AdvSeparate=T to return to previous form. |
| 325 |
|
|
| 326 |
|
checkpoint44g_post |
| 327 |
|
o fix surface correction term in multiDimAdvection (affects plume_on_slope |
| 328 |
|
results) ; ifdef missing in thermodynamics.F |
| 329 |
|
|
| 330 |
|
checkpoint44f_pre,checkpoint44f_post |
| 331 |
|
o added PTRACERS package |
| 332 |
|
This allows an arbitrary number of passive tracers to be integrated |
| 333 |
|
forward simultaneously with the dynamicaly model. |
| 334 |
|
+ Implemented so far: |
| 335 |
|
- basic forward algorithm (time-stepping, advection, diffusion, convection) |
| 336 |
|
- I/O and checkpointing |
| 337 |
|
- GM/Redi *but* using the GM/Redi coefficient of Salt |
| 338 |
|
+ Not implemented so far: |
| 339 |
|
- KPP |
| 340 |
|
- OBCS |
| 341 |
|
+ No specific example supplied (yet) but global_ocean.90x40x15 has the |
| 342 |
|
necessary data.ptracer file. Simply use -enable=ptracers and uncomment |
| 343 |
|
line in data.pkg. |
| 344 |
|
+ This package is disabled by default since it increases storage. |
| 345 |
|
|
| 346 |
|
|
| 347 |
|
o testscript: added -cleanup option |
| 348 |
|
o Shap_filt: a) new shap_filt S/R to use no-slip BC with S2 filter |
| 349 |
|
b) enable to filter 2D fields. |
| 350 |
|
|
| 351 |
|
checkpoint44e_pre,post |
| 352 |
|
o re-wrote convective adjustment to use pre-calculated weights |
| 353 |
|
Affects: model/src/convective_adjustment.F, convective_adjustment_ini.F |
| 354 |
|
Adds: model/src/convective_weights.F, convectively_mixtracer.F |
| 355 |
|
Original calls to convect() still in place. |
| 356 |
|
|
| 357 |
|
chkpt44d_pre,post |
| 358 |
|
o added missing EXCLUDE_MONITOR flags |
| 359 |
|
o changed "e" to "_d" in gmredi_slope_limit, gmredi_slope_psi |
| 360 |
|
(incompatible typ in MIN/MAX expressions caused problems |
| 361 |
|
on IBM SP3) |
| 362 |
|
o in genmake added variable MAKEDEPEND |
| 363 |
|
plus resetting for case SunOS |
| 364 |
|
o added timer_stats.c routine for IBM SP3 |
| 365 |
|
o removed variables in dynamics |
| 366 |
|
!!! I forgot to create tag chkpt44d_pre !!! |
| 367 |
|
!!! But chkpt44d_pre = chkpt44c_post !!! |
| 368 |
|
|
| 369 |
o real fresh water flux implemented with non-linear free-surface. |
o real fresh water flux implemented with non-linear free-surface. |
| 370 |
|
|
| 371 |
chkpt44c_pre,post |
chkpt44c_pre,post |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch