| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o tools: |
| 8 |
|
- add a hack in "remove_comments_sed" (used only with TAF) for recent |
| 9 |
|
version of OpenMPI (shipped with FC.23): comment out the new header |
| 10 |
|
file "mpif-sizeof.h" which causes problems to TAF. |
| 11 |
|
o pkg/seaice: |
| 12 |
|
- change the termination criterion so that maxits (SEAICElinearIterMax) |
| 13 |
|
has an effect |
| 14 |
|
- unify iteration parameters for implicit solvers (JFNK and Picard) |
| 15 |
|
SEAICEnonLinIterMax replaces SEAICEnewtonIterMax/NPSEUDOTIMESTEPS |
| 16 |
|
SEAICElinearIterMax replaces SEAICEkrylovIterMax/SOLV_MAX_ITER |
| 17 |
|
SEAICEpreLinIterMax replaces SOLV_MAX_ITER in preconditioner |
| 18 |
|
SEAICEpreNL_IterMax replaces NPSEUDOTIMESTEPS in preconditioner |
| 19 |
|
SEAICEnonLinTol replaces JFNKgamma_nonlin |
| 20 |
|
- remove S/R SEAICE_FGMRES_DRIVER and put content into S/R SEAICE_JFKN |
| 21 |
|
- add new Picard-Krylov solver, compile with SEAICE_ALLOW_KRYLOV and |
| 22 |
|
use with SEAICEuseKrylov |
| 23 |
|
o pkg/seaice/seaice_lsr.F |
| 24 |
|
fix small bug: replace zetaZ by zetaZloc in S/R seaice_lsr_rhsu/v; only |
| 25 |
|
relevant if SEAICEuseStrImpCpl=T, and even then it does not seem to have |
| 26 |
|
an effect. |
| 27 |
|
o pkg/seaice: add two new parameters |
| 28 |
|
- SEAICEuseLSR to simplify the logic in seaice_dynsolver (this |
| 29 |
|
parameter is not in any namelist) |
| 30 |
|
- SEAICEusrPicardAsPrecon to allow s/r seaice_lsr to be used as |
| 31 |
|
a preconditioner for the non-linear Newton iteration of the JFNK |
| 32 |
|
solver |
| 33 |
|
o pkg/seaice/seaice_lsr.F: replace 1./SEAICE_deltaTdyn with recip_deltaT, |
| 34 |
|
affects some experiments at truncation level: |
| 35 |
|
- global_ocean.cs32x15.icedyn (11 digits), |
| 36 |
|
- global_ocean.cs32x15.seaice (11 digits), |
| 37 |
|
- seaice_itd (12 digits), seaice_itd.lipscomb07 (12 digits) -> update |
| 38 |
|
- also update global_ocean.cs32x15 adjoint and tangent-linear results |
| 39 |
|
o pkg/salt_plume & pkg/thsice: |
| 40 |
|
- for coupled set-up with pkg/thsice in ATM and pkg/salt_plume in OCN, |
| 41 |
|
connect salt_plume flux to ATM-OCN coupling pkgs interface: |
| 42 |
|
a) in ATM: also cumulate salt-plume flux in S/R THSICE_SALT_PLUME (for |
| 43 |
|
export to coupler interface) even without compiling pkg/salt_plume ; |
| 44 |
|
b) in OCN: apply coupler salt-plume flux to pkg/salt_plume field |
| 45 |
|
(saltPlumeFlux) just before exch call, in S/R SALT_PLUME_DO_EXCH. |
| 46 |
|
o pkg/dic: |
| 47 |
|
- use the updated seaice fraction (from seaice pkgs) for DIC forcing |
| 48 |
|
instead of the previous time-step value ; done in new S/R DIC_FIELDS_UPDATE |
| 49 |
|
called from the top of dic_biotic_forcing.F (instead of in DIC_FIELDS_LOAD); |
| 50 |
|
- move also update with values from ATM-OCN Coupler in the same new S/R |
| 51 |
|
(previously done in ocn_apply_import.F). |
| 52 |
|
o tools/adjoint_options: |
| 53 |
|
- specify explicitly "-f77" or "-f90" in default ad-optfile since TAF (default) |
| 54 |
|
has been changed on Jan 14, 2016. |
| 55 |
|
o pkg/ggl90: |
| 56 |
|
- add alternative discretisation of vertical shear (off by default), to |
| 57 |
|
compute the mean (@ grid-cell center) of vertical shear compon instead of |
| 58 |
|
vertical shear of mean flow (@ grid-cell center). |
| 59 |
|
- add correct parenthesis in ALLOW_GGL90_SMOOTH code to get the same truncated |
| 60 |
|
results on adjacent faces of cubed-sphere grids; this fix the restart. |
| 61 |
|
o pkg/atm_compon_interf: |
| 62 |
|
- skip time-ave over cpl_atmSendFrq of Sea-Level Pressure (atmSLPr) and just |
| 63 |
|
pass last value, stored directly before export to CPL ; |
| 64 |
|
Note: this affects results of Coupled set-up where OCN uses useImportSLP=T |
| 65 |
|
- remove atmSLPr from pickup files (no longer needed for restart) |
| 66 |
|
|
| 67 |
|
checkpoint65s (2016/01/13) |
| 68 |
|
o eesupp & genmake2: |
| 69 |
|
- first check for Function ETIME and, if it fails, then check for subroutine |
| 70 |
|
ETIME and consistent output; use the appropriate version in timers.F |
| 71 |
|
o model/src: |
| 72 |
|
- to check restart test, add option to apply EXCH to U,V earlier, just before |
| 73 |
|
integr_continuity; turned off by default. |
| 74 |
|
o pkg/dic: |
| 75 |
|
- remove CPP option USE_ATMOSCO2 (not needed) |
| 76 |
|
o tools/tst_2+2: |
| 77 |
|
- try to ignore difference between +0 and -0 in binary file, causing test |
| 78 |
|
to fail for no good reason. |
| 79 |
|
o eesupp & genmake2: |
| 80 |
|
- put back modif from Jinxuan Zhu - change etime fct call to S/R call - |
| 81 |
|
but also check (in genmake2) that test runs and returns meaningful results. |
| 82 |
|
o coupler pkgs: |
| 83 |
|
Finish implementation of optionally-exchanged variable switches (started on |
| 84 |
|
Nov 12) with addition of few more exchanged fields, including 2-way exchange |
| 85 |
|
of pkg/thsice vars to allow to use seaice dynamics (from pkg/seaice) & seaice |
| 86 |
|
advection in OCN component; this requires sequential coupling. |
| 87 |
|
coupler (pkg/atm_ocn_coupler): |
| 88 |
|
- add argument msgUnit and iter number to CPL_SEND/RECV_ATM/OCN_FIELDS |
| 89 |
|
routines (useful for debug). |
| 90 |
|
- only export/import optionally exchanged fields (RunOff, seaice, Salt-Plume |
| 91 |
|
flux, DIC fields) if corresponding switch is ON. |
| 92 |
|
- add option to export RunOff to OCN (to use for DIC river input) |
| 93 |
|
- add 2-way thSIce vars exchange; check for consistent cpl_sequential setting. |
| 94 |
|
components (pkg/atm_ & ocn_compon_interf): |
| 95 |
|
- to save memory space in ATM & OCN comp: add some #ifdef ALLOW_${PKG} around |
| 96 |
|
optionally exchanged fields; add a check & stop in cpl_readparms.F. |
| 97 |
|
ATMCPL.h: ALLOW_LAND for RunOff, ALLOW_THSICE for seaice, ALLOW_AIM for CO2; |
| 98 |
|
OCNCPL.h: ALLOW_SALT_PLUME, ALLOW_DIC for DIC, ALLOW_THSICE for 2-W thsice. |
| 99 |
|
- rename: fracIce -> sIceFrac_cpl (<- match coupler var name) |
| 100 |
|
+ in ATM: fracIceTime -> sIceFracTime ; seaIceTime -> sIceMassTime. |
| 101 |
|
- add 2-way thSIce vars exchange ; add Salt-Plume flux ; add RunOff in OCN. |
| 102 |
|
- hold in common block coupler optionally-exchanged variable switch |
| 103 |
|
and set default "useImportXXX" accordingly; + check & stop if inconsistent. |
| 104 |
|
- add diagnostics for all imported fields from coupler (in both OCN & ATM); |
| 105 |
|
this is intended to replace the OCN cplFldtave.* output files (not updated |
| 106 |
|
to contain new additional exchanged fields). |
| 107 |
|
o pkg/dic: |
| 108 |
|
- fix an other bug (also from 2014/08/30) in dic_atmos.F, in dic_int1=3 code. |
| 109 |
|
o pkg/thsice: |
| 110 |
|
- when coupled to ATM (useCoupler) with thSIce_skipThermo=T, skip reading |
| 111 |
|
and writing of thsice pickup files. |
| 112 |
|
o pkg/atm_compon_interf: |
| 113 |
|
- move all field individual export-call out of cpl_export_import_data.F |
| 114 |
|
into new S/R ATM_EXPORT_FIELDS |
| 115 |
|
o pkg/atm_compon_interf: |
| 116 |
|
- store state variables (for now, only thSIce fields) just before beeing |
