| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o pkg/bling: |
| 8 |
|
- bling_production: Fix typo in calcium carbonate uptake. Also rearranging and cleaning |
| 9 |
|
o model/src: |
| 10 |
|
- move k loop inside correction_step.F (+ update calling S/R) |
| 11 |
|
- collect implicit pressure gradient tendencies in local 2-D array; |
| 12 |
|
affect results (at machine truncation level) when using 3-D solver; update |
| 13 |
|
output of exp.: exp4 (x1), global_ocean.cs32x15 (x2),hs94.cs-32x32x5.impIGW, |
| 14 |
|
rotating_tank, short_surf_wave,tutorial_deep_convection (x2) |
| 15 |
|
& tutorial_plume_on_slope. |
| 16 |
|
|
| 17 |
|
checkpoint66j (2017/08/15) |
| 18 |
|
o pkg/ptracers: |
| 19 |
|
- fix surface forcing under ice-shelf (useShelfIce=T). |
| 20 |
|
o pkg/atm_phys: |
| 21 |
|
- use hFac weighted average when applying A-grid atm-phys momentum tendency |
| 22 |
|
to C-grid dynamics; this ensures wind-stress match mom tendency. |
| 23 |
|
o eesupp/src: |
| 24 |
|
- change default for opening scratch files to avoid STATUS='SCRATCH' |
| 25 |
|
- close scratch files with STATUS='DELETE' |
| 26 |
|
- changes affect eeset_parms.F and open_copy_data_file.F and many routines |
| 27 |
|
that read namelists (because they close the scratch files) |
| 28 |
|
- you can revert to old default by setting USE_FORTRAN_SCRATCH_FILES |
| 29 |
|
in CPP_EEOPTIONS.h (tested in ideal_2D_oce/code & lab_sea/code_ad) |
| 30 |
|
o pkg/atm_phys: |
| 31 |
|
- fix unset arguments in (unused) S/R surface_flux_0d. |
| 32 |
|
o eesupp/src: |
| 33 |
|
- add argument "procId" to S/R EESET_PARMS to fix scratchFile1 & 2 setting |
| 34 |
|
(case TARGET_BGL or TARGET_CRAYXT) for early call (from eeboot_minimal.F): |
| 35 |
|
called with myProcId in eeboot.F and with mpiMyWId in eeboot_minimal.F |
| 36 |
|
o pkg/streamice: |
| 37 |
|
- remove blank between _EXCH_XY_RL and "()", which my Apple Xcode |
| 38 |
|
preprocessor /usr/bin/cpp cannot interpret; 3 instances in total in |
| 39 |
|
streamice_driving_stress_fd.F and streamice_advect_2dtracer.F |
| 40 |
|
o pkg/diagnostics: |
| 41 |
|
- new option to allow to cumulate thickness-factor (hFac) weighted field |
| 42 |
|
(if field permitted); implemented by changing sign of "jdiag" pointer |
| 43 |
|
(> 0: cumlulate field, < 0: cumulate hFac weighted field). |
| 44 |
|
a) restricted to 3-d (Nr levels), level centered diagnostics at U,V or Cell |
| 45 |
|
center position (gcode(2)='U','V','M' & gcode(3,5,9:10)='R',' ','MR') |
| 46 |
|
b) for now, activated by setting fileFlags 3rd character to 'h'. |
| 47 |
|
|
| 48 |
|
checkpoint66i (2017/07/18) |
| 49 |
|
o pkg/cheapaml: |
| 50 |
|
- add parameter "useRelativeWind", default set to False; this change the |
| 51 |
|
previous setting which was always using relative wind. |
| 52 |
|
o add hooks for the Parallel Data Assimilation Framework PDAF |
| 53 |
|
(http://pdaf.awi.de/) in eeboot_minimal.F and the_main_loop.F; |
| 54 |
|
to use set CPP-flag USE_PDAF in CPP_EEOPTIONS.h |
| 55 |
|
o pkg/exf: |
| 56 |
|
- interpolation of vector fields (exf_interp_uv.F): fix bug in calculation |
| 57 |
|
of South-Pole mean value. |
| 58 |
|
- interpolation near the poles, case where second additional row is at the |
| 59 |
|
