4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
5 |
============================== |
============================== |
6 |
|
|
7 |
|
o pkg/cal: |
8 |
|
- move calendar setting & summary (previously in cal_readparams.F) into new |
9 |
|
S/R CAL_INIT_FIXED to be called later in the initialisation process, once |
10 |
|
length of the simulation is set (after SET_PARMS call); This is an attempt |
11 |
|
to fix Pb when, e.g.: pkg/longstep, modifies the length of the run. |
12 |
|
- For safety: keep a record of pkg/cal parameter settings status |
13 |
|
to allow to STOP if one pkg/cal S/R is called too early. |
14 |
|
|
15 |
|
checkpoint63l (2012/04/04) |
16 |
|
o pkg/bbl: |
17 |
|
- fix call to FIND_RHO_2D in bbl_calc_rho.F (case multi-tiles per proc); |
18 |
|
update global_with_exf output.yearly.txt file. |
19 |
|
o pkg/cal: |
20 |
|
- start to clean-up cal_set.F ; fix for the case where baseTime is non-zero. |
21 |
|
- remove 2 unsused and very short (1 line only) functions ; |
22 |
|
- remove unsused subroutines from AD-Diff list. |
23 |
|
o model/src: |
24 |
|
- start to implement clean wait to stop in initialization S/R (config_check.F, |
25 |
|
packages_check.F, ini_parms.F and pkg/obcs/obcs_check.F + pkg/mypackage) |
26 |
|
o folowing PRINT_MESSAGE changes: |
27 |
|
- remove reminder multiple print to the same file (STDOUT) when multi-threads. |
28 |
|
o pkg/bbl: |
29 |
|
- modification so it can work with z* |
30 |
|
- replace deltaT with dTtracerLev(k) for asynchronous time stepping |
31 |
|
- global_with_exf/input.yearly verification; changes output.yearly.txt |
32 |
|
o eesupp/src: |
33 |
|
- PRINT_MESSAGE: use OpenMP critical instruction (similar to PRINT_ERROR); |
34 |
|
writing to unit 0 (if unit=errorMessageUnit) extended to the case myThid=1 |
35 |
|
- CHECK_THREADS: specific code if using OpenMP threading: improve checking of |
36 |
|
# of threads across processors (e.g., different env OMP_NUM_THREADS setting) |
37 |
|
- improve special preliminary call to EESET_PARMS (if eedata is read 2 times). |
38 |
|
- when using MPI, always set usingMPI=T ; set default accordingly and stop |
39 |
|
if not consistent. |
40 |
|
- ALL_PROC_DIE: add a BARRIER after MPI_FINALIZE call ; skip this when thread |
41 |
|
argument is zero (in case called within single-thread section). |
42 |
|
o pkg/fizhi: |
43 |
|
clean-up turbulence cold-start switch: decided in fizhi_init_vars.F, stored |
44 |
|
in common bloc (fizhi_coms.h) and then passed as argument up to S/R TURBIO. |
45 |
|
o pkg/exch2: |
46 |
|
- add argument "updateCorners" to S/R exch2_get_uv_bounds (enable to |
47 |
|
activate EXCH_IGNORE_CORNERS in vector EXCH S/R). |
48 |
|
- switch 1rst EXCH call to IGNORE_CORNERS (instead of UPDATE_CORNERS) |
49 |
|
but keep the 2nd with UPDATE_CORNERS: this prevents overwriting good data |
50 |
|
with bad ones coming from a not-yet-updated halo. |
51 |
|
This should fix the problem when sNx <> sNy and GCD(sNx,sNy) < 2*Olx. |
52 |
|
o pkg/frazil: |
53 |
|
move FrzForcT diag from frazil_diagnostics_state.F to frazil_calc_rhs.F |
54 |
|
http://mitgcm.org/pipermail/mitgcm-devel/2012-March/005384.html |
55 |
|
o model/src/config_check.F: add a check for OLx>sNx/OLy>sNy |
56 |
|
o verification/lab_sea |
57 |
|
* #undef SEAICE_EXCLUDE_FOR_EXACT_AD_TESTING |
58 |
|
* #define ALLOW_SST_CONTROL & ALLOW_SST_CONTROL |
59 |
|
* nTimeSteps 3 -> 4 |
60 |
|
* changed reference output accordingly |
61 |
|
o pkg/ctrl: |
62 |
|
* fix ctrl_get_gen for xx_sst, xx_sss |
63 |
|
o pkg/fizhi: |
64 |
|
- fix 2 calls to S/R GRD2MSC (argument igrd) in update_earth_exports.F |
65 |
|
update results of exp. fizhi-cs-32x32x40 (fail with only 6 digits for cg2d) |
66 |
|
- by default, use new version of S/R GETPWHERE (#define TRY_NEW_GETPWHERE); |
67 |
|
fix few out-of-bounds indices in referencing vars (initialise igrd to 1); |
68 |
|
clean-up fizhi_turb.F (remove unused & uninitialised qdiaglsm array). |
69 |
|
o pkg/gmredi: |
70 |
|
- add diagnostic of Redi (main diagnonal) vertical diffusive flux of Temp |
71 |
|
o pkg/fizhi: |
72 |
|
- fix few initialisation and multi-tile Pb: aqualev & gridalt now pass the |
73 |
|
(old)aces open64 tuv test. |
74 |
|
o pkg/icefront: fix bug and reorder loop in icefront_init_varia.F |
75 |
|
o pkg/mom_vecinv: |
76 |
|
- fix 1 loop range in mom_vecinv.F to be like all the other loops which |
77 |
|
set/update gu,gv/Diss (fix Pb of uninitialised fVerUkp in admom_vecinv |
78 |
|
with non-lin free-surf). |
79 |
|
o model/src: move k-loop outside of i/j-loops when (re-)computing R_low |
80 |
|
in ini/update_masks_etc.F in order to avoid -O3 optimization problems with |
81 |
|
some compilers (gfortran 4.6.0 on MacOS) |
82 |
|
o momentum: |
83 |
|
- separate fVer?(:,:,kUp) & fVer?(:,:,kDown) in argument list of MOM_FLUXFORM |
84 |
|
& MOM_VECINV subroutines (to help TAF). This fixes Pb of uninitialised |
85 |
|
fVer?(kUp) in admom_fluxform with non-lin free-surf or bottom control. |
86 |
|
|
87 |
|
checkpoint63k (2012/03/17) |
88 |
|
o model/src: |
89 |
|
- selectAddFluid (ALLOW_ADDFLUID): add missing term in horizontal momentum |
90 |
|
flux-form (mom_fluxform.F) and vertical momentum (calc_gw.F) equations. |
91 |
|
- SOLVE_DIAGONAL: switch default to use adjoinable & vectorized version |
92 |
|
(#undef SOLVE_DIAGONAL_LOWMEMORY); update output of exp. front_relax.bvp. |
93 |
|
o model/src: solve_*diagonal.F |
94 |
|
- replace division with multiplication by inverse (similar to original version) |
95 |
|
update output of AD exp. global_ocean.90x40x15. |
96 |
|
- fix pkg/gmredi/gmredi_calc_psi_bvp.F & pkg/ggl90/ggl90_calc.F |
97 |
|
to enable the use of the vectorized & differentiable solve_tridiagonal.F |
98 |
|
- test exp. global_ocean.90x40x15/input_ad: switch to advection scheme 20 for |
99 |
|
vertical advection (to get solve_tridiagonal tested in AD mode). |
100 |
|
o pkg/seaice: ECCO-CLIVAR merge: |
101 |
|
1. seaice AREA relaxation |
102 |
|
* CPP option: SEAICE_ALLOW_AREA_RELAXATION |
103 |
|
* runtime flag: SEAICE_tauAreaObsRelax |
104 |
|
2. modified global mean cost imbalance code: CPP option |
105 |
|
ALLOW_SSH_GLOBMEAN_COST_CONTRIBUTION |
106 |
|
o another iteration on solve_*diagonal.F |
107 |
|
Make choices independent of AUTODIFF |
108 |
|
Default will be the adjoinable vectorized version. |
109 |
|
Alternative choices are: |
110 |
|
* non-adjoinable original version with low memory footprint: |
111 |
|
#define SOLVE_DIAGONAL_LOWMEMORY |
112 |
|
* still adjoinable but does not vectorize (k as inner-most loop): |
113 |
|
#define SOLVE_DIAGONAL_KINNER (keeping SOLVE_DIAGONAL_LOWMEMORY undef) |
114 |
|
o testreport |
115 |
|
- add option to use a home made wrapper around "make" |
116 |
|
- remove MITGCM_IEEE env. variable (no longer active) |
117 |
|
o pkg/shelfice |
118 |
|
- improve computation of friction velocity for velocity dependent |
119 |
|
exchange coefficients: include velocities into SHELFICEBoundaryLayer |
120 |
|
block to average velocities vertically |
121 |
|
- add TAF store directives for SHI_ALLOW_GAMMAFRICT |
122 |
|
o Modify diagonal solvers, now three versions (upon request): |
123 |
|
#ifndef SOLVE_DIAGONAL_AUTODIFF |
124 |
|
1. default version (dating back a while ago) |
125 |
|
#else /* def SOLVE_DIAGONAL_AUTODIFF */ |
126 |
|
# ifdef SOLVE_DIAGONAL_KINNER |
127 |
|
2. adjoinable version with k-innermost loop (G. Forget) |
128 |
|
# else |
129 |
|
3. adjoinable vectorized version of 2. |
130 |
|
# endif |
131 |
|
#endif /* SOLVE_DIAGONAL_AUTODIFF */ |
132 |
|
All flags are currently set within routines, but could/should be moved to CPP-options. |
133 |
|
Array bounds for 1. are as before, for 2.,3. cover the entire domain |
134 |
|
(could be changed for 1. too). |
135 |
|
o verification/lab_sea (fwd): |
136 |
|
- update the 4 fwd lab_sea test experiment: no longer SEAICE_GROWTH_LEGACY, |
137 |
|
use 7 categories (except hb87), new solve4temp params (except salt_plume), |
138 |
|
constant salinity (=4) or variable salinity (salt_plume) done through |
139 |
|
SItracer. |
140 |
|
o pkg/seaice: |
141 |
|
- rename parameters: SIsalFrac to SEAICE_saltFrac & SIsal0 to SEAICE_salt0; |
142 |
|
- remove old parameters ( SEAICE_freeze, SEAICEturbFluxFormula, SEAICE_gamma_t, |
143 |
|
SEAICE_gamma_t_frz, SEAICE_availHeatTaper & SEAICE_availHeatFracFrz) |
144 |
|
from SEAICE_PARMS.h ; still need to fix seaice_cost_sst.F (broken when |
145 |
|
SEAICE_freeze got retired) but already better than previously with |
146 |
|
SEAICE_freeze=123456.7 deg.C; |
147 |
|
- move setting of facOpenGrow/Melt from SEAICE_INIT_FIXED to SEAICE_READPARMS |
148 |
|
(safer multi-threaded setting); always set SEAICEuseEVP; |
149 |
|
- setting of ocean-ice turb. flux coeff: moved from SEAICE_CHECK to |
150 |
|
SEAICE_READPARMS & SEAICE_INIT_FIXED (this fixes wrong summary report); |
151 |
|
stop if multiple specifications for the same coeff; make sure default |
152 |
|
SEAICE_mcPheePiston is compatible with drF(1) & deltaT. |
153 |
|
o pkg/obcs: |
154 |
|
- allow to switch to upwind 1rst order advection scheme for the advective |
155 |
|
flux computation at the open-boundary; disabled for adjoint built. |
156 |
|
o pkg/ptracers: |
157 |
|
- rename PTRACERS_RESTART.h -> PTRACERS_START.h |
158 |
|
- rename ptracers_fields_blocking_exchanges.F -> ptracers_fields_blocking_exch.F |
159 |
|
- add parameter to decide when to start a tracer; use internal flag to |
160 |
|
know which tracer to step-forward; switch this internal flag at the top |
161 |
|
of forward_step.F (new S/R PTRACERS_SWITCH_ONOFF); for now, not enabled |
162 |
|
for adjoint built. |
163 |
|
o pkg/seaice: |
164 |
|
- add SEAICE_SIZE.h anywhere in the code where SEAICE.h is included ; |
165 |
|
- move MULTDIM declaration and setting from SEAICE.h to SEAICE_SIZE.h ; |
166 |
|
- keep experiment offline_exf_seaice truly single category (set MULTDIM=1). |
167 |
|
o pkg/ecco: |
168 |
|
- fix writing of output file DiagnosCost_ClimSalt/Theta |
169 |
|
and DiagnosCost_DriftSalt/Theta (if multi-tiles per proc). |
170 |
|
o pkg/seaice: |
171 |
|
- add missing initialisation of COR_ICE array in dynsolver.F |
172 |
|
- merging SEAICE_MULTICATEGORY define and undef cases. Now the same code is |
173 |
|
compiled either way, and is tested in all exps (multidim was only tested |
174 |
|
in lab_sea ad before). |
175 |
|
- cosmetic change to prepare for future addition of ITD. no change to results. |
176 |
|
no defaults change. |
177 |
|
- added run time param SEAICE_multDim (7 for SEAICE_MULTICATEGORY, 1 otherwise). |
178 |
|
- by lack of adequate verification experiments, I further used custom versions |
179 |
|
of global_ocean.cs32x15 to test multicat or not, with pickup or not. All |
180 |
|
should be correct. |
181 |
|
- avoid recomputations in ad. |
182 |
|
o array initialisation in adjoint mode: |
183 |
|
1) pkg/mom_common/mom_calc_ke.F: |
184 |
|
add full array initialisation of KE inside mom_calc_ke.F (ALLOW_AUTODIFF_TAMC): |
185 |
|
The S/R which calls MOM_CALC_KE initialises KE but in some recomputation |
186 |
|
part this initialisation is gone, resulting in Floating Point Exception |
187 |
|
(caught by open64 compiler with debug option). |
188 |
|
2) model/src/convective_weights.F |
189 |
|
set output convectCount over the full array range (instead of doing |
190 |
|
an update): this fix a similar issue when called within recomputation (adjoint) |
191 |
|
which could by-pass the initialisation (normally done in the caller S/R). |
192 |
|
o pkg/seaice: |
193 |
|
- final version of turbulent flux params, as agreed at MIT ice meet: |
194 |
|
SEAICE_mcPheePiston (default is ~ 0.000875) |
195 |
|
SEAICE_mcPheeTaper (default is 0.) |
196 |
|
SEAICE_frazilFrac (defualt is 1.) |
197 |
|
SEAICE_mcPheeStepFunc (default is false) |
198 |
|
- changes results at trunction error level |
199 |
|
- ifdef SEAICE_EXTERNAL_FLUXES, disallow SEAICE_waterAlbedo |
200 |
|
o added pkg/frazil: |
201 |
|
- see frazil_description.tex for details |
202 |
|
- verification experiment is global_with_exf/input.yearly; changes output.yearly.txt |
203 |
|
o pkg/shelfice/shelfice_thermodynamics.F |
204 |
|
- fix exponents for velocity dependent exchange coefficients: change |
205 |
|
from 2/3(=0) to 0.66666666666666667 |
206 |
|
- move computation of velocity dependent exchange coefficients into |
207 |
|
separate loop, so that it applies to both thermodynamics |
208 |
|
- move some divisions by constants out of loops |
209 |
|
|
210 |
|
checkpoint63j (2012/02/16) |
211 |
|
o pkg/seaice: |
212 |
|
- overhaul of seaice_summary.F after feb 2012 clean-up of seaice_growth.F etc. |
213 |
|
- SItracer: added run time params, and allow coupling to ocn salinity. |
214 |
|
o pkg/seaice: remove (almost) all divisions by constants within loops |
215 |
|
in seaice_growth.F and seaice_solve4temp.F, this changes (only) the cg2d |
216 |
|
results of global_ocean.cs32x15.seaice from 16 to 11 digits (but only |
217 |
|
those, caused by 1/QI -> recip_QI) -> update output.seaice.txt, |
218 |
|
output_adm.seaice_dynmix.txt and also lab_sea: output_adm.evp.txt |
219 |
|
o pkg/seaice : |
220 |
|
- simplify McPhee tapering, and allow to use it with SEAICEturbFluxFormula |
221 |
|
= 1 or 2, using newly added run time parameter SEAICE_availHeatTaper. |
222 |
|
- fix d_AREAbyATM, d_AREAbyOCN, d_AREAbyICE diags I broke in r1.148. |
223 |
|
- add permanent SItflux diag that corresponds to TFLUX but includes |
224 |
|
ice+snow. Hence SItflux-TFLUX should match the ice+snow heat budget. |
225 |
|
- allow activation/testing of a fix for suspected missing term in |
226 |
|
ocn-ice heat budget (to be confirmed). To test this, you want to |
227 |
|
undef SEAICE_DISABLE_HEATCONSFIX, and then set the run time param |
228 |
|
SEAICEheatConsFix to .TRUE. that also allows the 'SIaaflux' diagnostic. |
229 |
|
- no change in results, fwd or ad. |
230 |
|
o global_ocean.cs32x15/results: 3 changes following from the pkg/seaice commit below: |
231 |
|
- switch to variable freezing point. |
232 |
|
Results change : cg2d -> 1 digits; ad -> 3-6-6 digits. |
233 |
|
- bug fix : the sublimation term that was missing is now activated. |
234 |
|
Results change : cg2d -> 2 digits; ad -> 8-6-6 digits. |
235 |
|
- bug fix : flooding sign term for SIsal0.NE.0. |
236 |
|
Result change : cg2d -> 8 digits; ad -> 11-12-9 digits. |
237 |
|
o pkg/seaice: |
238 |
|
- unified freezing point treatment : the old SEAICE_VARIABLE_FREEZING_POINT |
239 |
|
is now the default, and the old default constant freezing point is recovered |
240 |
|
with SEAICE_tempFrz0 = -1.96, |
241 |
|
SEAICE_dTempFrz_dS = 0., |
242 |
|
- retire SEAICE_freeze (= the old way of specifycing the constant freezing point) |
243 |
|
- remove ALLOW_SEAICE_FLOODING brackets; run time switch is already there. |
244 |
|
- bug fix (thanks to M. Losch) : the sublimation term that was missing is now |
245 |
|
activated. To reproduce old results with this bug define SEAICE_DISABLE_SUBLIM. |
246 |
|
- bug fix (silly me) : flooding sign term (seaice_growth.F r 1.149line 1211) |
247 |
|
was wrong for SIsal0.NE.0. Changes cs32x15 results (see above item for detail). |
248 |
|
- bug fix (thanks to O. Jahn) : area loss for melting got messed up in |
249 |
|
seaice_growth r149 for legacy branch (no results change). |
250 |
|
- introduce SEAICE_CAP_SUBLIM : caps sublimation heat flux in solve4temp (code |
251 |
|
from I. Fenty). |
252 |
|
- results did not change except for global_ocean.cs32x15, mostly due to the |
253 |
|
switch to variable freezing point (see above item for detail). |
254 |
|
o pkg/thsice: |
255 |
|
- fix and simplify interface to salt_plume pkg ; diagnose directly amount of |
256 |
|
seawater freezing in thsice_calc_thickn (not used). |
257 |
|
o pkg/seaice: turbulent flux term specification via run time params. |
258 |
|
- replace CPP switches with run time switches to facilitate testing |
259 |
|
- retired CPPs : MCPHEE_OCEAN_ICE_HEAT_FLUX, |
260 |
|
GRADIENT_MIXED_LAYER_TURBULENCE_FACTOR |
261 |
|
- added runtime parameter : SEAICEturbFluxFormula |
262 |
|
- move Mcphee constants to SEAICE_PARAMS.h |
263 |
|
- seaice_readparms.F : read turbulent flux term params (SEAICEturbFluxFormula; |
264 |
|
SEAICE_availHeatFrac/Frz, or SEAICE_gamma_t/_frz if apply); |
265 |
|
remove seaice_summary call (moved to seaice_init_fixed) |
266 |
|
- seaice_init_fixed.F : set SEAICE_availHeatFrac/Frz depending on |
267 |
|
SEAICEturbFluxFormula, and seaice_summary call (moved from seaice_readparms) |
268 |
|
- seaice_summary.F : add missing report of read turbulent flux term set up |
269 |
|
and SITRACER. |
270 |
|
- seaice_check.F : check range; inform and stop if retired CPPs are used |
271 |
|
(incl. SEAICE_AGE). |
272 |
|
- seaice_growth.F : unified use of SEAICE_availHeatFrac/Frz, without |
273 |
|
CPP switch, same computation as before for the different options. |
274 |
|
- change results at the truncation level. |
275 |
|
o pkg/seaice: |
276 |
|
- remove un-used parameters LAD & MAX_TICE ; add new run-time parameters: |
277 |
|
useMaykutSatVapPoly (default=F), postSolvTempIter (default=2) and |
278 |
|
SEAICE_wetAlbTemp (default=-1e-3). Remove option SEAICE_SOLVE4TEMP_LEGACY. |
279 |
|
- update verification/*/input*/data.seaice: |
280 |
|
to reproduce previous results with former #defined SEAICE_SOLVE4TEMP_LEGACY |
281 |
|
code, set useMaykutSatVapPoly=T & postSolvTempIter=0 ; |
282 |
|
and to reproduce previous results with former non-SOLVE4TEMP_LEGACY code, |
283 |
|
set SEAICE_wetAlbTemp=0 & SEAICE_snowThick=0. |
284 |
|
o pkg/seaice/seaice_solve4temp.F (SOLVE4TEMP_LEGACY): |
285 |
|
- remove LEGACY code for solving for tsurf (A1,A2,A3) but maintain the same |
286 |
|
algorithm (same choice: useMaykutPolySatVap=T, postSolvTempIter=0); |
287 |
|
difference in results only due to machine truncation. |
288 |
|
- remove MAX_TICE (tsurf is always =< TMELT anyway); keep MIN_TICE if using |
289 |
|
MaykutPolySatVap; keep MIN_LWDOWN. |
290 |
|
- adapt SEAICE_MODIFY_GROWTH_ADJ code (untested) to non-legacy formulation. |
291 |
|
- update output of verification exp. lab_sea (4 fwd: 11 cg2d digits except hb87, |
292 |
|
+ 3 AD: 6 digits except evp) and offline_exf_seaice.seaice (12 digits). |
293 |
|
o pkg/salt_plume: |
294 |
|
- isolate saltPlumeFlux array exch into a specific S/R called from |
295 |
|
do_oceanic_phys.F |
296 |
|
- add interface to pkg/salt_plume (filling of saltPlumeFlux) in pkg/thsice. |
297 |
|
o verification: |
298 |
|
- move untested set-up to MITgcm_contrib/verification_other. |
299 |
|
o pkg/seaice: |
300 |
|
- retire old way of SEAICE_AGE. Now replaced with ALLOW_SITRACER. |
301 |
|
o pkg/seaice: |
302 |
|
- seaice_solve4temp.F (keeping results unchanged): |
303 |
|
* add 2 future run-time params to select which saturation vapor pressure |
304 |
|
formula is used and how post-iteration fluxes are set. |
305 |
|
Temporary set these 2 params according to option SOLVE4TEMP_LEGACY; |
306 |
|
* New piece of code (not used) to update surf. atmos. heat flux F_ia using |
307 |
|
the linearized expression to stay consistent with tsurf finding. |
308 |
|
- seaice_freedrift.F avoid floating exception in sqrt(y) since y=sqrt(x*x)-x |
309 |
|
can be < 0. |
310 |
|
o pkg/seaice/seaice_solve4temp.F (SOLVE4TEMP_LEGACY undef): |
311 |
|
- remove from F_ia (and from it's derivative dFiDTs1) the contribution of |
312 |
|
conductive heat flux F_c , and add it explicitly when updating tsurf as |
313 |
|
solution of: Fc = Fia + d/dT(Fia - Fc) *delta.tsurf |
314 |
|
so that now F_ia has a consistent meaning through the entire routine. |
315 |
|
This causes differences at truncation level: update output of fwd exp. |
316 |
|
global_ocean.cs32x15.seaice. |
317 |
|
o pkg/seaice: |
318 |
|
- lots of editing/cleaning in seaice_solve4temp.F + better documentation |
319 |
|
of arrays, sign/direction of fluxes; no effect on results. |
320 |
|
- replace hard coded 273.16 (=previous value of celsius2K) by celsius2K |
321 |
|
in seaice_budget_ocean.F & seaice_solve4temp.F (SOLVE4TEMP_LEGACY part); |
322 |
|
note: this should have been done before changing celsius2K default. |
323 |
|
- to keep results unchanged, add back old value of celsius2K in verification |
324 |
|
exp. input*/data : lab_sea (AD), offline_exf_seaice (fwd+AD) & seaice_obcs |
325 |
|
- replace hard coded seawater freezing temp in seaice_solve4temp.F (part |
326 |
|
SOLVE4TEMP_LEGACY) of 271.20 (=previous value of celsius2K + default |
327 |
|
value of SEAICE_freeze) by celsius2K+SEAICE_freeze so that seawater |
328 |
|
freezing temp is consistent with value in seaice_growth.F ; this causes |
329 |
|
differences at machine truncation level: update lab_sea AD output (3/4). |
330 |
|
o pkg/seaice: |
331 |
|
- mask (in exf_init_fixed.F) atemp,aqh,evap,precip,runoff,lwflx,areamask |
332 |
|
and climsst/sss even when using seaice (to prevent calculation that might |
333 |
|
use SST on land); does not affect any verification exp. |
334 |
|
- apply mask to taux,tauy (in seaice_get_dynforcing.F); update output |
335 |
|
of lab_sea.fd & lab_sea.hb87 and global_ocean.cs32x15.seaice_dynmix. |
336 |
|
o pkg/seaice/seaice_growth.F: add mask to avoid non-zero ice thickness |
337 |
|
over land (non-legacy code) |
338 |
|
o vectorize salt_plume package: |
339 |
|
- add extra code to salt_plume_frac and salt_plume_tendency_apply_s |
340 |
|
- this code computes plumebk everywhere (also over land), thus requires |
341 |
|
code to avoid divisions by zero and catch the case of kLev=Nr. |
342 |
|
- efficiency of salt_plume_frac is still sub-optimal |
343 |
|
o tools (genmake2): |
344 |
|
- fix makefile NAME specification (option: -makefile NAME) by adding |
345 |
|
"-f $MAKEFILE" in few places where it was missing including makedepend |
346 |
|
command (contribution from Paul). |
347 |
|
|
348 |
|
checkpoint63i (2012/01/24) |
349 |
|
o model/src: |
350 |
|
- add factorized versions of some EOS code to find_rho.F, |
351 |
|
define USE_FACTORIZED_EOS to turn it on (default for TARGET_NEC_SX) |
352 |
|
o model/src: |
353 |
|
- add celsius2K to namelist PARM01; remove unused recip_rhoNil from PARAMS.h; |
354 |
|
- change default celsius2K from 273.16 to 273.15 |
355 |
|
- to keep results unchanged, add back old value of celsius2K in verification |
356 |
|
exp. input*/data : aim.5l_cs, 1D_ocean_ice_column, lab_sea (fwd), |
357 |
|
global_ocean.cs32x15(seaice) & cpl_aim+ocn(input_atm). |
358 |
|
o model/src: add a little initialisation trick to find_rho.F and find_alpha.F |
359 |
|
to make TAF generate vectorizable code |
360 |
|
o pkg/seaice/seaice_evp.F: add code (copied from CICE) that avoids underflows. |
361 |
|
Turn on by defining SEAICE_EVP_ELIMINATE_UNDERFLOWS. Most compilers |
362 |
|
have flags that do this more efficiently. |
363 |
|
o pkg/diagnostics/diagstats_calc.F |
364 |
|
- for TARGET_NEC_SX fix the treatment of the scaling factor tmpFac=scaleFact |
365 |
|
o pkg/obcs (with exch2): |
366 |
|
- derive OB indices in overlap region from OB-InsideMask (in obcs_init_fixed.F) |
367 |
|
and remove code (in obcs_readparms.F) which was based on EXCH of indices |
368 |
|
(through buffer) and did not account for exch2 (e.g., CS-grid) topology. |
369 |
|
o pkg/exf and seaice exp: |
370 |
|
- change offline_exf_seaice.seaicetd test exp. to use default DIFF1 value (=0) |
371 |
|
so that switching off area & heff advection flags does not affect results. |
