3. LFRic code and software#

In this section, LFRic is introduced from a software perspective, focusing on the LFRic Atmosphere model and its underlying components.

In this module you will learn about…

  • how LFRic Atmosphere fits within the wider LFRic software stack and the Momentum Framework;

  • how the LFRic codebase is split across LFRic Core, LFRic Apps, and larger science component repositories;

  • how the PSyKAl architecture and PSyclone separate scientific code from parallelisation, data movement, and performance concerns;

  • which development tools and repositories you need to work with LFRic source code, configurations, and tests;

  • where the LFRic working practices are documented, including guidance on issues, branches, testing, reviews, and multi-repository development;

  • where to find support when you are unsure how to apply the working practices;

  • how model output is controlled through XIOS and the iodef.xml configuration file.

At the end of this module you should be able to…

  • describe the main LFRic code repositories and identify which parts of the model they contain;

  • recognise the role of LFRic Apps, LFRic Core, science component repositories, dependencies, and interfaces in a model build;

  • explain, at a high level, how the algorithm, kernel, and PSy layers fit together in the PSyKAl approach;

  • find the appropriate working-practice guidance before creating issues, branches, tests, reviews, or multi-repository changes;

  • identify the support routes for questions about LFRic development and working practices;

  • follow the practical workflow for checking out code, compiling, running a simple configuration, making a small change, and using tests as evidence;

  • locate and modify iodef.xml to control which fields are written, at what frequency, and to which output files.

Platform and prerequisites#

The practical pages in this module assume access to a prepared Unix-like development platform that can build and run LFRic Atmosphere. They are not a from-a-clean-operating-system installation guide.

At the Met Office, use the managed LFRic training environment and load the supported module stack as shown in Practical 1: Run the model from command line:

ml use ~lfricadmin/lmod
ml lfric

On non Met Office platforms, follow your site’s LFRic Development Environment documentation to provide an equivalent environment before starting the practicals. The exact compiler and library versions are platform-specific and are owned by the supported LFRic build environment and repository dependency files rather than by this training course. The Non Met Office tabs in the practicals identify where local site instructions are expected.

The Met Office module stack selects the compiler, MPI implementation, Python tools, PSyclone, and supporting scientific libraries used by the practicals. On non Met Office platforms, the equivalent site-supported environment should provide those tools. Where a course uses Azure Spice, treat it as this prepared Met Office platform route rather than as a portable installation requirement.

Table 3 Tools and access expected by this module#

Requirement

Used for

How to obtain it

Unix-like shell environment

Checking out code, running commands, inspecting logs, and launching workflows.

Use the Met Office managed LFRic environment or an equivalent site-supported Linux or HPC environment.

Git and GitHub repository access

Cloning LFRic Apps, the example workflow, and any training branches.

Follow your site’s GitHub access guidance. Met Office and non Met Office tabs give the repository URLs used by each practical.

Fortran/C/C++ build toolchain, MPI, NetCDF/HDF5, and build utilities

Compiling the atmosphere executable and linked science components.

Use the lfric module stack at the Met Office, or the toolchain documented for your supported LFRic platform.

Python and PSyclone

Running build scripts and generating PSy-layer Fortran from LFRic source.

Use the supported LFRic build environment for your platform.

Rose, Cylc, and FCM

Running workflows, managing configurations, and running stem tests.

Use the Met Office module stack or your site’s workflow tooling installation.

MOSRS and, where required, SVN access

Working with shared Met Office repositories and legacy repository workflows used by some development practices.

Request access through your Met Office or partner support route before attempting those exercises.

Important

The development environment in this repository’s README is for building these training pages with Sphinx. It does not provide the full compiler, workflow, and runtime stack needed to build or run LFRic Atmosphere.

Before starting the practical exercises, check that your platform can clone the required repositories, load the LFRic build environment, and run git, python, rose, cylc, fcm, and the required compiler commands. If one of those checks fails, use your local platform documentation or support route before continuing.

Contents