We are an alliance of scientists, engineers, and applied mathematicians building a new generation of Earth system models (ESMs), entirely written in Julia. Our mission is to pioneer a new, data-informed and physics-based approach to climate modeling.

We are developing a model that:

• Learns from the wealth of Earth observations from space and from the ground.
• Runs efficiently on modern computing architectures with GPUs.
• Delivers actionable predictions, from sub-seasonal to decadal scales, to inform critical decisions in risk management and infrastructure and resource planning.

Our goal is to set a new standard in the accuracy, usability, and scalability of climate models.

Core Model Components

• ClimaAtmos.jl: An atmosphere model leveraging data assimilation and machine learning for modeling and calibrating subgrid-scale processes.
• ClimaLand.jl: A modular land surface model capable of standalone or integrated simulations (e.g., hydrology, canopy, snow, and soil CO2).
• ClimaCore.jl: The dynamical core of the atmosphere and land models, providing discretization tools to solve their governing equations.
• Oceananigans.jl: A fast and flexible dynamical core of the ocean model, solving the fluid dynamical equations on CPUs and GPUs.
• ClimaSeaIce.jl: Coupled and stand-alone simulations of sea ice freezing, melting, and horizontal motion for Earth system modeling.
• ClimaCoupler.jl: The coupler bringing atmosphere, land, ocean, and sea ice models together for global climate simulations (CMIP/AMIP).
• CloudMicrophysics.jl: A library of cloud microphysics and aerosol parameterizations for high-performance climate simulations.
• RRTMGP.jl: A Julia implementation of Radiative Transfer for Energetics (RTE) and the RRTMGP optics.

Shared Physics & Numerics

• ClimaTimeSteppers.jl: GPU-ready IMEX and multirate ODE solvers tailored for large-scale scientific models.
• Thermodynamics.jl: Moist thermodynamic formulations shared across the CliMA ecosystem.
• SurfaceFluxes.jl: Calculates surface exchange fluxes between model components (e.g., land-atmosphere, ocean-atmosphere).

Calibration & Uncertainty Quantification

• CalibrateEmulateSample.jl: A machine-learning-accelerated pipeline for calibration and uncertainty quantification of computationally expensive models.
• EnsembleKalmanProcesses.jl: Optimization and approximate uncertainty quantification algorithms based on Ensemble Kalman methods.
• ClimaCalibrate.jl: Calibration pipelines for CliMA model components.
• ClimaParams.jl: A single source of truth for physical constants and tunable model parameters across the CliMA ecosystem.

Tooling & Diagnostics

• ClimaComms.jl: A unified interface for diverse computing devices and distributed environments (CPUs/GPUs, single-process/MPI).
• ClimaUtilities.jl: Shared utilities for the CliMA project, supporting, for example, time management and MPI-safe artifact downloading and data handling.
• ClimaDiagnostics.jl: A framework to define and output observables and statistics from CliMA simulations.
• ClimaAnalysis.jl: An analysis library to post-process and visualize ClimaAtmos simulations and NetCDF files.

Each repository contains extensive documentation, including examples and tutorials. Visit the documentation links in individual repos to get started, or explore our website at clima.caltech.edu for an overview of the project.

A full list of publications is available at clima.caltech.edu/publications.

The Climate Modeling Alliance is a coalition of partners from Caltech, MIT, and NASA's Jet Propulsion Laboratory (JPL).

Our current work is generously supported by:

• Schmidt Sciences
• U.S. National Science Foundation (NSF)
• U.S. Office of Naval Research (ONR)
• Resnick Sustainability Institute (RSI)

Learn more about our mission, our team, and our science at clima.caltech.edu.