| 117 |
|
exported (in cpl_export_import_data.F, like on ocean side) instead of from |
| 118 |
|
atm_store_my_data.F like all exported fluxes. |
| 119 |
|
- new version of pickup_cpl: write field names into meta file and read meta |
| 120 |
|
file to find field postion in pickup. Also only write/read fields meeded for |
| 121 |
|
restart (same logic as main pickup file ; also using pickupStrictlyMatch) |
| 122 |
|
o pkg/dic: |
| 123 |
|
- fix bug (added on 2014/08/30) in dic_atmos.F for the case dic_int1= 2 or 3. |
| 124 |
|
o pkg/seaice: add new parameter SEAICEpressReplFac to choose between |
| 125 |
|
pressure replacement method (=1., default) or original Hibler (1979) (=0.) |
| 126 |
|
value can be from domain [0,1] to average between the two methods |
| 127 |
|
o pkg/atm_phys: |
| 128 |
|
- allow to provide 2-d map of albedo and mixed-layer depth from files (but |
| 129 |
|
hold fixed in time). |
| 130 |
|
|
| 131 |
|
checkpoint65r (2015/12/21) |
| 132 |
|
o pkg/seaice: |
| 133 |
|
fix bug in tridiagonal solvers for SEAICE_OLx/y>0, this affects the JFNK |
| 134 |
|
solver (offline.dyn_jfnk) and global_ocean.cs32x15.seaice -> updated |
| 135 |
|
o pkg/seaice: |
| 136 |
|
- fix a bug in computing areaS in (J-J -> J-1) that affects runs |
| 137 |
|
when SEAICEscaleSurfStress=.TRUE., |
| 138 |
|
- apply areaW and areaS in S/R seaice_calc_lhs |
| 139 |
|
o pkg/atm_phys: |
| 140 |
|
- add diagnostics for 3-D, Nr+1 levels diagnostics of radiative fluxes |
| 141 |
|
(shortwave & longwave, net & downward) and diagnostics of temp. tendency |
| 142 |
|
for individual contribution (convection, LSC cond, radiation) |
| 143 |
|
- add Ruth Geen etal, GRL 2016, radiation optionis (see paper supp. |
| 144 |
|
information); activated with: solar_exponent=0. and wv_exponent=-1. |
| 145 |
|
o pkg/ggl90: |
| 146 |
|
- within IDEMIX add upper limit for horizontal group velocity to |
| 147 |
|
avoid instability problems in the explicit horizontal diffusion term for |
| 148 |
|
long time steps or high resolution |
| 149 |
|
o pkg/diagnostics: |
| 150 |
|
- simplify setting of local-array size in diagstats_output.F: uses "numLevels" |
| 151 |
|
(from DIAGNOSTICS_SIZE.h) to allow to write out more than Nr levels. |
| 152 |
|
o testreport: |
| 153 |
|
- remove "-v" (verbose option) in default mpirun command. |
| 154 |
|
o pkg/seaice/seaice_growth: |
| 155 |
|
- catch potential division by zero in ITD code, does not change |
| 156 |
|
verification results |
| 157 |
|
o pkg/ecco: |
| 158 |
|
- retire old codes to the Attic. The calls to retired functions are |
| 159 |
|
left in a commented form in case they ever need to be re-activated. |
| 160 |
|
- ecco_check.F: issue warning if any old CPP option is |
| 161 |
|
defined that corresponds to retired codes -- then stop. |
| 162 |
|
o pkg/atm_compon_interf: |
| 163 |
|
- change seaice fraction (use for OCN dic surf.flux) stored and sent to |
| 164 |
|
coupler: use last time-step value instead of average over cplSendFrq_iter |
| 165 |
|
time-step, similar to what is done for seaIceMass. |
| 166 |
|
This will affect results (pkg/dic) of coupled ATM-OCN+dic set-up. |
| 167 |
|
o eesupp & genmake2: |
| 168 |
|
- from Jinxuan Zhu: change etime fct call to S/R call (fix gfortran ETIME); |
| 169 |
|
<= remove until a fix is found for some ifort version on ia32 machine. |
| 170 |
|
o pkg/seaice/seaice_evp: |
| 171 |
|
- refine residual computation |
| 172 |
|
|
| 173 |
|
checkpoint65q (2015/11/18) |
| 174 |
|
o pkg/seaice/seaice_evp: |
| 175 |
|
