Design a ’param_’ file

For designing an experiment : exhaustive list of all parameters that can be used in param files

Article mis en ligne le 3 octobre 2012
dernière modification le 28 août 2015

par senesi, voldoire

Defining a CNRM-CM experiment is a simple as designing a so-called ’param_’ file, which filename must be of the form ’param_EXPID’, where EXPID is the name you want to give to your experiment. This file, in bash syntax, sets values for ECLIS parameters and then sources the ECLIS install script

This article assumes you are familiar with the introduction article

All existing parameters are listed here for full documentation. However, we suggest the user to use an existing, tested, param file corresponding to an experiment close to the expected one and only to modify few lines in the param file. There are also examples there and in the ECLIS directory

Parameters index

Basic common mandatory parameters

ARPEGE parameters
 Basic parameters
 Forcing parameters
 Nudging parameters

SURFEX parameters
 SURFEX Offline parameters

NEMO parameters
 Basic parameters
 Forcing parameters in ocean only mode
 Nudging parameters
 Post-processing paramters

GELATO parameters

TRIP parameters

GRISLI parameters

Parameters for coupled simulations

Parameters for the IO Server

Optional parameters

Advanced user parameters

In the following, when we give examples of parameter possible values, but we do not indicate e.g. actual path for files and directory as this information could be obsolete. Instead we only indicate if a directory or a file name is needed. The reader is refered to existing param files given in the introductory page

Basic common mandatory parameters

Parameter Possible value Explanation
EXPID set otherwise The experiment name is actually usually set by the name of the parameter file, which shoud match pattern ’param_EXPID’. It must not exceed 16 characters (thanks to Nemo)
INSTALLER file From V5.4 : mandatory name of the script for installing the experiment ; usual value : script ’install’ from $ECLIS/scripts/ directory. The last line of the param_ file must be source $INSTALLER ; executing the file will actually install/prepare the experiment
INIDATE 19900101 Start date of the simulation
ENDDATE 19910615 End date of the simulation (necessary but the end date can be modified afterwards)
RUNMAIL E-mail address where error messages are sent
MACH_NAMELIST tori where to find all namelists (tori, yuki, cougar or another machine reachable via rcp)
MACH_RESTART tori where to find all restarts (tori, yuki, cougar or another machine reachable via rcp)

ARPEGE parameters

ATMEXE file ARPEGE binary path
ATMREST file Atmospheric initial restart. The machine is indicated in the parameter MACH_RESTART.
ATMNAMREF file Atmospheric namelist. This namelist will be and duplicated by the installation procedure in the relance directory with the name fort.4 (ARPEGE required name) and updated for each run as explained in Note on Arpege namelists. The machine where to find it is indicated in the parameter MACH_NAMELIST.

NB : The consistency between the namelists and the paramfile is not checked at installation stage. Thus the namelist has to be correcly prepared. For instance, if the user wants to use SURFEX, the key LMSE has to be activated in the namelist ! (quoting a Surfex namelist file is not enough). Further, from ECLIS V5.10, you must set LLSM to 0 if you run Arpege without Surfex

ATMRESARCH directory Directory for saving ARPEGE restarts on cougar. See also EXPARCHDIR
ATMOUTARCH directory Directory for ARPEGE outputs on cougar. See also EXPARCHDIR
UPDCLI file Binary used to update the ARPEGE (and eventually SURFEX) restart(s) before each simulation (update forcing such as aerosols, SSTs, surface conditions). This step is done consistently with the ARPEGE namelist flags, see UPDLCI versions for more information.
UPDOZO file Binary to update the ozone coefficients in the ARPEGE restarts (each month).From V5.6, running without ozone is also possible
NHH file Period for FullPos post-processing (in hours). Will be used only if your namelist sets NFRPOS to AUTO and you use the standard script for updating Arpege namelist

ARPEGE Forcing parameters

For mixed gases for which only a global concentration is considered, an ascii file should be used with each line starting with a reference tag (usually the year) and followed by the concentration of the different species considered in ARPEGE. For example of such files, please refer to files in the directory ˜mrgu548/forcages/ghg_et_solaire_cmip5/ where CMIP5 data files are stored.

All other forcings are provided by external files in binary format and the different forcing state have to be in separate files (each time and forcing in a different file).

In all cases, the forcing can be fixed or can evolve. If it evolves, it can be updated on a monthly/yearly/decadal time scale and, from ECLIS V5.6, it can be cycled. For a given forcing, the generic path of the files used has to be given in the variable FORXXX, where XXX depends on the forcing. In the generic path, the string YYYY refers to the place where the year is written, the string MM refers to the month. Each time the forcing is updated, YYYY and MM are replaced by the proper values. If the FORXXX parameter does not contain any YYYY (MM resp), then the forcing will not evolve according to the year (month resp) of the simulation. If none of the two patterns is specified then the forcing is fixed and FORXXX contains the name of the file to use. In order to accomodate your data filenames, which may include some Ys or Ms, default strings "YYYY" and "MM" can be changed using parameters YEAR_PATTERN and MONTH_PATTERN (see advanced use).

If YYYY is present in FORXXX, then parameter YEAR_XXX indicates at which time scale the forcing evolves :
 to be updated on a yearly basis use SCEN_YEARDEB_YEARFIN (or SCEC_YEARDEB_YEARFIN). For instance, with YEAR_XXX=SCEN_1950_2000 the forcing will be updated each year from 1950 to 2000.
 To be updated on a decadal basis use SCED_YEARDEB_YEARFIN. In this case the forcing of year 1970 will be used from 1966 to 1975 (included).

In both cases, if the simulation starts before YEARDEB and/or ends after YEARFIN then the forcing will be fixed to YEARDEB before YEARDEB and to YEARFIN after YEARFIN. An alternative to SCEN_YEARDEB_YEARFIN is SCEC_YEARDEB_YEARFIN. With SCEC, before YEARDEB and after YEARFIN, a climatological file is used. This file must be named as indicated in FORXXX, but YYYY has to be repaced by CLIM.

 Cycling the forcings is possible with YEAR_XXX=SCYC_YEARDEB_YEARFIN, in which case the forcings of YEARDEB will apply to the first year of simulation (say FYEAR), and those of YEARFIN to simulated year FYEAR - YEARDEB + YEARFIN, after which a new cycle will apply YEARDEB forcings to the next year and so on... For handling Surfex experiments with spinup, see Running Surfex offline with a spinup
 If YEAR_XXX does not start with SCEN, SCED or SCEC, then it can contain a specific year and in this case, this fixed year is used along the simulation (identical as if FORXXX does not contain the pattern YYYY, but directly the desired year).



Surface fixed conditions (orography, + vegetation if not SURFEX)


Sulfate aerosols


volcanic aerosols


Ozone coefficients


Black carbon aerosols


Organic aerosols


Dust aerosols


Sea salt aerosols


Greenhouse gases + chore + solar constant
NB : For this forcing, the file has to contain all forcing values (for the different dates). In case you want to use a constant GHG forcing, YEAR_GHG has to be specified as the tag of the first colomn (ie generally the year) in the data file.

In case ARPEGE is not used in coupled mode, then the user has to specify an additional SST forcing (method similar to the other forcings)



SST forcing

Specific parameters to nudge ARPEGE (before ECLIS V5.4)

From ECLIS V5.4, nudging Arpege is performed through a standard plugin. Please see there for setting its parameters

ONUDG 1 Key to include nudging in the simulation.
NB : The key LNUDG must be set to .TRUE. in ARPEGE namelist.
NUDGMASK file Nudging mask (in case the user wants to nudge all over the globe, then do not specify any file and put "no" as the parameter value). NB : The key LNDRIV must be set to .TRUE. in ARPEGE namelist if you want to use this mask.
DIRNUD directory path on cougar where nudging files are stored.
INDNUD file Small script that calculates the NNN value to use according to the date of the simulation in nudging files.

