| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o pkg/seaice adjoint of LSR: |
| 8 |
|
- substantially rework STORE dirs. |
| 9 |
|
- add new flag SEAICE_LSR_ADJOINT_ITER and parameter SOLV_MAX_FIXED |
| 10 |
|
this flag will do SOLV_MAX_ITERS but only store SOLV_MAX_FIXED fields |
| 11 |
|
and only perform SOLV_MAX_FIXED adjoint iters. |
| 12 |
|
If flag SEAICE_LSR_ADJOINT_ITER is undef-ined, the store in the iteration |
| 13 |
|
always goes to same record, meaning that the last forward field will be |
| 14 |
|
the one used in the adjoint - an approximation. |
| 15 |
|
- now produces non-zero LSR-related gradients, and acceptable gradients for |
| 16 |
|
An's zonal (x-z) test case (to be checked in) |
| 17 |
|
- lab_sea: produces acceptable gradients for lab_sea with forward/backward LSR |
| 18 |
|
(but these are screwed up by the thermodynamics in the first place); |
| 19 |
|
reduce from 4 to 3 timesteps to avoid t.d. issue as temporary solution; |
| 20 |
|
output_adm.txt also reflects changes to seaice_lsr |
| 21 |
|
- global_ocean.cs32x15: |
| 22 |
|
* the "default" input_ad.seaice has been re-defined to always exclude dynamics |
| 23 |
|
(output_adm.seaice.txt was updated) |
| 24 |
|
* new input_ad.seaice_dynmix tests a forward LSR / backward free-drift |
| 25 |
|
(output_adm.seaice_dynmix.txt) |
| 26 |
|
o test-exp. exp4+stevens: |
| 27 |
|
- switch off pTracers and nonHydrostatic (restart problems with NH); update output |
| 28 |
|
o pkg/seaice (lsr solver): |
| 29 |
|
- put coriolis term into FORCEX,Y (as in seaice_evp.F; previously in rhsU,V only); |
| 30 |
|
update output (change truncation) of FWD test exp lab_sea & global_ocean.cs32x15 |
| 31 |
|
- fix SEAICEuseFREEDRIFT for new OBCS version (ndef OBCS_UVICE_OLD) |
| 32 |
|
- new parameter "LSR_mixIniGuess" allows (if=2,4) to mix free-drift solution |
| 33 |
|
into initial guess for linearised solver; use weight ~ (1/err)^2,4. |
| 34 |
|
- compute (with new S/R) and print initial (+ free-driff sol. if available) |
| 35 |
|
and final residual of linearised system. |
| 36 |
|
o pkg/pp81: add diagnostics for vertical eddy viscosity and diffusivity |
| 37 |
|
(requires new subroutines pp81_init_fixed.F/pp81_diagnostics_init.F) |
| 38 |
|
o pkg/obcs: Stevens BCs (still only for east and west) |
| 39 |
|
- fix Stevens BCs by using intermediate velocities of previous |
| 40 |
|
time step as "baroclinic" contribution rather velocity of one grid cell |
| 41 |
|
inside the boundaries; more consistent but makes the entire treatment |
| 42 |
|
more complicated, requires new pickup files and two new S/Rs: |
| 43 |
|
obcs_copy_uv_n and obcs_save_uv_n |
| 44 |
|
keep old version with flag OBCS_STEVENS_USE_INTERIOR_VELOCITY |
| 45 |
|
(undefined by default) |
| 46 |
|
- silently move computation of some grid factors into separate loops to |
| 47 |
|
save some recomputations |
| 48 |
|
- add a hack for non-linear free surface |
| 49 |
|
- add a verification experiment to exp4 that tests this code |
| 50 |
|
o make gluemnc/xplodemnc work with (moron) non-GNU shell tools |
| 51 |
|
|
| 52 |
|
checkpoint63d (2011/10/25) |
| 53 |
|
o pkg/obcs + seaice-dynamics: |
| 54 |
|
- new implementation calls OBCS_APPLY_UVICE (in seaice_dynsolver.F) before |
| 55 |
|
solving for uIce,vIce and prevent SEAICE_LSR or SEAICE_EVP to modify OB |
| 56 |
|
values (consistent with an implicit method). |
| 57 |
|
- add option OBCS_UVICE_OLD (in OBCS_OPTIONS.h, defined by default) to |
| 58 |
|
return to older implementation (same results as before); note that none of |
| 59 |
|
the "experimental" OBCS_SEAICE options are implemented in the new version. |
| 60 |
|
- switch test exp. seaice_obcs to use new version (update output). |
| 61 |
|
o pkg/seaice: |
| 62 |
|
- add few stop (B-grid & CS-exch, B-grid & obcs, AUTODIFF & SEAICE_maskRHS) |
| 63 |
|
- add parenthesis in Water-Ice drag calculations (evp & lsr); |
| 64 |
|
- in seaice_lsr.F, simplify and fix AA3 setting in case sNx or sNy=1; |
| 65 |
|
and fix case SEAICE_waterDrag=0 (but unstable if no or little seaice) |
| 66 |
|
- update output of test exp lab_sea that uses evp or lsr (fwd+ad) |
| 67 |
|
and global_ocean.cs32x15 (icedyn & seaice fwd+ad). |
| 68 |
|
o pkg/layers: |
| 69 |
|
- Include potential density as new coordinate (Thanks to David Munday) |
| 70 |
|
o pkg/dic: |
| 71 |
|
- Remove subroutine CALC_PCO2_APPROX_CO3 from carbon_chem.F and add |
| 72 |
|
carbonate computation/output to CALC_PCO2_APPROX |
| 73 |
|
o pkg/generic_advdiff |
| 74 |
|
- make TAF generate code that vectorizes: |
| 75 |
|
gad_dst3fl_adv_*.F, gad_u3c4_impl_r.F |
| 76 |
|
o model/src: |
| 77 |
|
- improve vectorization also for adjoint code of |
| 78 |
|
solve_tridiagonal and solve_pentadiagonal. As this requires up to |
| 79 |
|
5 new 3D-fields, this code is only turned on for the combination of |
| 80 |
|
ALLOW_AUTODIFF_TAMC and TARGET_NEC_SX. Also remove error message |
| 81 |
|
for singular matrix check (only for TARGET_NEC_SX). |
| 82 |
|
- remove hard-wiring of CPP-flag ALLOW_SOLVERS_KLOOPINSIDE to |
| 83 |
|
ALLOW_AUTODIFF. This flag now needs to be set outside of |
| 84 |
|
solve_tri/pentadiagonal.F |
| 85 |
|
o pkg/exf: |
| 86 |
|
- fix previous modif in exf_readparms.F (revision 1.66: treatment of |
| 87 |
|
optional namelist EXF_NML_OBCS). |
| 88 |
|
- add many debug-print (to trace down where pkg/cal error msg comes from) |
| 89 |
|
o pkg/mom_vecinv/mom_vi_hdissip.F |
| 90 |
|
- rearrange loops so that vectorization is improved and in particular |
| 91 |
|
the adjoint routine vectorizes. The code is a little longer now, but I |
| 92 |
|
think easier to read. |
| 93 |
|
Note: CPP-flag MOM_VI_ORIGINAL_VISCA4 is not tested. |
| 94 |
|
o pkg/seaice: |
| 95 |
|
- allow to specify initial seaice velocity from files (uIceFile,vIceFile) |
| 96 |
|
o pkg/exch2: |
| 97 |
|
- alternative code (without tile x tile matrix) for cumulated (global) sum; |
| 98 |
|
used as default; keep using matrix version in adjustment.cs-32x32x1. |
| 99 |
|
o tools/build_options: |
| 100 |
|
- after merging some MPI and/or multi-threads specific optfiles so that |
| 101 |
|
the modified default optfile support both built (using genmake2 -mpi |
| 102 |
|
for the MPI built, using genmake2 -omp for OpenMP built), |
| 103 |
|
move now obsolete optfiles to tools/build_options/unsupported/ |
| 104 |
|
(add copy of email sent to support list in doc/optfiles_changes.txt) |
| 105 |
|
|
| 106 |
|
checkpoint63c (2011/09/28) |
| 107 |
|
o pkg/ebm: |
| 108 |
|
- fix ebm_zonalmean.F for several procs in Ydir (nPy > 1); update output of |
| 109 |
|
test exp. global_ocean_ebm (affects machine truncation, 12 digits for cg2d) |
| 110 |
|
o pkg/seaice (SItracer pickup) |
| 111 |
|
* Simple code for pickup of SItracer fields (#define ALLOW_SITRACER) |
| 112 |
|
To be done: |
| 113 |
|
- thorough testing of missing fields in S/R seaice_read_pickup |
| 114 |
|
- thorough matching between generic tracers and specific fields used |
| 115 |
|
(e.g. volume-weighted age, area-weighted age, salinity, ...) |
| 116 |
|
o pkg/grdchk: |
| 117 |
|
- all procs calculate and print finite-difference gradient (for testreport) |
| 118 |
|
- change description output used by testreport; |
| 119 |
|
- AD testing: also compare and report Finite-difference gradient; |
| 120 |
|
=> update all output_adm*.txt |
| 121 |
|
o test exp. global_ocean_ebm: |
| 122 |
|
- switch back useRealFreshWaterFlux (no problem anymore with run-off); |
| 123 |
|
and turn on exactConserve; update (fwd) output. |
| 124 |
|
o pkg/seaice (LSR): |
| 125 |
|
- print to STDOUT number of solver iterations at SEAICE_monFreq interval. |
| 126 |
|
- modify solver to deal cleanly with u/v switch at face edges (CS-grid): |
| 127 |
|
* identical results with non-CS topology (but save up to half of EXCH calls) |
| 128 |
|
* no much changes in cs-32 verification tests since no ice at face edges |
| 129 |
|
(differences due to convergence criteria, since nb of iterations changed) |
| 130 |
|
* expect some differences on lat-lon-cap grid. |
| 131 |
|
update FWD results of global_ocean.cs32x15 test exp. (icedyn & seaice) |
| 132 |
|
o pkg/exch2: |
| 133 |
|
- add new CPP option to disable Cumul-Sum matrix (to save storage when using |
| 134 |
|
many tiles); until alternative coding is added, just add a stop. |
| 135 |
|
o pkg/kpp: |
| 136 |
|
- move ptracer/salt correction to KPP diffusivity from calc_3d_diffusivity.F |
| 137 |
|
to S/R KPP_CALC_DIFF_PTR (+ add tracer-index to argument list). |
| 138 |
|
o optfiles: |
| 139 |
|
- merged few MPI specific (*+mpi*) and multi-threads specific (*+mth) optfiles |
| 140 |
|
into corresponding standard one, which now support different build: |
| 141 |
|
(when using "genmake2 -mpi" for MPI built, "genmake2 -omp" for OpenMP built) |
| 142 |
|
- add "darwin_amd64_gfortran" optfile (initially from Constantinos, tested by Gael) |
| 143 |
|
o pkg/obcs (seaice): |
| 144 |
|
- add missing EXCH for uIce,vIce after applying OBCS; |
| 145 |
|
- remove S/R EXF_SWAPFFIELDS_X/Y , disable S/R EXF_SET_OBCS_X/Y by adding |
| 146 |
|
argument = number of levels to S/R EXF_SWAPFFIELDS_XZ/YZ and S/R |
| 147 |
|
EXF_SET_OBCS_XZ/YZ ; update obcs_exf_load.F; |
| 148 |
|
- fix EXF_SET_OBCS_XZ/YZ for multi-threaded (assuming loaded arrays are shared) |
| 149 |
|
- verification exp. seaice_obcs: split domain in 2 tiles ; add SIZE.h_mpi |
| 150 |
|
and eedata.mth for MPI and mutlti-threaded testing. |
| 151 |
|
o pkg/flt: |
| 152 |
|