372 |
|
- change default EXF_LWDOWN_WITH_EMISSIVITY to #define in EXF_OPTIONS.h |
373 |
|
- keep exp. seaice_obcs & 1D_ocean_ice_column(ad) unchanged (using #undef |
374 |
|
EXF_LWDOWN_WITH_EMISSIVITY) and, using pkg/exf default emissivities, update |
375 |
|
results of verification experiments: 1D_ocean_ice_column(fwd), |
376 |
|
global_ocean.cs32x15.seaice(fwd+ad) & seaice_dynmix(ad), lab_sea (all 8) |
377 |
|
and offline_exf_seaice (all 4). |
378 |
|
o pkg/seaice: |
379 |
|
- remove array YNEG, RIVER, TMIX (replaced by local var) from common block; |
380 |
|
remove also areaNm1 & hEffNm1 when SEAICE_GROWTH_LEGACY is undef. |
381 |
|
- fix bug: areaNm1 & hEffNm1 are needed in seaice_growth (SEAICE_GROWTH_LEGACY) |
382 |
|
but were not filled when resp. SEAICEadvArea=F & SEAICEadvHeff=F and |
383 |
|
not multi-dim advection; |
384 |
|
- fix seaice_growth.F to compile with both SEAICE_GROWTH_LEGACY and |
385 |
|
SHORTWAVE_HEATING #undef. |
386 |
|
o optim: make Makefile more robust (?) |
387 |
|
o pkg/obcs: |
388 |
|
* fix treatment of obcs_read_pickup for TAF |
389 |
|
o pkg/shelfice: |
390 |
|
* add missing initialisation of num_shifwflx |
391 |
|
* new adjoint verification exp. for isomip |
392 |
|
* change Prandtl and Schmidt numbers from generic formulations |
393 |
|
(in terms of model diffus. and visc.) to separate runtime |
394 |
|
parameters, with Holland and Jenkins (1999) default values. |
395 |
|
o pkg/exf (interpolation): |
396 |
|
- extended input field (2 rows) near the N & S poles: |
397 |
|
* fill in with the symetric value (when even Nb of data in longitude) |
398 |
|
* add average value at the poles only for scalar quantities ; for vector |
399 |
|
component interpolation, skip the averaging and keep duplicated values, |
400 |
|
unless the 2 components have same location (uvInterp=T) in which case |
401 |
|
interpolate the 2 components together using S/R EXF_INTERP_UV. |
402 |
|
- new S/R EXF_INTERP_UV to interpolate the 2 components of a vector field |
403 |
|
together: this allows to account properly for local orientation when |
404 |
|
averaging at N & S pole. |
405 |
|
- add CPP option EXF_USE_OLD_INTERP_POLE to recover old results regarding |
406 |
|
exf-interpolation near N & S poles. Undef by default. |
407 |
|
- add debug check/print on 1rst iter if debugLevel >=2. |
408 |
|
o pkg/exf: |
409 |
|
- change masking flags definition (no longer "parameter"), put them in |
410 |
|
common block and set them in S/R EXF_INIT_FIXED. |
411 |
|
- change recognized masking flag from s,u,v to c,w,s since the mask |
412 |
|
to use (maskC,W,S) depends on position on model grid and has little to |
413 |
|
do with scalar/vector difference; fix masking if using P-coordinates. |
414 |
|
- fix masking of wind-stress (was wrong if interpolated or read on A-grid) |
415 |
|
and update results of exp. global_with_exf (x2). |
416 |
|
- fix setting of zenith-angle table when SHORTWAVE_HEATING is undef. |
417 |
|
o pkg/obcs |
418 |
|
- S/R obcs_calc_stevens: fix a small bug (gammat -> gammas where appropriate) |
419 |
|
o pkg/exf (interpolation): |
420 |
|
- assume periodicity in X only if input field cover full longitude range. |
421 |
|
- improve search for lat. index (supposed to be faster, in ~log2(ny) steps, |
422 |
|
and should vectorise). |
423 |
|
- fix input lat of the 2 added row (in case we provide N.pole data). |
424 |
|
|
425 |
|
checkpoint63h (2011/12/30) |
426 |
|
o pkg/seaice: |
427 |
|
- change declaration of UVM from _RL to _RS in order to get B-grid |
428 |
|
uIce,vIce monitor when _RS = real*4 |
429 |
|
o pkg/offline: |
430 |
|
- new S/R OFFLINE_RESET_PARMS for resetting main-model params (switch off flags); |
431 |
|
- compute vertical velocity if not read from file (switch on exactConserv |
432 |
|
& staggerTimeStep); |
433 |
|
- fix output of ptracer monitor iter & time for default PTRACERS_monitorFreq. |
434 |
|
o verification: |
435 |
|
- update grid-files (in tutorial_held_suarez_cs/input and fizhi-cs-32x32x40/input) |
436 |
|
with grid-angle generated using "utils/matlab/cs_grid/cubeCalcAngle.m": |
437 |
|
fix Pb in yG definition @ 2 missing corners (previously yG was averaged); |
438 |
|
- update output of exp. global_ocean.cs32x15.viscA4 after changing |
439 |
|
grid-file angles (used here for coriolis-3d); |
440 |
|
- remove (or gzip) few files not needed to run fizhi-cs experiments. |
441 |
|
o pkg/exf (vector interpolation): |
442 |
|
- rotate vector component toward local grid-axis orientation using grid-angles |
443 |
|
(stored in GRID.h) instead of locally computed rotation angle; keep old code |
444 |
|
available with exf option EXF_USE_OLD_VEC_ROTATION (undef by default). |
445 |
|
o pkg/seaice: |
446 |
|
- allow to compile with exf options ALLOW_ATM_TEMP or ALLOW_DOWNWARD_RADIATION |
447 |
|
undefined. Fix and remove the stop when ALLOW_RUNOFF is undefined. |
448 |
|
o pkg/PKG_OPTIONS.h files: |
449 |
|
- move #include "CPP_OPTIONS.h" just after #include "PACKAGES_CONFIG.h", |
450 |
|
outside of the #ifdef ALLOW_THISPKG / #endif block. |
451 |
|
- update verification/*/code/PKG_OPTIONS.h files. |
452 |
|
o add a verification experiment (offline_exf_seaice/input.seaice) that test |
453 |
|
the B-grid LSOR solver |
454 |
|
o pkg/seaice: sort out emissivity and long wave radiation step 2: |
455 |
|
- add emissivity for snow and ice as new run-time parameters |
456 |
|
(unfortunately) that default to the corresponding exf values; this |
457 |
|
changes results in most verification experiments so that for backward |
458 |
|
compatibility these parameter have to be set explicitly to inconsistent |
459 |
|
values (ugly); |
460 |
|
- the exf-flag EXF_LWDOWN_WITH_EMISSIVITY then gets rid off the hard-wired |
461 |
|
emissivities of 0.97 associated with lwdownloc in S/R seaice_solve4temp |
462 |
|
until we agree how to handle this bug properly |
463 |
|
o pkg/exf (interpolation): |
464 |
|
- add a set of run-time param ({inputfield}_interpMethod), one for |
465 |
|
each interpolated input field, to select the interpolation method |
466 |
|
with a value of zero switching off the interpolation; |
467 |
|
- print interpolation parameters to standard output; |
468 |
|
- refine exf_check.F accordingly ; stop if USE_NO_INTERP_RUNOFF is defined |
469 |
|
(to use instead runoff_interpMethod=0 in data.exf). |
470 |
|
- simplify default setting of interpolation parameters (for all fields, |
471 |
|
set position to match model grid cell-center position, leading to trivial |
472 |
|
interpolation in case of regular Lat-Lon grid) and fix for the case of |
473 |
|
non-uniform delY. |
474 |
|
o pkg/exf/exf_radiation.F: add the factor ocean_emissivity to lwdown, for |
475 |
|
within #ifdef EXF_LWDOWN_WITH_EMISSIVITY/#endif. Unless ocean_emissivity=1. |
476 |
|
this factor is absolutely required, because the lw exitance (= out-going |
477 |
|
long wave radiation) is emissivity*stefanBoltzmann*T^4 + rho*lwdown, |
478 |
|
where the reflectivity rho = 1-emissivity for conservation reasons. Once |
479 |
|
this is established, we can remove the CPP-Flag (if we want to). |
480 |
|
o pkg/seaice: |
481 |
|
- qualify a stop statement in S/R seaice_check: EXF_READ_EVAP and |
482 |
|
ALLOW_BULKFORMULAE are only required if SEAICE_EXTERNAL_FLUXES is defined |
483 |
|
- sort out emissivity and long wave radiation, step 1: make SEAICE_emissivity |
484 |
|
what the name implies (and not emissivity*BoltzmannConstant), this means |
485 |
|
the old default value of 5.5e-8 now means |
486 |
|
5.5e-8/5.670e-8 = 0.97001763668430343479 as ocean_emissivity in |
487 |
|
S/R exf_readparms |
488 |
|
- FV-discretization of the B-grid LSOR solver, turn on with |
489 |
|
SEAICE_LSRBNEW. This new solver also involves a slight re-organization |
490 |
|
of the calling routine dynsolver, so that B and C-grid code look a |
491 |
|
little more alike. |
492 |
|
- fix index bugs in ostres.F (B-grid code) |
493 |
|
o pkg/seaice/seaice_ocean_budget.F: change (unused) code to make it look better |
494 |
|
o pkg/seaice/seaice_readparms.F: make ocean_emissivity*stefanBoltzmann the |
495 |
|
default for SEAICE_emissivity, if useEXF; requires setting a wrong |
496 |
|
parameter value in global_ocean.cs32x15/input.seaice/data.seaice for |
497 |
|
backward compatiblity |
498 |
|
o model/src: use sigmaR in calc_ivdc |
499 |
|
|
500 |
|
checkpoint63g (2011/12/16) |
501 |
|
o pkg/seaice/seaice_growth.F: |
502 |
|
fix heat conservation of the coupled ocean-seaice system. |
503 |
|
when SEAICE_HEAT_CONSERV_FIX is defined, the mismatch of |
504 |
|
solid vs liquid water treatment is resolved via QNET. |
505 |
|
o pkg/seaice/seaice_growth.F: |
506 |
|
add missing term for sublimation in saltFlux, which was breaking conservation. |
507 |
|
o pkg/seaice/seaice_model.F: |
508 |
|
Remove this patholigical case treatment in evolution branch, |
509 |
|
where it is un-necessary and breaks conservation. |
510 |
|
Update global_ocean.cs32x15/results accordingly for |
511 |
|
output.seaice.txt output_adm.seaice.txt and output_adm.seaice_dynmix.txt |
512 |
|
o model/src: |
513 |
|
useRealFreshWaterFlux: remove masking of EmPmP (using maskInC) in |
514 |
|
external_forcing_surf.F and apply mask directly in S/R SOLVE_FOR_PRESSURE. |
515 |
|
o pkg/obcs: |
516 |
|
modify Stevens boundary conditions so that S/R obcs_calc_stevens computes |
517 |
|
tendency for temperature and salinity that is then applied to update |
518 |
|
theta and salt in S/R obcs_apply_ts; this requires proper masking of all |
519 |
|
horizontal advection and diffusion contributions on the boundary (in |
520 |
|
S/Rs gad_advection, gad_calc_rhs, gmredi_rtransport, gmredi_calc_diff), and |
521 |
|
removing the mask in S/R timestep_tracer. The advantage of this approach is |
522 |
|
that the updated tracer values include contributions from the surface |
523 |
|
forcing and vertical diffusion. |
524 |
|
This modification changes exp4.stevens because of vertical diffusion. |
525 |
|
o model/src: |
526 |
|
change non-curvilinear grid setting to also work with exch2 + blank tiles: |
527 |
|
- move delX,delY to new header file (SET_GRID.h) and adjust length to 1rst |
528 |
|
face dimensions. |
529 |
|
- new S/R INI_LOCAL_GRID to define model tile-local grid (using same units |
530 |
|
as delX,delY) and called from S/R INI_CARTESIAN_GRID, INI_CYLINDER_GRID |
531 |
|
and INI_SPHERICAL_POLAR_GRID. |
532 |
|
change test-experiment global_ocean.90x40x15: |
533 |
|
- split the domain in many more tiles (9x4 tiles of 10x10 each); |
534 |
|
- switch to exch2 and remove 1 blank-tile (but only for MPI test). |
535 |
|
- update results of both test (std & dwnslp). |
536 |
|
o pkg/diagnostics/diagstat_calc.F: fix problem with non-initialised halos in |
537 |
|
tmp-fields for TARGET_NEC_SX at the cost of extra loops and less vectorisation |
538 |
|
o model/src: |
539 |
|
- take sigma-coords related code out of update_etah.F into new S/R UPDATE_ETAWS |
540 |
|
and fix initialisation when using sigma-coords. |
541 |
|
o model/src: |
542 |
|
- improve Gael's changes from Jan 14, 2011 "addition of an 'update' of hfacc etc. |
543 |
|
to their current value at the beginning of forward_step.F": add a new run-time |
544 |
|
switch "doResetHFactors" (default=false) and use it to do (or not) the resetting. |
545 |
|
- set doResetHFactors=T in the following test exp: |
546 |
|
global_ocean.90x40x15/input (rStar) & input.dwnslp (linFS), |
547 |
|
global_ocean.cs32x15/input.seaice (rStar), hs94.cs-32x32x5/input.impIGW (linFS) |
548 |
|
and tutorial_global_oce_in_p (nonLinFS in p) |
549 |
|
o pkg/ecco |
550 |
|
- added IES cost term |
551 |
|
o verification/advect_xz |
552 |
|
- scale down the depth by 1/100 (from 200 km to just 2.km) |
553 |
|
- add secondary test using divergent flow, r-star and implicit vertical advection. |
554 |
|
- add note (check_conserve_TS.txt) and matlab script to check global |
555 |
|
conservation of T & S with NonLin-FreeSurf and Adams-Bashforth. |
556 |
|
o pkg/rbcs: |
557 |
|
- add missing relaxUFile & relaxVFile in namelist (in rbcs_readparms.F). |
558 |
|
o pkg/obcs (Stevens BCs): |
559 |
|
- fix various problems in the phase speed computations of the Stevens |
560 |
|
boundary conditions: |
561 |
|
- sign error for western and southern boundaries, |
562 |
|
- time tendencies were underestimated by factor of dTracerLev and some just |
563 |
|
zero (for multidim advection schemes) -> now estimate time tendencies |
564 |
|
from difference between previous (n) time level and the one before (n-1); |
565 |
|
this requires storing a field per tracer per boundary, also in the pickup |
566 |
|
files. |
567 |
|
- replace deltaTmom in cflMer/Zon computation with dTracerLev |
568 |
|
- set default phase velocity to cflMer/Zon |
569 |
|
- update verification experiment exp4.stevens |
570 |
|
o pkg/generic_advdiff (implicit_r): |
571 |
|
- fix implicit vertical advection conservation with AB and/or NonLin-FreeSurf |
572 |
|
(remove Tr*d/dz(w) in gad_implicit_r.F and add it in gad_calc_rhs.F); |
573 |
|
- fix implicit vertical diffusion or advection with NonLin FreeSurf |
574 |
|
(was using current recip_hFacC instead of future recip_hFacC): set future |
575 |
|
recip_hFac in thermodynamics.F and pass & use it in all vertical implicit |
576 |
|
solver S/R ; |
577 |
|
- fill diagnostics of vertical advective fluxes (computed from tendency) |
578 |
|
when using implicit vert. advection; |
579 |
|
- update output of exp that uses NonLin-FreeSurf & implicit vertical diffusion: |
580 |
|
global_ocean.90x40x15 (fwd+AD), global_ocean.cs32x15 (4 fwd +4 AD), |
581 |
|
tutorial_global_oce_in_p (fwd), tutorial_tracer_adjsens (AD) |
582 |
|
and cpl_aim+ocn (used in ocean); to be fixed: bottom_ctrl_5x5. |
583 |
|
o pkg/diagnostics: |
584 |
|
- add diags for T & S tendency which goes through Adams-Bashforth. |
585 |
|
|
586 |
|
checkpoint63f (2011/11/30) |
587 |
|
o pkg/obcs/obcs_calc_stevens.F: |
588 |
|
- join some loops to reduce number of if-statements |
589 |
|
- fix a bug in code for northern and southern boundaries. |
590 |
|
o pkg/seaice/seaice_init_varia.F |
591 |
|
- undo resetting UVM=0 that has slipped in unnoticed in r1.49 and |
592 |
|
caused zero ice-velocities for the B-grid code |
593 |
|
o pkg/seaice/seaice_init_fixed.F |
594 |
|
- fix a bug in the initialisation of the second metric terms |
595 |
|
k2AtC and k2AtZ: k2AtC was set to the value of k2AtZ and k2AtZ was 0. |
596 |
|
- this affects all seaice configurations with useCurviLinearGrid = .true. |
597 |
|
and SEAICEuseDynamics = .true. (default) and SEAICEuseMetricTerms |
598 |
|
= .true. (default) |
599 |
|
- the only verification experiment affected is global_ocean.cs32x15: |
600 |
|
icedyn, seaice, and seaice_dynmix (AD) |
601 |
|
o pkg/thsice/thsice_diagnostics_state.F |
602 |
|
- only for TARGET_NEC_SX: initialise all of tmpFld, so that overlap does |
603 |
|
not contain garbage that can lead to surprises in diagstats_calc.F |
604 |
|
o model/src: |
605 |
|
- move bi,bj loops inside INTEGR_CONTINUITY ; |
606 |
|
- call INTEGR_CONTINUITY directly from forward_step.F |
607 |
|
(previously called from momentum_correction_step.F) |
608 |
|
- call UPDATE_ETAH from integr_continuity.F and update initialise_varia.F, |
609 |
|
forward_step.F and pkg/ecco/the_main_loop.F |
610 |
|
o model/src/do_oceanic_phys.F: |
611 |
|
- reset addMass to zero only if useICEFRONT ; |
612 |
|
- separated k-loop for rhoInSitu calculation ; |
613 |
|
- add 2 diagnostics for wVel*delta^k(Rho) at constant pressure (= WdRHO_P) |
614 |
|
and at constant T,S (= WdRHOdP); |
615 |
|
NOTE: change diagnostics WRHOMASS : this is now simply the Vertical transport |
616 |
|
of In-Situ Density (Anomaly) (=wVel*rho^bar-k) (previously was Vertical |
617 |
|
transport of locally-referenced potential density); |
618 |
|
to recover old diag: (WRHOMASS)_old = (WRHOMASS)_new -1/2 * (WdRHOdP) |
619 |
|
o gluemnc: |
620 |
|
- replace ncrcat with ncks when adding variables to *.glob.nc |
621 |
|
- change the way variables without record dimensions are handled so that |
622 |
|
files that have variables with and without record variables can be glued: |
623 |
|
add a record dimension with a new unique name to be able to differentiate |
624 |
|
between variable with and without a-prior record dimensions. Remove |
625 |
|
this record dimension before adding the variable to the *.glob.nc file. |
626 |
|
This procedure automatically works also for grid.*.nc files that do not |
627 |
|
have any record dimension. |
628 |
|
o verification/cpl_aim+ocn: |
629 |
|
- generate new reference output on baudelaire (gfortran, mpich2) (previously |
630 |
|
from old aces-grid cluster, ao + ifort) and update 'run_cpl_test' script. |
631 |
|
o pkg/obcs/obcs_calc_stevens.F: |
632 |
|
- add code for northern and southern boundaries. |
633 |
|
- reorder some loops to save some operations and so that drFBar can be scalar |
634 |
|
o model/src: |
635 |
|
- re-arrange extrapolation calculation in Adams-Bashforth S/R (affects machine |
636 |
|
truncation) and add diagnostics for Adams-Bashforth tendency increment. |
637 |
|
- update test-exp. AD output: |
638 |
|
1D_ocean_ice_column, bottom_ctrl_5x5, lab_sea.noseaicedyn, obcs_ctrl; |
639 |
|
and FWD output: exp4 (x3), fizhi-cs-32x32x40, fizhi-cs-aqualev20, |
640 |
|
front_relax (x3), global_ocean.90x40x15 (x2), global_ocean.cs32x15 (x5), |
641 |
|
global_ocean_ebm, ideal_2D_oce, internal_wave, isomip (x3), lab_sea (x4), |
642 |
|
matrix_example, MLAdjust (x5), natl_box (x2), tutorial_advection_in_gyre, |
643 |
|
tutorial_global_oce_in_p, tutorial_global_oce_latlon and |
644 |
|
tutorial_plume_on_slope. |
645 |
|
o pkg/diagnostics/diagstats_calc.F: add special code for TARGET_NEC_SX |
646 |
|
that vectorises. This code uses 3 extra 2D fields, F90 array assignments |
647 |
|
and F90 intrinsic routines SUM, MINVAL, MAXVAL, but since TARGET_NEC_SX |
648 |
|
does not even come with an F77 compiler anymore, this should not be |
649 |
|
a limitation |
650 |
|
|
651 |
|
checkpoint63e (2011/11/06) |
652 |
|
o pkg/seaice adjoint of LSR: |
653 |
|
- substantially rework STORE dirs. |
654 |
|
- add new flag SEAICE_LSR_ADJOINT_ITER and parameter SOLV_MAX_FIXED |
655 |
|
this flag will do SOLV_MAX_ITERS but only store SOLV_MAX_FIXED fields |
656 |
|
and only perform SOLV_MAX_FIXED adjoint iters. |
657 |
|
If flag SEAICE_LSR_ADJOINT_ITER is undef-ined, the store in the iteration |
658 |
|
always goes to same record, meaning that the last forward field will be |
659 |
|
the one used in the adjoint - an approximation. |
660 |
|
- now produces non-zero LSR-related gradients, and acceptable gradients for |
661 |
|
An's zonal (x-z) test case (to be checked in) |
662 |
|
- lab_sea: produces acceptable gradients for lab_sea with forward/backward LSR |
663 |
|
(but these are screwed up by the thermodynamics in the first place); |
664 |
|
reduce from 4 to 3 timesteps to avoid t.d. issue as temporary solution; |
665 |
|
output_adm.txt also reflects changes to seaice_lsr |
666 |
|
- global_ocean.cs32x15: |
667 |
|
* the "default" input_ad.seaice has been re-defined to always exclude dynamics |
668 |
|
(output_adm.seaice.txt was updated) |
669 |
|
* new input_ad.seaice_dynmix tests a forward LSR / backward free-drift |
670 |
|
(output_adm.seaice_dynmix.txt) |
671 |
|
o test-exp. exp4+stevens: |
672 |
|
- switch off pTracers and nonHydrostatic (restart problems with NH); update output |
673 |
|
o pkg/seaice (lsr solver): |
674 |
|
- put coriolis term into FORCEX,Y (as in seaice_evp.F; previously in rhsU,V only); |
675 |
|
update output (change truncation) of FWD test exp lab_sea & global_ocean.cs32x15 |
676 |
|
- fix SEAICEuseFREEDRIFT for new OBCS version (ndef OBCS_UVICE_OLD) |
677 |
|
- new parameter "LSR_mixIniGuess" allows (if=2,4) to mix free-drift solution |
678 |
|
into initial guess for linearised solver; use weight ~ (1/err)^2,4. |
679 |
|
- compute (with new S/R) and print initial (+ free-driff sol. if available) |
680 |
|
and final residual of linearised system. |
681 |
|
o pkg/pp81: add diagnostics for vertical eddy viscosity and diffusivity |
682 |
|
(requires new subroutines pp81_init_fixed.F/pp81_diagnostics_init.F) |
683 |
|
o pkg/obcs: Stevens BCs (still only for east and west) |
684 |
|
- fix Stevens BCs by using intermediate velocities of previous |
685 |
|
time step as "baroclinic" contribution rather velocity of one grid cell |
686 |
|
inside the boundaries; more consistent but makes the entire treatment |
687 |
|
more complicated, requires new pickup files and two new S/Rs: |
688 |
|
obcs_copy_uv_n and obcs_save_uv_n |
689 |
|
keep old version with flag OBCS_STEVENS_USE_INTERIOR_VELOCITY |
690 |
|
(undefined by default) |
691 |
|
- silently move computation of some grid factors into separate loops to |
692 |
|
save some recomputations |
693 |
|
- add a hack for non-linear free surface |
694 |
|
- add a verification experiment to exp4 that tests this code |
695 |
|
o make gluemnc/xplodemnc work with (moron) non-GNU shell tools |
696 |
|
|
697 |
|
checkpoint63d (2011/10/25) |
698 |
|
o pkg/obcs + seaice-dynamics: |
699 |
|
- new implementation calls OBCS_APPLY_UVICE (in seaice_dynsolver.F) before |
700 |
|
solving for uIce,vIce and prevent SEAICE_LSR or SEAICE_EVP to modify OB |
701 |
|
values (consistent with an implicit method). |
702 |
|
- add option OBCS_UVICE_OLD (in OBCS_OPTIONS.h, defined by default) to |
703 |
|
return to older implementation (same results as before); note that none of |
704 |
|
the "experimental" OBCS_SEAICE options are implemented in the new version. |
705 |
|
- switch test exp. seaice_obcs to use new version (update output). |
706 |
|
o pkg/seaice: |
707 |
|
- add few stop (B-grid & CS-exch, B-grid & obcs, AUTODIFF & SEAICE_maskRHS) |
708 |
|
- add parenthesis in Water-Ice drag calculations (evp & lsr); |
709 |
|
- in seaice_lsr.F, simplify and fix AA3 setting in case sNx or sNy=1; |
710 |
|
and fix case SEAICE_waterDrag=0 (but unstable if no or little seaice) |
711 |
|
- update output of test exp lab_sea that uses evp or lsr (fwd+ad) |
712 |
|
and global_ocean.cs32x15 (icedyn & seaice fwd+ad). |
713 |
|
o pkg/layers: |
714 |
|
- Include potential density as new coordinate (Thanks to David Munday) |
715 |
|
o pkg/dic: |
716 |
|
- Remove subroutine CALC_PCO2_APPROX_CO3 from carbon_chem.F and add |
717 |
|
carbonate computation/output to CALC_PCO2_APPROX |
718 |
|
o pkg/generic_advdiff |
719 |
|
- make TAF generate code that vectorizes: |
720 |
|
gad_dst3fl_adv_*.F, gad_u3c4_impl_r.F |
721 |
|
o model/src: |
722 |
|
- improve vectorization also for adjoint code of |
723 |
|
solve_tridiagonal and solve_pentadiagonal. As this requires up to |
724 |
|
5 new 3D-fields, this code is only turned on for the combination of |
725 |
|
ALLOW_AUTODIFF_TAMC and TARGET_NEC_SX. Also remove error message |
726 |
|
for singular matrix check (only for TARGET_NEC_SX). |
727 |
|
- remove hard-wiring of CPP-flag ALLOW_SOLVERS_KLOOPINSIDE to |
728 |
|
ALLOW_AUTODIFF. This flag now needs to be set outside of |
729 |
|
solve_tri/pentadiagonal.F |
730 |
o pkg/exf: |
o pkg/exf: |
731 |
- fix previous modif in exf_readparms.F (revision 1.66: treatment of |
- fix previous modif in exf_readparms.F (revision 1.66: treatment of |
732 |
optional namelist EXF_NML_OBCS). |
optional namelist EXF_NML_OBCS). |