NOAAs National Air Quality Forecast Capability operational and experimental updates

Dorothy Koch (1), Ivanka Stajner (2), Jeff McQueen (2), Pius Lee (3), Jianping Huang (2,5), Ho-Chun Huang (2,5), Li Pan (2,5), Youhua Tang (3,6) Daniel Tong (3,6), Patrick Campbell (3,6), Ariel Stein (3), James Wilczak (4), Irina Djalalova (4,8), Dave Allured (4,8), Phil Dickerson (7), Jose Tirado (1,9)

(1) NOAA NWS/STI (2) NOAA NWS/NCEP (3) NOAA ARL (4) NOAA ESRL (5) IMSG (6) CICS, University of Maryland (7) EPA (8) CIRES, University of Colorado (9) Eastern Research Group

This presentation will provide an overview of recent updates to NOAA's operational air quality predictions and experimental updates in the works for future possible operational implementation. The NOAA National Air Quality Forecast Capability (NAQFC) provides predictions for ozone, fine particulate matter (PM2.5) and wildfire smoke over the United States (U.S.); and predictions of airborne dust over the contiguous 48 states. Ozone, smoke and dust predictions are available at airquality.weather.gov. They are also available, together with PM2.5 predictions, through a web service at idpgis.ncep.noaa.gov/arcgis/rest/services/NWS_Forecasts_Guidance_Warnings. Ozone and PM2.5 predictions are produced operationally using a system linking the Community Multiscale Air Quality model (CMAQ) with meteorological inputs from the North American Mesoscale Forecast System (NAM). In addition, smoke and dust predictions are separately produced by NOAA's HYSPLIT model. In our most recent operational upgrade, in December 2018, we updated the air quality model to add a post-processing scheme for ozone predictions based on the Kalman Filter Analog (KFAN) bias correction and improved the bias correction scheme for PM2.5. Both systems are now unified and use consistent training data sets and additional monitor sites. The focus for our next operational implementation is the coupling of CMAQ with the new NOAA's operational Global Forecast System version 15.1 that includes the Finite Volume Cubed-Sphere dynamical core. We are also working to extend the range of CMAQ predictions from 48 to 72 hours, including the KFAN/bias corrected products. Development is underway for a new probabilistic forecast product for ozone in collaboration with our partners at NOAA's Earth System Research Laboratory. Considerable attention is also being given to updates of wildfire smoke impacts. Emissions estimates for CMAQ based on NESDIS Blended Global Biomass Burning Emissions Product (GBBPx) are currently being tested and are the focus of extensive summer experiments. We will present impacts of these recent and planned updates and discuss future plans.

Implementation of new satellite-based source maps in the FENGSHA dust module and initial application with the CMAQ-based NAQFC system

Daniel Tong, Barry Baker, Kerstin Schepanski, Shobha Kondragunta, Pubu Ciren, Benjamin Murphy, Youhua Tang, Pius Lee, Patrick Campbell and Rick Saylor

Dust storms impose a variety of imminent risks on human health (e.g., infectious diseases), transportation safety (highway pileups and roadblocks), and economic impacts (topsoil loss, deposition on solar farms, etc.). Dust forecasting/early warning systems are one of key tools used to mitigate the risks for the impacted communities. A common challenge to all dust models is to accurately represent dust sources at different spatial and temporal scales. Here we present the development of two satellite-based global dust source maps within the FENGSHA dust module, and apply these maps to simulate one of the largest dust storms over North America. On April 10, 2019, a rare large dust storm, originated from northern Chihuahua Desert, swept over the central US, carrying soil particles all the way to Minnesota, an extent that was seen only during the "Dust Bowl" era. The CMAQ based National Air Quality Forecast Capability (NAQFC) system was unable to predict this event by both timing and magnitude. Two satellite source maps, one representing sediment supply hotspots using Moderate Resolution Imaging Spectroradiometer (MODIS) surface reflectance, and the other representing dust sources using an inverse of MODIS black-sky albedo, were developed and implemented in FENGSHA, which is coupled with CMAQ to reproduce this enormous dust event. CMAQ simulations using both new source maps were compared with that using the current static source map, and against Visible and Infrared Imaging Radiometer Suite (VIIRS) aerosol optical depth (AOD), GOES-16 Advanced Baseline Imager (ABI) dust mask product, and ground monitoring of PM2.5 and PM10.

Comparing Projected and Modeled Health Benefits of Alternative Energy Futures

Kristen E Brown

Emissions from energy production, conversion, and use are significant contributors to problematic air quality. It is possible to reduce emissions from the energy system. Calculating the health benefits of potential emissions reductions is often an important motivator to spur action. We calculate the benefits of emissions reductions in two different ways and compare the resulting values. An energy system model was used to determine how damage-based fees might impact technology choice and resulting emissions from the energy system. The health benefits of the emissions reduction should be equal to the change in emissions multiplied by the damage estimates used. Alternatively, air quality and health benefit models are used to quantify the effect of the resultant emission changes. If the damages were linearly related to the emissions, the benefit of the policy would be calculated the same using the initial damage estimates and the more model intensive procedure. The sources of difference between the calculation methods are examined. Major sources of difference include changing population, which can increase the value by 2-3% per year, and regional variations within the national scale analysis. An additional source of uncertainty comes from upstream emissions. Life cycle analyses usually provide only total emissions values without location information. A new treatment is needed for these emissions, because they can be significant, but cannot be modeled in the same way that combustion emissions are.

Exploring the Vertical distribution of Wildland Fire Smoke in CMAQ

Joseph Wilkins, George Pouliot, Thomas Pierce, Jeffrey Vukovich, Kirk Baker, James Beidler

The area burned by wildland fires (prescribed and wild) across the contiguous United States has expanded by nearly 50% over the past 20 years, now averaging 5 million ha per year. Chemical transport models are used by environmental decision makers to both examine the impact of air pollution on human health and to devise strategies for reducing or mitigating exposure of humans to harmful levels of air pollution. Since wildfires are increasing in size and burning more intensely, the exposure of humans to fine particulate matter (PM2.5) and ozone (O3) is projected to grow. Currently, there is little consensus on fire pollution vertical transport methods. The height to which a biomass burning plume is injected into the atmosphere, or plume rise, is not only difficult to qualitatively determine but also comes with quantitative difficulties due to poor understanding of physical constraints within models. Many air quality models rely on plume rise algorithms to determine vertical allocation of emissions using various input models or in-line plume height calculations to determine plume height vertical structures and invoke transport of emissions. In this work, we test basic plume rise methods currently being used in chemical transport modeling in order to determine where the Community Multiscale Air Quality (CMAQ) modeling system's current capabilities can be improved. We investigate proposed improvements for allocating the vertical distribution of smoke by separately characterizing the impacts of model grid resolution, emissions temporal profile, and plume rise algorithm.

Air Quality and Human Health Impacts of Prescribed Fire: Links to PM2.5 and Asthma in Georgia, USA

Ran Huang, Yongtao Hu, Armistead Russell, James Mulholland, and Talat Odman

Short-term exposure to fire smoke, especially PM_2.5, has been associated with adverse health effects. To quantify the impact of prescribed fire on human health, exposure fields of PM_2.5 from prescribed fires in Georgia, USA during the 2015-2018 burn seasons were generated combining air quality/sensitivity simulations with observations through a data fusion method. Then, a general health impact function was used with those exposure fields to estimate the health burden of prescribed fire. The sparsity of observation sites often leads to fire impacts on air quality remaining undetected. While those impacts can be estimated by air quality model simulations various uncertainties in the models lead to inaccuracies. Application of a data fusion method to adjust modeled fire impacts with observations generally improved the exposure fields. However, in some cases, limited observational data reduced the impact of smoke plumes successfully captured in model simulations. The dearth of monitoring sites can be alleviated, in part, by using low-cost sensors. The data withholding evaluation using only sensors data shows that low-cost sensors could be used to provide spatial and temporal information missed by both regulatory monitoring sites and model simulations. A method has been developed to identify the days and areas when and where prescribed burning has a major impact on local air quality, which can be used in exploring the relationship between prescribed burning and acute health effects. The results show a strong spatial and temporal variation of prescribed burning impacts. April 2018 had larger estimated daily health impact with more burned areas compared to Aprils in previous years, likely due to an extended burn season resulting from the need to burn more areas in Georgia. About 145 emergency room (ER) visits in Georgia were estimated for asthma due to prescribed burning impacts in 2015 during the burn season. This number increased by about 18% in 2018, compared to 2015. Although southwestern, central, and east-central Georgia have large fire impacts on air quality, the absolute number of estimated ER asthma visits resulting from burn impacts is small in those regions compared to metropolitan areas where population density is higher. Metro-Atlanta has the largest estimated prescribed burn-related asthma ER visits in Georgia with an average of about 66 during the reporting years.

Impact of renoxification on air quality over Northern Hemisphere

Golam Sarwar, Daiwen Kang, Wyat Appel, Christian Hogrefe, Rohit Mathur

National Exposure Research Laboratory, Environmental Protection Agency, RTP, NC, USA

Barron Henderson

Office of Air Quality Planning & Standards, Environmental Protection Agency, RTP, NC, USA

Recent field and laboratory experiments suggest that aerosol nitrate can undergo photolysis to generate nitrous acid and nitrogen dioxide which can influence the formation of secondary pollutants. In this study, we examine the effect of aerosol nitrate photolysis on air quality using the hemispheric Community Multiscale Air Quality (CMAQ) model. Consistent with other air quality models, CMAQ currently does not contain the photolysis of aerosol nitrate. We add the photolysis of aerosol nitrate to CMAQ using the recently published rate expression and perform three simulations over the Northern Hemisphere for 2016: (1) without the photolysis of aerosol nitrate (2) with the photolysis of aerosol nitrate occurring only over the marine environment (3) with the photolysis of aerosol nitrate occurring over all environments. The photolysis of aerosol nitrate decreases mean aerosol nitrate and enhances nitrous acid and nitrogen dioxide over the Northern Hemisphere. Preliminary results show that additional nitrous acid and nitrogen dioxide indirectly enhance mean aerosol sulfate, secondary organic aerosols, formaldehyde, and ozone over the Northern Hemisphere. Enhancements of long-lived species, like ozone, are not restricted to where the photolysis occurs, so implementing aerosol nitrate photolysis over the marine environment increases ozone over both the marine environment and the land. Enabling aerosol nitrate photolysis over all environments further increases ozone. Within the United States, the effect on ozone is larger over the West compared to the East. A comparison of model predictions with observed ozone from routine surface monitoring networks in the United States and Japan suggests that adding this reaction improves model performance for cooler months but deteriorates the performance for warmer months. The presentation will include a comparison of model predictions with vertical ozonesonde profiles from the World Ozone and Ultraviolet Radiation Data Centre and other available observations.

DISCLAIMER

The views expressed in this paper are those of the authors and do not necessarily represent the views or policies of the U.S. EPA.

Health and Economic Impacts of Air Pollution Induced by Weather Extremes over the Continental U.S.

Yang Zhang, Peilin Yang, Yang Gao, Ruby L. Leung, and Michelle L. Bell

Extreme weather events may enhance ozone (O3) and fine particulate matter (PM2.5) pollution, causing additional adverse health effects. This work aims to evaluate the health and associated economic impacts of changes in air quality induced by heat wave, stagnation, and compound extremes under the Representative Concentration Pathways (RCP) 4.5 and 8.5 climate scenarios. The Environmental Benefits Mapping and Analysis Program-Community Edition is applied to estimate such impacts causes by the changes in surface O3 and PM2.5 levels due to heat wave, stagnation, and compound extremes over the continental U.S. during current (i.e., 2001-2010) and future (i.e., 2046-2055) decades under the two RCP scenarios. Under the current and future decades, the BENMAP results show that the weather extremes-induced concentration increases may lead to several tens to hundreds all-deaths annually for O3 and several hundreds to over ten thousands for PM2.5. High mortalities and morbidities are estimated for populated urban areas with strong spatial heterogeneities. The estimated annual costs for these O3 and PM2.5 outcomes are $5.5-12.5 and $48.6-140.7 billion U.S. dollar for mortalities, and $9-48.6 and $20-113 million for morbidities, respectively. Of the extreme events, the estimated O3- and PM2.5-related mortality and morbidity attributed to stagnation are the highest, followed by heat wave or compound extremes. Large increases in heat wave and compound extremes events in the future decade dominate changes in mortality during these two extreme events whereas population growth dominates changes in mortality during stagnation that is projected to occur less frequently. Projected reductions of anthropogenic emissions under both RCP scenarios compensate for the increased mortality due to increased occurrence for heat wave and compound extremes in the future. These results suggest a need to further reduce air pollutant emissions during weather extremes to minimize the adverse impacts of weather extremes on air quality and human health.

Predicting the phase state and phase separation of atmospherically relevant aerosols and its impact on multiphase chemistry in a regional-scale atmospheric model (CMAQ)

Quazi Z. Rasool¹, Sara Farrell¹, Yue Zhang^1,4, Ryan Schmedding¹, Havala Pye², Haofei Zhang³, Yuzhi Chen¹, Jason D. Surratt¹, William Vizuete¹

¹ Department of Environmental Science and Engineering, Gillings School of Public Health, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599 United States ²National Exposure Research Laboratory, Office of Research and Development, Environmental Protection Agency, Research Triangle Park, Durham, North Carolina, 27709 ³Department of Chemistry, University of California at Riverside, Riverside, California, 92521

⁴ Aerodyne Research, Inc., Billeria, MA, 01821

Multiphase chemistry of isoprene-derived epoxydiols (IEPOX) is one of the major sources of tropospheric secondary organic aerosol (SOA). Our modeling work will first implement new laboratory findings that demonstrate pre-existing SOA coatings on acidified sulfate seed aerosol impede acid-catalyzed multiphase chemical reactions of IEPOX leading to new SOA. Accurate prediction of the phase state and mixing state of mixed inorganic-SOA particles is critical to quantify the impact on climate and air quality. The phase state of SOA can change more than 14 orders of magnitude from 10^-2 Pa s to 10¹² Pa s in the atmosphere, significantly impacting the multiphase reactions between the particles and gas-phase species. The mixing state of the aerosols, such as phase separation, can create a highly viscous organic shell that surrounds an inorganic core of the particle, which may potentially decrease both the partitioning of the semi-volatile species and increase or decrease the extent of acid-catalyzed multiphase chemical reactions. Prior studies have demonstrated that aerosols phase separate over 70% of the time at a rural site in the Southeast US by using observed compositions with thermodynamic models to predict organic and inorganic constituents as well as aerosol water content. However, such phase separation processes are not routinely treated in atmospheric models used for regulatory purposes such as, US EPA's CMAQ, especially for multiphase reactions of IEPOX leading to SOA. This work develops algorithms for phase separation of SOA in a regional-scale atmospheric model (CMAQ). These new algorithms determine phase state based on the mechanistic interactions between oxidation products and meteorology, which result in the particle water uptake in organic or inorganic components of SOA and corresponding particle morphology. We determine the phase separation of the aerosols both in the semi-solid/solid phase separation and Liquid-Liquid phase separation (LLPS) regime. For semi-solid/solid phase separation, our approach first estimates the glass transition temperatures (T_g) of SOA components by accounting the variability in composition of different organic compounds, aerosol water, and the atomic oxygen-to-carbon (O:C) ratio. Then the phase separation was incorporated based on the phase state of the aerosol particles, until formation of homogeneous nuclei at Efflorescence Relative Humidity (ERH) occurs. When T_g indicates a liquid state, we also included algorithms to determine whether there is LLPS as a function of O:C ratio, Organic matter: Inorganic sulfate ratio and Separation Relative Humidity (SRH). In agreement with recent literature, T_g provides a more accurate indication of phase state and correlates well with the viscosity of SOA. Sensitivity analysis with different T_g estimation methodologies based on: a) C:H:O and b) vapor pressure/saturation concentration of different compounds, will be compared to the method that relies solely on O:C ratios. This work enables predictions of phase separation frequencies across varied conditions (urban or rural) and subsequently examines the influence on IEPOX-derived SOA phase separation due to aerosol viscosity, morphology, and phase state.

Impacts of Medium-Range Transport of Biomass Burning Aerosols on Air Quality and Public Health in Colombia

Karen Ballesteros-Gonzalez, Thalia Alejandra Montejo-Barato, Juan Manuel Rincon-Riveros, Maria Alejandra Rincon-Caro, and Ricardo Morales-Betancourt

Due to the high degree of urbanization and rapid growth of mobile sources atmospheric modelling efforts in Colombia have often been focused on evaluating the contribution of local sources to air pollution in urban areas. However, recent studies have suggested that an important fraction of the seasonality in PM2.5 concentration can be explained by the large number of wildland and open fires in Northern South America (NSA). Those studies have shown through remote sensing data and back-trajectory analysis, that air quality in NSA, both at local and regional scales, can be impacted through medium-range transport of biomass burning plumes. In this work we use a regional chemical transport model (WRF-Chem) to study the processes associated to medium-range transport of aerosol particles from open biomass burning events over Colombia. We also use the model to assess its potential impacts in public health. Model was configured with nested domains covering the northern half of South America at a horizontal resolution of 27x27 km, 9x9 km, and 3x3 km, with 121x121, 127x127 and 133x133 grid cells for domains d01, d02, and d03, respectively, and 41 vertical levels. Numerical experiments where set-up to consider a NOFIRE scenario, where biomass burning emissions were neglected, and a FIRE scenario in which the Fire INventory from NCAR (FINN version 1) emission inventory was included. The chemical mechanism MOZART and the aerosol scheme MOSAIC with 4 aerosol bins were used. The simulation period was from 01 February 00h UTC to 28 February 23h UTC of 2018. We compered simulation results against PM2.5 and PM10 observations from Bogota's air quality networks. Speciated PM2.5 data obtained during a measurement campaigns, was also used to compare against modeled aerosol composition. Public health impacts of the biomass burning aerosols were estimated with a log-lineal concentration-response model, where health impacts were calculated under the assumption of equal toxicity and based on the PM concentration attributable to fires over Bogota according to modelling results. Simulations results show that fire plumes can be injected to upper layers, entering the free troposphere and PM can be efficiently transported over the Andean mountains reaching high altitude regions. Model simulations show an increment on PM10, and PM2.5 concentration over Bogota when biomass burning emissions are included in the simulations, with a substantial contribution from secondary organic aerosols. Emissions from biomass burning events at the eastern plains of NSA are the main regional influence in PM concentration over Colombia.

CAMQ-AI: A computationally efficient deep learning model to improve CMAQ performance over the United States

Ebrahim Eslami, Alqamah Sayeed, Yunsoo Choi, Yannic Lops

A new computationally-efficient deep learning (DL)-based model was proposed to improve numerical model results. The model, called CMAQ-AI, uses the Community Multiscale Air Quality (CMAQ) to forecast surface ozone. We used a deep convolutional neural network to map CMAQ outputs (used as DL inputs) with the observed hourly ozone concentration at the monitoring station location (as DL target). The CMAQ outputs are the meteorological parameters from the Meteorology-Chemistry Interface Processor (MCIP) and ozone precursors from CMAQ Chemistry-Transport Model (CCTM). The CAMQ-AI model domain covered the contiguous USA, and model data were verified against U.S. Environmental Protection Agency AIRNow ozone observations. These observations are measured at 1081 quality-controlled stations over the CMAQ domain (with a spatial resolution of 12 km). The model was trained using three years (2011 to 2013) of CMAQ forecasts of hourly ozone concentrations from April to October. The model was tested using 2014 CMAQ forecasts and observation data. The CMAQ-AI model significantly improved the performance of the CMAQ model both in accuracy (Pearson correlation coefficient and index of agreement) and bias (maximum daily ozone). Index of agreement and correlation coefficient (r) improved on average by 0.13 and 0.16 (up to 0.37 and 0.60), respectively. The CMAQ-AI model was found to improve the simulated ozone peaks for almost the entire year and the whole domain. The CMAQ-AI model reduced the CMAQ's prediction bias more than 20 ppb on average. The number of low accuracy and high bias days was also significantly decreased after using the CMAQ-AI model for almost all States. In addition, the model was able to statistically identify the shortcomings of the CMAQ model for cases that CMAQ performed poorly. The systematic improvements in the CMAQ simulations suggest that the deep learning model is a suitable technique to reproduce an accurate estimate of ground-level air quality concentration. While this study focused on ozone in the United States, the proposed approach can be applied for any measured air pollution parameters in a mesoscale resolution.

Source Attribution of Global PM2.5 Mortality Using the Adjoint of Hemispheric CMAQ

Yasar Burak Oztaner, Shunliu Zhao, Amir Hakami (Carleton University); Amanda Pappin (Health Canada); Rohit Mathur (US EPA); the CMAQ-Adjoint Development Team

PM2.5 human exposure which is directly related with mortality is a major global health concern. The Global Burden of Disease (GBD, 2015) assessments indicate that outdoor fine particulate matter causes over 4 million premature deaths annually. In this study, we present backward/adjoint analysis to provide location-specific source attribution of the long-term mortality from exposure to fine particles in the Northern Hemisphere. We will later extend the study to include mortality in the Southern Hemisphere. We apply U.S. EPA's (CMAQv5.0) and its adjoint to quantify the premature mortality associated with exposure to ambient PM2.5. Meteorological inputs are from the Weather Research and Forecasting (WRF v3.8.1) model, and emissions for the hemispheric-scale domain are taken from EDGAR/HTAPv2 inventory for the year 2010. Subsequently, these emissions are processed in (SMOKE) model to get hourly emissions. The simulations are carried out over a 108-km resolution for two-weeks of each season of 2010. We use the Global Exposure Mortality Model (Burnett et al., 2018) for chronic exposure mortality which assigns a generalized concentration-response function based on numerous cohorts across the globe. The global population data is obtained from NASA Socio-Economic Data and Application Center (SEDAC). Wherever possible, country-specific baseline mortality data will be applied (IHME, 2019). Mortality counts due to changes in PM2.5 emissions and its attribution to regional sources of various sectors will be estimated. The findings of this mortality source attribution will be used to form an evaluation matrix for control policy options for the northern hemisphere.

An efficient way in quantifying the nonlinear response of air pollution to emission changes using the indicator-based response surface model

Jia Xing^1,2, Dian Ding^1,2, Shuxiao Wang^1,2, Zhaoxin Dong^1,2, Carey Jang³, Yun Zhu⁴, Jiming Hao^1,2

¹ State Key Joint Laboratory of Environmental Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China

² State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China

³ Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, USA

⁴ College of Environmental Science & Engineering, South China University of Technology, Guangzhou Higher Education Mega Center, Guangzhou, China

Efficient prediction of the air pollutant responses to precursor emissions is a prerequisite for an integrated assessment model system in developing effective control policies. Representing nonlinear behavior of air pollutants in response to emissions control with accuracy remains a major challenge for policy makers and modelers. Our previous study suggested that the nonlinear response of PM2.5 and O3 to emissions can be represented using a set of polynomial functions (i.e., response surface model with polynomial function, pf-RSM) in which the 14 term coefficients are fitted by at least 20 randomly selected control scenarios simulated by the chemistry transport model (CTM). In this study, we investigated the correlations between the 14 term coefficients and ambient concentrations of chemical species related to PM2.5 and O3. The results show that the 14 term coefficients in the polynomial function could be successfully estimated from a linear combination of ambient concentrations of 16 and 18 species from the baseline, fully controlled and additional 5 control scenarios (denoted as indicators) for O3 and PM2.5 respectively. The indicator-based response surface model with polynomial function (denoted as i-RSM) significantly enhances the efficiency and practicality since it only requires 5 additional scenarios beside baseline and full-controlled scenarios simulated by CTM. The method of i-RSM was successfully applied in China with 27km resolution as well as Northern China Plain (NCP) with 9 km resolution. The performance of i-RSM was evaluated through comparisons with CMAQ brute-foruce simulations and pf-RSM predictions. The results showed that i-RSM predicted compatible responses and sensitivities of O3 and PM2.5 to precursor emissions with those simulated by CTM and pf-RSM. The mean normalized biases, mean fraction biases, mean normalized errors and maximal fractional errors in i-RSM are within 0.5%, 0.5%, 2% and 18% across all grid cells over 9km NCP domain. The nonlinearity of PM2.5 and O3 to precursor emissions was investigated and the principle of using indicator in i-RSM to represent the nonlinear response was also discussed.

Traffic-Related PM2.5 and NO2 Health Risk Assessment in the United States: A Fine Scale Hybrid Modeling Approach

Alejandro Valencia, Marc Serre, Dongmei Yang, Sarav Arunachalam

Residing near major roadways with significant emissions of traffic-related air pollutants (TRAPs) has been associated with a range of adverse health effects. Most TRAP concentrations are highest in the vicinity of roads and decrease dramatically to background levels within 150 m to 200 m of the road. Previous national scale approaches that quantify the TRAPs burden of disease use coarse resolution (110km, 36km, 12km, etc.) chemical transport models (CTMs) or statistical land-use regression (LUR) models. However, CTMs at these resolutions lack the spatial gradients necessary to assess the relationship between TRAP population exposure and health. The LUR models ignore chemical/transport process that can better quantify road source contributions. The goal of this work is to show an improved assessment of on-road emissions and show how they affect air quality and health in the continental U.S. Thus, we propose, a hybrid approach that utilizes the combined strength of air quality observations and regional/local scale models. The regional photochemical grid model CMAQ (Community Multiscale Air Quality) will predict the spatiotemporal impacts at local and regional scales. And the local scale dispersion model, R-LINE (Research LINE source) will estimate concentrations that capture the sharp TRAP gradients from roads. This work builds upon previous work where we quantified that PM2.5 related health risk was underestimated by 23% in Central North Carolina when near-road gradients were not incorporated. We further apply Bayesian Maximum Entropy (BME)-based data fusion techniques that integrate both model predictions and AQS observations and consider the model's error to improve exposure estimates. By combining observations, highly resolved local and regional scale models, we will estimate TRAP concentrations (specifically NO2 and PM2.5) at census block centroids level for the entire U.S. and conduct a detailed assessment of the health-risk related to these pollutants. Our results will provide an accurate characterization of total health risk due to traffic-related pollutants and specifically aid in identifying vulnerable populations near-road and avoiding possible exposure misclassification when near-road gradients are not included.

Estimating damages from climate change and air pollution for subnational incentives for green on-road freight

Rebecca K. Saari, Wilson Wang, Chris Bachmann, Ushnik Mukherjee

Subnational incentives to adopt zero emission vehicles (ZEVs) are critical for reducing the damages posed by transportation to air quality and the climate. Few studies estimate these damages for on-road freight, especially at scales relevant for subnational policies. Here, we assess the damages associated with emissions of air pollutants (PM2.5, NOx, SO2, NH3), and greenhouse gases (CO2, CH4, N2O) from freight trucking, and the benefits of ZEV adoption by census division in the Province of Ontario. We develop an integrated modelling framework that connects a travel demand model, a mobile emissions simulator, and reduced form model of the marginal damages of air pollutants and the social costs climate change. We estimate $1.8 billion (2005 USD) annual damages resulting from atmospheric emissions from medium and heavy duty trucks for Ontario in 2012. This implies $7,700 per truck per year in damages, which could inform an economic incentive to reduce these emissions. Meeting the provincial goal of 5% ZEV adoption would yield approximately $89 Million (2005 USD) in benefits annually from these trucks alone. This result varies by up to 25% according to the sensitivity analysis related to the travel and emissions models, though the economic damages are likely the largest uncertainty source. Such advances in subnational scale, computationally efficient, integrated modeling of the environmental impacts of freight can offer insights into the sustainable design of policy affecting this economically vital, growing, but polluting sector.

Hybrid Air Quality Modeling in West Oakland, California, to Support the Development of an Emissions Reduction Program

Stephen Reid, Bonyoung Koo, Yiqin Jia, James Cordova, Virginia Lau, Annie Seagram, Yuan Du, Minh Nguyen, David Holstius, Phil Martien

Bay Area Air Quality Management District, San Francisco, CA

Assembly Bill (AB) 617, adopted in California in 2017, continues and expands work undertaken at the Bay Area Air Quality Management District (BAAQMD) more than a decade ago to reduce exposures in communities most impacted by air pollution. Under AB 617, local air districts are tasked with partnering with community groups and other stakeholders to develop community emission reduction programs (Action Plans). For the first Action Plan in the San Francisco Bay Area, BAAQMD partnered with the West Oakland Environmental Indicators Project community group to reduce air pollution exposures in the West Oakland, a community surrounded by freeways and adjacent to the fifth largest container port in the U.S.

To support the West Oakland Action Plan, BAAQMD conducted a technical assessment of air quality in the West Oakland, work that included the development of a bottom-up emissions inventory for PM2.5 and air toxics, air quality modeling, and source apportionment analyses. First, BAAQMD applied the CMAQ model at 1-km grid resolution to characterize PM2.5 and air toxics concentrations at the regional scale; CMAQ results were used to provide an estimate of background pollutant concentrations in West Oakland concentrations that would occur in the absence of any local emission sources in the community. Next, the AERMOD dispersion model was applied to develop "community-scale" pollutant concentrations and estimate the contribution of local sources to annual average pollutant concentrations and cancer risk at finely-spaced receptor locations in West Oakland.

Sources modeled with AERMOD include permitted stationary sources, on-road motor vehicles, and port-related sources, such as ships, tug boats, locomotives, and cargo handling equipment. Results of the modeling analyses were used to map pollutant concentrations and cancer risk and quantify the contribution of specific sources to air quality impacts in different parts of the community. Findings from this technical assessment helped inform and set targets for West Oakland's Action Plan.

This presentation will focus on the hybrid CMAQ/AERMOD modeling approach and share key findings that will inform ongoing AB 617 technical assessments.

Improving SCICHEM Pre- and Post-Processing Setup Speed Using Cloud Computing

Amy McVey - AER

Jarrod Lewis - AER

Matthew J. Alvarado - AER

Prakash Karamchandani - Ramboll

Douglas Henn - Xator Corp

Eladio Knipping - EPRI

All air pollution dispersion models take time to set up. Learning to run new models can take significant training time and getting the appropriate data and inputting the information to the correct formatted input files for multiple model runs can be tedious. In this project, we used Amazon Web Service (AWS) cloud computing resources and AER's AQcast modeling platform to develop a scalable, automated, cloud-based system for running the SCICHEM model (v3.2) that will allow both expert and novice users to perform model simulations for a wide variety of sources. The SCICHEM model, which stands for SCIPUFF with Chemistry, is used to model the transport, dispersion and chemical reaction of gaseous and aerosol releases in the atmosphere using gas phase, aqueous phase and aerosol chemistry treatments that are comparable to those in photochemical grid models (PGMs). The user interface for the system is hosted on the web and uses intelligently-selected defaults in the absence of user input, allowing novice users to perform regulatory-quality modeling simulations with just the specification of the site location, stack height, and emission rates, but also allows for all of the options of the underlying SCICHEM model to be invoked and altered by expert users. The system includes all of the necessary pre-processors for each model, which are run automatically on the AWS cloud via AWS Batch. Needed input data for each pre-processor and model is automatically downloaded as needed from external data sources, stored as part of the system itself (e.g., WRF and CMAQ output), or supplied by the user via the web-based user interface. Model output is post-processed and displayed on the web, and the modeling protocol, including model input and output files, is archived for download and review by the user and the regulatory authority. Our presentation will demonstrate the system, discuss the implementation of the system on the AWS cloud, and present a few example simulations performed by the system.

Using mobile phone data to quantify the impact of spatiotemporal human mobility on air pollution exposure estimation

Haofei Yu. Department of Civil, Environmental, and Construction Engineering. University of Central Florida. Orlando, FL. USA

Cesunica Ivey. Department of Chemical and Environmental Engineering, University of California Riverside. Riverside, CA. USA

Xiaonan Yu. Department of Civil, Environmental, and Construction Engineering. University of Central Florida. Orlando, FL. USA

Lucas Henneman. T. H. Chan School of Public Health, Harvard University. Cambridge, MA. USA

Zhijiong Huang. Institute for Environmental and Climate Research, Jinan University, Guangzhou, China.

The spatiotemporal movement of human individuals has substantial implications on their air pollution exposure. However, mobility is usually neglected in exposure estimation due to lack of data, and exposure misclassification errors are likely. How mobility impacts the results of exposure estimation at population and individual levels remains under-investigated. In this study, we applied a large cell phone location dataset containing over 35 million location records collected from 310,989 subjects from Shenzhen, China to investigate how different levels of mobility impacted each subject's estimated exposures for five chosen ambient pollutants. Additionally, we applied and compared the results of exposure estimates based on concentration fields developed using two different methods: CMAQ model outputs, and the inverse distance weighting (IDW) method. Our results showed that including detailed mobility information does not have considerable impact on the estimated exposures averaged across the entire population, though the impact of mobility at the individual level can be substantial. We found that the errors in exposure estimates and exposure misclassifications generally increase with increased mobility when a exposure were estimated at each individual's home address. Neglecting mobility results in underestimated exposures to traffic-related pollutants, particularly during afternoon rush-hour and overestimated exposures to ozone during mid-afternoon. We found that pollutant concentration fields generated using the IDW method are smooth and not suitable for exposure estimation when detailed mobility data were considered. Our findings highlighted the tremendous potentials of using cell phone location data in air pollution exposure estimation for a large population, and our results have significant implications for future air pollution exposure and health studies.

CMAQ 5.3b PARALLEL PERFORMANCE WITH MPI AND OPENMP

George Delic, HiPERiSM Consulting, LLC, PO Box 569 Chapel Hill, NC 27514.

This presentation covers selected thread parallel performance results for CMAQ 5.3b from a forthcoming publication [1]. Attention is focused on the Gear and Rosenbrock solvers in the Chemistry Transport Model (CTM), for both FSparse [2], and the legacy JSparse [3] algorithms. The former implements OpenMP thread parallelism for which these two solvers are well suited. The results include execution performance and numerical precision results with a 24 hour scenario included with the CMAQ download. Both the legacy (EPA) JSparse and the FSparse thread parallel versions are compared. Results will be presented for both MPI and thread scaling on homogeneous and heterogeneous configurations in a cluster of 10 nodes with a total of 128 cores.

[1] G. Delic, Modern Environmental Science and Engineering, issue 9, 2019.

[2] G. Delic, Annual CMAS conference, 2012, 2013, 2016, 2017, 2019.

[3] M. Jacobson and R.P. Turco (1994), Atmos. Environ. 28, 273-284.

HyADS: A tool for estimating nationwide exposures to emissions from large numbers of sources

Lucas RF Henneman, Christine Choirat, Joan Casey, Irene Dedoussi, Cesunica Ivey, Kevin Cummiskey, Corwin Zigler

We endeavor to quantify national health and environmental impacts of changes in exposure to coal power plant SO2 emissions across periods of regulation-driven emissions changes. To do so, we employ a new reduced complexity model HyADS. The model is designed to flexibly capture air pollution exposure variability on multiple spatial and temporal scales from individual sources for use in air pollution epidemiology and population exposure studies. HyADS, which is accessible through an R package, averages hundreds of HYSPLIT transport and dispersion forward trajectories per source-day. In this talk, I will begin by introducing the HyADS model. I will discuss model evaluations that employ a variety of observed and modeled quantities. For instance, we found excellent agreement between HyADS-modeled exposure to emissions from all coal-fired power plants and PM2.5 sensitivities to coal emissions in the United States modeled with CMAQ-DDM. Similarly, we found high agreement with state-specific individual coal-fired power plant PM2.5 sensitivities modeled with the GEOS-Chem adjoint model. Next, I will illustrate HyADS spatial, temporal, and source flexibility relative to existing reduced complexity models by describing two recent applications of the HyADS model in epidemiological studies to identify improvements in health attributable to interventions on coal power plants. These studies include an accountability assessment of changes in nine hospitalization outcomes and all-cause mortality in the Medicare population between 2005 and 2012 and an assessment of changes in asthma outcomes in Louisville, KY with control installations and retirements at four nearby coal-fired power plants. Finally, I will provide recommendations and future uses of the model based on the evaluations and applications to date.

Global Sources of North American Ozone

Barron H. Henderson, Pat Dolwick, Carey Jang, Alison Eyth, Jeff Vukovich, Rohit Mathur, Christian Hogrefe, George Pouliot, Brian Timin, K. Wyat Appel

Air quality management practices often focus on local sources to address local air pollution problems. However, it is also important to consider the global contributions to local pollution in order to develop effective approaches to achieve healthy air quality. The Hemispheric Transport of Air Pollutants (HTAP) and Air Quality Model Evaluation and International Initiative (AQMEII) have highlighted intercontinental contributions to ozone in North America and Europe. The US EPA Policy Assessment for the 2015 National Air Quality Standard for ozone highlights the role of background ozone, which includes natural and anthropogenic international sources. In recognition of the potential air quality contributions from international emissions there is a need for credible large-scale air quality modeling to support regulatory assessments. This talk will highlight a recent large-scale model application study, consider the broad collaboration that enabled it, and close with proposed areas for future work.

Inter-continental transport of air pollution occurs at time scales of days to weeks within the northern or southern hemisphere. Quantifying transport of anthropogenic pollution between countries at this scale requires a modeling system that credibly represents global processes. Emissions and transport have independent uncertainties that need exploration and both affect US national composition. Our application uses updated emissions from the EPA 2016 modeling platform, Mexico, Canada, and China to improve the representation of emissions from domestic and foreign sources. Our results include updates to evaluation with surface monitors, sondes, and aircraft to include available satellite products.

The base simulation is supplemented by "zero-out" simulations to characterize ozone contributions from anthropogenic and natural sources. The anthropogenic sources are split into USA and International. The international results are further refined to provide country estimates from China, India, and Canada/Mexico. In addition to country-specific contributions, estimates of global shipping ozone and all fire-related ozone contributions are provided. This presentation will share 108km and 12km scale modeling results highlighting natural and international contributions around the northern hemisphere.

The results will be discussed in the context of the existing literature and relevant transport time and distance scales. Each emission source's spatial and temporal variability influences the ability of ozone to be transported to the US, and the spatial distribution over the US. These results provide insights into the need for further observation and modeling studies.

Version 1.0 of the National Emissions Inventory Collaborative 2016 Emissions Modeling Platform

A. Eyth, Z. Adelman, D. Boyer, C. Farkas, M. Janssen, S. Kayin, T. Manning, J. McDill, S. Roberts, J. Snyder, T. Richardson, J. Vukovich, E. Zalewsky

The National Emissions Inventory Collaborative is a partnership between state emissions inventory staff, multi-jurisdictional organizations (MJOs), federal land managers (FLMs), EPA, and others to develop a North American air pollution emissions modeling platform with a base year of 2016 for use in air quality planning. The Collaborative planned for three versions of the 2016 platform: alpha, beta, and Version 1.0. Version 1.0 was released in September 2019 and improved on the 2016 beta platform by incorporating updated data from state and local agencies, updated methods for many of the inventory sectors, and some data and methodologies compatible with the 2017 National Emissions Inventory (NEI). In addition to the base year 2016 emissions, the 2016 version 1.0 platform includes future year emissions for 2023 and 2028. This presentation will describe the Collaborative, detail the approaches used to the develop the 2016 version 1.0 platform, and compare the version 1.0 emissions to previous national modeling platforms.

Local to Global Air Quality Simulations using the NASA GEOS Composition Forecast Model GEOS-CF

Christoph A. Keller (Universities Space Research Association, Columbia, MD / NASA Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, MD)
K. Emma Knowland (Universities Space Research Association, Columbia, MD / NASA Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, MD)
Bryan N. Duncan (Atmospheric Chemistry and Dynamics Laboratory, NASA Goddard Space Flight Center, Greenbelt, MD)
Steven Pawson (Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, MD)

We give an overview of the NASA Global Earth Observing System Composition Forecast model (GEOS-CF), a high-resolution (roughly 25-km) global composition model developed by the NASA Global Modeling and Assimilation Office (GMAO). This system combines the GEOS weather and aerosol model with the GEOS-Chem chemistry module to provide a holistic view of atmospheric composition that captures a wide range of air pollutants such as ozone, nitrogen oxides, volatile organic compounds, and fine particulate matter. Given the global extent of the model, GEOS-CF captures large-scale processes such as long-range transport of air pollutants, across ocean basins and continents. At the same time, the ~25-km spatial resolution is fine enough to resolve local features such as nighttime ozone titration.

Comparisons against surface observations highlight the model's overall capability to reproduce the diurnal variability of air pollutants under a variety of meteorological conditions. In addition, we show how machine learning techniques can be used to correct for sub-grid-scale variability, which further improves model estimates at a given observation site.

The GEOS-CF system offers a new tool for scientists and the public health community alike and is being developed jointly with several government and non-profit partners. As an example, we show the use of GEOS-CF during the Satellite Coastal and Oceanic Atmospheric Pollution Experiment (SCOAPE). The campaign, conducted in collaboration between NASA and the Bureau of Ocean Energy Management (BOEM), aims to investigate the impact of offshore oil and gas exploration, development and production on onshore air quality. Detailed gas-phase chemistry, as provided by GEOS-CF, is critical to understand the formation of air pollution related to hydrocarbon emissions from offshore oil and gas activities. The accuracy of GEOS-CF can be further improved by incorporating detailed offshore emissions compiled by BOEM, demonstrating the potential feedback of such collaborative endeavors on future model developments.

Updates and Improvements to 2023 and 2028 Emission Inventory Projections

Caroline Farkas, Alison Eyth, Jeffrey Vukovich

Emission inventories projected to future years are used in air quality models to estimate future air quality concentrations. These projections often play a critical role in regulatory, planning, and research processes. As part of a collaborative effort to create a 2016 emissions inventory for air quality modeling, EPA worked alongside Federal, State, Local, and Tribal agencies and Multi-Jurisdictional organizations (MJOs) to update and improve on projection methods of a US emission inventory for years 2023 and 2028. We review updated methods for projecting and applying controls to emission sectors and their resulting changes temporally and spatially throughout the US. Further, we describe the mix of key source sectors that impact air quality in the future year inventories versus those in the base year.

CAMx Ozone Source Apportionment Application over the Northern Hemisphere

Pradeepa Vennam¹, Christopher Emery¹, Greg Yarwood¹, Jim Smith² and Shantha Daniel²

¹Ramboll, Novato, California.

²Texas Commission on Environmental Quality, Austin, Texas

It is increasingly important for air quality planning in the US to quantify the contribution of international transport of air pollution. Global and hemispheric models are available to study inter-continental transport but none include multi-pollutant source tagging, such as CAMx source apportionment technology (SA) for ozone and PM. Using a 2016 Northern Hemispheric dataset with 108-km grid resolution developed by US EPA for CMAQ, we demonstrate a hemispheric CAMx "proof-of-concept" application with ozone SA tagging for 6 areas (US, Canada, Mexico, Central and South America, East Asia and other). We also updated CAMx and an existing boundary condition interface so that SA tracers can be communicated from the hemispheric simulation to a 1-way nested North American simulation. We describe the hemispheric CAMx model and present source contributions to inter-continental ozone transport for several areas of Texas.

Development of a Year 2016 fire inventory for United States through a multi-agency inventory collaboration effort

Jeffrey M. Vukovich, USEPA

James Beidler, General Dynamics Information Technology

Wildland fire has a significant impact on air quality in the United States. In past National Emissions Inventories (NEIs), wildland fires within the United States have been shown to be the largest-emitting PM2.5 source category. An Inventory Collaborative effort for year 2016 was organized to generate emissions inventories for use in an emissions modeling platform. The Collaborative effort consisted of federal, Multi-Jurisdictional Organizations (MJOs), state and other agencies. This effort included generating a wildfire and prescribed burn emissions inventory for the entire year of 2016. This inventory combines multiple sources of ground reports of fire information obtained from federal, state and tribal organizations with satellite detections from the Hazard Mapping System. We describe the methods used to prepare the inventory as well as spatial and temporal patterns observed in the fire activity and emissions. The challenges that occurred when generating this emissions inventory and how some of these challenges may be addressed with upcoming new tools will be discussed.

Quantifying Economic Damages from Crop Productivity Loss Due to Ozone Precursor Emissions via Hemispheric Adjoint Analysis

Yasar Burak Oztaner, Shunliu Zhao, Amir Hakami (Carleton University); Rohit Mathur (US EPA); the CMAQ-Adjoint Development Team

Long-term exposure to the high ozone concentrations triggers destruction or loss to plants and crops. The global economic cost of wheat production loss is estimated to be $14-24 Billion for the year of 2000 (Van Dingenen et al., 2009).Many conducted studies for reducing of the concentrations were all scenario-based, but recently adjoint models have been used to estimate crop damages as well (Lapina et al., 2015; Capps et al., 2014). We aim to carry out economic benefit analysis of emission control impacts on crop potential productivity of corn and wheat in North America and Europe as well as across the Northern Hemisphere. Our approach will, for the first time, use the adjoint of the full CMAQ model in a hemispheric setting. We apply U.S. EPA's (CMAQv5.0) and its adjoint to quantify the impact of emission reduction of NOX and VOCs on crop potential productivity and its economic benefits. Meteorological inputs are from the Weather Research and Forecasting (WRF v3.8.1) model, and emissions for the hemispheric-scale domain are derived from EDGAR/HTAPv2 inventory for the year 2010. Subsequently, these emissions are processed in (SMOKE) model to get hourly emissions. The simulations are carried out over a 108-km resolution for four months (May-August 2010). ). Country-specific wheat and corn production data for the year of 2010 from FAO are obtained and used in this study. We use crop yield maps created by Monfreda et al. (2008) and Ramankutty et al. (2008) to distribute wheat and corn production across the modelling domain. Economic benefits are determined using country-specific annual producer prices for each crop type obtained from FAOSTAT. We implement W126 O3-vegetation exposure index as reported by EPA and AOT40 O3-vegetation exposure index (accepted by EEA) for each crop type separately. Economic benefits due to the reduced O3 and the subsequent impact on crop productivity (corn and wheat) are calculated. The adjoint model is used to attribute estimated crop damage and its associated economic valuation to individual sources across the hemisphere. The findings of potential productivity loss for the two types of crop and their economic benefits for emission control policy options for North America and Europe as well as Northern Hemisphere will be discussed.

Estimating Emissions from Wildland Fires for Air Quality Modeling: Status Update

George Pouliot, Joseph Wilkins, Tom Pierce, James Beidler, Jeff Vukovich, Venkatesh Rao

Biomass burning from wildfires, prescribed fires, grasslands, rangelands, and crop residue is an important contributor to the degradation of air quality because of its impact on ozone and particulate matter. During the past several years, there have been several updates and revisions to the estimation of emissions from fires and the methods used in regional air quality modeling systems. This presentation summarizes six recent advances and updates that have been or will be implemented in the emission inventory process (e. g. geospatial locations of fires, differentiating between flaming and smoldering), the emission modeling framework (e.g. calculation of emissions from fuel loading and emission factors), or the air quality modeling system (e.g. plume injection height) . The BlueSky framework, which is the primary tool for estimating emissions from wildland and prescribed fires, contains a fuel consumption model known as CONSUME. For the 2016 and 2017 fire inventories, the residual smoldering estimate from CONSUME has been separated from the flaming component so that air quality models can model these two combustion processes separately. Beginning with the 2014 NEI, emission estimates for crop residue burning has been updated and revised to provide a consistent methodology and to incorporate information from multiple sources along with updates to the VOC emission factors to ensure consistency between the Hazardous Air Pollutants and Criteria air pollutant inventories. For the 2016 and 2017 fire inventories, grassland fires in the Flint Hills region of Kansas have been assigned a unique source classification code, and grassland fires (both wildfires and prescribed) have been estimated in the BlueSky Framework to avoid any double counting with the crop residue inventory. The current set of fire processing tools including the Bluesky framework and SMARTFIRE have been somewhat difficult to maintain because of the proprietary nature. Efforts are underway to update these tools to the new Bluesky pipeline that is open-source and more modular. Some additional analysis of the Moderate Resolution Imaging Spectroradiometer (MODIS) fire detections has resulted in improvements to the geospatial location of MODIS fire detects from the NASA MODIS fire product archive. Combining this improved geospatial information with our Hazard Mapping System fire location information is ongoing work. Finally, we have implemented an alternative approach to the plume rise algorithm currently in the Community Multiscale Air Quality (CMAQ) modeling system using the Sofiev method. Comparisons and analysis of this method with the existing method highlights the need for further research into the plume height injection methods needed in regional air quality modeling systems.

Incorporation of Volcanic SO2 Emissions in H-CMAQ Modeling System and its Impacts on Sulfate Aerosol Concentration across the Northern Hemisphere

Syuichi Itahashi (Central Research Institute of Electric Power Industry, Japan) Rohit Mathur (US Environmental Protection Agency, USA) Christian Hogrefe (US Environmental Protection Agency, USA) Sergey L. Napelenok (US Environmental Protection Agency, USA) Yang Zhang (North Carolina State University, USA)

The Community Multiscale Air Quality (CMAQ) modeling system has been recently extended by the U.S. EPA to the hemispheric scale (H-CMAQ). The input emission data for H-CMAQ can be configured with available emission inventories; however, current applications have lacked inclusion of possible impacts of volcanic SO₂ emissions. Because volcanoes are mostly located in Pacific Rim region, their emissions could impact ambient sulfate aerosol concentration over both Asia and the U.S. We analyze the impact of volcanic SO₂ emissions on large scale air pollutant distributions during April 2010, a period during which trans-Pacific transport of photochemical oxidants (Ox) has been previously analyzed using the H-CMAQ modeling system. The results show increase in sulfate aerosol concentration of more than 0.5 g/m³ over coastal East Asia, eastern Pacific, and western Atlantic. This value corresponds to a 10-20% increase compared to the base-case simulation which did not include volcanic SO₂ emissions. This work will improve our understanding of the importance of volcanic SO₂ emissions on ambient sulfate aerosol concentration over northern Hemisphere based on the state-of-the-art hemispheric modeling system.

Evaluating Updated Tools for the Estimation of Wildland Fire Emissions

James Beidler General Dynamics/Information Technology beidler.james@epa.gov

George Pouliot Computational Exposure Division, National Exposure Research Laboratory U.S. Environmental Protection Agency, Research Triangle Park, NC 27711 pouliot.george@epa.gov

Wildland fires are a major contributor of primary particulate emissions in the United States. The default method used in the National Emissions Inventory (NEI) for estimating US wildland fire emissions is a multistep process where fire reports, perimeter, and satellite data are reconciled into fire activity using SmartFire2 and emissions are estimated from the activity using the BlueSky Framework (BSF). Since the default method was first applied in the 2011 NEI, new open source versions of these tools with updated methods have become publicly available. In this study, we highlight key feature changes and differences in estimated emissions between the BlueSky Framework and the BlueSky Pipeline. Additionally, we compare methodological differences and estimated activity between SmartFire2 and the R-based SmartFire3 project.

Evaluation of the Community Multiscale Air Quality (CMAQ) model version 5.3

K.W. Appel, C. Hogrefe, K. Foley, B. Murphy, H. Pye, J. Bash, J. Pleim, G. Sarwar, D. Wong, D. Luecken, W. Hutzell, K. Fahey, S. Roselle, L. Ran, F. Sidi, D. Kang, R. Gilliam and R. Mathur

More than a year ago, a beta version of the Community Multiscale Air Quality (CMAQ) model version 5.3 was released and evaluated. Since that time, additional development of CMAQv5.3 has taken place on the path to a final version of the model for release in the summer 2019. Updates to the model over the past year have addressed several model performance issues identified in the beta release evaluation, with these updates resulting in potentially significant changes in performance compared to the beta version of the model. In this study, we present an evaluation of the final release version of CMAQv5.3, comparing the model results to the previous version of the model (v5.2.1) and against an extensive array of observations. Simulations are performed for the contiguous United States using a regional configuration of the model and for the northern hemisphere using a version of the model configured for hemispheric simulations. The impact of the more extensive updates and/or new options in the modeling system (e.g. AERO7; STAGE) will be examined exclusively from the other model updates to provide a comprehensive assessment of the impact these larger updates have on model performance. Evaluation of gas, particle and wet deposited species using routine observations from U.S. networks (e.g. AQS, IMPROVE, NADP), Europe networks (e.g. EMEP), campaign measurement data (e.g. DISCOVERAQ), and, where applicable, global and satellite data will be presented. The presentation will also highlight some of the new analysis features available in the latest version of the Atmospheric Model Evaluation Tool (AMETv1.4), also scheduled for release in summer 2019.

Development of a Fast Fire Emission Processor and Its application with HMS-Bluesky and GBBEPx Inventories

Youhua Tang^1,2, Daniel Tong^1,2,3, Pius Lee¹, Barry Baker^1,2, Patrick Campbell^1,2, Jeff McQueen⁴, Ho-Chun Huang^4,5, Li Pan^4,5, Jianping Huang^4,5 , Jose Tirado^6,7, Shobha Kondragunta⁸, Xiaoyang Zhang⁹, and Ivanka Stajner⁴

1. NOAA Air Resources Laboratory, 5830 University Research Court, College Park, MD. 2. Cooperative Institute for Climate and Satellites, University of Maryland, College Park, MD. 3. Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA. 4. NOAA National Centers for Environmental Prediction (NCEP), College Park, MD 5. I.M. Systems Group Inc., Rockville, MD 6. NOAA NWS/STI 7. Eastern Research Group, Inc (ERG) 8. NOAA NESDIS/STAR 9. South Dakota State University, Brookings, MD

We developed a fast fire emission processor to replace the standard SMOKE procedure for wildfire emission to allow rapid emission processing and the versatility to digest multiple satellite-based fire products for the NOAA National Air Quality Forecasting Capability (NAQFC) system. Starting from HMS (NOAA hazard mapping system)-Bluesky emission inventory, this processor combined the traditional several steps (Smkinven/Temporal/Elevpoint) into a single step with certain tunable parameters, such as diurnal/daily profile for prescribed burning and forest fires, respectively. It directly generated the CMAQ-ready-input in-line emission file without intermediate files. The total processing time for the fire emissions was reduced to several seconds for CONUS domain, which enabled us to extend the NAQFC forecast to one more day. Due to some degrading issues of the near-real-time HMS, we expanded this fast processor's usage to the blended global biomass burning emissions product (GBBEPx) emission inventory, and use satellite fire radiation power (FRP) to estimate the heat flux. Unlike fuel-loading based HMS-Bluesky inventory that yields the total fire emission for certain period over certain areas, the GBBEPx estimates a snapshot of fire emissions based on satellite retrievals, and we need derive all other parameters, such as burning duration, for the NAQFC forecasts in this emission processing. We compared the performance of HMS-Bluesky versus GBBEPx inventories in NAQFC system (CMAQ 5.0.2 driven by FV3 meteorology) during the spring and summer 2019 with various observations, including in-situ measurements and satellite retrievals. The GBBEPx inventory showed overall better correlations with surface data, implying that it better captured the time and location of fire emissions.

Evaluation of CMAQ Estimated NOx from 2002 to 2016

Kristen Foley, Heather Simon, Claudia Toro, Kirk Baker, Wyat Appel, Barron Henderson, Alison Eyth, Deborah Luecken

In recent years, many published studies comparing ambient NO_X and NO_Y concentrations to modeled or inventory values found a high bias on the order of 1.4 - 2 times observed levels. Some researchers proposed reducing mobile-source NO_x emissions by 30-70% in their modeling applications. Many of these applications were based on evaluation of summer 2011 modeling to leverage the latest 2011 National Emissions Inventory and various summer field campaigns such as the 2011 DISCOVER-AQ Baltimore. Here, model estimates of NO_X from 2002 through 2016 from the Community Multiscale Air Quality (CMAQ) model were compared to routine surface network measurements to identify differences in NO_X bias across years, seasons, time of day, and regions of the country. Evaluation against NO_X observations across the U.S. show that the high summertime bias has significantly decreased across this period with decreasing ambient NO_X levels and improvements in the emissions inventories and the CMAQ system over time. Wintertime NO_X is found to be underestimated in many regions of the country in this set of simulations. Aircraft measurements taken as part of the 2011 DISCOVER-AQ Baltimore field campaign were compared to model estimates to further explore how emissions, model formulation (e.g., chemistry), and measurement uncertainty contribute to predictive skill. In-depth analysis using 2011 field measurements showed that the model bias in NO_X and NO_z components was sensitive to choices about pairing of model and measured values with disparate spatial and temporal resolution and to the chemical mechanism used. Estimates of daytime NO_Y normalized mean bias in the boundary layer aloft could vary from 76% to 28% depending solely on choice of model chemistry and measurement method.

Impacts of a National Survey on the Development of the 2017 National Emissions Inventory

Rich Mason, U.S. EPA and David Cooley, Abt Associates

The US EPA has updated the Residential Wood Combustion (RWC) tool via results from a Council for Environmental Cooperation (CEC)-funded national survey approach developed and implemented by NESCAUM and Abt Associates. The CEC-based survey targeted all RWC devices and accounts for variations in urban vs rural areas, land cover, access to utility natural gas for primary heat, housing types, and climate. Responses include information on burn rates and probabilities for each appliance type for primary vs secondary/aesthetic heating and state-total wood consumption estimates were compared to existing U.S. Energy Information Administration (EIA) Residential Energy Consumption Survey (RECS) data. We collaborated with State inventory developers to craft a national approach for converting the CEC-based survey results into new activity data for the RWC tool for the 2017 National Emissions Inventory. We also compare these results to the previous RWC tool-based estimates used for the 2014 NEI.

Improved estimation of background ozone and emission impacts using chemical transport modeling and data fusion

Nash Skipper, Petros Vasilakos, Yongtao Hu, Armistead G. Russell

US background ozone (BGO) is the ozone that would be observed if US anthropogenic emissions were zero. BGO originates from noncontrollable sources (e.g., wildfires, stratosphere-troposphere exchange, non-domestic pollution) and can vary significantly by region, elevation, and season, leading to high uncertainty in BGO contributions. BGO is typically quantified using a chemical transport model, such as the Community Multiscale Air Quality (CMAQ) model, with anthropogenic emissions for the region or country of interest set to zero. A method of adjusting for model bias in the estimation of BGO has been developed that fuses model results with observations. Our method uses observational and modeled data to develop non-linear functions of space, time, temperature, emissions, and other key factors that relate CMAQ-simulated base case (using estimated emissions) and CMAQ-modeled US BGO (no US anthropogenic emissions) to the observations. Separate adjustment factors are developed for locally formed and background ozone. This allows for calculating both adjusted US BGO and the amount of ozone formed from anthropogenic emissions that better align with observations and elucidation of the key influences and sources of bias for these two sources of ozone. The effects of boundary conditions on BGO estimates and model bias is also examined.

EPA's 2018 Emissions & Generation Resource Integrated Database (eGRID): Updates and Improvements

Jonathan Dorn, Marissa Hoer and David Cooley, Abt Associates, Durham, NC

Travis Johnson, U.S. Environmental Protection Agency, CAMD, Washington, DC

Electricity generation is the dominant industrial source of air pollutant emissions in the United States today. Whenever you switch on an electrical appliance, chances are you are contributing to air pollution and greenhouse gas emissions. By documenting the environmental attributes of electric power generation, the Emissions & Generation Resource Integrated Database (eGRID) can help consumers, policy analysts and researchers to better understand the relationship between electricity and the environment. eGRID integrates many different federal data sources on power plants and power companies, including, but not limited to data sources from: EPA, the Energy Information Administration (EIA), and the North American Electric Reliability Corporation (NERC). Emissions data from EPA are carefully integrated with generation data from EIA to produce emission rates in pounds per megawatt-hour (lb/MWh), which allows direct comparison of the environmental attributes of electricity generation. eGRID is used by EPA, other government agencies, nongovernmental organizations, and private industry to quantify the release of emissions (i.e., SO₂, NO_x, CO₂, CH₄, N₂O, and PM_2.5) to the air and subsequently directly assess the impacts of air pollutants on natural resources. eGRID provides a convenient source of data for states implementing policies such as emissions disclosure, output-based emissions standards, and renewable portfolio standards.

EPA is slated to release eGRID2018 by the end of 2019. This paper will: 1) discuss the procedures for developing eGRID and the recent improvements, such as the addition of PM_2.5 emissions rates, found in eGRID2018; 2) provide an overview of the current emission rates by region and state in the United States; and 3) discuss proposed additions and improvements for future editions of eGRID.

Evaluating Seasonality and Trends in Modeled PM2.5 Concentrations Using Empirical Mode Decomposition

Huiying Luo1, Marina Astitha1*, Christian Hogrefe2, Rohit Mathur2, S. Trivikrama Rao1,3

1University of Connecticut, Department of Civil and Environmental Engineering, Storrs-Mansfield, CT, USA

2U.S. Environmental Protection Agency, Research Triangle Park, NC, USA

3North Carolina State University, Raleigh, NC, USA

Regional-scale air quality models are being used for studying the sources, composition, transport, transformation, and deposition of PM2.5. The availability of decadal air quality simulations provides a unique opportunity to explore sophisticated model evaluation techniques rather than relying solely traditional operational evaluations. In this study, we propose a new approach for process-based model evaluation of speciated PM2.5 using Empirical Mode Decomposition (EMD) to assess how well version 5.0.2 of the coupled WRF-CMAQ model simulates the time-dependent long-term trend and cyclical variations in daily average PM2.5 and its species, including SO4, NO3, NH4, Cl, OC and EC. The use of the proposed approach for model evaluation is demonstrated at three monitoring locations. At these locations, the model is generally more capable of simulating the rate of change in the long-term trend component than its absolute magnitude. Amplitudes of the sub-seasonal and annual cycles of total PM2.5, SO4 and OC are well reproduced. However, the time-dependent phase difference in the annual cycles for total PM2.5, OC and EC reveal a phase shift of up to half year, indicating the need for proper temporal allocation of emissions during this study period and an update to the treatment of organic aerosols compared to the model version used for this set of simulations. Evaluation of several sub-seasonal and inter-annual variations indicates that model is capable of replicating the sub-seasonal cycles in terms of magnitude and phase shift.

Expansion of a Size Distribution Profile Library for Particulate Matter (PM) Emissions Processing from Three to 30 Source Categories

Junhua Zhang and Michael D. Moran

Air Quality Research Division, Environment and Climate Change Canada, 4905 Dufferin Street, Toronto, ON, M3H5T4, Canada

Elisa Boutzis

IT/Net-Ottawa Inc., 150 Elgin Street, Suite 1800, Ottawa, ON K2P 2P8, Canada

This presentation will describe a significant upgrade to the PM (particulate matter) size distribution profile library used for preparing emissions files for the Environment and Climate Change Canada chemical transport model GEM-MACH (Global Environmental Multiscale-Modelling Air-quality and CHemistry). This model uses a sectional (bin) approach to represent the PM size distribution. Two sectional configurations are commonly used: two-bin or 12-bin. For the two-bin configuration, PM10 emissions are separated into two size bins, PM2.5 (fine bin) and PMC (coarse bin, equal to PM10 - PM2.5), whereas for the 12-bin configuration, PM10 emissions are disaggregated into 10 size bins ranging from 0.01 to 10.24 m in diameter (there are two larger size bins for diameters greater than 10 μm). For the 12-bin size disaggregation step, a small library of three generic PM size distribution profiles is currently applied for three broad source types (area, mobile, and point). These profiles are based on 10 source-specific PM size distribution profiles discussed in Eldering and Cass (1996). However, as might be expected these generic profiles are not always representative: for example, emissions from two very different area sources paved road dust and residential wood combustion are disaggregated using the same generic PM size distribution profile. In addition, PM emissions for GEM-MACH are speciated chemically into six components: sulphate, nitrate, ammonium, black carbon, primary organic matter, and crustal material using 89 PM speciation profiles based on the simplified PM speciation profiles compiled from the U.S. EPA's SPECIATE4.3 database. In order to improve this PM size distribution profile library, a comprehensive literature review was conducted: over 100 relevant publications were identified and 30 source-specific PM size distribution profiles were selected and compiled. These 30 PM size distribution profiles were then combined based on process type with corresponding PM speciation profiles to create a SMOKE-ready table of chemically-speciated and size-disaggregated source-specific PM disaggregation profiles. This table can now be used by SMOKE for 12-bin emissions processing for GEM-MACH to perform PM chemical speciation and size allocation in a single step. Details of the compilation of the 30 PM size distribution profiles will be discussed in this presentation. Differences in processed PM emissions based on the current and updated PM size distribution profile library will also be shown.

Lightning Assimilation in WRF4.0.2: Impact of Parameter Options and Introduction of New Lightning Data

Daiwen Kang, Robert Gilliam, Jerry Herwehe, and Jon Pleim

Lightning assimilation has been proven to be effective in improving atmospheric convection simulations in the Weather Research and Forecasting (WRF) model. A few parameter options are associated with the Kain-Fritch (KF) convective scheme in the WRF model: kfeta_trigger -controls how convections are trigged with values of 0 (default) and 1 (moisture-advection modulated trigger function) and cudt - minutes between cumulus physics calls (for example, cudt = 10, 10 minutes, and cudt = 0, call every time step). Different combinations of these parameter options are recommended with/without lightning assimilation for WRF simulations. In this model exercise, the possible combinations of these parameter options are applied to the WRFv4.0.2 simulations with/without lightning assimilation and the impact on 2-m temperature, water vapor mixing ratio, wind speed, and wind direction is evaluated against ground observations. Precipitation is assessed against the PRISM (Parameter elevation Regression on Independent Slopes Model) product that is ingested from in-situ point measurement. In addition to using lightning flash data from the National Lightning Detection Network (NLDN), the lightning data from the World Wide Lightning Location Network (WWLLN) is introduced for the first time to perform lightning assimilation in the WRF model. The preliminary assessment of using the new data source for lightning assimilation will be presented.

The estimated impacts of volatile chemical products on particulate matter and ozone criteria pollutants in an urban atmosphere

Momei Qin(1,), Benjamin N. Murphy(1), Brian C. McDonald(2,3), Stuart A. McKeen(2,3), Lauren Koval(1), Kristin K. Isaacs(1), Quanyang Lu(4,5), Allen L. Robinson(4,5), Madeleine Strum(6), Jennifer Snyder(6), Christos Efstathiou(7), Chris Allen(7), Havala O.T. Pye(1)

(1) Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA

(2) Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA

(3) Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA

(4) Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

(5) Center for Atmospheric Particle Studies, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

(6) Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA

(7) General Dynamics Information Technology Research Triangle Park, North Carolina, USA

() Present address: School of Environmental Science and Engineering, Nanjing University for Information Science & Technology, Nanjing, China

Volatile chemical product (VCP) usage, including application of personal care products, paints, and adhesives, results in the release of volatile organic compounds (VOCs) to the ambient atmosphere. These VOCs oxidize and contribute to ozone and PM2.5 formation and are expected to be an increasing fraction of urban VOCs as combustion emissions are controlled. We find that CMAQ v5.3 with emissions based on the National Emission Inventory (NEI) underestimates daily-average PM2.5 and daily maximum 8-hour averaged (MDA8) ozone by 0.9 g m-3 and 6 ppb respectively for Los Angeles summer 2010 conditions. A factor of 3 higher VCP emissions than currently represented in the NEI is supported by recent literature and new simulations with the Stochastic Human Exposure and Dose Simulation Model (SHEDS) developed for high-throughput (HT) assessment of near-field exposure. These higher magnitude emissions drive higher ozone formation predictions. In addition, we find a factor of 1.4-2.9 times higher secondary organic aerosol (SOA) yields from VCPs compared to current CMAQv5.3 values are needed to reconcile observed and modeled SOA in Los Angeles. In this work, we estimate VCP usage accounts for ~41% of prompt photochemical SOA and ~17% of MDA8 O3 in observations from Los Angeles, making them important precursors to criteria pollutant formation.

Comparing Extreme Weather Events Generated by 36-km and 12-km WRF Simulations

T. L. Spero, J. H. Bowden, A. M. Jalowska, M. S. Mallard, and G. M. Gray

Extreme weather events, such as heat waves, drought, flooding, and tropical cyclones, can have catastrophic impacts on society. These extreme events can adversely affect the economy, infrastructure, ecosystems, agriculture, transportation, air quality, and human health. Governing organizations typically account for extreme events to some degree in their planning processes to minimize the impacts of extreme events.

There is a growing consensus in the literature that extreme events are likely to intensify through this century. Consequently, more emphasis has been placed on quantifying the changes to the frequency and intensity of these events. Several modeling techniques have been used to project potential changes to weather and climate across the next century, and there are several public data sets available that are currently used by federal, state, and local agencies to aid in the decision-making process. However, not all data sets can characterize the extremes associated with these events.

In this study, the Weather Research and Forecasting (WRF) model is configured as a regional climate model, and simulations are conducted on historical (verifiable) data sets to emulate the dynamical downscaling procedure that would be applied to refine global climate model projections. Here, we use 36-km and 12-km modeling domains, and we compare categories of extreme weather events that would influence governmental planning organizations. We evaluate these WRF simulations against observations to demonstrate the trade-offs between the computational expense of 12-km modeling domain versus developing a broader ensemble at 36-km on extreme event realization.

Initial Development of a NOAA Emissions and eXchange Unified System (NEXUS)

Patrick Campbell, Barry Baker, Rick Saylor, Daniel Tong, Youhua Tang, Pius Lee, Stuart McKeen, Gregory Frost, and Christoph Keller

The past decade has experienced rapid advances in global aerosols and atmospheric composition (AAC) model prediction capabilities. AAC models are key components of unified forecast systems that often employ an Earth System Model Framework (ESMF; i.e., a high-performance, flexible software infrastructure for building and coupling weather, climate, and related Earth science models) for weather and climate predictions. Emissions of trace gases and primary aerosols are a critical component of AAC models and are often the most important component to ensure accurate predictions of trace species distributions. However, developing these emissions inputs to AAC models is often a laborious, time-consuming process, especially to ensure that the datasets are suitable for a range of spatial scales and applications. Furthermore, inventory-based emission inputs are subject to a bottom-up approach that is prepared separately (offline) and suffers distinct time lags from the AAC models, which affects both the timing and accuracy of trace gas predictions. In this work, the Harvard-NASA Emission Component (HEMCO) is serving as the foundation of a new unified emissions modeling framework, which is already capable of utilizing numerous emissions datasets (both global and regional), can be run offline (inventory-based) or online (processed-based), is ESMF-compliant, and can be easily linked to satellite data sources. Here we present the initial development the NOAA Emissions and eXchange Unified System (NEXUS), which will interface with different NOAA AAC models, including both global and regional models for both operational and research-oriented applications. Preliminary developments of a comprehensive, adaptable emissions and air-surface exchange processing system for use in conjunction with NOAA AAC models will be shown. This includes examples of model-ready anthropogenic emissions using a combination of global and regional anthropogenic emission inventories with the NEXUS platform, and an initial assessment of NOAA AAC model simulations using these emissions. We will also present the first steps toward implementation of advanced inline dust and fire emissions using new NOAA products and emission models, as well as demonstrate the potential for combining ammonia emission inventories with satellite ammonia measurements and inline agroecosystem processes, to support a novel air-surface exchange model for ammonia fluxes on regional to global scales. Ideas on adapting NEXUS to a wider community of AAC models outside of NOAA will also be demonstrated.

Particulate matter sensitivity to local emissions and meteorology over a Latin American megacity for source apportionment and uncertainty analysis

James East, Jorge Pachon, Juan Montealegre, and Fernando Garcia Menendez

Bogota, Colombia, a megacity in Latin America, regularly experiences exceedances of particulate matter air quality standards, resulting in negative impacts on public health. In recent years, significant advances in air quality modeling research have aided air quality and health studies in the region. However, uncertainty stemming from emissions data, structural model uncertainty, and meteorological drivers limits model performance and the ability to use model results to inform environmental policy. Uncertainty in meteorological and emissions inputs are of great concern due to complex topography around the city and the relevance of emissions for policy. Here we identify the most influential meteorological and emissions parameters in simulated air quality over the city by evaluating the sensitivity of modeled particulate matter concentrations (PM_2.5 and PM₁₀) to variations in input data, addressing uncertainty from input data fields. We also assess the sensitivity of modeled PM_2.5 and PM₁₀ to changes in model parameterizations, addressing structural uncertainty. We use CMAQ to perform simulations of a 2-week air pollution episode in Bogota under varying emissions and meteorological drivers. Meteorological inputs are varied across multiple physical schemes available in the WRF model and evaluated for air quality modeling in Bogota. Emissions sources are scaled to quantify the contribution of each source to the total modeled PM burden and at 13 observation sites across the city. Our results reveal the model inputs with the largest influence on predicted concentrations and largest potential contributions to uncertainty, a finding which can guide future research efforts. Emissions sensitivities used for policy-relevant source apportionment show that for Bogota resuspended dust dominates PM concentrations and that on-road emissions are the next highest contributor. In addition, PM sensitivity to changes in model parameterizations suggest that addressing structural uncertainty can improve model performance.

Towards Refining Estimates of Ammonia Emissions: Modeling Framework Preparations

Mahmoudreza Momeni, Drexel University, Civil, Architectural, and Environmental Engineering, Philadelphia, Pennsylvania, USA

Shannon Capps, Drexel University, Civil, Architectural, and Environmental Engineering, Philadelphia, PA, USA, shannon.capps@drexel.edu

Shunliu Zhao, Carleton University, Civil and Environmental Engineering, Ottawa, Ontario, Canada

Amir Hakami, Carleton University, Civil and Environmental Engineering, Ottawa, Ontario, Canada

Daven Henze, University of Colorado, Mechanical Engineering, Boulder, Colorado, USA

Steven Thomas, University of Melbourne, School of Earth Science, Melbourne, Victoria, Australia

Jeremy Silver, University of Melbourne, School of Earth Science, Melbourne, Victoria, Australia

Peter Rayner, University of Melbourne, School of Earth Science, Melbourne, Victoria, Australia

CMAQ Adjoint Development Team

Ammonia (NH3) has a critical role to play in forming fine inorganic particulate matter (PM2.5) in the atmosphere. NH3 also affects the nitrogen cycle and climate change. Uncertainty in NH3 emissions is propagated into secondary PM2.5 simulated by model, which limits the precision with which issues related to inorganic PM2.5 may be addressed with models. One potential aid in revising emissions is to assimilate observations from satellites. In this study, a Python-based four-dimensional variational assimilation (py4dvar) framework integrated with the Community Multiscale Air Quality (CMAQ) Model and its adjoint is tested. First, the adjoint-based sensitivities of concentration with respect to emissions are evaluated against sensitivities calculated from the forward model using the finite difference method. Then, pseudo-observation tests with NH3 treated as a tracer are conducted to evaluate how much of a perturbation in emissions can be recovered with perfect observations. This work is a necessary, preliminary step to assimilating observations of NH3 recently made available.

Model-Measurement Comparison of Ozone and Precursors Along Land-Water Interfaces during the 2017 LMOS Field Campaign

Liljegren, J; Baker, K.; Valin, L; Szykman, J; Henderson, B.; Judd, L.; Al-Saadi, J; Janz, S.; Sareen, N.; Possiel, N

Several counties in the Lake Michigan region are currently exceeding the level of the Ozone National Ambient Air Quality Standards (O3 NAAQS). Precursor emissions from various source sectors (e.g., mobile, power plants, biogenic) and complex land-water meteorology lead to episodes of elevated O3 in late spring and early summer in this region. It is important to understand how well photochemical transport models represent regional emissions, meteorology, and chemistry to determine which emission control scenarios are most effective at improving air quality. Highly instrumented field studies provide a unique opportunity to evaluate multiple aspects of photochemical grid model representation of emissions, dispersion, and chemical evolution. Routine surface measurements coupled with airborne and remotely sensed measurements from the 2017 Lake Michigan Ozone Study (LMOS) provide information needed to constrain model predicted O3 formation, transport, and chemical evolution. Further, NOX and speciated VOC measurements provide a unique opportunity to better understand how well certain source sectors are being characterized in the modeling system. The Community Multiscale Air Quality (CMAQ) model was applied at 12 and 4 km grid resolution for the Lake Michigan region with source attribution for major emissions sectors. Surface measurements of O3, NOx, and speciated VOC including formaldehyde and aircraft measurements of O3 and NO2 over the lake and lakeshore were matched with model predictions. Sub-orbital remotely sensed NO2 column measurements from the field campaign provide a unique opportunity to evaluate spatial NO2 patterns of emissions in the Chicago urban core and ground-based NO2 column measurements made with PANDORAS provide high time resolution temporal constraints. In this talk we will present the results of our comparison of model predictions to the corresponding observations at the surface and aloft. We will also present the contributions from source sectors to ozone and precursor model-predicted concentrations.

Inter-comparison of Mobile Source Emissions from the CARS and CAPSS in Seoul Metropolitan Area, South Korea
Minwoo Park1, Jung-Hun Woo1*, Younha Kim1, Bok Haeng Baek2, Jinseok Kim1, Jinsu Kim1, Youjung Jang1, Rizzieri Pedruzzi2

1 Konkuk University, Seoul, Korea
2 University of North Carolina, Chapel Hill, USA

Air quality is getting worse because of the increase in population and energy in East Asia. In the case of a megacity like Seoul, Korea, which has a high density of population, air pollution is very serious due to anthropogenic emissions derived from human activities. In the Seoul Metropolitan Area (SMA), emissions from mobile source are account for 54% of total NOx emissions and 20% of total PM2.5 emissions. Moreover, in the SMA, the proportion of secondary fine particle generated by chemical reactions accounts for more than 60% of the total PM2.5 emissions. A new Korean Air Quality Model (KAQM) modeling system is being developed to improve ability to estimate and forecast fine particle concentrations over a complex terrain region like Korea. A new mobile source emissions model named Comprehensive Automobile emissions Research Modeling System (CARS) has been developed as a part of KAQM modeling system. The government of South Korea is making its tremendous efforts to improve air quality by applying policies to control emissions from mobile emission sources. The current Korean National Emissions Inventory (NEI) so called Clean Air Policy Supporting System (CAPSS) estimates mobile emission using static emission factor-based method originated from the European Environment Agency's (EEA) COPERT (COmputer Program to calculate Emissions from Road Transport) methodology. It is useful to estimate the total amount of mobile sources emissions by administrative unit level, using the total VKT and averaged vehicle speeds. However, there are limitations in CAPSS to consider dynamic change of emission distributions due to traffic behavior changes. Also, fast applications of emissions reduction from the control measures policy are limited due to its static emissions procedures. The CARS has the advantage of being more responsive to the applying traffic condition change because it is the county and link-based model. It is designed to perform effectively to apply recent policy measures because it is based on the real inspection data such as age of vehicle, daily VKT and county information from the Vehicle Safety institute of Korea. In this study, we inter-compare the mobile source emissions from CARS and CAPSS and performs air quality modeling to understand the benefits of the newly developed emissions model. Further estimation results and follow-up analysis will be presented at the conference.

ACKNOWLEDGEMENT

This research was supported by the National Strategic Project-Fine particle of the National Research Foundation of Korea(NRF) funded by the Ministry of Science and ICT(MSIT), the Ministry of Environment(ME), and the Ministry of Health and Welfare(MOHW) (NRF-2017M3D8A1092022). This work was supported by a grant from the National Institute of Environment Research (NIER), funded by the Ministry of Environment (MOE) of the Republic of Korea (NIER-2019-01-02-037).

* Correspondence : jwoo@konkuk.ac.kr

Application of ozone source apportionment using CMAQ-ISAM during LISTOS

Qian Shu, K.R. Baker, S.L. Napelenok, J. Szykman, L. Valin, T. Plessel

Ozone levels have exceeded the National Ambient Air Quality Standard (NAAQS) in the New York City (NYC) metropolitan area for many decades, which afflicts the health and well-being of millions of people living in the NYC metro area and downwind in Connecticut, Rhode Island, Massachusetts, and beyond. A key problem in addressing this regional ozone pollution is comprehensively understanding multiple emission source sector contributions to ozone production and downwind transport. The Integrated Source Apportionment Method (ISAM) ozone approach has been developed and implemented in the Community Multiscale Air Quality (CMAQ) model to characterize and quantify the relationship between emission sources and ozone concentrations in regional areas. The updated ISAM in the newest version CMAQ (v5.3) is more efficient to conduct either short- or long-term simulations for small or large domains. In this study, we apply the newest CMAQ-ISAM to investigate ozone and its precursors local to regional transport and source contribution features during the Long Island Sound Tropospheric Ozone Study (LISTOS). We first conduct annual 2018 CMAQ simulation in a 12km contiguous US platform for comparison to annual average satellite observations for feature characterization. We then refine CMAQ-ISAM simulation for a heavy ozone pollution episode in summer 2018 with a 4 km platform to compare with field study observations. We expect to demonstrate the capability of CMAQ-ISAM to comprehensively understand contribution to precursors and O3 production over the NYC area to inform optimal emissions control strategies.

Impact of anthropogenic emissions on urban air quality over the East African big conurbations of Addis Ababa, Kampala and Nairobi.

Mazzeo Andrea 1 3, Quinn Andrew 1, Burrow Michael 1, Marais Eloise 2, Singh Ajit 3, and Pope Francis 3

1 School of Civil Engineering, University of Birmingham, UK

2 Department of Physics, Earth Observations and Astronomy, University of Leicester, UK

3 School of Geography Earth and Environment Sciences, University of Birmingham, UK

Sub Saharan East Africa (SSEA), since 1990, is experiencing a population growth accompanied by an increase in industrial activities, private and public vehicle utilization and use of solid fuels for cooking and heating (WPP, 2015). This growth has consequently contributed to the worsening of urban air quality (NUA, 2017).

The absence of infrastructures for air quality monitoring and regulations relating to the mitigation of atmospheric emissions in many countries of SSEA contribute to keep contamination levels in East African urban areas unknown as well as the real extent of the problem of air pollution in the different urban districts of SSEA cities.

In this scenario, the ASAP project (ASAP, 2018) aims to bring together leading UK and East African researchers in air pollution, urban planning, economic geography, public health, social sciences, engineers and development studies to provide a framework for improved air quality management in three East African cities: Addis Ababa (Ethiopia), Kampala (Uganda) and Nairobi (Kenya).The research program, split into seven complementary work packages (WPs), employ a combination of multidisciplinary methodologies to study the East African cities as integrated systems investigating the complex theme of urban air quality.

The WP-4 of the ASAP project explores the air pollution levels of SSEA using a modelling system at high spatial resolution (up to 2x2km) for meteorology and atmospheric chemistry processes.The Weather Research and Forecast model (WRFv3.9, Skamarock et al., 2008) has been coupled with the chemistry-transport model (CTM) CHIMERE (version 2017, Menut et al., 2013; Mailler et al., 2017) to simulate the main meteorological and aerosols dispersion patterns over the domains of Addis Ababa, Kampala and Nairobi.

A new merged emission inventory developed for the purpose of the project, has been tested on the three domains. The most up-to-date and available global emission inventory, EDGARv4.3.2 (Crippa et al., 2018) has been merged with Diffuse and Inefficient Combustion Emissions in Africa inventory (DICE-Africa, Marais et al., 2016), both created originally for the year 2012. The final inventory (DICE-EDGAR) has been projected for the year 2017 by linear extrapolation of population density data from the Socioeconomic Data and Application Centre of the Columbia University and NASA (SEDAC, 2019).

The validation of the modelling system has been done for a period of 30 days - since the 14^th of February to the 15^th of March 2017 - using meteorological observations from MIDAS database (Met Office, 2012) from several regional sites of each country and particulate matter (PM₁₀ and PM_2.5) observations from the ASAP WP-2 field sampling in Nairobi (Pope et al, 2018) and from the U.S. Embassies of Addis Ababa and Kampala.

Results show that the modelling system is able to describe the main regional and local meteorological patterns at different altitudes as well as to simulate PM₁₀ and PM_2.5 levels in all three urban sites. High levels of air pollution found in the three domains have been compared with WHO limits (WHO, 2006) and Air Quality Index (EEA, 2018). This highlights for a serious risk for citizens health in several districts of each city taken in account and in the natural outskirts, calling for concrete actions to drastically reduce the urban contamination through appropriate mitigation policies. Finally, scenarios testing reduced emissions from road transport and wood burning sectors show effective reduction in the air contamination at city level.

