| 3 |
|
|
| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
o added CPP_EEOPTIONS.h flag DISABLE_MPI_READY_TO_RECEIVE. Speeds up |
| 7 |
|
1/6-deg Southern Ocean configuration by 8% on IBM Datastar and |
| 8 |
|
high_res_cube configuration by 4% on Columbia. |
| 9 |
|
o clean-up multi-threaded problems (reported by debugger tcheck on ACES); |
| 10 |
|
add (in global_with_exf/code) a version of EXF_INTERP_READ without |
| 11 |
|
dynamical allocation (seems to work in multi-threaded). |
| 12 |
|
o new function "MASTER_CPU_IO" (=T if master thread & master CPU, if using |
| 13 |
|
single CPU io) so that TAF does not have to see EESUPPORT.h |
| 14 |
|
o pkg/seaice: implement sea-ice loading (when using RealFreshWaterFlux) |
| 15 |
|
|
| 16 |
|
checkpoint58p_post |
| 17 |
|
o obcs: stop if obcs is used with useCDscheme = .true. |
| 18 |
|
o shelfice: |
| 19 |
|
- fix a sign error so that no slip at shelfice interface works |
| 20 |
|
- remove stop for combination non-hydrostatic code and shelfice. That does |
| 21 |
|
not mean, that the results are correct, but at least they are numerically |
| 22 |
|
stable |
| 23 |
|
o diagnoctics: modify UVELPHI and VVELPHI to include hfac (i.e. mass-weighted) |
| 24 |
|
o added adjoint exch routined for all exch*3d* routines |
| 25 |
|
and updated flow directives |
| 26 |
|
|
| 27 |
|
checkpoint58o_post |
| 28 |
|
o replace calls to EXCH_Z & EXCH_UV_AGRID: use the new _3D_ version ; |
| 29 |
|
remove old subroutines (_XY_ & _XYZ_); update the adjoint version. |
| 30 |
|
o change controling S/R for writing pickup file (new S/R do_write_pickup) |
| 31 |
|
and clean-up reading/writing of pickup files. |
| 32 |
|
o fix local version of read/write pickup (NH part) in rotating_tank/code |
| 33 |
|
o simplify EXCH interface: |
| 34 |
|
- change _XYZ_ interface to _3D_ subroutine (with 3rd dim in argument list) |
| 35 |
|
- not often used EXCH S/R (exch_z, exch_uv_agrid): keep only the _3D_ version |
| 36 |
|
o added the regrid package |
| 37 |
|
o shelfice: |
| 38 |
|
- add a simple boundary layer scheme to reduce noise |
| 39 |
|
- fix the indexing in shelfice_u/v_drag so that there will be actually drag |
| 40 |
|
|
| 41 |
|
checkpoint58n_post |
| 42 |
|
o fix globalArea multi-threaded calculation: need to apply GLOBAL_SUM to |
| 43 |
|
local variable (not in common block). |
| 44 |
|
o lot of small changes for multi-threaded. Most tests (using OMP & ifort) pass. |
| 45 |
|
- still problems with pkgs: mnc, exp2, dic, cal, exf (and probably more); |
| 46 |
|
- still BEGIN_MASTER/END_MASTER to remove around READ/WRITE_FLD S/R calls |
| 47 |
|
to work with SingleCpuIO. |
| 48 |
|
o add "_d 0" to real constant of pkg/cfc subroutines ; |
| 49 |
|
update exp. cfc_exemple & tutorial_cfc_offline output files. |
| 50 |
|
o multi-threading and IO: |
| 51 |
|
- all threads call READ_FLD S/R (needed if using SingleCpuIO): |
| 52 |
|
(no BEGIN_MASTER/ END_MASTER around the call, but inside MDSIO S/R. |
| 53 |
|
- barrier before and after reading. |
| 54 |
|
- stop if other than master is calling MDS_WRITELOCAL (if #undef IO_THREAD_SAFE) |
| 55 |
|
- avoid using MDS_WRITELOCAL for LAND pkg output. |
| 56 |
|