- make FLT EXCH working with multi-threads (and MPI + multi-threads); |
| 153 |
|
- use buffer for floats IO (trajectories & profiles); size set in FLT_SIZE.h; |
| 154 |
|
- fix pickup and trajectories + profiles output with multi-threads; |
| 155 |
|
- add parameter to select group of fields to write along trajectory and |
| 156 |
|
profile (default unchanged); |
| 157 |
|
- only call PORT_RAND* function when flt_noise <> 0; |
| 158 |
|
- add multi-threaded test for verification exp. flt_example. |
| 159 |
|
o eesupp/src: |
| 160 |
|
- fix S/R ALL_PROC_DIE for MPI+multi-threads. |
| 161 |
|
- USE_OMP_THREADING: check that number of threads (from eedata) matches |
| 162 |
|
actual number (from env.var. setting). |
| 163 |
|
o pkg/ebm: |
| 164 |
|
- fix multi-threading in ebm_area_t.F ; clean-up EBM.h header file. |
| 165 |
|
- add multi-threaded test for verification exp. global_ocean_ebm |
| 166 |
|
|
| 167 |
|
checkpoint63b (2011/08/29) |
| 168 |
|
o pkg/ebm: |
| 169 |
|
- fix sign of runoff; |
| 170 |
|
- update directly forcing fields (Qnet,EmPmR) to use standard |
| 171 |
|
external_forcing_surf.F (remove ebm_forcing_surf.F); |
| 172 |
|
- avoid loading forcing fields a 2nd time (1rst time from ini_forcing.F); |
| 173 |
|
- set either tauThetaClimRelax (in data) or tauThetaZonRelax (in data.ebm) |
| 174 |
|
and apply relaxation towards corresponding SST field (2-D map or ZonAver). |
| 175 |
|
- update global_ocean_ebm parameters file data and update output (fwd+ad). |
| 176 |
|
o verification/matrix_example: |
| 177 |
|
- split domain in 8 tiles (2x4); add SIZE.h_mpi and eedata.mth for |
| 178 |
|
MPI and mutlti-threaded testing. |
| 179 |
|
o pkg/matrix: |
| 180 |
|
- change name of S/R MATRIX_INIT to S/R MATRIX_INIT_VARIA ; |
| 181 |
|
- fix for multi-tiles (add bi,bj indices to exp/impMatrixCounter); |
| 182 |
|
- fix for multi-threaded (remove DATA statement); |
| 183 |
|
- use standard DIFFERENT_MULTIPLE function; |
| 184 |
|
- add MATRIX_OPTIONS.h |
| 185 |
|
o verification/matrix_example: |
| 186 |
|
To allow to split the domain in several tiles, change the domain size |
| 187 |
|
from 31x31 to 32x32 by adding 1 row (on Southern side) + 1 column (on |
| 188 |
|
Western side) of land points. |
| 189 |
|
o (*_)write_pickup: |
| 190 |
|
- fix few bugs in mypackage_read/write_pickup.F |
| 191 |
|
- in MDS_WR_METAFILES calls, replace "myTime" by a local copy into |
| 192 |
|
array of length 1 (was caught by compiler syntax-check). |
| 193 |
|
o pkg/diagnostics (velocity potential solver): |
| 194 |
|
- fix initialisation of main diagonal matrix (for overlap values) |
| 195 |
|
o tools/build_options: |
| 196 |
|
- towards single optfile for MPI and non-MPI built: merge few +mpi optfile |
| 197 |
|
into previously non MPI one (linux_am64 pgf77,open64,pathf90,ifort11) |
| 198 |
|
- linux_amd64_gfortran: merge +mpi_generic version into standard optfile: |
| 199 |
|
* move IEEE flags to DEVEL section; take IEEE flags from +mpi_generic ; |
| 200 |
|
* FAST case: for now, keep -O3 with ini_masks_etc.F in NOOPTFILES list; |
| 201 |
|
and use "-devel" testreport option for standard testing on baudelaire. |
| 202 |
|
o pkg/grdchk: |
| 203 |
|
- Assume that all gradient-check position are within the same tile |
| 204 |
|
corresponding to global indices (iGloTile,jGloTile) and that |
| 205 |
|
positions (i,j indices or nbeg) are local to this tile; |
| 206 |
|
- set tile local-indices and proc. number according to tile global-indices; |
| 207 |
|
- remove grdchkwhichproc from namelist (retired param); |
| 208 |
|
- send adjoint gradient to all procs so that all procs print it to STDOUT; |
| 209 |
|
- switch on MPI test for global_ocean.90x40x15 AD test experiment. |
| 210 |
|
o adding package bbl (Bottom Boundary Layer) |
| 211 |
|
description in MITgcm/pkg/bbl/bbl_description.tex |
| 212 |
|
example/test experiment in MITgcm_contrib/bbl |
| 213 |
|
o adjoint verifications |
| 214 |
|
- changed 5 experiments to multi-tiles |
| 215 |
|
* global_ocean.90x40x15 |
| 216 |
|
* obcs_ctrl |
| 217 |
|
* offline_exf_seaice |
| 218 |
|
* tutorial_dic_adjoffline |
| 219 |
|
* tutorial_tracer_adjsens |
| 220 |
|
changed reference output for those that crossed tile boundaries |
| 221 |
|
(changed choices of gradient check positions) |
| 222 |
|
- with same tiling, switch on MPI test for the last 4 (add SIZE.h_mpi) |
| 223 |
|
- temporary bug fix in grdchk_loc.F for obcs |
| 224 |
|
o genmake2 & testreport |
| 225 |
|
- add target "obj" (just for OBJFILES files) to Makefile; |
| 226 |
|
- add option to specify FEXTRAFLAGS on command line. |
| 227 |
|
o pkg/mypackage: |
| 228 |
|
- remove bi,bj arguments from argument list of S/R MYPACKAGE_CALC_RHS |
| 229 |
|
|
| 230 |
|
checkpoint63a (2011/08/02) |
| 231 |
|
o pkg/diagnostics (velocity potential): |
| 232 |
|
- fix velocity-potential calculation when using OBCS: balance net inflow |
| 233 |
|
at each level (new S/R OBCS_DIAG_BALANCE). |
| 234 |
|
- fix some of the problems with OBCS in stream-function calculation. |
| 235 |
|
o pkg/exf/exf_readparms.F |
| 236 |
|
- make treatment of optional namelists (EXF_NML_04, EXF_NML_SGRUNOFF, |
| 237 |
|
EXF_NML_OBCS) more consistent, so that you need not have these lists |
| 238 |
|
in data.exf when they are not used. |
| 239 |
|
o optim/optim_readdata.F: |
| 240 |
|
- add some output about obcs variables that might help debugging |
| 241 |
|
- add a check about nvarlength |
| 242 |
|
o pkg/ctrl: once again, fix computation of endrec in s/r ctrl_init_rec |
| 243 |
|
o pkg/diagnostics (velocity potential): |
| 244 |
|
- fix mask for OBCS (still problems in stream-function with OBCS); |
| 245 |
|
- add specific parameter (default = main code CG2D params) for solver; |
| 246 |
|
- in case of poor convergence, use solution corresponding to lowest residual. |
| 247 |
|
o pkg/seaice/seaice_growth.F: small changes that permit vectorizable |
| 248 |
|
adjoint code. In general statements like tmpscal=fld(i,j) are problematic |
| 249 |
|
and should be avoided where possible. |
| 250 |
|
o model/src (rStar): |
| 251 |
|
- include r* effect on reference-state (PhiRef) in totPhiHyd |
| 252 |
|
(so that it's really the potential anomaly at the cell center); |
| 253 |
|
affect solution only if z* with full pressure in EOS. |
| 254 |
|
- test-exp global_ocean.90x40x15: reduce viscA4 (was not stable) and |
| 255 |
|
update output (because of z* in totPhiHyd and viscA4 changes); |
| 256 |
|
- keep the previous diagnostics as "PHIHYDcR" (closer to potential anomaly |
| 257 |
|
@ constant r) until interpolation at constant r is implemented. |
| 258 |
|
o pkg/diagnostics |
| 259 |
|
- fix RCENTER diagnostics in case cell center is not at interface middle |
| 260 |
|
o add code for TEOS-10 (www.teos-10.org, McDougall et al. 2011). Use |
| 261 |
|
this eos with eosType = 'TEOS10', in data (PARM01). This eos implies |
| 262 |
|
that THETA and SALT are "conservative temperature" and "absolute |
| 263 |
|
salinity" |
| 264 |
|
o pkg/ctrl |
| 265 |
|
- fix computation of endrec in s/r ctrl_init_rec |
| 266 |
|
- ctrl_pack/unpack_*: remove unused variables and some obsolete code |
| 267 |
|
o pkg/seaice/seaice_growth.F |
| 268 |
|
- fix recomputation with FENTY_AREA_EXPANSION_CONTRACTION code, |
| 269 |
|
remove then obsolete STORE directives |
| 270 |
|
- fix STORE directives for tice and tices, so that only the |
| 271 |
|
appropriate part of the fields are stored (and not the entire |
| 272 |
|
field); update global_ocean.cs32x15.seaice AD output. |
| 273 |
|
- cosmetic changes: adjust CPP flags and indentation for better |
| 274 |
|
legibilty |
| 275 |
|
o pkg/autodiff: add mnc-output of adBottomDragFld (code contributed by |
| 276 |
|
Chris Wilson) |
| 277 |
|
o pkg/gmredi: |
| 278 |
|
Add Sub-Meso Eddies parameterisation from Baylor (for now, disabled with TAF), |
| 279 |
|
and test it in verification exp. "front_relax.bvp" (update output.bvp.txt) |
| 280 |
|
|
| 281 |
|
checkpoint63 (2011/07/11) |
| 282 |
|
o pkg/exch2: |
| 283 |
|
- rename+move: nTiles in W2_EXCH2_PARAMS.h --> exch2_nTiles in W2_EXCH2_TOPOLOGY.h |
| 284 |
|
- exch2 version of Cumul-Sum routine: S/R W2_SET_MAP_CUMSUM set up the |
| 285 |
|
(2*nTiles)x(nTiles) Cumul-Sum matrix and Master Proc collects all tiles X,Y |
| 286 |
|
increments, calculates Cumul-Sum @ tile origin and sends results to each tile. |
| 287 |
|
o pkg/diagnostics: |
| 288 |
|
- deal with pairs of post-processed diags which are computed together; in |
| 289 |
|
particular, re-order list of diags within 1 file to save 2nd computation. |
| 290 |
|
- add S/R to compute velocity stream-function (called after velocity-potential |
| 291 |
|
calculation) from divergence free transport -> new diagnostics "PsiVEL". |
| 292 |
|
- add parameter to select grid-point location where PsiVEL == 0. |
| 293 |
|
o eesupp/src: |
| 294 |
|
new S/R to calculate cumulated sum on tiled array, corner grid-cell location |
| 295 |
|
(specific S/R needed with exch2 topology - for now, just print an error msg) |
| 296 |
|
o pkg/shelfice |
| 297 |
|
- Implement friction velocity-dependent transfer coefficients following |
| 298 |
|
Holland and Jenkins, JPO, 1999 |
| 299 |
|
Original code by M. Losch with small modifs. |
| 300 |
|
Enable with CPP option SHI_ALLOW_GAMMAFRICT |
| 301 |
|
and runtime flag SHELFICEuseGammaFrict |
| 302 |
|
o verification lab_sea & global_ocean.cs32x15: |
| 303 |
|
- use tools/xmakedepend (instead of cyrus_makedepend, in genmake_local): |
| 304 |
|