- fix residual computations so that they no longer affect the actual |
| 176 |
|
simulation code |
| 177 |
|
- fix seaice_readparams to be able to actually set new parameters |
| 178 |
|
SEAICEaEVPcStar and SEAICEaEVPalphaMin, also fix logic a little |
| 179 |
|
o model/src: |
| 180 |
|
- fix wrong truncation (single prec) from "FLOAT" function in expressions |
| 181 |
|
like: deltaT*FLOAT(myIter) when myIter is large: |
| 182 |
|
a) replaced by DFLOAT (if not seen by TAF); |
| 183 |
|
b) remove "FLOAT" and use implicit type conversion (if seen by TAF). |
| 184 |
|
o pkg/aim_v23: |
| 185 |
|
- write/read the expected CO2 pickups (from S/R AIM_DO_CO2): fix reading |
| 186 |
|
when pickupSuff is set ; do write rolling pickups + at the end of the run. |
| 187 |
|
o pkg/atm_ocn_coupler + ocn_compon_interf: |
| 188 |
|
- in prep for using pkg/seaice Dyn in coupled set-up, implement sequential |
| 189 |
|
coupling option-1: ATM -(cpl)-> OCN and then OCN -(cpl)-> ATM ; |
| 190 |
|
add corresponding call in DO_OCEANIC_PHYS. |
| 191 |
|
o coupler pkgs: |
| 192 |
|
Start a new way to set-up optionally-exchanged variables (previously, only |
| 193 |
|
there for exchange_DIC): switch on/off from 1 place, at run-time, in coupler |
| 194 |
|
data.cpl (new) instead of at compile time in ATMIDS.h & OCNIDS.h ; |
| 195 |
|
send switch to ATM & OCN compon before reading their own "data.cpl" (so that |
| 196 |
|
compononent param default can be set accordingly). |
| 197 |
|
Note: this new switch (cpl_exchange_[xxx]) is not a per variable switch |
| 198 |
|
but rather applies to a group a fields (going both ways like for DIC) |
| 199 |
|
and corresponds to an option of coupling config. |
| 200 |
|
coupler (pkg/atm_ocn_coupler): |
| 201 |
|
- change parameter file from "data" to "data.cpl" (in cpl_read_params.F), |
| 202 |
|
and read namelist directly (without eessup S/R calls) instead of |
| 203 |
|
in the very un-flexible old way, 1 param per line, in fixed order; |
| 204 |
|
- add error-counter to track errors (e.g., missing "data.cpl") and to stop |
| 205 |
|
cleanly in CPL_CHECK_CPLCONFIG |
| 206 |
|
- add new coupling-exchange selectors (some not yet active) including |
| 207 |
|
cpl_exchange_DIC (to set to 3 to replace atmCpl_exchange_DIC=T in ATMIDS.h |
| 208 |
|
and ocnCpl_exchange_DIC=T in OCNIDS.h); |
| 209 |
|
- add 2 new S/R (CPL_SEND_ATM/OCN_CPLPARMS) to send coupler-params to each |
| 210 |
|
component ; called directly from coupler.F, after comp. registrations; |
| 211 |
|
- rename few variable-ID names (in ATMIDS.h & OCNIDS.h): |
| 212 |
|
atm/ocnSeaIceName --> atm/ocnSIceMassName ; atm/ocnFCO2Name -> |
| 213 |
|
-> atm/ocnCO2FluxName ; atm/ocnWSpdName --> atm/ocnWindSpdName |
| 214 |
|
components (pkg/atm_ & ocn_compon_interf): |
| 215 |
|
- add ATM/OCN_CPL_OPTIONS.h in component pkgs and include the corresponding |
| 216 |
|
CPP header file in all pkg source files. |
| 217 |
|
- move checking of comp. configs out of cpl_exch_configs.F into new S/R |
| 218 |
|
ATM/OCN_CHECK_CPLCONFIG; |
| 219 |
|
- add error-counter to track errors (e.g., in "data.cpl") and to stop |
| 220 |
|
cleanly in ATM/OCN_CHECK_CPLCONFIG; |
| 221 |
|
- add new S/R CPL_IMPORT_CPLPARMS to import coupling-exchange selectors |
| 222 |
|
from coupler ; called from initialise_fixed.F, after PACKAGES_BOOT and |
| 223 |
|
before PACKAGES_READPARMS; |
| 224 |
|
verification/cpl_aim+ocn/: |
| 225 |
|
- update ATMIDS.h & OCNIDS.h (in shared_code) and use new parameter file |
| 226 |
|
"data.cpl" for coupler (in input_cpl). |
| 227 |
|
o eesupp/inc: |
| 228 |
|
- change the MPI-global-sum default (in CPP_EEOPTIONS.h) to: |
| 229 |
|
#define GLOBAL_SUM_ORDER_TILES (only affects MPI built of global_sum_tile.F) |
| 230 |
|
to always cumulate tile local-sum in the same order (still uses |
| 231 |
|
MPI_Allreduce like standard code). |
| 232 |
|
- in verification experiment code, remove local CPP_EEOPTIONS.h_mpi files |
| 233 |
|
that are now identical to standard version (in eesupp/inc). |
| 234 |
|
o pkg/shelfice: |
| 235 |
|
- add SHELFICETransCoeffTFile for initializing spatially varying |
| 236 |
|
heat and salt transfer coefficients |
| 237 |
|
o pkg/atm_compon_interf & ocn_compon_interf: |
| 238 |
|
- combine into 1 (new) S/R CPL_EXPORT_IMPORT_DATA the two previous |
| 239 |
|
S/R CPL_EXPORT_MY_DATA & CPL_IMPORT_EXTERNAL_DATA |
| 240 |
|
- move bi,bj loops inside S/R OCN_STORE_MY_DATA |
| 241 |
|
o pkg/atm_ocn_coupler: |
| 242 |
|
- initialise to zero ATMVARS.h & OCNVARS.h arrays. |
| 243 |
|
o pkg/rbcs: |
| 244 |
|
- select RBCS-mask for Temp & Salt the same way as for pTracer ; this allows |
| 245 |
|
to compile with maskLEN=1 when all tracers use the same RBCS-mask. |
| 246 |
|
- add a short report (to STDOUT) about RBCS-mask settings. |
| 247 |
|
|
| 248 |
|
checkpoint65p (2015/10/23) |
| 249 |
|
o pkg/seaice/seaice_evp.F: fix store directives and key computations, since |
| 250 |
|
there is no adjoint test for EVP, we don't know what it does to the adjoint |
| 251 |
|
solution, but at least there are no more recomputation warnings left |
| 252 |
|
o pkg/seaice/seaice_evp.F: slight modification of averaging so that EVP stands |
| 253 |
|
a chance of giving the same results as a fully converged VP (JFNK) solver |
| 254 |
|
o tools/adjoint_options/adjoint_diva: |
| 255 |
|
- fix bug (typo) in Tang-Linear FLAGS setting for MPI + DIVA |
| 256 |
|
o pkg/streamice: |
| 257 |
|
- spotted bug in streamice_driving_stress using dyG where should have been dxG |
| 258 |
|
o pkg/exf: |
| 259 |
|
- remove STOP statement that prevented to use wind or (wind-)stress |
| 260 |
|
interpolation for constant-in-time forcing fields (wind/stress period = 0). |
| 261 |
|
o pkg/seaice: |
| 262 |
|
- replace SEAICE_EVP_USE_ADAPTATION by run time parameters, requires |
| 263 |
|
5 additionals 2D fields and 6 new store directives |
| 264 |
|
- change logic in seaice_readparms.F: EVP code can now be turned on in |
| 265 |
|
various ways, setting SEAICE_deltaTev is no longer required, |
| 266 |
|
alternatively set SEAICE_evpAlpha/Beta, SEAICEuseEVPstar, SEAICEuseEVPrev, |
| 267 |
|
or SEAICEaEVPcoeff |
| 268 |
|
o pkg/shelfice/shelfice_thermodynamics.F |
| 269 |
|
- redefining uLoc/vLoc to u/v, instead of 2u/v |
| 270 |
|
o pkg/seaice/seaice_evp.F: introduce "adaptive EVP" by Madlen Kimmritz |
| 271 |
|
- this is an intermediate state of a truely converging EVP that has |
| 272 |
|
the potential of outperforming implicit VP methods, I do not recommend |
| 273 |
|
using it yet. |
| 274 |
|
- for now, "adaptive EVP" is turned on by setting SEAICE_EVP_USE_ADAPTATION |
| 275 |
|
and all parameters are hardwired |
| 276 |
|
- for convergence it still requires very smooth regularisations, that means, |
| 277 |
|
if SEAICE_DELTA_SMOOTHREG is defined, deltaCreg=deltaC+deltaMin, which is |
| 278 |
|
different from the VP case with deltaCreg=sqrt(deltaC**2+deltaMin**2); also |
| 279 |
|
specific averaging appears to be required, still subject to tuning |
| 280 |
|
- still needs cleaning and I need to turn parameter into runtime parameters |
| 281 |
|
- add code to compute and print residuals of iteration, compile with |
| 282 |
|
defining ALLOW_SEAICE_EVP_RESIDUAL to enable it |
| 283 |
|
o pkg/seaice/seaice_evp.F: fix a bug (factor of four in front of e12Csq) |
| 284 |
|
that entered with r1.54 on Sep04, 2015, -> update results |
| 285 |
|
|
| 286 |
|
checkpoint65o (2015/09/14) |
| 287 |
|
o pkg/mom_vecinv & mom_common: |
| 288 |
|
- add deep-atmosphere geometry factor in mom_calc_visc.F + in routines |
| 289 |
|
called from there. |
| 290 |
|
- add anelastic and deep-atmosphere geometry factor in pkg/mom_vecinv ; this |
| 291 |
|
allows to use Vector-Invariant form in deep atmos and anelastic formulation |
| 292 |
|
o pkg/seaice: introduce fast ice parameterisation following Itkin et al. (2015) |
| 293 |
|
and Koenig-Beatty+Holland (2012) |
| 294 |
|
- two new parameters SEAICE_tensilFac and SEAICE_tensilDepth |
| 295 |
|
- global 2D field tensileStrength, computed in seaice_calc_ice_strength |
| 296 |
|
- requires an extra input argument for S/R seaice_calc_viscosities |
| 297 |
|
- tensileStrength's depth dependence is different from Itkin et al. (2015) |
| 298 |
|
(to be changed and played with, for now exponential) |
| 299 |
|
o pkg/seaice: introduce new parameter SEAICEscaleSurfStress (default = .FALSE.) |
| 300 |
|
- if TRUE scale ice-ocean and ice-atmosphere stress acting on sea ice |
| 301 |
|
by AREA according to Connelly et al. (2004), JPO. |
| 302 |
|
- For EVP and the LSR solver, the implicit part of the drag term is |
| 303 |
|
the only non-zero term in the denominator when there is no ice (which |
| 304 |
|
prevented running the model with zero ice-ocean drag). If |
| 305 |
|
SEAICEscaleSurfStress = .TRUE., this results in a division by zero |
| 306 |
|
(or zero main diagonals BU/BV) which need to be caught. A practical |
| 307 |
|
consequence is that for open water, the momentum equation reduce |
| 308 |
|
to trivially 0 = 0 + 0 (for EVP). For LSR, BU/BV are reset to 1, if |
| 309 |
|
they happen to be zero, often leading to a non-zero solution over |
| 310 |
|
open water. |
| 311 |
|
o pkg/seaice: revise EVP code (seaice_evp.F) in preparation for a more |
| 312 |
|
efficient EVP method. |
| 313 |
|
- introduce area-weighted averages for e12 (analogously to |
| 314 |
|
seaice_calc_viscosities.F); this totally changes lab_sea.hb87 (2 digits |
| 315 |
|
remain, but also changing the order of terms in the previous average |
| 316 |
|
changed the results as much) -> update experiment |
| 317 |
|
- add SEAICE_DELTA_SMOOTHREG code |
| 318 |
|
- disentangle computation of zetaC/deltaC and zetaZ/deltaZ, also for |
| 319 |
|
TEM-code |
| 320 |
|
- adjust loop bounds so that only required fields are computed |
| 321 |
|
- add new averaging code for zetaZ/deltaZ (again similar to |
| 322 |
|
seaice_calc_viscosities.F), inactive by default for now (will change |
| 323 |
|
results again), but will replace old code eventually |
| 324 |
|
o eesupp/src: |
| 325 |
|
- add an other alternative for multi-proc GLOBAL_SUM (in global_sum_tile.F) |
| 326 |
|
to always cumulate tile local-sum in the same order |
| 327 |
|
(option #define GLOBAL_SUM_ORDER_TILES in CPP_EEOPTIONS.h). |
| 328 |
|
Still uses MPI_Allreduce (like standard code) but on array of tiles |
| 329 |
|
and should be faster than GLOBAL_SUM_SEND_RECV code. |
| 330 |
|
- switch almost all verification experiments that use GLOBAL_SUM_SEND_RECV |
| 331 |
|
(defined in CPP_EEOPTIONS.h_mpi) to GLOBAL_SUM_ORDER_TILES instead. |
| 332 |
|
- add CPP_EEOPTIONS.h_mpi with GLOBAL_SUM_ORDER_TILES defined in 14 more exp.: |
| 333 |
|
aim.5l_cs, cheapAML_box, deep_anelastic, dome, fizhi-cs-32x32x40, |
| 334 |
|
fizhi-cs-aqualev20, fizhi-gridalt-hs, global_ocean_ebm, matrix_example, |
| 335 |
|
natl_box, solid-body.cs-32x32x1, tutorial_advection_in_gyre, |
| 336 |
|
tutorial_baroclinic_gyre and tutorial_held_suarez_cs. |
| 337 |
|
|
| 338 |
|
checkpoint65n (2015/07/29) |
| 339 |
|
o pkg/thsice: |
| 340 |
|
- add a warning when used with pkg/aim but without updating SST |
| 341 |
|
(stepFwd_oceMxL=F & useCoupler=F & tauRelax_MxL<>-1 ). |
| 342 |
|
o pkg/offline: |
| 343 |
|
- change where Offline-Fields are loaded, from load_fields_driver.F to |
| 344 |
|
directly within FORWARD_STEP, just after updating iter number and time; |
| 345 |
|
- move setting of GM and KPP diffusivity out of OFFLINE_FIELDS_LOAD into new |
| 346 |
|
S/R OFFLINE_GET_DIFFUS that is called towards the end of DO_OCEANIC_PHYS; |
| 347 |
|
These changes allow: |
| 348 |
|
a) to recover exactly on-line results using deltaToffline=deltaTClock, |
| 349 |
|
tested (using offlineIter0= nIter0 -1 and offlineTimeOffset= -deltaT/2) |
| 350 |
|
both with Lin & Non-Lin Free-Surf, both with staggerTimeStep & synchronous |
| 351 |
|
time-step, both with GM & KPP files and with recomputing these fields; |
| 352 |
|
b) no needs to switch on staggerTimeStep anymore; |
| 353 |
|
c) simplify rescaling of offline horiz. velocity with Non-Lin Free-Surf. |
| 354 |
|
- update output of experiment tutorial_dic_adjoffline (only 3 digits left). |
| 355 |
|
o pkg/offline: |
| 356 |
|
- change (rarely used) offset-iteration parameter to an offset time (in s): |
| 357 |
|
can easily recover previous offset-iter effect by setting offset-time |
| 358 |
|
(offlineTimeOffset) to offset-iter x deltaToffline; |
| 359 |
|
- do not turn-off Temp & Salinity surface relaxation when using KPP (used |
| 360 |
|
in KPP as part of surface buoyancy forcing) in offline run. |
| 361 |
|
- fix ptracers KPP non-local term in the case where KPP is re-computed. |
| 362 |
|
o OpenAD |
| 363 |
|
- Adding preliminary ggl90 test to OpenAD verification. |
| 364 |
|
Config. with ggl90 currently diverges after 64 time steps |
| 365 |
|
so will need more work to become more stable |
| 366 |
|
- Adding preliminary kpp test to OpenAD verification. |
| 367 |
|
required small tweakes in kpp_routines.F |
| 368 |
|
- Merge and update from Krishna Narayanan's contrib area: |
| 369 |
|
* Add one variable for preliminary OpenAD DIVA |
| 370 |
|
* Add replacement time-stepping for OpenAD DIVA |
| 371 |
|
* genmake2 flag -diva (but only for OpenAD; preliminary) |
| 372 |
|
* required modifs for OAD_support |
| 373 |
|
successful preliminary test for tutorial_global_oce_biogeo/ |
| 374 |
|
o pkg/seaice: |
| 375 |
|
- initialise deltaC in seaice_lsr, cosmetic changes in seaice_lsr |
| 376 |
|
- add cpp-flag SEAICE_DELTA_SMOOTHREG for option of regularising |
| 377 |
|
delta with a smooth function in s/r seaice_calc_viscosity |
| 378 |
|
(no effect on EVP) |
| 379 |
|
- rename local suffix "sqr" into "sq" for more consistent variable names |
| 380 |
|
o pkg/seaice: |
| 381 |
|
add new parameter SEAICE_deltaMin that is used ***only*** |
| 382 |
|
for regularizing Delta (and nothing else, like the parameter |
| 383 |
|
which is also used for all sorts of things SEAICE_EPS). |
| 384 |
|
Defaults to SEAICE_EPS for backward compatibility only. |
| 385 |
|
o pkg/seaice: replace erroneously storing e12Csqr on the wrong tape with |
| 386 |
|
the wrong key by inialising it before use -> fixes a recomputation |
| 387 |
|
problem but does not fix the gradient |
| 388 |
|
o pkg/layers: |
| 389 |
|
- move resetting of flux arrays to outside the iLa loop |
| 390 |
|
(this caused layers_diapycnal to not work with more than one layer coord.) |
| 391 |
|
- fix bug in calculation of layers_recip_delta in layers_init_fixed.F |
| 392 |
|
o pkg/layers: |
| 393 |
|
- remove from common block 3D layers fields (used to diagnose transport |
| 394 |
|
in tracer coords) and declare them as local arrays in layers_calc.F |
| 395 |
|
o pkg/seaice: |
| 396 |
|
- Rename SEAICE_VECTORIZE_LSR_ZEBRA to SEAICE_LSR_ZEBRA and define it |
| 397 |
|
in lab_sea forward experiment (changes results -> updated) |
| 398 |
|
- update adjoint experiment lab_sea after cleaning up seaice_lsr_tridiagu/v |
| 399 |
|
o pkg/seaice: modify seaice_lsr.F in order to improve the gradient |
| 400 |
|
computations; for SEAICE_VECTORIZE_LSR |
| 401 |
|
- move the loops over which the tridiagonal solvers (seaice_lsr_tridiagu/v) |
| 402 |
|
sweep to the outside, |
| 403 |
|
- remove store directives and add "CADJ loop sequential" directives |
| 404 |
|
in analogy to model/src/solve_tridiagonal.F |
| 405 |
|
- replace many "#ifdef SEAICE_VECTORIZE_LSR_ZEBRA" by variable loop |
| 406 |
|
boundaries to yield more readable code. This has also the charming side |
| 407 |
|
effect that your can use SEAICE_VECTORIZE_LSR_ZEBRA without |
| 408 |
|
SEAICE_VECTORIZE_LSR (i.e. adjoint without recomputations in |
| 409 |
|
seaice_lsr_tridiagu/v now requires either of these flags, vectorization |
| 410 |
|
still requires SEAICE_VECTORIZE_LSR with SEAICE_VECTORIZE_LSR_ZEBRA as |
| 411 |
|
an additional option) |
| 412 |
|
The tridiagonal solvers are now completely analoguous to solve_tridiagonal.F |
| 413 |
|
but the adjoint gradients (which are affected by this change) still explode. |
| 414 |
|
o coupler (atm & ocn _compon_interf): |
| 415 |
|
- avoid using Exch2 IO mapping if not suitable. |
| 416 |
|
|
| 417 |
|
checkpoint65m (2015/06/15) |
| 418 |
|
o pkg/layers: |
| 419 |
|
- clean-up old parameters setting (from Sep 2012): |
| 420 |
|
+ remove old parameters setting from common block (declared locally); |
| 421 |
|
+ stop if any old parameter is mixed with new parameters; |
| 422 |
|
+ check for valid or consistent parameter setting (e.g., layers_name) |
| 423 |
|
- check that layers_bounds for Pot.Density are in the right range; |
| 424 |
|
this is supposed to catch un-updated setting after recently shifting |
| 425 |
|
potential density field "prho" by -1000 (now in "sigma" scale). |
| 426 |
o pkg/layers: |
o pkg/layers: |
| 427 |
- Added missing call to layers_fill for keeping track of tottend in |
- Added missing call to layers_fill for keeping track of tottend in |
| 428 |
pkg/diagnostics/diagnostics_fill_state.F |
pkg/diagnostics/diagnostics_fill_state.F |
| 440 |
thermodynamics.F, impldiff.F) and gad to save the necessary fields |
thermodynamics.F, impldiff.F) and gad to save the necessary fields |
| 441 |
for layers_thermodynamics calculations |
for layers_thermodynamics calculations |
| 442 |
o pkg/ptracers: |
o pkg/ptracers: |
| 443 |
- add capability of stepping a passive tracer forward |
- add capability of stepping a passive tracer forward without advection |
| 444 |
without advection by setting PTRACERS_advScheme to 0. |
by setting PTRACERS_advScheme to 0. |
| 445 |
o pkg/diagnostics: |
o pkg/diagnostics: |
| 446 |
- add few (threads) _BARRIER around any update of "diag_pkgStatus". |
- add few (threads) _BARRIER around any update of "diag_pkgStatus". |
| 447 |
o pkg/layers: |
o pkg/layers: |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch