MODIPSL is a tool developed (in Korn Shell and Python) by IPSL
modelling pole engineers with the aim of providing a common access and
a common interface to each of the IPSL different models.
MODIPSL is structured around two main building blocks that propose :
- A working environment common to the different models and which can be deployed on any type of platform.
- A set of standard commands facilitating the use and functioning of the models.
This set of commands is based on the following principles :
- Mutualisation of the acces to the different model's sources
- Adaptation of these commands, to a given platform, in order to generate the appopriate executable files.
The common environment takes form through a uniformed directory
structure nonetheless respecting the specific characteristics of each
model, from the Coupled model IPSL-CM4
- IPSLCM4_v1 - (LMDZOR, ORCALIM and
ORCHIDEE). Note that prior to using the commands on a given platform, the functional validation of a
model, on this platform, is required. Currently, installation and
functioning of the Coupled model have been validated on the NEC SX5 at
IDRIS and on the Fujistu VPP5000 at CEA. Tests are currently
undergoing for moving the Coupled model on the NEC SX6 from CEA.
Considering the above framework, MODIPSL enables the extraction,
installation, compilation and execution of any model but also the
analysis of its output data.
Prior to using the MODIPSL
functionalities the following installations are required:
- CVS software
- netCDF library compiled with real*8
- Fortran 90 compiler
- Python software
- The CDAT library of Python
- Ferret software
- NCO and netCDF operators
MODIPSL must be extracted from a CVS server and then installed on the given platform.
This fitting provides the necessary tools for the
extraction and installation of the desired model.
Note that the mod.def file contains the description of all the
information concerning each IPSL model. Note also that the model
command extracts from one or several CVS servers :
So in order to access the ORCHIDEE sources on the CVS server, you need a password.
If you whish to obtain this password, please fill-in the form below and send it.
Then install the model components which name has been passed as
argument. In addition, the
model utility offers some maintenance functions for the installed models.
Once the installation is finished, compiling requires the use of the
ins_make command. It enables the installation and configuration of
the models makefiles dependent on the working platform. Before executing
the model, the final step consists in configuring the desired
simulation. This is done by editing physical parameterisation and
launching files:
- Activating or not physical parameterisation.
- Specifying time limits and simulation memory needs.
- Defining start time, end time and output frequency.
- Activating or not post-treatment flags.
Installing the submission job of the fully parameterized simulation
requires using ins_script command. This command can also be used to
configure the job depending on the simulation platform as well as
assign a name to the simulation. The following command series
illustrates the above-defined steps as well as the simulation
configuration. The example is given for the platforms rhodes (SGI2100)
and uqbar (NEC SX5) at the IDRIS enter. It is also possible to compile on a station
and to launch the simulation on another calculator. Note also that in
order to launch a simulation, the access to input files stored on
IDRIS or CEA file servers, or a DODS server is mandatory.
CNRS - Institut Pierre Simon Laplace
Université Pierre et Marie Curie
4, place Jussieu
casier 101
75252 PARIS Cedex 5
FRANCE
Feedback:
The W3C Validator Team