Ramboll Shair: Integrating real-time sensor measurements and regional/local-scale models in Richmond, California

Justin Bandoro, Tasko Olevski, Kurt Richman, Drew Hill, Mike Dvorak, Julia Luongo, Shari Beth Libicki, Chris Emery, Greg Yarwood

Low-cost air quality sensor networks are streaming data at both high temporal and spatial resolution. Simultaneously, cities are tracking more activity data that are related to air quality such as link-level traffic and congestion, port activity, and continuous emissions monitoring from stationary sources. With the growing abundance of data, there is pressing demand for better ways to synthesize all of this information, pull meaning from data, and present air quality insights to the public that are relevant to daily living. Ramboll developed a system that combines forward deterministic models executed in real-time with sensor measurements at 50 locations in Richmond, California to produce detailed air quality maps with source apportionment. The model, named Ramboll Shair, combines the Comprehensive Air Quality Model with Extensions (CAMx) grid model with a roadway dispersion model, ShairStreet, to compute PM2.5 and NO2 concentrations with source apportionment at every hour. ShairStreet was developed to resolve sub grid-level PM2.5 and NO2 concentrations from road traffic emissions by accounting for the effects of street canyon geometry, near-roadway dispersion of pollutants, and NO-NO2-O3 chemistry . The outputs of ShairStreet and CAMx are fused together to map urban PM2.5 and NO2 concentrations at a high resolution (currently ~10-meter) with source attribution. Data from the network of low-cost air quality sensors is fed into a geostatistical fusion methodology to merge information from the sensor network with model output and produce best-fit real-time maps of PM2.5 and NO2 that fill in information deficits of the model. By tracking trends in where bias is reduced between predictions and observations, we aim to refine source apportionment estimates over time. By using real-time data to model every hour, there is potential to identify unexpected pollution levels from observations that the deterministic models did not expect from the available activity data.

Incorporating Isotope into Atmospheric Chemistry Models

Huan Fang, Greg Michalski, Scott Spak

Accurately constraining N emissions in space and time has been a challenge for atmospheric scientists. It has been suggested that ¹⁵N isotopes (¹⁵N) may be a way of tracking N emission sources across various spatial and temporal scales. However, the complexity of multiple N sources that can quickly change in intensity has made this a difficult problem. ¹⁵N was incorporated into SMOKE emission model to test how emission sources would impact the ¹⁵N value of NO_x. ¹⁵N was incorporated into CMAQ but excluding its chemical module, to explore how atmospheric processes would alter the ¹⁵N of atmospheric NO_x. 15N was incorporated into RACM, one of the chemical mechanisms in CMAQ, to analyze how tropospheric photochemistry alter the ¹⁵N of atmospheric NO_x.

Air Quality Impacts of Oil and Gas Emissions in the United States

Saravanan Arunachalam, Srinivas Reka, Dongmei Yang, Institute for the Environment, The University of North Carolina at Chapel Hill, NC, USA

David Lyon, Hillary Hull, Ryan O Connell, Ananya Roy, Beth Trask, Environmental Defense Fund, Washington DC, USA

Jonathan Buonocore, Harvard TH Chan School of Public Health, Boston, MA, USA

Emissions from oil and gas production are presently a major and underappreciated contributor to air pollution. The 2014 U.S. EPA's National Emissions Inventory (NEI) indicates that oil and gas production was the largest anthropogenic source of VOCs, and the 6th highest and 7th highest anthropogenic emitter of SO2 and NOx respectively. However, limited efforts have examined the contribution of shale oil and gas extraction on local-to-regional scale air pollution and associated health impacts. We use the WRF-SMOKE-CMAQ modeling framework to assess the contribution of individual pollutant emissions from this sector in the continental U.S. at a spatial resolution of 12x12-km in 2016. Based on a recent study that quantified underestimates in nation-wide methane emissions, we adjusted the NEI-based VOC estimates from this source sector. We use CMAQ v5.2 instrumented with the Decoupled-Direct Method (DDM) - an advanced method to assess 1st order sensitivities of ozone (O3) and fine particulate matter (PM2.5) due to precursor emissions from both conventional and unconventional oil and gas activities. We configured the model to tag specific combinations of oil and gas emissions source regions and individual precursors from different activities (production, processing, transportation, storage and compressors) as individual sensitivity parameters. Recent updates to the CMAQ model have included formation of secondary organic aerosol (SOA) from both anthropogenic and biogenic VOCs. Specifically, long-chain alkanes (C6 - C20) SOA are predicted to be responsible for ~30% of SOA from anthropogenic VOCs with the largest absolute concentrations during summer in urban areas. Our analyses will thus be able to effectively quantify the air quality impacts of VOCs through both O3 and PM2.5 formation. We will present results from this study with a specific focus on air quality impacts in downwind regions from select individual or groups of shale plays in the U.S. for current and future year emissions scenarios.

Source Apportionment Modeling to Investigate Local and Non-Local Contributions to Ground-Level Ozone in Albuquerque, New Mexico

Kenneth Craig, Garnet Erdakos, Shih Ying Chang, Lynn Baringer

Sonoma Technology, Inc., Petaluma, CA

Ozone design values in Albuquerque, New Mexico, have decreased over the last 15 years, but have increased since 2016; the current ozone design value for Albuquerque (70 ppb) is on the cusp of exceeding the 2015 National Ambient Air Quality Standard (NAAQS). To support the Albuquerque Environmental Health Department with its air quality planning, a source apportionment modeling analysis was conducted for two high-ozone episodes in 2017 to develop a day- and episode-specific understanding of the effects of local emissions and meteorology, long-range pollutant transport, and wildland fires on ozone concentrations in Albuquerque. The modeling analysis was conducted with CAMx version 6.4 using ozone source apportionment technology for three nested domains, including a 4-km domain covering New Mexico. Meteorological fields were developed using the Weather Research and Forecast (WRF) model (version 3.9.1). Emissions were based on the U.S. Environmental Protection Agency's (EPA) 2014 version 7.2 emissions modeling platform, with hourly 2017 continuous emission monitoring system (CEMS) data for electrical generating unit (EGU) emissions, and day-specific 2017 fire emissions from the BlueSky Framework based on satellite hotspot and fire perimeter data. On-road mobile source emissions were projected to 2017 levels using scaling factors developed from MOVES to account for changes in vehicle activity, fleet turnover, and emission factors, with separate factors developed for Bernalillo County based on local fleet data.

A source tagging strategy was developed to evaluate the role of local and non-local emissions, specific emissions source sectors, and selected individual emissions sources on ozone concentrations in Albuquerque. The modeling tracked ozone contributions from nine source regions; boundary conditions; and five emissions source groups: (1) biogenic emissions, (2) on-road mobile sources, (3) wildland fire, (4) major EGUs in New Mexico, and (5) other anthropogenic sources. A separate sensitivity simulation was conducted to quantify potential impacts from oil and gas emissions throughout New Mexico.

The modeling analysis showed that high ozone concentrations in Albuquerque during the June 2017 ozone episode were largely driven by non-local emissions from outside the Albuquerque area, while the high ozone concentrations during the July 2017 episode were driven more strongly by local emissions from within Albuquerque. Anthropogenic emissions from Albuquerque and Bernalillo County contributed between 5 and 16 ppb of ozone in Albuquerque on high ozone days, depending on the day and episode. Half of the ozone generated by local emissions was from on-road mobile sources, but contributions from nonroad and non-mobile source sectors were also significant and are becoming increasingly important as emissions from on-road mobile sources continue to decrease. Ozone contributions from major EGUs in New Mexico (outside of Bernalillo County) were less than 0.5 ppb on most high-ozone days. Ozone contributions from wildfire emissions were as large as 2 ppb in Albuquerque on any given day, but as large as 20 ppb elsewhere in New Mexico. These results have important implications for air quality planning in the Albuquerque area.

Examining the air pollutant emission implications of electric vehicle market penetration scenarios
Dan Loughlin and Chris Nolte
U.S. EPA Office of Research and Development

Samaneh Babaee and Yang Ou
Oak Ridge Institute for Science and Education program participants

Aaron Sobel, Michael Shell, Chris Ramig, Susan Burke, and Meredith Cleveland
U.S. EPA Office of Transportation and Air Quality

Ryan Sims
U.S. EPA Office of Atmospheric Programs

In 2018, electric vehicle (EV) sales had grown to over 1.2% of new U.S. light-duty vehicle market. That percentage is predicted to increase dramatically over the coming decades. For example, the 2019 Annual Energy Outlook estimates that EVs will constitute nearly 19% of sales in 2050, and some other projections predict much greater EV sales. If such projections are realized, there could be implications for air quality. For example, while EVs would lead to reduced transportation emissions, there would also be accompanying emission changes in the electric sector and refineries, as well as in fuel extraction and transport. Understanding the overall emissions impacts of EVs will be important, particularly for areas that are currently in or near nonattainment of the air quality standards or that are considering incentivizing EV sales to meet climate and air quality goals. In this poster, we will discuss an emerging modeling platform for evaluating the emission impacts of EV scenarios. The Global Change Assessment Model with state-level resolution (GCAM-USA) includes representations of the energy, industry, buildings, and agricultural systems, simulating how those systems evolve and interact over the coming decades. Outputs of the model include technology and fuel choices, fuel prices, and emissions. We demonstrate the use of GCAM-USA to examine three different EV market penetration scenarios and evaluate the net emissions implications of each. We also discuss future directions, including linkage of GCAM-USA with a "load tool" for examining alternative charging profiles and with the Integrated Planning Model (IPM) for examining power sector dynamics in more detail.

Mutual comparison of source sensitivities and apportionments obtained by BFM, DDM, and ISAM on PM2.5 and ozone concentrations over Japan

Satoru Chatani¹, Hikari Shimadera², Syuichi Itahashi³, and Kazuyo Yamaji⁴

¹ National Institute for Environmental Studies

² Osaka University

³ Central Research Institute of Electric Power Industry

⁴ Kobe University

It is important to appropriately interpret differences between source sensitivities and apportionments for considering effective strategies aiming at better air quality. This study applied BFM, DDM, and ISAM of CMAQ version 5.0.2 to calculate source sensitivities and apportionments on PM_2.5 and ozone concentrations over Japan. The sum of the apportionments of all the sources obtained by ISAM coincided with the simulated concentrations in accordance with its principle. The sum of the sensitivities of all the sources obtained by BFM and DDM exceeded the simulated PM_2.5 concentrations by approximately 20% due to nonlinearities. Features of the differences were evident in PM_2.5 components. The source sensitivities and apportionments were identical for EC and primary OC. Significant differences appeared in secondary components, particularly NO₃^- and NH₄⁺. Mutual dependence on NH₃ and HNO₃ in NH₄NO₃ formation resulted in sensitivities of NH₃ sources on NO₃^- concentrations and those of NO_X sources on NH₄⁺ concentrations, which never appeared in the source apportionments. The source sensitivities and apportionments on ozone concentrations were overwhelmed by the long-range transport. Whereas the source sensitivities and apportionments of VOC sources were similar, the sensitivities of NO_X sources, whose apportionments were always positive, were mostly negative on average concentrations due to titration.

Impact of future electrification of passenger cars on air quality within the United States

Abi Lawal

Jooyong Lee

Yilin Chen

Huizhong Shen

Kara M Kockelman and Armistead G Russell.

Self-driving vehicles are expected to have a majority market share (> 60%) if not full share by 2050. For reasons that range from engineering practically, emissions standards and government policies such as tax incentives, these cars are wholly expected to be electric and will allow for more shared vehicle miles being driven (i.e Uber). The impact of this projection is not only expected to change vehicle ownership in households (which will go down) but could influence dynamic ride sharing fleets (DRS), giving vehicle access to different social economic groups that may otherwise not have access to such vehicles due to cost. In addition to changing the DRS market, automation of vehicles is projected also expected to increase the number of vehicle miles traveled as well, which are already projected to increase annually. As a result, the combination of electric cars in addition to vehicle miles traveled is expected to have a significant impact on emissions.

Assessing the Impacts of Emissions from Oil and Gas Extraction on Urban Ozone and Associated Health Risks

Congmeng Lyu, Civil, Architectural, and Environmental Engineering Department, Drexel University, Philadelphia, PA, USA

Shannon Capps, Civil, Architectural, and Environmental Engineering Department, Drexel University, Philadelphia, PA, USA

Daven Henze, Mechanical Engineering Department, University of Colorado, Boulder, Colorado, USA

Amir Hakami, Department of Civil and Environmental Engineering, Carleton University, Ottawa, Ontario, Canada

Shunliu Zhao, Department of Civil and Environmental Engineering, Carleton University, Ottawa, Ontario, Canada

Rene Nsanzineza, Mechanical Engineering Department, University of Colorado, Boulder, Colorado, USA

Jana B. Milford, Mechanical Engineering Department, University of Colorado, Boulder, Colorado, USA

Natural gas and oil exploration and production processes emit gases that contribute to tropospheric ozone formation, which negatively impacts human health and public welfare. Attaining the ozone National Ambient Air Quality Standard (NAAQS) has been challenging for some urban regions, including Colorado Front Range adjacent to oil and gas development. For these locations, understanding the relative contribution of emissions from oil and gas activities is important to evaluating emissions control strategies. In this investigation, we elucidate the influences of oxides of nitrogen (NOx) and volatile organic compound (VOC) emissions on these urban areas, and estimate the contribution of recent oil and natural gas activities to ozone exceedances and ozone-related premature mortality.

Specifically, we use the adjoint of CMAQ to calculate the sensitivity of urban ozone concentrations and NAAQS exceedances and of ozone-related health risks to emissions of precursor gases. The adjoint efficiently determines these relationships for each emitted species. We compare and contrast the results from the adjoint to positive matrix factorization (PMF) modeling results and chemical mass balance (CMB) modeling results.

Optimal use of grid-connected energy storage to reduce human health impacts

Qian Luo, Jeremiah Johnson, Fernando Garcia Menendez

Grid-connected energy storage can perform a variety of applications, yielding benefits to power system operations and costs. Current applications for energy storage, however, do not explicitly consider its potential to reduce adverse human health impacts from power generation. In this study, by taking advantage of energy storage's ability to shift both the time and location of power sector emissions based on their charging and discharging strategies, we propose a method that enables energy storage to cost-effectively reduce human health impacts from the power sector. To do this, we use dispersion modeling to determine the hourly health damage cost for each electricity generating unit. We then internalize these health damage costs in power plant dispatch decisions, re-optimizing the unit commitment and economic dispatch in light of these costs. We introduce two factors, energy storage, and health damage cost, and our preliminary results show that both can contribute to a health impact reduction: internalizing the time- and location- varying health damage costs into the unit commitment and economic dispatch model helps reduce the adverse health impacts from electricity generation and the addition of energy storage to the grid can help reduce additional health impacts when considering health damage costs, while also reducing the electricity generation costs.

Assessing Air Quality Impact on Non-attainment Regions in Ohio Resulting From Power Plant Closures and Shale Gas Activity

Saikat Ghosh, Kevin Crist

The shift from coal to natural gas for power generation has been accelerating over the last several years. While natural gas use for electric generation is expected to improve air quality, unconventional oil and gas production, involving a wide distribution of emission sources, can impact local and regional air quality. For Ohio this contrast is highlighted with the oil and gas development of the Marcellus shale and the closing/fuel switching of coal fired power plants occurring throughout the state. The current work presented here involved modeling the net impact of reduced coal power generation and the increased in emissions associated with unconventional oil and gas production shale on the regional ozone levels in Ohio. 2011 based model simulations were performed using EPA's Community Multiscale Air Quality (CMAQ v5.2) with updated Carbon Bond 6 (CB6r3) gas phase chemistry. 2011 base year model performance was evaluated for all the ozone monitors in Ohio. In addition, sensitivity simulations were performed with predictions of 2023 emissions of shale gas exploration and electricity generation obtained from USEPA's 2011v6.3 inventory. Simulations showed a 0.06 to 0.22 ppb increase in 8-hour average ozone due to oil and gas production in Ohio. However, the 8-hour average ozone significantly reduced by 4.8 ppb when combined with power plant closures forecasted for 2023. Calculation of ozone design values at the monitors also showed a maximum decrease of 4 ppb.

Impact of using the updated detailed-level marine emissions on Canadian Air Qualit

Rabab Mashayekhi1, Mourad Sassi1, Calin Zaganescu1 and Radenko Pavlovic1

1 Canadian Meteorological Center, Environment and Climate Change Canada, Montreal, Quebec, Canada

The Air Quality Policy-Issue Response section (REQA), within the Meteorological Service of Canada (Environment and Climate Change Canada ECCC) is responsible for preparing and distributing model-formatted emissions inventories based on the latest version of Canadian Air Pollutant Emission Inventory (APEI). As one of the most recent update, REQA processed a link based detailed-level marine emissions for the year 2015. These emissions are based on the Canadian national Marine Emissions Inventory Tool (MEIT) that is a bottom-up, activity-based inventory for all types of marine vessels operating in Canadian waters. The new marine emissions inventory contains hourly-based emissions for all marine regions of Canada down to a 1-km grid resolution. This high spatial-temporal resolution inventory is used to develop a set of detailed-level, better representative surrogates and temporal profiles for marine emissions processing in SMOKE. This work presents the impact of using the updated version of marine emissions on air quality in Canada. That includes a review of the methodology for processing model-ready marine emissions as point sources with stack parameters versus the gridded 10-km emissions compiled from provincial area sources and allocated using the up-to-date ship type-based surrogates. The impact of using each set of emissions on air quality simulations by GEM-MACH (Global Environmental Multi-scale Modeling Air Quality and Chemistry) will be discussed.

Relaxing energy policies on top of climate change will significantly undermine states efforts to attain U.S. ozone standards

Huizhong Shen, Yilin Chen, Yufei Li, Armistead G. Russell, Yongtao Hu, Lucas R. F. Henneman, Mehmet Talat Odman, Jhih-Shyang Shih, Dallas Burtraw, Shuai Shao, Haofei Yu, Momei Qin, Zhihong Chen, Abiola S. Lawal, Gertrude K. Pavur, Marilyn A. Brown, Charles T. Driscoll

The U.S. government has recently sought to relax energy policies (EPs), which is expected to increase emissions of not only greenhouse gases but also conventional air pollutants, resulting in a deterioration of local air quality, including ozone (O3) pollution. Such relaxation in EPs offsets efforts to attain gradually tightened U.S. O3 standards, potentially increasing costs of up to several billion dollars for abatement. Unfortunately, the complex long-standing impacts of EPs on the U.S. energy system coupled with the uncertainty associated with future climate change challenges understanding of the effects of these policy changes on O3 pollution. Here we apply an integrated modeling framework to show that relaxation of EPs coupled with climate change will increase the number of U.S. counties in nonattainment for O3 (NNA) by more than three-fourths in the 2050s. The effect of EP relaxation on NNA is projected to be magnified by up to 300% under a changing climate. This magnification due to climate change is the result of vast enhancement of O3 production efficiency associated with warming, whereby the enhanced OPE makes the increased NOX emissions from EP relaxation produce O3 more efficiently. Our study suggests that segregate pathways linking EPs to local pollutant emissions and to global climate would significantly and synergistically leverage O3 pollution if the future world was governed by relaxed EPs.

Air pollution control strategies directly limiting national health damages in the U.S.

Yang Ou, J. Jason West, Steven J. Smith, Christopher G. Nolte, Daniel H. Loughlin

We demonstrate a methodology for identifying cost-effective strategies for reducing health damages associated with fine particulate matter (PM_2.5). We directly specify future national PM_2.5 mortality cost reduction targets in a human-earth system model with state-level resolution (GCAM-USA), to identify the control actions, sectors, and locations that most cost-effectively meet these targets. We modified GCAM-USA to include mortality cost factors for sulfur dioxide (SO₂), nitrogen oxides (NO_x), and primary PM_2.5 from each state. A series of PM_2.5 mortality constraints are evaluated, targeting 10% to 50% reductions in national total PM_2.5 mortality costs by 2050, relative to a current legislation scenario.

Our results suggest that substantial health benefits can be achieved cost-effectively by targeting sources with higher primary PM_2.5 emission intensities, generally replacing these with electricity. The decreased use of industrial coal, building biomass, and industrial liquids contributes to 89% of the total PM_2.5 mortality reductions in the 50% constraint scenario in 2050, but the reduced fuel quantity is only equivalent to 2% of the total energy consumption. The marginal and total PM_2.5 health benefits in 2050 are approximately 2 and 6 times the marginal and total policy costs, respectively, in the 50% constraint scenario. Increasing the stringency of PM_2.5 constraints expedites the phaseout of high emission-intensity sources, leading to larger declines in major air pollutant emissions, but very limited co-benefits in reducing carbon dioxide (CO₂) emissions. Control strategies also tend to reduce emissions more in the East North Central and Middle Atlantic states, which have greater population density and higher base-year PM_2.5 mortality than national averages, while also having greater opportunities for low-cost controls. Our study illustrates how public health considerations can be integrated explicitly into the development of multi-sector, multi-pollutant, and multi-region air quality management.

Assessing PM2.5 Model Performance for the Conterminous U.S. with Comparison to Model Performance Statistics from 2007-2015

James T. Kelly, Shannon N. Koplitz, Kirk R. Baker, Barron H. Henderson, Norm Possiel, Heather Simon, Alison M. Eyth, Carey Jang, Sharon Phillips, and Brian Timin

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

Amara L. Holder, Havala O.T. Pye, Benjamin N. Murphy, Jesse O. Bash

Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, USA

Previous studies have proposed that model performance statistics from earlier photochemical grid model (PGM) applications can be used to benchmark performance in new PGM applications. A challenge in implementing this approach is that limited information is available on consistently calculated model performance statistics that vary spatially and temporally over the U.S. Here, a consistent set of model performance statistics are calculated by year, season, region, and monitoring network for PM2.5 and its major components using version 4.7.1-5.2.1 simulations from the Community Multiscale Air Quality (CMAQ) model for years 2007-2015. The multi-year set of statistics is then used to provide quantitative context for model performance results from the 2015 simulation. Additionally, results from the 2015 simulation are compared with results from five sensitivity simulations to help identify causes of model performance features. This presentation will illustrate the application of the model performance assessment method for major PM2.5 components and discuss its strengths and limitations.

The Fast Air Quality Responses from the Fine Particle Reduction Measures of South Korea
Jinseok Kim1, Younha Kim1, Jung-Hun Woo1*, Bok Haeng Baek2, Jinsu Kim1, Youjung Jang1, Minwoo Park1

* Correspondence : jwoo@konkuk.ac.kr
1 Konkuk University, Seoul, Korea
2 University of North Carolina, Chapel Hill, USA

To maintain environmentally sound and sustainable development while at the same time achieving a low-carbon society, it is advantageous to improve the effectiveness of integrated management of air pollution and climate change policies with lower cost. The GUIDE(GHGs and Air Pollutants Unified Information Design System for Environment) integrated assessment model has been developed to improve the climate change and the atmospheric environment at the same time. One of the major component module is the emissions - air quality impact analysis model which are to developed based on RSM (Response Surface Methodology) in the ABaCAS(Air Benefit and Cost and Attainment Assessment) modeling system. The RSM could be developed using the multiple iteration of CMAQ air quality model, WRF meteorological model, and SMOKE-Asia emissions processing system. In our research GUIDE-RSM module was developed based on the KORUSv5 emission inventory. The boundary condition is 27km grid domain and research area (South Korea) is 9km grid domain. Emissions of VOCs and PM were speciated using SAPRC99 / AERO5 mechanism.

Using the GUIDE-RSM, a case study was conducted on South Korea where various mitigation policy measures are being implemented in order to fight against serious fine particle pollution problem. The average PM2.5 concentration in South Korea in 2016 is 26 ug/m3, which is twice the WHO recommended standard (10 ug/m3) and the other mega city (13.8 ug/m3 in Tokyo and 11 ug/m3 in London, 15'). In addition, the average concentration of PM2.5 was high in spring and winter, and the number of the fine particle pollution warning has been increased. In order to improve the high PM2.5 pollution, the government established "The comprehensive fine particle management plan." The goal of the plan is to improve the PM2.5 concentration of 26 ug/m3 to 18 ug/m3 at Seoul by reducing 30% of domestic emissions until year 2022. In this presentation, we will present the effectiveness of policy measures in Korea by applying reduced emissions from the various measures to the PM2.5 air quality using GUIDE-RSM.

ACKNOWLEDGEMENT

This subject is supported by Korea Ministry of Environment as "Climate Change Correspondence Program (project no.2016001300001). This work was supported by a grant from the National Institute of Environment Research (NIER), funded by the Ministry of Environment (MOE) of the Republic of Korea (NIER-2019-01-02-037).

Modeling of U.S. and International Contributions to Visibility Impairment at Class I Areas

Brian Timin, Barron H. Henderson, Alison M. Eyth, Kirk R. Baker, Heather Simon, Sharon Phillips, Norm Possiel, and Shannon N. Koplitz

In support of regional haze rule requirements, previous modeling studies have examined future year projected visibility impairment at U.S. Class I areas (Federally protected national parks and wilderness areas). These mostly remote locations have relatively low concentrations of particulate matter, which makes modeling a challenge. One of the largest uncertainties in past visibility studies has been large contributions to visibility impairment from boundary conditions, especially in the West and near the edges of the regional scale modeling domain. This was particularly problematic because boundary conditions could not be specifically attributed to natural or international anthropogenic emissions. In this study, we track the natural and anthropogenic sources of regional haze using a multi-scale application of CMAQ (covering the northern hemisphere), and CAMx (covering most of North America).

We use a combination of hemispheric and regional photochemical grid models to project visibility impairment to 2028. High quality base year (2016) international emissions were developed to drive hemispheric modeling. The hemispheric version of CMAQ 5.2.1 was used to provide initial and boundary conditions to the CAMx model (version 7.0) using a regional scale 36km and 12km modeling domain. Emissions contributions to visibility impairment were calculated for 22 U.S., Canadian, and Mexican emissions sectors using CAMx particulate source apportionment technology (PSAT). In addition, a combination of hemispheric CMAQ zero-out sensitivity model runs and CAMx PSAT outputs were used to quantify international anthropogenic contributions to visibility impairment at Class I areas.

This talk will share model performance evaluation, projections and attribution results. Preliminary performance evaluations results show improvements at many monitors. Attribution results will provide clarity on the previous boundary conditions and make it possible to estimate international anthropogenic contributions to visibility impairment and compare observationally-derived natural conditions to the simulated quantity.

Dynamical downscaling of a global chemistry-climate model to study the influence of climate change on mid-21st century PM2.5 and Ozone distributions in the Continental US

Surendra B. Kunwar, Jared H. Bowden, George Milly, Michael Previdi, Arlene M. Fiore, J. Jason West

Anthropogenically induced climate change and natural feedback emissions (biogenic VOCs, wildfires) have the potential to alter PM_2.5 and O₃ levels in the coming decades. In this study, we aim to quantify the impacts of climate change on US air quality (PM_2.5 and O₃) at fine spatial resolution and as probability distributions for the 2050s, and to distinguish the climate change signal in air quality from natural climate variability.

GFDL-CM3 simulates global climate and air chemistry (at resolution 2.5^o X 2^o) for the period 2006-2100 under the IPCC-defined RCP8.5 climate change scenario. To isolate the impact of only climate change on air quality, GFDL-CM3 simulations fix aerosol and O₃ precursor emissions at 2005 levels. Empirical Orthogonal Function (EOF) analysis of GFDL-CM3 simulations have aided us in carefully selecting present (2006-2020) and future (2040-2060) years that represent the upper quartile and median of PM_2.5 probability distributions for different CONUS regions. We dynamically downscale the GFDL-CM3 meteorology and chemistry of the selected simulation years in the regional models WRF (Weather Research and Forecasting) and CMAQ (Community Multiscale Air Quality), respectively.

While the three ensemble members of GFDL-CM3, along with NCAR CESM simulations, provide unprecedented statistics to define PM_2.5 and O₃ probability distributions, the downscaled CMAQ simulations (with up-to-date organic aerosol chemistry) allow us to spatially (12km grid cells) and temporally (1-hour intervals) refine the coarse global model probability distributions. The calculation of present and future PM_2.5 and O₃ probability distributions from coarse global and fine regional models are described here, and analysis of PM_2.5 and O₃ distribution changes during the period 2006-2060 for both global and regional models can improve our understanding of meteorological drivers of future air quality change and associated changes in health impact and visibility. Here we also compare the present-day model meteorology and PM_2.5/O₃ levels with present day observations, and with future climate and chemistry from the global simulations.

Air Quality Modeling of a Typical Wintertime PM2.5 Pollution Event in Cache Valley, Utah: Implications for Emission Control Strategies.

Nancy Daher, Christopher Pennell

The Cache Valley in Utah is susceptible to elevated levels of fine particulate matter (PM2.5) associated with persistent cold air pool episodes. During these periods, PM2.5 levels often exceed the 24-hr PM2.5 national ambient air quality standard, with ammonium nitrate accounting for over 50% of PM2.5 mass. Nitrogen oxides (NOx), ozone (O3) and photochemically-produced oxidants from volatile organic compounds and chlorine species play an important role in particulate nitrate formation.

A typical wintertime PM2.5 air pollution event that occurred in the Cache Valley in 2011 was simulated using the Comprehensive Air Quality Model with extensions (CAMxv6.3). A comparison of measured and modeled PM2.5 chemical species showed that while the model captures well the temporal variation in PM2.5 mass, it underestimates ammonium nitrate. This underprediction in ammonium nitrate is accompanied by an underestimation of O3 and overestimation of NOx during daytime hours. Nitryl chloride, an important source of radicals, is also underpredicted in the model. Findings suggest that the photochemical production of oxidants, including ozone, is potentially underestimated in the model, leading to an underprediction of NOx conversion to nitric acid and ammonium nitrate. This has potential implications on the response of ammonium nitrate to changes in precursor emissions. Further investigation of the model performance at simulating free radical sources, such as aldehydes and chlorine species, is needed.

Urban-Scale Source Attribution of Greenhouse Gases Using an Air Quality Model

Mike Moran, Stephanie Pugliese Domenikos, Craig Stroud, Junhua Zhang, Felix Vogel, Shuzhan Ren, Qiong Zheng, Doug Worthy, and Jennifer Murphy

Eulerian air quality models are widely used to simulate the emission, transport, transformation, and removal of air pollutants such as SO₂ and NO_x as well as longer-lived species such as CO and low-reactivity VOCs. Such simulations require the model to be provided with detailed information about pollutant emissions and meteorological conditions. In principle, though, it should be possible to apply the same model to simulate atmospheric concentrations of long-lived greenhouse gases (GHGs) such as CO₂ and methane (CH₄) if their emissions can be described in similar detail. Recently, the GEM-MACH in-line air quality model has been used in two urban-scale GHG source attribution studies, one for CO₂ and one for CH₄. Both studies employed a high-resolution (2.5-km) regional grid centred over Toronto, Canada and a special version of GEM-MACH with an additional set of "tagged" tracer concentration fields to track GHG emissions from specific source sectors, source regions, and times of day. The sum of these tagged concentration fields yields the total GHG concentration field, and the ratio of a tagged GHG concentration field to the total GHG concentration field at a receptor provides an estimate of the relative contribution of that source sector-region-time at that place and time. Both studies required the development of a special regional GHG emission inventory that was then processed by the SMOKE emissions processing system to build a set of tagged GHG emissions fields for input to GEM-MACH. For the CO₂ attribution study, 11 tagged CO₂ emissions fields for 9 source sectors and 3 source regions were used while the CH₄ attribution study considered 136 tagged CH₄ emissions fields for 9 source sectors, 14 source regions, and 2 times of day. Both studies will be described in this presentation, with a focus on results from the first (CO₂) attribution study.

Estimation of Surface NO2 Using Airborne Remote Sensing Data and CMAQ Model Output from DISCOVER-AQ Campaigns

K. Pickering, L. Lamsal, M. Follette-Cook, D. Allen, W. Swartz, S. Janz, W. Appel, G. Pfister

Satellite-based observations of NO2 vertical column densities have enabled development of long-term global NO2 datasets at reasonably high spatial resolution. However, these column data are underutilized by air quality specialists, who need surface concentrations to augment sparse ground measurement networks. We describe a method for inferring estimates of surface-level NO2 from retrievals of NO2 tropospheric column density. This methodology is applied to airborne remotely sensed NO2 column data from the four NASA DISCOVER-AQ field campaigns. We use high-resolution CMAQ simulations to provide a priori information for the retrievals. The spatially and temporally varying relationship between surface and column from the CMAQ model is used in estimating surface NO2 from the column observations. These estimates are evaluated against in-situ surface concentrations during the NASA DISCOVER-AQ field campaigns. Uncertainties in the estimates are addressed through evaluation of the CMAQ NO2 profile shapes using in-situ profile data from the NASA P-3B aircraft. The method utilized here can be considered a prototype for potential use in estimating surface NO2 from column data observed by the future TEMPO geostationary air quality satellite instrument.

A Retrieval Closure Study to Determine the Best Metrics for Evaluating the CMAQ Model with CrIS NH3 Retrievals

Matthew J. Alvarado¹, Karen Cady-Pereira¹, Jeana Mascio¹, Chantelle Lonsdale¹, and Mark Shephard²

¹Atmospheric and Environmental Research (AER)

²Environment and Climate Change Canada (ECCC)

In order to use infrared (IR) satellite observations to evaluate modeled vertical profiles of pollutants and improve emission estimates, an observation operator must be applied to the modeled profile to estimate what profile the satellite instrument would have retrieved if the model profile were the true profile. However, the best formulation of the observation operator to use, and the best metrics to compare the transformed model profile with the satellite retrieved profile (i.e., vertical column density versus concentration at a specific vertical level or the surface) can depend on the retrieval approach, the species being retrieved, and the model being evaluated. A retrieval closure study where (a) known profiles are used to simulate the radiance measured by the satellite, (b) the satellite retrievals algorithm are applied to those radiances, and then (c) the consistency between the original and retrieved profiles is examined for different formulations of the observation operator can be used to determine the best formulation of the observation operator and the best evaluation metrics.

In this work, we used about 850 CMAQ simulated NH₃ profiles from the 2013 SENEX campaign in the Southeast US and the CrIS NH₃ retrieval algorithm to perform such a closure study; the objective was to determine the best formulation of the observation operator and the best evaluation metric to use in our study to improve NH₃ emission estimates across North America. We find that for evaluating CMAQ with CrIS NH₃ retrievals, a linear observation operator works better than the traditional logarithmic observation operator and the column density is a better metric than the surface concentration or other single pressure level concentration for the comparison. We also find that the three different a priori profiles used in the current CrIS retrieval (representing clean, moderately polluted, and highly polluted conditions) can lead to a sudden jump in concentrations when the a priori changes, especially under conditions of low thermal contrast. We discuss the implications of these results for improving NH₃ emission inventories in CMAQ with CrIS observations and discuss potential updates to the CrIS retrieval algorithm that would lead to improved consistency of the modeled and retrieved profiles.

The impact of the direct effect of aerosols on meteorology and air quality using aerosol optical depth assimilation during the KORUS-AQ campaign

Jia Jung¹, Amir H. Souri², David C. Wong³, Sojin Lee¹, Wonbae Jeon⁴, Jhoon Kim⁵, and Yunsoo Choi^1*

¹Department of Earth and Atmospheric Sciences, University of Houston, TX, USA

²Harvard-Smithsonian Center for Astrophysics, Cambridge, MA, USA

³US Environmental Protection Agency, Research Triangle Park, NC, USA

⁴Institute of Environmental Studies, Pusan National University, Busan, Republic of Korea

⁵Department of Atmospheric Sciences, Yonsei University, Republic of Korea

To quantify the impact of the direct aerosol effect accurately, this study incorporated the Geostationary Ocean Color Imager (GOCI) aerosol optical depth (AOD) into a coupled meteorology-chemistry model. We designed three model simulations to observe the impact of AOD assimilation and aerosol feedback during the KORUS-AQ campaign (May-June 2016). By assimilating the GOCI AOD with high temporal and spatial resolutions, we improve the statistics from the comparison AOD and AERONET data (RMSE: 0.12, R: 0.77, IOA: 0.69, MAE: 0.08). The inclusion of the direct effect of aerosols produces the best model performance (RMSE: 0.10, R: 0.86, IOA: 0.72, MAE: 0.07). AOD values increased as much as 0.15, which is associated with an average reduction in solar radiation of -31.39 W/m², a planetary boundary layer height (-104.70 m), an air temperature (-0.58 ), and a surface wind speed (-0.07 m/s) over land. In addition, concentrations of major gaseous and particulate pollutants at the surface increase by 7.87 - 34 % while OH concentration decreases by -4.58 %. Changes in meteorology and air quality appear to be more significant in high-aerosol loading areas. The integrated process rate analysis shows decelerated vertical transport, resulting in an accumulation of air pollutants near the surface and the amount of nitrate, which is higher than that of sulfate because of its response to reduced temperature. We conclude that constraining aerosol concentrations using geostationary satellite data is a prerequisite for quantifying the impact of aerosols on meteorology and air quality.

High-spatial resolution airborne mapping of NO2 column densities of New York City and Long Island Sound

Laura Judd, Jassim Al-Saadi, Scott Janz, Matthew Kowalewski, Lukas Valin, James Szykman, Paul Miller, Brad Pierce, Brian McDonald

In 2018, NASA's UV-VIS airborne mapping spectrometers were used to observe air quality over New York City and Long Island Sound as part of the Long Island Sound Tropospheric Ozone Study (LISTOS). Spanning June-October, the GeoCape Airborne Simulator (GCAS) and GEOstationary Trace gas and Aerosol Sensor Optimization (GeoTASO) flew on NASA Langley Research Center aircraft during the daylit hours on 17 flight days under a wide variety of clear-sky meteorological conditions. High resolution spectra from these instruments are used to retrieve NO2 column densities at 250 x 250 m via Differential Optical Absorption Spectroscopy. Gapless rasters were built over 2-4 hour periods by executing parallel flight lines spaced to ensure overlap between adjacent swaths - both these instruments operate in a push-broom configuration with a swath width of approximately 7 km at the nominal flight altitude of 8.5 km. Rasters were repeated 2-4 times per day capturing the impact of diurnally varying emissions and meteorology on the spatial heterogeneity of NO2 in this coastal urban environment. This work presents an overview of the NO2 column retrievals from LISTOS, which often spanned two orders of magnitude between the background and most polluted environments with highly variable day-to-day patterns. The NO2 raster datasets are spatially regridded to simulate coarser spatial resolution observations (e.g., TEMPO or regional air quality model analysis) to demonstrate the influence of spatial resolution on observed spatial patterns and comparisons to validation datasets. Data over known emission sources are also examined to begin the discussion on how these high spatial resolution observations can be used to evaluate emissions inventories.

Understanding the Impacts of Land Use and Land Cover Change on Regional Climate in Sub-Saharan Africa: A Cautionary Tale for Regional Climate Modeling

Timothy Glotfelty, Diana Ramirez, Adrian Ghilardi, Jared Bowden, and J. Jason West

Land use and land cover changes (LULCC) induce biogeophysical changes in surface albedo, evapotranspiration, and surface roughness that can have a substantial impact on regional climate by altering surface energy balances, hydrologic cycles, and regional to global scale circulations. Land surface models (LSMs) simulate these biogeophysical impacts of LULCC in weather and climate models, calculating surface energy, moisture, and momentum fluxes between the land surface and the overlying atmosphere.

This study uses the Weather Research and Forecasting (WRF) Model and tests five different LSM configurations to elucidate the impacts of LSM errors and uncertainty on LULCC induced changes in regional climate. We aim to evaluate the performance of different LSM configurations on regional climate, and determine whether these different configurations can alter the regional climate impact of LULCC. We model Sub-Saharan Africa because of its significant historical and expected future LULCC. The five LSM configurations used in this study include the Noah LSM, the Noah multi-parameterization (Noah-MP) LSM, the Noah LSM using satellite climatology derived albedo and leaf area index (LAI) (Noah-SAT), the default community land surface model (CLM-D), and the CLM model with albedo and LAI parameters that we updated (CLM-U). All five configurations were applied to an annual (2013) simulation over Sub-Saharan Africa at 36 km resolution. Noah, Noah-MP, CLM-D, and CLM-U were also applied over 2010-2015, once using static LULC fields from 2001, and using dynamic LULC from those years, the difference indicates the climate impact of LULCC over time.

Results illustrate that caution must be exercised for LULCC applications of WRF. We find that the Noah LSM can have a significantly larger albedo than the satellite climatology, resulting in overpredicted reflected surface radiation and underpredicted near surface temperatures. Noah-MP does not simulate the relationship between soil type and albedo, which results in significantly underpredicted albedo in arid regions. Noah, Noah-MP, and CLM-D all inaccurately represent the woody savanna land use category, leading to unrealistic spatial distributions for albedo and surface roughness length. Additionally, CLM-D was developed for the Northern Hemisphere (NH), using NH specific LAI monthly profiles, resulting in inaccurate biophysical parameters in the tropics and Southern Hemisphere. We corrected this in CLM-U by generating new monthly LAI profiles based on the WRF satellite climatology for 17 bio-climate regions in Sub-Saharan Africa. We also scaled up the albedo for sandy soils to better match the satellite climatology. Noah-Sat and the CLM-U configurations have the most accurate surface properties, but these configurations did not always lead to significant improvement in meteorological model performance across multiple variables compared to the other configurations. Hence, using standard model evaluation procedures, especially for regions like Sub-Saharan Africa where observations are sparse, can easily miss underlying problems within LSMs for LULCC applications. Our study indicates that special attention to LSM model configuration is needed before trying to use a regional climate model to understand the impacts of LULCC on the regional climate.

Overview of measurements and preliminary findings from the Long Island Sound Tropospheric Ozone Study 2018 (LISTOS)

Jassim A. Al-Saadi (NASA LARC), Pete Babich (CT-DEEP), Kirk Baker (EPA OAQPS), Tim Berkoff (NASA LARC), Roisin Commane (LDEO-Columbia), Russell Dickerson (UMD), Dirk Felton (NYS DEC), Pete Furdyna (NYS DEC), Drew Gentner (Yale), Scott J. Janz (NASA GSFC), Laura Judd (NASA LARC), Luis Lim (NJ DEP), John Mak (SUNY-Stony Brook), Brian McDonald, (NOAA-CIRES), Paul Miller (NESCAUM), Brad Pierce (UWISC SSEC), Jon Pleim (ORD/NERL), Xinrong Ren (UMD), Jim Schwab (ASRC SUNY Albany), Charlie Stanier (U Iowa), John Sullivan (NASA GSFC), Jim Szykman (EPA ORD/NERL), Robert Swap(NASA GSFC), Luke Valin (ORD/NERL), Andrew Whitehill (EPA ORD/NERL), David Williams (EPA ORD/NERL),

The Long Island Sound Tropospheric Ozone Study (LISTOS) of 2018 sought to better understand the emissions, chemistry and transport of ozone and ozone precursors upwind and over the Long Island Sound. NESCAUM, federal, state, local and academic partners coordinated a wide range of measurements through a grass-roots planning effort which included both short-term intensive measurements between mid-June and mid-October along with long-term enhanced monitoring activities, some of which are ongoing in support of the re-designed Photochemical Assessment Monitoring Station network. The variety of instrumentation deployed, both remote sensing and in situ, and the airborne, mobile, ship-based and fixed site platforms serve as a prototype of an atmospheric composition integrated observing system. Preliminary datasets and analysis indicate that the study provides valuable information on the timing, spatial patterns and quantity of NOx emissions; quantification of poorly characterized VOC sources in the region such as volatile consumer products, identification of gaps in our understanding of relative importance of biogenic vs anthropogenic sources of VOC, and an impressive characterization of vertical distributions of ozone, NO2, turbulence/mixing and aerosol backscatter.

Intensification of Extreme Precipitation in Eastern North Carolina Projected in Dynamically Downscaled CESM and CM3 Models Under Future Scenarios (2025-2100) Using WRF.

A. M. Jalowska, J. H. Bowden, T. L. Spero

The increasing trend in the frequency and intensity of extreme precipitation events has been well-documented within the eastern United States in historical climate records. Recent climate research suggests that the frequency and magnitude of extreme precipitation in the eastern United States. will continue to increase throughout the twenty-first century. Eastern North Carolina communities are particularly vulnerable to frequent and devasting storms and their associated flooding. Additionally, recent years showed that these communities and the infrastructure are not well prepared to face more intense precipitation events, under changing climate. To address arising challenges related to changing precipitation characteristics, governing and managing bodies need information to prepare for future weather, which can be provided by the modelling community. This study uses the Weather Research and Forecasting (WRF) model to dynamically downscale simulations from two global climate models for a highest greenhouse gas emission scenario (RCP8.5): the Community Earth System Model (CESM) and the Geophysical Fluid Dynamics Laboratory coupled climate model (CM3). In this study, we examine changes to both, mean and extreme precipitation within eastern North Carolina from 2025 to 2100. We apply a regional frequency analysis to the WRF model fields to better inform stakeholders of possible changes in extreme precipitation and their recurrence probabilities in these scenarios. Preliminary data from this study indicate up to a 30% increase in annual precipitation from 2025 to 2100. Changes in the intensity of extreme precipitation events are more pronounced in future climate, with a one-fold increase in the intensity of one-day precipitation maxima.

Improving Mexico Emissions Inventory Using Satellite Data
Tejas Shah, Lynsey Parker, John Grant, Greg Yarwood
Ramboll, Novato, California

Jim Smith and Chris Kite
Texas Commission on Environmental Quality, Austin, Texas

International transport of pollution has increased in importance as the US National Ambient Air Quality Standards for ozone and particulate matter (PM) have become more stringent in recent years. Mexican anthropogenic emissions contribute to ozone and PM transport into the continental United States. This project evaluated existing anthropogenic emissions inventories for Mexico and made targeted improvements by drawing upon several satellite-derived datasets. We compared the emissions inventory to satellite-derived emission estimates for large sulfur dioxide (SO2) sources, nitrogen oxide (NOx) emissions, and flaring associated with oil and gas production. Satellite data for SO2 are consistent with emission estimates for many sources but reveal that several large SO2 sources are not captured in the existing inventory and should be added. Tropospheric NO2 column retrievals from National Aeronautics and Space Administration (NASA) indicate that there is good spatial agreement between NOx emissions inventory for stationary industrial sources and satellite NO2 column with several large NOx point sources observed in the NO2 columns. Satellite-derived estimates of gas flaring indicate that upstream oil and gas sources are not well represented in the inventory, and a bottom-up emission inventory should be developed for this sector.

Quantifying the air quality and human health benefits of GHG mitigation Pathways in California

Shupeng Zhu¹, Michael Mac Kinnon², Brendan Shaffer², Owen Yang², G.S. Samuelsen²

¹Computational Environmental Sciences Laboratory, University of California, Irvine, CA 92697, USA
²Advanced Power and Energy Program, University of California, Irvine, CA 92697, USA

California has established greenhouse gas (GHG) emission goals that require significant transitions within energy sectors to provide mitigation required by 2050. The present study utilizes scenarios developed via a cost analysis modeling tool (PATHWAYS) that evaluates long-term GHG abatement scenarios including high electrification and transitions to renewable gaseous fuels. Electrification represents a key strategy towards GHG reduction, and will also achieve criteria pollutant reductions. Additionally, GHG reductions can occur through the production of low carbon gaseous fuels which can be used in zero- and low-emission strategies including fuel cell conversion or low-NOx engines. Conversely, we consider the potential impacts of biorefineries as an introduced source of criteria pollutant emissions that could degrade local air quality. Using PATHWAYS, a range of scenarios can be developed that achieve the 2050 GHG goals, but also have differing impacts on regional air quality concentrations of pollutants including ozone and PM2.5 due to the assumed technological mixes within end-use sectors. The goal of this research is to characterize and quantify the air quality and human health impacts for a set of long-term low-carbon scenarios achieving California's GHG emission targets in 2050 to provide insight into the air quality co-benefits of technological shifts within cases. Using the output from PATHWAYS, we develop spatially and temporally resolved characterizations of criteria pollutants for each scenario for all major end-use sectors in California, including all stationary and mobile sources. Next, we translate emission changes into impacts on atmospheric pollution levels, including ground-level ozone and PM2.5, via a 3-D photochemical air quality model that accounts for atmospheric chemistry and transport. Impacts on regional air quality are then used to conduct a health impact assessment which provides a quantitative estimate of the incidence and value of avoided harmful health outcomes associated with air pollution. We show that the health savings of GHG abatement measures are notable, and that impacts are seasonally and regionally-dependent. For example, HDV abatement yields the largest benefits to summer ozone in the South Coast Air Basin while electrification of the residential and commercial sectors has notable benefits in the Central Valley in winter. The addition of biorefineries does degrade air quality locally and could reduce health saving, but the overall impact of the scenario assumptions result in positive health benefits ~90 to 95% of the total without refineries.

Comparison of PM2.5 measured in urban North Carolina settings using a low-cost optical particle counter and Federal Equivalent Methods

Brian Magi, Department of Geography and Earth Sciences, UNC Charlotte, Charlotte, NC

Calvin Cupini, Clean Air Carolina, Charlotte, NC

We present the results of a multi-season field evaluation of a low-cost optical particle counting sensor (PurpleAir PA-II) that reports mass concentration of particulate matter with diameter less than 2.5 microns (PM2.5). The PurpleAir is a widely adopted example of the large and growing field of low cost internet-of-things (IoT) sensors. In our study, one PA-II was collocated with a Federal Equivalent Method (FEM) Beta Attenuation Monitor (MetOne BAM model 1022) in Charlotte, North Carolina, and another PA-II was collocated with an FEM Teledyne 640x in Winston-Salem, North Carolina. We analyzed over two years of PM2.5 data from the Charlotte collocation study, and over one year of data from Winston-Salem. In this presentation, we discuss quality-control steps we applied when analyzing the PA-II data, how hourly PA-II PM2.5 compares with the FEM PM2.5, and the results of a multiple linear regression (MLR) model that uses with the FEM PM2.5, relative humidity (RH), and temperature as predictors to model the reported PA-II PM2.5. The MLR model results are part of on-going research to understand whether hourly PA-II PM2.5 can be corrected to better compare with FEM PM2.5. Initial analysis of a subset of the Charlotte collocation datasets suggests a 27-57% improvement in the accuracy of the PA-II PM2.5 data relative to the reference data from the BAM 1022, with the highest percentage improvements for moderate to high RH. Specifically, the higher percentage improvements are a result of our MLR correction method accounting for variable RH conditions. Thus, at moderate to high RH, non-linear hygroscopic growth factors that are not accounted for in the PA-II processing algorithms are better accounted for after correction. Overall, our results show that corrected hourly PM2.5 data from the PA-II is accurate to within about 4 micrograms per cubic meter. While the accuracy of the corrected PA-II is less than a maintained FEM device, we suggest that microelectronic IoT technology is indeed paving a path towards a transformational dataset that would complement the existing regulatory monitoring network.

Estimation of nose-level NO2 in the Houston-Galveston area using airborne remote sensing data from DISCOVER-AQ and output from a 4-km resolution simulation with CMAQ

D. Allen, L. Lamsal, M. Follette-Cook, K. Pickering, W. Swartz, S. Janz

Satellite-based observations of NO2 vertical columns have resulted in an improved understanding of pollutant emissions and trends over the last 20 years. However, their use by the health- and air-quality communities is limited because the relationship between column NO2 and surface NO2 varies spatially and temporally. In this study, NO2 tropospheric columns from the Ozone Monitoring Instrument (OMI), remote sensing data from the DISCOVER-AQ field campaign, and model output from a 4-km simulation with CMAQ are used to infer surface concentrations of NO2 over the Houston-Galveston area during September 2013. In this presentation, the quality of the off-line WRF-CMAQ simulation of vertical mixing, PBL heights, NO2, and O3 will be evaluated through comparison with measurements from the DISCOVER-AQ campaign. The sensitivity of inferred surface layer concentrations to NO2 profile will be examined through the use of profile data from CMAQ and the NASA P-3B aircraft. The method utilized here is general and the OMI data used here could be replaced with data from TROPOMI and the future TEMPO geostationary air quality satellite instrument.

Modeling the Contribution of Long Range Transport to Nitrogen Deposition in U.S. Hydrological Regions

Sharmin Akter^[1], Michael Crowl^[2], and Kristina Wagstrom^[2]

The deposition of atmospheric nitrogen containing species to aquatic systems and watersheds can lead to excessive algae growth, particularly in coastal waterways. It increases the risk of acidification and hypoxia in aquatic systems by reducing oxygen levels for living organisms. It is important to determine the major species, source sectors, and source regions responsible for atmospheric nitrogen deposition to develop effective watershed management. We are using the Comprehensive Air Quality Model with extensions (CAMx) version 6.0, along with the Particulate Matter Source Apportionment Technology (PSAT), to identify and separate different source regions and sector contributions to atmospheric deposition. We are modeling the amount of atmospheric nitrogen depositing from different sources such as electricity generating units, biogenic emissions, dust, onroad and marine vehicles, residential combustion, agricultural ammonia and other point sources as well as source regions including the 48 contiguous states (plus Canada and Mexico ) and the 20 largest metropolitan areas. We are using emissions, meteorology, boundary conditions, and ozone column inputs from the United States Environmental Protection Agency's 2011 Modeling Platform. CAMx provides a complete framework to track the contributions to atmospheric nitrogen depositions from different source regions and sectors resolved contributions. This approach can also be used to estimate the contributions from source regions and sectors to atmospheric deposition in other regions. We evaluate the total deposited mass and ambient concentrations of atmospheric nitrogen species against measurements. It will aid environmental regulators in developing watershed management plans to protect the health of aquatic and terrestrial ecosystems.

Modelling the impact of aromatics on secondary pollutants over eastern China

Xiaoyang Chen, Qi Fan, Yiming Liu, Chong Shen and Xuemei Wang

We studied the uncertainty of vapor wall loss in chamber experiments and its impact on the two-product approach in CMAQ model referring to previous studies. By considering relative humidity, aerosol acidity and crystallization, we updated the parameterizations of glyoxal and methylglyoxal uptake coefficients, mainly based on the calculations of Curry et al., 2018. And a revised WRF-CMAQ modelling system, implemented with a tracing method for quantifying the contribution of aromatics to dicarbonyls, was used to simulate the Julys (summer) and Octobers (autumn) of 2010 to 2016. Together with some sensitive experiments of aromatics emissions and the vertical density columns of formaldehyde and glyoxal from OMI, we studied the temporal and spatial distribution of aromatics and its impact on its key productions like glyoxal and SOA.

Improving Air Quality Simulation for 2016 Summer Ozone Episode by Assimilating Geostationary Satellite Observations

Peiyang Cheng, Arastoo Pour-Biazar, Andrew T. White

Since clouds can greatly affect photolysis rates for chemical species by their impacts on incoming solar radiation, having an accurate estimate of cloud properties is critical for air quality numerical simulations. In this study, a technique which assimilates Geostationary Operational Environmental Satellite (GOES)-derived cloud fields developed by White et al. (2018) was implemented in the Weather Research and Forecasting (WRF) model coupled with Chemistry (WRF-Chem) simulation to improve the reliability of estimated photolysis rates and the distribution of chemical species. The simulation was carried out on a 12-km grid domain to study the 2016 summer ozone episode over the contiguous United States (CONUS). Preliminary results from this study will be presented.

Projected Wildfire Impacts on Regional Air Quality in The Western United States

Cheng-En Yang ¹ , Xinyi Dong ¹ , Yang Liu ² , Yongqiang Liu ³ , Jian Sun ^1* , and Joshua S. Fu ¹

¹ Department of Civil and Environmental Engineering, University of Tennessee, Knoxville, TN 37996

² Department of Environmental Health, Emory University, Atlanta, GA 30322

³ U.S. Forest Service, Southern Research Station, Athens, GA 30602

^*Now in Pacific Northwest National Laboratory, Richland, WA 99352

Wildfires emit smoke particles and pollutants that greatly affect the public health over the western United States (WUS) each fire season. With rising surface temperatures projected by climate models, increasing wildfire activities and severities are predicted because vegetated regions become more vulnerable to fires as a result of lower soil moisture and atmospheric humidity. In this study, we simulated future wildfires with the Representative Concentration Pathway 8.5 scenario defined by the Intergovernmental Panel on Climate Change and evaluated the impacts of projected future (2050-2059) wildfires on regional air quality over WUS during fire months (April to November). Results showed that wildfire-induced monthly mean PM 2.5 concentration increased by up to 58 g/m 3 within the study domain. In terms of fire months, August was influenced by wildfires the most with the mean monthly PM 2.5 change up to 58 g/m 3 while wildfire had the least impact on PM 2.5 concentration changes (~ 10 g/m 3 ) in a single gridcell. Changes of other major pollutants will also be discussed in this presentation.

Impact of dust emission on particulate matter in spring of 2018 in China

Song Liu, Jia Xing, Dian Ding, Hongliang Zhang, Fenfen Zhang, Bin Zhao, Shuxiao Wang

The Tarim Basin and the Hexi Corridor in western China are two major sources of dust emissions. Due to the prevailing westerly winds in the winter and spring, the contribution of particulate matter concentration from dust emissions in western China will affect eastern China after regional transmission. This study selected a time period of smog in Beijing in the spring of 2018. Based on the CMAQ model, the contribution of dust emissions to PM2.5 and PM10 concentrations in typical regions of China was estimated. The impact of dust emissions on the concentration of particulate matter in eastern China is not large, but the simulation results including dust emissions are more ideal. In the dusty weather, the contribution of sand in western China to particulate matter cannot be ignored. We further compared the simulation results of the effects of dust emissions using CMAQ and WRF-Chem models during the same time period.

Future ozone concentrations in the state of Delaware under climate change scenarios
Mojtaba S. Moghani, Cristina L. Archer

College of Earth, Ocean, and Environment, University of Delaware, Newark, Delaware

Although in the United States the emissions of ozone precursors have declined significantly in the past decades and led to improvements in air quality and human health, studies have shown that climate change itself will worsen ozone pollution by increasing the frequency of high ozone days by mid-century [1].

In this study, we compared current (2015-2017) and future mid-century (2049-2051) ozone levels in the U.S. under the high-emission Representative Concentration Pathway (RCP8.5). We applied the Comprehensive Air quality Model with extensions (CAMx) over the continental U.S. with a 12 km resolution, using: the downscaled Community Earth System Model (CESM) data to run the mesoscale Weather Research and Forecast (WRF) model, the Goddard Earth Observing System (GEOS-Chem) model to generate initial and boundary conditions for the CAMx runs, and the most recent 2016 Environmental Protection Agency (EPA) national emission inventory. We used the 2016 emissions also for the mid-century runs to study the effect of climate change alone on ozone, without including future possible environmental policy changes.

Preliminary results showed that two main effects are expected in Delaware by mid-century. The first is an increase in the number of high-ozone days (i.e., maximum daily 8-hr concentration above 70 ppb), with five extra days over the three-year period 2049-2051, or about 1.5-2 extra day each year in mid-century. Second, the ozone concentrations are expected to be higher than in the 2015-2017 base case, as the number of days with 8-hr ozone concentrations above 80 ppb is expected to increase from 22 days in the three-year period 2015-2017 base case to 48 in 2049-2051. In summary, without policy intervention, ozone pollution is expected to worsen in frequency and intensity in Delaware.

[1] Archer, C. L., Brodie, J. F., & Rauscher, S. A. (2019). Global warming will aggravate ozone pollution in the US Mid-Atlantic. Journal of Applied Meteorology and Climatology, 58(6), 1267-1278.

A multi-scale model analysis of ozone formation in the Bangkok Metropolitan Region, Thailand

Pornpan Uttamang¹, Patrick C. Campbell^2,3, Viney P. Aneja¹ and Adel F. Hanna⁴

¹Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC, 27695, USA

²National Oceanic and Atmospheric Administration, Air Resources Laboratory, College Park, MD 20740, USA

³Coorperative Institute for Climate and Satellites, University of Maryland, College Park, MD 20740, USA

⁴Institute for the Environment, University of North Carolina at Chapel Hill, Chapel Hill, NC, 27517, USA

Over the last three decades, Thailand's rapid industrialization and urbanization has led to an impact on urban air quality. A majority of the country's development has occurred within and around Bangkok (BKK), the capital city of Thailand; and Bangkok Metropolitan Region (BMR). The increase in emissions is due to accelerated growth in automotive and industrial activities. Since 1995, the BMR has begun to experience air quality degradation, in particular, enhanced ozone (O₃) owing to elevated ozone precursor emissions, strong solar radiation, high temperature and high humidity. We present an observational analysis over five years (2010-2014), which reveals that [O₃] exceeded the National Ambient Air Quality Standard (NAAQs) of Thailand, especially during the dry seasons. The O₃ episodes (hourly [O₃] > 100 ppb) in this area are due to atmospheric stagnant conditions (WS < 4 ms^-1), and local and regional contributions of O_x (O₃ + NO₂). To better quantify the local and regional contributions to increased [O₃] in the BMR, the Weather Research and Forecasting model with Chemistry (WRF-Chem), version 3.9.1, is applied in this study. The model is used to investigate O₃ formation and transport that contribute to elevated O₃ levels in the BMR, in particular, the role of O₃ precursors transport from China on O₃ over the BMR, as elevated O₃ concentrations are frequently observed during dry seasons when there is a predominant Northeast monsoon wind direction. The model configuration consists of a triple-nested domain (36-, 12-, and 4-km horizontal resolutions), where the outer domain covers a majority of China and Thailand, the second and inner domains are focused on Thailand and the BMR region specifically. Five simulations including a baseline and sensitivity simulations (10%, 20% and 40% NO_x emission reduction and 40% NO_x emission reduction along with 40% VOC emission reduction over China) are performed. The meteorological ICONs/BCONs are from NCEP FNL and anthropogenic emissions are based on EDGAR-HTAP. Our study shows that East Asia and Southeast Asia including the BMR are more likely to be VOC-limited which controlling only NO_x emissions enhance the concentrations of O₃ in these regions.

What's New with the I/O API
Carlie J. Coats, Jr.

There are a number of new capabilities and options available in the Models-3 I/O API, which are of interest and use to the CMAS community of modelers:

• Now Available on GitHub: go to https://github.com/cjcoats/ioapi-3.2 for download, https://cjcoats.github.io/ioapi/AA.html for documentation.
• Increased maximum numbers of files and variables. There are two versions:
  • For the default release after October 1, parameter MXVARS3=256, for convenience in various CMAQ and SMOKE applications and with chaining sequences of files ( see below). This change should be transparent to the vast majority of applications.
  • New version I/O API Large has MXFILE3=512 and MXVARS3=16384 for use in CMAQ DDM and ISAM applications. It needs to be built separately from the normal version, and kept carefully isolated from it; it will have slightly lower performance and substantially greater memory requirements, due to these size-increases. It can be downloaded from https://www.cmascenter.org/ioapi/download/ioapi-3.2-large.tar.gz at the CMAS web site.
• PnetCDF Distributed I/O for CMAQ. Note that this is pretty-much limited to CMAQ only
• Chaining Sequence-of-Files capabilities, e.g., to handle all the single-day outputs from a study as though they were a single file. For example, the following would do a unified m3stat run for all the files of a given type for a 31-day month:

setenv name_1 < path >
...
setenv name_31 < path >
setenv FOO "LIST:name_31,...,name_1"
m3stat FOO DEFAULT

Investigating aqueous production pathways of particulate sulfur in CMAQ with AQCHEM-KMT (version 2) and the sulfur tracking method

Kathleen Fahey and Shawn Roselle

In many locations around the globe, sulfate remains a major contributor to ambient PM_2.5. Most sulfate formed in the atmosphere is attributable to a few dominant pathways (i.e., aqueous reactions of S(IV) with H₂O₂ and O₃ and gas phase oxidation of SO₂ by OH); however the major oxidation pathways in existing models are unable to accurately represent high sulfate levels observed in some regions/seasons (e.g., wintertime Beijing, Fairbanks). Recently, it was suggested that the high sulfate observations in wintertime Beijing might actually be a combination of sulfate and hydroxymethanesulfonate (HMS), a S(IV) species that can be formed in atmospheric droplets via reaction of SO₂ and HCHO (Moch et al., 2018). Here we extend the sulfur tracking method capabilities in CMAQv5.3 to track the contributions of the additional sulfate production pathways available in CMAQ's AQCHEM-KMT cloud chemistry module, as well as update CMAQ to track HMS (the formation/decomposition of which is currently included in the AQCHEM-KMT mechanism) as a transported species. We apply the model for a 2016 hemispheric simulation and investigate the relative contributions of different pathways for sulfate production as well as identify any locations/seasons where HMS might be a significant contributor to particulate sulfur.

Assessment of Environmental Variables that Affect Dissolved Oxygen Concentrations in Lake Erie Using Multi-media Modeling and Machine Learning

Christina Feng Chang¹, Marina Astitha¹, Valerie Garcia², Chunling Tang², Penny Vlahos³, David Wanik¹, Jun Yan⁴

¹Civil and Environmental Engineering, University of Connecticut, Storrs, CT, USA

²National Exposure Research Laboratory, Office of Research and Development, US Environmental Protection Agency, Research Triangle Park, NC 27711, USA

³Department of Marine Sciences, University of Connecticut, Groton, CT, USA

⁴Department of Statistics, University of Connecticut, Storrs, CT 06269, USA

The ability to predict water quality in lakes is important since healthy lakes provide numerous environmental benefits that positively influence our quality of life and the strength of our economy. A combination of urban areas, industries, and agricultural activities have contributed to an increased loading of nutrient pollution into Lake Erie, particularly phosphorus and nitrogen. In this study, in-situ measurements of dissolved oxygen (DO) in Lake Erie from 2002-2012 provided by the Lake Erie Committee Forage Task Group (LEC) and the Great Lakes National Program Office (GLNPO) serve as a proxy to identify hypoxia. Meteorological weather variables from the WRF model, hydrological variables from the VIC model, nitrogen deposition from the CMAQ model, and agricultural management practice variables from the EPIC model are used to fit a random forest model to predict concentrations of the response variables. In the past, using random forest regression, it was possible to set up a predictive model using environmental factors that explained 57% of the variance in chlor- . Since then, a predictive model has been developed for DO which is capable of explaining 82% of the variance. In addition, the importance of explanatory variables is evaluated, and the contribution of each covariate in the model is analyzed with Accumulated Local Effect (ALE) plots to better understand the occurrence of hypoxia.

Development of PM2.5 short-term prediction model using Artificial Intelligence.

Jeong-Beom Lee, Geon-Woo Yun,, Hyun-A Oh, Hui-Young Yun, Dae-Ryun Choi, Ji-Seok Koo, Kyung-Hui Wang, Yoon-Seo Koo

The need for an air quality forecast of PM becomes more important as environmental authorities and citizens are requiring air quality information in advance to take appropriate measures and actions to protect their health during pollution episodes in Korea.

In 2002, PM air quality forecasting using numerical model was implemented in Seoul, the capital city of Korea. In 2014, the air quality forecasting is extended to 18 regions across the country. The air quality forecasting system consisted of numerical model such as chemical transport model, weather forecasting model, emission model. Recently the system was repaced and improved using artificial intellegnece such as DNN, CNN, RNN, LSTM.

In this study, we developed the PM2.5 short-term forecasting model using artificial intelligence with atmospheric big data such as air quality, weather, contribution, backtrajectories, anomaly, cosine similarity forecasting data and measurement data.

The forecasting period was 3 days(D+0, D+1, D+2), and the machine learning technique was used DNN.

The results of DNN showed better agreement comparing with numerical model respect to A, FAR in Seoul.

Therefore, artificial intelligence DNN model has a possibility to overcome the a defect of the numerical model that overestimates the measurement.

Modelling high-resolution NO2 concentrations over the United Kingdom using RapidAIR

Scott Hamilton, Nicola Masey

The RapidAIR dispersion model is able to produce high temporal and spatial resolution predictions of gridded concentrations in short run times. We will present gridded NO2 concentration estimates for the whole of the United Kingdom (approx. 240,000 km2) produced by the RapidAIR model. The developments we have made to the RapidAIR model in order to be able to produce these high-resolution concentration estimates at a national scale is presented. These include how to deal with differences in terrain roughness (for example) that occur when modelling a large domain.

The Effect of Source-Resolved Primary Size Distributions on Particle Concentrations and Size predicted CMAQ

Benjamin N. Murphy¹ and Francis Binkowski²

¹National Exposure Research Laboratory, Research Triangle Park, NC USA

²Institute for the Environment, University of North Carolina, Chapel Hill, NC USA

We have updated the Community Multiscale Air Quality (CMAQ) model v5.3 with new algorithms and data to represent the emission, secondary formation, and growth of ultrafine particles. This enhanced model, CMAQ-UF, is evaluated with measurements from several urban and rural sites throughout the US. We then analyze CMAQ results at the continental US level to determine the impacts of primary and secondary particle sources on ultrafine particle exposure.

We have shown previously that organic compounds and amines can participate with sulfuric acid to form new particles or, when sulfuric acid concentrations are low, generating particles on their own. These processes are an important regional driver of ultrafine particle concentrations whereas primary emissions are the fundamental source for ultrafine particles in urban and industrialized areas. We have incorporated our robust algorithm for new particle formation and growth along with state-of-the-science parameterizations of the size distribution of primary particle emissions from more than 40 source classes (e.g. vehicles, natural gas operations, coal-fired power plants, wildfires, etc.) into a detailed annual simulation of ultrafine particle sources and fate in the United States.

We apply the new model to observations made throughout the US including California (the CalNex and CARES 2010 campaigns), and the Denver DISCOVER-AQ campaign. The qualities of these and other observation sites allow us to constrain the model in urban, suburban, and rural locations, which is critical since the drivers of ultrafine particle concentrations are expected to change dramatically among these receptor types. Characterization of ultrafine particle pollution with CMAQ-NPF improves our understanding of the most effective ways to mitigate the highest concentrations in the US.

Improvements to the Integrated Source Apportionment Method (ISAM) for CMAQv5.3 release.

S.L.Napelenok, L. Zhou, W. Hutzell, B. Murphy, S. Roselle.

The Integrated Source Apportionment Method (ISAM) calculates source attribution information for user specified ozone and particulate matter precursors within the CMAQ model. CMAQ-ISAM has been substantially updated in the CMAQv5.3 release. It is currently released directly with the source code of the base model. Its algorithms for propagating tagged concentrations fields through the gas chemistry, aerosols, and cloud modules have all ben revised and updated. These changes led to substantially faster runtimes and better performance.

Development of a Visibility Forecasting Product using Canada's Wildfire Smoke Prediction System (FireWork)

Rita So and Bruce Ainslie

Poor horizontal visibility due to smoke or haze can have detrimental impacts on the day-to-day airport operations. Due to capacity issues at major Canadian airports, any visibility conditions between 8 to 10 statue miles (SM) can start to affect the spacing of aircraft and thus reducing the number of aircraft allowed to land and take off per hour. Within Environment & Climate Change Canada, the Canadian Meteorological Aviation Center provides aviation weather forecast services, which include visibility forecast, to Canadians. The ability to provide timely and reliable forecast for poor visibility conditions can be highly valuable, especially for firefighting efforts. This study uses hourly air quality and meteorological forecasts from Canada's Wildfire Smoke Prediction System (FireWork) to develop a 2D gridded hourly horizontal visibility product at 10 km over North America. Model performance was assessed using hourly METAR and SPECI weather observations for all Canadian airports form 2016 to 2018. Much of the model development and evaluation efforts were focused on poor visibility conditions (less than 6 statue miles) as a result of smoke or haze.

Kalman Filter Bias Correction for Improving Wildfire Smoke 24-hr Average PM2.5 Forecasting for AIRPACT5

Joe Vaughan, Nicole June, Yunha Lee and Brian Lamb

A Kalman Filter Bias Correction scheme was developed for the AIRPACT5 PM2.5 forecast. The AIRPACT5 air-quality forecasting system provides daily forecasts for PM2.5 and ozone for a 4-km grid covering the Pacific and Inland Northwest Region of the USA, including the states of Idaho, Oregon and Washington. AIRPACT5 uses WRF meteorological forecasts, provided by the Department of Atmospheric Sciences at University of Washington, to drive the simulations, and uses MCIP, SMARTFIRE2, BlueSky, SMOKE (with MOVES), and CMAQ modeling components to provide a two-day forecast each morning. Chemical and particulate boundary conditions are obtained from the global WACCM model operated by NCAR Wildfires have proved an increasingly important, and difficult, part of the forecasting challenge over the last decade. AIRPACT5 uses satellite fire detects to locate fires, assumes that wildfires persist for the day after detection, and injects smoke aloft by treating fires as point emissions. Plume rise is calculated using an approach by Idaho DEQ that utilizes heat flux and stipulates both smoldering and flaming contributions of smoke. Compared with the ozone forecast, the PM2.5 forecast is problematically inaccurate. In the work reported in this poster, the bias for the 24-hr Average PM2.5 forecast is estimated by applying a Kalman Filter technique. The Kalman Filter is applied at air-quality monitoring sites falling within the AIRPACT5 domain and reporting PM2.5 measurements which are then available through the AirNow system; each day, AIRPACT5 collects the prior day's monitoring results from AirNow. The bias at each monitoring site is calculated by applying the Kalman Filter to the four previous days of model and monitoring results. The site specific bias estimates are then interpolated throughout the domain and then applied as a correction to the entire gridded forecast for 24-hr Average PM2.5.

Evaluating the air quality impacts of an operational prescribed burning program

Sadia Afrin, Fernando Garcia-Menendez, Shannon Koplitz, Kirk Baker

Prescribed fire smoke contributes to fine particulate matter (PM_2.5) in the Southeastern U.S. Evaluating the impacts of prescribed burning is, therefore, an important research need to minimize air pollution and associated health costs in the region. A limited number of studies have attempted to assess the impacts of prescribed fire on ambient PM_2.5, likely due to the challenges associated with compiling of the burn data. In this ongoing study, we use chemical transport modeling to explore the air quality and potential public health impacts of an operational prescribed burning program in North Carolina. The NC Division of Parks and Recreation (NC State Parks) maintains an active and growing prescribed burning program with detailed information for all fires that have occurred within State Parks land over multiple years. We analyze all NC State Park fires that occurred during a full year and estimate the associated air pollution and smoke exposure over the state of North Carolina using the Community Multiscale Air Quality Modeling System (CMAQ). In 2016, the Division's land management program conducted 57 prescribed burns, treating over 4,700 acres of NC State Parks land. In addition, two major wildfires occurred within the NC Parks in 2016, rendering this year a unique opportunity to serve as a case study. We compare smoke exposure from these two 2016 wildfires to that of the prescribed fires within the burning program. Emissions for all fires are estimated using the BlueSky modeling framework based on NC State Parks' fire records. The results of this study can help in assessing smoke exposure of a complete prescribed burning program. The comparison between the air quality impacts of prescribed fires and wildfires will also be valuable in understanding the trade-offs between prescribed fire and wildfire risk. In addition, the findings of the analysis can serve as an initial step in developing more complete cost-benefit analyses of operational land management programs.

Development of Smartphone Application for Modeling Personal Exposures to Ambient PM2.5 and Ozone

Michael Breen,1 Vlad Isakov,1 Catherine Seppanen,2 Sarav Arunachalam,2 Miyuki Breen,3 James Samet,3 Robert Devlin,3 Wayne Cascio,3 David Diaz-Sanchez,3 Haiyan Tong3 1. National Exposure Research Laboratory, US Environmental Protection Agency, Research Triangle Park, NC, USA (breen.michael@epa.gov; isakov.vlad@epa.gov) 2. Institute for the Environment, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA (cseppan@email.unc.edu; sarav@email.unc.edu) 3. National Health and Environmental Effects Research Laboratory, US Environmental Protection Agency, Chapel Hill, NC, USA (Samet.James@epa.gov; breen.miyuki@epa.gov; Devlin.Robert@epa.gov; Cascio.Wayne@epa.gov; Diaz-Sanchez.David@epa.gov; Tong.Haiyan@epa.gov)

To better understand human exposure to air pollutants and their potential for adverse health effects, it is important to account for time spent in different indoor and outdoor locations, and the air pollutant concentrations in those locations. Currently available exposure models require substantial technical exposure modeling expertise, and near real-time exposure predictions are not possible since large and diverse input data must be collected, organized, and processed. To address these limitations, we developed TracMyAir, an iPhone application (App) that automatically estimates exposures (24 h average) to ambient PM2.5 and ozone based on several sources of input data available from iPhones, including: near real-time ambient air pollution measurements from local monitors, local weather, user's locations, and building characteristics of the user's home. TracMyAir uses an exposure model that accounts for time spent in different indoor and outdoor locations, and building-specific attenuation of ambient PM2.5 and ozone when indoors. In central North Carolina, we evaluated the App's ability to automatically obtain real-time measurements (1 h averages across previous 24 h) for (1) ambient PM2.5 and ozone concentrations from the nearest official air quality monitoring station, and (2) temperature and wind speed from the nearest weather station. We evaluated the App at various locations across the region, which has four air monitoring stations and two weather stations. The App successfully obtained real-time data from all the stations, and always selected the station closest to the iPhone location. We also performed a sensitivity analysis, which showed that the modeled exposures varied substantially with changes in daily home operating conditions (window opening, air cleaner usage), indoor-outdoor temperature differences, and time spent outdoors. The App is being applied for two epidemiological studies in central North Carolina. TracMyAir could also be used for developing public health strategies to help individuals at elevated risk to reduce their exposures.

Fusing CMAQ with Surface Observations to Improve Estimates of Air Quality & Health Impacts of Oct. 2017 California Wildfires

Stephanie Cleland, Marc Serre, Jacob Becker, Marissa DeLang, Jason West

While air quality models, such as the Community Multiscale Air Quality Modeling System (CMAQ), simulate the atmospheric physics and chemistry that affect air pollutants, geostatistical methods can combine modeled concentrations with observations to produce more accurate estimates of surface concentrations. Such methods have not been previously applied to episodic air pollution events like wildfires. Here we use geostatistical methods to estimate daily PM_2.5 concentrations across California during the October 2017 wildfires using both surface monitor observations and CMAQ output. The Bayesian Maximum Entropy (BME) Framework and the Constant Air Quality Model Performance (CAMP) Method are used to error correct CMAQ model estimations and fuse the CAMP-corrected CMAQ modeled PM_2.5 concentrations with surface PM_2.5 concentrations from both FRM/FEM and temporary monitoring stations. These daily PM_2.5 maps are then used to estimate several acute health impacts of the October 2017 wildfires.

Mapping Global Surface Ozone Concentrations through the Statistical Fusion of Observations and Models

Jacob Becker¹, Marissa DeLang¹, Kai-Lan Chang², Owen Cooper^2,3, Stephanie Cleland¹, Elyssa Collins¹, Marc Serre¹, J. Jason West¹

¹Department of Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, NC

²NOAA Earth System Research Laboratory, Boulder, CO

³Cooperative Institute for Research in Environmental Studies, University of Colorado, Boulder, CO

Tropospheric ozone is an EPA criteria pollutant known to be detrimental to human, animal and plant health. To support the 2019 Global Burden of Disease Assessment, we aim to produce fine resolution global surface ozone estimations for each year between 1990 and 2017. We use the Bayesian Maximum Entropy (BME) framework to integrate surface observations from 5,966 monitoring sites, provided by the Tropospheric Ozone Assessment Report database (TOAR), with seven global atmospheric chemistry models that provide information where monitoring networks are sparse. We produce ozone fields for the 6-month ozone season average of the 8-hr daily maximum ozone. Models were first combined using the M³Fusion method, which weights models based on their ability to predict observations in each region and year. The model combination was evaluated using the Constant Air quality Model Performance (CAMP) framework to remove bias from the model based on comparisons with observations. We estimated a global offset as the weighted sum of the bias-removed model and observations interpolated by exponential kernel smoothing. This offset was removed from observations and model values to lower variability and improve estimation during calculations. The observation data and the bias-corrected and weighted M³ model, both with the global offset removed, were combined using BME to obtain global ozone residual estimations. These residuals were added to the global offset to obtain the final BME estimation. The BME estimation matches observed ozone at measurement locations, with the influence of those observations falling off with distance from the monitoring sites based on the derived spatial and temporal correlation. Elsewhere, the BME estimation tends toward the global offset and M³Fusion output based on the uncertainty of those quantities spatially. After estimating ozone at 0.5 degree resolution, we added fine spatial detail based on the NASA G5NR-Chem model at 0.125 degree resolution, which was not included in the original M³Fusion model. Our final product estimates global surface ozone concentrations for each year at 0.1 degree resolution.

CMAQ-Enhanced Estimates of Personal Exposure to Diesel Particulates

Khanh Do¹, Haofei Yu², Cesunica Ivey¹

¹Center for Environmental Research and Technology, University of California Riverside

²University of Central Florida

In this work, we blend modeled source impacts and high-resolution personal PM_2.5 measurements to characterize exposure risk to diesel combustion sources in a heavily-burdened air basin. CMAQ-DDM is run for a 5-week period to generate diesel PM2.5 concentrations over Inland Empire, CA. Modeled diesel PM_2.5 estimates are then blended with high resolution personal exposure measurements made during the sampling period for 17 residents of the Inland Empire. Our diesel PM_2.5 exposure model takes into account time spent in each microenvironment, indoor and outdoor exposure differences, distance from the nearest major roadway, and measured PM_2.5 exposure. This robust approach leverages high resolution personal exposure measurements, chemical transport modeling, GPS data, and GIS tools to improve diesel PM exposure estimates. This same approach can be applied for any sources of interest. 

Fixed point versus time-activity based personal air pollution exposure assessment: Does the number of days of assessment matter

Eun-Hye Yoo, Xiangyu Jiang

Background: Time-activity based estimates of personal exposure to air pollution are generally recognized as superior to a fixed point (residence) approach. However, little is known about the impact of the number of daily location assessments used in the exposure estimation.

Objectives: This research is designed to determine if estimated levels of personal air pollution exposure based on a fixed point (residence address) and time-activity patterns (trajectories) varied depending on the number of days of assessment.

Methods: We estimated daily PM_2.5 concentrations at census tract levels over the Buffalo-Niagara region of western New York from January 3rd to 20th of 2014 by fusing ground observations from monitoring sites and CMAQ-modeled PM_2.5 concentrations. Time-stamped location data of 1,416 residents in the study region were retrieved from GPS-equipped iPhones. We also collected participants' home addresses. First, we estimated each participant's daily exposure to PM_2.5 based on their residence locations and trajectories, respectively, for 18 days. Second, we calculated the accumulated difference between each participant's residence- and trajectory-based exposure estimates at each day. Lastly, we detected breaking points in the accumulated differences using regression trees to identify assessment periods at which these exposure methods yielded substantially different exposure estimates.

Results: Daily exposure estimates from the two methods were similar in the range of 3.82 - 27.1 μg/m³ with the same mean of 12.6 μg/m³ and standard deviation (SD) of 4.75. The accumulated difference between two estimates for all participants increased from 0.28 (day 1) and 3.07 (day 18) with a mean of 1.73 and SD of 1.94. Analyses revealed substantial differences between their accumulated exposure estimates on the day 7 and day 16 for most participants.

Conclusions: Accounting for individuals' mobility is important for a proper assessment of personal air pollution exposure, but measuring exposure frequently enough to capture individuals' routine activity patterns is also necessary.

A comparison of land management modeling tools as used for estimating prescribed fire smoke impacts

Megan M. Johnson, Fiona A. Tennyson, Fernando Garcia Menendez

Prescribed burning is an important land management practice used across the U.S. by private and public landowners alike. While this technique is beneficial for many ecological and fuel management purposes, there are several perceived barriers to its use, including smoke management. Several modeling tools are available to prescribed fire practitioners to predict smoke dispersion from burn projects. The adoption of these by land managers depends on practical designs, fast simulation times and publicly available data sources. However, the tools vary in ease of use, especially for those not familiar with air quality modeling, and complexity, which may alter smoke plume predictions. There are few systematic evaluations and analyses of these tools. In this work, several modeling tools are evaluated for the purpose of estimating prescribed fire smoke impacts from the perspective of planning a burn. Using data from a year's worth of prescribed burning on North Carolina State Parks, we model and compare smoke plumes predicted with VSmoke-Web, HYSPLIT, and CALPUFF. Model output is compared using multiple metrics, including spatial plume overlap, population smoke exposure, and predicted PM2.5 impacts. The tools are further evaluated by considering ease of use for land managers when planning prescribed burns. This comparison of smoke dispersion from past burning activity provides insight into potential public impacts in areas that may frequently experience prescribed burning and also highlights the feasibility of using these modeling tools to support prescribed burning decision-making.

The Impacts of Prescribed Burning on PM2.5 Air Quality and Human Health: Application to Georgia, USA

Ran Huang, Yongtao Hu, Armistead G. Russell, James A. Mulholland, M. Talat Odman

Short-term exposure to fire smoke, especially PM_2.5, has been associated with adverse health effects. To quantify the impact of prescribed fire on human health, exposure fields of PM_2.5 from prescribed fires in Georgia, USA during the 2015-2018 burn seasons were generated combining air quality/sensitivity simulations with observations through a data fusion method. Then, a general health impact function was used with those exposure fields to estimate the health burden of prescribed fire. The sparsity of observation sites often leads to fire impacts on air quality remaining undetected. While those impacts can be estimated by air quality model simulations, various uncertainties in the models lead to inaccuracies. Application of a data fusion method to adjust modeled fire impacts with observations generally improved the exposure fields. However, in some cases, limited observational data reduced the impact of smoke plumes successfully captured in model simulations. The dearth of monitoring sites can be alleviated, in part, by using low-cost sensors. A data withholding evaluation using only sensors data showed that low-cost sensors could be used to provide spatial and temporal information missed by both regulatory monitoring sites and model simulations. A method has been developed to identify the days and areas when and where prescribed burning has a major impact on local air quality, and used in exploring the relationship between prescribed burning and acute health effects. The results showed a strong spatial and temporal variation of prescribed burning impacts. April 2018 had larger estimated daily health impact with more burned areas compared to Aprils in previous years, likely due to an extended burn season resulting from the need to burn more areas in Georgia. About 145 emergency room (ER) visits in Georgia were estimated for asthma due to prescribed burning impacts during the 2015 burn season. This number increased by about 18% in 2018. Although southwestern, central, and east-central Georgia have large fire impacts on air quality, the absolute number of estimated ER asthma visits resulting from burn impacts is small in those regions compared to metropolitan areas where population density is higher. Metro-Atlanta had the largest estimated prescribed burn-related asthma ER visits in Georgia with an average of 66 during the reporting years.

A Comparative Study of Two-way and Offline Coupled WRF and CMAQ over the continental U.S.: Performance Evaluation and Impacts of Chemistry-Meteorology Feedbacks on Air Quality

Kai Wang1, Yang Zhang1, Shaocai Yu2, David Wong3, Jonathan Pleim3, Rohit Mathur3 and James Kelly4

1Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC, 27695

2College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, Zhejiang, P.R. China, 310058

3Computational Exposure Division, National Exposure Research Laboratory, U.S. EPA, Research Triangle Park, NC, 27711

4Office of Air Quality Planning and Standards, U.S. EPA, Research Triangle Park, NC, 27711

The Community Multiscale Air Quality (CMAQ) modeling system developed by the U.S. EPA has been widely used by the research community to study various air quality issues such as tropospheric ozone (O₃) and fine particulate matter (PM_2.5) pollution, regional haze, and acid deposition and to provide scientific guidance to policymakers for air quality regulations. Traditionally CMAQ has been driven by 3-D meteorological field generated offline by meteorological models such as the Weather Research Forecasting (WRF) model. With the rapid development of computational resources and better understanding of meteorology-chemistry interactions, the two-way coupled WRF-CMAQ has been developed to better represent the real atmosphere by accounting for complex chemistry-meteorology feedbacks.

In this study, we present a comparative study of two-way (with consideration of both aerosol direct and indirect effects) and offline coupled WRFv3.4 and CMAQv5.0.2 over the continental U.S. Two sets of multiple-year (2008-2012) simulations using both versions of CMAQ are carried out with anthropogenic emissions based on the U.S. National Emission Inventories and chemical initial and boundary conditions derived from an advanced Earth system model (i.e., the NCSU's version of the Community Earth System Model). Our objectives are to 1) perform comparative model evaluations for key meteorological variables and chemical species between the two versions of CMAQ; and 2) investigate the impacts of chemistry-meteorology feedbacks on the U.S. air quality. The preliminary results show that both versions of CMAQ perform well for O₃ and PM_2.5. Domain-wide annual average modeled concentrations of CO, O₃, NO_x, anthropogenic VOCs, and PM_2.5 were reduced in the coupled WRF-CMAQ compared with the offline CMAQ by 3.1% (up to 27.8%), 4.2% (up to 16.2%), 6.6% (up to 50.9%), 5.8% (up to 46.6%), and 8.6% (up to 49.1%), respectively, partially due to reduced temperature and shortwave radiation. The representation of meteorology-chemistry feedbacks in the coupled model improved performance in terms of normalized mean bias and normalized mean error compared with the offline CMAQ model for O₃, sulfate, nitrate, ammonium, and black carbon. The improved performance for coupled WRF-CMAQ should be considered along with other factors in developing future model applications and policymaking.

Effects of grid resolution on the global mortality burden of fine particulate matter and ozone

Ruozhang Xu, J. Jason West

Human health impact assessments associated with exposure to air pollution often use air quality models to simulate exposure concentrations. However, models run at coarse resolutions can result in uncertainties in health impact estimates because of poorly aligning population and air pollutant concentrations, such as in urban areas. In this study, we aim to quantify the effect of grid resolution on estimates of global premature mortality attributable to ozone (O3) and fine particulate matter (PM2.5). We use output from a 2013-2014 simulation of global atmospheric chemistry from the NASA Goddard Earth Observing System Model version 5 Earth system model (GEOS-5 ESM) with GEOS-chem chemistry (G5NR-chem). The model was run at a fine resolution of about 12.5 x12.5 km ^2 and a coarse resolution of about 200 x250 km ^2. We estimate global premature mortality at both resolutions, based on the 6-month ozone season average of 8-hr daily maximum O3 concentrations and the annual average PM2.5 concentrations, to evaluate the bias due to the difference in resolution. We also re-grid the fine resolution to coarser resolutions through simple averaging of modeled concentrations and repeat the mortality calculation, to estimate the contributions to the overall bias at coarse resolution from model chemistry vs. exposure. The biases caused by coarse resolution and the contributions from model chemistry and exposure are evaluated globally and in each world region. Poster: Tammy Ruozhang Xu

A Quantitative Assessment of Vehicle-class and Precursor-specific Contribution from Onroad Emissions to PM2.5 and O3 Formation in the Northeast and Mid-Atlantic U.S.

Calvin Arter, Carlie Coats, Dongmei Yang, Sarav Arunachalam

In this study, we present a onroad transportation sector-specific modelling platform based on CMAQ-DDM to estimate the relationship between changes in transportation emissions at the county scale and air pollutants (O3 and PM2.5) in support of the Transportation and Climate Initiative (TCI) for a 12-state region in the Northeast U.S. and Washington D.C. This initiative seeks to reduce carbon emissions from the transportation sector in the TCI region through clean energy and fuels as well as other economic policies. While the main focus of the TCI is the reduction of carbon emissions, we are expanding this focus to quantify the impacts of these potential policies on air pollutant reduction from the transportation sector in downwind states. We instrumented CMAQ-DDM to enable the tracking of precursor-specific contributions from each of five vehicle classes of onroad vehicle emissions (motorcycles, passenger cars, light-duty trucks, heavy-duty trucks, and buses) in each TCI state to O3 and PM2.5 concentrations, and associated health effects in downwind states. We utilize the CMAQ-DDM sensitivity outputs to construct emission reduction scenarios equivalent to reductions from proposed investment scenarios for TCI. The reduction in pollutant concentrations will be subsequently used to assess potential benefits in health impacts from these proposed TCI investment scenarios by combining the concentrations with concentration response functions for PM2.5 and O3 with a focus on damage functions that are tied to specific combinations of vehicle classes and emissions precursors. Our results will provide quantitative evidence for the potential for clean transportation policies focusing on reducing greenhouse gas emissions in the region to provide additional health and economic benefits from reductions in criteria pollutants.

Implementation of a dynamical NH3 emissions parameterization in CMAQ for improving PM2.5 simulation in Taiwan

Chia-Hua Hsu, Fang-Yi Cheng

Department of Atmospheric Sciences, National Central University, Taiwan

Ammonia (NH3) is an important precursor of inorganic fine particulate matter (PM2.5). Without specific information on the temporal variation in NH3 emissions, a simplified temporal variation in NH3 emissions is often used in chemical transport models. To better characterize NH3 emissions in an air quality model simulation for Taiwan, a dynamical NH3 emissions parameterization based on Gyldenkurne et al. (2005) was applied to improve the temporal profile of NH3 emissions from livestock operations, synthetic nitrogen fertilizers, and standing crops. The Community Multiscale Air Quality (CMAQ) modeling simulation with a fixed NH3 emissions rate (the CONST experiment) presents a large positive bias in simulated nitrate and NH3, particularly during the nighttime and winter months. On the other hand, the CMAQ simulation with a dynamical NH3 emissions approach (the DYN experiment) improves the diurnal and seasonal variations and reduces the simulated bias. Moreover, according to the Taiwan emissions inventory, NH3 emissions from sewage accounts for a large portion (37%) of total NH3 emissions in Taiwan. The CMAQ simulation with the dynamical NH3 emissions approach and with a reduced level of NH3 emissions from sewage (the DYN1 experiment) was conducted to assess the possibility that the existing Taiwan emissions inventory may overestimate sewage NH3 emissions. The evaluations with observed NH3, nitrate, and ammonium wet deposition concentrations indicate that the DYN1 experiment performs better than the CONST and the DYN experiments.

U.S. Residential Wood Combustion (RWC) Survey: Results and Use in Improving RWC Emissions in the 2017 National Emissions Inventory

David Cooley and Jonathan Dorn, Abt Associates, Durham, NC

Paul Miller, Northeast States for Coordinated Air Use Management, Boston, MA

Rich Mason, U.S. Environmental Protection Agency, OAQPS, RTP, NC

Quantification of air pollutant emissions is key to the development of emission reduction initiatives in support of air quality targets and climate change mitigation. This project involved the design and implementation of a survey for the United States to collect data on residential wood use, including local-scale counts of appliances used and quantities of wood burned, to support improved estimation of PM_2.5 and black carbon emissions in North America. The results of this survey provide important information about the use of wood-burning appliances and the amount of wood burned throughout the United States. This information is being used to make detailed estimates of wood-burning activity and associated air pollution emissions, including BC. In the National Emissions Inventory (NEI), BC is computed as a speciation factor from inventory PM_2.5 and is thus directly correlated to PM_2.5 estimates. The improved activity data based on the appliance profiles and burn rates established through this survey are being used to improve estimates of PM_2.5, and hence BC, emissions. The data gathered under this project also highlight the residential wood combustion appliance types where uncertainty associated with PM_2.5 and BC emissions factors should be further evaluated.

EPA's National Emissions Inventory for Nonpoint Sources: Process and Improvements for the 2017 NEI

Jonathan Dorn, Hannah Derrick and David Cooley, Abt Associates, Durham, NC

Rich Mason and Jennifer Snyder, U.S. Environmental Protection Agency, OAQPS, RTP, NC

The U.S. Environmental Protection Agency's Emission Inventory and Analysis Group (EIAG) produces the National Emission Inventory (NEI), a comprehensive and detailed estimate of air emissions of criteria pollutants, criteria precursors, and hazardous air pollutants from air emissions sources. These data are needed by states to evaluate emissions trends in each state and to compare emission trends between states. The NEI is also used as a basis for various EPA modeling and regulatory analyses and to produce the National Air Pollutant Emission Trends report.

Abt Associates is currently supporting EPA with development of the National Emissions Inventory for nonpoint sources for the 2017 NEI. This paper will discuss the nonpoint source categories slated for inclusion in the 2017 NEI, the new process for developing emission estimation methodologies and a new integrated tool for developing state-level nonpoint emission estimates for submittal to EPA through the central data exchange.

Chemical Speciation of PM2.5 in Bogot, Colombia: Experimental Determination and Simulation with the Chemical Transport Model WRF-Chem

Sebastian Orlando Espitia Cano; Maria Paula Perez Pena; A.P Sullivan; Ricardo Morales Betancourt.

In Bogota, the capital of Colombia, the scarcity of studies on the chemical composition of PM2.5, has hindered efforts to properly attribute the contributions of various sources to particulate matter. Previous studies have analyzed the chemical composition of PM10, finding a dominant contribution of organic compounds and mineral material (Ramirez et al.2018). In this study, we determined the chemical speciation of PM2.5 samples collected in the city, and compared them with the results obtained in the simulation using the chemical transport model in line WRF-Chem v3.9.1. The modeling period, September 2018, was selected to coincide with the chemical speciation campaign. Different emission scenarios were configured in the model and compared with the observed chemical composition. Three nested domains were used with resolutions of 27 km, 9 km, and 3 km respectively and 41 vertical levels. Gas phase chemistry is described by the RACM mechanism, and aerosol physics by the MADE modal scheme. SOA formation is described with the Volatility Basis Set scheme. The anthropogenic emissions come from the global emissions inventory EDGARv4.3.1 and the local emissions inventory developed by the Secretaria Distrital de Ambiente of Bogota (2012). Biogenic emissions are also included using MEGAN. Several simulations were carried out. In the first one, the emissions of the EDGARv4.3.1 inventory were taken into account for the three domains, showing an underestimation of PM10 and PM2.5 concentrations in Bogota. In the second simulation, the local inventory was adapted to the Bogota domain, showing a strong overestimation of the concentrations of gaseous species and PM, this last may be due to the predominance of resuspended material over other sources. Finally, for the last simulation, the emissions of the gaseous species from the residential, industrial and transportation sectors of the EDGARv4.3.1 inventory were spatially located in the local inventory and resuspended material was reduce to 50%. PM, O3 and NOx were in good agreement with observations on average with the exception of SO2 and CO. Regarding the speciation of PM2.5, the samples collected in the city, showed that about 40% of PM2.5 is organic matter (OM), and 6% is elemental carbon (EC), while inorganic ions account for 10%. Modeled speciation of PM2.5 showed that 16%, 12% and 7% is OM, EC and inorganic ions respectively. In these simulations, resuspended PM2.5 emission is not spiced. We adapted the speciate of resuspended PM10 emission (Ramirez et al.2018) to PM2.5 emission. The simulation showed that 27% represent OM, 13% is EC and 9% is inorganic ions. None of the scenarios explored in the simulations could reproduce the significant contribution of organic aerosols to PM2.5. Even in scenarios with high amounts of gaseous organic precursors (VOC's), OM's contribution to PM2.5 was lower than observed. This suggests that there are significant primary sources of OM that are not being included. This may be due to the uncertainty that presents the speciation of emissions inventories.

Emissions of PCDD/Fs from 1960 to 2014 in China

Ye Huang^1,*, Min, Liu^1,*, Wenjun Meng², Shu Tao²

(1School of Geographical Sciences, East China Normal University, Shanghai, 200241; 2Laboratory for Earth Surface Processes, College of Urban and Environmental Sciences, Peking University, Beijing, 100871 *Corresponding author E-mail huangye@geo.ecnu.edu.cn

Persistent organic pollutants (POPs) are groups of pollutants widely concerned as they are semi-volatile, bioaccumulative, persistent and toxic. Unlike other POPs, Polychlorinated-p-dibenzodioxins and dibenzofurans (PCDD/Fs) are never intentionally produced. They are generated during fuel combustion or as byproducts during industrial processes. Emission estimation of PCDD/Fs helps in understanding their major sources and taking measures to control them. Previous studies have reported PCDD/Fs emissions for some years and for some sources like iron&steel industry, however, information about long term historical PCDD/Fs emissions is still scarce. In this study, PCDD/Fs emissions for 86 sources from 1960 to 2014 were estimated. The PCDD/Fs emissions increased from 2798 g in 1960 to 7331 g in 2014 with cement production, iron&steel industry and waste burning as the 3 largest sources contributed more than half of total emissions.

Hamilton Airshed Modelling System

Janya Kelly, Anthony Ciccone, Katie Armstrong

The Hamilton Airshed Modelling System (HAMS) was developed to provide a platform to better understand the processes and contributions to Hamilton's air quality and to inform future policy and human health impact decisions. As air quality in an urban airshed is influenced by many factors including local, regional, and transboundary sources, as well as by the geographical features and prevailing meteorological conditions, the CMAQ-WRF-SMOKE regional modeling platform was selected. The Hamilton airshed is part of the larger southern Ontario airshed, with influences from both southwestern Ontario and the United States. On a local scale, influences from transportation and industry are analyzed for their contribution. Analysis is focused on the source contribution of different emission source categories, as well as how the contribution varies across a local scale. Emphasis is focused on NO_X, SO₂, O₃ and particulate matter (PM_2.5 and PM₁₀), benzene and benzo(a)pyrene for health and regulatory reasons.

Using 15N in NOx emissions for source identification

Greg Michalski, Huan Fang, and Scott Spak

There remain significant uncertainties in NOx emissions at various scales.  A new way to evaluate NOx emissions is using stable N isotopes because different sources of NOx have different ¹⁵N/¹⁴N ratios. We have incorporated ¹⁵N into a emission inventory generated by the SMOKE model. The results show that there are significant differences in the isotope composition of NOx in space and time, suggesting ¹⁵N can be used to constrain the NOx budget.  These emissions will be incorporated into CMAQ and coupled to chemistry and transport in order to predict the ¹⁵N/¹⁴N ratios in atmospheric nitrate and compared to a growing data set of observations from the NADP and other networks.

Modeling and Evaluation of Quebecs Air Quality due to an Increase in the Electric Vehicle Fleet Combined with Off-Road Emission Reductions

Jessica Miville¹, Rabab Mashayekhi¹, Brett Taylor², Jacinthe Racine¹, Patrick M. Manseau¹, Annie Duhamel¹, Mourad Sassi¹, Calin Zaganescu¹, Sophie Cousineau¹, and Radenko Pavlovic^1
1 Air Quality Policy-Issue Response Section (REQA), Meteorological Service of Canada, Environment and Climate Change Canada
² Pollutant Inventories and Reporting Division, Science and Technology Branch, Environment and Climate Change Canada

In 2018, a standard came into effect in the province of Quebec, which states that car manufacturers are required to earn a certain number of credits through the sale and/or lease of zero and low emission vehicles. This standard, along with other social and economic incentives given to owners of electric vehicles (EV), were created with the aim of increasing the number of EV in the province. The government of Quebec's goal is to reach 100,000 electric and rechargeable hybrid vehicles registered in Quebec by 2020 and more than 300,000 electric vehicles by 2026. Many off-road equipment also have an electric equivalent, such as electric lawn mowers, and recreational vehicles. Due to the rising cost of fuel, it is logical to believe that consumers will opt for electric alternatives, as long as the price of the equipment is comparable. As such, this study's objective is to cover a quantitative analysis regarding the impact of electric vehicles and electric off-road on the air quality in the province of Quebec for the year 2025.

The Environment and Climate Change Canada (ECCC) regional air quality model AURAMS (A Unified Regional Air-Quality Modelling System) was used to model three scenarios: 1) a substitution of 250,000 conventional vehicles for EV in Quebec, 2) a reduction of certain off-road equipment by 40%, and 3) a combination of the previous two scenarios (250,000 EV + 40% reduction of certain off-road equipment). The scenario with reductions in the off-road sector showed higher reductions for both emissions and concentrations compared to the scenario with 250,000 EV, though the scenario that combines 250,000 EV + 40% off-road reduction showed the highest reductions for emissions and concentrations.

Modeling SOA Formation over Eastern China: Impacts of Updates in CMAQ v5.3b2 relative to CMAQ v5.1
Xiaoyang Chen1,2, Qi Fan2, and Yang Zhang1

1Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, 27695
2School of Atmospheric Sciences, Sun Yat-sen University, Guangzhou, 510275

Aqueous chemistry on the surface of aerosols and in cloud droplets has been found to be an important pathway for secondary organic aerosol (SOA) formation. The Community Multiscale Air Quality modelling system version 5.1 (CMAQ v5.1) includes two simple representations for aqueous SOA formation from dicarbonyls: the uptake onto cloud drops with a constant reaction rate of 0.04 (implemented in AQCHEM-KMT/KMTI aqueous chemistry module), and the uptake onto aqueous aerosols with a constant uptake coefficient of 0.0029 for both glyoxal and methylglyoxal (implemented into aero6i aerosol module). We previously conducted the simulations using the default and a revised version of CMAQ v5.1 to study variation of SOA in representative months of year 2014 over eastern China. In the revised CMAQ v5.1, we implemented a GEOS-Chem-based parameterization to account for the dependence of the aqueous aerosol uptake on aerosol acidity, relative humidity, and crystallization. However, both simulations largely underpredicted secondary organic carbon (SOC) (thus SOA and fine particulate matter (PM2.5)). In addition to other major updates in organic aerosol modules, CMAQ version 5.3 beta2 (CMAQv5.3b2) offers a detailed approach with extended 53 aqueous-phase reactions (implemented in AQCHEM-KMT2 module) for the in-cloud pathway. The uptake coefficient of methylglyoxal has also been reduced by 91% in aero6i/7/7i in CMAQ v5.3b2 based on recent laboratory studies compared to the value used in CMAQ v5.1. All these updates should help improve model's predictability for SOA and PM2.5. In this work, we will apply CMAQ v5.3b2 with both the simple and the detailed representations of the in-cloud SOA formation (AQCHEM-KMTI and AQCHEM-KMT2, respectively) over the same eastern China testbed for representative months in 2014 using the same model inputs as CMAQ v5.1. The two sets of CMAQ v5.3b2 predictions will be evaluated with available observations and intercompared with previous two sets of CMAQv5.1 predictions. These results will be used to assess (a) if the simple and detailed in-cloud representations lead to substantially different in-cloud SOA (AORGCJ) predictions and computational time, (b) if the updates in CMAQ v5.3b2 help improve overall model predictions, particularly in SOC/SOA and PM2.5, relative to CMAQ v5.1, and (c) how sensitive the aqueous SOA (i.e., AGLYJ) prediction is to the changes in uptake coefficient of dicarbonyls on aqueous aerosol. This work will shed light on the performance of the latest version of CMAQ, particularly the impacts of mechanism updates in CMAQ v5.3b2 on model predictions of O3, SOA, and PM2.5 over China where the air pollution is more severe than the U.S. and suggest additional areas of model improvement and further development.

Assessment of a weather pattern-dependent bias correction method to the PM2.5 prediction in Taiwan

Fang-Yi Cheng¹, Chih-Yung Feng², Zhih-Min Yang², Yi-Ju Lee¹, Chia-Hua Hsu^1,3, Shuenn-Chin Chang⁴, Yi-Chun Tsai⁴

¹Department of Atmospheric Sciences, National Central University, Taiwan ²Manysplendid Infotech, Taipei, Taiwan ³Department of Mechanical Engineering, University of Colorado, Boulder, CO, USA ⁴Environmental Protection Administration, Taipei, Taiwan

A real-time air quality forecasting (AQF) system using Weather Research and Forecasting (WRF) meteorological model and Community Multiscale Air Quality (CMAQ) model version 5.2 was developed in Taiwan. There is a systematic over-prediction of the wind speed in western Taiwan. The too strong wind can be attributed to both synoptic prevailing wind and the local sea breeze. The sulfate and organic carbon are under-predicted in western Taiwan while the nitrate is over-predicted in northern Taiwan. To improve the PM_2.5 forecast, a bias correction method based on the dynamic model output statistics (DMOS) is developed to reduce the predicted PM_2.5 bias. It was found that under the influence of the strong northeasterly monsoonal flow, AQF tends to underestimate the PM_2.5 concentration in southwestern Taiwan. On the other hand, when the synoptic weather condition is weak, there is systematic PM_2.5 over-prediction in northern Taiwan and under-prediction in southern Taiwan. As the PM_2.5 biases vary with the synoptic weather patterns, an alternative bias correction method to account for the influence of the synoptic weather types was developed (namely DMOS_W). A certain number of synoptic weather patterns were identified using the cluster analysis method. The bias correction method considers the systematic PM_2.5 biases associated with each weather type. The performance of the PM_2.5 prediction is further improved with the approach using the synoptic weather pattern based method (DMOS_W) compared to that from the original DMOS method. More details will be introduced during the workshop.

Performance evaluation of the updated and developed air quality forecasting system for South Korea.

Dae-Ryun Choi, Hui-Young Yun, Jeong-Beom Lee, Geon-Woo Yun, Hyun-A Oh, Ji-Seok Koo, Kyung-Hui Wang, Yoon-Seo Koo

The Korean AQFS forecasts air quality up to three days ahead of time, two times every day, and it has been in operation since 2007 (http://www.kaq.or.kr/). The initial configuration of Korean AQFS was run by MM5 v4.7 and CMAQ v.4.3 models, and used TRACE-P emission inventory for 2000 and the Korea emission data of CAPSS from 2003.

The forecast system was updated to WRF v3.1 and CMAQ v4.6 using Asian emission inventory of INTEX-B as well as the Korean emission data of CAPSS for 2007 in 2011.

In order to enhance the forecasting accuracy for PM, it is necessary to improve elemental technologies consisting of the forecasting system such as meteorological prediction model, emission inventories, observation and analysis methods, and data assimilation method. It is further required to develop the convergence forecasting method by integrating the key elemental methodologies mentioned above.

In this study, the yellow dust emission module, the inverse modeling to enhance the emission data, the data assimilation model, and the convergence model using the air quality measurements at the ground stations over China and Korea were developed and the KAQFS forecasts air quality using developed model up to five days ahead of time, two times every day, and it has been in operation since 2017.

Their PM forecasting performance tested in East Asia focusing on the Korea Peninsula for 2017. The performance of developed model forecasting showed better agreement comparing with based model in respect to Accuracy, Proability of Detection, False Alarm as well as NMB, IOA and RMSE in most regions in Korea for 2018.

Puget Sound Ports Air Quality Health Impacts Study

Mahshid Etesamifard, Joseph K. Vaughan, Yunha Lee, Brian K. Lamb

Laboratory for Atmospheric Research, Washington State University

The ports of Tacoma, Seattle and Vancouver B.C. created a join initiative to reduce air pollutant and greenhouse gas (GHG) emissions from port operations called Northwest Ports Clean Air Strategy (NWPCAS). The strategy sets overarching emission reduction targets for the ports in addition to activity based targets for each emission sector encompassed in port operations. The NWPCAS is getting updated for years 2020 and beyond. To decide upon the new targets, goals and methods of the new NWPCAS, the Northwest Seaport Alliance is looking into the impacts of port-related emissions in the Puget Sound Airshed. In this study we will provide a source apportionment analysis using alternate emission inventories that eliminate one source at a time. A high-resolution modeling framework applying CMAQ version 5.2, is used to assess pollutants contributions with the greatest health effects from five port-related emission sources. Concentrations are simulated for the Puget Sound Region on a domain, using 1.33 km x 1.33 km grid cells, centered on the Puget Sound Airshed. These simulations are conducted for a representative months from each season and extrapolated to the full year to report annual averages and maximum contributions for each source type.

Fine-scale meteorological modeling of the land-sea breeze circulation of Long Island Sound and surrounding coastal areas

Robert Gilliam, Ana Torres-Vazquez and Jonathan Pleim

In 2018, a comprehensive field study was conducted to examine the tropospheric ozone around the New York City area. The Long Island Sound Tropospheric Ozone Study (LISTOS) is a collaborative (Federal, State and Academia) observational campaign with modeling support, to better understand the processes behind what continues to be high observed surface ozone in the region. The main factors behind these high ozone episodes are the fact that the area has such high population and it has a unique coastline geography. Anthropogenic emissions are proportional to the sizable population and local land-sea circulations are driven by land-water temperature contrast and coastal geography. Local sea and land breeze circulations tend to recirculate pollution. Ozone and other pollutants have a tendency to concentrate along sea breeze fronts that align over high population areas and over adjacent bodies of water that remain stable with little vertical mixing. In this research we attempt to model this problem at a 1.33 km scale using the coupled WRF-CMAQ modeling system. The focus here is the performance of the WRF model in terms of replicating the land-sea breeze circulation. We present an overview of model performance for the LISTOS study period and a more in-depth evaluation for specific high ozone days. We use routine observations as well as more detailed observations from the field campaign for this evaluation.

Evaluation and Intercomparison of Modeled Atmospheric Deposition over North America and Europe An Overview of Phase 4 of the Air Quality Model Evaluation International Initiative

Christian Hogrefe, Stefano Galmarini, Johannes Bieser, Olivia Clifton, Jason Ducker, Lisa Emberson, Johannes Flemming, Christopher Holmes, Paul Makar, Martijn Schaap, Donna Schwede, and Sam Silva

This poster will present an overview of work planned under Phase 4 of the Air Quality Model Evaluation International Initiative (AQMEII). Since its inception in 2009, AQMEII has brought together a total of 37 modeling groups from 17 countries in North America and Europe to conduct coordinated research projects and model inter-comparison exercises aimed at advancing model evaluation practices and informing model development. The focus of Phase 4 is on assessing the ability of regional-scale air quality models to simulate dry and wet deposition of trace gases and aerosols. The poster will present the design of the coordinated model simulations, describe the common emission and boundary condition datasets to be used by all modeling groups, summarize the observational datasets to be used for model evaluation over North America and Europe, and discuss the data infrastructure aspect of the activity. Furthermore, we provide a description of planned analysis methods aimed at quantifying the impact of different process-level deposition algorithms and parameterizations on modeled deposition fields. Finally, we present a roadmap for how output fields generated by AQMEII Phase 4 participants could potentially be used in measurement-model fusion approaches for generating total deposition maps and in ecosystem impact assessment studies.

Evaluation of GFS-driven CMAQ predictions of PM2.5 and O3 at NOAA

Jianping Huang (1,2), Jeff McQueen (1), Perry Shafran (1,2), Ho-Chun Huang(1,2), Li Pan (1, 2), Youhua Tang(3,4), Patrick Campbell(3,4), Pius Lee(3), Ivanka Stajner(1), Jack Kain(1), Jose Tirado-Delgado(5,6), and Dorothy Koch (5).

(1) NOAA NWS/NCEP/EMC (2) IMSG (3) NOAA OAR/ARL (4) Cooperative Institute for Climate and Satellite, University of Maryland (5) NOAA NWS/STI (6) Eastern Research Group

NOAA's National Air Quality Forecasting Capability (NAQFC) has been providing numerical forecast guidance of surface ozone (O3) and particulate matter with diameter of 2.5 micro-meters ( m) or less (PM2.5) for over 15 years. Currently, the North American Model (NAM) Non-hydrostatic Multiscale meteorological Model with Arakawa B grid-staggering (NMMB) is used to drive the EPA Community Multiscale Air Quality (CMAQ) model for providing air quality numerical prediction in the NAQFC. The upgraded Global Forecasting System (GFS) V15 with Finite Volume Cubed-Sphere (FV3) dynamical core is being tested as a replacement to NMMB. NOAA's operational GFS V15 is running on a 13-km horizontal resolution. In this study, two one-month experimental predictions (for January and July 2019) with a grid-spacing of 12-km are completed to assess the performance of GFS/CMAQ for surface O3 and PM2.5 predictions in different seasons (i.e., winter vs. summer) and at different regions (e.g., Eastern US vs. Western US). The National Emission Inventory (NEI) with base year 2014 V2 (NEI2014) and the Blended Global Biomass Burning Emissions product (GBBEPx) are used to provide anthropogenic and fire emissions to drive CMAQ predictions. For this study the GFS-Aerosol predictions provide dynamic aerosol lateral boundary layer conditions for CMAQ. The experimental GFS/CMAQ predictions are compared with the operational NMMB/CMAQ predictions. The predictions from both modeling systems will be evaluated against the AirNow observational data. Preliminary results indicate that GFS/CMAQ improves the nighttime over-predictions of surface O3 especially over the the western US with overall performance. On the other hand, the GFS/CMAQ-predicted PM2.5 concentrations are more consistent with observations during nighttime, but show larger underprediction during daytime in summer as compared to the operational NMMB-CMAQ predictions. Furthermore, larger GFS/CMAQ over-predictions are found during the early morning as the boundary-layer develops. Detailed analyses will be presented to illustrate the causes for these GFS/CMAQ prediction biases.

Organic mass predictions during PM2.5 high pollution episodes in China and Korea: high resolution model experiments

Hyeon-Kook Kim¹, Eun-Ryoung Kim¹, Yesol Cha¹, Gang-Ho Bae¹, Jong-Jae Lee¹, Chang-Keun Song^1*, Myong-In Lee¹, Jin-Seok Kim², Jin-Soo Kim², Jung-Hun Woo², Jieun Park³, Youngkwon Kim³, Seung-Muk Yi³ ,Kwon-Ho Jeon⁴, Kwangjoo Moon⁴, Hyeok-Gi Chae⁴, Tang Wei⁵, Meng Fan⁵

¹Ulsan National Institute of Science and Technology (UNIST), School of Urban and Environmental Engineering, Ulsan, Korea

²Konkuk University, Department of Advanced Technology Fusion, Seoul, Korea

³Seoul National University, Graduate School of Public Health and Institute of Health and Environment, Seoul, Korea

⁴National Institute of Environmental Research (NIER), Climate and Air Quality Research Department, Incheon, Korea

⁵Chinese Research Academy of Environmental Sciences (CRAES), Beijing, China

*Correspondence to: Chang-Keun Song (cksong@unist.ac.kr)

Due to the serious concerns of health and environmental issues, improving the fine particulate matter (PM_2.5) predictions became very important in China and Korea, especially during the high PM_2.5 concentration period (hereafter, high episode). However, the state-of-the-art air quality models including the Community Multi-scale Air Quality (CMAQ) model tend to significantly underestimate PM_2.5 concentrations during high episodes. In a simulation over the Asian region with CMAQ version 4.7.1 for high episodes in October 2017, underestimations of organic materials (OM) were the dominant factor of the negative bias errors in the PM_2.5 predictions in the China and Korea monitoring sites. Here, we examine the potential causes of the OM underestimation for high episodes in October 2017 by conducting a series of high resolution (9km X 9km) model sensitivity test over the Asian domain. The focus of this test is on emissions (e.g., with vs. without biogenic VOC) and the choice of model versions (e.g., CMAQv4.7.1 vs. CMAQv5.x). The base case simulation system includes the Weather Research and Forecasting (WRF) Model version 3.9.1.1 (WRFv3.9.1.1)/the Sparse Matrix Operator Kernel Emissions for Asia version 1.3.0 (SMOKE-ASIAv1.3.0)/ CMAQv4.7.1. In both the base and other cases, the Comprehensive Regional Emissions inventory for Atmospheric Environment for the year 2015 (CREATE2015) is commonly used to derive gridded emissions of anthropogenic origins over most part of the Asian domain. Biogenic emissions were estimated by the Model of Emissions of Gases and Aerosols from Nature version 2.1 (MEGANv2.1). Starting from the base model case, we sequentially assess the effects of other model cases on the OM predictions and try to identify the causes of the OM underestimation for high episodes.

Acknowledgements

This research is supported by the National Research Foundation of Korea (NRF) grant funded by the Ministry of Education (NRF-2019R1I1A1A01045978) and the Korea-China Joint Research Team of National Institute of Environmental Research.

Changes in ozone production chemistry across the U.S. between 2007 and 2016: An integrated modeling assessment

Shannon Koplitz, US EPA

Heather Simon, US EPA

Kirk Baker, US EPA

Barron Henderson, US EPA

Luke Valin, US EPA

Benjamin Wells, US EPA

Anthropogenic nitrogen oxide (NOx) emissions in the U.S. have declined by ~60% over the last decade, altering the NOx-volatile organic compound (VOC) chemistry and ozone (O3) production characteristics of many areas. In this work we use multiple air quality analysis tools including satellite-derived estimates of column nitrogen dioxide (NO2) and formaldehyde (HCHO), surface observations of NOx and VOC species, and the high-resolution regional chemical transport model CAMx implemented with the Higher order Direct Decoupled Method (HDDM) to assess how these large reductions in NOx have affected ozone production regimes across the U.S. between 2007 and 2016. We first evaluate the ability of the CAMx model to capture the evolution of NOx and VOC concentrations over this time period. We then use sensitivity calculations from CAMx-HDDM combined with surface and satellite observations to characterize shifts in the variability of spatial and temporal (seasonal, weekly, and diurnal) patterns in NOx and VOC concentrations between 2007 and 2016 in different regions of the U.S. The analysis focuses particularly on three large urban areas with historically high ozone concentrations associated with complex patterns of NOx-VOC chemistry: Chicago, New York, and Los Angeles. This work 1) provides new insights into how O3 production chemistry is evolving in large population centers in the U.S., and 2) highlights areas where future scientific development is needed to improve model-based assessments. The goal is to demonstrate the utility of this type of analysis for informing the design of emissions controls strategies to most effectively reduce ozone concentrations in such areas.

Apportioning Emission Source-group Impacts among Individual Sources: Application to Prescribed Fires

Ran Huang, Momei Qin, Yongtao Hu, Armistead Russell, Talat Odman

Prescribed fire is a prominent source of PM_2.5 in the U.S. accounting for 14% of total primary PM_2.5 emissions in the 2014 National Emissions Inventory. To support concomitant fire and air quality management, we forecast prescribed fire activity and impact on PM_2.5 concentrations in the southeastern U.S. using the HiRes2 air quality forecasting system. However, since the HiRes2 forecast is the combined impact of all the fires in the region, it is not always useful for local land and air quality managers. Especially when there are too many fires close to each other, it is difficult to distinguish which ones have the larger impacts on air quality. Knowledge of individual fire impacts could help local land and air quality managers to decide which burns should be allowed or restricted for greater benefits on public health and air quality in areas of concern.

We developed a novel source apportionment method, Dispersive Apportionment of Source Impacts (DASI), and applied it to split the combined prescribed fire impact from a chemical transport model (CMAQ-DDM) by using concentration fields from a dispersion model (HYSPLIT). DASI is computationally much more efficient than simulating individual fire impacts with CMAQ-DDM. Comparisons of apportioned fire impacts with directly simulated individual fire impacts show that DASI performs well with large and small fires that do not have too much non-linear interaction with other fires. DASI could also be applied to conduct local-scale source apportionment for other source groups.

Long-range transport of air pollutants over Brazilian Southeastern: Interaction between metropolitan areas
Rizzieri Pedruzzi1,2*, Bok H. Baek2, Marcelo F. Alonso3, Taciana T. de A. Albuquerque1

1 Dept. of Sanitary and Environmental Engineering, Federal University of Minas Gerais - UFMG - Brazil
2 Institute for the Environment - University of North Carolina Chapel Hill
3 Faculty of Meteorology, Federal University of Pelotas - UFPel - Brazil

*Corresponding author: rizzieri.pedruzzi@gmail.com

Air quality is affected by meteorological conditions, topography, chemical process in atmosphere and emissions. Another atmospheric process which also affects air quality is the Long-range transport of air pollutants (LRTAP), which is responsible to carry on primary and secondary pollutants through long distances, affecting other areas which are not the emitters. The Brazilian Southeastern is the most populated area of Brazil, concentrating 42% of total population and 55% of gross domestic product, additionally, these areas are also surrounded by many industries as well as holds up to 12% of the national vehicle fleet. All these characteristics are deteriorating the air quality over the region, especially in the metropolitan areas. The metropolitan areas of Sao Paulo, Rio de Janeiro, Belo Horizonte and Vitoria are the greatest of each state of Sao Paulo, Rio de Janeiro, Minas Gerais and Esperito Santo, respectively, and they are not too far from each other, which means that LRTAP may play a role on air quality, transporting air pollutant from one metropolitan area to another. Even though there are air quality monitoring stations over Southeastern region, the number of stations is not enough to address the interaction between them, thus it is necessary rely on the air quality modeling. Therefore, the goal of this work is to develop a regional air quality modeling domain to cover all metropolitan areas in Brazil Southeastern and to understand the interaction between the metropolitan areas regarding their emissions. WRF-SMOKE-CMAQ system was used to develop the meteorological conditions, emissions and air quality modeling as a base scenario. Additionally, source apportionment using CMAQ-ISAM have been applied, aiming to address the interaction between the four metropolitan areas and to quantify the responsibility of each area on the air quality of the region.

Numerical evaluation of the impacts of emission and meteorological characteristics on air pollution in the Seto Inland Sea region, Japan

Hikari Shimadera¹, Shin Araki¹, Shota Iida¹, Tomohito Matsuo¹, Akira Kondo¹, Tatsuya Sakurai², Syuichi Itahashi³, and Hiroshi Hayami³

1 Osaka University

2 Meisei University

3 Central Research Institute of Electric Power Industry

The Seto Inland Sea (SIS) region is characterized by higher air pollution level compared to the other regions in Japan. The SIS region includes SIS with heavy ship traffic volume and its industrialized coastal areas with a number of large point sources. In addition, the SIS region is surrounded by mountains, which may suppress atmospheric ventilation. This study applied WRF v3.8 and CMAQ v5.2.1 to the SIS region in a year from April 2013 to March 2014 in order to assess the impacts of local emission and meteorological characteristics on air pollution. WRF-CMAQ successfully simulated higher SO2 and PM2.5 concentration levels in the SIS region than those in the surrounding regions. The impacts of local emission sources on air pollution were estimated with brute force zero-out runs for SOx emissions from ships and large point sources. The annual mean contributions of local SOx emissions to SO2 and PM2.5 concentrations in the SIS region were about 5 and 3 times higher than those in the surrounding regions, respectively. The impacts of local meteorological characteristic on atmospheric ventilation were evaluated with a run with constant passive tracer emission. The annual mean tracer concentration in the SIS region was about 1.3 times higher than that in the surrounding regions, indicating a lower ventilation efficiency in the SIS region. These results show that the local emission and meteorological characteristics mainly and partly contribute to higher air pollution level in the SIS region, respectively.

EPA-State 2016 Model Evaluation Forum

Heather Simon, Sharon Phillips, Norm Possiel, Zac Adelman, Alison Eyth, Kristen Foley, Wyat Appel, Shannon Koplitz, Christian Hogrefe

Over the past year there has been a collaborative effort between states, MJOs and EPA to develop a 2016 emissions modeling platform. A central purpose of this effort was to develop a common emissions platform that could be used by MJOs, states, and EPA as a starting point for various regulatory and non-regulatory applications. The beta version of the 2016 base year emissions was released in the spring of 2019. To support air quality modeling, EPA-OAQPS prepared 2016 data for other model inputs (WRF meteorology, initial and boundary conditions, etc.) and completed annual 2016 CAMxv6.5 and CMAQv5.2.1 model simulations using the beta emissions for 36 km and 12 km modeling domains. A central piece of the platform development effort is the evaluation of 2016 base year air quality model predictions using measured data. The results of the model evaluation are important for understanding the strengths and weaknesses of the platform which, in turn, help inform the use of the platform and guide research efforts to improve model inputs and science. In response to feedback from some MJOs and states, EPA has engaged broadly with these organizations to evaluate the 2016 platform air quality model predictions. The EPA-State 2016 Model Evaluation Forum was created with the goal of promoting collaboration between federal and state partners on characterizing and understanding model performance and identifying performance issues for possible further research by EPA and/or the modeling community. The forum is also intended to serve as a venue for sharing data and evaluation results. EPA has made public the model inputs, outputs and data analysis tools and graphics from the 2016 beta model simulations. This presentation will describe the model evaluation forum, ongoing projects and publicly available data and tools.

The Benefits of Using High-Resolution Sea Surface Temperatures for Simulating Historical and Future Climate Extremes
Jared Bowden, Tanya Spero, Kristen Foley, Anna Jalowska, Megan Mallard, Adam Terando, Ryan Emanuel, Jaime Collazo

Sea surface temperatures (SSTs) within global climate models (GCMs) are both biased and unable to properly resolve important SST features, such SST gradients along the coast and climatologically significant boundary currents. Traditional dynamical downscaling using regional climate models (RCMs) prescribes the GCM SSTs at the lower boundary; hence the RCM retains the GCM bias and is unable to resolve important SST features that influence both the mean climate and extreme weather. As computational resources increase, scientists are challenged with improving the fidelity of RCM simulations at higher-resolutions. High-resolution retrospective meteorological modeling studies have demonstrated that these higher-resolution SSTs can significantly improve the simulated meteorology. Unfortunately, the record length of the higher resolution SSTs (the 1-km Global High-Resolution SST available since 2011; GHRSST) is a limitation for high-resolution regional climate modeling needing decades of data.

Here we develop a methodology that produces a climatology-influenced high-resolution SST dataset for RCMs for retrospective (reanalysis simulations) as well as dynamical downscaling of GCMs for historical and future periods. The methodology also bias corrects the GCM data based on historical observations. These climatologically influenced, high-resolution SST datasets will be demonstrated with the Weather Research, and Forecasting (WRF) model at 12-km resolution to further understand the impact on simulating precipitation extremes.

Projecting Future Ground-Level Ozone Concentrations in Albuquerque, New Mexico

Kenneth Craig, Shih Ying Chang, Garnet Erdakos, Lynn Baringer

Sonoma Technology, Inc., Petaluma, CA

Ozone design values in Albuquerque, New Mexico, have decreased over the last 15 years, but have increased since 2016; the current ozone design value for Albuquerque (70 ppb) is on the cusp of exceeding the 2015 National Ambient Air Quality Standard (NAAQS). To support the Albuquerque Environmental Health Department with its air quality planning, base-case and future-year (2025) air quality modeling analyses were conducted for two high-ozone episodes in 2017. The future-year analysis involved projecting emissions from 2017 to 2025 under varying scenarios to assess how future ozone concentrations in Albuquerque could be impacted by (1) national, regional, and local changes in emissions that are expected to take place between now and 2025, and (2) changes in emissions as a result of an expanded Inspection and Maintenance program, tri-county emissions reductions, local peak-demand electrical generating units (EGUs) operating at permitted emission levels, and the electrification of the local gasoline vehicle fleet.

The modeling analysis was conducted with CAMx version 6.4 for three nested domains, including a 4-km domain covering New Mexico. Base-case (2017) emissions were based on the U.S. Environmental Protection Agency's (EPA) 2014 emissions modeling platform (version 7.2), with hourly 2017 continuous emission monitoring system (CEMS) data for EGU emissions, and day-specific 2017 fire emissions from the BlueSky Framework based on satellite hotspot and fire perimeter data. Base-case on-road mobile source emissions were projected to 2017 levels using scaling factors developed from MOVES to account for changes in vehicle activity, fleet turnover, and emission factors, with separate factors developed for Bernalillo County based on local fleet data. The approach to projecting the 2017 base-year inventory to year 2025 involved developing adjustment factors for each emissions source sector, based on differences between the 2017 base-year inventory and EPA's 2025 future-year inventory. Separate adjustment factors for New Mexico and Bernalillo County were developed for most sectors. The projections included assumptions regarding future EGU emission controls and unit shutdowns in New Mexico.

The modeling analysis showed that if projected reductions in local, regional, and nationwide emissions by 2025 materialize, these projected emissions reductions would reduce peak 8-hr average ozone concentrations by 3-7% at ozone monitoring sites in Albuquerque. For context, a 5% reduction of ozone concentrations by 2025 could reduce the future-year ozone design value in Albuquerque by 3-4 ppb. Future ozone concentrations in Albuquerque could increase by as much as 4 ppb if the two peak-demand EGU facilities in the Albuquerque area were operated at permitted emission levels during high-ozone episodes. The analysis also showed that emissions reductions in neighboring counties would be effective at reducing future ozone concentrations in Albuquerque. Finally, replacing the light-duty gasoline vehicle fleet with electric vehicles in Bernalillo County would reduce future ozone concentrations in Albuquerque by as much as 2 ppb. These results have important implications for air quality planning in the Albuquerque area.

Particulate Matter Emissions from Building Destruction in Armed Conflict Regions

Daniel Han

Particulate matter (PM) inhalation is a well-documented cause of long-term morbidity rates. An oft-studied source of PM is from fuel combustion in various contexts, but this proposed presentation will draw attention to a less considered source of PM: building collapse and destruction. While perhaps not as applicable to State Implementation Plans in the United States, this source of PM is more widespread in regions of the world experiencing armed military conflicts, evidenced by the large amounts of dust visible in many photographs and footage. Nevertheless, this long-term consequence of violent conflicts is often left by the wayside in the face of more immediate casualties and infrastructure needs like clean water and electricity.

Because of the lack of ground-level sensors in many armed conflict regions, this presentation will consider two different models. The first model is brought up in the literature with respect to the aftermath of 9/11 and is a simple back-of-the-envelope calculation using a direct proportion and constant of proportionality. The second draws on measurement techniques relying on satellite data, using aerosol optical depth (AOD) and other atmospheric and surface factors to predict PM levels. Simple yet reasonable numbers will be used to test model sensitivity along with subsequent estimates of the effects on life expectancy. Comparing these two models will not only illustrate the very real long-term health effects from this particular PM source, but also shows the challenges of and need for better measurement and regulation.

Influence of the emission sources on O3 and PM2.5 concentrations in Taiwan using CMAQ-HDDM

Yi-Ju Lee and Fang-Yi Cheng

In Taiwan, the major anthropogenic emissions are from metropolitan cities, coal-fired power plants, industrial parks, vehicle exhausts and burnings of agriculture wastes. According to previous studies, local emissions will result in severe pollution events, especially under the weak synoptic weather condition.

Sensitivity analysis methods, such as Brute Force Method (BFM) and High-order Decoupled Direct Method (HDDM), provide the relationship between the emission sources and the ambient air pollutant concentrations. This study used CMAQ version 5.0.2 with HDDM to investigate the impact of different emission sources on PM_2.5 and ozone (O₃) concentrations. The meteorological and emission data were from WRF two-nested domains and Taiwan Emission Data System version 9.0 (TEDS9.0), respectively.

The study episode (Nov. 6-8, 2018) was associated with a weak synoptic weather condition and the stagnant wind trapped the air pollutants near the emission source region. During the daytime, reducing the NO_X emissions from both point and mobile sources can increase O₃ concentrations near the emission source regions and the downwind area due to the decrease of titration effect, and reduce the PM_2.5 concentrations in the inland regions. During the nighttime, the PM_2.5 increase in coastal areas that can be due to the increase of O₃ concentrations. On the other hand, reducing the VOC emissions from both mobile and area sources can reduce O₃ concentration, especially in metropolitan cities. Reducing the SO₂ emission from point sources can effectively decrease PM_2.5 concentration in central Taiwan.

SCICHEM model performance evaluations for near-field regulatory applications

Ryan Cleary², R. Chris Owen¹, Clint Tillerson¹, George Bridgers¹, Chris Misinis¹, James Thurman¹

¹US EPA, Office of Air Quality Planning & Standards, RTP NC 27711

²Wood, RTP, NC, 27711

Major source permitting programs require new sources to demonstrate that any additional emissions will not cause or contribute to a violation of the National Ambient Air Quality Standards (NAAQS). The US EPA's Guideline on Air Quality Models (Appendix W to 40 CFR Part 51) currently specifies AERMOD as the preferred model for projecting near-field dispersion of emissions for most permit applications. SCICHEM is a Lagrangian puff model that uses three dimensional Gaussian puffs to determine concentration fields. Version 3.0b2 has two modes of operation, one focused primarily on dispersion with some limited-chemistry and the second is a full-chemistry mode. The full-chemistry mode requires detailed inputs of three-dimensional wind and chemical speciation fields (e.g., gridded output from an Eulerian chemistry model), while the other mode can be run with meteorological data at a single location and requires a limited set of spatially uniform background concentrations. The limited chemistry mode also supports meteorological and emissions input files that are in AERMOD equivalent formats. The limited-chemistry mode addresses modeling scenarios similar to those for which AERMOD is typically applied. To date, there has been little evaluation of SCICHEM for near-field impacts. As a result, its performance with NSR/PSD facilities and applications is generally unknown. In the study presented here, we evaluate SCICHEM using several field studies that have also been used in AERMOD near-field evaluations. SCICHEM performance is benchmarked against both the field measurement data and the AERMOD results. The results of this study will provide technical support in the determination of the suitability of SCICHEM to estimate concentrations for permit applications and help provide guidance for the appropriate application of the model for various source types.

Atmospheric Monitoring of Methane Leaks from Shale Oil and Gas Production using Airborne Near-Infrared Spectrometer

Saikat Ghosh, Chester Akers, Kevin Crist

Due to the distributed nature of shale production over wide geographic regions, on-site monitoring for emissions has been difficult and expensive. Emission estimates from these operations have been mainly obtained with emission factors applied to production. Remote sensing using aircraft offers an opportunity to monitor wide geographical areas in a cost-effective manner. Accessibility to remote areas and ability to cover large geographical regions in less time are some key advantages of detecting methane leaks from aircraft over land-based measurements. Airborne methane measurements were collected by flying an aircraft installed with a near-infrared spectrometer over the Marcellus shale development in Ohio. The spectrometer captures the absorption peak of methane by measuring the reflected solar radiation in the range of 1.61-168 m. The absorption level detected by the spectrometer corresponds with the concentration of methane in the atmospheric column below the aircraft. Methane concentration was modeled via a retrieval algorithm with a radiative model (MODTRAN) taking into account variations in meteorological parameters such as albedo, temperature, and pressure. This study presents geo-located spectral data from three flight experiments in Ohio. Column concentrations of methane calculated using MODTRAN shows the peak emissions near active source regions. This study demonstrates the application of airborne spectral measurements as a next-generation tool for air quality monitoring.

Community Participatory Mapping Crowdsourced Platform for Monitoring Air Pollution

Wansoo Im, Ph.D. (Meharry Medical College), Paul D. Juarez, Ph.D. (Meharry Medical College)

In previous years, South Korea struggled with severe air pollution at high levels. What sources exactly contribute to the high level of particulate matters (PM), which can be detrimental to our health is largely undetermined. The causes of PM air pollution are very complicated as there are external and internal factors. Investigating into these causes is still ongoing; however, the major culprits are automobile traffic, and energy & industrial activity from China. There is no clear understanding of which sources exactly contribute to PM2.5 in South Korea.  In South Korea, there were not enough monitoring locations to provide PM data to a specific location. Most recorded and analyzed data are from a limited number of monitoring stations and do not reflect data from where most people reside. Recent debates center around conflicting characterizations of air pollution that result from the relying on government air pollution measurements versus citizen reported air pollution measurements.  

The Community Mapping Center in S. Korea has developed DIY(Do It Yourself) PM monitoring kits (indoor and outdoor), and is currently working on developing the Particulate Matter Citizen Information Network (PMCIN) to educate the public and better understand what contributes to the levels of fine dust in the air. PMCIN is an open data platform that connects community collected air pollution data with existing air pollution monitoring data, and leverages existing infrastructure and research. The portal is accessible to the public and the interactive map has relevant GIS layers such as wind, air pressure, and traffic and industry point data. The Community Mapping Center engages the public from the outset, involving them in the collection and use of PM data to encourage grassroots public health campaigns for air pollution mitigation in South Korea. The cases of community mapping projects produced detailed spatial levels of PM air pollution will be presented. 

Ambient measurements of biogenic VOCs near Cannabis facilities in Denver, CO

Chi-tsan Wang, Kirsti Ashworth, Christine Wiedinmyer, John Ortega, Peter Harley, Quazi Z. Rasool, William Vizuete

Colorado was one of the first US state to legalize the industrial-scale cultivation of Cannabis spp. for recreational purposes. In 2017, there were 600 indoor cannabis cultivation facilities (CCFs) in operation in the greater Denver area with a recorded 550,000 mature plants (higher than 8 inches) under cultivation. Indoor measurements in CCFs have shown high concentrations of a group of reactive compounds, monoterpenes (C10H16). Laboratory measurements by our team of emissions from four cannabis strains identified a large variety of these compounds. We applied our measured emission rates to develop a cannabis emission inventory for Denver County, estimating an additional emission of 36-362 tons/year of monoterpenes being emitted over baseline emissions. There have been limited studies, however, that have measured for these compounds outside of CCFs to help support these findings. We therefore conducted a field campaign in August 2016 in Denver County focused on six different CCF clusters near the intersection of interstate highways I-25 and I-70. At each site, we collected air samples on 16-32 cartridges designed to adsorb monoterpenes. Sampling locations were selected to capture both spatial gradients and temporal changes in concentrations. A total 162 ambient air samples were analyzed by GC-MS/FID to identify and quantify monoterpenes. Our results showed that the monoterpene mixing ratios near CCFs ranged from 200-800 ppt. These levels are 4-8 times higher than samples collected from a "background" site located at the Denver City Park (50-100 ppt). Total monoterpene concentrations peaked at 8:00-9:00 AM when mixing ratios reached as high as 800 ppt. That is 60% higher than the predicted 500 ppt projected when the maximum CCF emissions of 362 tons/year were included in the Colorado regulatory air quality model (AQM). Even during the daytime at the height of photochemical destruction between 1:00-2:00 PM, monoterpenes concentrations were as high as 200-400 ppt, which is of the order of double the maximum predicted value of 150 ppt in the AQM. Our field sampling also found that the composition of monoterpenes near CCFs was dominated by d-limonene (30%), -myrcene (20%), and alpha-pinene (15%), similar to what was found in our laboratory emission factor study. These ambient sampling results are consistent with observations from the laboratory, and also suggest that monoterpenes from CCFs may be present in sufficient quantities to impact ozone and secondary organic aerosol formation in Denver.