pole (or beyond the pole and moved to the pole): change first addition row |
| 60 |
|
value to a linear interpolation between pole and 1rst (S.pole)/last (N.pole) |
| 61 |
|
row (instead of just a copy of it). |
| 62 |
|
o pkg/seaice: |
| 63 |
|
- prevent JFNK and KRYLOV solvers from trying to reduce a residual of 0 |
| 64 |
|
- add the option to compute no-slip du/dy, dv/dx (i.e. |
| 65 |
|
the offdiagonal strain rate e12) by a second order approximation |
| 66 |
|
on the boundary; works only with JFNK, KRYLOV, or EVP solvers, because |
| 67 |
|
it is too messy to implement for the implicit LSR matrix. |
| 68 |
|
SEAICE_2ndOrderBC = .FALSE. by default |
| 69 |
|
|
| 70 |
|
checkpoint66h (2017/06/02) |
| 71 |
|
o pkg/seaice: |
| 72 |
|
- add new routines to compute stress tensor components and stress tensor |
| 73 |
|
divergence to be reused in s/r seaice_lsr and s/r seaice_calc_lhs |
| 74 |
|
- change names of diagnostics SIsigI/II (commonly used for invariants) |
| 75 |
|
to SIsig1/sig2 (more common for principal components) to avoid confusion |
| 76 |
|
o pkf/exf: |
| 77 |
|
- add a stop if useRelativeWind with un-updated uwind/vwind (e.g., constant); |
| 78 |
|
similarly, stop if rotateStressOnAgrid with constant wind-stress. |
| 79 |
|
o pkg/seaice: |
| 80 |
|
- remove (comment out) diagnostics SIuwind,SIvwind,SIfu,SIfv (duplicate of |
| 81 |
|
EXFuwind,EXFvwind,oceTAUX,oceTAUY) and point them to their equivalent. |
| 82 |
|
- add diagnostics of wind-stress over seaice (SItaux,SItauy) and seaice |
| 83 |
|
fraction weighted wind stress over ocean+seaice (SIatmTx,SIatmTy). |
| 84 |
|
o pkg/diagnostics: |
| 85 |
|
- small piece of code to rename (old) diagnostics, when namelist is processed. |
| 86 |
|
o model/src: |
| 87 |
|
- new run-time param (useMin4hFacEdges) to select method for setting hFacW,S: |
| 88 |
|
originally, set as minimum of adjacent hFacC factor ; now (new default) |
| 89 |
|
computed from rSurfW,S and rLowW,S by applying same rules as for hFacC. |
| 90 |
|
Only matters when useShelfIce=T with particular ice-shelf cavity geometry. |
| 91 |
|
o pkg/seaice: |
| 92 |
|
- add momemtum advection, off by default, still in test phase |
| 93 |
|
most of the momentum advection code is adapted and reused from mom_vecinv |
| 94 |
|
o pkg/mom_vecinv: |
| 95 |
|
- pass these runtime flags: |
| 96 |
|
selectVortScheme, highOrderVorticity, upwindVorticity, useJamartMomAdv |
| 97 |
|
as formal parameters to s/r mom_vi_u/v_coriolis, mom_vi_u/v_coriolis_c4, |
| 98 |
|
so that these routines can also be used in pkg/seaice. |
| 99 |
|
|
| 100 |
|
checkpoint66g (2017/04/24) |
| 101 |
|
o model/src: |
| 102 |
|
- more robust ordering of column thickness integral (=Sum_k(drF*hFac) ) |
| 103 |
|
to get same expression (regarding machine truncation) in different places |
| 104 |
|
in ini_masks_etc.F (e.g., rSurfC,W,S is now exactly zero when it should). |
| 105 |
|
Update forward output of experiments: global_ocean.cs32x15 (all 5 except |
| 106 |
|
viscA4), isomip (x3), tutorial_global_oce_in_p & tutorial_plume_on_slope. |
| 107 |
|
o pkg/shelfice: |
| 108 |
|
- ALLOW_SHIFWFLX_CONTROL: move setting of maskSHI from shelfice_update_masks.F |
| 109 |
|
to shelfice_init_fixed.F |
| 110 |
|