- new S/R AIM_WRITE_PHYS (replaces AIM_WRITE_LOCAL) to write AIM physics |
| 57 |
|
common-block variables ; Allows multi-threading with master-thread IO. |
| 58 |
|
o change default MAX_NO_THREADS from 32 to 4 ; |
| 59 |
|
and stop cleanly when MAX_NO_THREADS is too small. |
| 60 |
|
o add S/R BAR_CHECK to check multi-threaded barrier synchronization |
| 61 |
|
|
| 62 |
|
checkpoint58m_post |
| 63 |
|
o add a Non-Hydroatatic test (with 3-D Coriolis, NHMTerms, biharmonic Visc. |
| 64 |
|
and side-drag) on CS-grid : global_ocean.cs32x15.viscA4 |
| 65 |
|
o make "tutorial_global_oce_latlon" from standard "global_ocean.90x40x15" exp. |
| 66 |
|
keep global_ocean.90x40x15 (for AD testing) with former secondary test |
| 67 |
|
(".qshyd") as the main test (mv output.qshyd.txt to output.txt) |
| 68 |
|
o CVS modules: update MITgcm_verif_basic; remove MITgcm_verif_ocean & _verif_atmos; |
| 69 |
|
add MITgcm_tutorials (= code + all the tutorials) |
| 70 |
|
o ecco, option to compute cost for total ssh (mean+anom) |
| 71 |
|
o consistent with addition of ptracer and ssh in pkg/profiles |
| 72 |
|
o profiles: adding ptracer and ssh |
| 73 |
|
o testreport: change multi-test output-file name |
| 74 |
|
(from output.txt.{name} to output.{name}.txt) |
| 75 |
|
o profiles: now ok for mutli-tiles case |
| 76 |
|
o seaice: add 4 parameters for different advection schemes for area, |
| 77 |
|
(effective) seaice thickness, enthalphy, and snow height. Their default |
| 78 |
|
is the standard parameter SEAICEadvScheme |
| 79 |
|
o calc_gw: |
| 80 |
|
- add cosFacU in zonal viscous flux (harmonic.visc.). |
| 81 |
|
- put back side-drag (call new S/R MOM_W_SIDEDRAG) and output.txt. |
| 82 |
|
of tutorial_plume_on_slope. |
| 83 |
|
o obcs_calc.F : fix index Pb in OBSeta & OBNeta (Thanks to Gianmaria). |
| 84 |
|
o calc_gw: true flux-form, account for horizontal grid spacing. |
| 85 |
|
update rotating_tank results, but not plume_on_slope (affected at |
| 86 |
|
truncation level, 11 digits). |
| 87 |
|
|
| 88 |
|
checkpoint58l_post |
| 89 |
|
o calc_gw: |
| 90 |
|
- take bi,bj loops outside calc_gw to fix Pb with local array KappaRU,V |
| 91 |
|
- remove side-drag for now (need to fix it); update plume_on_slope results. |
| 92 |
|
- fix for partial-cell & non-uniform dz; update plume_on_slope results. |
| 93 |
|
o commit code to make SST/SSS weights spatially variable |
| 94 |
|
(based on either variable in-situ or variable initial T/S weights) |
| 95 |
|
o pkg/ecco: |
| 96 |
|
* commit code to relax deviations of xx_theta, xx_salt to 5*sigma, |
| 97 |
|
now based on horizontally varying weights wthetaLev, wsaltLev |
| 98 |
|
* cost_averagesflags now returns year indices in addition to day, month |
| 99 |
|
* sflux balance is performed on annual basis |
| 100 |
|
(sflxumm, sfluxmm2, bal_sfluxmm) |
| 101 |
|
* for SSH anomaly, separate more cleanly contrib from T/P, ERS, GFO, |
| 102 |
|
including separate counting and multipliers |
| 103 |
|
o pkg/rbcs |
| 104 |
|
* add all necessary infrastructure for pkg/rbcs levN adjoint checkpointing |
| 105 |
|
|
| 106 |
|
checkpoint58k_post |
| 107 |
|
o testreport: |
| 108 |
|
- always build in dir. "build" (if not there, mkdir build instead of building |
| 109 |
|
& running in input) |
| 110 |
|
- run in dir run (if it exists), in build otherwise. |
| 111 |
|
- call prepare_run (if input/prepare_run exists) |
| 112 |
|
- run multi-threaded test (with new option: "-mth") if file input/eedata.mth exists. |
| 113 |
|
o pkg/exf modifications to interpolation and rotation routines for more |
| 114 |
|
reasonable treatment of North Pole singularity: for tracers North Pole |
| 115 |
|
value is set to northernmost zonal-mean value, for zonal velocity it is |
| 116 |
|
set to zero, and for meridional velocity it is set to northernmost value. |
| 117 |
|
o pkg/exf/exf_check_range.F |
| 118 |
|
- add tile index to diagnostics and move stop statement to the very |
| 119 |
|
end of the routine. |
| 120 |
|
o pkg/diagnostics: (diagnostics_main_init.F) |
| 121 |
|
define title of isomorphic variables according to the current config. |
| 122 |
|
o pkg/thsice with pkg/seaice dynamics: |
| 123 |
|
- fix domain of computation (interior only, required for exf) and do exchanges |
| 124 |
|
- seaice-dynamics forcing: add atmospheric and seaice loading to SSH gradient |
| 125 |
|
o towards a standard way of dealing with multi-threading in MONITOR |
| 126 |
|
and ${PKG}_MONITOR (all except ptracers_monitor) subroutines. |
| 127 |
|
o Pass variable viscosities (vertical and horizontal, e.g., KPP or Leith) |
| 128 |
|
to W momentum equation in nonhydrostatic and quasihydrostatic runs. |
| 129 |
|
o make exf_monitor & seaice_monitor independant of the main (dynamics) monitor. |
| 130 |
|
o pkg/seaice: fix SEAICEuseEVPpickup parameter (put it in a commom block) |
| 131 |
|
o generic_advdiff: DST advection S/R : use local copy of velocity to compute CFL |
| 132 |
|
(problems ? with gradient check in bottom_ctrl_5x5 adjoint test) |
| 133 |
|
o tracer advection: |
| 134 |
|
- clean-up and simplify calc_common_factors.F |
| 135 |
|
- make a local copy of velocity and pass it (as u,v,r_Trans) to DST tracer |
| 136 |
|
advection subroutine (but still not used within DST advection S/R) |
| 137 |
|
o rename pkg aim_ocn_coupler to atm_ocn_coupler (later on, will be used without aim) |
| 138 |
|
|
| 139 |
|
checkpoint58j_post |
| 140 |
|
o new driver S/R: LOAD_FIELDS_DRIVER to load external forcing fields |
| 141 |
|
from file (collects calls previously in FORWARD_STEP) |
| 142 |
|
o move call to SEAICE_MODEL from FORWARD_STEP to DO_OCEANIC_PHYS |
| 143 |
|
o adjoint store dir. adjusted accordingly |
| 144 |
|
o genmake2 |
| 145 |
|
replace target "ad_taf" by $(EXE_AD) ; add $(EXE_AD) to the make CLEAN list. |
| 146 |
|
|
| 147 |
|
checkpoint58i_post |
| 148 |
|
o pkg/seaice: add parameter that allows to scale/turn off effect of seaice |
| 149 |
|
on wind stress (SEAICEstressFactor defaults to 1.) |
| 150 |
|
o Total number of Timers is finite (<=40): comment out 2 non-essential one |
| 151 |
|
in the_main_loop.F to avoid Pb (e.g., when running fizhi test) |
| 152 |
|
o vector_invariant: fix highOrderVorticity(=4.th.Order) advection |
| 153 |
|
on CS-grid: no longer loosing mass. |
| 154 |
|
o exf & curvilinear-grid: no longer modify longitude xC & xG. |
| 155 |
|
o pkg/ggl90: fix sloppy output and fix a bug |
| 156 |
|
o pkg/seaice: added SEAICEuseEVPpickup parameter |
| 157 |
|
- Set to false in order to start EVP solver with |
| 158 |
|
non-EVP pickup files when using SEAICEuseEVP |
| 159 |
|
- fix sign error in diffusion in seaice_calc_rhs |
| 160 |
|
|
| 161 |
|
checkpoint58h_post |
| 162 |
|
* Modifications for bottom topography control |
| 163 |
|
o replace hFacC by _hFacC at various places |
| 164 |
|
o replace ALLOW_HFACC_CONTROL by ALLOW_DEPTH_CONTROL |
| 165 |
|
o add non-self-adjoint cg2d_nsa |
| 166 |
|
o update autodiff support routines |
| 167 |
|
o re-initialise hfac after ctrl_depth_ini |
| 168 |
|
o works for 5x5 box, doesnt work for global_ocean.90x40x15 |
| 169 |
|
* Adding verification for 5x5 box (4-layer) bottom topography control |
| 170 |
|
|
| 171 |
|
checkpoint58g_post |
| 172 |
|
o pkg/diagnostics: implement "periodic averaging diagnostics" (e.g., |
| 173 |
|
used for mean seasonal cycle, mean diurnal cycle); add an example |
| 174 |
|
in verification/global_ocean.cs32x15/input.thsice/data.diagnostics |
| 175 |
|
o pkg/ggl90: |
| 176 |
|
- add parameter (minimum surface kinetic energy) |
| 177 |
|
- fix bug (avoiding division by zero) |
| 178 |
|
o couple pkg/thsice to pkg/seaice: |
| 179 |
|
- modify advection in seaice to accomodate new scalar fields to be advected |
| 180 |
|
- only muliDimAdection scheme work with seaice/thsice |
| 181 |
|
- mapping of thsice-variables to seaice-variables, back and forth |
| 182 |
|
- does not break the verification experiments, but does not give the |
| 183 |
|
desired results either |
| 184 |
|
o pkg/seaice |
| 185 |
|
- make (default) second order advection a bit less confusing? |
| 186 |
|
|
| 187 |
|
checkpoint58f_post |
| 188 |
|
o couple pkg/thsice and pkg/exf (as a preparation for coupling pkg/thsice |
| 189 |
|
to pkg/seaice): |
| 190 |
|
- new routines thsice_map_exf and thsice_get_exf provide the interface |
| 191 |
|
- add an additional formal parameter to thsice_solve4temp |
| 192 |
|
o pkg/exf: |
| 193 |
|
- add new fields wspeed, snowprecip (to be able to couple to thsice) |
| 194 |
|
- correct Stefan-Boltzmann constant |
| 195 |
|
- new parameter for longwave surface emittance |
| 196 |
|
- separate some stuff from exf_bulkformulae into new routines |
| 197 |
|
exf_radiation, exf_wind |
| 198 |
|
- add new parameters and constants: longwave emissivity for ocean, ice |
| 199 |
|
and snow, evaporation parameters over ice |
| 200 |
|
- decouple wind/stress computation from ALLOW_BULKFORMULA: always compute |
| 201 |
|
either wind from stress or stress from wind |
| 202 |
|
o pkg/thSIce: |
| 203 |
|
- split thsice_step_fwd.F in 2 S/R: thsice_step_temp.F & thsice_step_fwd.F |
| 204 |
|
- put i,j loops inside S/R: THSICE_ALBEDO, THSICE_SOLVE4TEMP, THSICE_EXTEND |
| 205 |
|
and THSICE_CALC_THICKN |
| 206 |
|
- for now, call to THSICE_GET_BULKF remains inside i,j loops. |
| 207 |
|
- add CPP option: ALLOW_DBUG_THSICE and header file THSICE_DEBUG.h |
| 208 |
|
to make debugging easier. |
| 209 |
|
- affects results (@ truncation level) of global_ocean.cs32x15.thsice: update output |
| 210 |
|
o pkg/bulk_force: |
| 211 |
|
- clean-up comments and parameters ; |
| 212 |
|
- add new Bulk-Formulae from Large and Yeager, 2004, NCAR/TN-460+STR ; |
| 213 |
|
o fix bug in aim+thsice albedo: snow & ice thickness arguments were switched; |
| 214 |
|
generate new output for test-experiment: aim.5l_cs.thSI |
| 215 |
|
o results_ad : rename "output.txt_adm" to "output_adm.txt" and change |
| 216 |
|
testreport accordingly |
| 217 |
|
|
| 218 |
|
checkpoint58e_post |
| 219 |
|
o SEAICE_CGRID adjoint done |
| 220 |
|
o R_STAR adjoint implemented. |
| 221 |
|
* Seems to just work (based on global_ocean.90x40x15/code_ad_nlfs/ ) |
| 222 |
|
* need to sort out init. of h0facc, etc. |
| 223 |
|
o Make pkg/profile fully independent of ecco,cost, etc. stuff |
| 224 |
|
to be able to use it in pure forward. |
| 225 |
|
|
| 226 |
|
checkpoint58d_post |
| 227 |
|
o upgrade simple test-experiment to tutorial example: |
| 228 |
|
step.1: duplicate the experiment and (step.2) will remove the old copy. |
| 229 |
|
tutorial_barotropic_gyre <- copy of exp0 (R1:barotropic_gyre_in_a_box) |
| 230 |
|
tutorial_baroclinic_gyre <- copy of exp1 (R1:baroclinic_gyre_on_a_sphere) |
| 231 |
|
tutorial_global_oce_latlon <- empty for now (R1:global_ocean_circulation) |
| 232 |
|
tutorial_global_oce_biogeo <- copy of dic_example |
| 233 |
|
tutorial_deep_convection <- almost like exp5 (R1:nonhydrostatic_deep_convection) |
| 234 |
|
tutorial_plume_on_slope <- copy of plume_on_slope |
| 235 |
|
tutorial_tracer_adjsens <- copy of carbon (R1:tracer_adjoint_sensitivity) |
| 236 |
|
tutorial_global_oce_optim <- empty for now (David ?) |
| 237 |
|
tutorial_global_oce_in_p <- copy of global_ocean_pressure |
| 238 |
|
some tutorials (R1) were already checked-in under tutorial_examples in release1_branch |
| 239 |
|
o correct a sign problem in cost_scat.F |
| 240 |
|
o obcs and ptracers: fix boundary conditions for advections schemes |
| 241 |
|
with larger stencil |
| 242 |
|
o thsice adjoint: start |
| 243 |
|
o mom_vecinv: add ISOTROPIC_COS_SCALING flags to mom_vi_hdissip for |
| 244 |
|
completeness |
| 245 |
|
o mom_fluxform: fixing the mutual use of ISOTROPIC_COS_SCALING |
| 246 |
|
and COSINEMETH_III, add comments in default CPP_OPTIONS.h |
| 247 |
|
o ptracers adjoint: |
| 248 |
|
* avoid extensive recomputatations |
| 249 |
|
* fix missing re-init. of gptr (missed by TAF) |
| 250 |
|
o obcs: add a flag (OBCSprintDiags, defaults is true) that allows |
| 251 |
|
me to turn off the excessive output to STDOUT that this package produces |
| 252 |
|
o move NH metric terms & 3.D Coriolis S/R to pkg/mom_common |
| 253 |
|
and add those terms in Vector-Invariant formulation. |
| 254 |
|
o in ctrl pkg, if ALLOW_CAL, use monthly mean controls |
| 255 |
|
if the control period (xx_hfluxperiod etc) is 0 |
| 256 |
|
o in profiles pkg, clean routines and fix details |
| 257 |
|
o in ecco pkg, reading ssh data: |
| 258 |
|
fix problem in case the model does not start at iter0=0 |
| 259 |
|
o fix mnc-pickup/restart for ptracers |
| 260 |
|
o new package prifiles: |
| 261 |
|
code to compute profile-based cost by Gael Forget |
| 262 |
|
(modifs in ecco, ctrl, autodiff) |
| 263 |
|
o Benign bug fix in initialization (spotted by M. Mazloff) of |
| 264 |
|
ctrl_set_globfld_xz.F ctrl_set_globfld_yz.F |
| 265 |
|
|
| 266 |
|
checkpoint58c_post |
| 267 |
