The atmosphere model of the CliMA Earth System Model: a GPU-capable global atmosphere model designed for calibration with data assimilation and machine learning.

ClimaAtmos.jl solves the compressible equations of atmospheric motion on cubed-sphere and column grids, with physics parameterizations for turbulence and convection (EDMF), cloud microphysics, and radiation. It is built on ClimaCore.jl and runs on CPUs and GPUs from a single codebase.

Documentation
Version
License
Tests
Code Coverage
Downloads

Condensed water path from a global simulation initialized with ERA5 on 8-31-25 00Z. Output every 30 minutes; ran for ~4 days.

• Global and single-column configurations: cubed-sphere grids for global simulations, column grids for parameterization development and testing
• Turbulence and convection: eddy-diffusivity mass-flux (EDMF) schemes, designed for calibration with data assimilation and machine learning
• Cloud microphysics: 0-moment to 2-moment bulk schemes via CloudMicrophysics.jl
• Radiation: RRTMGP radiative transfer
• GPU support: runs on CPUs and NVIDIA GPUs from the same codebase
• Composable configuration: script and YAML-config interfaces for every aspect of a simulation

ClimaAtmos.jl is a registered Julia package (recommended Julia: v1.11):

using Pkg
Pkg.add("ClimaAtmos")

The simplest simulation uses all defaults — it solves the dry compressible Euler equations on a global cubed-sphere grid from a hydrostatically balanced state:

import ClimaAtmos as CA

simulation = CA.AtmosSimulation{Float32}(; t_end = "1days")
CA.solve_atmos!(simulation)

Every aspect of the simulation can be customized through keyword arguments, for example a single-column model:

grid = CA.ColumnGrid(Float32; z_elem = 30, z_max = 30000.0)
simulation = CA.AtmosSimulation{Float32}(; grid, t_end = "6hours")

See Your First Simulation in the documentation for a guided introduction.

• Stable docs — equations, parameterizations, configuration reference, and API
• Dev docs — latest development version
• Available diagnostics — output variables

ClimaAtmos.jl is a component of the CliMA Earth System Model:

• ClimaCore.jl — dynamical core and discretization tools
• ClimaCoupler.jl — coupling to ocean, land, and sea ice components
• Thermodynamics.jl — moist thermodynamics
• ClimaParams.jl — centralized, calibratable model parameters

If you're interested in contributing to ClimaAtmos, we welcome contributions of any size! Let us know by opening an issue if you'd like to work on a new feature.

Contributors should follow the shared CliMA engineering standards in docs/dev-guides/, which cover architecture, performance, code quality, documentation, and workflows. These are vendored from CliMA/DeveloperGuides. The repo's AGENTS.md is a starting point for AI agents with repo-specific guidance. See also the contributor's guide.