| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
checkpoint61i |
| 8 |
|
o autodiff: |
| 9 |
|
fix autodiff_sore/_restore indices for obcs |
| 10 |
|
(spotted by I. Hoteit, M. Mazloff) |
| 11 |
|
o pkg/autodiff, pkg/ctrl: |
| 12 |
|
Add flag to reduce all tapelev I/O to single-prec. |
| 13 |
|
In data.ctrl set |
| 14 |
|
doSinglePrecTapelev = .TRUE. |
| 15 |
|
o model/src/ini_curvilinear_grid.F: check in hack for SX8 compiler |
| 16 |
|
to make sure that recip_dxV and recip_dyU are never zero, until we |
| 17 |
|
have a proper exchange for them. Most compilers do not seem |
| 18 |
|
to mind divisions by zero in the overlaps, as long as the results are |
| 19 |
|
not used anywhere, but sxf90 does. |
| 20 |
|
o pkg/shelfice: fix bug with SHELFICEconserve flag, prepare for |
| 21 |
|
real fresh water flux (not yet) |
| 22 |
|
o pkg/mdsio & rw: |
| 23 |
|
- add argument & option to S/R MDS_WRITEVEC_LOC (avoid Open/Close & write-meta) |
| 24 |
|
- add simple interface S/R to MDS_WRITEVEC_LOC in pkg/rw: WRITE_GLVEC_RX. |
| 25 |
|
o pkg/flt: |
| 26 |
|
- replace FLT_INIT by 3-steps standard initialization S/R: readparms, |
| 27 |
|
init_fixed & ini_varia. |
| 28 |
|
- clean-up & simplify linear interpolation S/R: |
| 29 |
|
move flt_bilinear.F -> flt_interp_linear.F ; add arg. myThid. |
| 30 |
|
- fix some indices (mainly vertical index) ; add IMPLICIT NONE. |
| 31 |
|
- more efficient I/O, using new options of MDS_WRITEVEC_LOC S/R. |
| 32 |
|
- add vert. coordinate & horiz. indices to output files. |
| 33 |
|
- change vertical position in initial condition, from index to depth [m]. |
| 34 |
|
- store horizontal (decimal) indices i,j in common block (instead of coord. x,y) |
| 35 |
|
(easier to get to a curvilinear grid) ; delX,delY no longer used. |
| 36 |
|
o pkg/ggl90: a few modifications: |
| 37 |
|
- fix bottom boundary condition, and interpolation of KappaE |
| 38 |
|
near top and bottom |
| 39 |
|
- use same set of viscosity/diffusivity for TKE/momnetum/tracer |
| 40 |
|
- add smoothing option as in Blanke and Delecluse (1993) (--> ALLOW_GGL90_SMOOTH) |
| 41 |
|
- add a few options for mixing-length upper limit (--> mxlMaxFlag) |
| 42 |
|
o rename thetaMin,phiMin -> xgOrigin,ygOrigin |
| 43 |
|
(maintain temporary backward compatibility, until next checkpoint) |
| 44 |
|
|
| 45 |
|
checkpoint61h |
| 46 |
|
o pkg/diagnostics with "periodic averaging diagnostics": |
| 47 |
|
fix S/R DIAGNOSTICS_COUNT ; hack diagnostics_out.F to get NetCDF output. |
| 48 |
|
o pkg/my82: add missing "_d 0" ; update output of exp. vermix.my82 |
| 49 |
|
o pkg/obcs: fix loop ranges for balancing code |
| 50 |
|
o pkg/flt: |
| 51 |
|
- improve criteria for finding which tile owns a float. |
| 52 |
|
- fix FLT_EXCH for case without MPI ; apply cyclic BC + fix also the MPI case. |
| 53 |
|
o Add well-mixed CO2 box in aim_v23 (which includes changes in pkg |
| 54 |
|
aim_compon_interf, ocn_compon_interf, gchem, and dic to get coupling |
| 55 |
|
with ocean biogeo) |
| 56 |
|
o pkg/flt: |
| 57 |
|
- do some cleaning and fix the restart ; |
| 58 |
|
- add output.txt in flt_example/results to get this pkg tested. |
| 59 |
|
o genmake2: add an EXTENDED_SRC option and flag. |
| 60 |
|
