Estimating the radiative effect and constraining the free parameter space of BrC aerosols in GISS ModelE

Maegan DeLessio, Columbia University/NASA GISS, PhD Candidate

Brown carbon (BrC) is an absorbing organic aerosol, primarily emitted through biomass burning, that exhibits light absorption unique from both black carbon (BC) and other organic aerosols (OA). Despite many field and laboratory studies seeking to constrain BrC properties, the radiative forcing of BrC is still highly uncertain. To better understand it’s climate impact, we introduce BrC to the One-Moment Aerosol (OMA) module of the GISS ModelE ESM. We assess ModelE sensitivity to primary BrC processed through a unique chemical aging scheme, as well as secondary BrC formed from biogenic volatile organic compounds (BVOCs). Initial results show BrC typically contributes a TOA radiative effect of 0.039 W/m2. This addition of prognostic BrC to ModelE allows for better physical representation of organic aerosols with no apparent trade-off in model performance: evaluation of simulated optical depth against AERONET and MODIS retrieval data indicates similar skill as before when it comes to global means. Additional AERONET retrieval of BrC properties, as well as in-situ flight measurements, can be used to evaluate our scheme on a regional scale and constrain BrC parameters.

View PDF of Presentation          Watch Recording of Presentation

Idealized particle-resolved large-eddy simulations to evaluate the impact of emissions spatial heterogeneity on CCN activity

Samuel Frederick, University of Illinois Urbana-Champaign, Graduate Student, Department of Atmospheric Sciences

Regional-scale aerosol models experience structural uncertainty due to multiple factors, including the selection of aerosol representation and the spatial heterogeneity (SH) of primary emissions and gas-phase precursors. In this work we investigate the uncertainty due to these two factors on the prediction of cloud condensation nuclei (CCN). We use the particle-resolved model PartMC coupled to the Weather Research and Forecasting model configured for large-eddy simulations. The resulting modeling framework resolves turbulence-chemistry interactions and aerosol evolution on a per-particle scale, including gas phase chemistry, particle coagulation, and gas-particle partitioning. The sensitivity of CCN activity to emissions SH is evaluated for sulfate and ammonium nitrate precursors. To investigate a range of precursor gas and primary aerosol emission scenarios, the spatial collocation between emitted species is varied by modifying the spatial frequency of emissions. The particle-resolved treatment allows direct calculation of particle hygroscopicity and activation making it a useful platform to measure the structural uncertainty in coarser resolution aerosol representations. CCN concentrations are compared against results from simulations using the modal MAM3 model at coarser grid resolutions. This work is a first step in quantifying structural uncertainties imposed by the choice of aerosol representation and spatial resolution and their interaction.

View PDF of Presentation          Watch Recording of Presentation

Application of the hyperdual-step method in the Community Multiscale Air Quality Model (CMAQ) for the assessment of aerosol formation from volatile chemical products (VCPs)

Jiachen Liu, Drexel University, Research Assistant

The Community Multiscale Air Quality (CMAQ) model is utilized for studying the physicochemical processes in the atmosphere using meteorology and emissions as inputs. The model has helped researchers and policymakers to comprehend the complexities associated with aerosol formation processes. In policy-related scenarios, the first- and second-order partial derivatives of output variables, such as criteria pollutant concentrations, with respect to input variables, such as emissions, are often of interest to ascertain expected changes due to emissions control strategies. Several methods have been employed to address this challenge, such as the higher-order direct decoupled method (CMAQ-HDDM) and the adjoint method (CMAQ-adjoint). We recently developed a novel, augmented version of CMAQ (CMAQ-hyd) capable of computing numerically exact first- and second-order sensitivities of all modeled species concentrations with respect to select emissions. The EPA developed new emissions profiles to account for emissions from VCPs (Seltzer et al., 2021). We exploit the new emissions profiles and employ CMAQ-hyd to calculate sensitivities and evaluate the health impacts of VCP-related aerosol formation across the continental United States. This study demonstrates the novel application of the hyperdual-step method in CMAQ to enhance our understanding of aerosol formation from VCPs and the potential of the novel model being applied to answer other critical policy-related questions.

View PDF of Presentation          Watch Recording of Presentation

Temperature-dependent composition of summertime PM2.5 in observations and model predictions across the Eastern U.S.

Pietro Vannucci, University of California, Berkeley, PhD Candidate at UC Berkeley/ORISE Fellow at U.S EPA

Throughout the U.S., summertime fine particulate matter (PM2.5) exhibits a strong temperature (T) dependence. Reducing the PM2.5 enhancement with T could limit the public health burden of PM2.5 now and in a warmer future. To fully elucidate the PM2.5-T relationship and inform control strategies, atmospheric models are a critical tool for probing the specific processes and components driving the observed behaviors. In this work, we describe how observed and modeled aerosol abundance and composition varies with T in the present-day Eastern U.S. with specific attention to the two major PM2.5 components: sulfate (SO4) and organic carbon (OC). We find that both measured SO4 and OC aerosols increase with T; however, the model underestimates SO4 and its increase with temperature and overestimates OC and its increase with temperature. We explore model design interventions to target these discrepancies and find that for SO4, model bias is improved significantly by the introduction of an aerosol-phase pathway for SO2 oxidation, and for OC, model bias is improved by the introduction of a photolysis sink for monoterpene-derived organic aerosols as well as a hydrolysis sink for organic nitrates. Additionally, we find that interventions designed to target modeled SO4 also influence OC. We conclude that the PM2.5-T relationship is driven by coupled inorganic and organic systems, and efforts to understand overall behavior will necessitate a multi-component approach.

View PDF of Presentation          Watch Recording of Presentation

AerChemMIP2 : Deciphering the role of aerosols and chemically reactive gases in climate change

Duncan Watson-Parris, University of California, San Diego, Assistant Professor

Phase 2 of the Aerosol Chemistry Model Intercomparison Project (AerChemMIP2) focuses on the quantification of the climate, atmospheric composition and air quality responses to changes in emissions of aerosols and chemically reactive gases. Such short-lived climate forcers (SLCFs) play a fundamental role in regional and global climate change. AerChemMIP2 aims at a better understanding of the relative contributions of individual SLCF emissions to composition change, radiative forcing and the climate response from the pre-industrial to present-day as well as for projected future emission scenarios. Assessing feedbacks from natural emissions relevant to SLCFs on atmospheric composition and climate change is a key component in AerChemMIP2. The experimental protocol builds on methodological knowledge, e.g., gained through AerChemMIP, RFMIP and ScenarioMIP, which were endorsed by CMIP6. Specifically, AerChemMIP2 seeks to closely align with the CMIP7 core experimental design, namely the Diagnostic, Evaluation and Characterization of Klima (DECK), historical, and future scenario experiments. In so doing, AerChemMIP2 strives to generate policy-relevant information on modern climate change, contributing new knowledge and estimates on the role of individual SLCFs.

View PDF of Presentation          Watch Recording of Presentation

Developments and Applications of NOAA’s UFS-Aerosols and UFS-Chem for Global Aerosol Forecasts

Li (Kate) Zhang, CIRES University of Colorado Boulder & NOAA GSL, Research Scientist

There are two global chemical forecast systems under development and online coupled with the Unified Forecast System (UFS) at NOAA. 1) UFS-Aerosols: the second-generation of UFS coupled aerosol system was collaboratively developed by NOAA and NASA since 2021, which is planned to be implemented into Global Ensemble Forecast System (GEFS) v13.0 in the coming years. UFS-Aerosols embeds NASA’s 2nd-generation GOCART model in a NUOPC infrastructure. 2) UFS-Chem: an innovative community model of chemistry online coupled with UFS, which is a wide collaboration between NOAA Oceanic and Atmospheric Research (OAR) laboratories and NCAR. It utilizes the Common Community Physics Package (CCPP) infrastructure to link the gas and aerosol chemistry modules to the rest of the model that enhance the research capabilities to use different gas and aerosol chemical mechanisms to couple different physics options. One of the aerosol components based on the current operational GEFS-Aerosols, has been implemented into UFS-Chem with some updates to wet deposition, fire emission etc. Both UFS-Aerosols and UFS-Chem include the direct and semi-direct radiative feedback from online aerosols prediction. They also have the capability to be fully coupled with ocean, sea ice, and wave components for S2S forecasts. The capabilities of UFS-Aerosols and UFS-Chem in medium-range and S2S predictions are evaluated and compared using observations from reanalysis data, ground-based measurements, and satellite data.

View PDF of Presentation          Watch Recording of Presentation

Comparison between a sectional and modal aerosol model in CESM2 for present-day and future aerosol injection experiments

Simone Tilmes, National Center for Atmospheric Research, Dr.

The Community Aerosol and Radiation Model for Atmospheres (CARMA) is a sectional aerosol and radiation model framework that has been integrated into the Community Earth System Model Version 2 (CESM2) Community Atmosphere Model Version 6 (CAM6) with chemistry (CAMchem) and the Whole Atmosphere Community Climate Model Version 6 (WACCM6) middle atmosphere (MA) chemistry versions. These model versions include a complete replacement of the modal aerosol model (MAM4) with CARMA to represent aerosols in the troposphere and stratosphere. CARMA and MAM4 include a comprehensive representation of aerosol microphysical processes coupled to chemistry, radiation, optics, cloud-aerosol interactions, emissions, and wet and dry removal.  We evaluate the performance of different model versions using MAM4 and CARMA in CAMchem and WACCM-MA during the last 30 years (1990-2020), including a large volcanic eruption (Mt. Pinatubo in 1991), volcanically quiet periods (around the year 2000), and a period of small-to-moderate eruptions (2005-2020). Model results are compared to each other and to balloon satellite and aircraft observations and show the advantages and shortcomings of the two different aerosol schemes. We also present the initial result of future aerosol injection experiments using CARMA and MAM4 and compare aerosol size distributions and burden, with implications for stratospheric aerosol intervention studies. 

View PDF of Presentation          Watch Recording of Presentation

Modeling constraints of aerosol layer height and night-time aerosol optical depth from space

Jun Wang, University of Iowa, UC Riverside, Director of the Atmospheric and Environmental Research Lab

In this presentation, I will highlight the progress and insights we’ve made to shed light on the global mapping of aerosol transport at night and absorbing (smoke and dust) aerosol layer height at daytime. I will present the first results of retrieving aerosol optical depth over the land and over the ocean by using reflected moonlight measured by the Day-Night- Band (DNB) of Visible Infrared Imaging Radiometer Suite (VIIRS) aboard Suomi-NPP and JPSS-1, and showcase the aerosol layer height product retrieved from TROPOMI and EPIC.

I will share some thoughts that the community can work together to use these recent advances in satellite remote sensing of aerosols to constrain the chemical transport modeling. 

View PDF of Presentation          Watch Recording of Presentation

Impacts of aerosol dynamical processes on the early stage evolution of volcanic plumes

Julia Bruckert, KIT Karlsruhe

Volcanic aerosols can disturb the Earth’s radiation budget by interacting with solar and terrestrial radiation and clouds with implications for weather and climate. Forecasting the impacts of volcanic aerosol requires detailed treatment of their life cycle. Most models oversimplify the volcanic emissions and neglect plume dynamical processes during explosive volcanic eruptions. Furthermore, they ignore the contribution of volcanic ash to aerosol dynamical processes. We use the ICON-ART (ICOsahedral Nonhydrostatic model with Aerosols and Reactive tracer gases) couple to the 1-D volcanic plume model FPlume and including aerosol dynamic processes to understand the developments of volcanic plumes after explosive volcanic eruptions. Here, we show the impacts of plume and aerosol processes on the dispersion on different time scales for examples from the 2019 Raikoke and 2021 La Soufrière eruptions.

Air Quality Modeling for Health and Regulatory Assessments

Estimating Sector-Oriented Roadside Exposure to Ultrafine Particle Number Concentrations: An implication to covariates influences on the models performance

Sultan Abdillah, Chung Yuan Christian University, P.hD Candidate

Exposure to ultrafine particles (UFPs) have recently become the main highlights of air pollution issues due to their higher penetrability in respiratory systems and distinct spatiotemporal characteristics compared to larger PM. Although there is still no regulation for UFPs, WHO Air Quality Guideline 2021 has recently published recommended exposure level limit for UFPs. However, many challenges persist in terms of collecting reference data for UFPs due to instruments constraint and their intra-urban spatial variability characteristic. Therefore, an optimized machine-learning approach is proposed to estimate and explain the characteristic of UFPs PNC exposure level and their influencing covariates at three closely located roadside sectors: industrial (IN), residential (RS), and urban background (UB). The model datasets were based on simultaneous measurement campaigns that have been conducted for one year in Taiwan. 11 covariates were incorporated, including black carbon (BC), traffic, and meteorology. Among three algorithms (MLR, RF, XGBoost), XGBoost possessed the highest performance for three roadside sectors (R2= 0.92–0.96; Average exposure= 1.1–1.8x10^10 #/h; RMSE= 5.4–7.9x10^8 #/h). Furthermore, SHAP analysis showed that UFPs PNC exposure at IN & RS were heavily affected by BC & traffic, while UB was influenced by solar radiation & humidity. The results implied that XGBoost + SHAP analysis could be used as proxy to complement UFPs exposure assessment in urban areas.

Health Impact Assessment of per Ton of Air Toxics and Its Regulatory Applications

Xue Meng (Sue) Chen, California Air Resources Board, Air Pollution Specialist

Health risk assessment is typically required during the rulemaking process for Airborne Toxic Control Measures (ATCMs) and is presented in the Initial Statement of Reasons (ISOR) as part of the rulemaking record. Through a high-resolution statewide modeling study of health impacts from major air toxics, including diesel particulate matter (DPM), volatile organic compounds (VOCs) and heavy metals, this study examines two aspects in support of the regulatory needs: 1) whether emission source sectors contribute differently to ambient risk; and 2) whether their contribution is uniformly distributed across California. We use the concept of incidence-per-ton (IPT) to examine the cancer risk per ton of DPM emissions for 13 emission sectors and 6 air basins. Our analyses have shown that different emission sectors generate different IPT values. For example, IPT values are high for on-road mobile sources compared to other emission sectors such as OGV due to their close proximity to residential areas and having lower emission plumes in the atmosphere. These IPT values can be directly used to estimate the health benefits achieved by toxic pollutant regulations. They can also be used to compare and evaluate the relative effectiveness of different emission source control measures, as well as to conduct cost-benefit analyses and inform strategies for reducing toxic emissions and exposure, supporting the district Community Emission Reduction Plans (CERPs) under the AB617 program.

View PDF of Presentation          Watch Recording of Presentation

Particulate Matter (PM2.5) precursor emission sensitivities and the impact on human health in California

Sarika Kulkarni, California Air Resources Board, Staff Air Pollution Specialist

Fine Particulate Matter (PM2.5) is known to reduce atmospheric visibility and adversely impact climate, ecosystems, and human health. PM2.5 is emitted directly from emission sources and formed via chemical reactions of precursor gases involving nitrogen oxides (NOx), sulfur oxides (SOx), reactive organic gases (ROG) and ammonia (NH3) in the atmosphere. The relative importance of these key precursors in the formation of PM2.5 exhibits distinct spatial, seasonal, and regional variability due to inherent non-linear chemistry and complex relationships between the PM2.5 precursors and PM2.5 concentrations.  This work examines the sensitivity of PM2.5 concentration to the decrease in the NOx, SOx, ROG and NH¬3 precursor emissions over the major air basins in California including San Joaquin Valley, Sacramento Metropolitan Area, and South Coast with the Community Multiscale Air Quality (CMAQ) version 5.4 (CMAQ54) simulations and presents the preliminary findings of the simulated health benefits from the predicted changes in PM2.5 concentration. Additional sensitivity analysis focused on understanding the impact of global boundary conditions and biogenic emissions on the simulated PM2.5 over California and the subsequent impact on the associated health impacts will also be evaluated.

View PDF of Presentation          Watch Recording of Presentation

Comprehensive Accounting for Reactive Organic Carbon Emissions from Residential Wood Combustion Processes

Benjamin Murphy, U.S. EPA, Physical Scientist

Residential wood burning is one of the largest sources of organic carbon particles and vapors to the atmosphere. The impact of these emissions on air quality is profound, and it is expected to increase in the future. Emission inventories and photochemical air quality models often use an outdated conceptual model of organic particulates and vapors. Specifically, regulatory test methods quantify particulate matter emission factors with an operational definition (i.e. mass captured on a Teflon filter) that is susceptible to systematic biases corresponding to the temperature and dilution conditions of each test. Meanwhile, total hydrocarbon vapors are characterized using flame-ionization detection, which provides an uncertain measure of gas mixtures containing oxygenated molecules. Finally, the speciation of residential wood burning emissions needs to be revised with new understanding of semivolatile and intermediate volatility compounds, which are organic aerosol precursors. This presentation discusses the methodology used to revise the organic emission factors and speciation in EPA emissions modeling tools and quantifies the impact of these updates organic aerosol concentrations with the Community Multiscale Air Quality (CMAQ) model. We evaluate the improved model predictions with measurements during the WINTER campaign (East U.S. Jan-Mar 2015). The new approach provides the most rigorous and complete translation of organic compounds emitted from residential wood burning.

The Impact of Air Pollution on the Health of Inhabitants in the City of Douala: CAMEROON.

Mbiake Robert, University of Douala, Professor

Some preliminary studies on the state of air pollution in the city of Douala; using AERMOD to simulate its dispersion, then after using direct and indirect method to analyse the same pollutants showed that the concentrations of PM2.5, PM10 et SO2 can reach at 183µg/m3; 194µg/m3 and 168µg/m3 respectively. These results come out clearly that, Douala is a polluted city and the immediate consequence that it affects the health of inhabitants primarily those exposed. To assess the impact of this pollution on the populations, we leaded research by submitting 1723 persons that are permanently exposed to this air. The results reveal that among the many clinical symptoms experienced, six (06) were frequently cited that is cold (77.19±1,6%), dry cough (71.51±1.6%), headache (70.93±1.7%), general tiredness (63.20±1.7%), eye pains (61.28%±1.5%), pungent nostril (53,34±1,1%). Using Pearson’s Exact Independence Test and Cochran-Mantel and Haenszel statistical method, it appears that alcohol and cigarettes are risk factors for 03 among the 06 that are dry cough (2=20.94), headaches (2=15.22) and cold (2=11.35). When age was associated with these clinical manifestations, it becomes a really confounding factor with these Odd Risk values (Cold  ;  =1,19), (Headache  ;  =1.56); Dry cough ( =1.46);  =0.35) for tobacco and alcohol respectively. In order to confirm these results, we intend to continue these studies with the assistance of doctors through biological and epidemiological analyzes.

View PDF of Presentation          Watch Recording of Presentation

Environmental Health, Racial/Ethnic Health Disparity, and Climate Impacts of Inter-Regional Freight Transport in the United States

Maninder Thind, California Energy Commission/ Formerly University of Washington, Seattle, Research Manager/ Formerly PhD Candidate

Atmospheric emissions from freight transportation contribute to human health and climate damage. In this research, we quantify and compare three environmental impacts from inter-regional freight transportation in the contiguous United States: total mortality attributable to fine particulate matter (PM2.5), racial−ethnic disparities in PM2.5-attributable mortality, and climate impacts (CO2 emissions). Our analyses use high spatial resolution air quality and health impact model, the Intervention Model for Air Pollution (InMAP) to model pollutant exposure and mortality impacts. We compare all major freight modes (truck, rail, barge, aircraft) and routes (∼30,000 routes). Our study is the first to comprehensively compare each route separately and the first to explore racial−ethnic exposure disparities by route and mode, nationally.

Impacts (health, health disparity, climate) per tonne of freight are the largest for aircraft. Among non-aircraft modes, per tonne, rail has the largest health and health-disparity impacts and the lowest climate impacts, whereas truck transport has the lowest health impacts and greatest climate impacts – an important reminder that health and climate impacts are often but not always aligned. Level of exposure and disparity among racial−ethnic groups vary in urban versus rural areas. This research can be used to inform, for a given origin and destination, which freight mode offers the lowest environmental health, health-disparity, and climate impacts.

Integrating Earth-System Modeling and Multi-Scale Observations to Support Health Studies in California

Minghui Diao, San Jose State University, Associate Professor

The increasing frequency and duration of wildfires and heatwaves in California in recent years has raised more concerns regarding the health impacts of wildfire smoke and extreme heat. More accurate estimates of surface PM2.5 and high temperatures at community scales are needed to support decision making activities by various stakeholders. In this work, I will present several approaches that integrate observational and modeling data at various scales to support health studies regarding PM2.5 and extreme heat in California. These datasets include ground-monitored data, mobile lidar and radar observations, satellite data, and Earth-system model simulations. The new capability of Department of Energy (DOE) Energy Exascale Earth System Model (E3SM) allows storm-resolving simulations at 3 km resolution for the entire state of California. Such new modeling capability offers the opportunity to assess health impacts of extreme events at both high-resolution spatial scales and long-term temporal span over several decades. A framework that aims to link the global climate model with public health studies will be discussed.

View PDF of Presentation          Watch Recording of Presentation

Formation of Reactive Oxygen Species by Atmospheric Particulate Matter

Manabu Shiraiwa, University of California, Irvine, Professor

Oxidative stress mediated by reactive oxygen species (ROS) is a key process for adverse aerosol health effects. Atmospheric particulate matter (PM) emitted from various sources including wildfires and tailpipe and non-tailpipe emissions, as well as secondary organic aerosols (SOA) contain various reactive and redox-active chemical compounds. Inhalation and deposition of such PM into the respiratory tract causes the formation of ROS by chemical and cellular processes. We have conducted laboratory experiments to quantify kinetics and elucidate mechanisms of ROS formation by various atmospheric PM. Such mechanistic and quantitative understandings provide a basis for further elucidation of adverse health effects and oxidative stress by fine particulate matter. We have also developed a kinetic model that can derive chemical exposure-response relations between ambient concentrations of air pollutants and the production rates and concentrations of ROS in the epithelial lining fluid of the human respiratory tract. The developed model and exposure-response relationship has been integrated into epidemiological studies for a quantitative assessment of the relative importance of different PM components on specific health endpoints such as respiratory and cardiovascular diseases.

View PDF of Presentation          Watch Recording of Presentation

Development, Application, and Reduction of Gas- and/or Particle-Phase Chemical Mechanisms for Aerosol Predictions

Modeling the seed-dependent particle growth via multiphase reactions with the particle-resolved model PartMC-CAMP

Yicen Liu, University of Illinois Urbana-Champaign, Graduate student

Secondary organic aerosol (SOA) contributes significantly to the total organic particulate mass. Recent laboratory studies found that the growth rate of particles due to SOA formation depends on the composition and phase of the seed particle and can be either enhanced or inhibited, which points to the occurrence of reactions in the particle phase. The observed processes may have important implications in determining the ambient CCN concentrations and can further influence global climate. To explore how important these effects are for the real atmosphere, the objective of this study is to integrate experimental results into the aerosol modeling framework (PartMC-CAMP) and to demonstrate how the aerosol mixing state impacts the particle growth due to SOA formation. We considered a system where products from α-pinene ozonolysis condense on pre-existing aerosols consisting of organics or ammonium sulfate. We systematically designed simulations going from simple mixing states to more complex cases, in which different subpopulations of seed particles and different combinations of initial size distributions compete for condensing organic vapors. We then evaluated how including the seed-dependent growth affects the size distribution, aerosol optical properties and CCN concentrations. Our results highlight the role of aerosol mixing state on the growth of aerosol particles due to SOA formation.

View PDF of Presentation          Watch Recording of Presentation

Atmospheric salt particle formation and hydration

Nanna Myllys, University of Helsinki, Academy research fellow

Molecular-level mechanisms of salt particle formation and hydration have been studied using state-of-the-art quantum chemical calculations together with cluster dynamics simulations. The hydration ability of a base follows in the order of gas-phase base strength whereas the proton transfer ability of a base in a water cluster is related to the aqueous-phase basicity (Myllys et al., PCCP 2021). Hydration ability of acid–base clusters seem to be related to the number of hydrogen binding sites. Increased relative humidity can enhance the particle formation several orders of magnitude when the smallest clusters are stabilised by water molecules, but it can also slightly reduce the particle formation in the case of strongly bound acid–base systems (Myllys, PCCP 2023).

General parameterization to compute particle formation rates from a heterodimer concentration has been derived. Parametrization was found it to be accurate within two orders of magnitude against an experimental data (Chee et al. ACP 2021). Effect of relative humidity in salt particle formation varies many orders of magnitude depending on the acid and base molecules. Therefore, neglecting hydration, or using same value for different systems may introduce remarkable inaccuracies in large-scale models. Thus, computationally cheap but accurate molecular-level parametrization to compute hydration factors is needed.

View PDF of Presentation          Watch Recording of Presentation

Machine Learning-Based Emulation of Secondary Organic Aerosol (SOA) Formation: An Overview of Ongoing Efforts

Alma Hodzic, NCAR, Scientist

Accurately predicting SOA formation is critical for gaining insights into air quality and climate systems. Chemistry-climate models have to rely often on incomplete representations of SOA chemistry given the computationally prohibitive cost of representing detailed chemistry of individual species. This presentation reviews the potential of machine learning approaches, specifically random forest, feed-forward and recurrent neural network architectures, to replicate the behavior of detailed SOA mechanisms while significantly reducing computational costs. The training data for the machine learning models are generated from thousands of individual simulations using an explicit chemistry model. They are randomly initialized to capture the range of  atmospheric chemical environments, and SOA precursors. The results demonstrate the effectiveness of machine learning in accurately emulating SOA formation, and a computational speedup up to 100,000 times compared to explicit models, enabling their integration into chemistry-climate models. We will present the methodology employed to generate the training dataset, implement the machine learning algorithms, and evaluate their performance. The influence of initial conditions and chemical regimes on model performance will be discussed. A comparative analysis of different machine learning algorithms will be presented, shedding light on their respective strengths and limitations.

Watch Recording of Presentation

Investigating Anthropogenic Emission Mitigation Effects on Biogenic SOA Formation using Simplified and GENOA-Generated Mechanisms in 3-D Modeling

Zhizhao WANG, CEREA/INERIS, Postdoctoral researcher

Due to computational limitations, 3-D Chemical Transport Models (CTMs) widely use simplified SOA mechanisms that may not fully capture the complexity of VOC chemistry on SOA formation. To address this issue, the GENerator of Reduced Organic Aerosol Mechanisms (GENOA) has been developed. This tool reduces detailed VOC mechanisms (e.g., MCM) into condensed SOA mechanisms, preserving the complex VOC chemistry suitable for large-scale modeling.

Using the 3-D CHIMERE CTM model coupled with the 0-D SSH-aerosol model, we investigated the impact of anthropogenic emission mitigation on biogenic SOA formation over Europe with simplified and condensed SOA mechanisms. Comparing the results with the simplified mechanism H2O (Hydrophilic/Hydrophobic Organics), the GENOA-generated mechanism GBM improved the agreement with measurements and showed greater sensitivity to anthropogenic emission reductions. Notably, biogenic SOAs decrease in response to anthropogenic VOC emission mitigation, but increase when NOx emissions are reduced, or when both VOC and NOx emissions are reduced. This increase was attributed to the promotion of auto-oxidation when reducing NOx emissions, resulting in the formation of more monoterpene SOAs.

Our work highlights the significance of incorporating detailed SOA mechanisms into 3-D air quality models, as they could provide valuable insights into the accurate SOA responses to anthropogenic emissions mitigation and the assessment of emission regulations.

View PDF of Presentation         Watch Recording of Presentation

Evaluating an Isoprene SOA Kinetic Model Using Laboratory and Field Measurements

Haofei Zhang, University of California, Riverside, Associate Professor

Isoprene is the largest global non-methane hydrocarbon emission, and the chemical reactions of isoprene with atmospheric oxidants play a crucial role in the formation of secondary organic aerosols (SOA). Two primary pathways contribute to SOA formation from isoprene low-NOx OH oxidation: (1) the reactive uptake of isoprene epoxidiol (IEPOX) on acidic or aqueous particles, resulting in the formation of 2-methyltetrols and organosulfates (IEPOX pathway) and (2) the production of low-volatility compounds through OH oxidation of intermediates such as ISOPOOH (LV pathway). In this study, we develop a condensed chemical mechanism for isoprene oxidation based on recent observation-constrained chemistry and use a zero-dimensional box model to simulate SOA formation from both pathways. We apply our revised isoprene mechanism to simulate chamber SOA data and compare its performance against existing mechanisms. Simulations of the SOAS field measurements using this model indicate that the LV pathway contribute to 15-20% of total isoprene SOA, with the rest from the IEPOX pathway. But the modeled relative contributions of the two pathways vary largely with product volatility and IEPOX reactive uptake kinetics. The new condensed isoprene chemical mechanism will be further incorporated into regional-scale air quality models such as CMAQ to assess the influence of the LV pathway on a larger scale.

View PDF of Presentation          Watch Recording of Presentation

Fundamental Aerosol Processes from Nano- to Microscale

The Effect of Atmospherically Relevant Aminium Salts on Water Uptake

Noora Hyttinen, University of Jyväskylä, Postdoctoral Researcher

Hygroscopic properties of aerosol constituents affect the water uptake and cloud activation of aerosol particles. Especially various salts increase the hygroscopic growth of aerosol due to lowered water activities. I have computed water activities in different salts consisting of atmospheric small acids and amines using the conductor-like screening model for real solvents (COSMO-RS). This method allows for the prediction of water activities in atmospherically relevant salts that have not been included in other thermodynamics models. Water activities were calculated for binary aqueous salt solutions, as well as ternary solutions containing proxies for organic aerosol constituents. The order of the studied cation species regarding water activities is similar in sulfate, iodate and methylsulfonate, as well as in bisulfate and nitrate. Predicted water uptake strengths (in mole fraction) follow the orders: tertiary > secondary > primary amines, and guanidinos > amino acids. The addition of water soluble organic to the studied salts generally leads to weaker water uptake compared to pure salts. On the other hand, water-insoluble organic likely phase separates with aqueous salt solutions, leading to minimal effects on water uptake. 

If you would like to learn more about this presentation and topic, you can contact Dr. Noora Hyttinen at noora.x.hyttinen <at> jyu.fi

Assessing the impact of unresolved particle characteristics on climate-relevant aerosol properties

Laura Fierce, Pacific Northwest National Laboratory, Scientist

For the past two decades, aerosol effects have been the largest source of inter-model variability in radiative forcing among climate simulations. Climate-relevant aerosol properties depend critically on the distribution in size, shape, and chemical composition of particle populations that evolve in time as they are transported through Earth’s atmosphere. However, tracking such particle-level details is computationally impractical for large-scale, long-running climate simulations. Instead, aerosol modules in Earth System Models necessarily simplify the representation of particle characteristics, leading to errors in climate-relevant aerosol properties that have not been well quantified. Here we present a framework for using particle-resolved simulations of aerosol-cloud-chemistry interactions to quantify structural errors from the numerical representation of particle populations. Particle-resolved models track per-particle composition among evolving aerosol populations and are, therefore, not subject to many of the numerical errors that plague reduced aerosol schemes. Through this approach, we aim to inform the development of aerosol schemes that balance model accuracy with computational efficiency, while also characterizing uncertainty from reduced representations of particle populations.

View PDF of Presentation          Watch Recording of Presentation

Molecular mechanism of gas phase oxidation of select volatile vapors

Siddharth Iyer, Tampere University, Academy research fellow

Gas-phase oxidation reactions of volatile compounds are a major source of atmospheric aerosol that have implications on health and on climate. To form aerosol, the participating vapors need to have sufficiently low volatility, which in practice implies molecules with multiple oxygen containing polar functional groups called highly oxygenated organic molecules (HOM). While the formation of HOM has been shown to be efficient for many biogenic and anthropogenic vapors, the underlying molecular level mechanism has been challenging to accurately elucidate because of the sheer number of potential pathways. In this presentation, I will describe our recent progress in understanding the formation of HOM from two key aerosol forming volatile vapors in the atmosphere, alpha-pinene and toluene. We used quantum chemical calculations and master equation simulations that account for energy non-accommodation of key reaction intermediates and targeted flow reactor experiments with nitrate chemical ionization mass spectrometry in our work. Our results may be vital for the accurate chemical modelling of the atmosphere.

View PDF of Presentation          Watch Recording of Presentation

Modeling uncertainties of aerosol properties and processes

Kari Lehtinen, University of Eastern Finland, Professor

In current scientific literature on aerosol science, it is very typical that when aerosol size distributions and/or estimations of process rates from measurements are presented, there is no accompanying estimation of uncertainties, at least not such that are done in a rigorous way. The aim of this work is to change how such measurement data and/or modeling results are presented by using probability density functions instead of single points or lines.

As a first practical test case we have investigated the effect of uncertainties associated with charging ions on the charging probabilities in instruments in which the measurement is based on particle charging and further measurement of electrical mobility, as well as how these uncertainties propagate further to the inverted size distributions. In addition, we have been studying how uncertainties in a time series of size distribution measurements affect aerosol microphysical process rate estimation.

As methodologies we have applied various sampling, Markov Chain Monte Carlo (MCMC) as well as Extended Kalman filter and smoother methods, coupled with aerosol and ion dynamics modeling.

View PDF of Presentation          Watch Recording of Presentation

Machine Learning and Data Science

Quantum Chemical Modelling of Atmospheric Molecular Clusters Enhanced by Machine Learning

Jakub Kubecka, Aarhus University, PostDoc

The formation and growth of molecular clusters in the atmosphere have significant implications for global climate dynamics. However, due to their small sizes and low concentrations, understanding these mechanisms presents challenges in terms of measurement and explanation. To explore this phenomenon, we employed computational quantum chemistry to investigate the dynamics of cluster populations. Previous studies have already highlighted the importance and cooperative nature of various multi-component systems in atmospheric new particle formation using this methodology. We additionally incorporate machine learning (ML) techniques into the traditional multi-level, funnel, configurational-sampling approach, thereby reducing the computational burden associated with quantum chemistry calculations. In this presentation, we discuss validation of the ML methods used for studying the formation of molecular clusters, along with the potential improvements facilitated by ML techniques. Importantly, ML not only accelerated the configurational sampling process but also allowed for rapid molecular dynamics simulations at the level of accuracy provided by quantum chemistry methods. As a result, we present an innovative approach that offers fresh insights into the thermodynamic stability of these atmospheric molecular clusters.

View PDF of Presentation         Watch Recording of Presentation
 

Characterizing Atmospheric Molecules for Machine Learning

Hilda Sandström, Aalto University, Postdoctoral researcher

Aerosols in the atmosphere impact air quality and contribute to climate change. A great number of atmospheric molecules and chemical processes lead to aerosol formation. This combinatorial complexity is well-suited for study using machine learning. However, for many applications, large datasets of atmospheric molecules, which are needed for machine learning, are not available.

In this contribution, I present an investigation into the similarity, as seen through molecular descriptors and fingerprints, between a dataset of atmospheric molecules and several available molecular datasets from other research domains. Preliminary results demonstrate little overlap between the atmospheric dataset and the others. The implication of this study is that the atmospheric science community needs a joint effort to collect and curate large atmospherically relevant datasets.

This work was supported by the VILMA (Virtual laboratory for molecular level atmospheric transformations) center of excellence funded by the Academy of Finland under grant 346377.

Watch Recording of Presentation

Combining Earth system modeling and machine learning to investigate volcanic sulfate deposition in polar ice cores

Kostas Tsigaridis, Columbia University and NASA GISS, Research Scientist

Volcanic eruptions emit large amounts of sulfur dioxide (SO2), water, and other chemicals into the atmosphere, both in the troposphere and the stratosphere. Most of the SO2 is converted to sulfate aerosol, which is eventually deposited following long-range transport. The deposits from large eruptions are potentially detectable in ice cores, but there are many cases in which sulfate layers have not been linked to their source volcanoes. To narrow down the search, we performed 140 simulations of volcanic eruptions using the GISS ModelE Earth system model. We varied the latitude, longitude, julian day, plume top, plume bottom, and injected SO2 and H2O amounts using a latin hypercube sampling approach, and analyzed correlations between these parameters and sulfate depositions at ice core sites in Antarctica and Greenland. Using machine learning and parameter estimation, we generated probability distributions for the parameters given sulfate deposition data. We find that the volcano latitude and SO2 content are best correlated with sulfate depositions at each pole, while longitude, julian day, and H2O have small or insignificant effects. Plume altitude and thickness are important because they determine how much of the SO2 is injected into the stratosphere, which has implications for sulfur transport and lifetimes.

View PDF of Presentation          Watch Recording of Presentation

Emulating Aerosol Optical Properties Using Machine Learning

Andrew Geiss, Pacific Northwest National Laboratory, Data Scientist

Accurate representation of the direct radiative effects of atmospheric aerosols is a crucial component of modern climate models. Direct computation of the radiative properties of aerosol populations is far too computationally expensive to perform during climate simulations however, so optical properties are typically approximated using a parameterization. This work develops artificial neural networks (ANNs) to replace the aerosol optics parameterization currently used in the Energy Exascale Earth System Model’s (E3SM) Atmosphere Model (EAM). The ANNs are trained using a large dataset of aerosol optical properties computed using accurate Mie scattering code. Two ANN models are developed, one that directly replaces the current parameterization while achieving significantly higher accuracy, and a second that represents a more sophisticated core-shell scattering model. Optimal neural architectures for this problem are identified by evaluating ANNs with randomly generated wirings. We show that randomly generated deep ANNs that include many skip connections can consistently outperform conventional multi-layer perceptron style architectures with comparable parameter counts. Finally, we discuss results of running E3SM with the new parameterization.

View PDF of Presentation          Watch Recording of Presentation

Physics-Constrained Learning of Aerosol Microphysics

Paula Harder, Fraunhofer ITWM, PhD student

Aerosol particles play an important role in the climate system by absorbing and scattering radiation and influencing cloud properties. They are also one of the biggest sources of uncertainty for climate modeling. Many climate models do not include aerosols in sufficient detail due to computational constraints. In order to represent key processes, aerosol microphysical properties and processes have to be accounted for. This is done in the ECHAM-HAM global climate aerosol model using the M7 microphysics, but high computational costs make it very expensive to run with finer resolution or for a longer time. We aim to use machine learning to emulate the microphysics model at sufficient accuracy and reduce the computational cost by being fast at inference time. The original M7 model is used to generate data of input-output pairs to train a neural network on it. We are able to learn the variables' tendencies achieving an average R2 score of 77.1%. We further explore methods to inform and constrain the neural network with physical knowledge to reduce mass violation and enforce mass positivity. On a GPU we achieve a speed-up of up to over 64x compared to the original model. Additionally, we investigate challenges for an implementation in Fortran and runs in a GCM.

View PDF of Presentation          Watch Recording of Presentation

Data driven futures: From stakeholder development to model development

David Topping, University of Manchester, Professor

Data science is still affecting changes across all areas of science, from accelerated discovery to new ways of working when it comes to translating theory to code. In aerosol detection and modelling, we see continual demonstration of the benefits it can bring. This includes adoption of powerful image classifiers, hybrid numerical solvers and more accessible simulation tools. With the increased adoption of such methods, we are already ‘baking in’ dependencies on platforms and skillsets. This is particularly true as we look towards placing ‘improved’ modelling capability at the hands of policy driven organisations and/or newly formed user communities. As part of the UKs Natural Environment Research Environment (NERC) Digital Solutions programme, we have been holding workshops across the nation to better understand perceptions, barriers and perceived opportunities around adoption of these powerful tools.  In this talk I will present outcomes from this research programme, but place user needs in the context of academic development and stakeholder adoption of aerosol models/tools we develop.

View PDF of Presentation          Watch Recording of Presentation

Process and Box Models of Aerosol Chemistry and Physics

PyPartMC: A Pythonic interface to a particle-resolved Monte Carlo aerosol simulation framework

Zachary D'Aquino, University of Illinois Urbana-Champaign, Graduate Research Assistant

PyPartMC is a free, libre, and open source Python interface to PartMC, a particle-resolved Monte-Carlo aerosol model implemented in Fortran. It simulates the microphysical processes that aerosol particles undergo during their life cycles in the atmosphere. These processes include new particle formation and emission from primary sources, Brownian coagulation, and removal by dry deposition and nucleation scavenging. The development of PyPartMC was motivated by the fact that running a PartMC simulation requires the use of shell and a netcdf-Fortran library, presenting a significant hurdle for those less experienced in computation. PyPartMC is published on pypi.org and can be installed at the command line via `pip install PyPartMC` to grant the user access to the unmodified and versioned Fortran internals of the PartMC codebase on Linux, macOS, and Windows. The Python packaging infrastructure eases distribution and maintenance while ensuring code version traceability for PyPartMC bindings, PartMC itself, and dependencies, such as SuiteSparse and SUNDIALS numerical solvers, CAMP (Chemistry Across Multiple Phases), and various input-output libraries. PyPartMC is written in Fortran and C++ and uses the pybind11 framework to build the Python API, which is callable from other languages, such as Julia. Here we show the utility of PyPartMC in facilitating valuable comparisons between modeling frameworks and its applications in reproducible computational research and active learning.

View PDF of Presentation          Watch Recording of Presentation

Water activity and surface tension of aerosol nanoparticles composed of aqueous ammonium sulfate and D-glucose aqueous solution of aerosolized nanoparticles

Eugene Mikhailov, Saint-Petersburg state university, professor

Water activity and surface tension are key thermodynamic parameters to describe the hygroscopic growth of aerosol particles. Due to size effects and high solute supersaturations, however, these parameters are not well constrained for nanoparticles composed of organic and inorganic compounds. In this study, we determined water activity and surface tension for aerosol nanoparticles composed of ammonium sulfate (AS) and D-glucose (Gl) by Ddifferential Köhler analysis (DKA) of hygroscopic growth measurement data in comparison to bulk measurements and thermodynamic model results. Hygroscopic growth experiments were performed for aerosol particles with diameters in the range of 17-100 nm using a humidified tandem differential mobility analyzer (HTDMA). We show that the HTDMA data complemented by the DKA, makes it possible to determine water activity and surface tension from moderate to highly supersaturated aqueous solutions. The high concentration range allowed observing a non¬-monotone change in the surface tension of nanodroplets. In particular, for 1:1 AS/Gl mixed particles a strong reduction of surface tension was observed up to a minimum value of 56.5 mN m-1 and its subsequent increase due to solidification. This non-monotonic change in the surface tension of mixed particles has not previously been observed. We suggest that D-glucose molecules surrounded by AS ions tend to associate, forming non-polar aggregates, which lower the surface tension at the air-droplet interface.

Watch Recording of Presentation

The Role of Interfacial Energy and Size-Dependent Morphology of Atmospheric Aerosol Particles

Ryan Schmedding, McGill University, Department of Atmospheric and Oceanic Sciences, PhD Student

Atmospheric aerosol may exhibit size ranges which span several orders of magnitude from the nanometer scale to the micrometer scale. This leads to aerosol particles having size-dependent yet high surface area to volume ratios. Furthermore, atmospheric aerosols may exhibit two or more liquid phases within a single particle. It has recently been found that some aerosol particles may undergo liquid–liquid phase separation at larger (submicron) particle sizes while ultrafine particles of similar composition, but smaller diameter, may not undergo liquid–liquid phase separation at the same conditions. It is thought that the energetic penalty found at the boundary between two condensed phases, better known as the interfacial tension, may play a role in this size-dependent liquid–liquid phase separation behavior. Interfacial tension is analogous to the surface tension at the liquid–gas phase boundary, yet some of the theoretical treatment differs. As such, we have extended the thermodynamically rigorous treatment of bulk–surface partitioning developed in Schmedding and Zuend (2023) to predict the interfacial tension of atmospheric aerosol particles as well. We also have included comparisons to other simplified treatments of interfacial tension in multi-component systems. The interplay between interfacial tension, surface tension, and geometric morphology will be discussed.

View PDF of Presentation         Watch Recording of Presentation

Investigating impact of surfactants on cloud condensation nuclei activity with a particle-resolved aerosol model

Xiaotian Xu, University of Illinois Urbana-Champaign, PhD student

Surfactants are organic compounds that can affect cloud condensation nuclei (CCN) activity by forming a film at the gas-liquid interface, reducing surface tension and impeding water transport. Previous studies used κ-Köhler theory to estimate CCN activity of organic/inorganic mixtures, assuming the surface tension of water instead of considering reduced surface tension caused by surfactants, which is crucial when studying realistic aerosol mixing states. This work combines an effective surface tension model with the particle-resolved aerosol model PartMC-MOSAIC to depict changes in surface tension as aerosol particles take up water. This approach avoids assumptions about aerosol mixing state that traditional modal or sectional models require and allows investigation of influence of this method on CCN concentrations at a per-particle level. Additionally, we integrate this method into the WRF-PartMC model to assess regional-scale impacts on CCN concentrations over California. Initial findings reveal a significant underestimation of CCN concentrations when disregarded. The error in CCN concentration averages around 25% over model domain with specific locations showing substantial underestimation of 2.5 times than the actual values. Notably, particles sized 40-70 nm are particularly sensitive to this error. These results highlight the crucial role of surfactants in accurately assessing CCN concentrations and emphasize the need for their inclusion in modeling studies.

View PDF of Presentation          Watch Recording of Presentation

Understanding the Formation of Organic Acids via Cloud Chemistry Box Modeling

Mary Barth, NCAR, Senior Scientist

Chemical processes in clouds can substantially alter atmospheric oxidant budgets affecting aerosol mass formation. However, many models exclude detailed aqueous-phase chemical mechanisms due to incomplete understanding of the processes and increased computational burden. The formation of organic acids is especially challenging as they have complex chemistry in both the gas and aqueous phases. Recent findings from a model intercomparison at Whiteface Mountain, New York (WFM) on cloud chemistry box models highlighted substantial variability in results caused by different equilibrium constants, aqueous-phase reaction rate constants, and chemical mechanisms. These results highlight the continued need for recommended equilibrium and aqueous-phase rate constants, much like what is done with gas-phase chemistry. The fixed location in the model intercomparison study prevented proper modeling of air parcel flow through clouds. Recent work has shifted to conducting box model simulations along trajectories using two gas-phase mechanisms followed by a cloud chemistry box model calculation. The results from this study reveal that formation of formic and acetic acids in the gas-phase is not consistent among mechanisms, suggesting a need to improve gas-phase mechanisms for organic acid formation. The cloud chemistry box model calculations substantially underpredicted formic and acetic acids yet reasonably predicted oxalic acid production compared to WFM observations.

View PDF of Presentation          Watch Recording of Presentation

Process-Level, Kinetic Models to Study the Formation, Physicochemical Properties, and Experimental Artifacts for Secondary Organic Aerosol

Shantanu Jathar, Colorado State University, Associate Professor

Secondary organic aerosol (SOA), formed from the oxidation of volatile organic compounds, is an important fraction of the atmospheric aerosol mass. The concentrations, composition, and properties of SOA are governed by a host of kinetic processes, many of which are inadequately represented in models. Over the past five years, our group has developed a kinetic, process-level model called SOM-TOMAS (Statistical Oxidation Model - TwO Moment Aerosol Sectional) that simulates the oxidation chemistry and microphysics of SOA. In this talk, I will present three case studies where we have applied the SOM-TOMAS model to study SOA formation and evolution in laboratory experiments. In the first case study, I will describe how the measured evolution of the aerosol size distribution can be leveraged to constrain the phase state and oligomer fraction of the SOA formed ozonolysis of alpha-pinene.  In the second case study, I will demonstrate that a detailed accounting of various processes can simultaneously explain SOA formation from alpha-pinene photooxidation in environmental chambers and oxidation flow reactor experiments, using a single set of SOA parameters. Finally, in the third case study, I will discuss wall loss artifacts in environmental chamber experiments and how artifact-corrected parameters can improve the aerosol performance in chemical transport models. Overall, we advocate for and share insights on the process-level representation of SOA in box and atmospheric models.

View PDF of Presentation          Watch Recording of Presentation

MultilayerPy: a python package for creating and optimising multi-layer models of aerosol and film processes

Adam Milsom, University of Birmingham, Research Fellow

Heterogeneous processes such as aerosol-gas chemical reactions and vapour uptake are key to understanding the behaviour of aerosols in our environment. Kinetic multi-layer models such as the kinetic multi-layer model for aerosol surface and bulk chemistry (KM-SUB) and gas-particle interactions (KM-GAP) are state-of-the-art models used to describe these processes on the particle and film level. These models are useful but cumbersome to write and there is a need for an open-source tool to assist researchers in creating and optimising them. We have developed MultilayerPy, an open-source Python package which facilitates the creation and optimisation of kinetic multi-layer models. This software is written such that the user uses building blocks (i.e. reaction scheme, bulk diffusion parameterisations, and model components) to automatically generate model code which can be ran and the output presented in a reproducible manner. This reduces the time needed to develop model descriptions of aerosol processes and allows the user to focus on the scientific issues rather than coding the models. I will present some use cases to showcase the key features of the package.

View PDF of Presentation          Watch Recording of Presentation

Poster Presentations