does not miss some dependencies (but is slower) and also works with many files |
| 305 |
|
(default makedepend limitation for AD built). |
| 306 |
|
- use cyrus_makedepend in adjustment.128x64x1 (so that it still tested). |
| 307 |
|
o pkg/diagnostics: |
| 308 |
|
- add run-time parameter "useMissingValue" (def=False) to fill land-point |
| 309 |
|
(i.e., where mask=0) with MissingValue ; used only in MNC output file. |
| 310 |
|
This replaces CPP-option DIAGNOSTICS_MISSING_VALUE. |
| 311 |
|
- Extend the use of "missing_value" setting from data.diagnostics for the case |
| 312 |
|
of counter-diagnostics (where fraction-weight is zero). This applies both to |
| 313 |
|
MNC and MDS output files. |
| 314 |
|
o pkg/seaice/seaice_growth.F: |
| 315 |
|
- add 'a priori' cap of latent heat flux based on actual mass considerations |
| 316 |
|
when SEAICE_ADD_SUBLIMATION_TO_FWBUDGET is used in conjunction with the |
| 317 |
|
non-legacy seaice_solve4temp. |
| 318 |
|
- add diagnostic of the residual sublimation freshwater flux (the freshwater |
| 319 |
|
flux taken from the ocean if the total actual latent heat fluxes exceed |
| 320 |
|
the maximum) |
| 321 |
|
- remove the CPP switch around the *_FWbySublim unit change and move diagnostic |
| 322 |
|
fill for SIatmQnt, SIfwSubl, SIatmFW to the end of seaice_growth.F. |
| 323 |
|
- add diagnostic of the actual sublimation freshwater flux (that is 0. ifndef |
| 324 |
|
SEAICE_ADD_SUBLIMATION_TO_FWBUDGET) and of the latent heat flux (evap+sublim). |
| 325 |
|
- bug fix : frWtrAtm needs a_FWbySublim ifdef SEAICE_ADD_SUBLIMATION_TO_FWBUDGET. |
| 326 |
|
- bug fix : missing part to Martin's r_FWbySublim pathological case fix. |
| 327 |
|
o verification 1D_ocean_ice_column: |
| 328 |
|
made model domain truly 1x1x23 instead of 5x5x23 with only center cell (3,3,:) wet |
| 329 |
|
update results |
| 330 |
|
o pkg/diagnostics: |
| 331 |
|
- Implement setting of "Post-Processed" diagnostics (corresponding to gdiag(5)='P') |
| 332 |
|
which are not filled-up but computed from other diags ; In this case, the mate |
| 333 |
|
diag points to the primary (filled-up) diag to used for post processing. |
| 334 |
|
- Update Velocity-potential computation using post-processed diag framework. |
| 335 |
|
==> Change the ordering of fields and time-periods in MDS output file |
| 336 |
|
when using periodic averaging: now writes one time record with the full set |
| 337 |
|
of fields for this time period, then the next time reccord (it used to be |
| 338 |
|
one field with the full set of time records then the next field). |
| 339 |
|
Note: this is now similar to NetCDF (MNC) output file. <== |
| 340 |
|
|
| 341 |
|
checkpoint62z (2011/06/21) |
| 342 |
|
o adjoint monitor: |
| 343 |
|
- Add adjoint monitor output for packages exf, seaice |
| 344 |
|
and update relevant testreport output |
| 345 |
|
o pkg/seaice: |
| 346 |
|
- re-order diags inits so that they appear in order in available_diagnostics.log |
| 347 |
|
- added diags: hsnow thermo. rates of change, area preceeding ridiging process, |
| 348 |
|
area/heff/hsnow preceeding thermodynamic growth/melt. |
| 349 |
|
- modified diag: SIdA; output actual effect, as the description states. |
| 350 |
|
- modified description: SIqneti, SIqneto, SIuheff, SIvheff, SIdAb*, HSNOW. |
| 351 |
|
- retired redundant diags: SIyneg, SIfice, SIsnwice, SIthdgrh. |
| 352 |
|
(SIyneg->SIdHbOCN; SIfice->SIdHbATC; SIsnwice->SIdHbFLO; |
| 353 |
|
and SIthdgrh->SIdHbOCN+SIdHbATC+SIdHbATO) |
| 354 |
|
- modified diags (2) : switch all SIdA*, SIdH*, SIdS* from delta to rates. |
| 355 |
|
- added diagnostics for seaice generic tracers (SItr*). |
| 356 |
|
- added SItrUnit and SItrNameLong run time param (for SItr* diags). |
| 357 |
|
- in diag names, replaced 'PrTh' abbrev. of 'preceeding thermo' with 'PT'. |
| 358 |
|
I. Fenty: 18-Jun-2011: |
| 359 |
|
Minor changes to seaice package: |
| 360 |
|
1) Retired old variables (A22, SEAICE_lhsublim, areaMax, areaMin, hiceMin) and |
| 361 |
|
added some new ones (SEAICE_area_reg, SEAICE_hice_reg, SEAICE_area_floor) |
| 362 |
|
- Differentiated "regularization variables" from "floor variables" |
| 363 |
|
* areaMin became SEAICE_area_reg (old A22) and SEAICE_area_floor |
| 364 |
|
* hiceMin became SEAICE_hice_reg (old hiceMin) |
| 365 |
|
(with _reg meaning regularization variable) |
| 366 |
|
- SEAICE_lhSublim becomes lhSublim, the sum of SEAICE_lhEvap and SEAICE_lhFusion |
| 367 |
|
so as to ensure energy conservation when going between phases |
| 368 |
|
- A22 was not used anywhere |
| 369 |
|
2) Changed regularization procedure for heffActual and hsnowActual to ensure |
| 370 |
|
well-boundedness and smooth adjoint in seaice_growth.F |
| 371 |
|
3) Fixed a bug where seaice_solve4temp would not recognize ice-free grid cells |
| 372 |
|
because the old regularization always set heffActual >= 0.05 cm |
| 373 |
|
4) Changed the model so that the default behavior is to put a small (10^-5) "floor" |
| 374 |
|
on AREA when HEFF > 0. |
| 375 |
|
- went from requiring ALLOW_PRECLUDE_INFINITESIMAL_AREA to be defined to |
| 376 |
|
requiring that DISABLE_AREA_FLOOR *not* be defined |
| 377 |
|
Affected verifications: |
| 378 |
|
- 1D_ocean_ice_column (output.txt output_adm.txt) |
| 379 |
|
- global_ocean.cs32x15 (output.seaice.txt output_adm.seaice.txt) |
| 380 |
|
o pkg/diagnostics: |
| 381 |
|
- move MNC code from diagnostics_out.F into new diagnostics_mnc_out.F (2 S/R). |
| 382 |
|
- rename S/R GETDIAG to DIAGNOSTICS_GET_DIAG and change type of 1rst argument |
| 383 |
|
(was _RL, now integer) with option = 0 to retrieve all levels. |
| 384 |
|
- add pieces of code to compute velocity potential at each level (new diagnostics |
| 385 |
|
PhiVEL, works only when UVELMASS & VVELMASS are also turned on); |
| 386 |
|
for now, using same params as CG2D (target-resisual, max Nb of iter ...). |
| 387 |
|
- fix few issues with counter diagnostics in diagnostics_setdiag.F & |
| 388 |
|
diagstats_setdiag.F (if diagnostic counter is itself a counter diag) |
| 389 |
|
and in diagnostics_fract_fill.F (do not need a valid pointer for the |
| 390 |
|
diagnostics counter-mate); check that definition of counter-diagnostic |
| 391 |
|
includes a valid counter-mate (in diagnostics_set_levels.F). |
| 392 |
|
o model/src (rStar): |
| 393 |
|
- add rStar scaling factor in diags_phi_hyd.F (similar to calc_grad_phi_hyd.F); |
| 394 |
|
affect solution only if z* with full pressure in EOS; |
| 395 |
|
update output of test-exp global_ocean.90x40x15. |
| 396 |
|
o pkg/seaice: |
| 397 |
|
- introduce siEps=1e-5 parameter |
| 398 |
|
- use areaMax in AREA update (part 4), consistent with ridging step (part 2.5). |
| 399 |
|
- re-organize SEAICE_ADD_SUBLIMATION_TO_FWBUDGET code. |
| 400 |
|
- introduce ALLOW_SITRACER and seaice_tracer_phys.F to handle generic seaice tracer. |
| 401 |
|
For now it covers, and was tested for, ice salinity and age (attached to HEFF), |
| 402 |
|
as well as ice cover age and ridging index (attached to AREA). |
| 403 |
|
For now the run time parameters (in data.seaice/SEAICE_PARM03) are |
| 404 |
|
SItrName (e.g. 'age') and SItrMate ('HEFF', which is the default, or 'AREA'). |
| 405 |
|
Up to SItrMaxNum=3 tracers can be used simultaneously. |
| 406 |
|
o pkg/salt_plume: |
| 407 |
|
-introduce SPsalFRAC run time parameter that sets the fraction of the |
| 408 |
|
salt by-product of ice growth (SPsalFRAC*(1-SIsalFRAC)*salt typically) |
| 409 |
|
that gets vertically redistributed (fluxed down) by pkg/salt_plume. |
| 410 |
|
o model debug options: |
| 411 |
|
- Switch reading of debugMode from S/R INI_PARMS (parameter file "data", stored |
| 412 |
|
in PARAMS.h) to S/R EESET_PARMS (parameter file "eedata", stored in EEPARAMS.h). |
| 413 |
|
- Refine debug-msg printing selection (3 new levels): |
| 414 |
|
debugMode : controls printing of debug messages (sequence of S/R calls). |
| 415 |
|
debugLevel : controls printing of algorithm-relevant auxiliary/debug messages |
| 416 |
|
debLevZero = 0 ; /* level of disabled aux. msg printing */ |
| 417 |
|
debLevA = 1 ; /* level of minimum aux. msg printing */ |
| 418 |
|
debLevB = 2 ; /* level of low aux. print (report read-file opening)*/ |
| 419 |
|
debLevC = 3 ; /* level of moderate debug prt (most pkgs debug msg) */ |
| 420 |
|
debLevD = 4 ; /* level of enhanced debug prt (add DEBUG_STATS prt) */ |
| 421 |
|
debLevE = 5 ; /* level of extensive debug printing */ |
| 422 |
|
New debugLevel default: with default debugMode value (=F), |
| 423 |
|
= 2 (debLevB) or = 1 (debLevA) in case ALLOW_AUTODIFF is defined.; |
| 424 |
|
and if debugMode=T, default debugLevel = 4 (debLevD). |
| 425 |
|
Relation with previous settings: |
| 426 |
|
debLevB (=2) corresponds to former debLevA (=1) |
| 427 |
|
debLevD (=4) corresponds to former debLevB (=2) |
| 428 |
|
Note: |
| 429 |
|
a) if debugLevel=0, skip WRITE_GRID call when nIter0 > 0 (same as before); |
| 430 |
|
b) default debugLevel=1 is convenient in AD runs (avoid many file opening msgs); |
| 431 |
|
to recover previous default, set debugLevel=2 (e.g. AD-test hs94.1x64x5) |
| 432 |
|
- Add new parameter (printResidualFreq, read from PARM02 namelist) to print |
| 433 |
|
detailed CG solvers convergence; off (=0) by default, except if debugLevel=5. |
| 434 |
|
o pkg/obcs: |
| 435 |
|
- change OBCSfixTopo default from True to False (should not change the results); |
| 436 |
|
keep OBCSfixTopo=T in verification-exp. seaice_obcs. |
| 437 |
|
o pkg/thsice: |
| 438 |
|
- fix bug in thsice_calc_thickn.F (was introduced in version 1.18) in |
