Skip to content

bacy (short for basic cycling) is a collection of scripts that run assimilation cycles with ICON. It is modular in structure with each module performing steps like computing the dynamical model evolution (more), data assimilation (core), surface, snow and sea surface analysis.

BACY_1.0

This document gives you a short introduction into the usage of BACY_1.0. A detailed userguide and further documentation is available at git@gitlab.dkrz.de:dace/dace_doc.git under dace_doc/org/dace_bacy.git and is continuously being updated.

How to get BACY_1.0?

BACY_1.0 is under version control using git. To obtain BACY_1.0 from the remote repository git@gitlab.dkrz.de:dace/dace_bacy.git the following steps on your local machine are necessary:

Text Only
1
2
3
git clone git@gitlab.dkrz.de:dace/dace_bacy.git <my_bacy_name>
cd <my_bacy_name>/modules
./copy_defaults.sh
If you want to get access to the DKRZ-gitlab, someone at DWD or DKRZ needs to approve, maybe someone from the DA or Climate group can help you.

How to use BACY_1.0? (Overview)

Modules

BACY_1.0 pursues a MODULAR CONCEPT. Each MODULE is an self-sufficient unit. The only contact points to its environment are realized by INTERFACES. These are the directories iodir/input and iodir/output which are part of each module.

The prep_<module> and save_<module> scripts take care of PROVIDING the module with input data from and of STORING its output data to a central place, respectively. In this way the COMMUNICATION between modules is realized. Unless the input directory is not filled with input data, the module itself cannot be run. Otherwise, the module can start its work and it uses its working directory run.

The behaviour of a module can be controlled via its INPUT ARGUMENTS during the <module> script call and via its PARAMETERS defined in the configuration file <module>_conf.sh. Parameters which are needed by more than one module are outsourced to the COMMON configuration file common/bacy_conf.sh.

For a more detailed module-setup please read modules/module/README. This module is a dummy module and serves as TEMPLATE to help you to create your own modules. In particular, in its README file you can find some valuable hints which rely on NAMING CONVENTIONS, classification of variables, etc., which are IMPORTANT to maintain the modular concept.

In summary: to run a module stand-alone, the following steps are necessary.

Text Only
1
2
3
prep_<module> [-h] [-d] arg1 ... argn
<module>      [-h] [-d] arg1 ... argn
save_<module> [-h] [-d] arg1 ... argn

The option [-h] gives you an explanation of the usage of the script and the option [-d] calls the script in a dry-run, meaning that no jobs will be submitted.

A cycle

The module cycle plays a special role and cannot be understood as a module in the usual sense (see above). The main difference is the absence of the input, output and run directories. The purpose of cycle is the provision of an environment in which the order of individual module calls can be controlled. Thus, this is the central place to arrange all modules to set up an experiment.

In summary: to start an experiment, the following steps are necessary.

Text Only
1
2
prep_cycle [-h]      arg1 ... argn
cycle      [-h] [-d] arg1 ... argn

For the options [-h] and [-d] see the explanations above.

Default files

The configuration files <module>_conf.sh as well as namelists and job scripts are not under version control. Instead we use default-files, e.g., <module>_conf.sh.default. This has the advantage that subsequent calls of git pull will not change the current configuration.

The script copy_defaults.sh creates the normal files from the default versions. This must be done whenever you clone a fresh BACY from the repository!

How to make your own developments on BACY_1.0?

The BACY_1.0 MAIN DEVELOPMENT BRANCH is called bacy-dev. This branch provides the BACY_1.0 REFERENCE which has to be the BASIS for ALL users! Only a few people can push to bacy-dev to keep this branch away from accidental or not quality proven commits.

If you want to contribute to a NEW FEATURE for BACY_1.0, first of all create a new branch <project>-<feature>-<team>, which you can setup as remote to share results inside your team.

During your development it is IMPORTANT to regularly pull from the bacy-dev branch to keep your basic BACY_1.0 up to date.

Once you have FINISHED your contribution and you have tested, tested and again tested your work and you are sure that you match all naming conventions for BACY_1.0 (see above), then you are ready to contact one person responsible which merges (together with you) your contribution into bacy-dev.


Last update: December 7, 2023
Created: December 7, 2023