- change S/R SHELFICE_UPDATE_MASKS to SHELFICE_INIT_DEPTHS (still called from |
| 111 |
|
ini_masks_etc.F for now) to update ocean upper boundary position according |
| 112 |
|
to ice-shelf topography (instead of directly updating hFacC). |
| 113 |
|
o model/src: |
| 114 |
|
- move call to CTRL_INIT & CTRL_ADMTLM from the middle to the bottom of |
| 115 |
|
packages_init_fixed.F to allow other pkgs to set mask and weight fields. |
| 116 |
|
|
| 117 |
|
checkpoint66f (2017/04/06) |
| 118 |
|
o model/src: |
| 119 |
|
- add specific run-time param to select level of printed plot-field-maps, |
| 120 |
|
set by default to debugLevel. Remove internal param "printDomain". |
| 121 |
|
o pkg/seaice: |
| 122 |
|
- add flag to add snow mass to seaiceMassC/U/V, the default is |
| 123 |
|
SEAICEaddSnowMass=.FALSE., not to change previous results |
| 124 |
|
o pkg/bling |
| 125 |
|
- Bug fix in bling_production regarding unit conversion for diagnostics |
| 126 |
|
being put in the wrong place |
| 127 |
|
o model/src: |
| 128 |
|
- add back recomputation of rSurfW & S to ensure consistent column thickness |
| 129 |
|
(needed for rStar) between Sum_k(hFac*drF) and rSurf-rLow, for now only |
| 130 |
|
when using pkg/shelfice; was initially added on 2012/06/01 (ini_masks_etc.F |
| 131 |
|
revision 1.51) but was removed accidentally on 2014/02/08 (v1.53). |
| 132 |
|
o pkg/autodiff: |
| 133 |
|
- clean-up "checkpoint.flow" |
| 134 |
|
o pkg/rw (Contribution from Jody Klymak): |
| 135 |
|
option to set mds file suffix as function of time in simulation: |
| 136 |
|
*) new input parameter "rwSuffixType" (read from PARM03 namelist) |
| 137 |
|
to select file suffix format; by default (=0), use iteration number; |
| 138 |
|
*) add simple routine to return file suffix according to rwSuffixType; |
| 139 |
|
*) in model/src and in most pkgs (except: atm2d, fizhi, regrid & streamice), |
| 140 |
|
use new S/R to get file suffix, except for debug output file (diagFreq) |
| 141 |
|
and pkgs offline & rbcs input files. |
| 142 |
|
o pkg/rw: |
| 143 |
|
- uniform processing of "I10" suffix across writing fld routines; used to |
| 144 |
|
simplify debug writing of cg2/3d_b/x in solve_for_pressure.F & post_cg3d.F |
| 145 |
|
(similar to cg3d.F, cg3d_ex0.F) |
| 146 |
|
o pkg/mom_vecinv: |
| 147 |
|
- highOrderVorticity in selectVortScheme=2 case: fix c4 coeff in splitted |
| 148 |
|
vorticity flux (keep same sum in case vBarXm=vBaXp, uBarYm=uBarYp). |
| 149 |
|
o pkg/bling: |
| 150 |
|
- Many small changes intended to clean the code. |
| 151 |
|
Unnecessary ADJOINT_SAFEs removed ; |
| 152 |
|
removed BLING_COST.h BLING_CTRL.h bling_cost.F ; |
| 153 |
|
diagnostics added. |
| 154 |
|
|
| 155 |
|
checkpoint66e (2017/03/14) |
| 156 |
|
o pkg/exf: |
| 157 |
|
- interpolation with #undef EXF_INTERP_USE_DYNALLOC: move buffer size |
| 158 |
|
definition outside EXF_PARAM.h in new header file: EXF_INTERP_SIZE.h; |
| 159 |
|
use fixed size buffer (passed to S/R EXF_INTERP & EXF_INTERP_UV) to |
| 160 |
|
avoid any dynamic allocation ; |
| 161 |
|
- S/R EXF_CHECK: add new check & stop when file-name is defined but not |
| 162 |
|
read-in ; add warning when file is read-in but not used within pkg/exf ; |
| 163 |
|
+ check for buffer-size (if #undef EXF_INTERP_USE_DYNALLOC). |
| 164 |
|
- update exf_summary.F (print {FLD}RepCycle). |
| 165 |
|
o verification/global_with_exf: |
| 166 |
|
- switch on exf interpolation in AD test (just like the forward test) |
| 167 |
|
o pkg/mom_vecinv: |
| 168 |
|
- add storage directives to avoid recomputations within mom_vecinv.F |
| 169 |
|
which were generating seg. fault with gfortran -devel |
| 170 |
|
o pkg/ctrl, pkg/smooth: |
| 171 |
|
- initialize array halos which were generating seg. fault with |
| 172 |
|
gfortran -devel in adjoint mode |
| 173 |
|
o pkg/exf: |
| 174 |
|
- for each input field, enable to specify a field specific repeat-cycle |
| 175 |
|
({FLD}RepCycle) which defaults to pkg/exf main "repeatPeriod". |
| 176 |
|
o verification/lab_sea: |
| 177 |
|
- update AD results of primary test and noseaicedyn test after switching |
| 178 |
|
to new version of TAF (5.3.13 vs 5.3.2 before); agreement with gradient |
| 179 |
|
check stays about the same. |
| 180 |
|
o pkg/exf: |
| 181 |
|
- fix bug in exf_zenithangle.F for cases select_ZenAlbedo > 0 with useCAL=T. |
| 182 |
|
- exf_check: MAX_LAT_INC need only be "loc_nlat-1", not equal to loc_nlat |
| 183 |
|
o pkg/bling, USE_EXFCO2 code: |
| 184 |
|
- add parameter "apco2RepCycle" (set to pkg/exf "repeatPeriod" by default); |
| 185 |
|
- update calls to EXF S/R (use new version: EXF_INIT_FLD & EXF_SET_FLD). |
| 186 |
|
o pkg/exf: |
| 187 |
|
- rename S/R EXF_INIT to EXF_INIT_VARIA ; |
| 188 |
|
- add arguments: fldName (for debug msg) and fldRepeatCycle (set to |
| 189 |
|
repeatPeriod) to S/R EXF_SET_UV and to S/R EXF_SET_GEN (renamed EXF_SET_FLD) |
| 190 |
|
and update calls in pkg/exf calling S/R (exf_getffields.F & exf_getclim.F) ; |
| 191 |
|
- add argument: fldName (for debug msg) to S/R EXF_INIT_GEN (renamed |
| 192 |
|
EXF_INIT_FLD) and update calls in exf_init_varia.F |
| 193 |
|
- For now, for backward compatibility, keep a simplified version of S/Rs |
| 194 |
|
EXF_SET_GEN & EXF_INIT_GEN in file exf_set_gen.F |
| 195 |
|
- add and pass new arguments: fldName (for debug msg) and fldRepeatCycle |
| 196 |
|
to S/R EXF_GetFFieldRec ; update calls in pkg/obcs/obcs_exf_load.F |
| 197 |
|
o tools/genmake2: |
| 198 |
|
- replace "bc -l" (only used once) by "expr" (already used). |
| 199 |
|
o verification/global_ocean.cs32x15: |
| 200 |
|
- turn off pkg/cal (useCAL=F) in secondary FWD exp. "icedyn". |
| 201 |
|
Note: exf setting (data.exf) can be compared to other exf secondary exp |
| 202 |
|
"seaice" (kept with useCAL=T) which has same time dependent forcing. |
| 203 |
|
|
| 204 |
|
checkpoint66d (2017/02/13) |
| 205 |
|
o pkg/exf: |
| 206 |
|
- remove setting of useCAL=T when using pkg/exf (useEXF=T); add warning |
| 207 |
|
early on (in packages_boot.F) to stress this change in pkg setting. |
| 208 |
|
- In (most) verification exp. that use pkg/exf, add useCAL=.TRUE. to data.pkg |
| 209 |
o model/src: |
o model/src: |
| 210 |
- move call to ecco_phys to end of time step; this may induce minor |
- move call to ecco_phys to end of time step; this may induce minor |
| 211 |
cost function changes by shifting time averages by one time step |
cost function changes by shifting time averages by one time step |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch