2024 Conference Agenda

Loknath Dhar, Dr. Seth Lyman

In Utah’s Uinta Basin, elevated ozone concentrations can occur during the winter months and form under conditions of thermal inversion, snow cover, and the emission of pollutants from the oil and gas industry. Carbonyl compounds are important precursors to winter ozone, but their chemistry during winter ozone episodes has not been thoroughly investigated. Box models can be useful in air quality studies because they simplify the atmosphere into a single box, allowing for detailed analysis of atmospheric chemistry. In this study, (1) the F0AM Box Model was used along with four different chemical mechanisms to identify the specific carbonyl species that act as important precursors to winter ozone formation and determine how they form; (2) modeled emission factors for carbonyl compounds were determined, (3) the ozone formation potential of different carbonyl compounds was assessed, and (4) the sensitivity of various alkanes, alkenes, and aromatic species to carbonyl compounds was analyzed to provide information about how different hydrocarbon species regulate the production of ozone and carbonyl compounds. These analyses were conducted using the MCMv331, RACM2, SAPRC07, and CB06 chemical mechanisms, and the results of the different mechanisms were compared. Emission factors were adjusted until the modeled carbonyl concentrations matched the measured values. Final emission factors were zero or nearly zero, which indicates carbonyl compounds were mostly secondary pollutants, created in the atmosphere from photochemistry. Throughout this study, MCMv331, RACM2, and SAPRC07 provided nearly similar results, showing formaldehyde to be the most important carbonyl species, contributing 33–42% of the total impact of carbonyls on ozone production. The same models agreed that acetaldehyde was the second most important contributor, with around 15–28%. However, the CB06 mechanism resulted in KET (a model species signifying ketone groups) as the most important carbonyl contributor to ozone formation (65%) and formaldehyde as the second most important with 17%. The CB06 mechanism uses functional groups as a stand-in for specific compounds, and this might cause the observed differences in the results compared to other chemical mechanisms. By incorporating heterogeneous chemistry in the F0AM model script with MCM, the ozone level decreased by less than 2 ppb, and the overall contribution of different carbonyl compounds to ozone production remained the same. The next step in this project will be to use CMAQ for a comprehensive, regional-scale analysis. The impact of different emission sources will be evaluated, and the results from the F0AM box model will be compared against CMAQ results. Baseline simulations will be run to understand the model's default behavior, followed by sensitivity simulations with different mechanisms and modified emissions to study the impacts of organic emissions on ozone formation, particularly focusing on carbonyl compounds. After the completion of simulations, results will be analyzed using tools like VERDI to compare different chemical mechanisms, helping to understand factors influencing winter ozone formation and the impact of various precursors and emission sources. This comparison will highlight the strengths and weaknesses of the mechanisms, guiding future improvements in atmospheric chemistry modeling and validating the F0AM box model results.

Tanya Spero, Megan Mallard, Jeff Willison, Jared Bowden, and Chris Nolte

The EPA Dynamically Downscaled Ensemble (EDDE) is a collection of physics-based regional climate data that represent atmospheric conditions for historical and future periods under different greenhouse gas emission scenarios. The EDDE datasets were developed using the Weather Research and Forecasting (WRF) model driven by multiple global climate models. The EDDE datasets can be used to explore regional climate change and its influence on extreme weather events out to 2100 across the contiguous United States to quantify potential impacts on human health and the environment. Specifically, the EDDE datasets have been used with CMAQ to explore plausible impacts of climate change on air quality, pollutant deposition to sensitive ecosystems, and human health.

This presentation will provide an overview of EDDE Version 1, a subset of which is available through Amazon Web Services Open Data Project. The technical details and utility of the dataset will be covered. In addition, EDDE Version 2 is under development, and details of that dataset and timelines will be presented.

Golam Sarwar, William T. Hutzell, Jesse Bash, George Pouliot, David Wong, Robert Gilliam, Christian Hogrefe, Kristen Foley, Fahim Sidi, T. Nash Skipper, Havala Pye, Rohit Mathur, Jeff Willison, Ben Murphy, Barron Henderson, Kevin Talgo, William R. Stockwell, Alfonso Saiz-Lopez

The Community Regional Atmospheric Chemistry Multiphase Mechanism (CRACMM) is a newly developed chemical mechanism that integrates reaction pathways for formation of ozone, secondary organic aerosol, and hazardous air pollutants and contains 508 chemical reactions. To extend the applicability of CRACMM to broader spatial and temporal scales, we implement 370 additional halogen related reactions and the photolysis of particulate nitrate into CRACMM and perform three different simulations for the year 2018 using the Community Multiscale Air Quality (CMAQv5.4) model over the Northern Hemisphere with 108-km horizontal grid spacing. The first simulation uses the base CRACMM without halogen chemistry and no particulate nitrate photolysis. The second simulation uses base CRACMM augmented with halogen chemistry, and the third simulation uses base CRACMM with both halogen chemistry and the particulate nitrate photolysis. The halogen chemistry generally reduces ozone with a larger reduction occurring over seawater than over land. In contrast, the photolysis of particulate nitrate enhances ozone and compensates the halogen mediated ozone loss with a larger enhancement occurring over seawater than over land. The halogen mediated ozone loss is the largest near the surface layer and decreases with altitude. In contrast, the particulate nitrate photolysis initiated ozone enhancement is larger aloft than near the surface. We compare model ozone mixing ratios with measurements from the Clean Air Status and Trends Network (CASTNET) and the Air Quality System (AQS) over the United States. The mean bias (MB) of the model without the halogen chemistry is generally negative in cooler months and positive in warmer months. The halogen chemistry deteriorates the MB in cooler months but improves it in warmer months. The photolysis of aerosol nitrate improves MB in cooler months but deteriorates it in warmer months. We also compare model predictions with available ozonesonde data and find that halogen chemistry generally deteriorates the comparison while the particulate nitrate photolysis improves the comparison with observed data.

Disclaimer: The views expressed in this abstract are those of the authors and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency (EPA).

Kristen Foley¹, Adam Reff², Jeannette Reyes², Sharon Philips², Barron Henderson², Christian Hogrefe¹

¹ US Environmental Protection Agency Office of Research and Development

² US Environmental Protection Agency Office of Air and Radiation

As part of the EPA’s Air Quality Time Series (EQUATES) project, we created measurement-model fused estimates of PM2.5 and maximum daily 8-hr average ozone (MDA8 O3) using the set of consistent CMAQv5.3.2 simulations and a Bayesian space-time Downscaler (DS) statistical model. We compare the 2002-2019 EQUATES DS surfaces to another set of DS fused surfaces that EPA developed to inform the CDC’s National Environmental Public Health Tracking Network. The CDC DS surfaces were developed over the last decade based on the best air quality modeling data available at the time. For example, the fused estimates for the early 2000s are based on older versions of the CMAQ system and use a courser grid resolution than estimates for more recent years. We used 10-fold cross-validation (CV) of both the EQUATES and CDC surfaces to quantify differences in estimation error at unmonitored locations from the two fused datasets. We find that the EQUATES DS and CDC DS have very similar aggregate cross validation errors (i.e., averaged across seasons and regions), even when the underlying CMAQ estimates in EQUATES had better performance than the older CMAQ versions used for the CDC timeseries, reinforcing the benefits of using data fusion as a bias adjustment tool. However, looking at individual days we see large differences can occur between the two datasets, highlighting that biases in the underlying air quality output can still be carried through to the fused surfaces. We also use a modified cross validation approach to investigate how the 1-in-3-day sampling at many PM2.5 monitoring sites impacts the fused estimates, i.e., how the CV error changes on days with less observational data available to adjust the model surfaces. The EQUATES and CDC DS cross validation metrics show that biases in PM2.5 are within ±4 µg/m3 across seasons and regions except for a persistent underestimation in the fall and winter in northwestern states. The seasonal/regional CV mean biases in MD8 O3 for both EQUATES and CDC are typically within ±3ppb. The EQUATES DS cross validation correlation is typically slightly higher than the CDC DS correlation. The fused cross validation metrics are substantially better than the metrics for the unadjusted EQUATES and CDC CMAQ output for most regions and seasons. The EQUATES DS fused PM2.5 and MD8 O3 offer the advantage of consistency in the underlying CMAQ modeling which is particularly important in the early years of the time series. We discuss the implication of these results on the use of Downscaler fused CMAQ surfaces for epidemiological studies linking health outcomes with pollutant concentrations.

Duncan Quevedo, Khanh Do, George Delic, José Rodríguez Borbón, Bryan M. Wong, and Cesunica E. Ivey

The Community Multiscale Air Quality (CMAQ) model simulates atmospheric phenomena including advection, diffusion, gas-phase chemistry, aerosol physics and chemistry, and cloud processes. Gas-phase chemistry traditionally presents the major computational bottleneck due to its representation as large systems of coupled nonlinear stiff differential equations. We leverage the parallel computational performance of graphics processing unit (GPU) hardware to accelerate the numerical integration of these systems in CMAQ’s CHEM module. Our implementation, dubbed CMAQ-CUDA in reference to its use of the Compute Unified Device Architecture (CUDA) general purpose GPU (GPGPU) computing solution, migrates CMAQ’s Rosenbrock solver from Fortran to CUDA Fortran. CMAQ-CUDA accelerates the Rosenbrock solver by a factor of at least 3.5 and maintains an acceleration factor of at least 2.25 when compared to Euler Backwards Integration, CMAQ’s fastest solver. CMAQ-CUDA maintains fidelity to the standard Fortran implementation with discrepancies well below 0.1% throughout the simulation domain for the duration of the simulation period. Our results demonstrate that CMAQ is amenable to GPU acceleration and provide a Rosenbrock solver implementation that relieves the computational bottleneck imposed by the CHEM module. This work facilitates timely air quality modeling studies that can ultimately protect human and ecosystem health.

I am a 4th year environmental engineering PhD student in Dr. Cesunica Ivey's Air Quality Modeling and Exposure Lab at the University of California, Berkeley. My background is in applied math and public health. I received my BS in Applied Mathematics and my BSPH in Environmental Health Sciences from the University of North Carolina at Chapel Hill. My research interests include air quality modeling with numerical and statistical tools to help reduce the burden of air pollution exposure around the country.

Youhua Tang^1,2, Patrick Campbell^1,2, Beiming Tang^1,2, Wei Li^1,2, Cory Martin³, Hongli Wang^4,5, Yunyao Li², Tianfeng Chai¹, Mariusz Pagowski^4,5, Daryl Kleist³, Barry Baker¹, Kai Wang^3,6, Jianping Huang³, Jeff McQueen³, Raffaele Montuoro³, Daniel Tong^1,2, Ivanka Stajner³, Youngsun Jung⁷, Shobha Kondragunta⁹, Ken Aikin¹⁰, Andrew Rollins¹⁰, Eleanor Waxman^5,10, John Crounse¹¹, Paul Wennberg¹¹, Ryan Bennett¹², Ilana Pollack^5,10,13, Emily Lill¹³, Matt Coggon¹⁰, Kelvin Bates^5,10, Amy Sullivan¹³, Mike Robinson¹⁰, Carsten Warneke¹⁰, Matthew Coggon¹⁰, Lu Xu^5,10, Kristen Zuraski^5,10, Jeff Peischl^5,10, Chris Jernigan^5,10, Wyndom Chace⁵, Steven S. Brown¹⁰, Caroline Womack¹⁰, Gordon Novak¹⁰, Patrick Veres¹⁰, Chelsea Stockwell¹⁰, Ann Middlebrook¹⁰, Rebecca Schwantes¹⁰, Charles Gatebe¹⁴

1. NOAA Air Resources Laboratory (ARL), College Park, MD. 2. Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA. 3. NOAA/NCEP/EMC 4. NOAA Global Systems Laboratory 5. CIRES, University of Colorado, Boulder, CO 6. Lynker, Leesburg, VA 7. NOAA NWS/STI 8. National Center for Atmospheric Research, Boulder, CO 9 NOAA/NESDIS/STAR 10 NOAA Chemical Sciences Laboratory 11 California Institute of Technology 12 Bay Area Environmental Research Institute, California 13. Colorado State University, Fort Collins, CO 14. NASA Ames Research Center, MS 245-5, Moffett Field, CA

The recently implemented NOAA National Air Quality Forecasting Capability (NAQFC) uses chemistry based on the Environmental Protection Agency (EPA)’s Community Multiscale Air Quality (CMAQ) model version 5.2. CMAQ serves as the basis for NOAA’s Unified Forecast System (UFS) air quality model (UFS-AQM), which is now online-coupled with the Finite Volume Cubed-Sphere Dynamic Core (FV3), and Common Community Physics Package (CCPP). Here we are working to upgrade UFS-AQM with multiple changes, including the newer CMAQ version ( 5.4+), increased resolution, physics changes, and data assimilation with Joint Effort for Data-Assimilation Integration (JEDI). This JEDI-based chemical DA system has been applied to constrain the UFS-AQM’s chemical initial conditions over an expanded North American domain at 13 km resolution. The assimilation of in-situ AIRNow data consistently improves the model's predictions for surface ozone and PM2.5. The change of UFS-AQM to use CMAQ version 5.4 mainly affects the dry deposition rates of ozone and water-soluble tracer gases, such as SO2 and NOx, and leads to lower concentrations of these species and higher concentrations of volatile organic compounds (VOCs). When compared to the aircraft data of Atmospheric Emissions and Reactions Observed from Megacities to Marine Areas (AEROMMA), the difference before and after the CMAQ 5.4 changes mainly appears in low altitude (<4km). Besides the primarily emitted species, the chemical mechanism change also affects the secondary oxidized compounds, such as acetaldehyde and glyoxal. Some systemic model biases are likely due to emissions and chemical background concentrations.

Youhua Tang is senior research scientist in George Mason University and NOAA Air Resources Laboratory. His work focuses on air quality modeling, its associated data/process analysis and chemical data assimilation. He is one of major scientists to support the NOAA National Air Quality Forecasting Capability (NAQFC) system

Simon Rosanka, Bill Hutzell, Kathleen Fahey, Ann Marie Carlton

Clouds and in particular aerosol-cloud interactions represent one of the largest uncertainties in projections of future climate. Aqueous phase chemistry in clouds produces sulfate and carbonaceous aerosol mass aloft, where particles have a longer lifetime and scatter incoming solar radiation. While aqueous phase cloud chemistry occurs simultaneously with partitioning and gas-phase reactions, historically, most atmospheric models represent multiphase chemistry sequentially. In general, due to numerical reasons, most physical and chemical processes are conducted as consecutive rather than concurrent steps by using operator splitting. For example, the “SCIPROC” subroutine in the Community Multiscale Air Quality (CMAQ) model first calls aqueous phase processes for resolved and convective clouds in series, followed by the representation of gas and heterogeneous chemistry, and lastly aerosol processes. This fundamental assumption, however, contradicts the strong interconnection between these processes.

In this work, we reduce CMAQ’s chemical operator splitting and develop a submodel that integrates chemical processes across gas- and aqueous-phases in parallel. This is achieved by representing gas and aqueous phase, and heterogeneous reactions in one single system of ordinary differential equations using the Kinetic PreProcessor. We evaluate the performance of the new chemical operator, with a particular focus on the representation of model variability of soluble gases and aerosols. Overall, the new development has the potential to improve the adjoint modelling capabilities of CMAQ as well as representing processes governing gas and aerosol species.

Dazhong Yin, Majiong Jiang, Zhan Zhao, Maybelline Disuanco, Chenxia Cai, Jeremy Avise

The Re-evaluating the Chemistry of Air Pollutants in California (RECAP-CA) 2021 field campaign gathered rich ambient air quality and meteorological data via aircraft, special ground site, and other measurements. Researchers from UC Berkeley conducted airborne in-situ and flux measurements of NOx using a thermal-dissociation laser-induced fluorescence (TD-LIF) and volatile organic compounds (VOCs) using a proton transfer reaction time-of-flight mass spectrometer (PTR-TOF-MS) on board a Twin-Otter aircraft from June 1 to June 22, 2021. The NOAA Chemical Sciences Laboratory deployed a trailer which housed the gas-phase instruments, including NOAA gas chromatography–mass spectrometry (GC-MS) from August 2 to September 7, 2021 at the ground site on the Caltech campus in Pasadena. Two NOAA LIDAR systems were set up at the site as well. The NOAA mobile laboratory periodically measured the spatial distributions of NOx and VOCs in the LA basin during that time.

Air quality modeling for 2021 was conducted using the Community Multiscale Air Quality Modeling System (CMAQ). We will evaluate the model performance for ozone, particulate matter and their precursor species in the South Coast and San Joaquin Valley (SJV) against the measurement data collected from the RECAP field campaign as well as the routine monitoring network. Given the availability of observational data, we will be able to explore various issues. These include assessing the contrast between model capabilities in reproducing measured NOx and VOCs concentrations at the ground level and within the planetary boundary layer. We will also exam the emissions deficiencies due to the lack of understanding related to the volatile consumer products (VCPs); motor vehicle emissions deficiencies due to an influx of ecommerce warehouses in the LA basin; the growing contribution of VOC emissions from VCPs; and cooking emissions in urban areas. Moreover, the 2021 modeling platform will serve as a testbed for the model attainment demonstrations required for development of State Implementation Plans under the new lower annual PM2.5 NAAQS. In California, this means upwards of 15 non-attainment regions, many of which are smaller rural air districts that have never required a comprehensive emissions inventory or modeling evaluation.

Katie Tuite¹, Chris Emery¹, Ling Huang², Pradeepa Vennam¹, Greg Yarwood¹

¹Ramboll Americas Engineering Solutions 7250 Redwood Blvd., Suite 105 Novato, CA 94945

²School of Environmental and Chemical Engineering, Shanghai University, Shanghai, 200444, China

The U.S. Environmental Protection Agency (EPA) recently lowered the National Ambient Air Quality Standard for particulate matter with a diameter of 2.5 micrometers or smaller (PM2.5). Because of this, state air quality agencies are preparing for PM modeling that may be required to demonstrate compliance with the new standard. Secondary organic aerosol (SOA), which is formed chemically in the atmosphere from precursor emissions, can contribute a large fraction of PM2.5. In this study, we implement and test new SOA schemes in the Comprehensive Air quality Model with extensions (CAMx), a photochemical grid model which is used worldwide for modeling PM2.5, oxidants, and air toxics. Three new SOA schemes were developed and tested in CAMx alongside the existing scheme, SOAP2. Two of the new schemes have SOA yield parameters that emulate the AERO7 and CRACMM schemes in the Community Multiscale Air Quality (CMAQ) model. The third scheme (referred to as SIMPLE) assumes all SOA is non-volatile with SOA yields that align with multi-model means. Initial box model tests quantified the response of SOA concentrations to varying anthropogenic VOC and NOx emissions. The SOAP2, AERO7-based, and SIMPLE schemes respond similarly to variations in VOC and NOx emissions. The CRACMM-based scheme predicted a decrease in total SOA when NOx was reduced, which is opposite of the other schemes and experimental findings. Because this behavior is unique to CRACMM, we did not proceed to 3-D testing of the CRACMM-based scheme. CAMx 3-D simulations were conducted with the SIMPLE, AERO7-based (re-named to COMPLEX), and SOAP2 schemes and results were evaluated against observed organic carbon (OC) and PM2.5 from monitoring networks across the US. A statistical and graphical analysis showed comparable model performance among each of the SOA schemes. The SIMPLE and COMPLEX SOA schemes have replaced the SOAP2 scheme in CAMx version 7.3 because they are based on more recently published data for SOA yields. The SIMPLE scheme has additional practical advantages, including the ability to calculate emission-weighted SOA potential and the potential for efficiency gains with CAMx’s source apportionment probing tools by reducing the number of tracer species. The SOA schemes developed through this work were chosen considering that models used for State Implementation Plan (SIP) purposes must predict reasonable SOA concentrations and, importantly, that the SOA response to precursor reductions should be traceable to data that is known with some confidence.

Lakshmi Pradeepa Vennam¹, Prakash Karamchandani¹, Tejas Shah¹, Ralph Morris¹, Karim Hamza², Fred Turrati², Yamada Taiga² and Toru Kidokoro²

¹Ramboll Americas Engineering Solutions, Novato, CA

²Toyota Motor Corporation

Recent reductions in the annual fine particulate matter (PM_2.5) standards from 12 to 9 µg/m³ have created interest in understanding the contributions of specific source categories to PM_2.5, particularly in areas in the US that are already in non-attainment or may potentially become nonattainment areas due to the new regulations. This understanding can be obtained by conducting future year photochemical grid modeling to quantify the contributions of various source types to PM_2.5. In this study, we used the Comprehensive Air quality Model with extensions (CAMx) with the Particulate Source Apportionment Technology (PSAT) probing tool for a future year (2032) to calculate PM_2.5 design values and the contributions of various source categories to primary, secondary, and total PM_2.5 at monitoring locations in 24 key cities throughout the US. The future year design values indicated that the new PM NAAQS is estimated to be achieved in most of the 24 cities investigated in this study except for a few locations in California. In terms of source category contributions, our modeling showed that the other anthropogenic emission sector (consisting of EGU, locomotive, marine, agriculture/prescribed fires, fugitive dust emissions and other area sources) is the major contributor to the total PM_2.5 followed by the off-road sector. The primary PM_2.5 showed relatively higher contribution than secondary PM_2.5 to the total PM_2.5. The source apportionment results also showed that the other anthropogenic source sector is the dominant contributor to primary PM_2.5, indicating that total PM_2.5 is largely influenced by primary anthropogenic fugitive dust emissions and emissions from agricultural/prescribed fires.

Jianping Huang (1), Fanglin, Yang (1), Ivanka Stajner (1), Raffaele Montuoro (1), Jeff McQueen (1), Cory Martin (1), Ho-Chun Huang (1, 2), Kai Wang (1,2), Brian Curtis (1, 2), Hyundeok Choi (1,2), Hiaxia Liu (1,2), Barry Baker (3), Daniel Tong (4), Youhua Tang (3, 4), Patrick Campbell (3,4), James Wilczak (5), Dave Allured (5, 6), Irina Djalalova (5,6), Shobha Kondragunta (7), Chuanyu Xu (7,8), and Fangjun Li (9))

1) NOAA/NWS/NCEP/EMC, 2) Lynker, 3 NOAA/OAR/ARL, 4) GMU, 5) NOAA/OAR/PSL, 6) CIRES, University of Colorado Boulder, 7) NOAA/NESDIS/STAR, 8) IMSG, 9) South Dakota State University

Wildfires are significant sources of numerous air pollutants in the United States (US), posing a major challenge to air quality predictions. In response, the National Oceanic and Atmospheric Administration (NOAA) has developed a regional air quality online prediction system (AQPS) within the Unified Forecast System (UFS) framework. This initiative aims to improve the representation of wildfire emissions and their impact on air quality. The online system integrates the Environmental Protection Agency's (EPA) Community Multiscale Air Quality (CMAQ) model as a column atmospheric chemistry model with the UFS-based atmospheric model. Wildfire emissions are computed utilizing real-time satellite-derived hourly high-resolution fire products, the Regional Hourly Advanced Baseline Imager (ABI), and the Visible Infrared Imaging Radiometer Suite (VIIRS) Emissions (RAVE).

This presentation will provide an overview of the UFS-AQM online prediction system, emphasizing major scientific advancements, followed by a comprehensive evaluation of model performance across various seasons, regions, and intense fire events. The model has been extensively evaluated using various observational datasets, including surface monitoring networks and satellite retrievals. Additionally, a bias correction post-processing procedure has been implemented to refine prediction accuracy. Results show that the newly developed online system substantially improves predictions of daily 8-hour maximum ozone (O3) and 24-hour average particulate matter with diameters less than 2.5 micrometers (PM2.5), as well as alerting their exceedance events.

Further ongoing efforts include the development of a data assimilation approach to constrain chemical initial conditions, using observations of PM2.5 from AirNow, Aerosol Optical Depth (AOD) retrievals from the Visible Infrared Imaging Radiometer Suite (VIIRS), and NO2 retrievals from the TROPOspheric Monitoring Instrument (TROPOMI). Additionally, efforts are directed towards developing machine learning emulators to improve computational efficiency and address the challenges posed by high-resolution predictions.

Air pollution has become one of the most significant environmental risks to humans in Southeast Asia (SEA), detrimental to public health and the natural ecosystem. Ozone (O₃) has recently risen to one of the top of the list among air pollutants. This study advances and applies the CMAQ-adjoint model to quantify the source-receptor (S-R) relationship between O₃ concentration and the precursors’ emissions (i.e., VOC and NO_x) as well as determines the contribution of each precursor’s emission source over different receptor regions in SEA. Eleven country-level receptor regions are defined, including Singapore, Malaysia, Indonesia, Brunei, Philippines, Vietnam, Cambodia, Thailand, Laos, Myanmar, and East Timor. The simulation results show high-level O₃ concentration in Singapore, Jakarta, Bangkok and the Malacca Straits. Meanwhile, O₃ has an overall NO_x-limited formation regime in SEA due to high temperatures and large biogenic emissions. In addition, transboundary O₃ is a large pollutant, while transportation, shipping and industry are the three dominant contributing sectors to the O₃ in most countries in SEA.  The study is anticipated to give a complete understanding of the S-R relationship of O₃ in SEA. It may provide policymakers with the information to evaluate a variety of emission control schemes to reduce O₃.

Hu Jie is a PhD student at the Earth Observatory of Singapore, Nanyang Technological University. Her research interests include surface ozone (O3) and its related health impacts, especially the sensitivity of O3 and O3-induced mortality to precursor emissions.

B.H. Baek, Siqi Ma, Irena Ivanova, Chi-Tsan Wang, Patrick Campbell, Youhua Tang, Daniel Tong 

Department of Atmospheric, Oceanic and Earth Sciences, George Mason University, Fairfax, VA 22030, USA

The dynamic coupling of meteorological conditions and emissions of ammonia (NH₃) and volatile organic compound (VOC) from livestock sources with the Community Multiscale Air Quality (CMAQ) model version 5.4 could provide a more accurate temporal representation of air quality impacts. Traditional emission inventories often use static temporal profile factors that do not account for the variability in meteorological conditions, such as temperature and precipitations, leading to potential inaccuracies in air quality predictions in NOAA’s National Air Quality Forecasting Capability (NAQFC) system. This study integrates a dynamic meteorologically-induced emission coupler, known as MetEmis, considering temperature and precipitations, to better capture the emissions from livestock activities.

Following up on the success with the CMAQ-dynamic coupler for onroad mobile emissions, we have developed the CMAQ-coupler for livestock emission sources, called CMAQ-MetEmis_LV. We conducted simulations over a region with significant livestock farming to evaluate the effects of this dynamic coupling on NH₃ and VOC emissions and their subsequent impacts on air quality.

The outcome indicates that Incorporating meteorologically-driven emission variations has led to significant differences in the predicted emissions of NH₃ and VOCs, especially during the summer season and over the eastern U.S. Unlike the current static temporal allocation approach, the dynamic emissions modeling reflected the decrease of emissions during the warmer and rainy conditions, aligning better with observed concentration patterns. These changes are expected to impact the ambient concentrations of fine particulate matter (PM2.5) and ozone (O₃). We conduct a comparative analysis between static and dynamic emission scenarios to demonstrate that the dynamic coupling reduces the uncertainties in emissions and improves the performance of air quality models in predicting pollutant concentrations and their temporal distributions.

These findings underscore the importance of considering meteorological variability in emission inventories to enhance the reliability of air quality forecasts and highlight the necessity for continuous advancements in emission modeling techniques to address the complexities of air pollution sources and their interactions with atmospheric processes.

Dr. Baek is a research professor from George Mason University, and he has extensive experience in atmospheric chemistry monitoring and modeling as well as emissions modeling. Last 18 years, he has worked on developing the SMOKE modeling system with USEPA, and focusing on enhancing the emissions input for air quality modeling community. Especially, Dr. Baek has been recently working on enhancing the NOx, NH3, PM2.5 annd VOC emissions inventory using the machine-learning CTM model with satellites and other surface observations.

Emma L. D’Ambro, Benjamin N. Murphy, Havala O. T. Pye, and Ivan R. Piletic

Per- and polyfluoroalkyl substances (PFAS) are a class of thousands of industrially-produced compounds that have been released into the environment for decades. They are long-lived and many do not readily degrade chemically, resulting in their distribution throughout the globe in all environmental media, and earning them the nickname “forever chemicals”. Several atmospheric chemical transformations have however been documented, and as our understanding of the present and historical atmospheric emissions of PFAS grows, so too does our understanding of important atmospheric chemical pathways. One such pathway is the hydrolysis of acyl fluoride-containing compounds into carboxylic acid-containing compounds. The hydrolysis of acyl fluorides to carboxylic acids is rapid in bulk condensed water, and the resulting carboxylic acids exhibit higher solubility in atmospheric particles resulting in decreased atmospheric lifetimes due to enhanced deposition. This process is relevant for one compound of particular regulatory and public interest: HFPO-DA (GenX). We apply the Community Multiscale Air Quality (CMAQ) model version 5.4 to a case study in Eastern North Carolina to model the PFAS emissions, transport, and chemical transformations at fine scale (1 km) from the Chemours Inc. Fayetteville-Works facility. We implement condensed-phase hydrolysis of hexafluoropropylene dimer acyl fluoride (HFPO-DAF) to hexafluoropropylene dimer acid (HFPO-DA) in aerosol and cloud liquid water. We evaluate the model’s deposition predictions with deposition measurements in North Carolina and discuss the impact of this chemical transformation on results.

Disclaimer: The views expressed in this presentation are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA.

Kiran Alapaty1, Jesse Bash1, Rob Gilliam1, Christian Hogrefe1, Daiwen Kang1, Barron Henderson1, Alan Vette1, Saravanan Arunachalam2, and Chris Nolte1

For the past half century similarity functions have been widely used as universal functions in atmospheric boundary layer modeling for meteorology and air quality models. Many box models that are driven by field-scale measurements don’t require usage of similarity functions since field measurements provide natural variability of the atmosphere. However, in 3-D models similarity functions are necessary for accurate model predictions of boundary layer evolution and associated processes. A proliferation of similarity methods has contributed to a divergence of formulations, which has contributed to differences in the outcomes among different 3-D atmospheric models. To address this cause of model divergence, we have developed a seamless universal formulation for use in box as well as 3-D atmospheric models. This innovative methodology avoids the usage of similarity functions and has the potential to unify models in representing turbulence effects in atmospheric boundary layer modeling. The basis of the new methodology was the development of a 3-D turbulence velocity scaling, which is parameterized using the surface turbulence kinetic energy and was validated using field scale measurements. Meteorology and air quality models were then used at 108 km grid spacing covering the northern hemisphere to perform two sets of numerical simulations: (1) Use standard similarity profile methods (SPM); and (2) Replace the SPM with the newly developed 3-D turbulence velocity-based method (TVM) in both WRF and CMAQ models (while prior work was limited to the implementation of TVM only in WRF model). Simulated meteorological parameters that drive an air quality model which simulates ozone and particulate matter (PM2.5) were intercompared and evaluated against measurements for the year 2018. Preliminary results indicate that the simulations using TVM are at least as accurate as those using similarity functions in both the meteorology and air quality models.

Kiran Alapaty is the Senior Science Advisor in the Atmospheric & Environmental Systems Modeling Division in the ORD of US EPA. He is a part-time researcher for the past 20 years focusing on developing scale-aware cumulus convection scheme, development of new ways to model atmospheric boundary layer processes. Before joining EPA, Kiran was directing the Dept of Energy's national climate programs: ARM and ASR programs, and the global climate program at National Science Foundation. Kiran holds a Ph.D. degree from NC State and a M.S. from Indian Institute of Sciences.

Colleen B. Baublitz, Christian Hogrefe, Olivia E. Clifton, Paul Makar, Jesse O. Bash, Kathleen Fahey, Rohit Mathur, John T. Walker, Philip Cheung, Stefano Galmarini, Alma Hodzic, Aurelia Lupascu, Jon Pleim, Young-Hee Ryu, Roberto San Jose

Reactive nitrogen and sulfur deposition can degrade natural resources and affect the lifetime of air pollutants that harm human health. The National Trends network offers long-term measurements of wet deposition to facilitate constraining regional chemical transport models over North America. Here we evaluate an ensemble of simulations from the fourth round of the Air Quality Model Evaluation International Initiative (AQMEII4) for their representation of the seasonal and spatial spread in nitrogen and sulfur wet deposition. A larger inter-model spread could indicate conditions or regions for which expanded measurement constraints may facilitate evaluation and prioritize directions for model development. Preliminary results using the 2010 simulations suggest that summer has the largest model spread (defined here as the standard deviation across models in space and/or time) in wet deposition for each of nitrate, ammonium and sulfate, corresponding to their summertime peak. Between spring and fall, the relative variation (across models, the standard deviation divided by the mean) is largest between the Baja Peninsula and California. Inter-model differences in the representation of aerosol species’ wet deposition could influence their representation of the aerosol lifetime. The inter-model spread in precipitation and each species’ wet deposition correlate for several months of the year, especially in the summer, suggesting that model differences in precipitation fields may contribute to the wet deposition spread. Future work will extend this analysis to an additional year (2016) to consider a potential role for interannual variability. We will compare these simulations with NTN data to evaluate individual ensemble member performance, with a special focus on the summer and in California, where the intermodel relative variation is largest.

Colleen Baublitz is a Physical Scientist in the Air Quality Modeling Group in the EPA's Office of Air Quality Planning and Standards.

Robert Gilliam, Michael Morton and Wyat Appel

The Atmospheric Model Evaluation Tool (AMET) was developed and first presented as a model evaluation option for air quality and meteorology at the 4th Annual CMAS conference in 2005. Since that time, AMET has evolved with periodic version updates, and more recently the distribution of new releases and more frequent bug fixes via Git-Hub (https://github.com/USEPA/AMET). In a recent effort to make AMET more user friendly, the US EPA has developed a Java-based Graphical User Interface (GUI) to broaden the analysis capabilities that are currently performed using Unix C shell scripts. The GUI essentially provides all the query criteria and analysis options in a single unified interface. Furthermore, an Amazon Web Services (AWS) AMET server instance has been configured including the AMETGUI with test datasets and distributed as an Amazon Machine Image (AMI). Traditionally AMET required users to install external software and other configuration steps on their local servers or systems that often require IT administrators and higher degree of Unix expertise. The AMI can instantly allow users to become familiar with AMET matching and analysis steps with a populated database and underlying model and observation datasets and quickly transition to evaluating their own model simulations.

Robert has been involved in meteorological modeling at the US EPA for 20+ years with a primary focus on developing the Atmospheric Model Evaluation tool. Additionally, Robert has been closely involved in running, evaluating and improving model simulation for air quality applications using models such as MM5, WRF and MPAS from global to fine-scales. 

K. Wyat Appel, Christian Hogrefe, Kristen Foley, Fahim Sidi, Benjamin Murphy, Christopher Nolte

Particulate matter (PM) is an important atmospheric pollutant comprised of particles of varying size and chemical properties. The Community Multiscale Air Quality (CMAQ) model uses three lognormal modes to represent the atmospheric particle distribution. These three modes correspond to the Aitken, accumulation, and coarse particle sizes. When summed together, these modes represent the full distribution (e.g., total PM) of respirable particles in the atmosphere. Often, however, the interest is in particles within a specific size range, such as PM2.5, which consists of only those particles less than 2.5 micrometers (µm) in diameter. Early versions of CMAQ did not provide a simple method to subset the three particle size modes in CMAQ into particles smaller than 2.5 µm; instead, PM2.5 was calculated by summing of the particles in the Aitken and accumulation modes. This approach may not accurately represent particles smaller than 2.5 µm, as the upper tail of the accumulation mode can contain particles larger than 2.5 µm. Additionally, the lower tail of the coarse mode can contain particles smaller than 2.5 µm, resulting in particles being erroneously excluded from the sum. Conceptually, a 'sharp-cut' PM2.5 approach is more appropriate when comparing to observed PM2.5 values. In practice, the three particle modes in CMAQ may not accurately represent the particle size distributions in the atmosphere, adding uncertainty to the model PM2.5 estimates. Nolte et al. (2015; https://gmd.copernicus.org/articles/8/2877/2015/) compared CMAQ PM2.5 values based on the modal sum and sharp-cut approach and found that while there could be relatively large differences (> 2 µgm-3) between the two calculations for a single day and site, the difference between the two values became relatively small (0.04 – 0.2 µgm-3) when aggregated over space and time. This indicated that aggregate evaluation metrics were likely to lead to the same conclusions about model bias and error using either approach. Based on the uncertainty in the particle size distributions and these findings, many subsequent CMAQ evaluations against observations continued to use the sum of the Aitken and accumulation modes to represent PM2.5. Since publication of that work in 2015 there have been significant updates to the CMAQ modeling system. This work will utilize output from the EPA’s Air QUAlity TimE Series (EQUATES) to reexamine the comparison of the modal sum and sharp-cut PM2.5 estimates using a newer version of the CMAQ modeling system (v5.3.2). The EQUATES data also allow for analysis over a long time period (2002-2019) to explore whether the differences between the two methods for computing PM2.5 have changed over time because of emissions-driven changes in aerosol composition. Comparisons of the two model PM2.5 values against observations from several air quality networks across the United States will also be presented to assess how the use of each method for computing PM2.5 compares to observed values.

Baker, Kirk

It is useful to understand the potential air quality impacts related to new construction and operation/maintenance emissions from Outer Continental Shelf (OCS) wind energy projects located off the Northeast coast of the United States given the proximity to areas not compliant with the O3 and annual PM2.5 National Ambient Air Quality Standards. The Community Multiscale Air Quality (CMAQ) photochemical grid model was applied for 1) a representation of current pre-construction conditions, 2) a future year that includes construction emissions from offshore wind energy projects, and 3) a future year that reflects operation emissions for these same projects. An alternative future year representing operating emissions was also modeled to include projected reductions in onshore electrical generation resulting from the increased supply of offshore energy. Human health effects for each future year scenario were quantified based on the change in population-weighted air quality related to seasonal (April to September) average maximum daily 8-hr average (MDA8) O3 and annual average PM2.5.

Jeff Vukovich Office of Air Quality Planning and Standards U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

James Beidler Center for Environmental Modeling and Measurement U.S. Environmental Protection Agency, Research Triangle Park, NC 27711

Christine Allen General Dynamics Information Technology 79 T.W. Alexander Drive, Research Triangle Park, NC 27709

Rhonda Payne Western States Air Resources Council (WESTAR) Helena, MT 59601

A year 2022 wildland fire emissions inventory was developed to support the 2022 Emissions Modeling Platform as part of a Collaborative effort that involved various federal, state, and local agencies. Over 30 government agencies submitted fire activity information that was organized and ingested into the Python-based SMARTFIRE2 tool to generate daily acres burned data. The daily acres burned data was input into the US Forest Service’s Bluesky Pipeline system to generate daily emissions for prescribed burns and wildfires for all 50 states. A summary of acres burned and emissions, lessons learned, and key areas needed for improvement will be presented.

James joined the emissions systems analysis branch of EPA's ORD earlier this year after nearly two decades as an emissions modeling contractor. He has contributed to the development of wildland fire emissions for the 2017 and 2020 National Emissions Inventories along with multiple intermediate platform years.

Dan Loughlin, Joyce Kim, and Trinity White

Historically, low-income disadvantaged communities have been disproportionately impacted by environmental stressors, including exposure to poor air quality. As the United States seeks to decarbonize our energy system, air quality is expected to improve since carbon dioxide and air pollutants are co-emitted by many of the same sources. However, not all low-carbon technologies are also low in air pollutants. For example, biomass and fossil fuels, paired with carbon capture, could have significant particulate matter emissions that would impact downwind communities. Thus, when developing decarbonization strategies, there are important considerations about who would be impacted positively and negatively, as well as how these changes impact historical disparities. In the work presented here, we use EPA’s GLIMPSE air-climate-energy planning tool to simulate the emissions implications of several decarbonization pathways. The air quality impacts are then evaluated using the reduced-form air quality model within EPA’s COBRA tool. County-level impacts on ozone and fine particulate matter concentrations are compared with demographic data from EPA’s EJScreen and from the U.S. Census, and categorical regression trees are applied to develop insights about the environmental justice implications. Among the results that will be presented are an assessment of how these implications differ across pathways and for different areas of the country.

Dan Loughlin is a senior scientist in the U.S. EPA’s Office of Research and Development, where he has conducted energy system and integrated assessment modeling for nearly twenty-three years. Dan leads the GLIMPSE project ( https://epa.gov/glimpse), an effort to develop a decision support tool with state-level resolution for long-term air quality, climate, and energy planning. GLIMPSE was released to the public in the summer of 2023. Dan was a contributing author to the Energy Chapter of the recent 5th National Climate Assessment. He has a doctoral degree from NC State University and is an adjunct professor at Duke University where he teaches a course on Integrated Assessment Modeling.

James Beidler, Jeff Vukovich, George Pouliot, Venkatesh Rao

The burning of piled woody biomass, or pile burns, is a major method for disposing of cut and downed wood from logging operations and for wildland hazardous fuel reduction. Pile burning is the predominant prescribed fire practice in the northwestern United States and is extensively utilized throughout the west and southeast. Historically, pile burn emissions have not been included in the National Emissions Inventory because they have not been reported by the states, cannot be classified or detected through older methods of remote sensing, and/or have been difficult to estimate. For the first time in the 2022 collaborative emissions modeling effort, we have developed a method to estimate pile burn emissions, with assistance from a few key states and improved remote sensing capabilities to detect these fires. Over a dozen states, federal agencies, and tribes submitted pile burning activity for the 2022 modeling platform. The method for estimating pile burning emissions, utilizing various activity data sources, was applied as part of the EPA 2022 emissions modeling platform. Using this method pile burns were estimated to contribute approximately 11% of the total prescribed fire PM2.5 in the western United States. Here, we will describe the emissions estimation method and the resulting 2022 estimates, as well as discuss improvements for application in the 2023 NEI.

James joined the emissions systems analysis branch of EPA's ORD earlier this year after nearly two decades as an emissions modeling contractor. He has contributed to the development of wildland fire emissions for the 2017 and 2020 National Emissions Inventories along with multiple intermediate platform years.

Yuhong (Ruby) Tian¹, Winston Hao¹, Eric Zalewsky¹, Jeongran Yun¹, Alexandra Karambelas², and Kevin Civerolo¹

¹Division of Air Resources, New York State Department of Environmental Conservation

²Northeast States for Coordinated Air Use Management

Approximately 30 million people in the mid-Atlantic and Northeast regions experience unhealthy levels of ozone (O₃) and sometimes PM_2.5. High ground-level concentrations of these pollutants can decrease lung function and trigger asthma attacks, which is a serious human health concern. US EPA’s tighter standards pose persistent challenges for our air quality management. In this study we used the Community Multiscale Air Quality (CMAQ) model v5.4 to examine the response of O₃ and PM_2.5 due to emission reductions resulting from Whole Home Electrification (WHE). The WHE refers to electrification of residential space heating and cooling, water heating, and appliances, and eliminates fossil fuel consumption and emissions for these needs.

We ran CMAQ for summer (June and July) and winter (January and February) months using 2026 projected emissions from the EPA 2016 V3 Emission Platform. The Integrated Planning Model (IPM) projected electric generating unit (EGU) emissions (IPM PTEGU) were replaced with the Eastern Regional Technical Advisory Committee (ERTAC) projected EGU emissions (PTERTAC) in this study. The sensitivity tests contained two steps. Step 1 was to eliminate fossil fuel consumption and emissions for residential fuel combustion use. The residential fuel includes distillate oil, residual oil, natural gas, liquified petroleum gas and kerosene. Step 2 was to modify EGU emissions to account for the increased electricity demand due to WHE by applying adjustment factors. Such extra demand was split between heating season and cooling season with water heating spread evenly throughout the year based on studies from Northeast States for Coordinated Air Use Management (NESCAUM) using National Renewable Energy Laboratory (NREL) ResStock tool. The adjustment factors were applied to all model species in the PTERTAC EGU output files generated from the Sparse Matrix Operator Kernel Emission (SMOKE) processing system.

The sensitivity tests showed that impacts on elevated summertime O₃ are generally mild. The daily maximum 8-hour average (MDA8) O₃ decreased by about 0.5 ppb on high (>60 ppb) O₃ days, with isolated O₃ increases near NYC due to reduced NOx titration. The added electricity caused some additional decreases in regions of elevated O₃. Impacts on PM_2.5 mass and species were larger during the winter months. PM_2.5 decreased by as much as 0.5-1 ug/m³ and >4 ug/m³ in New York City. NO₃ decreased by up to 1.5 ug/m³, which accounted for a substantial portion of the PM_2.5 decrease.

Yuhong (Ruby) Tian received her Bachelor's and Master's degrees in Meteorology in China and her Ph.D. in Geography in the United States. She is a research scientist at the New York State Department of Environmental Conservation. Her research focuses on understanding ozone and particulate matter (PM) formation through remote sensing data analysis and photochemical modeling.

M. Talat Odman, Zongrun Li, Kamal J. Maji, Yongtao Hu, Ambarish Vaidyanathan, Armistead G. Russell

Prescribed burning is an effective land management tool for reducing wildfire risks. However, smoke from both wildfires and prescribed fires degrades air quality and impacts human health. Additionally, prescribed burning does not necessarily prevent wildfires; therefore, air quality impacts from post-prescribed burn wildfires should also be considered in evaluating the tradeoffs between wildfires and prescribed fires.

In this study, we focused on the 2016 Gatlinburg fires and used a modeling framework to compare the impacts of actual wildfires to those of counterfactual prescribed fires followed by post-prescribed burn wildfires. While this framework has been used before, we improved on the design of counterfactual prescribed burns by considering fire breaks in setting the boundaries of burn units, practicability in deciding on fire sizes, and meteorological conditions in selecting burn days. We estimated all fire emissions based on high resolution fuel maps considering the heterogeneity of fuels within the fire boundaries. These fire emissions were then input to a chemical transport model to simulate the air quality impacts. Finally, population-weighted smoke concentrations and smoke impacted populations under different fire scenarios were analyzed to evaluate the tradeoffs between wildfires and prescribed fires.

For the 2016 Gatlinburg fires, total emissions from prescribed fires and post-prescribed burn wildfires were less than the emissions from wildfires. This led to lower daily average PM2.5, MDA8-O3, and 1-hr max NO2 concentrations and exposures. In the county where the fires occurred the daily average PM2.5, MDA8-O3, and 1-hr max NO2 decreased by 5.28 g/m3, 1.68 ppb, and 0.26 ppb, respectively. Smoke exposure times-per person also decreased. Hence prescribed burning was beneficial for air quality and public health in this case study. The benefits came primarily from preventing the long-distance transport of smoke and reducing smoke concentrations.

Chris Nolte, Dan Loughlin, Ben Murphy, Jesse Bash

Combustion sources that emit carbon dioxide (CO2) also emit reactive nitrogen species and other atmospheric pollutants. Thus, recent decarbonization targets at the national and state levels have the potential to bring about substantial co-benefits for air quality. Here we apply the Global Change Analysis Model with state-level resolution of the U.S. energy system (GCAM-USA) to project future emissions under the following two scenarios: a reference scenario that includes provisions of the Inflation Reduction Act and an alternative scenario that reaches net-zero CO2 emissions in 2050. We quantify the air quality impacts of these emissions changes associated for the years 2035 and 2050 using the Community Multiscale Air Quality (CMAQ) model v5.4. The Detailed Emissions Scaling, Isolation, and Diagnostic (DESID) module applies state, sector, and pollutant-specific scaling factors from GCAM-USA to emissions from EPA’s Air Quality Time Series (EQUATES) project for 2016. This presentation will discuss impacts on modeled ozone and PM2.5 concentrations and compare results to those from the EPA’s Avoided Emissions and Generation Tool (AVERT) reduced-complexity model. These scenarios will also be used with CMAQ’s Integrated Source Apportionment Method (ISAM) to isolate the impacts of specific economic sectors on nitrogen deposition to the Chesapeake Bay watershed.

Amber Soja, Emily Gargulinski

In the Central and Southeastern U.S. (C& SE), prescribed, small, and short-lived fires are underestimated by satellites. Accurately detecting and quantifying fires in the C& SE is challenging due to size, time of burning, cloud cover, and quick regrowth/plowing soon after burn. Satellite data are critical to defining fire in our National Emissions Inventory (NEI), because historically, ground-based inventories are incomplete and can be geographically inaccurate. Quantifying burned area (BA) and emissions in the C& SE will provide the basis for Health Impact Assessments, with the goal being to inform communities of potential health impacts. As part of NASA’s Health and Air Quality Applied Sciences (HAQAST) team, we are building a C& SE BA inventory, focusing on these often ‘missing’ small fires. We have sampled 100 random 5 x 5 km tiles from the Flint Hills and have digitized burn scars from the March – April 2022 burning season using Sentinel-2 10 meter data. Preliminary results of 355 burning events show 66% of the burn scars had at least one detection between VIIRS and MODIS satellites. Burn scars without detections averaged 57 acres and burn events that had detections averaged 750 acres.

Huazhen Liu, Jared H. Bowden, Timothy Glotfelty, Jordan Kern, J. Jason West

Drought is a complex and multivariate phenomenon influenced by diverse physical and biological processes. It occurs naturally, but climate change can lead to more frequent and intense drought events in some regions. The strong perturbation of drought on the atmospheric water cycle can influence air quality. The high electricity demand and low streamflow, which decreases hydropower, during drought periods lead to more power generated by fossil fuels, which exacerbates air pollution. Possible increases in wildfires during drought events also exacerbate air pollution. In this study, we model the effects of drought on air pollution in California, contrasting drought and non-drought years, separating the influences of atmospheric and natural processes, fire emissions, and emissions from fossil fuel electricity generation. We compare a severe drought year in California (2021) with a non-drought year (2019). According to observations in the dry season (April to September), the average temperature in 2021 is 0.68℃ higher than in 2019, and the relative humidity is lower (54.08% vs. 59.78%). The total power generated by fossil fuels is 39.90% higher than in 2019, leading to higher emissions of SO₂ (63.18% higher) and NOx (50.23%). The total wildfire-burned area in 2021 is 14.32 times that in 2019. The averaged daily PM_2.5 (particulate matter with diameter less than 2.5 μm) concentration in 2021 is 12.61 μg/m³ and the averaged ozone MDA8 (maximum daily 8-hour average) concentration is 48.88 ppb, both of which are higher than in 2019 (PM_2.5: 6.41 μg/m³; O₃: 45.95 ppb). To simulate meteorology and air quality, we used the WRF-Chem model with nested grids at 12 km (the outer domain, over the Western US and part of the Pacific Ocean) and 4 km (the inner domain, over California and Nevada) resolution. The high resolution allows for more detailed representation of key processes, including the effect of topography. WRF-Chem uses the MOZART-MOSAIC gas phase and aerosol mechanisms and the Noah land surface model. In simulations of the dry season in the drought year 2021 and the non-drought year 2019, the model generally performs well for meteorological parameters as well as PM_2.5 and ozone. By replacing the meteorology of 2021 with 2019, we determine the contribution of meteorological changes on exacerbated air pollution. Similarly, the contributions of changes in power plant emissions and fire emissions are also determined.

Azimeh Zare, ShihMing Huang, Samantha Kramer, Farnaz Hosseinpour, Kayla Besong, Charles Scarborough, Naresh Kumar, and Tim Brown

we developed a turnkey, online decision-support tool named SmokePath Explorer for prescribed fire planning and wildfire response. Sponsored by CAL FIRE, the tool allows users to specify a time window and a location for a planned prescribed fire or current wildfire and view the fire weather statistics and the probability of potential downwind smoke impacts. Based on 20 years (2001-2020) of high-resolution (2 km) climatological data for California and Nevada, fire weather statistics were computed, and air transport data were created through hourly air parcel trajectory modeling. For user-specified location, years, and months or weeks of the year, the SmokePath Explorer: a) Determines the probability of air parcel transport downwind, b) Displays the historical summary of weather parameters relevant to fire and smoke conditions, and c) Identifies the potential receptor locations by integrating air parcel transport probability with population, education facility, healthcare facility, and ZIP code data. Air and land managers can use the SmokePath Explorer to plan prescribed fires, allowing for more timely public messaging regarding potential smoke impacts, and proactive mitigation efforts to assist at-risk populations. In addition to the tool, the project team has identified multiple case studies based on past prescribed fire projects and wildfires to evaluate the transport modeling results and establish guidance for utilizing the tool. These cases include pile burns, broadcast burns, and typical wildfires in different air basins across California. Results of these case studies and the tool will be presented.

Jordan French; University of Texas at Austin; Department of Civil, Architectural and Environmental Engineering
Sarah Alverson; University of Texas at Austin; Department of Civil, Architectural and Environmental Engineering
Pedro A. Sánchez-Pérez; Environmental Systems Graduate Program, School of Engineering; University of California Merced
Martin Staadecker; Division of Engineering Science, University of Toronto; Mechanical and Aerospace Engineering, University of California San Diego
Patricia Hidalgo-Gonzalez, Mechanical and Aerospace Engineering, University of California San Diego
Sergio Castellanos; Department of Civil, Architectural and Environmental Engineering; University of Texas at Austin

Capacity expansion models have proven valuable for developing cost-optimized decarbonization portfolios for grid systems; however, air pollution and environmental justice impacts are often neglected or oversimplified. This work attempts to address this gap in the literature by pairing SWITCH, an open-source capacity expansion model, with pollutant dispersion modeling through an iterative procedure. The Texas ERCOT grid is modeled as an example, with analysis conducted to determine the siting and emissions of future natural gas plants. NOx emissions dispersion is then modeled with disperseR and InMAP, two reduced complexity pollutant dispersion models. The results of these models are used to define environmental justice constraints, which are incorporated into SWITCH for equitable decarbonization pathways. In developing this framework, the results of disperseR and InMAP are compared to determine the sensitivity of environmental justice metrics to monthly wind patterns, particulate matter formation, interstate considerations, and facility siting decisions. This methodology lays a foundation for applying environmental justice constraints in the development of equitable grid decarbonization pathways, and highlights potential paths forward for more effective constraints.

Jordan is a graduate researcher in the Rapid, Equitable, and Sustainable Energy Transitions (RESET) Lab at the University of Texas at Austin. His work applies a combination of pollutant dispersion modeling, air quality monitoring, and community-based research methods to understand the environmental justice implications of the energy transition.

Heather Simon, James Beidler, Kirk Baker, Barron Henderson, Loren Fox, Chris Misenis, Jeff Vukovich, Norm Possiel, Alison Eyth

The summer of 2023 was a record-breaking season for wildfire acres burned in Canada. Many of the wildfire plumes reached portions of the central and eastern United States that have not experienced frequent and severe wildfire impacts in recent years. At the same time, much of the US experienced ozone concentrations that were unusually high compared to the last decade. Photochemical modeling is a valuable tool for assessing the contribution of fire emissions to ozone formation. Here we present the Expedited Modeling of Burn Events Results (EMBER). EMBER is a rapid-turnaround screening-level CMAQ modeling dataset covering the summer of 2023. The modeling uses 36-km grid resolution and leverages existing publicly available datasets to obtain meteorology and boundary condition inputs. Fire emissions across the US and Canada were developed for this effort using the BlueSky pipeline modeling system. Emissions from a recent 2018 emissions modeling platform are used for emissions other than wildfires. Zero-out modeling was conducted to estimate the ozone concentrations resulting from Canadian Wildfire emissions as well as from the combination of US and Canadian prescribed and wildfire emissions.

Dr. Heather Simon is a physical scientist in the US EPA’s Office of Air Quality Planning & Standards where she performs photochemical modeling and ambient air quality data analysis to support national air pollution regulatory efforts. Dr. Simon has worked on EPA rules relating to ozone National Ambient Air Quality Standard (NAAQS) standard setting and implementation. Dr. Simon has also worked on characterizing air quality impacts as part of regulatory impact assessments for power sector regulations. Some of her technical interests include atmospheric chemistry, ambient ozone trends, ozone impacts on human health and ecosystems, ClNO2 chemistry, organic aerosols, emissions inventories and PM2.5 speciation. Dr. Simon received a BS in Earth Systems from Stanford University and MSE and PhD degrees in Environmental Engineering from the University of Texas at Austin.

Shreya Guha, Ting Zhang, Patrick L. Kinney, Lucas R.F. Henneman

Fine particulate matter (PM_2.5) and surface ozone air pollution have severe health consequences. Since China implemented the nationwide Air Pollution Prevention and Control Action Plan (APPCAP) in 2013, annual average PM_2.5 concentrations have decreased while O₃ concentration have increased in Beijing. It is unclear, however, to the extent that short-term meteorological variability has influenced these trends. To separate the influence of meteorology from long-term emissions-induced changes, we quantify the portion of the long-term PM_2.5 and O₃ concentration signals associated with short-term meteorological variability. Building off recent literature highlighting the regional nature of pollutant variability, we incorporate both locally observed meteorology and average regional reanalysis. We isolated sub-seasonal variability using the Kolmogorov-Zurbenko filter. Next, we trained and compared the utility of three statistical models of varying complexity (a linear model, a general additive model with splines, and a random forest) in their ability to predict out-of-sample daily concentrations. The random forest model yields the best predictive capability in holdout tests and predicts changing meteorological contributions to PM_2.5 and ozone concurrent with changing concentrations across the study period. Results show an overall decrease in meteorology induced PM_2.5 concentration variability at daily scales since 2013, but variability from meteorology has increased relative to mean PM_2.5 concentrations. In contrast, O₃ shows the opposite trend, with increasing meteorology-induced variability (meteorology-induced variability normalized by mean observed O₃ has decreased). Our results show the importance of including regional meteorology in explaining local PM_2.5 and O₃ variability and quantify links between long-term policy implementation and short-term meteorological contributions to pollutant concentrations.

Keywords: PM_2.5, Ozone, Meteorological Detrending, Machine learning

Taylor Y. Wilmot, Heather A. Holmes, Derek V. Mallia, Kerry E. Kelly, Ross T. Whitaker

In light of the public health burden presented by particulate and ozone pollution, nationally and especially in the western U.S., it is evident that there is a need for high-quality and timely physics-based air quality forecasts. The state of Utah has elevated ozone and particulate matter (PM) concentrations and is currently in nonattainment or maintenance for the PM2.5, PM10, and ozone National Ambient Air Quality Standards (NAAQS). The sources contributing to these elevated pollutant concentrations vary significantly in space and time (e.g., dust, wildfire smoke), often resulting in acute air pollution episodes. Leveraging an NSF CIVIC project titled Community Resilience through Engaging, Actionable, Timely, high-rEsolution Air Quality Information (CREATE-AQI) we developed an air quality forecast system for Utah communities. Through CREATE-AQI our goal was to co-develop community-scale, actionable air-quality products for the state of Utah to help reduce exposure to air pollutant hazards, particularly for children. The CREATE-AQI public facing user interface displays both low-cost sensor observations (nowcasting) and three-day air quality predictions (forecasting) using visualizations that were co-developed with our community partners. Our objective is to provide actionable information about current and future air quality to local stakeholders.

This talk will present results from the CREATE-AQI modeling platform showing the integration of the nowcasting and forecasting air quality products. The focus will primarily be on the development and evaluation of the numerical air quality forecasting platform. The forecasting framework uses the Weather Research and Forecasting (WRF) model to forecast meteorology, the U.S. EPA National Emissions Inventory (NEI) and the Sparse Matrix Operator Kernel Emissions (SMOKE) to provide and process emissions, and the Community Multiscale Air Quality (CMAQ) model to forecast air quality. In its first iteration, this air quality forecast provides outputs for the continental United States at 12 km horizontal resolution and hourly temporal resolution out to 72 hours. In future iterations the Utah forecasts will come from a nested 4km horizontal resolution domain to improve the simulation quality over mountainous terrain. The forecast system incorporates novel developments to improve wildfire smoke forecasting. Specifically we (1) forecast the temporal fire emissions to improve on the wildfire persistence assumption that is often used to forecast wildfire emissions, (2) incorporate plume top heights from the Freitas plume rise scheme, and (3) incorporate a forecasted vertical distribution of smoke emissions below the plume top. This vertical distribution of emissions is based on simulated wildfire energetics and atmospheric structure at the time of emission. Combined, these model developments improve the air quality forecasting capabilities in the western U.S.

Taylor (Kai) is a postdoc in the chemical engineering department at the University of Utah, working under advisor Heather Holmes. He earned his PhD in atmospheric sciences from the University of Utah in 2022, where the research for his dissertation focused on wildfire smoke transport across the western US. His current research focusses on forecasting wildfire smoke emissions, wildfire plume rise, and the associated downwind air quality impacts. To date, most of Taylor’s research has utilized the HYSPLIT-STILT trajectory model, he is a relatively new CMAQ user and looks forward to learning more from the CMAQ community.

Xiaorong Shan1, Joan A. Casey2, Lucas Henneman1

1 Department of Civil, Environmental, and Infrastructure Engineering, George Mason University, Fairfax, VA, USA; xshan2@gmu.edu; lhennem@gmu.edu

2 Department of Environmental and Occupational Health Sciences, School of Public Health, University of Washington, Seattle, WA, United States

Objectives

Our team aims to test the relationship between 1940 air quality and dementia onset in the 1990s-2000s, but this work is limited by a lack of air quality data, including emissions information and historical measurements for evaluation, prior to 1970 in the United States. Here, we devised and implemented advanced methodologies to quantify the air quality in 1940 and compared our estimates to present-day where we could validate our methods using air quality monitors.

Materials & Methods

We developed several county-level source-specific exposure metrics across the contiguous United States in 1940 and 2010. We quantified exposure to traffic-related air pollution (TRAP), oil/gas wells, power plants, and fine particulate matter (PM2.5). For TRAP, we incorporated information on statewide gasoline sales and county-level population to estimate gasoline consumed. Using oil/gas well locations, we employed Inverse Distance Weighting (IDW). For fossil fuel power plant locations, we employed IDW. We quantified PM2.5 concentrations using climate model output from the Coupled Model Intercomparison Project Phase 6 (CMIP6). We evaluated these source-specific exposure estimates against observed 2010 annual concentrations from EPA Air Quality System monitors, using Pearson correlation coefficients and mean differences as evaluation metrics.

Results

The initial evaluation revealed that CMIP6-based simulations correlated moderately with observed PM2.5 concentrations (Pearson: 0.40). In contrast, automobile emissions showed a weak correlation with PM2.5 (Pearson: 0.04), but a stronger one with NOx (Pearson: 0.41). Well activity presented stronger correlations with VOCs (Pearson: 0.34). The IDW approach to power plant emissions indicated a fair correlation with PM2.5 (Pearson: 0.29) and a higher correlation with sulfate (Pearson: 0.33).

Conclusions

Positive correlations suggest expected relationships between the sources of interest and observed air pollutant concentrations. The somewhat lower than expected correlations, however, suggest some refinement to the exposure methods may be necessary. In addition, there may be utility in comparing the concentrations at finer spatial scale than county level.

Zongrun Li, Yongtao Hu, Armistead G. Russell, M. Talat Odman

Coupled fire-atmosphere models that simulate the fire propagation, smoke plume formation and transport under different ignition patterns are useful tools for the management of prescribed fires. Emissions and meteorology are the primary sources of uncertainty in these models, especially for fire simulations at high spatial and temporal resolution. Understanding the impacts of these uncertainty sources is important for improving model performance and increasing confidence in the model used for air quality forecasting.

We designed and implemented numerical and statistical methods to quantitatively analyze the impacts of wind uncertainties on downwind smoke concentrations in prescribed fire simulations. Our numerical methods aim to mitigate wind uncertainty during meteorological simulations. Data assimilation with different intensities and modifying initial and boundary conditions are the numerical methods. The statistical methods provide an uncertainty range by considering the differences between observed and simulated winds in the post-analysis of modeled smoke concentrations.

To illustrate and evaluate the designed methods, we simulated prescribed fire events with precise ignition information and downwind smoke measurements using the coupled fire-atmosphere model WRF-SFIRE at high spatiotemporal resolution. Among the numerical methods, modifying initial and boundary conditions by incorporating the regional observed winds is the most effective way to reduce the wind bias, especially the wind speed bias. The performance of the smoke concentration simulation is enhanced by improving the accuracy of the wind simulation. As algorithms for post-analysis of numerical modeling results, the statistical methods provide computationally efficient sensitivity analysis and uncertainty estimation for smoke concentrations by considering the simulated wind bias at monitoring locations. The statistical methods estimated the probability of smoke impacting the monitor location under uncertain wind simulations and demonstrated that the modeled concentrations are highly sensitive to wind conditions. The methods proposed in this presentation can also be applied to other chemical transport models for fire simulations as well as other situations involving point or area sources.

Zongrun Li is currently a PhD student at Georgia Tech. His research interests include empirical modeling and numerical modeling related to air quality. His PhD project is focused on developing and improving air quality modeling frameworks for wildland fires, with a particular emphasis on prescribed fires.

Patrick Campbell, Daniel Tong, Yunyao Li, Shinkuang Chang, Siqi Ma, B.H. Baek, Youhua Tang, Rick Saylor, and John Walker

A consensus has emerged that climate-driven increases in the frequency and severity of wildfires can have serious ramifications for enhanced downwind nitrogen (N) emissions and deposition and associated ecosystem impacts like critical load exceedances (see the September 2023 WMO Air Quality and Climate bulletin: https://library.wmo.int/records/item/62090-no-3-september-2023). There is a need in the community to better understand the long term (~ multi-decadal) fire activity and implications on changes in N emissions, atmospheric deposition, and the impacts on downwind ecosystems across the contiguous U.S. (CONUS). Thus, to aid in this important effort, here we use two 20-year (2002 - 2021) coupled Weather Research and Forecasting (WRF) - Community Multiscale Air Quality (CMAQ) model simulations for both “with-fire” and “without-fire” sensitivity runs to quantify the change in relative trends and contributions of fire sources to the total emissions and deposition trends of reactive (i.e., oxidized) versus reduced forms of N over CONUS. The meteorological conditions are constrained by downscaling the ECMWF Reanalysis v5 (ERA-5) dataset, and the emissions (both for anthropogenic and fire sources) are based on the U.S. EPA’s Air QUAlity TimE Series Project (EQUATES) for 2002-2017, the 2020 National Emissions Inventory (NEI), and 2018v2, 2019, and 2021 emissions modeling platforms. Preliminary model results show that fire activity has increased substantially in the western U.S. fire climate regions, especially in the northwest U.S., and that this increase is associated with positive annual near-surface temperature and vapor pressure deficit anomalies compared to the 2002-2021 average. This has led to pronounced increases in both oxidized (i.e., NOx = NO + NO2) and reduced (i.e., NHx = NH3 + NH4+) forms of N emissions from fires, which can dominate the total amount of N emissions in parts of the western fire climate regions, especially after year 2012. We will further show that the increasing trend of enhanced N emissions from fires lead to increases in dry deposition of N compounds, especially NHx, which has implications for trends of increasing critical load exceedances for major vegetation types of the western U.S. between 2002 - 2021. We anticipate that future climate change may exacerbate such trends in the western U.S.

Dr. Patrick Campbell is an Earth systems scientist and research associate professor at the George Mason University Center for Spatial Information Science and Systems and is the associate director of GMU’s Satellite and Earth System Studies (SESS) program. Dr. Campbell is also a NOAA Air Resources Laboratory Affiliate under the Cooperative Institute for Satellite Earth System Studies (CISESS). Dr. Campbell is working closely with the NOAA-Air Resources Laboratory Chemical Modeling and Emissions Group on scientific research, development, and application of next-generation atmospheric composition and air quality forecasting models that help protect human and ecosystem health.

Xue Meng Chen, Shuming Du, Abdullah Mahmud, Yiting Li, Yunle Chen, Pingkuan Di, and Jeremy Avise

In this study, we quantified the disparity that exists across California in terms of exposure and health impact from major air toxics such as diesel particulate matter (DPM), volatile organic compounds (VOCs) and heavy metals, and how it has changed from 2012 to 2017. To help achieve environmental justice, disadvantaged communities that are overburdened by air pollutants have been identified by various programs such as the Community Air Protection Program (AB617) and CalEnviroScreen (SB535). CARB’s California Air Toxics Assessment (CATA) provides estimates of population exposure and associated cancer risks from major air toxics at the census tract level. Using CATA’s high resolution modeling results, this analysis aims to answer two main questions: 1) how do air toxics related cancer risks in AB617 and SB535 communities compare with the rest of the state, i.e., how bad is the disparity in California in terms of exposure to air toxics, and 2) has that disparity changed over time. Our results indicate that in both 2012 and 2017, most disadvantaged communities still bear a disproportionate cancer risk burden from air toxics. We also found that even though the average cancer risk from air toxics decreased across the state, disparity has not decreased substantially, especially in the southern part of the state.

Abdullah Mahmud is a trained air quality modeler and analyst. He earned his PhD in Civil & Environmental Engineering from the University of California, Davis. He currently works at the California Air Resources Board. Abdullah is primarily interested in understanding the health impacts of exposure to air toxics and promoting environmental justice at both community and regional levels.

Tsengel Nergui, Victor Geiser, and Zac Adelman

The Midwest is a land-locked mid-latitude geographical setting where complex atmospheric processes take place in conjunction with local emissions and transported air pollutants. Periodically, upwind wildland and prescribed fire smoke is transported into the region and results in unhealthy concentrations of fine particulate matter (PM2.5) and ozone (O3) at the surface. Meteorological conditions associated with typical high pollution days, compared to those of fire smoke influenced days, are often less apparent to air quality forecasters and planners. To better understand the meteorological setting and pollutant transport pathways bringing wildland fire smoke into the region, LADCO applied a spatial classification technique called a Self-Organizing Map (SOM) to daily average PM_2.5 and daily maximum 8-hour average O₃ concentrations using the 3-km resolution High Resolution Rapid Refresh (HRRRv4) reanalysis dataset for 2019-2023. The objective of the analysis is to identify the primary features of the atmospheric conditions associated with air pollution episodes with and without the influence of fire smoke. We will present the results of SOM analysis of high pollution episodes caused by wildland fires originating in the southwestern US and southeastern Canada. We used SOM to identify the synoptic-scale meteorological conditions and the corresponding increases in PM_2.5 and O₃ during these fire events. In addition, we investigated a key aspect of whether the long-range transported smoke aloft reached the surface. Vertical atmospheric characteristics such as wind shear, stability, and 24-hour changes in geopotential height and temperature for the fire-influenced SOM nodes highlight key upper-air features for vertical mixing, and indicate whether air masses ascend or descend along the transport path between the fire smoke source and receptor monitors. Our study offers two practical applications for air quality forecasters. First, using SOM to identify the weather patterns associated with typical high-pollution days provides historical data for similar-day analysis for exceptional event applications. Second, the identified synoptic weather patterns linked to fire smoke-influenced days provide insights into the expected increases in PM_2.5 and O₃ concentrations due to the fire smoke in the Midwest.

Kai Fan, Zhen Qu

Air pollution is the fifth leading risk factor for mortality globally. In the U.S., it disproportionately affects marginalized communities, with monitoring stations primarily situated in high-income, predominantly white regions. This leaves vulnerable communities unaware of their exposure levels. To address these challenges, we assessed air pollution exposure disparities in the top 50 populated counties in the contiguous U.S. (CONUS). Our approach combines high spatial-temporal air pollution predictions with health assessments to improve equity and environmental justice (EEJ) in the U.S. We utilized a U-Net-like model with three-dimensional spatiotemporal partial convolutions to fill nighttime AOD and spatial gaps in the missing GOES satellite AOD. Subsequently, a random forest model predicted hourly air pollutant concentrations at a 1-km horizontal resolution, using gap-filled AOD, ground measurements (AQS, PurpleAir), and other geographical data (e.g., population, terrain height, land cover). The gap-filled AOD was evaluated against AERONET observations, achieving a correlation coefficient of 0.8. Site-based cross-validation for daily PM2.5 predictions yielded an R2 value above 0.7. The health assessment estimated annual premature deaths due to PM2.5 exposure over CONUS using our air pollution predictions and the GBD Exchange Results tool. The overall premature death rate for the Black population is 6.1% higher than for the non-Hispanic white population, and low-income census tracts face a 7.0% higher premature death rate. In the top 50 most populated counties, premature death rates vary by racial groups due to the different residential locations of these groups. The premature death rate for the Black population is 12% higher than for the white population because the Black population resides in areas with higher PM2.5 concentrations compared to areas where the white population lives. Our research provides high spatial-temporal resolution air pollution predictions and offers insights into air pollution exposure disparities among races/ethnicities and locations, highlighting environmental justice issues in local communities.

Sambit Bhattacharya, Taylor Brown, Matthew Wilkerson

Wildfire emissions significantly impact air quality, posing health risks and contributing to climate change. Early detection and rapid response are crucial for mitigating these effects. This study presents an image-based wildfire detection system, which is being developed under a project that is executed by a collaboration of three universities and the Jet Propulsion Laboratory (JPL) and is supported by the NASA funded Institute for Multi-agent Perception through Advanced Cyberphysical Technologies (IMPACT). Our approach leverages advanced machine learning algorithms and remote sensing technologies to detect wildfires with high accuracy and minimal latency. By integrating data from the UAVSAR platform and the IMPACT Early Exit framework, we are working on close to real-time monitoring capabilities that are essential for proactive wildfire management. Our system utilizes a combination of Synthetic Aperture Radar (SAR) imagery and optical data to identify wildfire signatures, even under challenging conditions such as dense smoke or cloud cover. The integration of these datasets enhances detection reliability and provides comprehensive situational awareness. This research underscores the importance of cutting-edge image processing techniques in wildfire detection, offering a promising tool for emission control and air quality improvement. Early wildfire detection not only helps in immediate response efforts but also plays a vital role in long-term air quality management, highlighting the intersection of technology and environmental stewardship.

Sambit Bhattacharya is a Computer Scientist with more than 15 years of experience in teaching and research. He has a PhD in Computer Science and Engineering from the State University of New York at Buffalo. He is a tenured Full Professor of Computer Science at Fayetteville State University, North Carolina, USA. In 2023 he was honored by the University of North Carolina (UNC) Board of Governor’s Award for Teaching Excellence. He was a winner of NASA’s Space Tech Catalyst prize in 2024 which was awarded in recognition of work to increase diversity in the workforce. Dr. Bhattacharya is experienced in developing and executing innovative and use-inspired research in Artificial Intelligence and Machine Learning (AIML) with a broad range of techniques and applications, and with multidisciplinary teams. He has several peer reviewed publications and has delivered many oral presentations, including keynote lectures at conferences. As PI he leads projects funded by the National Science Foundation, the US Department of Defense (DoD), and NASA. While on teaching leave from the university, Dr. Bhattacharya has served as faculty research fellow in research labs of the DoD and in the National Geospatial Intelligence Agency (NGA). He is currently a Visiting Scientist part time at NGA. Recently he has started working as Co-PI and AI lead with the UNC Radiation Oncology department on an NIH funded project. He directs the Intelligent Systems Lab (ISL), which supports faculty and students, and collaborations with external partners.

Abdullah Mahmud, Pingkuan Di, and Jeremy Avise

Exposure to toxic Volatile Organic Compounds (VOCs) through inhalation poses significant health risks. Some of the toxic VOCs such as formaldehyde, acetaldehyde, benzene, 1,3-butadiene, perchloroethylene, and p-dichlorobenzene have been identified as human carcinogens. The primary objectives of the current study were to assess the population health risks and, to some extent, to support identifying areas that are likely to experience disproportionately high concentrations of toxics as mandated by the California Community Air Protection Assembly Bill 617 (AB 617). CMAQv5.3.2 was utilized to simulate the physical and chemical processes that the toxic VOCs undergo in the atmosphere. The simulations were carried out at a horizontal resolution of 2-km over an area encompassing major air basins including the Sacramento Valley (SV), San Francisco Bay (SFB), San Joaquin Valley (SJV), South Coast (SC), San Diego (SD), and Imperial (IMP) in California for the entire 2017. The meteorological fields and the speciated and gridded emissions from anthropogenic, biogenic, and wildfire sources that drove the air quality simulations were generated by the Weather Research Forecast version 3.9.1 (WRFv3.9.1) model and the SMOKE emissions processing model, respectively. Gridded annual average concentrations of toxic VOCs were applied in calculating potential cancer risk and burden at the 30-yr and 70-yr exposure levels according to the California Office of Environmental Health Hazard Assessment (OEHHA) guidelines, respectively. These risks were further aggregated to census tract levels. The SC, SJV, SV, and SFB air basins showed to have high levels of cancer risks. Significant risks were also seen in areas in SD and IMP near the California-Mexico border due to transport of emissions from the Mexican side. The population weighted cancer risk in the South Coast was approximately 129 per million which was the highest followed by the risk of about 103 per million in San Diego. The cancer risks in the Sacramento Valley, San Francisco and San Joaquin Valley were 68, 74, and 65 per million, respectively. The comparisons between modeled and monitored annually averaged concentrations for four major VOCs (1,3-butadiene, benzene, formaldehyde, and acetaldehyde) were carried out using available data from monitors located throughout the state. The modeled results for 1,3-butadiene and benzene showed good agreements with the observed data, with index of agreements of 0.7 and 0.6, respectively. CMAQ underestimated the concentrations of formaldehyde and acetaldehyde, particularly during extreme events. The cause of such under-estimation could be attributed to a lack of appropriate representation of chemical processes of formation for secondary acetaldehydes and formaldehyde during wildfire events that occurred in significant number in 2017. The modeled concentrations of formaldehyde will further be evaluated against satellite measurements in the future.

Abdullah Mahmud is a trained air quality modeler and analyst. He earned his PhD in Civil & Environmental Engineering from the University of California, Davis. He currently works at the California Air Resources Board. Abdullah is primarily interested in understanding the health impacts of exposure to air toxics and promoting environmental justice at both community and regional levels.

Xin He, Yakun Zhou, Armistead G. Russell, Jennifer Kaiser

In this study, we employ the Community Multiscale Air Quality (CMAQ) CTM to characterize fine resolution at 1 km x 1 km, investigating how EV adoption would impact emissions of traffic related air pollutants PM2.5, NO2, O3, and estimating health impact disparities between racial/ethnic population at census tract-level. We investigate the effect of the spatial resolution on estimated health burdens. We focus on the most densely populated regions in Georgia, centered around the Atlanta metro area. We consider two EV adoption scenarios (no-EV and 30-EV) and our approach accounts for excess emissions from electricity generation units (EGUs) owed to additional battery charging demands by utilizing a power dispatch model. 

Johnson, J.¹, Goldberg, D.², de Foy, B.³, Nawaz, M.O.², Yarwood, G.¹, Judd, L.⁴

¹ Ramboll Americas Engineering Solutions 7250 Redwood Blvd., Suite 105 Novato, CA 94945

² Department of Environmental and Occupational Health George Washington University Washington, DC 20052

³ Saint Louis University St. Louis, MO 63103

⁴ NASA Langley Research Center Hampton, VA 23681

Ozone formation is limited by nitrogen oxide (NOx) emissions in most North American cities and further reducing ozone production rates within cities will require accurate quantification of NOx emissions with source-sector precision. One major limitation of our current observing networks is the inability to accurately quantify NOx emissions on a sector-by-sector basis. In this study, we leverage spatially continuous column NO₂ measurements from the NASA GeoCAPE Airborne Simulator (GCAS) at 250 × 560 m² resolution, and coarser Tropospheric Monitoring Instrument (TROPOMI) NO₂ columns, to validate a finely resolved 444 x 444 m² CAMx model simulation of the Houston metropolitan area during the September 2021 NASA/TCEQ Tracking Aerosol Convection ExpeRiment - Air Quality (TRACER-AQ) field campaign. We compared CAMx NO₂ columns with observations from the GCAS instrument, Pandora instruments, and TROPOMI to identify gaps in our understanding of NOx emissions within CAMx. We find that column NO₂ from GCAS has excellent agreement with Pandora measurements (r²=0.81 and NMB of +2.4%), column NO₂ from TROPOMI had good correlation with Pandora (r²=0.62), but with a low bias that was unresolved even when we applied a CAMx-based air mass factor (NMB of –11%). Column NO₂ from CAMx showed a low bias when compared with Pandora (NMB of –20%) and GCAS measurements (NMB of –31%). We exploited the ability of CAMx to tag NO₂ concentrations (and columns) by source sector, tagging 15 emission source categories, including 9 EGUs individually. We applied a flux divergence method to the GCAS measurements and CAMx source-apportioned NO₂ columns to better distinguish emission sources, e.g., the linear shape of major highways is revealed by flux-divergence. Using a multiple linear regression model, we found that some tagged sources matched well with the GCAS measurements, including individual EGUs that have continuous emission monitors (CEMs). However, on-road, railyard and “other” NOx emissions were underestimated relative to the GCAS measurements and contribute to the overall low NO₂ bias. Additionally, we identified a likely overestimate of marine shipping NOx emissions. We modified on-road and shipping NOx emissions in a new CAMx simulation and increased the background NO₂ and found better agreement with GCAS measurements: bias improved from –31% to –10% and r² improved from 0.78 to 0.80. We will also present preliminary findings from similar work for the New York City and Chicago areas. This study shows how remote sensing data, including fine spatial information from newer instruments such as Tropospheric Emissions Monitoring of Pollution satellite (TEMPO), can be used in combination with chemical transport models with source apportionment to provide actionable information for air quality managers to identify further opportunities to improve NOx emissions inventories.

Jeremiah Johnson is a Managing Consultant at Ramboll’s Novato, California office. He has extensive experience applying the WRF meteorological model for regional and local-scale air quality modeling applications in the U.S. and internationally. Additionally, Jeremiah specializes in ozone forecasting, exceptional event analysis, evaluation of emission inventories using remote sensing, and wildland fire emissions.

Surendra Kunwar¹, Jared Bowden², George Milly³, Michael Previdi³, Arlene Fiore⁴, J. Jason West¹

¹University of North Carolina at Chapel Hill, Chapel Hill, NC, USA

²North Carolina State Climate Office, Raleigh, NC, USA

³Lamont-Doherty Earth Observatory and Columbia University, Palisades, NY, USA

⁴Massachussetts Institute of Technology, Cambridge, MA, USA

Climate variability can confound the effects of climate change on future US PM_2.5, leading to uncertainty in estimates of climate change induced mortality via PM_2.5 changes. Many past studies have averaged atmospheric model output for fewer than 10 years, which can be insufficient to distinguish the effects of climate change on PM_2.5 from natural variability. Here, we present a novel method to estimate the effects of both climate change and variability on U.S. PM_2.5 under the RCP8.5 climate change scenario at 12km resolution, and estimate associated human mortalities.

In this study, fine scale annual PM_2.5 probability distributions (assumed Gaussian) of two periods of interest - present-day (2006-2020) and midcentury (2051-2065) - are defined by statistics from the combination of a large dataset of global chemistry-climate simulations (GFDL-CM3 model, 2 degree spatial resolution, years 2006-2100, 3 initial condition ensemble members for RCP8.5 climate change), and high resolution downscaling. We downscale eight selected GFDL years representative of medium and high PM_2.5 in different regions of the US using the regional climate model WRF and regional air quality model CMAQ, to 12km spatial resolution. Across the models, anthropogenic PM_2.5 and O₃ precursor emissions are fixed at present day levels, thereby isolating the influence of climate change (and variability) on PM_2.5. Natural feedback emissions of seasalt, biogenic VOCs and lightning NOx are activated in the regional chemistry simulations.

At each 12km gridcell of the regional model US domain, a Monte Carlo simulation (1000 draws) of the difference between present-day and midcentury mean annual PM_2.5 is used to estimate the impacts of climate change on PM_2.5, in light of natural climate variability, in the form of a probability distribution. There is a wide range of PM_2.5 changes due to climate change in the 2050s, with most areas within the US experiencing a mean increase of up to 1.25 μg/m³, throughout major cities of the US. The Southeastern US sees a large increase in mean annual PM_2.5 from boosted OM levels in the future.

We then use BenMAP (Benefits Mapping and Analysis) to assess probability distributions of premature deaths related to PM_2.5 and attributable to climate change. Estimated deaths are represented as probability distributions that reflect the uncertainty in concentration-response functions, and the year-to-year meteorological variability. Due to the natural variability in PM_2.5, there is uncertainty in health impacts at the grid cell level and aggregate (county, state, national) levels. Presenting the health impacts of PM_2.5 from climate change in the form of probability distributions to incorporate natural climate variability is very informative and relevant for policymaking.

Hantao Wang, Marc Serre, Kazuyuki Miyazaki, Juan Cuesta, J. Jason West

Satellite observations have been widely used to monitor the global distribution of tropospheric ozone, but ground-level ozone is difficult to detect. This study proposes a method to infer global ground-level ozone concentrations by combining satellite Lowermost Tropospheric (LMT) observations, a chemical reanalysis, and ozonesonde profiles. We then aim to use this approach to better understand ozone in regions with few ground monitors and include the output in data fusion of global ozone concentrations. We use ozonesonde observations at 43 sites globally to fit a monthly trend of 0-3 km ozone concentrations by modeling time series of ozone concentrations at discrete pressure levels with associated seasonal and interannual variability, with a generalized additive model. Our analysis shows an inverse linear relationship with very high certainty between pressure and ozone concentration over 0–3 km altitude, and a geographic autocorrelation over a 15° range. We then use the Bayesian Maximum Entropy (BME) data fusion method with multispectral satellite LMT observations from IASI+GOME2 as the global offset, and monthly ozonesonde fitting data as observations, to produce monthly estimates of LMT ozone. Next, we generate a global map of the ratio of ground-level to LMT ozone, also using the BME method, where ratios from the Tropospheric Chemistry Reanalysis version 2 (TCR2) are input as the global offset, and ratios from monthly ozonesonde fitting data as the observations. Finally, this BME ratio map is combined with the monthly LMT ozone data output from the BME to produce a global 1° x 1° monthly ground-level ozone map. The resulting monthly ground-level ozone map was evaluated with respect to monthly average DMA8 (daily maximum 8 h average) of ground-level observations at 5228 sites in 2017 from the Tropospheric Ozone Assessment Report (TOAR-II) database. The global map demonstrates a robust R2 of 0.47 and a Root Mean Square Error (RMSE) of 8.77 ppb, significantly surpassing the R2 of 0.21 and RMSE of 10.17 ppb from the original satellite LMT data. When focusing on TOAR-II stations north of 30°N during the summer (June to August), the performance of the derived ozone map is more accurate than that of satellite LMT observations and the chemical reanalysis.

I'm in third year of my PhD. My research primarily focuses on global aerosol mapping and utilizing satellite observations. Also, I am good at modeling using statistical methods.

Fleischer, Daniel; Penaloza Gutierrez, Juan

This presentation explores environmental modeling with an emphasis on Computational Fluid Dynamics (CFD) applications, drawing on case studies from our nesting projects in West Oakland, CA, Dallas, TX, Louisville, KY, and our National Science Foundation (NSF) grant development efforts, which are pivotal to the development of the AdaptOS platform. The primary objective of our research is to enhance urban environmental quality through the implementation of green infrastructure.

Our methodology encompasses extensive monitoring and modeling to establish baseline metrics, followed by strategic design interventions aimed at improving air quality and mitigating heat. The CFD modeling process requires inputs such as a domain model, meteorological data, and traffic data. The domain model, or digital twin, represents the urban environment's structure and may depict existing, historical, or proposed conditions. CFD can estimate the effects of design decisions on air pollution metrics by comparing different digital twins while keeping other inputs constant.

Urban areas in particular, are characterized by complex interactions between various built and natural features, influencing the dispersion of pollutants. Our research employs CFD to analyze pollutant dispersion in urban environments, focusing on micro-scale simulations that capture detailed flow dynamics around buildings, streets, and trees. Understanding these local variations is crucial for mitigating pollution exposure and optimizing urban design. In our studies, we aim to address urban flows at the meter to sub-meter scales, resolving the interactions between airflow and urban features including buildings, trees, and vehicles.

As we well know, trees play a significant role in pollution reduction by acting as physical barriers and facilitating deposition processes. Conversely, vehicle-induced turbulence enhances pollutant dispersion, creating flow patterns necessitating detailed analysis. A key consideration is the dual role of vegetation: by slowing down wind, trees can trap air pollution at ground level, increasing exposure, which might otherwise disperse pollutants vertically into less inhabited atmospheric strata.

We will also examine the implications of this work on urban environmental justice, emphasizing potential enhancements and benefits to the community. Additionally, we will discuss the complexities of collaborating with regulatory agencies and the critical importance of democratizing environmental modeling via software development. This study entails a comparative analysis of various methods, objectives, and outcomes, uncovering diverse results from different modeling approaches and monitoring techniques. The findings highlight the importance of integrating detailed micro-scale simulations in urban planning to balance the benefits and drawbacks of various design features on pollutant dispersion -- as well as considering how our CFD can be utilized in partnership with existing CMAS models.

Juan Penaloza Gutierrez, PhD

CFD Specialist

Hyphae Design Laboratory 456 8th St, Oakland, CA 94607

Email: juan@hyphae.net

Joshua Kumm^1,2, Barron Henderson³, Sergey Napelenok⁴, Heather Simon³, Zhen Qu²

1. ORISE Research Fellow @ US EPA

2. North Carolina State University, Department of Marine, Earth, and Atmospheric Sciences

3. US EPA Office of Air Quality Planning and Standards

4. US EPA Atmospheric & Environmental Systems Modeling Division

This work explores the ability of satellite data assimilation to improve early modelling of the 2023 North American wildfire season. We assimilate Tropospheric Ozone Monitoring Instrument (TROPOMI) NO2 in the CMAQ chemical transport model using 3D variational data assimilation (3DVAR) and assess the changes in CMAQ simulation skill for ozone associated with the fires. Nitrogen oxides (NOx = NO + NO2) and volatile organic compounds (VOC) are emitted from forest fires and can be carried long distances in smoke plumes by atmospheric wind patterns, where they can react photochemically to form secondary pollutants such as tropospheric ozone and fine particulate matter. This process leads to increased concentrations of ozone and fine particulate matter even in remote locations, which was evident in the Eastern US as large wildfires raged in Canada during the summer and early fall of 2023. Proper estimation of NO2 and its byproducts is critical to understanding overall fire air quality impacts and informing policy decision making; however, fire emission inventories for 2023 are still under development due to the labor-intensive and time-consuming nature of fire inventory development. Preliminary inventories with high uncertainty have been incorporated into a rapid modeling effort known as Expedited Modeling of Burn Events Result (EMBER). Here, we attempt to improve the EMBER air quality state by developing and running a framework that uses CMAQ coupled with Gridpoint Statistical Interpolation (GSI) software for assimilating TROPOMI NO2 data. The resulting simulation is then compared to a CMAQ EMBER base simulation (with no assimilation) to analyze changes in NO2 and ozone, and how these changes compare to observations. We focus our analysis on the assimilation impact within the simulated fire plume and at downwind locations. Overall, this work seeks to aid the scientific understanding of wildfire air quality impacts, inform fire emissions inventory corrections, and enable swift regulatory decision making.

Stephanie Bachman, Dr. Nicholas Meskhidze, Jonah Hazelwood, Lintong Cai, Dr. Markus Petters

Ambient aerosol measurements conducted on the rooftop of Jordan Hall at North Carolina State University (NCSU) revealed episodic events where sub-10 nm particle number concentrations reached 373,086 #/cm^3. The particle modal diameter remained constant over time, suggesting these particles were not associated with new particle formation events but likely originated from primary sources. Increases in sub-10 nm particle number concentrations correlated with nighttime conditions and/or overcast skies. To investigate potential sources of these particles, simulations were performed using EPA's Gaussian plume-based model, AERMOD. Through campus drives and analysis of Google Earth images, three possible sources capable of emitting large numbers of sub-10 nm particles were identified in the vicinity of the measurement site: Cates Avenue Steam Plant, Yarbrough Central Utility Plant, and Centennial Campus Central Utility Plan. In AERMOD, different point sources were created for each power plant. Each plant was assigned an emission factor of 500 g/s, and concentrations were modeled at the Jordan Hall location at various heights from ground level to 20 meters above ground. The study explored source contributions and the effects of various meteorological conditions on modeled particle numbers, including wind speed, wind direction, solar radiation, and the Planetary Boundary Layer (PBL) height. This approach allowed for a comprehensive analysis of both the potential sources and the environmental factors influencing the observed sub-10 nm particle concentrations.

Preliminary results show a good correlation between the model-assigned and measured particle number concentrations. Based on wind speed and direction, it was possible to identify plumes originating from different power plants, reaching the Jordan Hall location at varying concentrations. The model also successfully reproduced the observed relationship between particle numbers and the PBL height. The model accurately replicated elevated numbers of particles during periods with shallow PBL heights.

Sub-10 nm-sized particles are very harmful, as they efficiently deposit in the nasopharyngeal, tracheobronchial, and alveolar regions. Within these regions, the particles cross the mucous membrane and reach sensitive organs, including the liver, heart, spleen, brain, lungs, and gastrointestinal tract. Given the proximity of both the NCSU student population and the wider Raleigh area to the power plants, further investigation is warranted. Additional measurements and modeling studies are necessary to more accurately determine the power plants’ contribution to the measured nanoparticle concentrations and analyze their potential impacts, especially within the exhaust plume.

Stephanie Bachman is a master's candidate in atmospheric sciences at North Carolina State University. Her research focuses on aerosols and ultrafine particles and utilizing different modeling approaches for backtracking source identification. Stephanie received bachelor's degrees in physics and sociology from Indiana University of Pennsylvania. She also works at the Environmental Protection Agency in the Office of Chemical Safety and Pollution Prevention on various projects to support the risk assessment process for pesticides with an emphasis on model evaluations, especially dietary exposure models.

Barron H. Henderson, Halil Cakir, Brett Gantt, Ben Wells, Janica Gordon, Phil Dickerson, Hai Zhang, Alqamah Sayeed, Shobha Kondragunta, Yang Liu, Meng Qi, and the rest of the HAQAST AirNow team

This presentation provides an update on the AirFuse system that was introduced last year at CMAS. AirFuse integrates monitors, sensors, models, and satellites to make the best air quality estimates available for AirNow. Last year, we showed results from a historical period and highlighted the strengths and weaknesses of the system. This year will provide evaluation of a real-time pilot application that began in March of 2024, but only uses monitors, sensors, and models. We will also share updates on the more complete integration of GOES-R satellite PM25 data for the next release.

AirFuse has been piloted the EPA running every hour of every day using just the monitors and sensors. As a standard part of AirFuse, it outputs a 10-fold cross validation that allows us to assess the performance. In October, we will have 8 months of live performance data. This is a good test of the data screening and will highlight how moving to AirFuse will increase the fidelity of air quality in AirNow.

We will also share the updates on integration of GOES-R into the AirFuse system. The GOES-PM25 satellite data fusion has also been running every daylight hour of every day. The GOES-PM25 system has been updated to improve pixel quality assurance and averaging techniques on the AOD that improves the weighted regression-based PM25. In addition, the machine learning bias-correction algorithm has been updated to further improve the transition between subdomains. Finally, we’ve been improving the weighting methodologies to better integrate GEOS-PM25 with the monitors and sensors to create a final fused product.

This talk will cover the updates to the methodology, real-time validation results, evaluate the new integration of GOES-PM25, and outline a plan for a next version in AirNow.

Barron Henderson is a Physical Scientist at the United States Environmental Protection Agency. He uses theory, computer simulation, and satellite data to explore scientific and societal issues related to air pollutants. Barron's body of work advances process-level understanding and quantitative constraints within air quality models, and uses those models to quantify integrated impacts of air pollution at local, regional and global scales.

Junhua Zhang, Qiong Zheng, and Ali Katal

Air Quality Research Division, Environment and Climate Change Canada, 4905 Dufferin Street, Toronto, ON, M3H 5T4, Canada

junhua.zhang@ec.gc.ca

Laurent Bruneau Cossette

Air Quality Policy-Issue Response Section, Meteorological Service of Canada, Environment and Climate Change Canada, 2121 Rte Transcanadienne, Montreal, Quebec, H9P 1J3, Canada

On-road motor vehicles are one of the important sources of pollutants that affect air quality. The current Environment and Climate Change Canada (ECCC) Regional Air Quality Deterministic Prediction System (RAQDPS) has been found to overestimate NO2 concentrations for Canada, particularly in large Canadian cities.

This overestimation may be due to the spatial surrogates used for processing model-ready on-road emissions, which allocate excessive emissions to the urban areas. To address this, extensive efforts have been made to improve the on-road spatial surrogates, such as creating different off-network surrogates for different types of vehicles based on building types, e.g., residential building for passenger vehicles, commercial and industrial buildings for heavy duty vehicles, and truck rest stops for heavy duty vehicle hoteling.

Additionally, on-road emissions reported in previous Canada’s Air Pollutant Emission Inventory (APEI) might have been overestimated. Recently in 2023, two major updates were made for APEI for estimating on-road vehicle emissions: 1) corrections to diesel fuel oil consumption determined from the Report on Energy Supply and Demand (RESD) in Canada, and 2) an upgrade of the MOtor Vehicle Emission Simulator (MOVES) model from MOVES2014 to MOVES3 (see, Table A3–4 of https://publications.gc.ca/collections/collection_2023/eccc/En81-30-2021-eng.pdf).These two updates resulted in approximately a 50% reduction in emissions for the major pollutants, such as NOx, PM2.5, and VOC.

The updated spatial surrogates and revised on-road emission inventory will be evaluated using ECCC’s GEM-MACH (GEM–Modeling Air quality and CHemistry) model, which serves as the online one-way-coupled chemical transport model in ECCC’s RAQDPS. This presentation will discuss the impacts of those updates on model performance in predicting NO2, O3, and PM2.5 concentrations over different Canadian regions and large cities.

Dr. Junhua Zhang is a physical scientist working in the Air Quality Modelling and Integration Section, Air Quality Research Division, Environment and Climate Change Canada. One of his main responsibilities is to obtain and process emissions for supporting regional/global air quality modeling.

Jia Xing^1,2, Bok H. Baek¹, Siwei Li³, Chi-Tsan Wang¹, Ge Song³, Siqi Ma¹, Daniel Tong¹, Joshua S. Fu²

¹ Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA 22030, USA

² Department of Civil and Environmental Engineering, the University of Tennessee, Knoxville, TN 37996, USA

³ Hubei Key Laboratory of Quantitative Remote Sensing of Land and Atmosphere, School of Remote Sensing and Information Engineering, Wuhan University, Hubei, 430000, China

Accurate and efficient retrieval of atmospheric chemical concentrations across space and time is crucial for health and climate assessments. However, existing model-measurement fusion methods suffer from limitations due to imbalanced samples from ground measurements or less effective assimilation of satellite data along with numerical modeling. To address these limitations, this study introduces a novel Deep-learning Measurement-Model Fusion method (DeepMMF) constrained by physical and chemical laws inferred from numerical chemical transport models (CTM). This method is applied to NO₂ species over the Continental United States (CONUS) domain for the years 2019 and 2020. By pre-training with abundant CTM simulations, fine-tuning with satellite and ground measurements, and employing a novel optimization strategy for selecting proper prior emissions, the retrieved spatiotemporally continuous surface NO₂ concentrations present consistent values and daily variations with observations (NMB reduced from -0.3 to -0.1 compared to original CTM simulation). Importantly, the corresponding emissions have been simultaneously adjusted, showing good agreement with changes reported in the national emission inventory (NEI) between 2019 and 2020. Interpretation analysis suggests that the DeepMMF model effectively identifies the importance of satellite data at the regional level and ground measurements at the city level, which is scientifically sound. It exhibits consistent prediction of ground measurements while successfully avoiding the sample imbalance problem that leads to overestimation (up to +100%) of downwind/rural concentrations compared to other existing methods. These results demonstrate the great potential of DeepMMF in data assimilation and retrieval studies for other pollutants and regions, to better support health and climate studies.

Maha Shehadeh, Vahid Hosseini

This presentation explores advanced methodologies for improving the accuracy of vehicle emissions inventories through machine learning. Using a case study from an urban area in Canada, we employ traffic camera data to achieve detailed temporal and spatial monitoring of fleet composition at the community level. The study compares observed fleet composition with the assumed homogenous fleet composition derived from insurance records. This comparison is particularly crucial for heavy-duty vehicles (HDVs), which frequently transit through the region without local insurance coverage. This work aims to improve the accuracy of emissions inventory inputs.

Vahid Hosseini is an associate professor of Sustainable Energy Engineering at Simon Fraser University, Vancouver Canada. He has a Ph.D. in mechanical engineering - combustion generated air pollution from the University of Alberta, Canada and 10 years of previous experience as a professor at Sharif University of Technology in Tehran, Iran and head of Tehran's air qulaity monitoring network for 5 years.

Chi-tsan Wang, B.H. Baek, Yunyao Li , Charles Chang , Jia Xing

The oil and gas industry significantly impacts local and regional air quality, particularly in regions experiencing extensive unconventional oil and gas development (UOGD), such as the Marcellus Shale area. From 2008 to 2020, natural gas production in the Marcellus Shale region increased dramatically, growing from approximately 2 billion cubic feet per day to over 32 billion cubic feet per day, marking more than a 15-fold increase. In 2021, gas production in Marcellus Shale counties accounted for approximately 44% of nationwide gas production. Long- term trends of emissions from UOGD, including hazardous air pollutants (HAPs) and criteria air pollutants (CAPs) such as NOx, ozone, and PM2.5, are not well characterized. Therefore, understanding the long-term impact of these emissions is crucial for developing effective regulatory and industry practices to mitigate adverse environmental and public health impacts. To address these long-term air quality change issues, the Comprehensive Air Quality Model with Extensions (CAMx) was applied to simulate four target years (2008, 2014, 2019, and 2021) using the reactive tracer method in a 4km x 4km flexi-nesting domain. The CAMx model results include CAPs (such as ozone, NOx, and PM2.5) and selected HAPs concentrations (including styrene, benzene, toluene, ethylbenzene, xylenes, naphthalene, and butadiene). These four years of CAMx input and output data were then used as training data for the deep-learning enhanced CTM system (DeepCTM). After completing the training process, the DeepCTM model, which mimics the CAMx model processes but saves considerable simulation time, was used to generate 20-year (2002-2021) CAPs and HAPs concentrations. The input data for the DeepCTM model are the same as for the CAMx model, including daily emission, meteorological, and land data. In current preliminary results for HAPs hourly data, the DeepCTM model results closely matched the CAMx output (R-square: 0.72-0.79).

Seongeun Jeong, Eunhye kim, Yoon-Hee Kang, Soontae Kim

In Northeast Asia, PM2.5 concentrations had been rapidly changing due to stringent emission reduction policies and socioeconomic impacts. To accurately capture these changes, continuous efforts to update emission inventories were essential. However, keeping pace with these updates had been challenging. Previous studies recommended adjusting gaseous emissions using ground-based or satellite observations, but they had been limited in addressing primary particulate emissions, which directly influenced PM2.5 concentrations. To enhance the accuracy of emissions used in the air quality model, we implemented an emission adjustment method for both gaseous and particulate emissions using ground observations and simulation results from the Community Multiscale Air Quality (CMAQ) model. Considering the source-receptor relationship on a regional scale, we recognized that PM2.5 concentrations are transported to transboundary regions. Initially, we adjusted emissions in China and South Korea sequentially, given that South Korea is located downwind of China. To validate the simulation results using these adjusted emissions, we compared simulated concentrations with satellite and aircraft observations from the ASIA-AQ project. This comparison confirmed the accuracy and reliability of the adjusted emissions inventories.

Hossein Alizadeh, Ali Azimi, Hamid Delbari, Vahid Hosseini

Accurate meteorological fields are crucial for effective air quality modeling due to their known impact on the dispersion, reaction, and deposition of atmospheric pollutants. This study uses a nudging method to enhance the accuracy of the Weather Research and Forecasting (WRF) model's outputs and demonstrates their impact on the Community Multiscale Air Quality (CMAQ) model outputs. A deep learning architecture is developed to downscale low-resolution (108 km2) ambient data of the National Centers for Environmental Prediction (NCEP) to a high-resolution (1 km2) grid using an encoder-decoder architecture with skip connections. The model employs bilinear interpolation to produce a preliminary high-resolution estimate, which is then refined by predicting residuals, resulting in more accurate and detailed outputs. An attention mechanism within the model captures complex patterns and variations, particularly in high-frequency variables like wind speed. The resulting high-resolution (1 km2) data is used to nudge the WRF model's state across the domain. This approach helps to rectify any discrepancies in the WRF model’s predictions, leading to more accurate representation of atmospheric conditions. The enhanced meteorological fields are utilized for emissions processing and air quality simulations using the CMAQ. The findings highlight that nudging the WRF results with high-resolution data significantly improves the model's accuracy. Furthermore, the use of these more precise meteorological data in CMAQ enhances the model's capability to predict pollutant concentrations. This study underscores the importance of high-resolution data assimilation using deep learning in improving the reliability of air quality predictions.

Vahid Hosseini is an associate professor of Sustainable Energy Engineering at Simon Fraser University, Vancouver Canada. He has a Ph.D. in mechanical engineering - combustion generated air pollution from the University of Alberta, Canada and 10 years of previous experience as a professor at Sharif University of Technology in Tehran, Iran and head of Tehran's air qulaity monitoring network for 5 years.

Daiwen Kang, Yuling Wu, Arastoo Pour-Biazar, Rohit Mathur, David Wong, and Wyat Appel

Lightning produced NOx emissions (LNOx) provide the most abundant NOx source in the mid-to-upper troposphere, a critical precursor for ozone and hydroxyl radical (OH), the primary tropospheric oxidants. Accurately quantifying LNOx emissions in time and space in chemistry transport models is challenging due to various uncertainties associated with lightning data and LNOx production rates. Currently, LNOx yield is mainly based on lightning flashes either detected by lightning detection networks or parameterized by lightning parameterization schemes based on modeled convective parameters (such as cloud-top height, cloud ice fluxes, or convective available potential energy). However, the definition of lightning flash varies from one network to another, confounding the use of datasets from different networks. In addition, it has long been recognized that LNOx yield is related to lightning energy levels, which varies across lightning types (e.g., cloud-to-ground, cloud-to-cloud) and exhibits large geographical variations. Along with flash and stroke counts, most networks also report information on energy associated with lightning strikes. Even though the registered energy values are only a tiny fraction of the sampled lightning strikes (either electromagnetic or optical) by different detection techniques, they represent the same lightning activities (if the networks are similarly purposed), thus some relationship should exist among them. Exploratory analysis of the lightning datasets from the World Wide Lightning Location Network (WWLLN) and the Geostationary Lightning Mapper (GLM) aboard the GOES-R satellites have indicated that even though both networks detect total lightning, the WWLLN network is far better at detecting cloud-to-ground lightning strikes, while the GLM technique is good at detecting total lightning. The energy aspects detected by the two networks display strong relationship when these energy values are aggregated over time and space, suggesting that synergies between the two datasets could be exploited to improve LNOx emissions estimate across temporal and spatial scales broader than those covered individually. An energy-based LNOx model will be presented and assessed for LNOx emissions estimate using a modeling domain covering the contiguous United States and surrounding areas discretized with a 12-km resolution grid.

Jonathan Brunton, Christian Hogrefe, K. Wyat Appel, Robert Gilliam, Fahim Sidi, Kristen Foley

Meteorology and air quality modeling data from the EPA’s Air QUAlity TimE Series Project (EQUATES), spanning 2002-2019, was used for diagnostic evaluation of bias in modeled summertime ozone. Using paired modeled-observed data from the Weather Research and Forecasting (WRF) model and paired modeled-observed MDA8 O3 from the Community Multiscale Air Quality (CMAQ) model for the contiguous U.S., we conduct an exploratory statistical analysis of the impact of model bias in temperature, mixing ratio, windspeed, and precipitation on top ten modeled MDA8 O3 days for each year in the time series. To study this relationship further, we developed a machine learning workflow using Random Forest (RF) regression to predict MDA8 O3 bias and its trends based on meteorological and spatiotemporal variables. The spatiotemporal variables (year, month, latitude, longitude, elevation) are used to represent drivers of ozone bias not explicitly included in the statistical model, e.g., seasonal and regional patterns in emissions biases. An individual RF Regressor model was trained for each NOAA climate region in the EQUATES CMAQ domain, to optimize the model for regional characteristics. The RF models allow for exploration of complex relationships between the ozone bias and multiple model variables, or features. A Feature Importance analysis identifies regions where bias in meteorological variables had a comparable impact on MDA8 O3 bias as the non-meteorological features. Partial Dependence analysis shows whether the relationship between the ozone bias and the meteorological, spatial, and temporal features are linear, monotonic, or more complex. This workflow and RF model allow for future studies of the same nature involving additional CMAQ model inputs, such as emissions or additional meteorological data.

Laurent Bruneau-Cossette, David Niemi, Maxim Bulat, Mourad Sassi

The TREND and PROJ inventories are AQM-ready inventories of respectively past and future Canadian pollutant emissions. They are created and used by REQA for regulatory work and provided to various domestic and foreign governments and regulatory agencies. Recently, the work of updating, automating, and porting the existing methodology into a PostgreSQL + PostGIS database was undertaken. The resulting implementation greatly improves the quality control capabilities, scalability and production efficiency of TREND+PROJ inventories. TREND+PROJ21 is the first available inventory produced by this implementation. This increase in productivity also provides opportunities for further improvement.

Nana Wu, Zhen Qu, Amir Ghajari

Bottom-up emission inventories face uncertainties at fine temporal scales (e.g., daily, hourly), posing challenges in chemical transport modeling. The newly launched TEMPO instrument has unprecedented hourly resolution over the US and therefore offers the capability to enhance our understanding of the temporal variations in NOx emissions from space. Here, we set up a pseudo-observation experiment with known “true” emissions to evaluate how the finite difference mass balance (FDMB) inversion methods and TEMPO NO2 retrievals improve day-to-day NOx emissions. TEMPO-constrained NOx emissions are then used for the weekend effect quantification of air pollution in different regions. We find that FDMB methods have the potential to recover day-to-day emissions at 12 km resolution grids. Integrating highly resolved TEMPO observations, the daily NOx emissions in the third week of December 2023 increased by an average of 27%, leading to a 33% rise in NO2 concentration. Compared to the prior emissions, the posterior emissions result in a 2.0 ug/m3 decline in the mean bias (MB) of simulated daily NO2 concentrations across the US against ground observations, a 1.5 ug/m3 reduction in the root mean square error (RMSE), and a 13.2% decrease in the normalized mean error (NME) by 13.2%. We will incorporate satellite measurements of additional species (e.g., formaldehyde) to further improve chemical transport modeling and support policy formulation.

Siqi Ma, Daniel Tong, Fei Liu, Jia Xing, Bok H. Baek, Chi-tsan Wang, Yunyao Li

Emissions of air pollutants and greenhouse gases are the main driver of many Earth system processes behind some of the grandest environmental challenges today, such as air pollution and global warming. Creating emission inventories (NEI) typically lags by a few years, so that these datasets may not reflect the current emission status, especially during the periods of Economic Recession or COVID-19. Updating emissions from the base year to the specific forecast year using observational datasets, a process known as emission rapid refresh, is an effective approach to reduce the uncertainties of outdated emission inventories to increase forecasting accuracy. In this study, we applied four refresh methods to update the base year emission: state-level emission adjustment (ADJ), ADJ with budget-aware urban adjustment (BA), ADJ with TROPOMI city emissions (SatNOx), and adjustment with daily Variational AutoEncorder (VAE) physically-informed inverted emissions. Then we conducted summertime (July 2020) sensitivity experiments in a 12-km CMAQ modeling system using these three methods, alongside two no-adjustment runs—simulations with previous year (2019, control) and current year (2020, N2020) emissions—to evaluate the effectiveness of each approach. The results show that O3 generally decreases, with increases observed in the suburban and rural areas of New Mexico, southern Texas, and western New York in the adjusted runs. NO2 shows a consistent change pattern with NOx emissions. The highest O3 changes occur in suburban areas, while NO2 changes are most pronounced in urban areas. Compared to the control run, the simulations with adjustments generally reduce bias by 0.4–1.4 ppb in the eastern and southern US, while increasing by 0.2–0.8 ppb in the western US. The adjusted runs perform comparably to N2020 in the O3 simulation and better reproduce surface NO2 due to a more significant NOx emission reduction in N2020. However, N2020 outperforms others in NO2 vertical concentration density (VCD) simulation. Our findings indicate that emission refresh methods are capable of improving the O3 predictability, achieving performance comparable to simulations using current-year emissions.

Douglas Lapham

Artificial Intelligence (AI) is more than a single technology breakthrough. It's a complex integration of processes and algorithms that deliver transformational effects to organizations' missions. Now is the time to harness and integrate this incredible technology as it transforms our understanding of the environment at multiple scales.

This session provides a high-level overview of AI and how it can integrate with, and enhance, current detection and risk prevention technologies. Potential use cases include how AI can be used to efficiently run and deploy environmental models, and the utilization of AI for understanding complex natural processes needed to formulate recommendations on new emerging air and water related risks. Emerging technologies, such as quantum sensors, may both accelerate and supplement the value added by AI, unlocking new sensor capabilities and analytic insights relevant to EPA's mission. Additionally, the session will discuss AI’s role in supporting model development through applications, such as cyber security, policy auditing, software and data quality assessment, data policy enforcement, and data model usage compliance, and make it useful for operations and regulatory purposes.

For example, we will explore the use of machine learning (ML) to ingest, extract, and normalize data from Coupled Model Intercomparison Project (CMIP6) and the United States Geological Survey (USGS) REST water model. The platform follows repeatable patterns that build scalable, repeatable data pipelines to process climate data. This speeds redeployment and integration of additional services, ready for enterprise expansion as new Climate Intelligence use cases arise.

Finally, the presentation will survey AI integration in other Federal Agencies also focused on climate science and modeling, including USGS, National Aeronautics and Space Administration (NASA), National Oceanic and Atmospheric Administration (NOAA), and the Department of Defense (DoD). For example, in support of a recently awarded Broad Agency Announcement (BAA) at NOAA we are exploring the feasibility and capability of emerging technologies focusing on data processing and dissemination of weather data. We are leveraging an AIOps framework and automated data pipelines to ingest NOAA data, process simultaneous AI/ ML algorithms at scale, and implementing a foundation for modernized, efficient research to operations workflows. Understanding and implementing the AI of today is key to unlocking the effective climate modeling of tomorrow.

Marie Chen Nguyen is a Senior Associate at Booz Allen Hamilton, leading technical delivery and strategy on Artificial Intelligence/Data Science applications in the climate domain.  The models and applications developed by her team have been used to integrate long term climate trends into resource planning and risk management. Marie has a Master of Science in environmental engineering from Johns Hopkins University and a Bachelor's degree in Biochemistry and Molecular Biology from Boston University.

Zhen Qu, Nana Wu

Accurate quantification of air pollutant emissions from each human activity is critical for the planning and verification of emission reduction efforts. Top-down estimates with satellite data provide important information on the sources of air pollutants. We apply a newly developed sector-based inversion method based on 4D-Var and mass balance to quantify NOx, SO2, and VOC emissions from various activities, including transportation, industry, residential, aviation, shipping, energy, and biomass burning over the US. We incorporate TEMPO NO2 and HCHO, TROPOMI SO2, and MOPITT CO observations and leverage the co-emission of these gases to identify the source sectors. The framework improves emission estimates at the process level by optimizing emission factors and activity rates without relying on explicit knowledge of their values and resolves discrepancies with bottom-up inventories at the sector level. Previous applications of this approach show reduced normalized mean bias (NMB) by 7% (NO2)–15% (SO2) and normalized mean square error (NMSE) by 8% (SO2)–9% (NO2) compared to a species-based inversion when evaluated with surface measurements. We will evaluate the sectoral profile and temporal variations of our posterior emissions with bottom-up inventories (e.g., EPA NEI and CEDS) and EPA AQS measurements.

Dr. Zhen Qu is an assistant professor in the Department of Marine, Earth and Atmospheric Sciences at North Carolina State University. She received her Ph.D. in Mechanical Engineering from the University of Colorado Boulder in 2019, B.S. in Physics from Peking University in 2014, and postdoctoral training at Harvard University. Her research interests include air quality, greenhouse gases, remote sensing, data assimilation, and inverse modeling. Her research aims to address the current challenges in air pollution and climate change by (1) exploiting observations to elucidate air pollutant and greenhouse gas emissions, concentrations, and chemistry; and (2) developing science-based strategies to improve air quality and abate climate forcing.

V. Rao1, K. Seltzer1, M. Strum1, Y. Dietrich1, R. Mason1, L. Dayton1, A. Eyth1, J. Vukovich1, A. Diem1, J. Bash2, A. Holder2, I. George2, H. Pye2, J. Beidler2, G. Pouliot2, 1 US EPA, OAQPS, EIAG 2 US EPA, ORD, CEMM

The National Emissions Inventory (NEI) program within the EPA Office of Air Quality Planning and Standards continually works to update science related to emissions inventories. Efforts includes developing emission estimation methods for previously uncharacterized sources, upgrading methods for existing sources (including improved activity and emission factor data) and other ancillary data (e.g., modeling-related emissions transformations such as chemical speciation, temporalization, spatial allocation, and emission projections), and improving emission trends information. In this talk, we provide an overview of the new methods introduced for the 2020 NEI, planned updates for the 2023 NEI, and discuss the drivers of priorities in this arena as we move beyond the 2023 NEI. In doing this, we will summarize some of the impacts of past methods improvements on emissions, including added pollutant coverage, improved/added emissions source coverage, and ancillary data improvements for modeling. Specifically, this presentation will include emissions methods improvements related to: fires, solvents/asphalts, sources of NH3, residential wood combustion, VOC and PM2.5 speciation, and spatial/temporal upgrades.

Tesh has worked at the US EPA for over 32 years.  Tesh has worked for both the office of tranportation and air quality as well as the office of air quality planning and standards, both in EPA's OAR Office.  Tesh's work has mainly involved working on emission inventories and associated atmospheric chemistry sectors.  Current work that Tesh is engaged in at EPA/OAQPS include emissions methods development and working on specific sectors related to ammonia, fires, and emerging new emission sources.

Lexie Wilson, Rachel Edie Ph.D., Mark Sghiatti

Understanding the impact of biogenic emissions on air quality modeling is crucial for effective environmental management in urban areas in the intermountain west. This study investigates the sensitivity of our photochemical model (Comprehensive Air Quality Model with extensions, CAMx) to different biogenic emissions inputs in Utah’s Northern Wasatch Front ozone nonattainment area. Building on previous work by the Utah Division of Air Quality (UDAQ) to improve model performance, we evaluate Biogenic Volatile Organic Compound (BVOC) emissions from Biogenic Emissions Inventory System version 4 (BEIS 4) against the Model of Emissions of Gases and Aerosols from Nature (MEGAN 3.2). MEGAN inputs were tailored to our region using satellite observations and local tree surveys.

Our analysis focuses on comparing modeled ozone and BVOC concentrations from CAMx with MEGAN and BEIS inputs, and to available observations in the nonattainment area. Observational analysis will be both qualitative and quantitative due to PAMS expansion after our 2017 modeling year.

Lexie Wilson is an air quality modeler, researcher, and lead GIS analyst at the Utah Division of Air Quality. Lexie's current work is focused on improving emissions modeling for Utah's ozone SIP using SMOKE.

Karl Seltzer, Zac Adelman, Michael Aldridge, Christine Allen, James Beidler, Andrew Bollman, Art Diem, Yijia Dietrich, Lindsay Dayton, Alison Eyth, Janice Godfrey, Mark Janssen, Byeong Kim, Tom Moore, Rhonda Payne, Matt Roark, Sarah Roberts, Tom Richardson, Kevin Talgo, Mary Uhl, Jeff Vukovich

The 2022v1 emissions modeling platform (EMP) is being developed by the National Emissions Collaborative, a constellation of staff at regional, state, local, and tribal air agencies, and the U.S. EPA. The platform will support both regulatory modeling and air quality related studies. As with the 2016 Emissions Inventory Collaborative, transparency is fundamental to the development and overall confidence in the platform.

Emissions for the year 2022 were developed using state / local / and tribal submitted data for 2022, along with EPA-developed data based on the 2020 National Emissions Inventory and other data sources. In addition, ancillary data for the platform were reviewed and some improvements were implemented. Here, the methods and data sources used to develop the 2022 air quality model-ready emissions for the 2022v1 EMP will be discussed. We will also illustrate the magnitude of emissions from various sectors, discuss ancillary data updates made, and describe how updates were made to ensure the transparency of this development process. The development and values of the analytic year emission for 2026, 2032, and 2038 will be discussed in a separate talk.

Karl Seltzer is a Physical Scientist in the U.S. EPA’s Emissions Inventory and Analysis Group within the Office of Air Quality Planning and Standards. He contributes to the development of the tri-annual National Emissions Inventory, the Agency’s photochemical emissions modeling platforms, is the speciation lead for the Agency’s emissions group, and helps develop the chemistry used in the Community Multiscale Air Quality modeling system.

Jeongran Yun¹, Winston Hao¹, Yuhong (Ruby) Tian¹, Eric Zalewsky¹, Jin-Sheng Lin², Doris McLeod², Susan McCusker³, and Kevin Civerolo¹

1 New York State Department of Environmental Conservation

2 Virginia Department of Environmental Quality

3 Mid Atlantic Regional Air Management Association

In this study, we considered two sets of projected emissions for Electricity Generating Unit (EGU) point sources: one based on the Integrated Planning Model (IPM), and another based on the Eastern Regional Technical Advisory Committee (ERTAC) EGU model. Air quality modeling based on the two EGU projection approaches has been used to guide air quality planners in control strategy development and policy decision-making. EPA uses emissions from the IPM model, while member states of the Ozone Transport Commission use emissions from the ERTAC model for regulatory modeling.

EPA performed CAMx ozone source apportionment modeling to determine the upwind contributions of total anthropogenic emissions from each state to projected ozone design values at downwind monitor sites for the Good Neighbor Plan to address attainment of the 2015 ozone National Ambient Air Quality Standard. In this study, we obtained EPA’s 2016 base year and 2026 analytical year version 3 modeling platform used for the Good Neighbor Plan, replaced the IPM-based EGU and non-EGU point sources with the ERTAC-based EGU and non-EGU point sources, and performed CAMx ozone source apportionment modeling. We discuss the spatial differences in our comparisons of the CAMx modeling results using the two projected EGU/non-EGU point source emissions and the impacts of using ERTAC-based emissions in place of IPM-based emissions on upwind ozone contributions by states to downwind monitors.

Z. Adelman, M. Aldridge, C. Allen, A. Bollman, M. Collins, Y. Dietrich, L. Dayton, A. Eyth, J. Godfrey, M. Janssen, S. Kayin, B. Kim, S. McCusker, T. Moore, E. Murray, R. Payne, M. Roark, S. Roberts, T. Richardson, K. Seltzer, K. Talgo, M. Uhl, J. Vukovich

The 2022v1 emissions modeling platform includes emissions for the base year of 2022 and for the analytic years of 2026, 2032, and 2038. These future years were selected for inclusion in this platform due to the near-term importance of regulatory modeling for ozone, particulate matter, and regional haze. In a separate talk, development of the year-2022 inventory was discussed. Here, development of the analytic year emissions is discussed. We cover the anticipated changes in activity data and emissions from anthropogenic sources from 2022 through 2038, the methods used to compute analytic year emissions, including proper reflection of applicable control programs and closures of facilities and units, when models were run to reflect changes between the base (i.e., 2022) and analytic years, and when emissions were held constant. It is important to note that the analytic year air quality modeling for these types of analyses uses the same 2022 meteorological data as the base year air quality modeling.

Zac Adelman has been the Executive Director of LADCO since September 2017. He works at the interface of air quality modeling, ambient monitoring, and air quality planning. Before joining LADCO, he worked for 15 years as an air quality researcher and project manager at the University of North Carolina.

Zac holds degrees from the UNC Dept of Environmental Sciences and Engineering. He is an atmospheric scientist with expertise in air pollution modeling and ozone chemistry.

Cam Phelan, Abi Lawal, William Vizuete, Jacob Boomsma, Heather Holmes, Cesunica Ivey

The 2024 update to the annual PM_2.5 standard will require both state and local entities to submit updated SIPs and potentially perform demonstration modeling for proposed emission controls. Many of the areas projected to be out of attainment for this new standard are in the western U.S., where CMAQ traditionally does not perform well. Previous work has shown that the choice of chemical mechanism significantly impacts organic carbon, nitrate, and ammonium PM_2.5 concentrations, especially during wintertime temperature inversions. This creates an issue for regulatory modeling applications, where there is no official guidance on chemical mechanism choice, and we aim to characterize how and why different chemical mechanisms in CMAQ 5.3.3 and 5.4 impact simulated PM_2.5.  We use inline Integrated Reaction Rate (IRR) and Integrated Process Rate (IPR) in monthlong simulations to identify science processes and reactions of interest that can explain the discrepancies between mechanisms and explore the impact of meteorology inputs on these processes and model performance.