Learning to Simulate Aerosol Dynamics with Graph Neural Networks
Open Access
- 14 July 2025
- journal article
- Published by American Chemical Society (ACS) in ACS ES&T Air
- Vol. 2 (8) , 1426-1438
Abstract
Aerosol effects on climate, weather, and air quality depend on characteristics of individual particles, which are tremendously diverse and change in time. Particle-resolved models are the only models able to capture this diversity in particle physiochemical properties, and these models are computationally expensive. As a strategy for accelerating particle-resolved microphysics models, we introduce Graph-based Learning of Aerosol Dynamics (GLAD) and use this model to train a surrogate of the particle-resolved model PartMC-MOSAIC. GLAD implements a graph-network-based simulator (GNS), a machine learning framework that has been used to emulate particle-based fluid dynamics simulations. In GLAD, each particle is represented as a node in a graph, and the evolution of the particle population over time is simulated through learned message passing. We demonstrate our GNS approach on a simple aerosol system that includes the condensation of sulfuric acid onto particles composed of sulfate, black carbon, organic carbon, and water. A graph with particles as nodes is constructed, and a graph neural network (GNN) is then trained by using the model output from PartMC-MOSAIC. The trained GNN can then be used for simulating and predicting aerosol dynamics over time. Results demonstrate the framework's ability to accurately learn chemical dynamics and generalize across different scenarios, achieving efficient training and prediction times. We evaluate the performance across three scenarios, highlighting the framework's robustness and adaptability in modeling aerosol microphysics and chemistry.Keywords
All Related Versions
Funding Information
- Pacific Northwest National Laboratory
This publication has 29 references indexed in Scilit:
- Chemical Insights into the Molecular Composition of Organic Aerosols in the Urban Region of Houston, TexasACS ES&T Air, 2024
- Quantifying structural errors in cloud condensation nuclei activity from reduced representation of aerosol size distributionsJournal of Aerosol Science, 2024
- Confronting the Challenge of Modeling Cloud and Precipitation MicrophysicsJournal of Advances in Modeling Earth Systems, 2020
- Aerosol Mixing State: Measurements, Modeling, and ImpactsReviews of Geophysics, 2019
- The super‐droplet method for the numerical simulation of clouds and precipitation: a particle‐based and probabilistic microphysics model coupled with a non‐hydrostatic modelQuarterly Journal of the Royal Meteorological Society, 2009
- Simulating the evolution of soot mixing state with a particle‐resolved aerosol modelJournal of Geophysical Research: Atmospheres, 2009
- Measurement of the mixing state, mass, and optical size of individual black carbon particles in urban and biomass burning emissionsGeophysical Research Letters, 2008
- Model for Simulating Aerosol Interactions and Chemistry (MOSAIC)Journal of Geophysical Research: Atmospheres, 2008
- Global anthropogenic aerosol direct forcing derived from satellite and ground‐based observationsJournal of Geophysical Research: Atmospheres, 2005
- Global estimate of aerosol direct radiative forcing from satellite measurementsNature, 2005