SURFEX parameters

If running Surfex coupled to Arpege, do not forget to put the key LMSE=.T. in ARPEGE namelist

SFXNAMREF file SURFEX namelist. The machine is indicated in the parameter MACH_NAMELIST
SFXREST file SURFEX restart :
- for version up to 6 : without the extension ".fa" as restarts are composed of two files that should have the same name but one is .txt and the other one .fa.
- from version 7 : full name of Surfex restart (PREP). Can be a compressed (.gz) file
The machine is indicated in the parameter MACH_RESTART.
SFXRESARCH directory Directory for saving SURFEX restarts on cougar
SFXOUTPATT string Pattern for matching the diagnostic file that shoud go to SURFEX outputs archive ; defaults to "SURFOUT.*.fa ISBA_* SURF_ATM.*"
SFXOUTTAR 0 or 1 If set to 0, Surfex output will not be grouped in tar files before saving on archive achine. Defaults to 1
SFXOUTARCH directory Directory for SURFEX outputs on cougar
SFXPGD file From V5.7 : Surfex PGD file. Optional. Can be a compressed (.gz) file Defaults to PGD.fa.gz in the same directory as the restart file
LSFXPGD 0 or 1 From V5.7 : flag telling if your version of Surfex do read a PGD file (i.e. Sfx version >= V7.2). Can be omitted if SFXPGD or DATA_SFX is specified
DATA_SFX file pattern From V5.7 : pattern for files to be linked before Surfex launch (e.g. Ecoclimap files). Applies only for Surfex version from 7.2. Optional. Default value is decharme/ecoclimap_gmgec/ecoclimap*.bin
UPDNAMLSM file Script called to update Land Surface Model namelist, if necessary, at each model launch. Will be called as $UPDNAMLSM ${SFXNAM} $YEAR $MONTH $DAY $NDAY
LSFXCPL 1 Does Surfex handles the coupling ? default is 1 since Eclis 6.4
POSTSFX std processing Path for a user script for post-processing Surfex outputs. Such a script must : choose files to process in local dir (Eclis ensures that it contains only those Surfex outputs matching SFXOUTPATT parameter), echo (on stdout) the list of files to save on archive machine, striclty avoid to echo anything else on stdout : rather redirect it to stderr for getting feedback in job output - It receives EXPID, YEAR and MONTH as arguments - No need to clean up local dir - New from ECLIS V6.6
LDAYNUDG 0 Should updcli be provided with namelist key LDAYNUDG=.T.

SURFEX Offline parameters

From ECLIS V5.8, Surfex can be run in stand-alone (so-called ’offline mode’), and next parameters apply :

SFXEXE file SURFEX Offline binary. Can be parallel (MPI)
SFXSTEPOUT 86400. Period of diagnostics output (in seconds, floating point). If not set, if running Arpege and NFRSFXHIS is set to AUTO in Arpege namelist, will use Arpege value
NSPINUPS 10 Number of spinup cycles ; applicable when cycling forcing is required by YEAR_ATM (see below). See detailed article
LSFXSPINOUT 1 Should we save model diagnostics during the spinup period
YEAR_ATM YYYY or SCYC_... Pattern for evolution of atmospheric forcing ; see atmosphere
FORATM filename pattern Pattern for atmospheric forcing filenames (on cougar) ; see atmosphere
YEAR_SST YYYY or SCEN_... Pattern for evolution of SST forcing ; optional (depending on your Surfex binary) ; see atmosphere
FORSST filename pattern Pattern for SST forcing filenames (on same machine) see atmosphere
YEAR_SSS YYYY or SCEN_... Pattern for evolution of SSS (Sea Surface Salinity) forcing ; optional (depending on your Surfex binary) ; see atmosphere
FORSSS filename pattern Pattern for SSS forcing filenames (on same machine) see atmosphere
YEAR_SIC YYYY or SCEN_... Pattern for evolution of SIC Sea-Ice Cover forcing ; optional (depending on your Surfex binary) ; see atmosphere
FORSIC filename pattern Pattern for SIC forcing filenames (on same machine) see atmosphere
YEAR_SIT YYYY or SCEN_... Pattern for evolution of SIT Sea-Ice Thickness forcing ; optional (depending on your Surfex binary) ; see atmosphere
FORSIT filename pattern Pattern for SIT forcing filenames (on same machine) see atmosphere
SFXFRC file Directory (preferably on FTDIR) for storing atmospheric forcings. Optional. Useful in combination with SFXFRC_KEEP (see below). Defaults to $FTEXP/forcings
SFXFRC_KEEP 0 or 1 Should we keep the forcing files on $SFXFRC after use ? This may speed up other experiments. Defaults to 0 (i.e. do not keep)

NEMO parameters

OCEEXE file NEMO binary
OCENAMREF file NEMO namelist. The machine is indicated in the parameter MACH_NAMELIST.
IOXNAMREF file Namelist to manage NEMO output (useful with or without ioserver, but not for Nemo with old output scheme). The machine is indicated in the parameter MACH_NAMELIST.
IOSNAMREF file NEMO IOserver namelist. Automatically updated w.r.t. IO server launch. The machine is indicated in the parameter MACH_NAMELIST.
LOCEREST 0 ou 1 If 0, then the simulation starts from Levitus state and there is no need of an ocean restart.

If 1, the simulation starts from the restart file indicated in OCEREST

OCEREST file NEMO restart, only used if LOCEREST=1. The machine is indicated in the parameter MACH_RESTART.
OCE_REBUILD_ON_INSTALL 0 or 1 Put 0 if your Nemo restart is piecewise and you want to rebuild it for your simualtion (e.g. because it has a wrong number of pieces)
OCERESARCH directory Directory for NEMO restarts on cougar. See also EXPARCHDIR
OCEOUTARCH directory Directory for NEMO outputs on cougar. See also EXPARCHDIR
DATA_OCE directory Where input files for NEMO are stored (must be on the local machine where simulations are performed. It contains files that do not change for a given configuration (bathymetry, grid files, etc...)

Forcing data parameters in ocean only mode

Note : in forced mode, NEMO is run over years of 365 days (leap years are not considered - LEAP=0 in script_nemo).

For forcing variables, as for ARPEGE forcings, there are two parameters driving the forcing : one indicating how it evolves (YEAR_XXX, where XXX is the nickname for a variable, or a set of variables grouped in a forcing file), the other providing a pattern for the forcing data filenames (FORXXXGEN). In the case of ocean forcing, files are stored on the archive machine and FORXXXGEN must contain the full path on the archive machine. See the description in the section ARPEGE forcing for a documentation on the way FORXXXGEN and YEAR_XXX can be specified.

From V5.6, , in order to handle the various forcing configurations manageable by NEMO, none of the parameter describing forcing is mandatory ; also, you can freely choose which are the XXX nicknames, by listing them in the (string) parameter OCEFORCINGS. For these nicknames : U10 V10 Q2M T2M PCP SNW LWR SWR TDF SLP CSW CLW RNF, ECLIS will find in Nemo’s namelist which are the filenames under which Nemo expects to find the data. If you use other nicknames, or if your Nemo does not have such information on local filenames, you will have to set, for each of them, a parameter named FORXXX which tells how to rename the corresponding data file for Nemo (see relevant parameter)

In the file param_OCEAN you will find an example of use of these parameters.

FORCING directory where forcing files are locally stored - usually on FTDIR. If the user has several experiments that use the same forcing, this directory can be set to a shared location so as not to retrieve the same files in different directories (save ftget time and local disk space). The default is FTDIR/EXPID/FORCING
OCEFORCINGS "U10 V10" String listing those of the forcings described by a FORXXX that will actually feed Nemo. Default value is "U10 V10 Q2M T2M PCP SNW LWR SWR TDF SLP CSW CLW RNF". See above for explanations
YEAR_U10 FORU10GEN, FORU10TAR scenario patter and forcing data filename pattern for 10 meter zonal wind ; optional forcing tar file pattern, if forcing files are grouped in tar file (file fetching is smart enough to avoid duplicate fetches)
YEAR_V10 FORV10GEN, FORV10TAR 10 meter meridional wind
YEAR_Q2M FORQ2MGEN, FORQ2MTAR 2 meter specific humidity
YEAR_T2M FORT2MGEN, FORT2MTAR 2 meter temperature
YEAR_LWR FORLWRGEN, FORLWRTAR Long-wave net radiation
YEAR_SWR FORSWRGEN, FORSWRTAR Short-wave net radiation
YEAR_SLP FORSLPGEN, FORSLPTAR mean sea level pressure, this forcing is not used by default in NEMO, it has to be specified in NEMO namelist.
YEAR_CSW FORCSWGEN, FORCSWTAR Correction on short wave radiation, this forcing is not used by default in NEMO, it has to be specified in NEMO namelist.
YEAR_CLW FORCLWGEN, FORCLWTAR Correction on long wave radiation,this forcing is not used by default in NEMO, it has to be specified in NEMO namelist.

Nudging parameters for NEMO

Before ECLIS V5.4, nudging was only coded in the ocean-only configuration. From V5.4, nudging and lateral boundary conditions are handled through the plugin feature and apply also to the coupled case ; please refer to that page.

The following applies to pre-V5.4 ECLIS. Thereafter, similar parameters are used in the relevant plugin
As for NEMO forcing files, NEMO nudging files have to be given with their full path on the archive machine.

LDAMP 0 ou 1 Key to activate the nudging (retrieve nudging files and make them available when running the model). 1 is for nudging. Note that the NEMO namelist used has to be set accordingly.
FTUD directory where nudging files are locally stored - usually on FTDIR. If the user has several experiments that use the same nudging files, this directory can be set to a common directory so as not to retrieve the same files in different directories (save ftget time and local disk space). The default is FTDIR/EXPID/NUDG
NUDSOGEN 3D salinity reference data file
NUDTOGEN 3D temperature reference data file
NUDSOSGEN Surface salinity reference data file
NUDTOSGEN Surface temperature reference data file

Post-processing parameters for NEMO

A default post-processing of Nemo diagnostic files includes : fixing the time_counter, rebuilding the outputs with a given version of the rebuild script and binary, and tar-ing all outputs for one month in a signle tar file ; Next parameters allow to tune these aspects

POSTOCE path to a script Path to a postprocessing script. It receives EXPID, YEAR and MONTH as arguments. It must process files in current directory and output the file names to save on archive. See e.g. postoce_with_tar in directory toolbox
REBUILD path to a script Path to a an alternate rebuild script that the standard ocean post-processing should use in replacement for the standard rebuild. See script toolbox/ to see how it is called

Parameters for the IO Server

The IO Server is a separate process which use is optional, and which can handle NEMO I/O instead of having it handled by NEMO itself. As a by product, this allows to have NEMO outputs in one piece instead of assembling ("rebuilding") it after the run. For now, it has been successfully tested only on tori and in a configuration where it uses only one process. It is not mandatory if Nemo runs on one proc only

IOSEXE file name of an IO Server binary

GELATO or LIM parameters

From V5.6, ECLIS can run LIM instead of Gelato

LLIM 0 or 1 Are we using LIM instead of GELATO ? (from V5.6)
ICENAMREF file GELATO or LIM namelist. The machine is indicated in the parameter MACH_NAMELIST.
ICEEXE file GELATO stand-alone binary, if applicable.
DATA_ICE directory GELATO / LIM fixed data directory, if applicable.
LICEREST 0 or 1 LIM and Gelato can run without a restart. For that case, set LICEREST to 0.
ICE_NC 0 or 1 Put 0 if your Gelato uses NetCDF restarts. Useless if an initial restart is indicated
ICE_REBUILD_ON_INSTALL 0 or 1 Put 0 if your Ice restart is piecewise and you want to rebuild it for your simualtion (e.g. because it has a wrong number of pieces)
ICEREST file GELATO or LIM restart. The machine is indicated in the parameter MACH_RESTART.
ICERESARCH directory Directory for saving GELATO or LIM restarts on cougar See also EXPARCHDIR
ICEOUTARCH directory Directory for GELATO or LIM outputs on cougar. See also EXPARCHDIR
UPDNAMICE file Name of a script called to update Gelato or LIM namelist, each month. Optional. Will be called as $UPDNAMICE ${ICENAM} $YEAR $MONTH $DAY $NDAY $ICE_RESTART where ICE_restart= .false. for first month of simulation if LICE_RESTART=0

