4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
5 |
============================== |
============================== |
6 |
|
|
7 |
|
o tools/adjoint_options/adjoint_diva: |
8 |
|
- fix bug (typo) in Tang-Linear FLAGS setting for MPI + DIVA |
9 |
|
o pkg/streamice: |
10 |
|
- spotted bug in streamice_driving_stress using dyG where should have been dxG |
11 |
|
o pkg/exf: |
12 |
|
- remove STOP statement that prevented to use wind or (wind-)stress |
13 |
|
interpolation for constant-in-time forcing fields (wind/stress period = 0). |
14 |
|
o pkg/seaice: |
15 |
|
- replace SEAICE_EVP_USE_ADAPTATION by run time parameters, requires |
16 |
|
5 additionals 2D fields and 6 new store directives |
17 |
|
- change logic in seaice_readparms.F: EVP code can now be turned on in |
18 |
|
various ways, setting SEAICE_deltaTev is no longer required, |
19 |
|
alternatively set SEAICE_evpAlpha/Beta, SEAICEuseEVPstar, SEAICEuseEVPrev, |
20 |
|
or SEAICEaEVPcoeff |
21 |
|
o pkg/shelfice/shelfice_thermodynamics.F |
22 |
|
- redefining uLoc/vLoc to u/v, instead of 2u/v |
23 |
|
o pkg/seaice/seaice_evp.F: introduce "adaptive EVP" by Madlen Kimmritz |
24 |
|
- this is an intermediate state of a truely converging EVP that has |
25 |
|
the potential of outperforming implicit VP methods, I do not recommend |
26 |
|
using it yet. |
27 |
|
- for now, "adaptive EVP" is turned on by setting SEAICE_EVP_USE_ADAPTATION |
28 |
|
and all parameters are hardwired |
29 |
|
- for convergence it still requires very smooth regularisations, that means, |
30 |
|
if SEAICE_DELTA_SMOOTHREG is defined, deltaCreg=deltaC+deltaMin, which is |
31 |
|
different from the VP case with deltaCreg=sqrt(deltaC**2+deltaMin**2); also |
32 |
|
specific averaging appears to be required, still subject to tuning |
33 |
|
- still needs cleaning and I need to turn parameter into runtime parameters |
34 |
|
- add code to compute and print residuals of iteration, compile with |
35 |
|
defining ALLOW_SEAICE_EVP_RESIDUAL to enable it |
36 |
|
o pkg/seaice/seaice_evp.F: fix a bug (factor of four in front of e12Csq) |
37 |
|
that entered with r1.54 on Sep04, 2015, -> update results |
38 |
|
|
39 |
|
checkpoint65o (2015/09/14) |
40 |
|
o pkg/mom_vecinv & mom_common: |
41 |
|
- add deep-atmosphere geometry factor in mom_calc_visc.F + in routines |
42 |
|
called from there. |
43 |
|
- add anelastic and deep-atmosphere geometry factor in pkg/mom_vecinv ; this |
44 |
|
allows to use Vector-Invariant form in deep atmos and anelastic formulation |
45 |
|
o pkg/seaice: introduce fast ice parameterisation following Itkin et al. (2015) |
46 |
|
and Koenig-Beatty+Holland (2012) |
47 |
|
- two new parameters SEAICE_tensilFac and SEAICE_tensilDepth |
48 |
|
- global 2D field tensileStrength, computed in seaice_calc_ice_strength |
49 |
|
- requires an extra input argument for S/R seaice_calc_viscosities |
50 |
|
- tensileStrength's depth dependence is different from Itkin et al. (2015) |
51 |
|
(to be changed and played with, for now exponential) |
52 |
|
o pkg/seaice: introduce new parameter SEAICEscaleSurfStress (default = .FALSE.) |
53 |
|
- if TRUE scale ice-ocean and ice-atmosphere stress acting on sea ice |
54 |
|
by AREA according to Connelly et al. (2004), JPO. |
55 |
|
- For EVP and the LSR solver, the implicit part of the drag term is |
56 |
|
the only non-zero term in the denominator when there is no ice (which |
57 |
