This study integrates machine learning and particle‐resolved aerosol simulations to develop emulators that predict submicron aerosol mixing state indices from the Earth system model (ESM) simulations. The emulators predict aerosol mixing state using only quantities that are predicted by the ESM, including bulk aerosol species concentrations, which do not by themselves carry mixing state information. We used PartMC‐MOSAIC as the particle‐resolved model and NCAR's CESM as the ESM. We trained emulators for three different mixing state indices for submicron aerosol in terms of chemical species abundance (χa), the mixing of optically absorbing and nonabsorbing species (χo), and the mixing of hygroscopic and nonhygroscopic species (χh). Our global mixing state maps show considerable spatial and seasonal variability unique to each mixing state index. Seasonal averages varied spatially between 13% and 94% for χa, between 38% and 94% for χo, and between 20% and 87% for χh with global annual averages of 67%, 68%, and 56%, respectively. High values in one index can be consistent with low values in another index depending on the grouping of species and their relative abundance, meaning that each mixing state index captures different aspects of the population mixing state. Although a direct validation with observational data has not been possible yet, our results are consistent with mixing state index values derived from ambient observations. This work is a prototypical example of using machine learning emulators to add information to ESM simulations.
3 September 2020 in Aerosol Science and Technology
This study evaluates the error that is introduced in quantifying observed aerosol mixing states due to a limited particle sample size. We used the particle-resolved model PartMC-MOSAIC to generate a scenario library that encompasses a large number of reference particle populations with a wide range of mixing states quantified by the mixing-state index χ. We stochastically sub-sampled these particle populations using sample sizes of 10 to 10 000 particles and recalculated χ based on the sub-samples. The finite sample size led to a consistent overestimation of χ, with the 95% confidence intervals ranging from –70 to 30 percentage points for sample sizes of 10 particles, and decreasing to ±10 percentage points for sample sizes of 10 000 particles. These findings were experimentally confirmed with single-particle measurements from the Pittsburgh area using a soot-particle aerosol mass spectrometer.
Quantifying aerosol impacts on climate is an inherently multiscale problem since macro-scale impacts are determined by processes on the micro-scale. This poses unique modeling challenges, since these microscale processes lead to a continuously evolving aerosol mixing state, which is difficult to represent in large-scale models. This presentation will show how high-detail particle-resolved simulations can be used to predict the evolving aerosol mixing state on the regional scale. In contrast to traditional aerosol models that use bins or modes to represent the aerosol, the particle-resolved approach uses individual computational particles that evolve in size and composition as the particles undergo aging processes in the atmosphere. This approach is therefore not limited by assumptions about particle composition within a given size range and can represent the full aerosol mixing state without simplifying assumptions. I will show results that illustrate the spatio-temporal evolution of aerosol mixing state, going beyond the traditional definitions of “externally” or “internally” mixed populations. I will conclude with a framework to synthesize a picture of the ambient aerosol from models and observations. This focuses on suitable metrics to quantify mixing state and sampling strategies to determine these metrics that are accessible for both models and observations. Together, these provide a unique opportunity for “getting the right answer for the right reasons”.
1 August 2026 in IEEE Transactions on Nuclear Science