for now, cannot be turned on on command line (only through genmake_local) |
| 61 |
|
o pkg/mdsio: read/write vector: |
| 62 |
|
- singleCpuIO: fix RL/RS in MDSWRITEVECTOR by calling new S/R MDS_BUFFERtoRL/RS |
| 63 |
|
- singleCpuIO=F: fix arrType & filePrec & _BYTESWAPIO in both MDSREADVECTOR |
| 64 |
|
& MDSWRITEVECTOR by calling new S/R MDS_RD/WR_VEC_RL/RS. |
| 65 |
|
- affects gradient of test-exp "bottom_ctrl_5x5" (not due to _BYTESWAPIO) |
| 66 |
|
(drop to 4 digits matching for grad). But this test does not look very |
| 67 |
|
reliable (was already failling on eddy, fc5, at 6.digits). Update output. |
| 68 |
|
|
| 69 |
|
checkpoint61g |
| 70 |
|
o pkg/flt: |
| 71 |
|
- move FLT_CPPOPTIONS.h to FLT_OPTIONS.h (standard name) |
| 72 |
|
+ make OPTIONS file more standard. |
| 73 |
|
- rename S/R MDSREADVECTOR_FLT to FLT_MDSREADVECTOR (to mach src file name) |
| 74 |
|
- fix _BYTESWAPIO CPP-Flag |
| 75 |
|
- standard reading of file-parameter "data.flt" |
| 76 |
|
- fix argument list of S/R FLT_RESTART |
| 77 |
|
still need to fix _BYTESWAPIO in pkg/mdsio/mdsio_writevector.F |
| 78 |
|
o pkg/obcs: fix a CPP-flag, bug reported by Jeff Blundell |
| 79 |
|
o gather_2d.F/scatter_2d.F: suppress vectorisation on NEC SX in respond to |
| 80 |
|
a still unfixed compiler/optimization bug |
| 81 |
|
|
| 82 |
|
checkpoint61f |
| 83 |
|
o pkg/seaice, LSR solver: fix bug in GLOBAL_MAX calls ; |
| 84 |
|
update results of global_ocean.cs32x15.icedyn test. |
| 85 |
|
o move getcon.F from model/src to pkg/fizhi. |
| 86 |
|
o pkg/exf: add parameter for extrapolation of temp up to the surface |
| 87 |
|
and option to prevent negative evap (in bulk_largeyeager04 only). |
| 88 |
|
o Prather Advection scheme: fix tracer conservation for divergent flow |
| 89 |
|
(e.g., when using NonLin Free-Surface). |
| 90 |
|
o kpp: retire KPPmld diagnostics --> use MXLDEPTH instead |
| 91 |
|
o gmredi (Visbeck-variable-K): |
| 92 |
|
- add more parameters to Visbeck-variable-K. |
| 93 |
|
- use a 5 points average for dSigmaR in Visbeck-K calculation |
| 94 |
|
(more "homogenous": same stencil and same weights as in dSigmaH) |
| 95 |
|
- update output of ideal_2D_oce experiment. |
| 96 |
|
o fix CPP-flag REAL4_IS_SLOW, when _RS expands into real*4 |
| 97 |
|
(#undef REAL4_IS_SLOW) |
| 98 |
|
- pkg/mnc/mnc_cw_cvars.F: write _RS-grid variable as netcdf-real*4 |
| 99 |
|
coordinate variable |
| 100 |
|
- pkg/monitor/monitor.F: call mon_printstats_rs for _RS variables |
| 101 |
|
(surface forcing fields), removes some of the functionality |
| 102 |
|
- model/src/write_grid.F: write _RS variables as RS-netcdf variables |
| 103 |
|
- model/src/diags_oceanic_surf_flux.F: fix a few diagnostics, but |
| 104 |
|
probably not all: copy _RS surface flux fields to RL field before |
| 105 |
|
calling DIAGNOSTICS_FILL |
| 106 |
|
- model/inc/CPP_EEMACROS.h: make the _EXCH_*_RS macros expand into |
| 107 |
|
real*4 (CALL EXCH_*_RS) versions. |
| 108 |
|
- tested in 2 experiments: inverted_barometer & solid-body.cs-32x32x1 |
| 109 |
|
(generate new output). |
| 110 |
|
|
| 111 |
|
checkpoint61e |
| 112 |
|
o gmredi (Visbeck-variable-K): |
| 113 |
|
- fix the "OLD_VISBECK_CALC" option + move the #undef to GMREDI_OPTIONS.h |
| 114 |
|
- change computation of Visbeck-K where Slope > Smax : |
| 115 |
|
now: N*min(Slope,Smax) (similar to OLD_VISBECK_CALC with gkw91 taperFct) |
| 116 |
|
previously was: M*sqrt(min(Slope,Smax)) ; update ideal_2D_oce output. |
| 117 |
|
o bottom-drag in P-coord.: change units of bottomDragLinear & bottomDragQuadratic |
| 118 |
|
to be the same as with Z-coord. (i.e., m/s & no-units respectively). |
| 119 |
|
o calc_viscosity: fix index range (was wrong for CD-Scheme & explicit Visc) |
| 120 |
|
o pkg gmredi, linear & fm07 tapering: put an upper limit on Slope^2 so that Kwz |
| 121 |
|
does not become too large (to avoid precision problems in implicit-diffusion |
| 122 |
|
solver); update ideal_2D_oce output. |
| 123 |
|
o pkg ggl90: |
| 124 |
|
- in ggl90_cal.F: compute only one set of visc/diff, used to |
| 125 |
|
step forward TKE/T/S/momentum |
| 126 |
|
- default value of GGL90viscMax/GGL90diffMax larger and equal (to |
| 127 |
|
have Prandlt =1) |
| 128 |
|
o gmredi (fm07): |
| 129 |
|
- add lower limit on Transition layer thickness + lower limit on 1/lambda |
| 130 |
|
- allow to set few more params from data.gmredi |
| 131 |
|
o pkg ggl90: |
| 132 |
|
- add missing sqrt(2) factor in computation of mixing length |
| 133 |
|
scale in ggl90_calc.F |
| 134 |
|
(see Eq. (2.35) in Blanke and Delecluse, JPO, 1993) |
| 135 |
|
|
| 136 |
|
checkpoint61d |
| 137 |
|
o adjoint modifs |
| 138 |
|
- adapt to new field rhoInSitu |
| 139 |
|
- allow to revert from StoreDynVars2D, StoreDynVars3D |
| 140 |
|
to individual fields |
| 141 |
|
(switches AUTODIFF_USE_OLDSTORE_2D, AUTODIFF_USE_OLDSTORE_3D) |
| 142 |
|
o store 3-D (in-situ) density in commom block (DYNVARS.h) |
| 143 |
|
- save 1 rho computation. |
| 144 |
|
- no longer modify T & S inland when using down_slope pkg. |
| 145 |
|
o gad_som : allow to switch to SOM advection for T & S without corresponding |
| 146 |
|
pickup files (using new S/R: MDS_CHECK4FILE). |
| 147 |
|
o keep removing _BEGIN/_END_MASTER between MDSIO S/R call |
| 148 |
|
(for useSingleCpuIO to work in multi-threaded) |
| 149 |
|
o pkg/shelfice: |
| 150 |
|
- disentangle parameters: retire shelfIceFile in data/PARM05 and substitute |
| 151 |
|
with SHELFICEtopoFile in data.shelfice |
| 152 |
|
- move code for reading ice shelf topography and modifying hFacC into separate |
| 153 |
|
subroutines of the shelfice package |
| 154 |
|
- separate initialisation of "fixed" and "variable" variables |
| 155 |
|
- kTopC is not (yet) replaced by kSurfC, although this is possible |
| 156 |
|
o fix broken diagnostic SIatmFW and changed the units from m/s to kg/m^2/s |
| 157 |
|
o add initialisation of deepFac if using Pcoords (new S/R set_grid_factors) |
| 158 |
|
and rename: |
| 159 |
|
ini_phiref.F --> set_ref_state.F (+ set anelastic factors) |
| 160 |
|
ini_reference_state.F --> load_ref_files.F (- set anelastic factors) |
| 161 |
|
o change FIND_RHO_SCALAR : return rho (instead of rho - rhoConst) |
| 162 |
|
|
| 163 |
|
checkpoint61c |
| 164 |
|
o add option ALLOW_ADDFLUID and run-time selector "selectAddFluid" |
| 165 |
|
to add mass source/sink of fluid in the interior (3-D generalisation |
| 166 |
|
of oceanic real-fresh water flux). |
