Rust Simple Climate Model (RSCM)

RSCM is a framework for building reduced-complexity climate models in Rust and Python. Models are assembled from components that are coupled through named variables, and solved on a common time axis. It is also a testbed for modularising MAGICC and reimplementing it in Rust.

Design goals

• Fast
• Easy to extend
• Understandable by non-Rust developers
• Can be driven from Python

What ships

• A two-layer energy balance model (rscm.two_layer).
• Example components for a one-box carbon cycle, CO2 forcing and four-box ocean heat uptake (rscm.components).
• MAGICC7 components (rscm.magicc) for:
  • CH4, N2O and halocarbon chemistry.
  • The terrestrial carbon cycle with nitrogen limitation, ocean carbon, CO2 budget and permafrost.
  • GHG, ozone and aerosol forcing, and forcing aggregation with efficacies.
  • The upwelling-diffusion climate and sea level rise.
• rscm.magicc.full.run_magicc7, which runs the full MAGICC7 chain from emissions. It matches MAGICC7 v7.5.3 on its default SSP245 run.
• Simpler MAGICC runners for concentration-driven and emissions-driven experiments.
• MCMC calibration and point estimation (rscm.calibrate).

Getting Started

This project comprises a Rust workspace in crates/ and a Python package (rscm) which includes a Python extension module to provide access to the Rust library.

Dependencies

The following dependencies are required to build the project:

• Rust
• uv (Python package management)

After these dependencies have been installed the local Python environment can be initialised using:

make virtual-environment

Rust is a compiled language, so the Python extension module must be recompiled after any changes to the Rust code. This can be done using:

make build-dev

Tests

Rust unit tests are embedded alongside the implementation files. The rscm crate holds the Python bindings, so it is excluded when running the Rust tests directly:

cargo test --workspace --exclude rscm

make test rebuilds the extension and runs both the Python and Rust test suites.

Documentation

A mkdocs-based documentation site is in the docs/ directory. These docs provide an overview of the whole project and provide reference documentation for the Python interface with some example notebooks.

These docs can be built using:

make docs

The documentation for the Rust codebase can be built using:

make docs-rust