SIParCS 2026 - Obin Sturm

Simulating atmospheric chemistry in a Julia-based climate model

Recorded Talk

Atmospheric composition is an essential part of the Earth system, with chemical processes determining the lifetime of potent greenhouse gases, the extent of the UV-protecting ozone layer, the ability of clouds to form, and the quality of the air we breathe. Computational models of atmospheric composition are essential tools used in air quality forecasts and climate scenarios and form a key part of Earth system models. The CliMA (Climate Modeling Alliance) initiative is a modern Earth system model written in the Julia scientific programming language that leverages recent computing advances, including machine learning and simulations on GPUs. However, the atmospheric component of CliMA does not yet include any representation of the chemistry that determines atmospheric composition. In this work, we build a representation of atmospheric chemistry in CliMA using the Multi-Scale Infrastructure for Chemistry and Aerosols (MUSICA). MUSICA facilitates flexible representation of chemistry, where different chemical mechanisms are chosen based on use case.  We maintain this flexible mechanism-agnostic feature in CliMA, allowing different chemical mechanisms to be chosen depending on the use case and processes of interest (e.g. equilibrium chemistry for testing, ozone photochemistry, sulfate aerosol formation, among others). We demonstrate the chemistry-enabled CliMA with a MUSICA backend in both a single column model setup as well as global simulations. Beyond the capability to represent chemistry, this work also contributes increased flexibility of the CliMA code base to handle new process variables programmatically.

Mentors: Anna Jaruga, Kyle Shores, Jian Sun

Slides and poster