| 167 |
|
o pkg/monitor: fix recip_dr argument in mon_advcflw2 call (-> advcfl_W_hf_max) |
| 168 |
|
o vermix test experiment: switch from LINEAR EOS to MDJWF (not tested before) |
| 169 |
|
o new pkg: down_slope |
| 170 |
|
density-driven downsloping flow parameterisation (Tellus, 1999, 412-430). |
| 171 |
|
- presently, modify T & S inland (dwnslp_calc_rho.F). |
| 172 |
|
can be avoided if rho was stored in common bloc and used in calc_phi_hyd. |
| 173 |
|
- implementation for ocean in P-coords not tested. |
| 174 |
|
- add secondary test to global_ocean.90x40x15 (input.dwnslp) to check this pkg. |
| 175 |
|
o move kSurfC,kSurfW & kSurfS from SURFACE.h to GRID.h (next to kLowC) |
| 176 |
|
(to facilitate conversion of oceanic pkg to p-coordinate). |
| 177 |
|
o change FIND_RHO to FIND_RHO_2D : |
| 178 |
|
pass arguments tFld & sFld as local (tile) 2D-arrays ; +fix EOS_CHECK. |
| 179 |
|
o new pkg: cheapaml |
| 180 |
|
|
| 181 |
|
checkpoint61b |
| 182 |
|
o exch2: |
| 183 |
|
change index bounds in rx2_cube exchanges (new S/R: EXCH2_GET_UV_BOUNDS) |
| 184 |
|
- no longer depend on the order sequence (N,S,E,W). |
| 185 |
|
- 3rd exchange no longer needed (tested with 24 tiles). |
| 186 |
|
- same modif to hand-written adjoint S/R (global_ocean.cs32x15: zero diff) |
| 187 |
|
- exch_UV_A-grid readily available (but not yet tested). |
| 188 |
|
comment out 3rd exch call and remove copy of V to U at SE & NE face-corners |
| 189 |
|
(both are no longer needed) in exch_uv: no effect in forward run; but |
| 190 |
|
get truncation error differences with adjoint exch. |
| 191 |
|
o exch2: |
| 192 |
|
- change index-bounds storage (moved from target to local tile, |
| 193 |
|
more intuitive this way) ; rename/remove some variables. |
| 194 |
|
- change utils/exch2/matlab-topology-generator & code_mods accordingly. |
| 195 |
|
|
| 196 |
|
checkpoint61a |
| 197 |
|
o exch2/matlab-topology-generator: |
| 198 |
|
fix bug in index bounds for N-S internal exchange (hard to tell if this |
| 199 |
|
had any effect on the solution). |
| 200 |
|
o pkg/mnc: small fix in mnc_handle_err: make sure that length of msgbuf |
| 201 |
|
is exceeded |
| 202 |
|
o testreport: modify syntax of awk, so that it works also with |
| 203 |
|
awk, and not only with nawk and gawk |
| 204 |
|
|
| 205 |
|
checkpoint61 |
| 206 |
|
o autodiff |
| 207 |
|
Change default for TAF usage by removing argument '-nonew_arg' |
| 208 |
|
This changes TAF subroutine argument list (ordering of active var.) |
| 209 |
|
Omission of this flag leads to incompatibility w.r.t. TAMC |
| 210 |
|
To restore TAMC-compatibility, need following modifs: |
| 211 |
|
* use adoptfile tools/adjoint_options/adjoint_tamc_compatibility |
| 212 |
|
* use CPP option #define AUTODIFF_TAMC_COMPATIBILITY |
| 213 |
|
Tested TAF version is 1.9.22 |
| 214 |
|
N.B.: exch2 hand-written adjoint code currently not TAMC compatible |
| 215 |
|
|
| 216 |
|
checkpoint60 |
| 217 |
|
--->>> This is the last checkpoint for which generating the adjoint |
| 218 |
|
--->>> is done via the TAMC-compatibility TAF-flag "-nonewarg" |
| 219 |
|
--->>> by default. |
| 220 |
|
--->>> Future checkpoints will ommit this flag by default. |
| 221 |
|
