Standard Compile Environment
- SCE Workflow (dashed/solid arrows: process/data flow)
The SCE workflow
The workflow in the SCE includes five steps:
- the model is made SCM compliant
- the script and necessary source code is retrieved from the repository
- (the model is modified)
- the Makefiles are updated
- the compile scripts are created
- the compilation is launched.
There are no constraints on the model source code besides the directory structure and the single-module/single-file rule.
Thus the critical work is the porting of the model to the platform envisaged. Running the model in the SCE then implies an upgrade of the SCE script code with the compiler options for the model on the platform.
Modularity
The infrastructure IMDI supports modular design of ES models. Components of coupled models can not only be the traditional ocean, atmosphere, sea ice, land surface, marine or air chemistry models, as defined by PRISM, but any other software package representing a subsystem of the ES (e.g. cloud or runoff scheme). The figure below shows some ES coupled models combined from the MPI-M's (Max Planck Institute of Meteorology) component models of the physical atmosphere, ECHAM5, the ocean, MPIOM, the marine bio-geo-chemistry, HAMOCC, the surface model, JSBACH, and of the aerosol model HAM.
- Earth System models assembled from the MPI-M component models.
Compilation
The SCE is a toolkit which can be used to assemble scripts for compilation of a specified coupled model on a specified platform from a common base of script code.
All scripts needed, i.e. for the library compilation and for the compilation of all component models, can be created in one step. Since they are all built from the same base of script code, the user is given a common look&feel with all models on all platforms interfaced with the SCE. The toolkits are designed in a way to allow an easy extension of the infrastructure to accommodate new models and platforms.
The system can be set up as a central installation where some components of the system are built initially and shared by different users (e.g. libraries and the coupler executable). Or, it can be run as a local installation, where a user builds all components herself. Depending on what is locally maintained, during the compilation process all targets are checked whether they need to be updated. I.e. when an executable is made, all submodel which are assembled into the executable and all libraries linked to either the main or any of its submodels are updated. The compilation process reacts on changes of cpp flags in a way that ensures a recompilation of the minimum only.
A coupled ES model can be made up of more than one executable (each possibly containing more than one component) in which case the information exchange between the executables is by the PRISM coupling software, which was also developed in the PRISM project. Both versions of the PRISM coupling software, OASIS3 and OASIS4 are supported, as well as single-executable (standalone) models.
For details please consult the 
PRISM SCE handbook.