TRIP parameters

RIVEXE file TRIP binary
RIVNAMREF file TRIP namelist. The machine is indicated in the parameter MACH_NAMELIST.
RIVREST file TRIP restart. The machine is indicated in the parameter MACH_RESTART.
RIVRESARCH directory Directory for TRIP restarts on cougar See also EXPARCHDIR
RIVOUTARCH directory Directory for TRIP outputs on cougar. See also EXPARCHDIR
DATA_RIV directory Directory where grid files for TRIP are stored (must be on the local machine)
UPDNAMRIV file Optional. Script called to update River model namelist, if necessary, at each model launch. Will be called as $UPDNAMRIV ${RIVNAM} $YEAR $MONTH $DAY $NDAY
POSTRIV file Optional. Script called to post-process River model outputs. It must echo the list of files to save and will be called as $POSTRIV $EXPID $YYMMDATE $DATEF

For GRISLI paramter, see Adding a model in coupled or forced mode

Parameters for coupled simulations

CPLEXE file OASIS binary
CPLNAMREF file OASIS namelist. The machine is indicated in the parameter MACH_NAMELIST.
CPLATMRES file OASIS atmospheric restart. The machine is indicated in the parameter MACH_RESTART.
CPLOCERES file OASIS oceanic restart. The machine is indicated in the parameter MACH_RESTART.
CPLRIVRES file OASIS runoff (TRIP) restart. The machine is indicated in the parameter MACH_RESTART.
CPLRESARCH directory Directory for OASIS restarts on cougar
DATA_CPL directory Directory where grids and weights data files are stored (must be on the local machine)

Optional parameters

TITRE string Long name for the experiment, written in the history file
CREDAS string Name a of directory holding a CNRM-CM Reference Data Set. See CREDAS definition and Easy experiment design
NMONTH 12 Number of months simulated in a macro-job. This parameter can be tuned for various reasons. Ensure consistency with requested CPU (see parameter ELAPS),and on e.g. vargas, with available space on WORKDIR
INITIME 0 Hour for starting the simulation. Default is 0 ; used to be 12 in ancient times
Period on which restarts are saved.
  • 12 : restarts saved once a year (default)
  • 1 : restarts saved every months
  • ’all’ : all restarts written by ARPEGE are saved (the number then depends on parameter NFRHIS in ARPEGE namelist)
  • ’m15’ : restarts are saved twice a month (on the first and fifteenth of the month)
GROUP group1/subg2 path segment for organizing outputs ; allow to group experiments in file system hierarchies ; new with V5.4, yet used only on workstations by tool EM ; you shuld use it also when setting RELDIR
EXPARCHDIR string Allows to simplify the specification of outputs hierarchy, by setting default value of ATMOUTARCH to EXPARCHDIR/atm/output, ATMRESARCH to EXPARCHDIR/atm/restart,and so on for sfx, oce, ice and riv.
NEXTEXP PICTL Name of an experiment that should be launched (using relan) at the end of the current experiment ; useful for experiment showing a dependency either physically or due to ressources constraints
EXP_END_CMD string A command string which will be executed at the end of the experiment. e.g ; launch the install and run of a another experiment. You can get a mail using "echo last date processed was \$DATF | mail -s '$EXPID successful' $RUNMAIL"

Advanced user parameters

script file Optional from ECLIS V5, which can guess it from the binaries and namelists parameters. Used to tell which script to use to run the model. Scripts are different for forced ARPEGE, NEMO and coupled run. Default value will be set as the script relevant to your case, from the same directory than the install script. You may provide you own script, but please also consider using the plugin feature if you need few changes
TOOL directory Tool Script directory (common for all scripts and configurations). Defaults to ECLIS subdirectory ’toolbox’
OTHER_FILES a string with file paths The files listed are additional, user-defined fixed input files for the run ; they will be copied to the experiment relance dir at install stage, under a basenmae derived from the parallel list in parameter OTHER_FILES_BASENAMES if provided, or otherwise with an unchanged basename ; later on, they will be linked from the run directory at each run loop
OTHER_FILES_BASENAMES a string with file basenames You may list here, parallel to OTHER_FILES, the basenames under which each additional input file wil be made available to the run
ASCEXE directory Optional path for a script or binary (of your own) used to complement Arpege output restart with some Surfex output restart fields ; Will be called as ’$ASCEXE surfex_restart arpege_restart’ on all archived restarts
UPDNAMARP directory Path for an alternate, user-defined script that will update Arpege namelist, each month. See Note on Arpege namelists
NN 1 Number of nodes used on the vector machine. Not used with LoadLeveler (i.e. on vargas). Default is 1.
NNP 3 Number of processors used on each node on the vector machine. Not used with LoadLeveler (i.e. on vargas). Default is 4.
NPROC_ARP 24 Indicate the number of processors to be used by ARPEGE.
NPROC_OCE 96 as NPROC_ARP for NEMO. Default is depednant on the cofniguration (coupled / forced). NB : until NEMO V3.3, you have to make sure that your NEMO binary fits NPROC_OCE.
NPROC_LSM as NPROC_LSM for stand-alone Land SUrface Model
NPROC_ICE as NPROC_ARP for stand-alone ICE model
NPROC_RIV as NPROC_RIV for the river routing model
NPROC_CPL as NPROC_CPL for the coupler
ELAPS 05:00:00 CPU time for computing on vector/parallel machine for each macro-job.
MEM 30gb Maximum memory used on vector/parallel machine.
STACK 30gb Maximum stack memory used. Needed only for vargas.
ELAPSFRONT 01:00:00 CPU time for managing file on the front-end / scalar machine (common to step 1 and step 3, queue ft on NEC).
MEMFRONT 8gb Maximum memory used on the front-end machine (queue ft).
QUEUE vector On NEC, name of the NQS queue on which the computing job is sent (vector/test/giec). Not used on IBM
RUNDIR Directory used for running the simulation. Used if KEEPTMP=yes, instead of using a $TMPDIR directory that the user can not see and/or that is cleaned after the job. This option is useful for debugging (default is yes from ECLIS V6.1)
KEEPTMP 1 If 1, execute the job in the directory $RUNDIR/$EXPID .
KEEPFT 0 If 1, does not erase the diagnostic files afetr archiving (keep it in the file transfer directory) ; you should then clean it up manually ; however, a periodic system clean up may occur based on file age .
STORE_LOC $FTDIR Directory where outputs and restarts are saved during a computing step before being archived to cougar by the front machine (step 3 of each macro-job).
YEAR_PATTERN %YEAR% String used in forcing filename patterns, which will be replaced by a year number. Defaults to YYYY.
MONTH_PATTERN MOIS String used in forcing filename patterns, which will be replaced by a month number. Defaults to MM.
RELDIR $HOME/relances/$EXPID Directory for automatic relance. Namelists and scripts output are posted here. A good practice is to set it to $HOME/relances/$GROUP/$EXPID (see GROUP above)
LICE 0 Is there a sea-ice model ? Default is 1 (for : true) ; option exists from V5.6
LLIM 0 Are we running LIM-2 sea-ice model ? Default is 0 (false). If value is 1, LICE is automatically set to 1. Option exists from V5.6
LATMOUT 1 Key to desactivate the storing of ARPEGE outputs. Default is 1 if 0, then ARPEGE outputs are not saved.
LOCEOUT 1 As LATMOUT for NEMO outputs.
LRIVOUT 1 As LATMOUT for TRIP outputs.
The default is not to store output listings from models. Only the last one is kept in the relance directory (ex : last_NODE .001_01 for ARPEGE, last_ocean.output pour NEMO). If the user wants to save all listings, put this key to "all". It is also possible to save all listings for one model only in coupled mode, in this case put the capital letter corresponding to the model in the parameter (A for ARPEGE...). The directory where these listings are store is indicated via the parameter DIR_LISTING.
DIR_LISTING directory The default is to save listing in the directory listings in the relance directory. But the user can specify another directory on the local machine that will be created during the installation.
SAVE_CPL_FILES 1 Allows to save weights file created by OASIS. Default is 0. If OASIS weights files are not already saved in DATA_CPL, then OASIS will generate these files and this parameter allows the user to save them (this calculation takes lots of time and should be done only once).
FRCPL 86400 Coupling period in seconds (default is 86400). NB : there are bugs to use this functionnality in pure CMIP5 configuration, the user has to swich to a newer version of instexp and HOM_TOOL. ([see this specific page<-#install_version]
ATMNAM, ATMRESIN, ATMRESOUT, ATMLISTOUT Name of namelist, input restart, output restart, listing as expected by the atmospheric model. These parameters should not be changed unless using a new ARPEGE version used new file names.
RESTARTS_BUFFER_LENGTH 4 When running with limited space on FTDIR (e.g. on vargas), one may want to clean restarts there. But if the simulation crashes, and only yearly restarts are saved, one may want to have a ’buffer’ of monthly restarts, in order to restart it from one of the latest ones ; Ths parameter defines its length (in months)
ARCHIVING BEFORE By defautlt, ARCHIVING=BEFORE, and transfering model(s) outputs to the archive machine occurs at the end of each macro-job (every NMONTH). In order to speed up the simulation, one may set ARCHIVING to DURING, which means that next model integration will be launched in parallel with the archiving step. However, archiving of the next integration will stop if the previous archiving was not successful ; and the user will then have either to spoil computation time by restarting the experiment at an early date, or to be expert in steps relaunch.
LCOMPAT_64 0 or 1 Version 6.4 of ECLIS introduced a break in compatibility for easing the definition of runs which includes the MCT version of Oasis and version 2 of Trip, and coupling by Surfex rather than Arpege. Set this parameter to 1 if you use this Eclis version (or a later one) with old model versions. See the install script if you want to get finer control
FORU10, FORV10, FORQ2M, FORT2M, FORPCP, FORSNW, FORLWR, FORSWR, FORSLP, FORCSW, FORCLW Ocean forcing local filenames as expected by the ocean model. From ECLIS V5.6,, these parameters are namelist-dependant and seeked in the namelist and should not be set anymore. If you use other nicknames than U10, Q2M ... for forcing variables, say XXX, you must define FORXXX. Note : to not mix FORU10 (local names for Nemo) with FORU10GEN (data filename on archive), etc.
NUDSO, NUDTO, NUDSOS, NUDTOS Before V5.4 : Ocean nudging filenames expected by the ocean model. These parameters are model binary dependent (and not forcing data files dependent)
From V5.4 : similar parameters are used by the standard plugin oce_lbc_nudging
To change the filename pattern for the nudging data files fetched from archive machine, you should rather refer to parameters NUDSOGEN, etc.
YEAR_PATTERN, MONTH_PATTERN yyyy, aaaa, mois In forcing filename patterns, you may include placeholders which will be replaced by year and month of actual models run as necessary. Default values for these placeholders are "YYYY" and "MM". You may change it using parameters YEAR_PATTERN and MONTH_PATTERN

Dans la même rubrique

le 31 janvier 2019
par senesi
Note on Ozone
le 20 novembre 2013
par senesi
Note on Arpege namelists
le 5 août 2013
par senesi
Moving outputs to archive media
le 3 octobre 2012
par senesi
How to make a simulation ?
le 3 octobre 2012
par voldoire ,