o pkg/generic_advdiff, os7mp: simplify and avoid division by zero. |
| 222 |
|
o pkg/grdchk: Add gradient check to etan0 perturbations. |
| 223 |
|
o pkg/exf: read ice area from file (#ifdef ALLOW_ICE_AREAMASK). |
| 224 |
|
o pkg/autodiff: more flexibility in zeroadj. |
| 225 |
|
model/src: modifying do_oceanic_phys.F accordingly. |
| 226 |
|
o pkg/ctrl: Bounds for (kapgm,kapredi,eddypsi) parameter estimation. |
| 227 |
|
o model/src and pkg/gmredi: GMREDI_WITH_STABLE_ADJOINT CPP option |
| 228 |
|
special setup of gmredi for which the adjoint is stable |
| 229 |
|
enough for parameter optimization. |
| 230 |
|
o bridging the gap between eddy stress and GM. |
| 231 |
|
-> eddyTau is replaced with eddyPsi (eddyTau = f x rho0 x eddyPsi) |
| 232 |
|
along with a change in CPP option (now ALLOW_EDDYPSI). |
| 233 |
|
-> when using GM w/ GM_AdvForm: |
| 234 |
|
The total eddy streamfunction (Psi = eddyPsi + K x Slope) |
| 235 |
|
is applied either in the tracer Eq. or in momentum Eq. |
| 236 |
|
depending on data.gmredi (intro. GM_InMomAsStress). |
| 237 |
|
-> ALLOW_EDDYPSI_CONTROL for estimation purpose. |
| 238 |
|
The key modifications are in model/src/taueddy_external_forcing.F |
| 239 |
|
pkg/gmredi/gmredi_calc_*F pkg/gmredi/gmredi_*transport.F |
| 240 |
|
|
| 241 |
|
checkpoint59r |
| 242 |
|
o new S/R (pkg/mdsio/mdsio_facef_read.F) reading of sNx+1 x sNy+1 array |
| 243 |
|
from per-face file (take S/R READSYMTILE_RS out of ini_curvilinear_grid.F); |
| 244 |
|
fix option to read coriolis(@ cell-corner) from files with CS-grid. |
| 245 |
|
o pkg/mnc: add some attributes to standard 1D-coordinate variables |
| 246 |
|
o pkg/diagnostics: enforce missing_value attribute for scalar variables at |
| 247 |
|
mass points, when using mnc for output |
| 248 |
|
- add runtime parameter (missing_value, missing_value_int) to override |
| 249 |
|
default (which is the value of getcon('UNDEF')) |
| 250 |
|
o pkg/mnc: prepare mnc for the actual use of missing values with the |
| 251 |
|
diagnostics package: small fix in mnc_cw_readwrite.template and more |
| 252 |
|
initialisation of never used fields |
| 253 |
|
- replace mnc_common.h and mnc_id_header.h with corresponding upper case |
| 254 |
|
versions |
| 255 |
|
o pkg/thsice/thsice_ini_vars.F added default initial values so that |
| 256 |
|
thermodynamics sea ice solver does not blow up when some initial fields |
| 257 |
|
are missing. |
| 258 |
|
o move from external_forcing_surf.F to ini_forcing.F |
| 259 |
|
the p-coord. hack consisting in using pLoad to set phi0surf ; |
| 260 |
|
update output of tutorial_global_oce_in_p (affected @ truncation level). |
| 261 |
|
o Freshen up a bit hs94.1x64x5, notably: |
| 262 |
|
remove rotationPeriod and tRef=5*0. (adjoint only), and add |
| 263 |
|
exactConserv=.TRUE. |
| 264 |
o pkg/generic_advdiff: |
o pkg/generic_advdiff: |
| 265 |
change option GAD_ALLOW_SOM_ADVECT to GAD_ALLOW_TS_SOM_ADV which only |
change option GAD_ALLOW_SOM_ADVECT to GAD_ALLOW_TS_SOM_ADV which only |
| 266 |
applies to files where Temperature & Salinity 2nd Order moments are used. |
applies to files where Temperature & Salinity 2nd Order moments are used. |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch