SIParCS 2025 - Angela Pak

Advancing Atmospheric Chemistry Workflows in Python

Recorded Talk

MusicBox is a component of MUSICA (Multi-Scale Infrastructure for Chemistry and Aerosols), a flexible, Python-based framework designed to simulate atmospheric chemistry across spatial scales. While the broader MUSICA project includes global modeling of large-scale atmospheric processes, it is unique in its ability to resolve chemical processes at scales relevant to emissions and exposure. MusicBox serves as a streamlined, user-friendly entry point to this system, offering a simplified interface to simulate atmospheric chemistry as a box model.
In their current states, both MUSICA and MusicBox lack instructions for new users to get a grasp of them. Thus, to support wider adoption and usability, we enhance MusicBox with comprehensive documentation and interactive Jupyter Notebook tutorials that walk through the workflow from the simplest example to customized workflows with increased complexity. These learning resources incorporate practical examples, including Latin Hypercube Sampling for exploring input variability, strategies for parallelizing simulations across multiple grid cells, and training different machine learning models for a given chemical system. As users become more familiar with the MUSICA system and their modeling needs grow in scale and complexity, these resources will help ease the transition to more advanced applications. In addition, we introduce a community discussion forum to encourage knowledge sharing, support, and ongoing feedback. Together, these efforts aim to strengthen the MusicBox and MUSICA user experience, expand their accessibility, and support scalable, high-performance atmospheric chemistry modeling within the broader MUSICA ecosystem.

Mentors: Jian Sun, Kyle Shores, Matthew Dawson

Slides and poster