3.0 Execution of model runs

Given that the model was installed by the user (look at chapter 5.0) and that it is clear, which phenomenon is to be simulated, there is a whole set of questions which must be answered before executing the actual application run, e.g.:

The answers to these questions substantially determine the choice of values for the most important model parameters, with which the model run is steered. All possible parameters are described in chapters 4.1, 4.2 and 4.3. The basic steering of the model with these parameters is explained in the introduction to chapter 4. The user will rarely use all model parameters, since many are preset with consistent default values and therefore do not have to be changed by the user. Some of the most important parameters are not preset with default values and must be adjusted by the user in each case. Such a typical, minimum parameter set is described in chapter 4.4.1. For the subsequent analysis of model runs, graphical visualization of model data is particularly important. Chapter 4.5 describes, how such outputs are produced with the model.

The selected parameters and their values have to be supplied to the model as an input file. The structure of this file is described in the introduction to chapter 4. The data, produced by the model, are located in a number of different output files. The model works with relative file names (i.e. without absolute path information). Thus all input and output files are expected and put in the directory, in which the actual model (the executable program created from the FORTRAN source code) is started (but also see next paragraph). All user-relevant input and output files are described in chapter 3.4.

The actual model (the executable program) is not started directly via the user, but with the help of the shell script mrun,which can be called by the user. This script takes over the entire pre- and postprocessing work, which is necessary for a model run. In addition to this, it generates e.g. automatically batch jobs for model runs on remote computers, provides for the appropriate file transfers of the input and output files to these computers, secures data on existing archives systems, starts if necessary restart jobs, and much more. mrun allows to specify individual path names for each simulation and for the different types of input/output files. The start of model runs with mrun is described in chapter 3.1.

The implementation of user code extensions in the model is made possible by a set of defined interfaces. You will find references to the programming of such extensions in chapter 3.5. Frequently errors arise during the execution of the model after such changes, which can often be found only by interactive model steering and error tracing (debugging), which are described in chapter 3.6.




Last change: $Id: chapter_3.0.html 97 2007-06-21 08:23:15Z kanani $