Microphysics box model intercomparison

Predicting how stratospheric aerosol injection (SAI) will work in practice requires accurately modeling the evolution of sulfur dioxide (SO₂) plumes released by aircraft. Current climate models simplify this injection, assuming SO₂ instantly mixes throughout large grid cells (roughly 100 km across) rather than evolving in the narrow, concentrated plumes that aircraft produce. Because these concentrated aircraft plumes evolve through tightly coupled chemical, dynamical, and physical processes, the resulting aerosol size distribution (the range of particle sizes the plume produces) ultimately dictates both the radiative efficacy (cooling potential) and stratospheric residence time of the particles.

Because observational data for aircraft-injected SO₂ in the stratosphere does not exist, research cannot yet definitively validate a single aerosol model. Different aerosol microphysics models (simulations of particle-level interactions) predict different aerosol properties from the exact same injection scenario, sometimes differing by a factor of two or more in predicted cooling. To address this gap, this research activity funds a systematic aerosol box model intercomparison to quantify the structural uncertainties between competing models such as Modal Aerosol Module (MAM), Community Aerosol and Radiation Model for Atmospheres (CARMA), Global Model of Aerosol Processes (GloMAP) and TwO-Moment Aerosol Sectional (TOMAS).

While aerosol microphysics schemes directly simulate nucleation (the formation of new particles), condensation (particle growth as vapor collects on them), and coagulation (particles merging together), these mechanisms remain highly sensitive to surrounding chemistry (oxidation) and plume dynamics (dilution). Oxidation and dilution indirectly, but significantly, shape the microphysical outcomes. By comparing models across a range of realistic conditions, this project will isolate systemic differences in how models represent these coupled processes and will clarify which processes — nucleation, condensation, and coagulation — govern the final aerosol size distribution. These findings will directly inform the design and interpretation of future physical field experiments. This research contributes to understanding the aerosol size distribution uncertainties.

This research activity ports a single box-model framework into Python, building on the TwO-Moment Aerosol Sectional (TOMAS) model originally developed by (Adams and Seinfeld, 2002). This publicly available tool enables users to simulate how SO₂ plumes evolve and transform into aerosol particles in the stratosphere. While this framework provides a baseline for understanding plume dynamics at the microphysical level, the broader community lacks a systematic evaluation of variability across different aerosol microphysics models.

Building on this foundation, this research activity executes a systematic intercomparison with MAM, CARMA, and GloMAP to evaluate how uncertainties in key environmental parameters affect aerosol evolution and the resulting radiative impact of SAI. The project examines the sensitivity of size distribution evolution and radiative forcing to background hydroxyl radical (OH) concentrations, background aerosol size distributions, temperature, pressure, dilution rates, and initial SO₂ concentration. Model comparisons span realistic injection scenarios at one and two weeks post-release, testing both a clean stratospheric background and conditions where SAI has already occurred. Each model simulation relies on plume modeling efforts that provide realistic initial conditions and physical evolution constraints.

As an extension beyond the core intercomparison, the project incorporates vertically coupled layers into select models to represent aerosol settling and atmospheric transport more completely. This capability captures expected differences in particle size distributions between aerosols near the injection altitude and those that have settled to lower stratospheric levels. These enhancements align box-model outputs more closely with how global climate models represent transport and microphysics.