Learning Aerosol Coagulation Dynamics in Latent Space with a Scale-Covariant Neural ODE
Abstract
Particle-resolved aerosol models preserve the joint size-composition structure that governs cloud activation, optical properties, and freezing, but their computational cost limits their use in large-scale atmospheric models. We extend AeroMELD from a compact representation of aerosol populations to a prognostic model of coagulation. The resulting AeroMELD-Coag advances nine learned coordinates for population shape and one for total particle number through a scale-covariant neural ordinary differential equation that incorporates the known concentration dependence of coagulation. Across 2,000 held-out trajectories spanning 48 h, median symmetric errors are 3.5% for latent shape and 0.52% for total number. This ten-coordinate state also retains the evolving size-resolved composition and associated cloud-condensation-nuclei (CCN) activation, optical properties, and frozen fraction. Compared with the evaluated 20-bin sectional model with 320 prognostic coordinates, AeroMELD-Coag lowers pooled mean CCN activation error from 1.98% to 1.38% and frozen-fraction error from 1.72% to 0.56%. In a matched integration benchmark, GPU integration time per trajectory, amortized over a batch, is more than three orders of magnitude lower than for both single-core CPU references. These coagulation results provide a proof of concept for advancing aerosol microphysics efficiently in a compact learned state while retaining information about aerosol mixing state. They establish a foundation for future large-scale atmospheric models to represent aerosol microphysics with greater mixing-state detail within practical computational budgets.
Related
- Interpretable machine learning quantifies composition and size influences on aerosol spectral absorption
- Reconstructing particle number size distributions via aerosol history–informed multiple deep learning models
- Aerosol vertical distributions shaped by boundary layer dynamics in a coastal urban environment: insights from the TRACER campaign
- Resilient Stratospheric Aerosol Injection is Cooperative
- Automated small-scale devices and robotic systems for analyzing the cellular impact of ambient aerosol particles
BibTeX
@misc{tang2026learning,
title = {{Learning Aerosol Coagulation Dynamics in Latent Space with a Scale-Covariant Neural ODE}},
author = {Wenhan Tang and Ruqi Yang and Jeffrey H. Curtis and Ehsan Saleh and Lekha Patel and Peter A. Bosler and Nicole Riemer and Matthew West},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.31271},
eprint = {2609.31271},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.31271},
}
RIS
TY - UNPB
TI - Learning Aerosol Coagulation Dynamics in Latent Space with a Scale-Covariant Neural ODE
AU - Wenhan Tang
AU - Ruqi Yang
AU - Jeffrey H. Curtis
AU - Ehsan Saleh
AU - Lekha Patel
AU - Peter A. Bosler
AU - Nicole Riemer
AU - Matthew West
PY - 2026
DA - 2026/09/25
PB - arXiv
DO - 10.48550/arxiv.2609.31271
UR - https://arxiv.org/abs/2609.31271
AB - Particle-resolved aerosol models preserve the joint size-composition structure that governs cloud activation, optical properties, and freezing, but their computational cost limits their use in large-scale atmospheric models. We extend AeroMELD from a compact representation of aerosol populations to a prognostic model of coagulation. The resulting AeroMELD-Coag advances nine learned coordinates for population shape and one for total particle number through a scale-covariant neural ordinary differential equation that incorporates the known concentration dependence of coagulation. Across 2,000 held-out trajectories spanning 48 h, median symmetric errors are 3.5% for latent shape and 0.52% for total number. This ten-coordinate state also retains the evolving size-resolved composition and associated cloud-condensation-nuclei (CCN) activation, optical properties, and frozen fraction. Compared with the evaluated 20-bin sectional model with 320 prognostic coordinates, AeroMELD-Coag lowers pooled mean CCN activation error from 1.98% to 1.38% and frozen-fraction error from 1.72% to 0.56%. In a matched integration benchmark, GPU integration time per trajectory, amortized over a batch, is more than three orders of magnitude lower than for both single-core CPU references. These coagulation results provide a proof of concept for advancing aerosol microphysics efficiently in a compact learned state while retaining information about aerosol mixing state. They establish a foundation for future large-scale atmospheric models to represent aerosol microphysics with greater mixing-state detail within practical computational budgets.
ER -