|
o adjoint: finish NLFS adjoint and provide reference output in |
| 268 |
|
verification/global_ocean.90x40x15/results_ad_nlfs |
| 269 |
|
o seaice: add an EVP solver following Hunke and Dukowicz (1997) and the |
| 270 |
|
documentation of CICE. Turn on by defining SEAICE_ALLOW_EVP in |
| 271 |
|
SEAICE_OPTIONS.h and SEAICEuseEVP=.true. in data.seaice. Works only |
| 272 |
|
with SEAICE_CGRID defined. |
| 273 |
|
o move calendarDumps from "data" to "data.cal" and clean-up the code |
| 274 |
|
with a simple call to pkg/cal S/R: CAL_TIME2WRITE |
| 275 |
|
o add useEXF & useCAL flags (for now, set in hard-coded way) |
| 276 |
|
o add S/R to compute NH-metrics terms & Coriolis f' in vertical momentum |
| 277 |
|
(were missing in NH mode, and wrong in quasihydrostatic mode). |
| 278 |
|
waiting for Chris updated version of calc_gw to add those CALLs. |
| 279 |
|
o fix bug in quasihydrostatic S/R ; |
| 280 |
|
add a test for this option: global_ocean.90x40x15.qshyd |
| 281 |
|
o new S/R "SET_PARMS": set parameters that might depend on the use of some pkgs. |
| 282 |
|
(called from INITIALISE_FIXED, after INI_PARMS & PACKAGES_READPARAMS) |
| 283 |
|
o add on/off flag useOffLine for package OFFLINE |
| 284 |
|
o seaice: |
| 285 |
|
- modify advection so that it works also on the C-grid |
| 286 |
|
- extend loop ranges in seaice_lsr and seaice_dynsolver; fixes |
| 287 |
|
problem in multi-tile runs with C-grid |
| 288 |
|
- add flags for clipping ice velocities and/or masking solver forcing |
| 289 |
|
- fix drag (ocean on ice) formulation and a bug that caused noise in the |
| 290 |
|
C-grid solver |
| 291 |
|
- added a few comments to seaice_diffusion.F |
| 292 |
|
- separate C-grid and B-grid versions more cleanly (UVM and seaiceMaskU/V |
| 293 |
|
are now exclusive) |
| 294 |
|
- add a new ocean-ice stress coupling method (Hibler and Bryan, 1987) which |
| 295 |
|
required re-ordering the code a little: |
| 296 |
|
+ make DAIRN global variable (defined in SEAICE) |
| 297 |
|
+ move computation of the viscosities eta and zeta into a new separate |
| 298 |
|
routine (because I need to recompute them in seaice_ocean_stress.F) |
| 299 |
|
|
| 300 |
|
checkpoint58b_post |
| 301 |
|
o AIM+thSIce: fix sign of Short-Wave through the Ice ; |
| 302 |
|
update output.txt of exp. aim.5l_cs.thSI |
| 303 |
|
o various mnc cleanups and improvements: |
| 304 |
|
+ shrink lookup tables by factor of ~4 |
| 305 |
|
+ better error reporting when running out of lookup space |
| 306 |
|
+ able to handle longer path/file names (up to 500 chars) |
| 307 |
|
o adjoint: |
| 308 |
|
- Another overhaul of store dirs. for NLFS to eliminate "hidden" recomputations. |
| 309 |
|
- TBD: "hidden" mom_vecinv recomp. in dynamics |
| 310 |
o change forcing_In_AB to affects both T,S forcing and Momentum forcing |
o change forcing_In_AB to affects both T,S forcing and Momentum forcing |
| 311 |
(allow to differentiate between forcing components using new integer flags: |
(allow to differentiate between forcing components using new integer flags: |
| 312 |
momForcingOutAB=1/0 & tracForcingOutAB=1/0) |
momForcingOutAB=1/0 & tracForcingOutAB=1/0) |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch