wiki:Documentation/UserGuide/Config

Version 5 (modified by nvuilsce, 12 years ago) (diff)

--

Documentation for ORCHIDEE_OL scripts

When you have installed ORCHIDEE and ORCHIDEE_OL source code for trunk version as described in the course, you will find in "modipsl"/config/ORCHIDEE_OL tree some directories for scripts used to run ORCHIDEE in an off-line mode.

We strongly invite you to read other wiki and pdf documentation in addition of this one :

  1. ModipslBeginner
  2. Documentation for IPSLCM5 current release of IPSL coupled model gives a lot of links on other documentations.
  3. libIGCM users and developpers documentation.

Some of those documentations are still in french and deal with the global couple model of IPSL (namly IPSLCM5 for CMIP5 exercise). But ORCHIDEE scripts use the same philosophy for executions, tree of directories, etc...

Other documentations can be found in other page of this wiki site :

Remark : there is a copy of all ORCHIDEE script directories in model/ORCHIDEE_OL. This is for subversion management and it won't work correctly if you use those directories. You must use ones in config/ORCHIDEE_OL.

libIGCM scripts

libIGCM scripts are used as an SRE (Sytem Running Environment). They are used to manipulate input and output files for an executable (or a set of executables in the couple configuration).

  • libIGCM : Set of scripts to launch simulations and posttreat. Modularity and

portability make you jobs work on a lot of machine (super-computer) in the same way.

Basic ORCHIDEE scripts

ORCHIDEE for CMIP5 and trunk (beginning of 2011) has three off-line drivers :

  1. orchidee_ol that use all ORCHIDEE source (dim2_driver, SECHIBA, STOMATE and LPJ) and may run the full ORCHIDEE model or just a part of (juste the driver and SECHIBA with prescibe LAI).
  2. teststomate is the driver to run only STOMATE and LPJ part of the model (daily). It is used to accelerate the SPINUP of the vegetation (but without feedback to the physic : soil temperature, hydrology).
  3. forcesoil is the driver to accelerate the convergence of the soil carbon pools only.

There are four basic directories (configuration) to run ORCHIDEE in a simple mode :

  1. OOL_SEC: used to run orchidee_ol without STOMATE and LPJ (SECHIBA with prescibe LAI).
  2. OOL_SEC_STO : run orchidee_ol with the full ORCHIDEE model.
  3. TESTSTOMATE : run teststomate and only STOMATE and LPJ part of the model
  4. FORCESOIL : run forcesoil to converge the soil carbon pools.

Each of those directories contains :

  • a config.card file to describe the configuration (see JobName, LongName, TagName, CalendarType, ListOfComponents), the experiment specifications (!Restarts and !Post) and informatic parameters (parallelization, WriteFrequency).
  • in COMP directory, you may find "card" and "driver" files for each components list in ListOfComponents section of config.card.
  • in PARAM directory you may find default parameters files for each components. It gives a list of all parameters required by the model with the routine "getin". Those files contains only default parameters (documented). You may change them with care because it can give an eratic behaviour of the model and some parameters are changed automaticly by the script (see remark 2 below).
  • Finaly, POST directory gives configurations files for SECHIBA and STOMATE/LPJ component outputs. Please, see intranet wiki page on Monitoring, if you want to add your own variables in MONITORING web pages. See remark 3.

Some important remarks :

  1. The type of the calendar (CalendarType) and the dates (DateBegin and DateEnd) are only use by the scripts to get the right input (forcings, maps ...) files, but they are not passed to the model. For ORCHIDEE model, the forcing file must give all informations on the calendar.
  2. The options in the cards always gives priority parameters regards to those of the parameter files because the driver use those options to overwrite the parameter value in the last input parameter files in the RUN_DIR directory.
  3. You may add you own TS (TimeSeries) monitoring in your DODS web page for your simulations. For that, you must verify that the variable is correctly output (right histdef and histwrite are created and calls in your version of the source code). You must control that the variable is listed in the TimeSeriesVars? option (2D or 3D list depends on the dimension of the variables) in the Post_1M_"your_component_history_file" section. Then it will be automaticly created if you have add the right line in your POST component file (with cfg extension).

SPINUP job?

Spin-up are the algorithms to get the convergence of a part of the surface model. They are used to give correct initial state for historical simulations.

The equilibrium of the physic (SECHIBA) of the land surface should be reach in less than 10 years with a fix climatology and a prescribe LAI. You may take care of the year of the forcing file because Nino or Nina specific year may give bad results for global spin-up.

The equilibrium of the LAI (seasonnal variation of the vegetation) less than take two decade to converge. There is a big feedback between the LAI and the hydrology and energy budget.

But the soil carbon pools take thousand of years to converge, but there is no feedback between those pools and the LAI and the hydrology yet.

important remark :
This configuration use OOL_SEC, OOL_SEC_STO, TESTOMATE, FORCESOIL basic configurations as skeletons to launch each spinup stages, depends of the algorithm you have chosen.
You may not change those initial directories (or with great care) because it may change all your spin-up !

See page SpinUpJobs? for additionnal informations.

FLUXNET job?

FLUXNET forcing files give a performent tool to validate ORCHIDEE model on a grid of sites (one point for each simulation) all over the world. It give fast spinup and simulation on a various set of vegetation and climate distributions.

This configuration use SPINUP configuration as a skeleton to launch an ensemble of jobs over the selected sites in FLUXNET or LATHUILE input files.

For FLUXNET jobs, the vegetation repartition is prescribed. Then default parameter IMPOSE_VEG is automaticly set to 'y' and the veget_max must be given in the list for each sites.

See page FluxnetValidation? for a complete documentation on FLUXNET sites validations.

ENSEMBLE job

ENSEMBLE configuration give a generalization of FLUXNET sites script to run global ENSEMBLE spinup and experiments.

With this configuration, you can change the forcing files and all parameters (by vegetation - PFTs - or not) to build all ENSEMBLE experiments you want with ORCHIDEE model.

This scripts give the same tools as FLUXNET configuration, but there is no IMPOSE_VEG set (initial vegetation is not prescribed).