Report of activity between 15/12/2008 and 15/05/2009

Tests

Tested till :

  • Tested 10 years
  • NEMO trunk version revision 1407
  • libIGCM trunk version revision 110
  • calendar type 360 : we can do restart every year or every month, and we can have output data every year or every month
  • calendar type noleap : we can do restart every year or every month, but we can have output data only every month ( because ioipsl need to have in input number of days per month ).

( I'm waiting for a tag of libIGCM )

Modifications at the code :

1. "Dynamic monitoring"

We can say "dynamic monitoring" monitoring of variables choosen by user (see : https://forge.ipsl.jussieu.fr/nemo/wiki/private/libIGCM/postprocessingpage#HowtoactivatedynamicMonitoring25022009 )

changed files :

monitoring01_opa9_ORCA2.cfg
monitoring01_lim2_ORCA2.cfg

( in which, coherently whit I want to do I put same variables of timeseries that I put in opa9.card )

monitoring.job

( in which I can't use ${FER_ATLAS} path, but I put monitoring01_opa9.cfg and monitoring01_lim2.cfg in directory of submission of the job )

opa9.card
lim2.card

( in which I put timeseries of variable that I want to monitoring.
NOTE : if I don't do it monitoring won't be available, because in monitoring.job there is an "exit if timeseries does not exist". )

( NOTE : tested with libiGCM tag version libIGCM_v1_2, revision 50. Now monitoring is changed a lot, I tested almost all but when there is a new tag I'll do another test)



2. "common accounts"

To store boundary conditions and forcing files we can use common account of ipsl ( see : https://forge.ipsl.jussieu.fr/nemo/wiki/private/libIGCM/common_account#Usingcommonaccounttostoreboundaryconditions )

changed files :

add in ~comptecommun/dmf/IGCM/BC/OCE/

all directories that correponds to existing configurations : ORCA2_LIM2
ORCA2_LIM2_PISCES
IPSLCM5
IPSLCM4_v1
IPSLCM4_v1_OASIS3
IPSLCM4_v2
IPSL_ESM_V1

and ORCA2_LIM2/v3
and ORCA2_LIM2/v3_1

to separate different boundary conditions, linked to NEMO version ( versions differ only for bathymetry now )

added:

~commonaccount/dmf/IGCM/STORAGE/BC/OCE/NEMO/ORCA2_TEST/

waiting for Arnaud Caubel test ( then it will be named only ORCA2 )



3. "new cards"

To use common accounts and to be coherent with old version of opa we changed nemo.card in opa9.card ( like suggested by Sébastien Denvil )

changed files :

opa9.card
opa9.driver
in config.card "Tag_Name" ( ORCA2_LIM it was replaced by ORCA2_LIM2 )



4. "like coupled"

After a meeting of engenieers and researcher to point out direction for IPCC we decided that "NEMO forced" has to have same output data, monitoring and atlas of coupled ( IPSLCM5 ).
So I add same variables of coupled used in monitoring and in ZNL atlas. ( NOTE : I add these variables also because if I want to use monitoring.job by default I can't do it; because in monitoring.job there is an "exit if timeseries does not exist", so I'm obliged to have same timeseries if I want to use monitoring!)

5. "new diaptr"

To have same variables, monitoring and atlas of coupled model ( IPSLCM5 ) I had to use diaptr.f90 ( see : https://forge.ipsl.jussieu.fr/nemo/wiki/private/libIGCM/ZNLatlaspage#MinistartguidetousediaptrinMPImode ). After changes I see that ZNL atlas had not same profile whit mono and MPI version, so Claude understood that there was a problem in mpi part of diaptr.F90. ( https://forge.ipsl.jussieu.fr/nemo/ticket/421 )

changed files :

diaptr.F90



6. "interannual"

Test whit noleap calendar type, that means that we have to do restart every month, so that means that we have to interpolate in January of current year of simulation whit December of previous year of simulation, and in December of current year of simulation whit January of next year of simulation. ( Note : documentation ( see : http://www.nemo-ocean.eu/About-NEMO/Reference-manuals NEMO_book_v3_1.pdf ) is not correct, after that all tests are ok I give to Sébastien Masson all changes, but he already knows, he's not enough time, I note all and then he will change ).

working in progress; problems now : it is stopped at first time step, probably boundary conditions files are not good (? to big velocity? )

probably are not good 6hours data, now we build a 1-day datas and run again

7. "monitoring comparison"

To compare two simulation in mono and MPI version ( like suggested by Marie-Alice Foujols ) it is useful to use "fegg" utility ( see : http://igcmg.lsce.ipsl.fr/monitoring/ ).
I didn't know that to use it there are 3 parameters in namelist that I've to change ( like suggested by Arnaud Caubel ).

changed files :

namelist

( changed parameters :

nbit_cmp = 1
n_cla = 0
nsolv = 2

)

I compared two run but in monitoring I've some difference, and this is to be investigated, I think first of all with Claude, to understand if all technical steps are ok, and then whit Sebastien (?) to understand if all scientific steps are ok.

changed also :

BB_make.ldef ( removed key_tradmp )
namelist : ln_rnf=.false.
namelist : ln_ssr=.false.



8. "monitoring & atlas"

All tests lead to the fact that we can done run of NEMO whit libIGCM, whit calendar of 360 or 365 days. And we arrive to have monitoring and atlas of simulation.
But "oceanographers" are not satisfied by their colors, their scales, so they said that in this way monitoring and atlas are not useful.
After meeting of Juliette Mignot we understood that Patrick Brockmann does not change monitoring for Ocean component. So now we are at the point in which:
we can leave in this way and to analyse data for run IPCC atlas and monitoring it will not be used ,
or we can work on it.
To work on monitoring and atlas to customize them we have to do modifications and then send all to patrick that is charged to put it in libIGCL and to see all dependencies.

9. "gallery"

I changed size of images in gallery in NEMO web site, but now (26/05/09) I see them again too small! I don't know now, probably there was something changed that I don't know. ( to monitoring it )
Added possibility to download pdf image.

Open points

* we have to point out inter annual runs :

what works : opa can find surface boundary conditions' files for December of previous year in January of current year, and for January of next year in December of current year. This is means that opa can do a correct interpolation.

what does not work :

  • at first time step opa is stopped because abs(U) is greater then 20 we have still to understand why
  • it doesn't read in a correct way input data (t2,q2,u10,v10) probably because data at 6hours ( ongoing )

* monitoring comparison :

what works : we can have monitoring comparison.

what does not work : monitoring comparison is not good for some variables; like: ( see : http://igcmg.lsce.ipsl.fr/monitoring/tmp/fegg_plot01_r5te4X_prod/ )

  • ice concentration north
  • north Atlantic deep water
  • Antarctic bottom water
  • sea surface salinity
  • net downward heat flux
  • sea surface heigh ( global )
  • sea surface temperatue ( global )
  • now changed tradmp key, and ln-rnt and ln_ssr parameters ( ongoing )

* monitoring and atlas :

what works : we can have default monitoring and atlas

what does not work : if we want to have different colors, different scales, write levels in atlas we have to work on monitoring files. But these files are in common accounts, and so we have to work in a way that is useful also for IPSLCM5. And in this moment we don't have expertise on it so we will take more time then Patrick.

* to test with calendar type : leap ( not having priority now )

not tested at all.

To see in more detailed way all developing see also https://forge.ipsl.jussieu.fr/nemo/wiki/private/libIGCM

Last modified 11 years ago Last modified on 2009-05-27T16:39:38+02:00