|
prevented running the model with zero ice-ocean drag). If |
58 |
|
SEAICEscaleSurfStress = .TRUE., this results in a division by zero |
59 |
|
(or zero main diagonals BU/BV) which need to be caught. A practical |
60 |
|
consequence is that for open water, the momentum equation reduce |
61 |
|
to trivially 0 = 0 + 0 (for EVP). For LSR, BU/BV are reset to 1, if |
62 |
|
they happen to be zero, often leading to a non-zero solution over |
63 |
|
open water. |
64 |
|
o pkg/seaice: revise EVP code (seaice_evp.F) in preparation for a more |
65 |
|
efficient EVP method. |
66 |
|
- introduce area-weighted averages for e12 (analogously to |
67 |
|
seaice_calc_viscosities.F); this totally changes lab_sea.hb87 (2 digits |
68 |
|
remain, but also changing the order of terms in the previous average |
69 |
|
changed the results as much) -> update experiment |
70 |
|
- add SEAICE_DELTA_SMOOTHREG code |
71 |
|
- disentangle computation of zetaC/deltaC and zetaZ/deltaZ, also for |
72 |
|
TEM-code |
73 |
|
- adjust loop bounds so that only required fields are computed |
74 |
|
- add new averaging code for zetaZ/deltaZ (again similar to |
75 |
|
seaice_calc_viscosities.F), inactive by default for now (will change |
76 |
|
results again), but will replace old code eventually |
77 |
|
o eesupp/src: |
78 |
|
- add an other alternative for multi-proc GLOBAL_SUM (in global_sum_tile.F) |
79 |
|
to always cumulate tile local-sum in the same order |
80 |
|
(option #define GLOBAL_SUM_ORDER_TILES in CPP_EEOPTIONS.h). |
81 |
|
Still uses MPI_Allreduce (like standard code) but on array of tiles |
82 |
|
and should be faster than GLOBAL_SUM_SEND_RECV code. |
83 |
|
- switch almost all verification experiments that use GLOBAL_SUM_SEND_RECV |
84 |
|
(defined in CPP_EEOPTIONS.h_mpi) to GLOBAL_SUM_ORDER_TILES instead. |
85 |
|
- add CPP_EEOPTIONS.h_mpi with GLOBAL_SUM_ORDER_TILES defined in 14 more exp.: |
86 |
|
aim.5l_cs, cheapAML_box, deep_anelastic, dome, fizhi-cs-32x32x40, |
87 |
|
fizhi-cs-aqualev20, fizhi-gridalt-hs, global_ocean_ebm, matrix_example, |
88 |
|
natl_box, solid-body.cs-32x32x1, tutorial_advection_in_gyre, |
89 |
|
tutorial_baroclinic_gyre and tutorial_held_suarez_cs. |
90 |
|
|
91 |
|
checkpoint65n (2015/07/29) |
92 |
|
o pkg/thsice: |
93 |
|
- add a warning when used with pkg/aim but without updating SST |
94 |
|
(stepFwd_oceMxL=F & useCoupler=F & tauRelax_MxL<>-1 ). |
95 |
|
o pkg/offline: |
96 |
|
- change where Offline-Fields are loaded, from load_fields_driver.F to |
97 |
|
directly within FORWARD_STEP, just after updating iter number and time; |
98 |
|
- move setting of GM and KPP diffusivity out of OFFLINE_FIELDS_LOAD into new |
99 |
|
S/R OFFLINE_GET_DIFFUS that is called towards the end of DO_OCEANIC_PHYS; |
100 |
|
These changes allow: |
101 |
|
a) to recover exactly on-line results using deltaToffline=deltaTClock, |
102 |
|
tested (using offlineIter0= nIter0 -1 and offlineTimeOffset= -deltaT/2) |
103 |
|
both with Lin & Non-Lin Free-Surf, both with staggerTimeStep & synchronous |
104 |
|
time-step, both with GM & KPP files and with recomputing these fields; |
105 |
|
b) no needs to switch on staggerTimeStep anymore; |
106 |
|
c) simplify rescaling of offline horiz. velocity with Non-Lin Free-Surf. |
107 |
|
- update output of experiment tutorial_dic_adjoffline (only 3 digits left). |
108 |
|
o pkg/offline: |
109 |
|
- change (rarely used) offset-iteration parameter to an offset time (in s): |
110 |
|
can easily recover previous offset-iter effect by setting offset-time |
111 |
|
(offlineTimeOffset) to offset-iter x deltaToffline; |
112 |
|
- do not turn-off Temp & Salinity surface relaxation when using KPP (used |
113 |
|
in KPP as part of surface buoyancy forcing) in offline run. |
114 |
|
- fix ptracers KPP non-local term in the case where KPP is re-computed. |
115 |
|
o OpenAD |
116 |
|
- Adding preliminary ggl90 test to OpenAD verification. |
117 |
|
Config. with ggl90 currently diverges after 64 time steps |
118 |
|
so will need more work to become more stable |
119 |
|
- Adding preliminary kpp test to OpenAD verification. |
120 |
|
required small tweakes in kpp_routines.F |
121 |
|
- Merge and update from Krishna Narayanan's contrib area: |
122 |
|
* Add one variable for preliminary OpenAD DIVA |
123 |
|
* Add replacement time-stepping for OpenAD DIVA |
124 |
|
* genmake2 flag -diva (but only for OpenAD; preliminary) |
125 |
|
* required modifs for OAD_support |
126 |
|
successful preliminary test for tutorial_global_oce_biogeo/ |
127 |
|
o pkg/seaice: |
128 |
|
- initialise deltaC in seaice_lsr, cosmetic changes in seaice_lsr |
129 |
|
- add cpp-flag SEAICE_DELTA_SMOOTHREG for option of regularising |
130 |
|
delta with a smooth function in s/r seaice_calc_viscosity |
131 |
|
(no effect on EVP) |
132 |
|
- rename local suffix "sqr" into "sq" for more consistent variable names |
133 |
|
o pkg/seaice: |
134 |
|
add new parameter SEAICE_deltaMin that is used ***only*** |
135 |
|
for regularizing Delta (and nothing else, like the parameter |
136 |
|
which is also used for all sorts of things SEAICE_EPS). |
137 |
|
Defaults to SEAICE_EPS for backward compatibility only. |
138 |
|
o pkg/seaice: replace erroneously storing e12Csqr on the wrong tape with |
139 |
|
the wrong key by inialising it before use -> fixes a recomputation |
140 |
|
problem but does not fix the gradient |
141 |
|
o pkg/layers: |
142 |
|
- move resetting of flux arrays to outside the iLa loop |
143 |
|
(this caused layers_diapycnal to not work with more than one layer coord.) |
144 |
|
- fix bug in calculation of layers_recip_delta in layers_init_fixed.F |
145 |
|
o pkg/layers: |
146 |
|
- remove from common block 3D layers fields (used to diagnose transport |
147 |
|
in tracer coords) and declare them as local arrays in layers_calc.F |
148 |
|
o pkg/seaice: |
149 |
|
- Rename SEAICE_VECTORIZE_LSR_ZEBRA to SEAICE_LSR_ZEBRA and define it |
150 |
|
in lab_sea forward experiment (changes results -> updated) |
151 |
|
- update adjoint experiment lab_sea after cleaning up seaice_lsr_tridiagu/v |
152 |
|
o pkg/seaice: modify seaice_lsr.F in order to improve the gradient |
153 |
|
computations; for SEAICE_VECTORIZE_LSR |
154 |
|
- move the loops over which the tridiagonal solvers (seaice_lsr_tridiagu/v) |
155 |
|
sweep to the outside, |
156 |
|
- remove store directives and add "CADJ loop sequential" directives |
157 |
|
in analogy to model/src/solve_tridiagonal.F |
158 |
|
- replace many "#ifdef SEAICE_VECTORIZE_LSR_ZEBRA" by variable loop |
159 |
|
boundaries to yield more readable code. This has also the charming side |
160 |
|
effect that your can use SEAICE_VECTORIZE_LSR_ZEBRA without |
161 |
|
SEAICE_VECTORIZE_LSR (i.e. adjoint without recomputations in |
162 |
|
seaice_lsr_tridiagu/v now requires either of these flags, vectorization |
163 |
|
still requires SEAICE_VECTORIZE_LSR with SEAICE_VECTORIZE_LSR_ZEBRA as |
164 |
|
an additional option) |
165 |
|
The tridiagonal solvers are now completely analoguous to solve_tridiagonal.F |
166 |
|
but the adjoint gradients (which are affected by this change) still explode. |
167 |
|
o coupler (atm & ocn _compon_interf): |
168 |
|
- avoid using Exch2 IO mapping if not suitable. |
169 |
|
|
170 |
|
checkpoint65m (2015/06/15) |
171 |
|
o pkg/layers: |
172 |
|
- clean-up old parameters setting (from Sep 2012): |
173 |
|
+ remove old parameters setting from common block (declared locally); |
174 |
|
+ stop if any old parameter is mixed with new parameters; |
175 |
|
+ check for valid or consistent parameter setting (e.g., layers_name) |
176 |
|
- check that layers_bounds for Pot.Density are in the right range; |
177 |
|
this is supposed to catch un-updated setting after recently shifting |
178 |
|
potential density field "prho" by -1000 (now in "sigma" scale). |
179 |
|
o pkg/layers: |
180 |
|
- Added missing call to layers_fill for keeping track of tottend in |
181 |
|
pkg/diagnostics/diagnostics_fill_state.F |
182 |
|
o pkg/layers: |
183 |
|
- added CPP flux LAYERS_FINEGRID_DIAPYCNAL to disable fine-grid binning |
184 |
|
of diapycnal fluxes |
185 |
|
o pkg/layers: |
186 |
|
- reverted changes to interpolation to previous treatment; there |
187 |
|
was never actually a problem |
188 |
|
o pkg/layers: |
189 |
|
- fix interpolation of layers tracers in the presence of partial cells |
190 |
|
- add lots of new functionality to compute water mass transformation |
191 |
|
(enabled by CPP flag LAYERS_THERMODYNAMICS) |
192 |
|
- add new calls from main code (diags_oceanic_surf_flux.F, |
193 |
|
thermodynamics.F, impldiff.F) and gad to save the necessary fields |
194 |
|
for layers_thermodynamics calculations |
195 |
|
o pkg/ptracers: |
196 |
|
- add capability of stepping a passive tracer forward without advection |
197 |
|
by setting PTRACERS_advScheme to 0. |
198 |
|
o pkg/diagnostics: |
199 |
|
- add few (threads) _BARRIER around any update of "diag_pkgStatus". |
200 |
|
o pkg/layers: |
201 |
|
- shift potential density field "prho" by -1000 for density layer averaging so |
202 |
|
that it corresponds to usual "sigma" definition (i.e., rho(p=const)-1000); |
203 |
|
Note that density bin boundaries (layers_bounds in data.layers) have |
204 |
|
to be set accordingly (i.e., shifted, in sigma scale). |
205 |
|
o pkg/layers: |
206 |
|
- fix snap-shot output of layer thickness @ V.pt and output of pot.density |
207 |
|
- fix calculation of layer non-weighted velocity and layer probability |
208 |
|
- add diagnostics for pot.density used for layer averaging |
209 |
|
o pkg/seaice: |
210 |
|
- seaice_calc_viscosity.F: change computation of deltaC**2 to ensure |
211 |
|
positiveness, modify a few comments and improve variable names, |
212 |
|
- the changed computation of deltaC affects most seaice-related |
213 |
|
experiments at the truncation level. |
214 |
|
seaice_itd.thermo is affected by far the most, with |
215 |
|
only 7 digits of agreement remaining; update experiments. |
216 |
|
|
217 |
|
checkpoint65l (2015/05/04) |
218 |
|
o pkg/shelfice: |
219 |
|
- in shelfice_thermodynamics.F, if SHELFICEBoundaryLayer=.true., then set |
220 |
|
factor drKp1 to zero where it is negative (ie. do not use |
221 |
|
SHELFICEBoundaryLayer if hFacC>1) |
222 |
|
- the same for factor drLoc in shelfice_forcing.F |
223 |
|
o tools/genmake2: move FEXTRAFLAGS from FFLAGS to FOPTIM to avoid a problem |
224 |
|
with modules in ptracers, when using testreport option "-small_f" |
225 |
|
o pkg/seaice/seaice_growth.F (in case of SEAICE_ITD) |
226 |
|
- replace tmpscal1**1.36 by faster exp(1.36*log(tmpscal1)) |
227 |
|
o verification/testreport |
228 |
|
- add new option -small_f |
229 |
|
with this option testreport first makes the target small_f (*.f files) |
230 |
|
before running make (all). This is useful for explicit inlining when |
231 |
|
all sources need to be full available. |
232 |
|
o pkg/exf/exf_interp.F: |
233 |
|
remove previous changes related to reverse (decreasing) input grids latitude |
234 |
|
|
235 |
|
checkpoint65k (2015/04/01) |
236 |
|
o pkg/ecco: |
237 |
|
fix missing cosphi initialization; use eccoiter in place of optimcycle; |
238 |
|
if autodiff is not compiled then use READ_REC_XY_RL/READ_REC_XYZ_RL |
239 |
|
instead of active read/write |
240 |
|
o pkg/ctrl: |
241 |
|
if autodiff is not compiled then use READ_REC_XY_RL/READ_REC_XYZ_RL |
242 |
|
instead of active read/write. |
243 |
|
o pkg/autodiff: make autodiff_readparms.F and autodiff_whtapeio_sync.F |
244 |
|
return if .NOT.useAUTODIFF |
245 |
|
o pkg/seaice: add CPP brackets (cost function related). |
246 |
|
o model/src: |
247 |
|
add useAUTODIFF run time switch. |
248 |
|
o pkg/exf/exf_interp.F: |
249 |
|
add code to allow input grids with latitude starting in the north (i.e. when |
250 |
|
j=1 corresponds to northern edge of field); modify global_with_exf to test |
251 |
|
this feature. |
252 |
|
o verification/testreport: |
253 |
|
- replace "grep -A3" with POSIX conformal sed command, because there are |
254 |
|
implementations of grep that cannot do contextual searches |
255 |
|
- replace "cat -n" by "nl" (two occurrences), because not all implementations |
256 |
|
of "cat" have the option "-n" |
257 |
|
o model/src: |
258 |
|
add new file gsw_teos10.F that contains 3 functions, more or less a copy |
259 |
|
of the gsw_toolbox.f90 of the TEOS-10 fortran interface, that are used to |
260 |
|
convert from conservative to potential temperature. For now they are not |
261 |
|
called anywhere. |
262 |
|
o pkg/ggl90: make "useIDEMIX" useful |
263 |
|
- re-organize code in s/r ggl90_calc a little to move computations that |
264 |
|
are different for idemix into separate loops (changes verification |
265 |
|
experiment idemix at truncation level, 12 digits remain -> updated) |
266 |
|
- requires 2 new 2D fields KappaM, verticalShear |
267 |
|
- make hFacI and recip_hFacI local 3D variables that are passed to |
268 |
|
s/r ggl90_idemix |
269 |
|
- fix and improve some comments |
270 |
|
|
271 |
|
checkpoint65j (2015/02/25) |
272 |
o pkg/openad |
o pkg/openad |
273 |
- synchronize seawater.F with version in model/src: replace routine |
- synchronize seawater.F with version in model/src: replace routine |
274 |
body of sw_temp by calling sw_ptmp and fix/add a few comments |
body of sw_temp by calling sw_ptmp and fix/add a few comments. |
275 |
o pkg/mnc: |
o pkg/mnc: |
276 |
- change default value of pickup_read/write_mnc params to FALSE |
- change default value of pickup_read/write_mnc params to FALSE |
277 |
(since few features are missing in reading MNC pickup files). |
(since few features are missing in reading MNC pickup files). |