| 439 |
|
2nd layer seaice sublimation (unlikely to happen) |
| 440 |
|
- use Interior mask for OBCS in thsice_advection.F (similar to what is |
| 441 |
|
done in gad_advection & seaice_advection). |
| 442 |
|
o adjoint: |
| 443 |
|
bug fix spotted by Gael Forget: |
| 444 |
|
seaice_advection and thsice_advection still had |
| 445 |
|
legacy CPP options to avoid CS corner updates for adjoint |
| 446 |
|
o pkg/seaice: |
| 447 |
|
=> Change some of the defaults (2011/05/29): |
| 448 |
|
- change SEAICEuseFluxForm default (from False to True) |
| 449 |
|
- change DIFF1 default (from 0.004 m/s to 0. m/s) |
| 450 |
|
And print warnings if DIFF1 is not set in data.seaice and centered |
| 451 |
|
avection scheme is used without any diffusion. |
| 452 |
|
also: |
| 453 |
|
- fix seaice_diffusion calls in default AdvScheme part of seaice_advdiff.F; |
| 454 |
|
- set SEAICEdiffKh[] default in the same way as SEAICEadvSch[] |
| 455 |
|
- stop if DIFF1<>0 with SEAICEadvScheme <> 2, if DIFF1<>0 with any |
| 456 |
|
other non-zero SEAICEdiffKh[] diffusivity, if SEAICEadvSalt=T with |
| 457 |
|
#undef SEAICE_VARIABLE_SALINITY, if SEAICEadvAge=T with #undef SEAICE_AGE. |
| 458 |
|
update results of test exp.: |
| 459 |
|
lab_sea: output.fd (switch to AdvScheme=33 with diffKh=20) |
| 460 |
|
output.hb87, output_adm.evp (SEAICEuseFluxForm=T and diffKh=200) |
| 461 |
|
output_adm.noseaicedyn (SEAICEuseFluxForm=T) |
| 462 |
|
offline_exf_seaice: output.seaicetd, output_adm (SEAICEuseFluxForm=T) |
| 463 |
|
seaice_obcs: output (SEAICEuseFluxForm=T) |
| 464 |
|
and prescribe old default for SEAICEuseFluxForm (=F) and DIFF1 (=0.004) |
| 465 |
|
for standard fwd & AD lab_sea tests (to keep the same output). |
| 466 |
|
o pkg/seaice/seaice_growth.F: |
| 467 |
|
- use Ian Fenty's capping formulas for actual ice |
| 468 |
|
thickness thoughout EVOLUTION branch. |
| 469 |
|
- put treatment pathological case #2) in CPP brackets |
| 470 |
|
(ALLOW_AVOID_INFINITESIMAL_AREA, undef by default). |
| 471 |
|
- update global_ocean.cs32x15 and 1D_ocean_ice_colum results accordingly. |
| 472 |
|
o pkg/seaice: |
| 473 |
|
- seaice_diffusion.F : remove the min(DX,DY) factor permanently; |
| 474 |
|
pass the Laplacian diffusivity (in m^2/s) diffKh as a parameter; |
| 475 |
|
also pass fac as a param, which can be 1. (to update the tendency |
| 476 |
|
field) or the time step (to update the field itself). |
| 477 |
|
- seaice_advdiff.F: for multidim, call seaice_diffusion with |
| 478 |
|
e.g. SEAICEdiffKhHeff and with fac=1; for single-dim, call |
| 479 |
|
seaice_diffusion with e.g. SEAICEdiffKhHeff and with fac=time step. |
| 480 |
|
- SEAICE_PARAMS.h, seaice_readparms.F, seaice_check.F : |
| 481 |
|
hooks to handle SEAICEdiffKhHeff, SEAICEdiffKhArea, etc. |
| 482 |
|
o pkg/seaice: |
| 483 |
|
diffusion with centered advection scheme (diffus.F & advect.F): |
| 484 |
|
- simplify S/R DIFFUS (put deltaT & DIFF1 factor outside, move EXCH outside) |
| 485 |
|
to just compute the laplacian of input-field. |
| 486 |
|
- unroll k loop (harmonic,biharmonic diff) in advect.F to save 1 exch call. |
| 487 |
|
- fix flux-form version in diffus.F to conserve tracer. |
| 488 |
|
update output of lab_sea.fd & lab_sea.hb87 (1rst set of modif affects |
| 489 |
|
machine truncation) and lab_sea.salt_plume (conservative flux-form modif) |
| 490 |
|
|
| 491 |
|
checkpoint62y (2011/05/26) |
| 492 |
|
o pkg/seaice: |
| 493 |
|
- seaice_growth.F : in EVOLUTION branch, always include |
| 494 |
|
a_QSWbyATM_cover in qnet (as it should be). |
| 495 |
|
- verification/global_ocean.cs32x15 : update results accordingly |
| 496 |
|
(output.seaice.txt output_adm.seaice.txt). |
| 497 |
|
o pkg/seaice + obcs: |
| 498 |
|
- use maskInW & maskInS in multi-dim advection and biharmonic diffusion: |
| 499 |
|
equivalent of assuming no tracer gradient outside the OB region. |
| 500 |
|
update output of seaice_obcs (uses DIFF1 <>0 => biharmonic diffusion) |
| 501 |
|
- remove calls to OBCS_COPY_TRACER after EXCH calls (no longer needed |
| 502 |
|
after changes above); |
| 503 |
|
- obcs+seaice cleaning: apply seaice-tracer OB values to tracer field |
| 504 |
|
only at OB location (instead of over a band of OLx width). |
| 505 |
|
o pkg/seaice: |
| 506 |
|
- fix wrong indices in old (non Flux-Form) diffusion default S/R diffus.F |
| 507 |
|
(dxG(I+1, ...) changed to dxG(I, ...), no impact on Lat-Lon grid) |
| 508 |
|
o pkg/obcs: |
| 509 |
|
- split header file "OBCS.h" into 4 separated files: |
| 510 |
|
OBCS_PARAMS.h, OBCS_GRID.h, OBCS_FIELDS.h & OBCS_SEAICE.h |
| 511 |
|
Note: OBCS_PARAMS.h & OBCS_GRID.h can be included without OBCS_OPTIONS.h |
| 512 |
|
(contain no conditional statement which depends on OBCS options). |
| 513 |
|
- move "useOBCSYearlyFields" from "data.obcs" (hold in OBCS_PARAMS.h) |
| 514 |
|
to parameter file "data.exf", namelist EXF_NML_OBCS (in EXF_PARAM.h). |
| 515 |
|
o ctrl, cost, grdchk & autodiff pkgs: |
| 516 |
|
- make options file more standard (like other {PKG}_OPTIONS.h file but |
| 517 |
|
with AD_CONFIG.h); clean-up some of the included header files. |
| 518 |
|
o pkg/obcs: |
| 519 |
|
- do not mask tracer when applying OB (tracer == 0 could generate problems) |
| 520 |
|
- apply tracer OB values to tracer field only at OB location (instead of |
| 521 |
|
over a band of OLx width) |
| 522 |
|
- clean-up obcs_apply_w.F & obcs_apply_eta.F (since less calls to these S/R) |
| 523 |
|
- remove OBCS_MULTIDIM_OLD_VERSION code in gad_advection.F and remove |
| 524 |
|
obcs_apply_tloc.F & obcs_apply_sloc.F |
| 525 |
|
o obcs in model/src: |
| 526 |
|
- call OBCS_APPLY_W only if NonHydrostatic. |
| 527 |
|
- call OBCS_APPLY_ETA only if NonLinFreeSurf. |
| 528 |
|
- reset dEtaHdt and wVel to zero outside OB interior region ; |
| 529 |
|
- leave tracer unchanged outside OB interior region: This has no effect |
| 530 |
|
on the solution but just to prevent unrealistic tracer value outside OB. |
| 531 |
|
- zero out velocity outside OB interior region: This has no effect on |
| 532 |
|
the solution but just to prevent unrealistic velocity outside OB. |
| 533 |
|
o obcs and non-lin free-surf: |
| 534 |
|
move call to OBCS_APPLY_R_STAR from update_r_star.F to calc_r_star.F |
| 535 |
|
and call to OBCS_APPLY_SURF_DR from update_surf_dr.F to calc_surf_dr.F |
| 536 |
|
+ add argument "etaFld" to OBCS_APPLY_R_STAR & OBCS_APPLY_SURF_DR: |
| 537 |
|
this fixes missing EXCH + get consistent eta OB value (+ fix restart). |
| 538 |
|
update results of test-exp. exp4(nlfs). |
| 539 |
|
o exp4 (nlfs) test-exp: |
| 540 |
|
use OBC for Eta at Western & Eastern OB (to test prescribing Eta |
| 541 |
|
with NonLin-FreeSurf); update output.nlfs.txt |
| 542 |
|
o obcs and 2d/3d solver norm: |
| 543 |
|
use interior masks (maskInC) to cancel out cg2d & cg3d matrix coeff at OB |
| 544 |
|
and outside OB interior region (will allow to place OB further inside the |
| 545 |
|
domain). Depending on the domain, can affects cg2d-Norm (previous |
| 546 |
|
cancellation outside OB was only partial), e.g. in exp4 & obcs_ctrl, |
| 547 |
|
and cg3d-Norm (previous Norm was computed before canceling OB terms), |
| 548 |
|
e.g. in tutorial_plume_on_slope. |
| 549 |
|
- update output of exp.: exp4 (x2), obcs_ctrl and tutorial_plume_on_slope. |
| 550 |
|
o pkg rbcs: |
| 551 |
|
- split RBCS.h into 3 files: RBCS_SIZE.h, RBCS_PARAMS.h & RBCS_FIELDS.h |
| 552 |
|
- add capability to apply relaxation to horizontal velocity uVel & vVel. |
| 553 |
|
o pkg/grdchk: add gradient check for control variable xx_shifwflx |
| 554 |
|
o pkg/shelfice/shelfice_init_fixed.F |
| 555 |
|
fix CPP-flag combinations |
| 556 |
|
o pkg/seaice: Merge iceage code of Pierre Rampal |
| 557 |
|
* update/change/correct ice age calculations |
| 558 |
|
* add age updates for pathological cases |
| 559 |
|
Probably needs more testing... |
| 560 |
|
|
| 561 |
|
checkpoint62x (2011/05/13) |
| 562 |
|
o pkg/mom_vecinv: |
| 563 |
|
high-order vorticity scheme: use interior masks (maskInW & S) |
| 564 |
|
to zero-out (outside OB) gradient of vorticity. |
| 565 |
|
o verification exp.: |
| 566 |
|
turn on "useEnergyConservingCoriolis" (better discretisation) in exp. |
| 567 |
|
hs94.128x64x5 and seaice_obcs ; update results. |
| 568 |
|
o pkg/obcs: |
| 569 |
|
fix bug in OBeta for YZ sections in obcs_exf_load.F |
| 570 |
|
o pkg/ctrl, pkg/shelfice, and optim: |
| 571 |
|
- add new control variable xx_shifwflx (fresh water flux underneath ice |
| 572 |
|
shelves). This is almost as tedious as obcs-ctrl, because the |
| 573 |
|
variables needs its own mask. |
| 574 |
|
- add a cost function term that penalizes the deviation of xx_shifwflx |
| 575 |
|
from the first guess (zero) in analogy to other control variables |
| 576 |
|
o pkg/shelfice: |
| 577 |
|
- start basic adjoint |
| 578 |
|
o pkg/ctrl: |
| 579 |
|
- remove extra 2D field "for good measure" when initializing in |
| 580 |
|
ctrl_set_globfld_[xy/xz/yz].F |
| 581 |
|
- replace hard coded precision 32 (a bug) in ctrl_map_ini_gen by ctrlprec |
| 582 |
|
- improve error messages in ctrl_set_unpack_* |
| 583 |
o OBC implementation in momentum: |
o OBC implementation in momentum: |
| 584 |
use interior masks (maskInC,W & S) in momentum S/R: |
use interior masks (maskInC,W & S) in momentum S/R: |
| 585 |
- multiply divergence, tension and some del^i.u or del^j.v by maskInC |
- multiply divergence, tension and some del^i.u or del^j.v by maskInC |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch