Estimating Social Impacts of Air Pollution Using Regression (EASIUR): A Reduced-Complexity Model Derived from CAMx

Peter Adams

Jinhyok Heo

Current methods of estimating the public health effects of emissions are computationally too expensive or do not fully address complex atmospheric processes, frequently limiting their applications to policy research. Using a reduced-form model derived from tagged chemical transport model (CTM) simulations, we present PM2.5 mortality costs per tonne of inorganic air pollutants with the 36 km - 36 km spatial resolution of source location in the United States, providing the most comprehensive set of such estimates comparable to CTM-based estimates. Our estimates vary by 2 orders of magnitude. Emission-weighted seasonal averages were estimated at $88,000 130,000/t PM2.5 (inert primary), $14,000 24,000/t SO2, $3,800 14,000/t NOx, and $23,000 66,000/t NH3. The aggregate social costs for year 2005 emissions were estimated at $1.0 trillion dollars. Compared to other studies, our estimates have similar magnitudes and spatial distributions for primary PM2.5 but substantially different spatial patterns for precursor species where secondary chemistry is important. For example, differences of more than a factor of 10 were found in many areas of Texas, New Mexico, and New England states for NOx and of California, Texas, and Maine for NH3. Our method allows for updates as emissions inventories and CTMs improve, enhancing the potential to link policy research to up-to-date atmospheric science. This presentation will also describe a source-receptor version of this model.

Sensitivity analysis using the Community Multiscale Air Quality (CMAQ) adjoint to study health and ecosystem impacts of ground-level ozone

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

Ground-level ozone is considered to have a significant effect on risk of death from respiratory and cardiopulmonary causes, and the effect of long-term exposure to ozone on air pollution-related mortality has been quantified [Jerrett et al., 2016]. Ozone is also a pollutant that causes reductions in vegetation biomass, or potential productivity losses. Exposure-response models are established for 5 crop species and 11 tree species using a concentration-based metric, W126 [Lehrer, 2007]. This study will assess the effect of ozone on respiratory illness mortality as well as its effect on potential productivity losses (PPL) of select crops and trees.

Sensitivity analysis is necessary to understand the effects of ozone precursor emissions on mortality and ecosystem impacts. The adjoint method efficiently estimates sensitivities of concentration-based metrics with respect to all input parameters simultaneously, which is especially helpful for models with orders of magnitude more input parameters than output parameters, such as the Community Multiscale Air Quality (CMAQ) modeling platform. After benchmarking the adjoint sensitivities against finite difference sensitivities, the adjoint of CMAQ will be applied to the continental U.S. to analyze the sensitivities of respiratory illness mortality and PPL with respect to ground-level ozone concentration as well as emissions and concentrations of the most relevant ozone precursors, nitrogen oxides (NOx) and volatile organic compounds (VOCs). This study provides information to support the development of the most efficient control strategy of ground-level ozone with respect to different locations and species.

This study is performed for the continental U.S. from May 1st to August 31st, 2007. The horizontal resolution of the model is a 36 km 36 km resolution.

InMAP: A reduced-complexity model for air pollution interventions

Christopher W. Tessum, Jason D. Hill, Julian D. Marshall

Mechanistic air pollution modeling is essential in air quality management, yet the extensive expertise and computational resources required to run most models prevent their use in many situations where their results would be useful. Here, we present InMAP (Intervention Model for Air Pollution), which offers an alternative to comprehensive air quality models for estimating the air pollution health impacts of emission reductions and other potential interventions. InMAP estimates annual-average changes in primary and secondary fine particle (PM_2.5) concentrations - the air pollution outcome generally causing the largest monetized health damages - attributable to annual changes in precursor emissions. InMAP leverages pre-processed physical and chemical information from the output of a state-of-the-science chemical transport model and a variable spatial resolution computational grid to perform simulations that are several orders of magnitude less computationally intensive than comprehensive model simulations. In comparisons run here, InMAP recreates comprehensive model predictions of changes in total PM_2.5 concentrations with population-weighted mean fractional bias (MFB) of 17% and population-weighted R² = 0.90. Although InMAP is not specifically designed to reproduce total observed concentrations, it is able to do so within published air quality model performance criteria for total PM_2.5. Potential uses of InMAP include studying exposure, health, and environmental justice impacts of potential shifts in emissions for annual-average PM_2.5. InMAP can be trained to run for any spatial and temporal domain given the availability of appropriate simulation output from a comprehensive model. The InMAP model source code and input data are freely available online under an open-source license.

Evaluation of the Community Multiscale Air Quality (CMAQ) Model Performance during Persistent Cold Air Pool Events

Xia Sun¹, Heather A. Holmes¹, Cesunica E. Ivey²

1. Atmospheric Sciences Program, Department of Physics, University of Nevada, Reno, NV, USA

2. Chemical and Environmental Engineering, University of California, Riverside, CA, USA

Persistent cold air pools (PCAPs) are associated with stagnant conditions that last for days or even weeks and have been identified in many mountainous areas worldwide. The stable atmospheric stratification can lead to air pollutant accumulation due to limited vertical mixing. The PM_2.5 concentration reached 72 μg m^-3 during one PCAP that occurred in Salt Lake Valley (SLV), Utah in the winter of 2010 to 2011. This value is more than two times of the 24-hr average National Ambient Air Quality Standard (NAAQS, 35 μg m^-3). The elevated air pollutant concentrations can cause harmful effects on the environment and human health. Thus, it is important to have realistic simulations of air quality not only for the State Implementation Plan (SIP) development, but also for public health alerts.

Meteorology fields, especially those in the planetary boundary layer (PBL), play an important role in predicting air quality in chemical transport models. We present an inter-comparison of the Community Multiscale Air Quality (CMAQ) model v5.2 simulations using input meteorology from the Weather Research and Forecasting (WRF) model with different PBL parameterizations. Three CMAQ model simulations were conducted utilizing outputs from WRF with the Asymmetric Convective Model v2 (ACM2), Yonsei University (YSU), and Mellor-Yamada-Janjic (MYJ) PBL schemes for January 2011, when five PCAPs (including three strong ones) were identified in SLV, Utah. The other model configurations and options were kept the same. The simulated primary air pollutant concentrations (e.g., PM_2.5, PM₁₀, O₃) were compared with observations. The CMAQ model sensitivity to WRF PBL parametrizations during PCAPs was investigated.

Evaluating reduced-form modeling tools for simulating annual average PM2.5 impacts

Kirk Baker, Heather Simon, Gobeail McKinley, Neal Fann, Elizabeth Chan

Reduced-form modeling approaches are an increasingly popular way to rapidly evaluate air quality and human health impacts from changes in air pollution. However, comprehensive comparison of the predictive capability of these tools compared with complex photochemical grid models for different types of emission control scenarios is lacking in literature. Such comparisons of air quality and health impacts will inform our understanding of the strengths and limitations of reduced-form approaches and provide insights into the situations where reduced-form approaches may be most appropriate and useful. There are several publicly available reduced-form modeling approaches that have been developed to provide estimates of annual PM2.5 and monetized health benefits. Here, we compare the Intervention Model for Air Pollution (InMAP) and Air Pollution Emission Experiments and Policy Analysis (APEEP) reduced form models against photochemical grid models to estimate change in annual average PM2.5 for multiple EPA sector-based regulatory policy scenarios focused on on-road mobile and electrical generating units (EGUs). Reduced-form model predictions of annual average PM2.5 were similar for certain areas of the U.S. but were notably different for many areas in both the eastern and western U.S. This variability in predicted air quality was seen in each of the different emission control scenarios. These spatial discrepancies in the predicted locations of PM2.5 changes mean that the air quality impacts of some region-specific control plans may not be realistically approximated by each of these tools.

Using Photochemical Models to Assess the Exceptional Event Rules Q/D Screening Guidance

Matthew Alvarado, Benjamin Brown-Steiner, Chantelle Lonsdale, and Jennifer Hegarty

Atmospheric and Environmental Research, Lexington Massachusetts

The Exceptional Events Rule determines the conditions under which the US EPA will forgo comparison of policy-relevant air monitoring data to a relevant National Ambient Air Quality Standard (NAAQS). Currently, a "Q/D" approach is used to screen for the impacts of biomass burning on ozone (O₃), where the emissions of NO_x and VOCs are summed (Q) and divided by the straight-line distance of between the fire and the monitor (D). In this work we used the ASP and STILT-ASP Lagrangian photochemical models and the CAMx Eulerian photochemical model to simulate the O₃ formation in biomass burning plumes during an exceptional event in El Paso, Texas on June 21, 2015. ASP and STILT-ASP were also used to examine two events where Yucatan fires impacted O₃ in Houston. The model simulations generally show O₃ increasing with time and distance downwind and that the straight-line distance is frequently a poor proxy for the travel distance of the wildfire emissions from source to receptor, suggesting that the Q/D metric has difficulty screening for the potential impacts of biomass burning on ozone. For example, analysis of the CAMx results for the El Paso exceptional event shows that average fire impacts on MDA8 O₃ tend to increase with distance within 200-300 km of the fire, with some evidence of the Hog fire VOC emissions causing additional O₃ production due to interaction with the anthropogenic NO_x in the El Paso-Ciudad Juarez urban area. We also show that the Q/D metric is inconsistent with most reports of O₃ formation from biomass burning in the scientific literature. We present an alternative, literature-based screening approach that uses published ratios of biomass burning O₃ enhancement to the enhancement of CO and/or NO_y in the plumes and discuss the strengths and weaknesses of this approach.

Evaluating health benefit estimates from reduced-form modeling tools

Neal Fann, Kirk Baker, Elizabeth Chan, Heather Simon, and Gobeail McKinley

Beginning in the 1990s, the U.S. EPA began using so-called "reduced-form" techniques to more quickly evaluate the estimated incidence and economic value of health endpoints resulting from air quality changes due to environmental policies. Over the past decade, interest in these approaches has grown in the research community and there now exist several publicly available reduced-form modeling tools that model some combination of air quality and health impacts. The U.S. EPA recently began a project to critically assess alternative reduced-form tools, evaluating the sensitivity of these models to factors including the level and distribution of precursor emissions or the sectors affected. In conjunction with an analysis of air quality modeling also presented in this session, we evaluate four tools: the U.S. EPA's source apportionment (SA) Benefit per Ton; the Intervention Model for Air Pollution (InMAP); the Estimating Air pollution Social Impact Using Regression (EASIUR); and, two versions of the Air Pollution Emission Experiments and Policy Analysis (APEEP) models (AP2 and AP3). We assessed (1) the extent to which each tool yields estimated health benefits similar to those generated by "full-form" techniques; and, (2) the types of policy questions each tool is best able to answer. We find considerable national and regional differences in the size and distribution of benefits quantified by each tool across policy scenarios affecting point and mobile emission sources. These results suggest that the utility of each tool will depend in part on the sector to be assessed.

Contributions of International Emissions to Ozone Attainment in the United States

Maria Zatko and Ralph Morris

With the lowering of the Ozone National Ambient Air Quality Standard (NAAQS) to 75 ppb in 2008 and 70 ppb in 2015, the contributions of background ozone is of increasing importance. Ozone exceedances are due to local sources, transport from upwind states, transport from international sources and global background natural sources (e.g., stratospheric ozone and biogenic emissions). EPA has the authority to control mobile source emissions and has national programs that reduce emissions from mobile sources and other source categories. In addition, under Section 110 of the Clean Air Act Amendments (CAAA) there is a "good neighbor" provision that allows EPA to mandate control on upwind states that have been demonstrated to contribute significantly to nonattainment in a downwind state (e.g., Cross-State Air Pollution Rule). When developing State Implementation Plans (SIPs) to demonstrate ozone attainment, State's typically implement controls on local sources that they have jurisdiction over (e.g., local industrial facilities) but rely on EPA controls for mobile sources and for reducing ozone transport from U.S. sources. Some areas may have large contributions from sources outside of the U.S. (international sources), which can make demonstrating attainment of the ozone NAAQS difficult. In this study we examined the contribution of international emissions to ozone attainment across the U.S. The GEOS-Chem global chemistry model was run for 2011 base case conditions to provide boundary conditions (BCs) for CAMx regional photochemical grid model 2011 base case simulations using 36 km resolution continental U.S. (CONUS) and 12 km resolution western U.S. (WESTUS) domains. GEOS-Chem was then run for a 2011 no-international emissions scenario that eliminated all global non-U.S. anthropogenic emissions. The GEOS-Chem no-international emissions results were used to provide BCs for a CAMx 2011 36/12 km simulation that also eliminated all anthropogenic international (Mexico and Canada) emissions. The results from the 2011 base case and no-international emissions scenarios were processed to estimate the contributions of international emissions to 2011 ozone Design Values across the U.S. International emissions have a much greater contribution to ozone in the western U.S. than the eastern U.S., especially at high terrain locations. In some cases, the elimination of the contributions of international emissions is sufficient to attain the ozone NAAQS, which could be part of a "but for" ozone attainment demonstration allowed under section 179B of the CAAA.

Representing the nonlinear responses of tropospheric ozone and fine particulate matter to precursor emission changes in a response surface model with polynomial functions

Jia Xing^1,2, Dian Ding¹, Shuxiao Wang^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 The 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

Quantifying the relationship between the responses of tropospheric ozone (O3) and fine particulate matter (PM2.5) to precursor emission changes is a prerequisite for developing an accurate and optimized control strategy to achieve pollution and climate targets. In this study we developed a new set of polynomial functions in a response surface model (hereafter called "pf-RSM") to represent the responses of ambient O3 and PM2.5 concentrations to changes in the precursor emissions of NOx, SO2, NH3, volatile organic compounds (VOC) + intermediate IVOC (denoted as "VOCs"), and primary organic aerosol. Forty training samples with marginal processing are recommended to meet the criteria of a mean normalized error within 2% and a maximal normalized error within 10%. The out-of-sample validation suggested that the pf-RSM-predicted PM2.5 and O3 responses were in good agreement with a Community Multi-scale Air Quality Modeling System simulation over a large spatial and temporal scale, with normalized errors lower than 5.6% and 2.0% for PM2.5 and O3 respectively, across the domain, and within 12.7% and 6.5% respectively, throughout the simulated period. Indicators used to represent nonlinearity in PM2.5 and O3 responses to precursor emission changes were calculated by using the new pf-RSM.

Ammonia Measurements from the SNPP CrIS Instrument: Validation and Applications

Karen Cady-Pereira, Mark Shephard, Enrico Dammers, Cynthia Whaley, Paul Makar, Shailesh Kharol, Chantelle Lonsdale, Rui Wang, Matthew Alvarado, Mark Zondlo

We will very briefly describe the characteristics (sensitivity, information content, error estimates) of the satellite-based NH3 retrieval obtained from the Cross-Track Infrared Sounder (CrIS) instrument using our CrIS Fast Physical Retrieval (CFPR) algorithm (Shephard and Cady-Pereira, 2015). CrIS was launched on 28 October 2011 on board the USA NOAA/NASA/DoD the polar-orbiting Suomi National Polar-orbiting Partnership (NPP) satellite and provides twice daily (~01:30 and 13:30 standard local time) global observations. The CrIS instrument has low noise and an ideal daytime observation time, and thus has greater sensitivity to near surface NH3 concentrations than other currently flying TIR instruments. We will then present evidence that CrIS NH3 compares well against ground-based observations that include: (i) profiles and column measurements from FTIR sites in the Network for the Detection of Atmospheric Composition Change (NDACC) around the globe (Dammers et al., 2017), (ii) surface in-situ Ammonia Monitoring Network (AMoN) observations across North America, including 3 Canadian Air and Precipitation Monitoring Network (CAPMON) sites, and (iii) aircraft observations from the California and Colorado DISCOVER-AQ campaigns.

Having established the characteristics of CrIS NH3, we will demonstrate several applications:

- determining the contributions of different sources to ambient ammonia in the Athabasca Oil Sands and north-western Canada by comparing the GEM-MACH output against CrIS NH3 (Whaley et al., 2017)

- constraining NH3 emissions at the resolutions typical of regional air quality modeling (e.g.. CMAQ) using a finite-difference mass-balance approach using CrIS NH3 observations from the SENEX campaign

- computing the dry deposition flux of reactive nitrogen from ammonia across North America (Kharol et al., 2017)

- deriving NH3 emissions from the 2016 Fort McMurray fires from the CrIS observations (e.g. Shephard et al., 2017)

An operational version of the CrIS CFPR research algorithm has been implemented at the NASA SNPP SIPS and is now the providing the official NASA CrIS NH3 product. We will conclude our presentation with some clips of the seasonal variability over regions with strong NH3 signals, e.g., the US Midwest, Central Africa and northern India.

A new reduced-form model based on SRSM and DDM for efficient uncertainty analysis of Atmospheric Chemical Transport Models

Zhijiong Huang

Uncertainty analysis is an effective means to quantify the uncertainty in atmospheric chemical transport model (CTM) simulations, diagnose model simulations and thus guide model improvements. However, uncertainty analysis of CTMs is often challenging due to the enormous computational cost. This study proposed a new reduced-form model based on a conventional Stochastic Response Surface Modeling (SRSM) and Decoupled Direct Method (DDM). Compared with the traditional SRSM, the new method can save approximately 60% of the up-front computational cost. The new method was applied to assess parametric uncertainties in PM2.5 simulations in the Pearl River Delta (PRD) region using CMAQv5.0.2 and to demonstrate its ability in model diagnosis. The uncertainty analysis revealed that much of the PM2.5 simulation bias in the PRD region is related to parametric uncertainties. PM2.5 emissions, PM2.5 concentrations in LBCs, wind speed, NH3 emissions, and temperature are the major parametric uncertainty sources. Therefore, the enhancement of these sources can effectively promote PM2.5 simulations in PRD. The uncertainty analysis also reveals that the limited secondary organic aerosol (SOA) module is the primary reason for the underestimation of SOA by CMAQv5.0.2. Although VOC emissions have high uncertainty, the VOC emission enhancement only is ineffective for SOA simulation improvement if the module has a sizeable systematic bias.

Application of Low-cost PM Sensors to Quantify the Impact of Prescribed Burning on Air Quality and Social Vulnerability in Southwestern Georgia

Ran Huang, Raj Lal, Yongtao Hu, Armistead G. Russell, M. Talat Odman

Prescribed burning (PB) is a prominent source of PM2.5 in the southeastern US. As the demand for burning increases and stricter controls are applied to other pollution sources, PB emissions will be responsible for a larger fraction of PM2.5 concentrations. Exposure to PB fire smoke is a severe health risk. In order to quantify the effect of PB on air quality, low-cost PM sensors will be used to measure the PM2.5 concentration in southwestern Georgia. Here, the feasibility of using low-cost sensors as a supplemental measurement tool will be evaluated by comparing the measured PM2.5 concentrations with reference instruments (TEOM and E-BAM). A chemical transport model (Community Multiscale Air Quality (CMAQ)) will also be used to simulate the contribution of PB on PM2.5 concentrations with the decoupled direct method (DDM, a sensitivity analysis technique for computing sensitivity coefficients simultaneously while air pollutant concentrations are being computed) to understand the impact of PB on the local air quality. Measurements from low-cost sensors will provide the local PM2.5 concentrations to evaluate the simulated concentrations. Simulations from the air quality model will be fused with observations from the nearby reference instrument and the low-cost sensors to generate exposure-to-smoke fields. The exposure fields, which integrate the spatial details of the air quality model with the temporal accuracy of observations, will be used to explore the relationships between PB and social vulnerability and will help us better understand the effect of PB on public health.

Quantifying impacts of emission reductions on environmental justice and human health in New York City

Robyn Chatwin-Davies, Burak Oztaner, Shunliu Zhao, Melanie Fillingham, Marjan Soltanzadeh, Amir Hakami (Carleton University); Amanda Pappin (Health Canada); Iyad Kheirbek, Kazuhiko Ito, Thomas Matte (New York City Department of Health and Mental Hygiene); Jay Haney, Sharon Douglas (ICF International); Matt Turner, Daven Henze (University of Colorado); Shannon Capps (Drexel University); Peter Percell (University of Houston); Jaroslav Resler (ICS Prague); Jesse Bash, Sergey Napelenok, Kathleen Fahey, Rob Pinder (USEPA); Armistead Russell, Athanasios Nenes (Georgia Tech); Jaemeen Baek, Greg Carmichael, Charlie Stanier (University of Iowa); Adrian Sandu (Virginia Tech); Tianfeng Chai (University of Maryland).

We use an adjoint model to examine inequity in exposure to fine particulate matter (PM2.5) in New York City and surrounding areas across various income groups. The objective is to characterize air pollution inequity and to identify emission control measures that can improve environmental equity in this region. By contrasting the sensitivities of public health and equity measures to emissions reductions on a location-by-location basis, this study offers novel yet practical suggestions to coordinate air quality management strategies that prioritize different policy endpoints.

Simulations were run using the CMAQ v5.0 and its adjoint. Simulations cover a area of New York City at 1 km resolution to characterize air pollution inequity. The forward CMAQ model was first run to estimate concentration surfaces for quantifying exposure to PM2.5. Second, the adjoint of CMAQ was run to estimate the sensitivity of domain-wide inequity parameters to pollution emissions, on a location-by-location basis for different sectors.

Our results show that lower income populations in New York City tend to be exposed to higher concentrations of PM2.5. Adjoint sensitivity analysis shows that emissions control measures differ in their impacts on the landscape of environmental justice in New York City. We find that emission reductions in Brooklyn, Harlem, and the Bronx would be most beneficial for reducing domain-wide inequity, while reductions in Manhattan would result in increased inequity. We also calculate monetized health benefits per ton of primary PM emissions across the domain, with benefits estimated to be significant (as high as $10M/ton) and spatially variable across the domain. We propose a novel approach for monetizing air pollution inequity, and find that monetized benefits from inequity reductions are comparable in magnitude to health benefits from avoided mortality. We examine hypothetical scenarios where quantified benefits with regards to health an inequity are considered in tandem to coordinate policies that target both endpoints simultaneously. Our results demonstrate that focusing emission reductions on sources that are influential on both environmental justice and public health can yield improvements across multiple policy goals simultaneously.

Quantifying the error in satellite AOD retrievals associated with surface reflectance uncertainties in the semi-arid western U.S.

S. Marcela Lorea-Salazar,

Heather A. Holmes,

Jayne M. Boehmler,

Patrick W. Arnott,

James D. Long

Atmospheric Sciences Program, Department of Physics, University of Nevada Reno, Reno, Nevada, U.S.A.

Satellite characterization of aerosol optical depth (AOD) is desired because of the enhanced spatial resolution compared to the monitor stations. However, previous research has found that NASA satellite aerosol algorithms (e.g. collection 6 deep-blue [DB]) have large uncertainties in AOD when retrieving over bright salt pans, a typical geographical feature in the western U.S. To address this, the new collection from the DB algorithm (Col. 6.1) was improved by detecting geographical areas where very bright surfaces affected AOD retrievals, where the bright surfaced led to an over estimation in aerosol levels. In addition, the Multi-angle implementation of atmospheric correction (MAIAC) algorithm offers a better characterization of surface reflectance that can help to retrieve AOD. The aim of this investigation is to develop a propagation of error analysis that quantifies the impact of surface reflectance uncertainties in the NASA aerosol satellite retrievals during periods with low aerosol loading. Previous literature has reported that an error in surface reflectance of 0.01 can affect AOD from satellite retrievals by 0.1. Preliminary results show improvements in the new DB product. However, surface reflectance is still affecting the monthly averages by overestimating AOD with a normalized mean bias of 100% in the new DB Col. 6.1 and MAIAC algorithms over bright surfaces in the western U.S. during specific satellite overpasses. Satellite overpasses near the backscatter direction are found to be especially problematic, suggesting need of a filtering algorithm to eliminate their artificially elevated AOD values. References 1. An enhanced VIIRS aerosol optical thickness (AOT) retrieval algorithm over land using a global surface reflectance ratio database - Zhang - 2016 - Journal of Geophysical Research: Atmospheres - Wiley Online Library. Available at: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1002/2016JD024859. (Accessed: 5th July 2018)

Direct estimates of health improvements from reduced coal power plant emissions exposure

Lucas RF Henneman, Christine Choirat, Cesunica Ivey, Kevin Cummiskey, Corwin Zigler

We endeavor to quantify and compare changes in 10 health outcomes among US Medicare beneficiaries attributable to two different-but related-metrics for changing air quality. Using long-term records of US coal-fired power plant emissions, air quality measurements, and ZIP code level Medicare data, we relate changes in the rates of adverse health outcomes to 1) changes in exposure to overall PM2.5 (all-source PM2.5) and 2) changes in exposure to emission from over 1,000 coal electricity generating units (coal exposure) from 2005 to 2012. Changes in coal exposure are estimated using a reduced complexity air quality approach based on the HYSPLIT model. We use HYSPLIT to simulate the dispersion of emissions from each electricity generating unit every six hours for each year and link the resulting exposure field to ZIP codes. We use a difference-in-difference regression approach designed to mitigate the threat of observed and unobserved confounding and use estimates from the models to compare how the change in health outcome rates from 2005 to 2012 is associated with changes in all-source PM2.5 and coal exposure during the same period. Results show general consistency between changes in health impacts associated with changing all-source PM2.5 exposure and coal exposure, but we find that changes in rates of chronic obstructive pulmonary disease (COPD) are not associated with reduced total PM2.5 exposure while they are associated with reduced coal exposure.

Air quality, social vulnerability, and health of the communities exposed to intense prescribed burn activity in the Southeastern U.S.

Sadia Afrin, Fernando Garcia Menendez

Although prescribed burning is an important strategy to reduce wildfire risk, assessing the impacts of controlled burns on air quality and public health has become a major research need as the amount of land subjected to fire treatments increases each year. Remote sensing products have been frequently used as a source of prescribed fire data in prior air quality studies due to challenges associated with compiling ground-based fire information. Here we rely on a permit-based prescribed fire database to quantify the influence of burning activity on PM2.5 concentrations recorded at the monitoring sites across the Southeastern US and find significantly stronger correlations between measured PM2.5 and permit-based fire records compared to the satellite-derived data. We use this prescribed fire and air quality from monitoring networks in Georgia and Florida to characterize communities exposed to the high levels of burning and related air pollution. Specifically, we determine the spatial association between prescribed fire and social vulnerability using Anselin's bivariate local indicators of spatial association (LISA) between burn permits and the Centers for Disease Control and Prevention's census block group-level social vulnerability index (SVI) dataset. We find that burn-intensive areas have more socially vulnerable populations compared to those with less burn activity, and identify significant spatial clustering of prescribed fire and vulnerability hotspots in Southwest Georgia, Northwest Florida and central regions of Southern Florida. In addition, we also examine parcel data describing land value and ownership around these communities. Finally, we use data from Georgia Department of Public Health and Florida Department of Health to further characterize public health at the communities most exposed to prescribed fire smoke.

Spatial variability in air quality social costs and the implications for policy

Nicholas Z. Muller, Elisabeth A. Gilmore, Jinhyok Heo, Christopher W. Tessum, Jason D. Hill, Julian D. Marshall, Peter J. Adams

Information on the benefits of pollution abatement is useful both for policy design and policy evaluation. Multiple models now exist that can produce empirical benefit estimates. This paper employs marginal ($/ton) benefits developed by three different models - AP2, Estimating Air pollution Social Impacts Using Regression (EASIUR), and the Intervention Model for Air Pollution (InMAP) to examine the robustness of the benefits estimates as well as to identify opportunities to improve the efficiency of extant regulations. First, we find that despite differences in the representation for the air quality chemistry in each model, the national level average benefits per ton are similar in magnitude. Further, the rank ordering of benefits per ton across pollutants is the same across models. However, each model predicts large variability in the benefits across the US with population density being a major driver of these spatial differences. The models predict significantly different degrees of cross-sectional variance in the marginal benefits across pollutants. From the perspective of policy design, the primary implication of this is a different distribution of emissions across source locations. The analysis concludes with an exploration of the projected benefits from a standardized set of emission scenarios to show how model choice affects levels and the distribution of benefits.

Evaluation of Novel NASA Aerosol Products during the Yosemite Rim Fire

S. Marcela Lorea-Salazar1,

Heather A. Holmes1,

Neil Lareau1,2,

James D. Long1

1Atmospheric Sciences Program, Department of Physics, University of Nevada Reno, Reno, Nevada, U.S.A.

2Department of Meteorology and Climate Science, San Jose State University, San Jose, California, U.S.A.

Smoke emissions represent a public health problem that is increasingly impacting vulnerable populations worldwide. An escalating body of research suggests that the number of wildfires in the western U.S. (and their associated implications) will continue to rise due to the changing climate. Another compounding factor is the severe drought in this region over the past years that has driven an increasing number of wildfires. Over extreme fire events, satellite characterization of aerosol pollution is desired because they are able to capture the horizontal extent of the fires at a reasonable spatial resolution (e.g., 10km horizontal grid resolution). Previous studies have shown that former algorithms related to the retrieval of aerosol products from satellite remote sensing have limitations in characterizing aerosol loading and plume injection height. However, new versions of aerosol products from NASA MODIS and VIIRS instruments promise a better characterization of pollution from wildfires. By understanding the scope and limitations of these satellites products it is possible to improve exposure estimates in the western U.S. to inform epidemiology studies that aim to investigate when people suffer from cardio-respiratory problems from wildfire smoke plumes. This study will present a review of the new NASA MODIS collection 6.1, MAIAC, and ASHE satellite algorithms over an extreme fire event to evaluate aerosol loading and plume injection height products using ground-based sunpotometry and aerosol Lidar techniques. Because of the complex physical and topographical features throughout the western U.S., NASA satellite algorithms are currently the best quality satellite products available for air quality modeling and health effects studies. Preliminary results show an improvement in fire detection and aerosol concentrations from the new algorithms.

Air pollution exposures attributable to U.S. economic activity: relationships among economic consumption, exposure, and race-ethnicity

Christopher W. Tessum, Joshua S. Apte, Andrew L. Goodkind, Nicholas Z. Muller, Kimberley A. Mullins, David A. Paolella, Nathaniel P. Springer, Julian D. Marshall, Jason D. Hill

Outdoor fine particulate matter (PM_2.5) air pollution is a major cause of death in the United States. For year-2014, we explore relationships among PM_2.5 exposures and health impacts, sources of the economic sectors responsible for atmospheric emissions, and the human economic activities that directly and indirectly induce those emissions. In addition, we investigate how economic consumption varies by racial-ethnic group. We compare estimated average exposure for each racial-ethnic group against what the average exposure would be in a hypothetical scenario where exposures are proportional to average personal consumption. We find, for example, that individuals identifying as Black or African-American and as Hispanic or Latino were exposed to 65% more pollution than they would in the hypothetical situation where exposures for all groups are proportional to average personal consumption. These results are useful for framing considerations of "responsibility" for pollution exposures and for defining and investigating environmental justice concerns.

Reconstruction of surface NO2 concentrations using a conservative downscaling of OMI, GOME-2, and CMAQ

Hyun Cheol Kim1,2, Sang-Mi Lee3, Tianfeng Chai1,2, Barry Baker1,2, Fong Ngan1,2, Li Pan1,2, and Pius Lee1

1Air Resources Laboratory, National Oceanic and Atmospheric Administration, College Park, MD

2Cooperative Institute for Climate and Satellites, University of Maryland, College Park, MD

3South Coast Air Quality Management District, Diamond Bar, CA

We introduce an approach to reconstruct surface NO2 concentrations over urban areas by combining satellite-observed column densities and simulated surface-to-column ratios from a chemical transfer model. The spatial inhomogeneity of urban NO2 plumes becomes a practical problem as the focus of satellite data analysis moves from global distribution to regional- or local-scale urban plumes. The need for fine-scale modeling, smaller than the scale of satellite footprint pixels, is urgently required now. A conservative downscaling was designed to enhance the spatial resolution of satellite measurements by applying the fine-scale spatial structure from the model, with strict mass conservation at each satellite footprint pixel level. With the downscaling approach, NO2 column densities from the Ozone Monitoring Instrument (OMI; 13x24 km nadir footprint resolution) and the Global Ozone Monitoring Experiment-2 (GOME-2; 40x80 km) show excellent agreement with the Community Multiscale Air Quality (CMAQ; 4x4 km) NO2 column densities, with R = 0.96 for OMI and R = 0.97 for GOME-2. We further reconstructed surface NO2 concentrations by combining satellite column densities and simulated surface-to-column ratios from the model. Compared with the Environmental Protection Agency's (EPA) Air Quality System (AQS) surface observations, the reconstructed surface concentrations show a good agreement; R = 0.86 for both OMI and GOME-2. This study demonstrates that the conservative downscaling approach is a useful tool to conduct a fair comparison between the satellite and model data for air quality and emissions studies.

Air quality health co-benefits of CO2 reduction in China

Marjan Soltanzadeh, Burak Oztaner, Shunliu Zhao, Amir Hakami (Carleton University); Amanda Pappin (Statistics Canada); Matt D. Turner, Daven K. Henze (University of Colorado); Shannon L. Capps (Drexel University); Peter B. Percell (University of Houston); Jaroslav Resler (ICS Prague); Jesse O. Bash, Sergey L. Napelenok, Kathleen Fahey, Rob W. Pinder (USEPA); Huizhong Shen, Yongtao Hu, Armistead G. Russell, Athanasios Nenes (Georgia Tech); Jaemeen Baek, Greg R. Carmichael, and Charlie O. Stanier (University of Iowa); Adrian Sandu (Virginia Tech); Tianfeng Chai (University of Maryland).

Reducing combustion-based CO2 emissions often entails significant co-benefits for public health by reducing emissions of criteria co-pollutants and precursors. In our previous work, we used the adjoint of CMAQ to estimate location-specific co-benefits for mobile and Electric Generation Units (EGUs) sectors in Europe, Canada and the United States. We have found that location and sector dependencies, captured in adjoint sensitivity analysis, are important factors that can change the overall source and sector neutrality of GHG reduction benefits, and can provide a measure of uncertainty/variability associated to their magnitudes. The rapid increase of CO2 emissions and growing air pollution challenges in most countries in East Asia motivated us to apply our methodology to calculate the co-benefit of reduction of CO2 in this region. This study aims to evaluate how the health co-benefits vary spatially and by sector across different countries.

The adjoint of the US EPA's CMAQ 5.02 was applied to quantify the health benefits associated with emission reduction of NOX as an O3 precursor and PM2.5 primary and precursor emissions on a location-by-location basis across the East Asia. These health benefits are then converted to CO2 emission reduction co-benefits by accounting for source-specific emission rates of criteria pollutants in comparison to CO2. We integrate the results from the adjoint of CMAQ with emission estimates from 2010 Emissions Database for Global Atmospheric Research (EDGAR) v4.3.1. The meteorology data and emissions are processed using WRF and SMOKE.v4.0, respectively. Our preliminary results for China show that the monetized health benefits (due to averted chronic mortality) associated with reductions of 1 metric ton of CO2 emissions is up to $1000/ton-CO2 and $5000/ton-CO2 for mobile on-road and EGU sources, respectively. We estimate more than $700B in total damage from fossil fuel power generation in China, and find that single power plants can account for up to $10B in damage. Implications of such spatial variability in devising control policy options that effectively address both climate and air quality objectives will be discussed.

Evaluation of the CMAQ Version 5.3 Beta Release

K.W. Appel, C. Hogrefe, K.M. Foley, G. Pouliot, R.C. Gilliam, and B. Murphy

In this study we evaluate the beta release of the Community Multiscale Air Quality (CMAQ) model version 5.3, scheduled for public release in fall of 2018. This latest version of the CMAQ model will contain a number of important updates, including a vastly updated aerosol module (AERO7), updated halogen and over-ocean chemistry, a new deposition scheme (STAGE) and updates to the M3Dry deposition scheme, as well as a number of other updates. This evaluation will also utilize simulations using the most recent version of the Weather Research and Forecast (WRF) model available, incorporating lightning assimilation to greatly improve the location and timing of precipitation in the model. The operational evaluation of the modeling system will be extensive, utilizing annual simulations from both 2011 and 2016 to fully assess the model performance across a range of meteorological and air quality conditions. The impact of the more extensive updates to 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 large updates have on model performance. Evaluation of gas and particle species using routine observations from U.S. networks (e.g. AQS, IMPROVE, CASTNET), campaign data (e.g. DISCOVERAQ), and, where applicable, regional and global data, will be presented. This work will provide an initial summary of the model performance expected from the final release of the CMAQv5.3 model in 2019.

State-level contributors to present and future fine particulate matter health costs in the United States

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

Fuel combustion adversely affects air quality and human health by contributing to fine particulate matter (PM2.5) concentrations in the United States. Future PM2.5 concentrations will be determined by energy use, emission controls, state and national policies, and other factors. The Global Change Assessment Model (GCAM) is a human-earth system model that analyzes the US and global economy, energy system, buildings, transportation, land use, and climate system. GCAM-USA is an extension of GCAM in which US energy supply and demand markets are disaggregated to state-level resolution. In previous work, we have integrated national-level year-, pollutant- and sector-specific PM2.5 mortality cost factors into GCAM-USA. These national-average factors, in units of dollars-per-ton, translate nitrogen oxides (NOx), sulfur dioxide (SO2), and direct PM2.5 emissions into PM2.5 mortality costs using relationships derived through air quality and health effects modeling. Projections of these factors account for national-level changes in population and economic growth.

In this study, we improve upon that work by applying PM2.5 mortality cost factors that are differentiated by state, accounting for the spatial heterogeneity of pollutant transport and chemistry, population, and baseline mortality rates. We demonstrate this new representation by examining the evolution of the energy system and emissions under a reference scenario shaped by current energy and environmental policies. Here we seek to understand the contributions of different emission sectors in each state to overall PM2.5 mortality costs, which will aid in more effectively reducing PM2.5 mortality. For example, preliminary results show that emissions from conventional coal-fired power plants were the largest contributor to national total PM2.5 mortality in 2015. However, the contributions from industrial sector coal use increase significantly from 2015 to 2050, which will be examined further. While the leading contributor to PM2.5 mortality varies by region in 2015, industry is the leading contributor for almost all regions in 2050. Differences between 2015 and 2050 are mainly driven by state-specific technology portfolios and air quality regulations.

Hemispheric-CMAQ Application and Evaluation for 2016
Henderson, Dolwick, Jang, Eyth, Vukovich, Mathur, Hogrefe, Pouliot, Possiel, Timin, Appel

Inter-continental transport of air pollution occurs at time scales of days to weeks within the northern hemisphere. Quantifying transport of anthropogenic pollution between countries requires a modeling system that credibly represents global processes. Our work evaluates two models, Hemispheric CMAQ and GEOS-Chem that cover the northern hemisphere. Our application uses updated emissions from the EPA 2016 modeling platform, Mexico, Canada, and China to improve the accuracy of domestic and foreign sources. This presentation will cover evaluation with surface monitors, sondes, aircraft, and satellites. Preliminary results show important systematic differences as well as differences that have a latitudinal gradient. We use contrasts in model performance and process-level differences between the models to identify influential uncertainties and recommend future research. The results from this modeling effort will ultimately be coupled with regional modeling to estimate fine space and temporal scale international contribution estimates.

GLIMPSE decision support system for air quality management: Overview and demonstration

Daniel H. Loughlin 1, Samaneh Babaee 1,2, Carol S. Lenox 1, Christopher G. Nolte 1, Yang Ou 1,2,3, Wenjing Shi 1,2, Steven J. Smith4, and Tai Wu 1

1 Office of Research and Development, U.S. Environmental Protection Agency, RTP, NC; 2 Oak Ridge Institute for Science and Education; 3 Environmental Sciences & Engineering, University of North Carolina at Chapel Hill; 4 Joint Global Change Research Institute, Pacific Northwest National Laboratory, College Park, MD

The GCAM-based Long-term Interactive Multi-Pollutant Scenario Evaluator (GLIMPSE) is a prototype decision support tool that is intended to inform national, regional, and state-level air quality management decisions. The "computational engine" for GLIMPSE is the Global Change Assessment Model, or GCAM, which has been developed by the Department of Energy's Pacific Northwest National Laboratory (PNNL). GCAM is a human-earth systems model that includes representations of the energy system, agriculture, land use, and the atmosphere and simulates the evolution of these interconnected systems through 2100. Users can explore wide-ranging scenarios, including those with alternative assumptions regarding population growth and migration, economic growth and transformation, technology costs, and environmental and energy policy. From an air quality management perspective, key outputs of GCAM include emissions of criteria pollutants, estimates of the health effects associated with emissions, and water demands from energy and agriculture.

The base or "core" GCAM model has global coverage, broken into 32 geographic regions. Two variants of GCAM have been developed that have additional spatial resolution. GCAM-USA represents the 32 global regions, but the U.S. region has been disaggregated into 50 states and the District of Columbia. Similarly, GCAM-China disaggregates the representation of China to China's 31 provinces. Thus, GCAM-USA and GCAM-China can support the analysis of state- and province-level air quality management strategies within the context of a global scenario.

While GCAM is very flexible and has the potential to positively affect air quality decision making, the complexity of the model and associated data system has limited its use to a small community of developers and "power users." However, the model is well suited to be integrated into a decision support tool that automates complicated functions (e.g., setting up input files, aggregating outputs over regions, and examining differences from one scenario to another), making the model accessible to a much broader set of users. Developing such a decision support tool is the primary goal of the GLIMPSE project. Over the past several years, we have implemented a GLIMPSE prototype and are engaged with beta testers.

In this presentation, we will provide an overview of GLIMPSE. We will then demonstrate its use for a real-world problem: assessing the potential air pollutant emissions implications of expanding the Regional Greenhouse Gas Initiative (RGGI) region to include Virginia, New Jersey, and Pennsylvania. Results to be explored include the state-level air pollutant co-benefits for the RGGI states.

Evaluation and Intercomparison of CMAQ-Simulated Ozone and PM2.5 Trends Over the United States

Christian Hogrefe, Kristen M. Foley, K. Wyat Appel, Shawn J. Roselle, Donna Schwede, Jesse O. Bash, and Rohit Mathur

In this study, we compare ozone and PM_2.5fields from two sets of long-term WRF/CMAQ simulations against available observations and against each other. The two simulations were performed for 1990 - 2010 and 2002 - 2014, respectively, and differed in terms of emission inputs, chemical boundary conditions, and horizontal grid spacing (36 km vs. 12 km). The evaluation of modeled ozone and PM_2.5levels and trends utilizes observations from AQS, IMPROVE and CASTNET and is performed on a seasonal basis for nine different geographic regions considering different percentiles of the pollutant distributions. A special emphasis is placed on evaluating the time period common to both simulations (2002 - 2010) to determine commonalities and differences between the simulations. This analysis also includes a comparison of the emission inputs and boundary conditions used in both simulations to investigate to which extent these factors may have caused differences in the simulated pollutant levels, variations, and trends. The results are discussed in the context of designing and interpreting dynamic model evaluation studies.

Plausible energy scenarios for use in robust decision making: Evolution of the US energy system and related emissions under varying social and technological development paradigms

Kristen E. Brown, Troy A. Hottle, Rubenka Bandyopahyay, Samaneh Babaee, Rebecca S. Dodder, P. Ozge Kaplan, Carol S. Lenox, Daniel H. Loughlin

The energy system is the primary source of most air pollution emissions. Thus, evolution of the energy system into the future will affect society's ability to maintain air quality. Anticipating this evolution is difficult because of inherent uncertainty in predicting future energy demand, fuel use, and technology adoption. We apply Scenario Planning to address this uncertainty, developing four very different visions of the future. Stakeholder engagement suggested technological progress and social attitudes toward the environment are critical and uncertain factors for determining future emissions. Combining transformative and static assumptions about these factors yields a matrix of four scenarios that encompass a wide range of outcomes. We implement these scenarios in the U.S. EPA MARKAL model. Results suggest that both shifting attitudes and technology transformation may lead to emission reductions relative to present, even without additional policies.An important outcome of this work is the scenario implementation approach, which uses technology-specific discount rates to encourage scenario-specific technology and fuel choices. End-use energy demands are modified to approximate societal changes. This implementation allows the model to respond to perturbations in manners consistent with each scenario. In all scenarios electricity continues to be generated by a mix of sources, with the fraction of natural gas, coal, wind, and solar differing between scenarios. Hybridization and electrification dominate light duty transportation with transformational technology, while paradigm shifts lead to enhanced use of biofuel. The benefit of this approach is that the scenarios can be utilized with other modeling approaches, creating multiple baselines upon which to model the impact of a new technology or policy.

Dynamic evaluation of WRF-CMAQ PM2.5 simulations over the Contiguous United States for the period 2000-2010

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

1 University of Connecticut, 2 US Environmental Protection Agency, 3 North Carolina State University.

Dynamic evaluation of the 2000-2010 fully coupled Weather Research and Forecasting (WRF) - Community Multiscale Air Quality (CMAQ) model fine particulate matter simulations over the contiguous United States (CONUS) is conducted to assess how well the changes and trends in observed PM2.5 and its speciation are simulated by the model. The changes induced by variations in meteorology and/or emissions are also evaluated during the same timeframe using Kolmogorov-Zurbenko (KZ) spectral decomposition of observed and modeled time series with the aim of identifying the underlying forcing mechanisms that control particulate matter exceedances. Apart from KZ filtering, mode decomposition methods such as Empirical Mode Decomposition (EMD) and Variational Mode Decomposition (VMD) are also employed to study the intrinsic time scales embedded in the observed and simulated time series and to critically assess the model's capability in reproducing the spatio-temporal features seen in observations. Results of the dynamic evaluation are presented for total PM2.5 and its chemical components for multiple time scales across six different geographical regions in the CONUS.

Linking Energy Sector and Air Quality Models through Downscaling

Emily Bartholomew Fisher, Benjamin F. Hobbs, Joseph Hugh Ellis, Shen Wang, Xiaoxue Hou, Shenshen Li, Xinrui Zhong (Yale-JHU Center Solutions for Energy, Air, Climate and Health; The Johns Hopkins University)

Air emissions from fossil energy use can have a significant impact on air quality. There are many tools and models for conducting sophisticated energy sector projections and analysis. Similarly, there are complex tools for conducting air quality research. There is a gap however in linking these tools. The national- or regional-scale models tend to provide combustion and emission results at low resolution in terms of time and geography (e.g., annually by census division), whereas air quality models need higher resolution emission inputs. Also, the way sources are defined in energy models is not well-matched to sources in air quality models and inventories. For these and other reasons, using the output of energy models as the input to air quality models is challenging.

The research summarized here develops methods for processing results from the National Energy Modeling System (NEMS) for subsequent use emissions processing (via SMOKE) then CMAQ to produce projections of future air quality as a result of long-term changes in the energy sector. Methods are developed for area sources (residential, commercial and industrial), point sources (electric generation), transportation, and oil and gas exploration and production for a set of criteria air pollutants (including NOx, SO2, PM, and VOC).

We calculate regional emission growth and reduction (hereafter "change") factors for 2020, 2030, 2040 and 2050 by multiplying sectoral fuel use from NEMS by emission factors, and dividing the future year emissions by NEMS or NEI emissions from the base year 2011. To downscale emissions from the region to point- or county-scale sources, we use different procedures for different source types.

For area and transportation sources we use "grow-in-place" assumptions and multiply the base year emissions inventory by the associated emissions change factors.But because of the rapid technology changes in the power sector, future siting and operating patterns for new electric generators will differ from the past. Therefore, we developed a three-step method to site new generator point source capacity identified by NEMS. Using optimization methods, new regional capacity is first allocated to sub-region based on dispatch feasibility, taking into account transmission limits between regions, sub-regional differences in fuel and capacity cost, load, existing capacity and renewable energy availability. Second, land is then screened for suitability, taking into account factors such as steepness, national park classification, existence of wet lands, etc. Finally, capacity is sited within subregions at locations that have not been disqualified by the screening step based on a least-cost optimization.

For oil and gas exploration and production, we will base changes in NEI emissions on changes in production levels reported by NEMS. We compare emissions change factors across geographic regions, sources, and pollutants from two NEMS scenarios: a base case and a future with high natural gas production.

Assessing the Meteorological Aspect of High Ozone Over Lake Michigan
Richard T. McNider, Arastoo Pour Biazar, Kevin Doty, Andrew White, Yuling Wu, Momei Qin, Yongtao Hu, Talat Odman, Bright Dornblaser, Patricia Cleary, Eladio Knipping, Stuart McKeen, Pius Lee

For the past few decades, high ozone concentrations along the shores of Lake Michigan have been a recurring theme. Yet, models continue to have difficulty in replicating ozone behavior in the region. While emissions and chemistry play an important role in model performance, the complex meteorological setting of the relatively cold lake in the summer ozone season and the ability of the physical model to replicate this environment may contribute to air quality modeling errors. Furthermore, land/water temperature contrasts are critical to lake and land breezes which impact mixing and transport.

In the current study, the role of model mixing and lake surface temperatures are examined in terms of changing stability over the lake. In particular, an analysis is presented that shows excessive mixing under stable conditions in the meteorological model may lead to an under-estimate of mixing in the chemical transport model if the mixing coefficients are re-diagnosed by the chemical transport model. The analysis is based on the results from WRF/CMAQ air quality forecasts for the summer of 2009 and 2011. A CMAQ sensitivity simulation for 2011 using results from two WRF simulations were performed to test the impact of more mixing under stable conditions. WRF simulations were identical except for the way mixing coefficients were calculated. The base WRF simulation used a short-tail (or minimum) mixing form, and the other simulation used the Louis stability function (long-tail mixing form) for more mixing. The results indicated that more mixing in WRF led to smaller diagnosed mixing coefficients in CMAQ (less mixing), causing increased surface ozone concentrations over the lake.

Additionally, the impact of nudging strategies was also examined. It was concluded that performing nudging above the boundary layer reduces the strength of nocturnal jet in the model. However, limiting nudging to above 1-2 km, might be a better strategy as it allows the model to develop a more realistic nocturnal jet that impacts nighttime transport.

Python as a data science platform for air quality studies

Andy Lewin, Dr Scott Hamilton, Dr Nicola Masey

The python language has great potential for practical applications in air quality modelling and assessment. The huge volumes of data from air quality observations, meteorological measurements, and air quality models can present a data management challenge to air quality practitioners, making it difficult to extract insights with which to aid decision makers. Python's many numerical and statistical libraries provide an open source basis for the development of complex and reproducible workflows from quite simple code- for example the 'pandas' library can be used for handling of large time series based measurement datasets, for inventory compilation, and for parameterising and controlling the USEPA model AERMOD in a reproducible manner. We will cover all of these in our presentation.

We will demonstrate several practical use cases based on our organisation's day to day use of the language and will offer a practitioner's guide to its fundamental libraries such as pandas, numpy, statsmodels, matplotlib and more. We will then present the following real cases based on our work and provide useful code snippets to help practitioners:

1) Handling of several million traffic data observations for emissions inventory compilation

2) High resolution mapping of traffic emissions based on GPS observations

3) Aligning hourly meteorological observations with air quality observations in an automated workflow

4) Setting up and controlling dispersion models such as AERMOD

5) Developing a complete road traffic emissions model from python functions (the RapidEMS system) 4) Building a complete urban modelling platform - (the RapidAIR sytem (which we presented at CMAS 2017)

6) Building a complete urban modelling platform - (the RapidAIR sytem (which we presented at CMAS 2017)

Investigating the formation of the high ozone exceedance on 25th September 2013 during the NASA DISCOVER-AQ Texas campaign

Shuai Pan, Yunsoo Choi, Wonbae Jeon, Anirban Roy, David A. Westenbarger, Hyun Cheol Kim

The highest maximum daily surface eight-hour O3 concentration (MDA8) of 124 ppb and the highest 1-hour value of 151 ppb during the NASA DISCOVER-AQ Texas campaign in 2013 were observed on September 25th. Under-estimation of this high O3 episode by air quality models has been reported consistently by several previous studies. In this study, a WRF-SMOKE-CMAQ air quality modeling system was used to investigate the causes of this event by comprehensive comparisons using measurements from various platforms (e.g., ground, buoy, aircraft, and mobile lab). Our analyses suggest that episodic flare emissions, dry sunny postfrontal stagnated conditions, and land-bay/sea breeze transitions could be the potential causes of the high O3. With "better" emission and wind, our model could reproduce this event in term of magnitude, location, and timing. The results suggest that improving the model capability to reproduce small-scale meteorological conditions favoring O3 production, such as stagnation and wind reversal, is needed.

Feedback between anthropogenic factors and climate: Energy consumption for space heating

Matthias Berger

Jurg Worlitschek

Anthropogenic factors influence global and local climate. Acting as a feedback, emissions from energy consumption are on the other hand driven by changes in climate. We analyze the relationship between climate and energy demand for space heating in Switzerland for the period 1980 to 2017. Results indicate a significant feedback, which will influence future heat demand, conversion technology mix and local emissions.

Searching for missing carbonaceous aerosols sources in Bogota: Sensitivity analysis using WRF-Chem

Perez-Pena, Maria Paula. Morales B, Ricardo

Bogota is the largest city of Colombia, with approximately 8 million inhabitants. It characterizes by a growing population and important sources of pollutants well within the city such as industrial facilities, as well as a massive vehicular fleet of more than 2 million vehicles. With a public transportation system consisting mostly of diesel power fleet, air quality pollution in Bogota is an increasing concern. Several studies in the city have identified particle matter as the main airborne species. Previous modelling simulation efforts using off-line chemical models, have highlighted the importance of resuspended particles from paved and unpaved roads to the particle matter in Bogota. On the other hand, few studies have investigated the chemical composition of particles in the city. Some measurement campaigns have found that crustal material is the most significant species in the coarse aerosols. Additionally, it has been identified a prevalent presence of organic matter, accounting for ~50% to the total aerosol mass, and higher than usual elemental carbon (between 12% and 30%). OC/EC ratios suggest an important contribution of secondary formation to the organic aerosols. Secondary inorganic aerosols (SO42- and NO3-) have shown little presence in the sampled PM10 mass. No fine particles studies are found to date. In this work, we use the on-line chemical transport model WRF-Chem to explore how different emissions inventories represent the known composition of atmospheric particles observed in the city of Bogota. Special focus is made to elucidate the composition of fine aerosols (PM2.5). We tested the spatial accuracy of two global emissions datasets, EDGAR-HTAP and EDGARv4.3.1, as well as a local inventory for Bogota. A base case emissions inventory was built merging global emission inventories with locally estimated re-suspended particulate matter emissions. The emissions were speciated according to model requirements for the gas phase chemical mechanism RACM. The MADE model was used for aerosol physics, while VBS was used to estimate secondary organic aerosols. The biogenic emissions included correspond to on-line calculated ones using MEGAN. Wild-fires and prescribed fires were also added to the simulations. The results of our simulations indicate that PM2.5 and PM10 levels can be represented well with the emission inventories explored in the study. However, the contribution of both, Elemental Carbon and Organic Aerosols is highly underestimated. Similarly, sulfate and nitrate contribution are overestimated by all model simulations. Organic aerosol contribution is only well represented when using the local inventories, which grossly overestimates other species such as CO and VOC. This suggests a necessary revision of current speciation of official local inventories in Colombia. Further gas phase chemistry and aerosol chemical composition are necessary to better constrain emissions of secondary aerosol precursors, both from biogenic and anthropogenic sources.

Projection of Wildfire Impacts on Regional Climate and Air Quality under Changing Climate

Cheng-En Yang, Joshua S. Fu, Xinyi Dong, Jian Sun, Qingzhao Zhu, Yang Liu, and Yongqiang Liu

Wildfires emit smoke particles and other pollutants that greatly threat public health over the western United States each fire season. Rising surface temperature due to climate change leads to lower soil moisture and atmospheric humidity that increases the possibilities of wildfires. The fire size and total burned area are reported to increase since vegetated regions become more vulnerable to fires. As continued worsening of soil moisture deficits in much of the western United States are predicted by Earth system models, it is essential to evaluate future wildfire impacts on regional climate and air quality and to provide crucial information for decision makers. Our study will utilize the Community Multiscale Air Quality model to explore the impacts of current (2005-2010) and projected future (2054-2059) wildfires on regional climate and air quality over the western United States from April to November. Major pollutants will be projected in response to the enhanced future fire emissions in addition to the Representative Concentration Pathway 8.5 scenario defined by the Intergovernmental Panel on Climate Change. With the simulated future pollutant concentrations due to wildfires, policymakers and governmental services can benefit from this information for making decisions.

Evaluation of Dynamically Downscaled, 36- and 12-km Simulations of the WRF Model to Develop Intensity-Duration-Frequency (IDF) Curves for U.S. Urban Areas

Anna Jalowska and Tanya Spero

Extreme precipitation events influence watershed, agriculture and urban management decisions. Intensity-Duration-Frequency (IDF) curves are a common tool used to project extreme precipitation events in urban and environmental planning. This study evaluates development of IDF curves using dynamically downscaled 36- and 12-km simulations of the Weather Research and Forecasting (WRF) model. The evaluation is performed on three data sets: 1) dynamically downscaled 36-km WRF model forced with 2.5-degree (~200-km) NCEP-DOE AMIP-II Reanalysis (R-2) data; 2) dynamically downscaled 36-km WRF model forced with 0.75-degree (~79-km) ERA Interim data; and 3) dynamically downscaled 12-km WRF model forced with ERA Interim data. The modeled data is verified with historical observational data for 3 cities in the U.S. using a 23- year historical period (1988-2010). IDF curves developed from both of the 36-km simulations provide good representation of 6-18-hour duration events. However, both 36-km simulations fail to predict short-duration precipitation events (1- 3-hour). Increasing the resolution of the model's driving data by changing from R-2 to ERA-Interim did not improve the model's representation of short-duration precipitation events. Using the WRF model to downscale to a 12-km horizontal grid enhances the model's ability to reproduce weather and IDF curves. Finer resolution allows better reproducibility of the frequency and intensity of 1- 3-hour rain events and improves representation of 6- 24- hour rain events. However, better performance of 12- km WRF data did not apply equally to all study sites, suggesting further modifications to the model and/or methodology are necessary.

Development of an Integrated Assessment Model for Korea : the GHGs and Air pollutants Unified Information Design System for Environment(GUIDE)
Younha Kim1, Jung-Hun Woo1*, Bok Haeng Baek2, Jinseok Kim1, Jinsu Kim1, Youjung Jang1, Minwoo Park1, Yungyeong Choi1, Eunji Lee1, Hyunjin Park1

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

East Asia is one of the most important regions to control Green House Gases (GHGs) emission to mitigate climate change. The Integrate Assessment Models(IAMs) have been extensively used to understand future emission pathways and impacts of Climate change. Combined analysis of air pollution and climate change, however, could reveal more important synergies of emission control measures, which could be of high policy relevance. Insight into the multiple benefits of control measures could make emission controls economically more viable. In this research, we have developed a prototype of the GUIDE(GHGs and air pollutants Unified Information Design system for Environment), a new decision support system for integrate GHGs and air pollutants. The GUIDE modeling system can interactively project and manage various climate change and air quality controls against baseline status. The special features of the GUIDE are; 1) the new macro economy-based Benefit-Cost(B-C) model for decision making, 2) a source-receptor relationship matrix which can examine impacts of emissions control in realtime, 3) implementation of integrated GHGs and Air Pollutants(APs) emissions inventory for Korea, and 4) incorporation of transboundary emission inventories to quantify out-of-region contribution. The simultaneous optimization for bi-directional co-benefits (i.e. co-benefits of APs and GHGs control) is the ultimate function of GUIDE system. We will present a more detail progress on developing a prototype version of the modeling framework.

KEYWORDS
GUIDE, IAM, GHGs, Climate, Air Pollution, Emission, Korea

ACKNOWLEDGEMENT
This subject is supported by Korea Ministry of Environment as "Climate Change Correspondence Program (project no.2016001300001). This work is 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).

Potential Performance differences of the National Air Quality Forecasting Capability when upgrading the Chemical Transport Model

Pius Lee1, Youhua Tang1,2, Daniel Tong1,2,3, Barry Baker1,2 , Jeff McQueen4, Jianping Huang4,5, Ho-Chun Huang4,5, Jose Tirado-Delgado6,7, and Ivanka Stajner4 1 NOAA/Air Resources Laboratory, College Park, MD, 2 UMD/Cooperative Institute for Climate and Satellites, College Park, MD, 3 Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA 4 National Oceanic and Atmospheric Administration (NOAA), National Centers for Environmental Prediction (NCEP), College Park, MD 5 I.M. Systems Group Inc., Rockville, MD 6 NOAA, Office of Science and Technology Integration, Silver Spring, MD 7 Syneren Technologies Corporation, Arlington, VA

The National Oceanic and Atmospheric Administration (NOAA) National Air Quality Forecasting Capability (NAQFC) is a vital service that NOAA provides to safeguard public health as well as environmental resilience through information-driver mitigation, and remedial and adaptation actions. The NAQFC system is under a study to potentially upgrade its Chemical Transport Modeling component from using the Community Multiscale Air Quality Model (CMAQ) version 5.0.2 to version 5.2. This is a major upgrades in chemistry and their corresponding emission sciences. The following lists the major science upgrades: (a) upgrade of the gas chemistry for the Carbon-Bond Mechanism version 5 (CB05) to version 5 Revision1 (CB05R1); (b) Inclusion of Halogen chemistry; (c) Employment of more explicit speciation for isoprene and monoterpenes from biogenic sources; (d) Upgrade of the aerosol module using a more sophisticated secondary aerosol production suite of multi-generational oxidation mechanism; and (e) Application of a fuller set of National Emission Inventory (NEI) that aligns better with CMAQ version 5.2 from the base year of 2014. We tested the new system for a retrospective summer case and compared its forecast performance with the real-time operational NAQFC. The U.S. Environmental Protection Agency (EPA) AIRNow monitoring network was used to verify the forecast accuracy. We noticed considerable discrepancies in the performance of the two realization of forecasting simulations. Their performance statistical metrics were compared and ranked to provide a basis for implementation recommendation.

Land use-related emission impacts on reactive nitrogen deposition and stream water acidification in sensitive regions

Yilin Chen, Shuai Shao, Huizhong Shen, Armistead G. Russell, Charles T. Driscoll

While SO₂ and NO_x emissions from energy, industrial and mobile sources have been reduced significantly by traditional regulations, land use-related reactive nitrogen emissions are playing a more dominant role in affecting atmospheric particulate matter formation as well as surface water acidification through deposition. Development of land use and climate-based policies to improve both air and water quality requires comprehensive assessments of future reactive nitrogen deposition and its associated impacts on water quality, especially in sensitive regions (National Parks and other Class I regions). In this study, we present the utilization of CMAQ-PnET-BGC model system to evaluate the atmospheric deposition change of reactive nitrogen species for the continental US (CONUS) and the corresponding responses in stream water pH and acid neutralizing capacity (ANC) in the Great Smoky Mountains National Park and the Adirondack region of New York in eastern US. The impacts of land use changes are emphasized by comparing future (2050) scenarios with and without considering land use-related emission changes in various sectors (agriculture, biogenic and forest fire emissions). Our results show that driven by climate warming, future cropland NH₃ emissions, biogenic VOC emissions, and NO_X emissions from fires are likely to increase, which increases both reduced (NH₃ and NH₄⁺) and oxidized (HNO₃ and NO₃^-) reactive nitrogen deposition and has important implications for sensitive ecosystems.

Improving Spatial Resolution of Wildland Fire Location and Fuel Biomass Data Inputs to NOAAs National Air Quality Forecast Capability.

Kenneth Craig,¹ Stacy Drury,² ShihMing Huang,¹ Nathan Pavlovic,¹ Garnet Erdakos,¹ Shih Ying Chang,¹ Anthony Cavallaro,^1
1Sonoma Technology, Inc., Petaluma, CA
²United States Forest Service Pacific Southwest Research Station, Davis, CA

The NOAA National Air Quality Forecast Capability (NAQFC) provides important information about air quality conditions that may pose a significant risk to human health. The NAQFC system includes smoke forecasts based on the HYSPLIT dispersion model, and leverages the BlueSky Smoke Modeling Framework. Wildfires can contribute a significant fraction of observed PM2.5 during severe smoke episodes. Wildfire activity has been increasing in the western United States in recent decades, and this trend is expected to continue because of climate change and past fire management practices. A quantitative description of fire emissions and their impact on air pollution remains a significant challenge for air quality forecasting, as estimating fire emissions requires the synthesis of numerous data and tools for estimating fire activity, burn acreage, fuel loading, consumption, and emissions.

The goal of this project is to reduce the uncertainty of smoke emissions estimates and improve HYSPLIT smoke forecasts through improved characterization of biomass burning conditions. A prototype fire modeling pathway was developed using fire activity data derived from the VIIRS 375 m Active Fire Product, daily fire perimeter progression maps from the Geospatial MultiAgency Coordination (GeoMAC) wildfire support system, and 30 m FCCS fuel loading data from LANDFIRE to develop improved estimates of fire size and fuel available to burn. This fire modeling pathway improves upon the current NAQFC process by providing more spatially resolved fire size information and biomass fuel loading estimates. These data are then used in the BlueSky Framework to provide improved estimates of fuel consumption, fire emissions, and smoke dispersion.

This prototype fire modeling pathway was evaluated for a July 2016 test case, and the resulting burn acreage, fuel loading, and fire emissions were compared to predictions from the operational NAQFC system. Burn acreage estimates from the prototype pathway were a factor of four higher than estimates from the operational NAQFC system, in part because the GeoMAC data stream captured several fires in the West that were missed in the NAQFC inventory. Fuel loading, fuel consumption, and PM2.5 emissions from the prototype pathway were a factor of two higher nationwide compared to the NAQFC system.

The resulting fire emissions were input into a HYSPLIT modeling simulation driven by meteorology from the 12-km resolution NAM model. The predicted ground-level and column-average PM2.5 concentrations due to fires were compared to smoke predictions from the operational NAQFC system, evaluated qualitatively against satellite imagery and AOD retrievals, and evaluated quantitatively against surface PM2.5 observations near large fires in the western United States. Results from these analyses showed that the use of improved fuel loading and fire emission estimates improved forecast performance during the July 2016 test period. Based on these results, we anticipate these improvements would have long-lasting benefits for the NAQFC and other operational smoke prediction systems.

Numerical study of the interaction of synoptic weather anomalies and the transportation of biomass burning emission from Indochina to elevated ground in Taiwan

Maggie Chel Gee Ooi^1,2, Ming-Tung Chuang², Steven Soon-Kai Kong^1,2, Wei-Syun Huang¹, Neng-Huei Lin¹

¹ Department of Atmospheric Sciences, National Central University, Taoyuan, Taiwan.

² Graduate Institute of Energy Engineering, National Central University, Taoyuan, Taiwan.

The frequent occurrences of biomass burning (BB) in north Indochina as a result of farmland clearing and wildfires have found to be transported to southern China, Taiwan and even across the Pacific Ocean. The fire hot spots are detected as soon as December and last until May, however, the transboundary BB plume is not commonly apparent until the month of March and April at the Lulin Atmospheric Background Station, Taiwan (LABS; 2862m AMSL; 23.28.07° N, 120.52.25° E). The India-Burma trough has brought in the dry air to sustain the fire, while the higher terrain over the northcentral Indochina has lifted the BB plume into the subtropical Pacific high (700 to 800 hPa) and transported eastward to LABS. The higher pollutant levels (CO > 300 ppb, O3 >75 ppb, PM2.5 > 50 ug m-3, PM10 > 30 ug m-3) are recorded at LABS when the previous year experienced stronger El-Nino Southern Oscillation (ENSO) winter (Dec to Feb; MEI > 0.5), as seen in 2007, 2010 and 2012. With the CMAQ model, the emission and transportation of BB plume and their mutual interaction are scrutinized under the different ENSO conditions. The plume is more likely to be transported to LABS under the strong influence of ENSO compared to La-Nina condition. Such phenomenon, however, does not apply under the extreme ENSO condition where the plume is transported further north of the station. It is interesting to note that the BB emission in Indochina (shown from the MODIS fire counts and MEYYA2 AOD) has indirectly correlated with ENSO. The westerlies is greatly affected by the presence of ENSO system, so does the India-Burma trough. This work has also attempted to answer the additional anomaly effect to ENSO system on the BB aerosol formation.

Year-long Simulation of PM2.5 in Pearl River Delta using WRF-SMOKE-CMAQ System

Xuguo ZHANG; Jimmy FUNG; Alexis LAU; Wayne HUANG

The booming of the economy in Hong Kong and PRD results in severe air pollution problem. Although county based monitoring network for criteria pollutants has been constructed for the whole China, large concentration gradient still exists. The temporal and spatial coverage is still limited and the PM components remain unknown, limiting further studies on control strategies and associating human exposure. In this study, a yearlong air quality simulation using the Weather Research and Forecasting (WRF) model and the Community Multi-scale Air Quality (CMAQ) model was conducted to provide detailed temporal and spatial map of ozone O3, total PM2.5, and chemical components. MIX Emission Inventory for outside PRD was used for anthropogenic emissions and observation data obtained from the national air quality monitoring network were collected to validate model performance. The model successfully reproduces the PM2.5 concentrations at most cities for most months, with model performance statistics meeting the performance criteria proposed by US EPA. Under prediction of PM2.5 occurs for the whole year. The upper wind bring pollutants to the downwind direction accumulating itself. Seasonal variations and reasonable spatial distribution were captured successfully by this new system. The model performance at different cities differs due to the differences in emission, topography, and meteorological conditions. The performance on these species can be used as indicator for emission uncertainties.

Trans-Pacific transport of tropospheric ozone from East Asia to the U.S. based on the CMAQ extended for hemispheric application (H-CMAQ) with the higher-order decoupled direct method (HDDM)
Syuichi Itahashi (North Carolina State University, USA; 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)

As anthropogenic emissions from East Asia increase, they affect air quality not only on the regional scale but also on global scales. The Community Multiscale Air Quality (CMAQ) modeling system has been recently extended by the U.S. EPA to the hemispheric scale (H-CMAQ). Based on the H-CMAQ, we have performed a comprehensive analysis to study the trans-pacific transport of tropospheric ozone (O3) from East Asia to the U.S. We focus on April 2010 when exceedances of the U.S. National Ambient Air Quality Standard (NAAQS) for O3 were widely observed across the U.S. Comparisons with surface observation network, ozonesonde, and satellite showed that the H-CMAQ can capture O3 during April 2010. To investigate the trans-Pacific transport of O3, in this work, we employed the higher-order decoupled direct method (HDDM) technique in H-CMAQ. We found that ozone formation during April 2010 was limited by VOCs over East Asia whereas it was limited by NOx over the U.S. except California, the Great Lakes, and New England. We will analyze the sensitivities of the model predictions to region-specific emissions (i.e., East Asia and the U.S.) to quantify the impacts of air pollutants from East Asia on the U.S. air quality. This work will improve our understanding of the role of trans-Pacific air pollution based on the state-of-the-art hemispheric modeling system.

Model Evaluation of Environment and Climate Change Canada's FireWork Air Quality Model: a Retrospective Analysis for the 2017 British Columbia Wildfire Season
Bruce Ainslie and Rita So

A retrospective analysis of the 2017 wildfire season will be conducted over British Columbia to examine the effects of improved spatial resolution (2.5 vs 10km), chemistry (zero, 2-bin, 12-bin) and wildfire emission modules on the model performance for fine particulate matter and ozone. Preliminary analysis indicated that air quality models, such as FireWork and BlueSky, cannot outperform a persistence model, despite the fact that various user groups have found the model forecasts to be useful. As a result, a user survey was conducted to better understand how the forecasts were used by the regional air quality and health agencies when issuing advisories. Based on the survey results, a novel set of model evaluation metrics was developed to better reflect how the model guidance is used, in addition to examining model performance at the individual stations. An event-based model evaluation method, aggregated at the local health area level, will be used to assess model performance. Feature-based model evaluation methods will also be investigated. A gridded PM2.5 product, which combines observations and MODIS Aerosol Optical Depth, was developed and will be used as part of model evaluation. This study will also examine the ozone over-prediction problem that appears to be common for various air quality modeling systems.

Quantifying the influence of boundary conditions and anthropogenic emissions to ozone concentrations towards estimating the modeled controllable portion of the ozone burden in continental United States

M. Astitha1, H. Luo1, C. Hogrefe2, R. Mathur2, and S.T. Rao1,3

1University of Connecticut, 2US Environmental Protection Agency, 3North Carolina State University

This study investigates contributions from boundary conditions and anthropogenic emissions to ozone exceedances and infers the modeled controllable portion of the tropospheric ozone burden across the continental United States. We use a set of simulations performed with the WRF-CMAQ model with a horizontal grid spacing of 12 km as part of the third phase of the Air Quality Model Evaluation International Initiative (AQMEII3) (Hogrefe et al.,2018). The simulations cover the year 2010 as the base simulation and are augmented by three simulations with the same meteorology: one with lateral boundary conditions for all species set to zero; the second with all anthropogenic emissions as well as wildfire emissions within the domain set to zero and a third one with a 20% reduction of anthropogenic emissions both in the global model simulations that provide lateral boundary conditions and within the WRF-CMAQ modeling domain. With the help of these simulations, we examine the drivers behind the baseline ozone concentration, defined as the long-term component of the spectral decomposition of ozone maximum daily 8hr concentration time series that exhibits significant influence on ozone exceedances. We determine the impacts on the ozone design value and assess the modeled controllable portion of ozone for various regions across CONUS.

Model Intercomparison and Evaluation of Particular Matter - study for MICS-ASIA III

Jiani Tan1, Joshua S. Fu1, Kan Huang1, Syuichi Itahashi2, Kazuyo Yamaji3, Tatsuya Nagashima4, Yu Morino4, Xuemei Wang5, Yiming Liu5, Hyo-Jung Lee6, Jeong-Eon Kang6, Chuan-Yao Lin7, Baozhu Ge8, Mizuo Kajino9, Zhining Tao10, Jia Zhu8, Meigen Zhang8

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

2 Central Research Institute of Electric Power Industry, Abiko, Chiba, Japan

3 Graduate School of Maritime Sciences, Kobe University, Kobe, Hyogo, Japan

4 National Institute for Environmental Studies, Tsukuba, Ibaraki, Japan

5 School of Atmospheric Sciences, Sun Yat-Sen University, Guangzhou, China

6 Department of Atmospheric Sciences, Pusan National University, Busan, South Korea

7 Research Center for Environmental Changes Academia Sinica, Taiwan

8 Institute of Atmospheric Physics, Chinese Academy of Science, China

9 Meteorological Research Institute, Japan Meteorological Agency, Japan

10 USRA at GSFC, Code 614, NASA Goddard Space Flight Center, Greenbelt, MD, USA

The main purpose of the Model Intercomparison Study Asia Phase III (MICS-ASIA III) is to understand the strength and weakness of models in simulating air quality in East Asia. Fourteen regional chemistry-transport models (CTM) participated in this project, including WRF-CMAQ (v4.7.1 and v5.0.2), WRF-Chem (v3.6.1 and v3.7.1), GEOS-Chem, NHM-Chem, NAQPMS and NU-WRF. This study inter-compares the model performances on simulating the spatial distribution and monthly variation of particular matter over East Asia. The modelled PM10, PM2.5, sulfate, nitrate, ammonium (SNA) and Aerosol Optical Thickness (AOT) are evaluated with observation from Asian networks including Acid Deposition Monitoring Network in East Asia (EANET), Air Pollution Indices (API), Aerosol Robotic Network (AERONET), Moderate Resolution Imaging Spectroradiometer (MODIS) satellite products and local monitoring data. We also study the inter-model variations among models on both spatial and temporal scales to identify the model uncertainties. The model bias with observation and inter-model differences can be attributed to various model processes related to model inputs and model mechanisms. In this study, we identify 4 main possible model processes/mechanisms that contribute to model bias: (1) model sources of particles: model input and boundary conditions, (2) model process related to the formation of fine particles: choice of gas and aerosol module and gas-aerosol phase equilibrium, (3) model production of coarse particles: effects of dust emission and mechanism, (4) model removal of particles: wet and dry deposition and deposition velocity. This study gives an insight into the model application of CTMs in East Asia, and offer suggestions for future model development.

Modeling the effects of climate change on US mid 21st century PM2.5 and O3 by dynamical downscaling of meteorology and chemistry from a global model

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

Human induced climate change, land use change and the resulting changes in natural emissions (fires, biogenic VOCs) are anticipated to impact PM_2.5 and O₃ levels in the coming decades. An important question in this context is distinguishing the climate change signal in air quality from natural climate variability. Here we study the effects of climate change (under the RCP8.5 scenario) by dynamically downscaling from the global chemistry-climate model GFDL-CM3 (2.5^o - 2^o resolution) using regional climate and chemistry models WRF and CMAQ (12 km resolution).

To isolate the impact of RCP8.5 climate change from 2006-2100, the GFDL-CM3 simulations fix aerosol and ozone precursor emissions to the year 2005. Based on an Empirical Orthogonal Function (EOF) analysis of the GFDL-CM3 simulations, we carefully select current (2006-2015) and future (2040-2060) years that represent the GFDL-CM3 upper quartile and median of the probability distributions of PM_2.5 and O₃ for each US region (NE, SE, SW, NW, Midwest, Intermountain West). For the selected years (one present year and one future year as examples), we downscale the meteorology in WRF with updated RCP8.5-compatible land use land cover in the future, and the chemistry in CMAQ. We also compare the modeled meteorology, PM_2.5 and O₃ with the global simulations, and the present day simulation with observations.

Our analysis to date is a first step towards combining statistical analysis of three ensemble members from the GFDL-CM3 21^st century simulation with the downscaled CMAQ simulations, allowing us to both spatially refine PM_2.5 and O₃ probability distributions from the global model and examine feedbacks (from biogenic VOC emissions and fires) that are not included in the global chemistry-climate model. Specifically, we evaluate the extent to which conclusions about mean and high seasonal/annual PM_2.5 and O₃ drawn from the global model for the selected years are consistent with those from the WRF and CMAQ simulations. Analyzing the US regional PM_2.5 and O₃ distribution changes with both global and regional models will improve our understanding of the meteorological drivers of future changes in air quality, and aid air quality, health and visibility planning for different US regions in the coming decades.

Influence of Brazilian Land Use Data on WRF Modeling Performance

Rizzieri Pedruzzi, Willian Lemker Andreko, Bok H. Baek, Jared Bowden and Taciana Toledo de Almeida Albuquerque.

Land use influences the energy balance, friction velocity, albedo and heat flux, and it can affect meteorological variables, such as ambient temperature, specific humidity and wind speed, which may bias the representativeness with real data. Thus, it is an important input for Weather Research and Forecasting (WRF) simulations.

WRF has its own database for land use, and for Brazil it can be chosen between USGS land use (24 categories) or MODIS (20 or 21 categories). The USGS database was developed based on satellite images from 1992 to 1993 and MODIS database was developed with satellite images from 2001 to 2005. It is known that land uses changes over the years and sometimes these two databases are not representative for the study area, including Brazil, which can have a huge impact in WRF model performance.

In 2014, the Instituto Brasileiro de Geografia e Estatestica (IBGE) created the land use dataset over Brazil based on satellite images from MODIS and LANDSAT, with 14 classes: artificial area; agricultural area; planted pasture; mixed agriculture/forest remnants; forestry; forest vegetation; mixed forest vegetation/agricultural areas; campestral vegetation; wetland; natural pasture; mixed agricultural area/campestral vegetation remnants; continental water bodies; coastal water bodies; and discovered area. In addition to images, it was used state reports and database from Instituto Nacional de Pesquisas Espaciais (INPE) to improve the accuracy of land use classification.

Comparing the land use classification from MODIS and the developed by IBGE, it was observed that there are some differences between them. Since land cover plays a key role in WRF simulations and a Brazilian land use classification is available, we have focused on WRF modeling performances with the reclassified IGBE land use data, also we compared the outcomes against the ones using MODIS land use datasets.

Estimating public health impacts of PM2.5 using fine-scale hybrid-modeled concentrations

Parvez, Fatema, Wagstrom, Kristina

Growth of the global motor-vehicle fleet due to population growth, economic improvement, metropolitan expansion, and increasing motor vehicles dependence has increased the number of people living and working near busy roads. As elevated concentrations near major roads typically reach background levels within a few hundred meters, conventional regional-scale modeling approaches cannot predict these sharp concentration gradients. This leads to a failure to effectively estimate health impacts from roadway emissions. In this study, we develop a hybrid-modeling framework combining a regional chemical transport model, CAMx, and a road dispersion model, R-LINE, to estimate hourly combined pollutant concentrations at 40mx40m resolution. We implement this modeling framework and estimate PM2.5 concentrations in three major cities in Connecticut: Hartford, New Haven, and Windham. We use these updated concentration estimates to calculate the difference in the estimated health impact from PM2.5 in the year 2011 using BenMAP-CE. We compare the estimated health impacts from the high resolution concentrations to those from regional concentrations estimates and using a nearest monitor approach. This provides an estimate of the likely under-prediction of health impacts resulting from not accounting for the sharp concentration gradients. In urban areas a significant fractions of the population lives in close proximity to roadways. Our fine scale modeling technique captures the elevated concentrations near the roadways. This leads to increased estimates of mortality and morbidity. However, at locations farther from major roads, estimates of morbidity and mortality decrease. We also observe an increase in the likelihood of emergency department visits, deaths from cardiovascular disease, and deaths from respiratory disease in the urban core where population density is highest. These results indicate that using regional air pollutants concentrations could lead to an under-prediction of the health impacts from air pollution exposure.

Model inter-comparison study for urban scale secondary atmospheric pollutants in Japan

Kazuyo Yamaji, Satoru Chatani, Kyo Kitayama, Syuichi Itahashi, Hiroshi Hayami, Tatsuya Sakurai, Hikari Shimadera, Masayuki Takigawa

A model inter-comparison study in Japan, Japan's study for reference air quality modeling (J-STREAM) project were started to investigate gaps in simulated secondary atmospheric pollutant concentrations due to differences between models and/or model settings and then to propose a favorable model setting. Each participated model, CMAQ, CAMx and WRF-Chem with several types of model settings performed nested air quality simulation from Asia to urban scale in Japan. All the model settings overestimated the ozone concentrations observed over Japanese urbanized areas in summer time. On the other hand, most model settings tended to underestimate total PM2.5 concentrations observed in Japanese urbanized areas, however the models could capture temporal day-to-day change of PM2.5. The concentration gaps resulting from differences of models and model settings were affected by both regional scale chemical and meteorological descriptions.

WUDAPT's next generation of Urban Canopy Parameters for advanced multi-scale Weather, Climate and Air Quality models
Jason Ching and Adel Hanna

The World Urban Database and Access Portal Tools (WUDAPT) project objective is to capture consistent information on urban form and function for cities worldwide that can support multiscale "Fit for Purpose" urban modeling. This information is captured in the form of urban canopy parameters (UCPs) that are used by models to simulate the effects of urban surfaces on the overlying atmosphere. WUDAPT's approach is to acquire, store and disseminate this information at different levels of detail. The lowest level employs the Local Climate Zone (LCZ) scheme that provides ranges of UCP values at a relatively crude scale. However, advanced modeling applications need data at higher levels of precision to describe the spatial heterogeneity present in cities and at different urban scales. This paper presents a pathway to generating such data. For this we introduce in the WUDAPT Portal, a highly innovative Digital Synthetic City (DSC) tool, which simulates the 3D building and road elements from Google type imagery that make-up an entire city landscape. This tool requires readily available data to perform the simulation and comparisons with real-world urban data are very encouraging. The UCPs derived from this simulated landscape can be generated at any desired scale, meeting the fit-for-purpose goal of WUDAPT. We further discuss developments to customize the DSC based on implementing a unique TESTBED concept to address building typology and architectural variation across the world based. Finally, the outcomes from these efforts will be detailed UCP data that can be generated at selected grid sizes; these data itself will be stored using a coded string, where each of the parts represent values for sets of key UCPs for each and all grid cells.

Evaluation of CMAQ simulated NH3 and PM2.5 concentration in Taiwan with dynamical NH3 emission parameterization

Chia-Hua Hsu and Fang-Yi Cheng

Ammonia (NH3) is an important precursor to the inorganic PM2.5 such as ammonium sulfate, ammonium bisulfate, and ammonium nitrate. Recent studies also show that soil-atmosphere exchange of NH3 is an important process of the NH3 cycle. In the 2010 Taiwan emission inventory (TEDS8.1), the NH3 emission is about 200 kilo-tons per year, and is majorly emitted from livestock operation, urban sewage and agricultural fertilizers. Most air quality simulations in Taiwan assume the fixed NH3 emission rate due to the lack of the detailed temporal information such as the seasonal and diurnal variation. With the fixed NH3 emission rate, the CMAQ simulation exhibits a large positive bias in simulated NH3 concentration with mean bias around +25.52 ppb (+206.93%) in western Taiwan. The highest bias occurs in the metropolitan cities, such as in Taipei, where the bias reaches +243.82%. The over-prediction of the NH3 also leads to the excess of nitrate formation with mean bias about +5.16 μg/m3 (+265.47%).

To reduce the simulated bias, we applied dynamical NH3 emission parameterization on the emission from livestock and agricultural fertilizer to better represent the temporal evolution of NH3 emission. With the application of the diurnal and monthly temporal variation of the NH3 emission, the CMAQ simulation is improved but still overestimate the NH3 (+18.38 ppb). The possible reason for the simulated NH3 bias could be the excess of the sewage NH3 emissions estimated in the urban areas. In the 80% seweage NH3 emission reduction experiment, the simulated NH3 (nitrate) concentration in the urban region is much closer to the observation with mean bias of -0.88 ppb (3.56μg/m3) but still overestimate in the agricultural region.

In the future we will apply ammonia bidirectional model in Taiwan region to better simulate atmosphere-land interaction of NH3 in the agriculatural region.The detailed analysis and simulation results will be presented during the conference.

Using MODIS AOD and WRF-Chem to infer daily PM2.5 concentrations at 1 km resolution in the Eastern United States

Dan Goldberg, Pawan Gupta, Kai Wang, Chinmay Jena, Yang Zhang, Zifeng Lu, and David Streets

To link short-term exposures of air pollutants to health outcomes, scientists must use high temporal and spatial resolution estimates of PM2.5 concentrations. In the air quality modeling, measurement, and satellite communities, data are assembled over very large spatial extents, but at coarse spatial resolutions. Thus, there is a mismatch between the needs of the epidemiology community, the computational limits of chemical transport models, the spatial coverage of ground monitoring networks, and the physical constraints of satellite data. In this work, we develop a daily PM2.5 product at 1 x 1 km² spatial resolution across the eastern United States (east of 90° W) with the aid of MODIS aerosol optical depth (AOD) data, 36 x 36 km² WRF-Chem output, 1 x 1 km² land-use type from the National Land Cover Database, and 0.125 x 0.125 ERA-Interim re-analysis meteorology. A gap-filling technique is applied to MODIS AOD data to construct robust daily estimates of AOD when the satellite data are missing (e.g., areas obstructed by clouds or snow). The input data are incorporated into a multiple-linear regression model trained to surface observations of PM2.5 from the EPA Air Quality System (AQS) monitoring network. Of the inputs to the statistical model, WRF-Chem output is the most important contributor to the skill of the model, while MODIS AOD is also a strong contributor. Daily PM2.5 output from our statistical model can be easily integrated into county-level epidemiological studies. The novelty of this project is that we are able to simulate PM2.5 that is constrained to ground monitors, satellite data, and chemical transport model output at high spatial resolution (1 x 1 km²) without sacrificing the temporal resolution (daily) or spatial coverage (>2,000,000 km²).

Scientific and Structural Developments in CMAQv5.3

Ben Murphy, H. Pye, J. Bash, K. Fahey, B. Hutzell, D. Kang, C. Nolte, G. Sarwar, T. Spero, J. Pleim, R. Mathur, and the CMAQ Model Development Team

The Community Multiscale Air Quality (CMAQ) model has undergone substantial improvements in its representation of the sources and sinks of atmospheric pollutants. This paper will motivate and explain the major updates in the context of the overall model framework for predicting speciated particulate matter, ozone and other regulated pollutants. Examples of important updates include a new aerosol processing module (AERO7), implementation of halogen chemistry (simple and detailed) and dimethyl-sulfide chemistry, a new version of AQCHEM-KMT taking into account more inorganic/organic chemistry, support for multiple deposition approaches including the new Surface Tiled Aerosol Gas Exchange (STAGE) scheme for pollutant deposition, updates to the M3Dry deposition scheme, thorough revision of infrastructure for emissions input and scaling, improvements to standard and diagnostic outputs, a new release of MCIPv5.0, and many other advances. The extensive additions to AERO7 allow CMAQ to treat the source-specific volatility of combustion-derived OA compounds (e.g. gas and diesel vehicles, wildfires, residential wood combustion, prescribed burns, etc), include the partitioning of water to the particulate organic phase, and represent the high degree of secondary organic aerosol (SOA) production observed from monoterpene precursors in the ambient atmosphere and laboratory studies.

We will show preliminary impacts of the most important changes on model performance, but discussion will primarily focus on the assumptions that are needed or relaxed by the inclusion of the new model techniques. The developments we discuss will be available to the community prior to the meeting via the release of a Beta version of CMAQv5.3 (CMAQv5.3.b1), and we encourage feedback and testing by the community. We will also lay out the current and ongoing priorities for further development and characterization that are identified by the EPA CMAQ model development team and the schedule for the public release of CMAQv5.3.

Projecting PM2.5 Spatial Fields to Correspond to Just Meeting National Ambient Air Quality Standards

James T. Kelly, Carey Jang, Brian Timin, Brett Gantt, and Adam Reff

PM_2.5 concentrations that correspond to just meeting existing or potential alternative National Ambient Air Quality Standards (NAAQS) have been used in risk assessments conducted during previous NAAQS reviews. Measured PM_2.5 concentrations were projected according to prescribed spatial patterns to just meet each NAAQS under consideration. In this study, we make two improvements on previously used methods. First, PM_2.5 is projected using response factors developed by combining Community Multiscale Air Quality (CMAQ) model predictions with ambient observations. Second, gridded spatial fields of PM_2.5 are projected according to the emission reductions required for PM_2.5 monitors to just meet the NAAQS in an area. In this presentation, we will discuss the development of PM_2.5 spatial fields, the development of PM_2.5 response factors from CMAQ sensitivity simulations and observations, and the software developed to implement methods. A case study will be used to illustrate techniques.

NOx emission reduction co-benefits for secondary organic aerosol formation

Havala O. T. Pye, Emma L. D'Ambro, Ben H. Lee, Siegfried Schobesberger, Masayuki Takeuchi, Yue Zhao, Felipe Lopez-Hilfiker, Jiumeng Liu, John E. Shilling, Jia Xing, Rohit Mathur, Ann Middlebrook, Jin Liao, Andre Welti, Martin Graus, Carsten Warneke, Joost de Gouw, John Holloway, Thomas Ryerson, Ilana Pollack, and Joel A. Thornton

Organic aerosol is a significant component of PM2.5 and is an increasing fraction of the total in many locations. Organic aerosol is largely secondary in nature and forms from the gas to particle conversion of low volatility or semi-soluble species. In the southeast US, monoterpene oxidation is responsible for half of the total organic particulate matter. In this work, we examine an efficient pathway to organic aerosol that is expected to account for an increasing fraction of monoterpene chemistry as NOx declines. This unimolecular pathway to secondary organic aerosol (SOA) poses a potential penalty for reducing NOx in terms of PM2.5. Using ambient observations downwind of Atlanta, an explicit mechanism, and CMAQ model calculations, we show that this NOx penalty can be fully offset by reductions in oxidants that come with reducing NOx. As a result, we predict a positive coupling between anthropogenic NOx and regional biogenic SOA from unimolecular monoterpene oxidation pathways.

Modeling PM2.5 concentrations from wildland fires for health impact assessments

Xiangyu Jiang1, Eun-Hye Yoo1

1Department of Geography, University at Buffalo, Buffalo, NY, 14260

Wildfire has become intense and frequently occur across the U.S. in recent years due to climate change. Increasing number of studies reported that wildfire contributed to poor air quality and had adverse impacts on public health. While some epidemiological studies have demonstrated the association between wildfire-related air pollution, such as fine particulate matter (PM2.5), and increased rates of hospital admissions and mortality, these findings are inconsistent from the challenges in the accurate estimation of air pollution concentrations specific to wildland fires. In the present study, we aim to investigate their causal associations by predicting daily air pollution concentrations from wildland fires and estimating adverse health outcomes during wildland fire periods. To illustrate our point, we modeled wildfire-related PM2.5 concentrations using Community Multiscale Air Quality (CMAQ) v5.2 over New York State for the year 2014. We ran the CMAQ model with and without fire emissions, respectively, and calculated the difference between two simulations as the air pollution contributed by wildland fires. Additionally, we quantified excess mortality and hospital admissions attributable to exposure to wildfire-related PM2.5 using the Benefits Mapping and Analysis Program (BenMAP).

Introducing the Surface Tiled Aerosol and Gaseous Exchange (STAGE) dry deposition option in CMAQ v5.3

Jesse O. Bash¹, Donna Schwede¹, Patrick Campbell¹, Tonya Spero¹, Wyat Appel¹, Rob Pinder²

1. U.S. EPA/NERL

2. U.S. EPA/OAQPS

There has been active development of the Community Multiscale Air Quality (CMAQ) model to better estimate atmospheric deposition for terrestrial and aquatic ecosystem health applications. A new tiled, land use specific, dry deposition scheme has been developed to improve simulations of critical loads, total maximum daily load (TMDL) and research applications to evaluate the impact of dry deposition on ambient air quality in CMAQ v5.3. This new scheme explicitly supports Weather Research and Forecasting (WRF) simulations with the Noah and Pleim-Xiu land surface schemes. A model resistance framework that parameterizes air-surface exchange as a gradient process and is consistent between bidirectional exchange and dry deposition will be introduced. A brief evaluation of box model HNO3, NH3, SO2, and O3 fluxes against micrometeorological observations will be presented. Continental-scale simulations using a beta version of CMAQ v5.3 are evaluated against network wet deposition and ambient concentrations to indicate where this option improves or degrades model performance. Preliminary simulations using the STAGE option provide more consistent coupling with water quality and ecosystem models as well as areas of improved model performance when evaluated against wet deposition and ambient concentrations, particularly coastal ozone. Insights gleaned into the processes involved in model improvements will be presented. Additionally, the impact of including subgrid air-surface exchange processes on grid resolution and the mapping of deposition results to high-resolution land use data will be discussed.

Forecasting Exposure to Prescribed Fire Smoke for Health Protection in Southeastern USA

M. Talat Odman¹, Ha H. Ai¹, Yongtao Hu¹, Armistead G. Russell¹, Ambarish Vaidyanathan¹ and Scott L. Goodrick²

1. School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0512, USA

2. Center for Forest Disturbance Science, Southern Research Station, US Forest Service, Athens, GA 30602, USA

Prescribed fire is the leading source of PM_2.5 emissions in southeastern USA. We developed the HiRes-X modeling system to forecast prescribed fires and their impact on air quality and human health. We disseminate these forecasts through the Southern Integrated Prescribed Fire Information System (SIPFIS) HiRes-X uses innovative modeling and statistical learning approaches to identify areas most impacted by prescribed fires. Specifically, a regression tree model is built using meteorology and prescribed fire data from recent years to generate highly resolved prescribed fire forecasts. Smoke emissions are calculated using satellite enhanced fuel maps, fuel consumption estimates and region specific emission factors. Community Multiscale Air Quality (CMAQ) model is used to compute the contribution of smoke emissions to local and regional air quality. HiRes-X modeling system is integrated with measures of cardiorespiratory health impacts and social vulnerability to identify communities vulnerable to smoke from prescribed fires. SIPFIS dashboard provides map and chart visualization tools that are built using open source software and features interactive capabilities to respond dynamically to user selections. Analyses that can be performed with SIPFIS include comparisons of prescribed fire forecasts to burn permit records or satellite fire detections, and air quality forecast to observations. SIPFIS can be used for managing prescribed burning operations to reduce human exposure to fire smoke and for emergency planning and preparedness purposes.

FEST-C v1.4: An integrated agriculture, atmosphere, and hydrology modeling system for ecosystem assessments

Limei Ran¹, Ellen Cooter¹⁶, Dongmei Yang², Yongping Yuan¹, Verel Benson³, Jimmy Williams⁴, Jonathan Pleim¹, Ruoyu Wang⁵, Adel Hanna², Val Garcia¹, Rohit Mathur¹

¹Computational Exposure Division, ORD NERL/USEPA, Research Triangle Park, NC, USA

²University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA

³Benson Consulting, Columbia, Missouri, USA

⁴Blackland Research & Extension Center, Texas A&M University, Texas, USA

⁵University of California, Davis, California, USA

⁶Retired

We have developed an agricultural modeling system using the Fertilizer Emission Scenario Tool for CMAQ (FEST-C) and Environmental Policy Integrated Climate (EPIC) agricultural cropping model for any defined conterminous United States CMAQ domain and grid resolution. Integrated with the WRF/CMAQ meteorology and air quality models, the system simulates plant growth, fertilization, production, hydrology, and complete soil biogeochemical properties under various management practices. The system was originally designed for generating required input for CMAQ simulations with the bi-directional NH3 option. Since the first release of FEST-C v1.0 in October 2013, the system has gone through many updates and enhancements up to the recent release of FEST-C V1.4 in July 2018. FEST-C v1.4 is enhanced to better integrate the Soil and Water Assessment Tool (SWAT) modeling system with EPIC and WRF/CMAQ output for improving our understanding of hypoxia in the Gulf of Mexico and other applications. These enhancements have advanced the system capabilities from simply supporting CMAQ simulations, to becoming a valuable tool for integrated, one-biosphere assessments of air, land, and water quality in light of social drivers and human and ecological outcomes. Therefore, it is important to evaluate all modeling components in this newly released integrated system in order to understand their strengths and identify issues for future improvements and applications. We will present the updates and enhancements on the interface and EPIC model in FEST-C v1.4. The presentation will focus on evaluating simulated yields and nitrogen fertilization from EPIC against USDA and USGS reports. We will also present and evaluate simulated NH3 flux and NH4+ wet deposition for 2011 by CMAQ v5.3 with a new bi-directional NH3 flux model which is directly coupled with EPIC. Simulated runoff and water quality by the integrated modeling system in the Mississippi River Basin will also be presented to demonstrate the capability of FEST-C V1.4 for facilitating hydrology and water quality assessment.

An integrated modeled and measurement-based assessment of particle number concentrations from a major US airport

Chowdhury G. Moniruzzaman, Kevin Lane, Jonathan I. Levy, Chloe Kim and Saravanan Arunachalam

Airports are an important source of ultrafine particulate matter (UFPM) that affects human heath in nearby communities through cardiovascular and pulmonary diseases. Aircraft emits primary UFPM and precursor gas of secondary UFPM formed in the atmosphere. The smaller size of UFPM increases its residence time in air and advection further downstream. The mass concentration of UFPM is very small comparing with particles having size up to 2.5 micron (PM2.5) in diameter and measurement of mass concentration of these small particles is sometimes challenging. Most of the assessment to-date of aircraft emissions to air quality has focused on PM2.5 mass. But increasing evidence has pointed to the role of UFPM in the vicinity of airports. Particle number concentration (PNC) has been used as a measure of UFPM in modeling and health impact studies. Dispersion modeling can provide airport attributable PNC at fine spatial resolution locations with close proximity to airports that can be used in potential health impacts study. PNC exposure rates change through emissions and deposition as well as aerosol microphysics such as nucleation and coagulation. The Second-order Closure Integrated Puff model with chemistry (SCICHEM) is a Lagrangian transport and diffusion model with gas and aqueous phase chemistry and aerosol microphysics which can be used to study the dispersion of UFPMs at high resolution spatial locations from airport emission sources. The unique features of SCICHEM model is that it has chemistry and aerosol dynamics which can better simulate the pollutants dispersions comparing with other dispersion models, although at a very high computational cost. The present study is focused on an integrated assessment of UFPM from Boston Logan International Airport using a measurement and a modeling approach. We used SCICHEM, a reactive puff model for Boston Logan airport to predict UFPM from landing and take-off (LTO) cycles from aircraft operations at very fine scales in the vicinity of the airport. The model predicted concentrations will be compared with observations from the field campaign at multiple locations along the aircraft flight paths near Boston. We will present results from this integrated assessment, from both modeled and measurement-based approaches, with specific focus on understanding aircraft contribution to UFPM.

Mechanistic representation of Soil N emissions in CMAQ v5.1

Quazi Ziaur Rasool1,2, Jesse Bash3 and Daniel S. Cohan1

1Department of Civil and Environmental Engineering, Rice University, Houston, TX

2Now at Department of Environmental Sciences and Engineering at UNC-Chapel Hill, NC

3Atmospheric Modeling and Analysis Division, National Exposure Research Laboratory, Office of Research and Development, US Environmental Protection Agency, RTP, NC, USA

Soils are a major and long overlooked source of reactive nitrogen emissions. These emissions include species like nitric oxide (NO), nitrous acid (HONO), nitrous oxide (N2O), and ammonia (NH3), generated by biogeochemical transformations of soil nitrogen reservoirs. They are more pronounced in the summer ozone season (growing season) and may become increasingly important as fertilizer use grows, while fossil fuel combustion sources of nitrogen decline. Soil NOx emissions can contribute up to 40% of tropospheric NO2 columns over the agricultural plains in the continental U.S. during growing season. Mechanistic process models of soil N emissions are used in the earth science and soil biogeochemical modeling community, on a plot or site scale. Regional scale air quality models like CMAQ have been using a mechanistic approach only for NH3, while using simpler parametric approaches for NO and typically neglecting soil emissions of HONO and N2O. Even our previous update to the parametric soil NO scheme (BDSNP) in CMAQ used a total weighted soil N and did not track the biogeochemistry generating different soil N emissions. Thus, there is a need to more mechanistically and consistently represent the soil N processes that lead to emissions to the atmosphere. Our work adds a new mechanistic scheme for modeling soil N emissions in CMAQ, integrating nitrification and denitrification mechanistic processes that generate NO, HONO, and N2O under different soil conditions and meteorology. For agricultural soil, our mechanistic scheme uses daily soil N pools from the EPIC simulations. Unlike BDSNP, the new mechanistic model tracks different forms of soil N pool as NH4, NO3, and organic N for different soil layers and vegetation types to be consistent with requirements of nitrification and denitrification modules from DAYCENT. As an update over the constant soil emission factors used in parametric schemes for non-agricultural soil, our new mechanistic scheme uses a global soil nutrient dataset in an updated C and N mineralization framework. This enables tracking the conversion of organic soil N to NH4 and NO3 pools on a daily scale for non-agricultural soils. Comparisons of the model estimated NO and HONO emissions from the new mechanistic scheme with other existing schemes during the growing season are presented. Model results from parametric schemes and the new mechanistic scheme are also evaluated against observed aerosol and ozone concentrations and satellite observations of NO2.

CMAS Users Special Forum: CMAS Data Warehouse and Modeling Platform Development Discussion
Coordinated by B.H. Baek, Adel Hanna, and Selected Panelists

The CMAS center is working with EPA on developing a new data warehouse and modeling platform concept to support community users. The highlight of the data warehouse is to provide users with modeling data with more download options and also potentially providing modeling platform for users who wish to develop modeling applications for their specific needs. CMAS would like to present the data sharing / modeling support concept to get feedbacks to help develop this valued resource to the air quality community.

Points for presentation / discussion:

1. Data/Model Sharing Warehouse Platforms on Google Drive and Amazon Web Server (AWS) using Dataverse.org Project
2. Modeling Platform Support on AWS and Google Cloud Platform (GCP)
3. Case examples in collaboration with OAQPS and ORD, U.S. EPA
4. Open Forum Discussion with the Community

Special Tutorial Session: An Introduction to Reduced-Complexity Models for Air Quality
Peter Adams (Carnegie Mellon University) and Serena Chung (US-EPA)

Chemical transport models (CTMs) are the gold standard for predicting how changing emissions will influence future air quality. However, they are computationally intensive, meaning that only a limited number of scenarios can be simulated in detail. Furthermore, they require considerable expertise, limiting the communities of researchers that can assess how emissions changes will affect ambient concentrations of air pollutants and human health. Reduced-Complexity Models (RCMs) fill this gap by providing tools that are better suitable for screening and uncertainty analyses and useable by policy and energy systems analysts that lack resources and expertise to use CTMs.

This training will provide an introduction to three RCMs now publicly available for air quality assessments: APEEP, EASIUR, and InMAP. We will summarize the inner workings and assumptions in each of these models. We will demonstrate where to obtain these models, introduce how to use them, including how to adjust assumptions about dose-response and economic valuation. An intercomparison of the results of the models will be summarized.

Quantification of Lightning NOX and its Impact on Air Quality over the Contiguous United States

Daiwen Kang, Rohit Mathur, Limei Ran, George Pouliot, David Wong, Kristen Foley, Wyat Appel, and Shawn Roselle

As one of the largest sources of natural NOX, it is estimated that lightning-induced NOX (LNOX) contributes 10-15% of the total global NOX emissions budget. Lightning activity exhibits strong spatial and temporal variations, and consequently so does the tropospheric distribution of NOX from lightning flashes. To assess the impact of LNOX on air quality, the Community Multiscale Air Quality (CMAQ) modeling system quantifies LNOX based on hourly gridded lightning strikes. The relative impact of LNOX on ambient O3 depends not only on the extent and magnitude of lightning activity, but also on NOX emissions from other sources, such as anthropogenic NOX and soil NO emissions. In this study, simulations with/without LNOX for April - September 2011 are performed using CMAQv5.2. Total column lightning NOX and its relative contributions to total NOX emissions are quantified by region and time of year. The impact of LNOX on air quality is assessed by region and season based on evaluation against gas phase measurements. Vertical profiles will be examined against available ozone-sonde data and data collected from the Deriving Information on Surface Conditions from Column and Vertically Resolved Observations Relevant to Air Quality (DISCOVER_AQ; http://www.nasa.gov) 2011 campaign.

Recent developments of FENGSHA dust emission module: Lessons learned from multi-year real-time dust forecasting over North America

Daniel Tong1,2,3, Barry Baker2,3, Shobha Kondragunta4, Li Pan5, Youhua Tang2,3, Pius Lee3, Rick Saylor3, Jianping Huang5, Jeff McQueen5, Georg Grell6, and Ivanka Stajner7

(1) GMU; (2) UMD; (3) NOAA ARL; (4) NOAA NESDIS/STAR; (5) NOAA NWS/NCEP; (6) NOAA ESRL; (7) NOAA NWS/STI

As drought becomes the new normal in many western states, the frequency of windblown dust storms increases rapidly. Long-term dust observations show that large dust storms have increased by 240% from 1990s to 2000s. In regions frequented by dust storms, there is an 800% increase in the infection rate of Valley Fever, an infectious disease caused by inhaling soil-dwelling fungi. In light of the enormous air quality and health burden imposed by dust storms, it is vital to provide accurate early warnings to mitigate the detrimental effects of dust storm hazards. We present here the latest developments of the FENGSHA (Windblown dust in Mandarin) dust emission module, which has been incorporated into the CMAQ model to support the National Air Quality Forecast Capability (NAQFC) PM2.5 forecasting over North America. Lessons learned to apply FENGSHA to regional dust forecasting are also highlighted. Since its initial implementation into NAQFC in 2015, the FENGSHA module has been updated with several improvements, including 1) new threshold friction velocities from a reanalysis of wind tunnel measurements conducted by Gillette and colleagues; 2) a high resolution global soil texture map; and, 3) emission data assimilation capability to assimilate satellite observations of land surface conditions. Case studies of these updates will be presented through comparison against multi-platform observations, including ground networks (AIRNow and IMPROVE) and satellite remote sensing (MODIS, VIIRS and GOES-R AOD and dust mask). Finally, we will discuss plans of future FENGSHA development and its application to global aerosol forecasting with the NOAA Next-Generation Global Prediction System (NGGPS).

Coupling MOVES mobile emissions modeling with Air Quality Modeling System

Bok.H. Baek, Rizzieri Pedruzzip, and Carlie Coats

One of the key advances in developing the next generation air quality modeling system is in dynamically coupling its three major modeling components: meteorology, emissions, and pollutant chemistry-transport. Recent versions of the Community Multiscale Air Quality (CMAQ) model integrate several meteorologically-driven emission processes "inline" with its chemistry and transport processes to dynamically calculate the emissions within the model using daily meteorology, rather than being estimated a priori and provided as model inputs. Simulating emissions "inline" in CMAQ is crucial for real-time air quality forecasting because it allows the model to include the influences of the most recently forecast meteorological fields on emissions from key sources such as vegetation, fertilizer applications, energy-generating stationary units, and wildfires. While biogenic emissions, bi-directional NH3 from fertilizer applications, and point-source plume rise are calculated inline in CMAQ, other important emissions sources have little or no accounting of meteorological influences in current air quality forecasts. This is especially true of the mobile sector, one of the highest emitters of NOX, PM2.5, and CO, particularly in metropolitan areas; mobile emissions estimates in forecast are based on monthly total inventory and standard weekly/daily temporal profiles. Although current SMOKE modeling system can estimate on-road mobile emissions in "offline" based on the state-of-the-science Motor Vehicle Emission Simulator (MOVES) emission factors, its computational requirements are prohibitive in real-time air quality modeling applications. We have developed the module in SMOKE that can dynamically present the meteorological influences on on-road mobile emissions using the high-order numerical MOVES emission factors (EF) algorithms without any computational bottlenecks for later "inline" application in CMAQ. The MOVES EF algorithms have been developed based on the latest MOVES EF lookup tables from U.S. EPA using the Best Curve-Fit Algorithm (BCFA) numerical method. We will first present the accuracy of emissions estimates as well as the significant computational improvement using the BCFA method to demonstrate the feasibility of "Inline" MOVES module development in air quality modeling system.

Recent improvements to FireWork: Environment and Climate Change Canada's operational air quality forecast system with near-real-time wildfire emissions

Jack Chen, Radenko Pavlovic, Kerry Anderson, Peter Englefield, Hugo Landry, Rodrigo Munoz-Alpizar, Mike Moran, Setigui Keita, and Daniel Thompson

Although intermittent in nature, wildfires can be major contributors to air quality issues in North America. Environment and Climate Change Canada has been running FireWork, a comprehensive operational air quality forecast system with near-real-time (NRT) biomass-burning emissions, since spring 2016. This system is the result of years of development in collaboration with the Canadian Forest Service, whose Canadian Wildland Fire Information System (CWFIS) provides NRT information on wildfire locations and characteristics. The CWFIS uses remote sensing data from AVHYY, MODIS, and VIIRS satellite instruments combined with vegetation and weather data to estimate biomass-burning fuel consumption and other fire activity information. Emissions estimates are then calculated for FireWork using the Fire Emission Production Simulator (FEPS), a component of the U.S. Forest Service's BlueSky Modeling Framework.

More recently the Canadian Forest Service developed the Canadian Forest Fire Emissions Prediction System (CFFEPS) to predict wildfire emissions and smoke plume development through integration with the Canadian Forest Fire Behaviour Prediction System (FBP). CFFEPS can use different treatments of fire growth and different pollutant emissions factors specific to different fire stages, and it estimates wildfire plume rise based on input forecast meteorology. CFFEPS has now been incorporated into FireWork as an alternative to the FEPS module. Tests for several recent fire seasons have shown that FireWork predictions of PM2.5 and other pollutants such as ozone using CFFEPS are more accurate than those from the current operational version of FireWork using FEPS.

Development of the 2016 Collaborative Emissions Modeling Platform

Alison Eyth and Zac Adelman

The 2016 Inventory Collaborative is a partnership between state emissions inventory staff, multi-jurisdictional organizations (MJOs), federal land managers (FLMs), EPA, and others to develop an emissions modeling platform for use in air quality planning. The Collaborative is structured around workgroups organized by emissions inventory sectors, plus a modeling workgroup. The workgroups are developing emission inventories in pursuit of the creation of 2016 base year and future year emissions inventories to be used for regulatory and non-regulatory air quality modeling. The Collaborative is currently working on the future years of 2023 and 2028. A coordination workgroup provides logistical support and facilitation to the sector workgroups as they move toward the goal of well-documented model-ready emissions for use in air quality planning. Three versions of the 2016 platform are planned: the alpha version included a draft version of year 2016 emissions and was released in spring of 2018; the beta version will include future years along with an updated year 2016 emissions and is planned for fall of 2018, and Version 1.0 is planned for winter-spring of 2019.

Exploring the Vertical Distribution of Wildland Fire Smoke in CMAQ

Joseph L. Wilkins, George Pouliot, Thomas Pierce, 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 (PM_2.5) 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 smoldering and flaming fires, identifying agricultural vs prescribed fires, and adjusting the diurnal profile of smoke emissions.

KORUS Ver. 2.0 : An Emissions Inventory in Support of the KORUS-AQ Field Campaign
Younha Kim1, Jung-Hun Woo1*, Bok Haeng Baek2, Jinseok Kim1, Jinsu Kim1, Youjung Jang1, Minwoo Park1, Yungyeong Choi1, Eunji Lee1, Hyunjin Park1

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

Air quality over the Northeast Asia region has been deteriorated despite of more stringent air pollution control policies by the governments. The needs of more scientific understanding of inter-relationship among emissions, transport, chemistry over the region are much higher to effectively protect public health and ecosystems. Two aircraft filed campaigns targeting year 2016, MAPS-Seoul and KORUS-AQ, have been organized to study the air quality of over Korea and East Asia relating to chemical evolution, emission inventories, trans-boundary contribution, and satellite application. We developed a new East-Asia emissions inventory, named KORUS Ver. 1.0, based on NIER/KU-CREATE (Comprehensive Regional Emissions inventory for Atmospheric Transport Experiment), in support of the filed campaigns. For anthropogenic emissions, it has 54 fuel classes, 201 sub-sectors and 13 pollutants, including CO2, SO2, NOx, CO, NMVOC, NH3, PM10, and PM2.5. Since the KORUS emissions framework was developed using the integrated climate and air quality assessment modeling framework (i.e. GAINS) and is fully connected with the comprehensive emission processing/modeling systems (i.e. SMOKE, KU-EPS, and MEGAN), it can be effectively used to support atmospheric field campaigns for science and policy. During the field campaigns, we are providing modeling emissions inventory to participating air quality models, such as CMAQ, WRF-Chem, CAMx, GEOS-Chem, MOZART, for forecasting and post-analysis modes. Based on initial assessment of those results, we are improving our emissions, such as VOC speciation, biogenic VOCs modeling to generate Version 2.0 inventory. From the 2nd iteration between emissions and modeling/measurement, further analysis results will be presented at the conference.

KEYWORDS
KORUS-AQ, Emissions, Inventory, Air Pollution, Korea

ACKNOWLEDGEMENT
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-2018-01-02-027). This work is supported by Korea Ministry of Environment(MOE) as Graduate School specialized in Climate Change .

Coupling Meteorology-sensitive Emissions with Air Quality Forecasting System Enhancement

B.H. Baek, SoonTae Kim, and Changhan Bae

The National Institute of Environment Research (NIER) in Korea has developed its operational National Air Quality Forecasting System (NAQFS) using the Community Multiscale Air Quality (CMAQ) model developed by the U.S. Environmental Protection Agency (EPA) to forecast concentrations of ozone and particulate matter below 2.5 m (termed PM2.5) over the contiguous East Asia region. In this application, the CMAQ model can use "inline" emissions processing, wherein emissions from meteorologically driven air pollutant emission processes are calculated dynamically within the model, rather than estimated a priori and provided as model inputs. Simulating emissions inline is crucial especially for real-time air quality forecasting because it allows the model to include the influences of the most recently forecast meteorological fields on emissions from key sources such as power plants, vegetation, fertilizer applications, and wildfires. While biogenic emissions, bi-directional NH3 from fertilizer applications, and point-source plume rise are calculated inline in CMAQ, other important emissions sectors have little or no accounting of meteorological influences in current air quality forecasts. The goal of this study is to improve the predictive ability of the NAQFS by enhancing the temporal representation of key air pollutant emissions sensitive to local meteorological conditions in the CMAQ. We have identified three major emissions inventory sources (i.e., onroad mobile, residential wood combustion, and agricultural practices) that are sensitive to local metrological conditions, and developed the statistical algorithms as a function of ambient temperature. We will share the outcome from CMAQ using a new pre-processor called "Meteorological Adjustment (MetADJ)" tool that can adjust CMAQ-ready gridded/hourly emissions with forecasted local meteorological conditions.

The predicted impact of VOC emissions from Cannabis spp. cultivation facilities on ozone concentrations in Denver, CO.

Enabling Sensitivity Analysis in CMAQ v.5.2.1 via Implementation of the Complex Variable Method

Isaiah Sauvageau, Bryan Berman, Shannon Capps

Sensitivity calculations with atmospheric models quantify the relationships between emissions and pollutant concentrations. When choosing among methods to calculate sensitivities, one must consider the accuracy, computational cost, and ease of implementation of the method. The finite difference method and decoupled direct method (DDM) of calculating sensitivities are available for the Community Multiscale Air Quality (CMAQ) model. By implementing the complex variable method, the accuracy of sensitivities would improve compared to the finite difference method by eliminating implicit truncation errors and subtractive cancellation errors. The ease of implementation of the complex variable method exceeds that of DDM. The increase in computational cost is reasonable. A tool that automates the implementation of the complex variable method has been used here. These tools redefine real variables as complex ones and overloads operators to handle complex arguments. The simplicity and accuracy of the complex variable method will be demonstrated here in CMAQ v.5.2.1 as an attractive option for calculating sensitivities.

An Uncertainty for Clean Air: Air Quality Modeling Implications of Underestimating VOC Emissions in Urban Inventories

Shupeng Zhu, Michael Mac Kinnon, Brendan P. Shaffer, G.S. Samuelsen, Jacob Brouwer, Donald Dabdub

Recent literature has shown that volatile organic compound (VOC) emission inventories for urban regions may be significantly underestimated. In particular, non-transportation sources including volatile chemical products (VCP) are increasing in relative importance due to both the current and historical focus on controlling transportation emissions. These findings have major implications for photochemical air quality modeling used to determine appropriate and effective regulatory controls to meet regulatory limits for primary and secondary pollutants, including fine particulate matter (PM2.5) and ozone. Using a regional air quality model, we quantify the changes in ozone and PM2.5 simulated for enhanced VOC emissions reported in a recent study by McDonald et al. relative to a baseline inventory for California. Results show that in summer, simulated maximum 8-hr ozone concentrations could increase by 17.8 ppb. In winter, simulated maximum 8-hr ozone concentrations could increase by 15 ppb, and 24-hr maximum PM2.5 increases by 7.8 μg/m3. Impacts reflect differences in the spatial location of VCP source emissions relative to those for transportation including increases in pesticides, consumer products, and architectural coatings. Compared to measurement data, model performance is not significantly improved by the adjustment of VOC emissions. Overall, augmented VOC emission inventories impact simulated concentrations of pollutants but may not affect the performance of models used for the design of emission control policy.

Demonstrating Sensitivity Analysis with the Multi-Complex Variable Method in ISOYYOPIA

Bryan Berman, Isaiah Sauvageau, Shannon Capps, Ryan Russell

Sensitivity analysis using atmospheric chemical transport models provides a deeper understanding of how specific emissions affect pollutant concentrations. Given a model with emissions as inputs and pollutant concentrations as outputs, this analysis is achieved by computing the partial derivatives of the underlying functions with respect to their input variables. Implementing higher-order sensitivity calculations can be quite difficult even though they are important to understanding nonlinear processes. A novel approach to sensitivity analysis leverages multi-complex variables to improve accuracy, computational cost, and ease of implementation over the finite difference method and decoupled direct method (DDM).

Here, the multi-complex variable method (MCVM) is first utilized in the inorganic aerosol thermodynamic equilibrium model, ISOYYOPIA, to serve as proof of concept before implementation in the Community Multiscale Air Quality (CMAQ) model. Implementing MCVM into ISOYYOPIA is strategic not only because ISOYYOPIA is used in CMAQ, but also because there are enough inputs and outputs to demonstrate a main advantage of the method, which is simultaneous calculation of the sensitivities of all outputs with respect multiple inputs. Furthermore, MCVM enables calculation of higher order derivatives, which is useful when the functions are nonlinear as it reduces error when using sensitivities to predict changes in model output with increments in the input parameters. Finally, the availability of analytically calculated sensitivities from DDM and higher-order DDM with ISOYYOPIA makes testing of this novel method feasible. Therefore, this project will demonstrate the multi-complex variable method in ISOYYOPIA as well as discuss its utility.

Development of 2016 Hemispheric Emissions for CMAQ

Jeffrey M. Vukovich (USEPA), Alison Eyth (USEPA), Barron Henderson (USEPA), Chris Allen and James Beidler

Emissions for a year 2016 case on a northern hemispheric grid was prepared for CMAQ. The emissions included Criteria Air Pollutants (CAPs) only, a year-long period spin-up and were processed through an updated version of SMOKE. The input emissions for the United States, Canada, and Mexico were based on EPA's 2016 alpha version platform emissions. The United States emissions rely on data from the recently release 2014 National Emissions Inventory, Version 2, updated to year 2016 for point, onroad, nonroad, oil and gas, fire and biogenic sectors. Fire emissions for the modeling period outside the USA were derived from the FIre INventory from NCAR (FINN) data set. Anthropogenic emissions for areas outside of North America were taken from the the 2010 Hemispheric Transport of Air Pollution (HTAP) version 2 dataset and projected to the year 2014. For China, a year 2015 dataset was used. Double counting of emissions, including those from commercial marine vessels, between the emission inventory sources was prevented. Spatial surrogates were developed for the 108km resolution hemispheric grid. New features in SMOKE were used including the ability to zero out and apply factors to emissions by country. The computation of hourly emissions took into account time zones around the world.

Simulation of the Meteorological Conditions by Introducing the Urban Physics within the PX-LSM and ACM2-PBL Scheme in the WRF Model over the PRD Region in Southern China

Utkarsh Prakash BHAUTMAGE^a; Chun Yin DY^b; Jimmy Chi-Hung FUNG^ab Jonathan PLEIM^c; Alexis LAU^a; Adel HANNA^d

^aDivision of Environment and Sustainability, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong

^bDepartment of Mathematics, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong

^cEnvironmental Protection Agency, Research Triangle Park, North Carolina, USA

^dInstitute for the Environment, The University of North Carolina at Chapel Hill, USA

Contact Person: upbhautmage@connect.ust.hk

Ph: +852-59379122

Better numerical simulation and forecasting of the regional meteorological conditions over the urban areas is always associated with the proper modeling of the planetary boundary layer's (PBL) true structure and its periodic evolution. There are various PBL schemes developed so far and the same have been integrated in the Weather Research and Forecasting (WRF) model to incorporate the governing turbulence within the PBL. Also, these PBL schemes are tied to the different land surface models (LSMs) to account the heat and moisture fluxes evolved from the earth's surface. The urban models such as UCM (Urban Canopy Model), BEP (Building Effect Parameterization) and BEM (Building Energy Model) have been developed and included within the WRF model in the past few years. The motive behind this was to enhance the representation of the urban surface phenomenon such as the urban heat island (UHI) effect and the turbulence caused due to the complex-built urban structures these days e.g. street canyon. It has improved the PBL structure creation over the cities worldwide. However, these models are associated with some challenges in terms of specifying the detailed urban morphological parameters and initialization of the detailed spatial distribution of state variables. Moreover, the widely used BEP-BEM model is computationally expensive to run it at its full advantage to obtain the real-time forecasts. The PRD region spanning over 39,380 km² area in the Southern China with a population around 120 million is expected to become a mega city in the coming future days. Recently, the performance of the four PBL schemes, namely MYJ, Boulac, YSU and ACM2 have been analyzed over the PRD region. From the results, it was concluded that the non-local (Asymmetric Convective Model V2) ACM2 scheme has shown good agreement with the observation of the meteorological field variables. In this research, an attempt has been made to include the urban morphology and the governing physics, mainly consisting the momentum drag within the ACM2 PBL scheme and thermal physics within the PX-LSM (land surface model). It is expected that performing simulations with the newly developed model over the cities having dense urban structures along-with many high-rise buildings for e.g. Hong Kong city, will forecast the meteorological field variables such as 10m-windspeed and 2m-temperature and their trends reasonably well and within the less computational time.

Contributions of condensable particulate matter to atmospheric organic aerosol over Japan

Yu Morino,1 Satoru Chatani,1 Kiyoshi Tanabe,1 Yuji Fujitani,1 Tazuko Morikawa,2 Katsuyuki Takahashi,3 Kei Sato,1 and Seiji Sugata1

1 National Institute for Environmental Studies, 16-2, Onogawa, Tsukuba, Ibaraki, 305-8506, Japan

2 Japan Automobile Research Institute, 2530 Karima, Tsukuba, Ibaraki 305-0822 Japan

3 Japan Environmental Sanitation Center, 10-6 Yotsuyakami-Cho, Kawasaki, Kanagawa, 210-0828, Japan

Because emission rates of particulate matter (PM) from stationary combustion sources have been measured without dilution or cooling in Japan, condensable PM has not been included in Japanese emission inventories. In this study, we modified an emission inventory to include condensable PM from stationary combustion sources based on the recent emission surveys using a dilution method. As a result, emission rates of organic aerosol (OA) increased by a factor of seven over Japan. Stationary combustion sources in the industrial and energy sectors became the largest contributors to OA emissions over Japan in the revised estimates (filterable-plus-condensable PM), while road transport and biomass burning were the dominant OA sources in the previous estimate (filterable PM). These results indicate that condensable PM from large combustion sources makes critical contributions to total PM2.5 emissions. Simulated contributions of condensable PM from combustion sources to atmospheric OA drastically increased around urban and industrial areas, including the Kanto region, where OA concentrations increased by factors of 2.5 - 6.1. Consideration of condensable PM from stationary combustion sources improved model estimates of OA in winter but caused overestimation of OA concentrations in summer. Contributions of primary and secondary OA should be further evaluated by comparing with organic tracer measurements.

National Air Quality Forecast Capability prediction updates: bias-correction for ozone

Ivanka Stajner¹, Jeff McQueen², Pius Lee³, Jianping Huang^{2, 5}, Ho-Chun Huang^{2, 5}, Li Pan^{3, 6}, Youhua Tang^3,6, Daniel Tong^3,6, Ariel Stein³, James Wilczak⁴, Irina Djalalova^4,8, Dave Allured^4,8, Phil Dickerson⁷, 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) Syneren Technologies

This presentation will provide an overview of recent updates to NOAA's operational air quality predictions. 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 http://airquality.weather.gov. 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). Predictions of PM2.5 include intermittent contributions from wildfire and dust sources. Anthropogenic emission upgrades included recent monitoring data for point sources and oil-and-gas industrial activities. In addition, smoke and dust predictions are separately produced by NOAA's HYSPLIT model.

The most significant recent update is the addition of a post-processing scheme for ozone predictions. The approach we are taking is unified with that used for bias-correction of PM2.5 predictions, except that it additionally relies on NOx, NOy and ozone as parameters to identify analogs. Updates to the PM 2.5 bias correction system include use of consistent training model predictions for the unified Kalman Filter Analog (KFAN) method, an increased number of observation sites for PM2.5 bias correction to an average of over 900 monitors. The scheme is modified for the prediction of extreme events for ozone and PM2.5 by adding the difference between the current raw model forecast and historical analogs mean to the KFAN bias-corrected predictions. Another area of recent focus has been modification of the way wildfires are identified and their emissions included in NOAA's air quality predictions. We will present impacts of these recent updates and discuss future plans.

Estimating smoke emission using GOES satellite observations and HYSPLIT model

Tianfeng Chai ^1,2,3, Hyuncheol Kim^1,2,3 and Ariel Stein¹

1: NOAA/OAR/Air Resources Laboratory (ARL)

2: Cooperative Institute for Climate & Satellites-Maryland (CICS-MD)

3: University of Maryland, College Park, MD

An inverse modeling technique has been developed to estimate wildfire smoke emissions using NOAA's HYSPLIT model and GOES Aerosol/Smoke products (GASP). In this top-down approach, a cost function is defined to mainly quantify the differences between model predictions and satellite measurements of column integrated air concentrations, weighted by the model and observation uncertainties. Minimizing this cost function by adjusting smoke sources provides wildfire smoke emissions that agree well with the satellite observations. The HYSPLIT-based Emissions Inverse Modeling System for wildfire (HEIMS-fire) aims to resolve the smoke source strength as a function of time and vertical levels. A wildfire event that took place in the Southeast US during November 2016 has been used as an example to test the HEIMS-fire system. Hindcasts with the emission estimates by the HEIMS-fire system have been performed. Comparison between the new emission estimation system and the current operational BlueSky emission prediction has been conducted as well. In addition, the effect of introducing modeling uncertainty terms in the inverse modeling will be discussed.

Development and evaluation of offline coupling of FV3-based GFS with CMAQ at NOA
Jianping Huang (1,2), Jeff McQueen (2), Li Pan (1,2), Perry Shafran (1,2), Ho-Chun Huang(1,2), Jack Kain(2), Pius Lee (3), Youhua Tang (3,4), Ivanka Stajner (2,5), and Jose Tirado-Delgado (5,6)

1) I.M. Systems Group Inc., Rockville, MD
2) National Oceanic and Atmospheric Administration (NOAA), National Centers for Environmental Prediction (NCEP), College Park, MD
3) NOAA Air Resources Laboratory, Silver Spring, MD
4) University of Maryland, College Park, MD
5) NOAA, Office of Science and Technology Integration, Silver Spring, MD
6) Syneren Technologies Corporation, Arlington, VA 22201

The Geophysical Fluid Dynamics Laboratory (GFDL) Finite Volume Cubed-Sphere (FV3) was selected as the dynamical core for the Next Generation Global Prediction System (NGGPS) at the NOAA National Weather Services (NWS). NOAA/NWS National Centers for Environmental Predictions (NCEP), Environmental Modeling Center (EMC) has been running FV3GFS, which is the upgraded Global Forecasting System (GFS) with the FV3 core and GFS physics package at 13-km horizontal grid spacing and 64 hybrid vertical levels and providing experimental weather predictions since September 2017. In this study, the FV3GFS is coupled with the Community Multiscale Air Quality (CMAQ) modeling system in an off-line mode. The interface coupler for meteorology and air quality forecasting models is revisited for mapping the FV3GFS outputs from the longitude and latitude grids and hybrid vertical levels of FV3GFS to the C-grid and sigma levels of CMAQ. FV3GFS/CMAQ predictions of surface ozone (O3) and fine particulate matter with diameter less than 2.5 μm (PM2.5) are evaluated with the USEPA AirNow observational data for one winter and one summer months. Results are further compared to the NOAA operational National Air Quality Forecast Capability (i.e., CMAQ driven by the Non-hydrostatic Multi-scale Model on the Arakawa staggered B-grid (NMMB) off-line) predictions. Evaluations of the off-line coupled system will provide a benchmark for evaluating the in-line coupled system under development at NOAA.

HERMESv3: a stand-alone multiscale atmospheric emission model

Marc Guevara Vilardell, Carles Tena Medina, Manuel Porquet, Oriol Jorba Casellas, Carlos Perez Garcia-Pando

The High-Elective Resolution Modelling Emission System version 3 (HERMESv3) is an open source, parallel and stand-alone multiscale atmopsheric emission model that processes and estimates emissions for global, regional and urban air quality modelling.

The model, which is coded in Python, consists on two main modules that can be combined or executed separately: (i) the global_regional module and (ii) the bottom_up module.

The global_regional module is a highly customizable emission processing system that calculates emissions from different sources, regions and pollutants on a user-specified model grid. This module applies an automatic combination of a set of already existing emission inventories, which are individually processed using detailed vertical, temporal and speciation profiles as well as scaling and masking factors defined by the user. The generated emission fields can be used as inputs for different chemical transport models (i.e. CMAQ, WRF-Chem, NMMB-MONARCH) at mutiple spatial (up to 1km2) and temporal (up to 1h) resolutions.

The bottom_up module is an emission model that estimates atmospheric emissions at the source (e.g. road link industrial facility, crop type) and hourly level combining state-of-the-art methods with local activity and emission factors as well as meteorological parameters (e.g. temperature and wind speed). This model covers the estimation of bottom-up emissions from road transport, point sources, residential combustion and agriculture. The road transport emission outputs are also adapted for their application with the R-LINE urban dispersion model.

Status and Plans for the Next Generation Air Quality Modeling System Development

Jonathan E. Pleim, David Wong, Robert C. Gilliam, Jerold A. Herwehe, O. Russell Bullock Jr., George Pouliot, Christian Hogrefe, Daiwen Kang, Bill Hutzell, Rohit Mathur, Shawn Roselle, and Limei Ran

A next generation air quality modeling system is being developed at the U.S. EPA to enable modeling of air quality from global to regional to local scales. The system will have three configurations: 1. Global meteorology with seamless mesh refinement and online atmospheric chemistry; 2. Regional (limited area) online meteorology and chemistry; 3. Offline (sequential) regional meteorology and chemistry. A one-dimensional air quality (AQ) component, built from state of the science chemistry and aerosol modules from the Community Multiscale Air Quality (CMAQ) model will be used in all three configurations. For the Global online configuration, the AQ component will be coupled to the Model for Prediction Across Scales - Atmosphere (MPAS-A), which is a global meteorological model with seamless mesh refinement developed at the National Center for Atmospheric Research (NCAR). The regional configurations will be coupled with WRF or a regional version of MPAS that has recently been developed at NCAR.

In the presentation we will describe the coupled MPAS-CMAQ, which is an operational prototype of the Next Gen Model. We will describe the addition of physics schemes to MPAS that are particularly designed for air quality applications, as well as the addition of four dimensional data assimilation (FDDA). We will also show the latest tests of the MPAS-CMAQ model where we have incorporated CMAQ modules for atmospheric chemistry, deposition, cloud processes, and vertical diffusion into MPAS. In initial one month simulations (June 2013) MPAS-CMAQ shows good skill in simulating ozone in the populated areas of the world but tends to underpredicted ozone in high elevation areas. Longer simulations for the full year of 2016 show that Including ozone from the GFS analysis for all layers above the tropopause helps to reduce this underprediction.

Mixing of Mineral Dust and Biomass Burning Aerosol over East Asia

Xinyi Dong^a, Joshua S. Fu^a, Kan Huang^a,b

^a Department of Civil and Environmental Engineering, The University of Tennessee, Knoxville, Tennessee, 37996, USA

^b Department of Environmental Engineering, Fudan University, Shanghai, 200438, China

Natural wind-blown dust and biomass burning aerosol are investigated in this study. Dust from Taklamkan and Gobi deserts in China and Mongolia transported southeastward towards the West Pacific in spring. Meanwhile, biomass burning plume from Southeast Asia moved eastward towards the West Pacific in the same time. The two different sources of aerosol mixed with each other over East Asia at Taiwan. The mixing cases occurred during 2006-2010 are investigated with both satellite observations and model. The twoway model WRF/CMAQ modeling system is applied to evaluate the impact of dust and biomass burning on radiative forcing change. We evaluated DRE of biomass burning and dust over East Asia under both clear and all sky conditions. Using a regional coupled climate and chemical transport model WRF/CMAQ, we found that biomass burning has negative DRE of -20W/m2 at surface layer (SFC) at source region over Indochina under both clear and all sky conditions, but has significant positive DRE of 10w/m2 at top of atmosphere (TOA) at downwind area over South China under all sky conditions. The perturbation of upwelling short wave radiation (SWR) at TOA introduced by biomass burning aerosol was significantly greater under all sky than under clear sky condition, indicating that biomass burning aerosols, if they exist, are always on top of the cloud over South China and absorbing more SWR than scattering back. On the other hand, Dust aerosol increases the upwelling SWR at TOA by +15W/m2 (10%) under clear sky condition and +5W/m2 (1%) under all sky condition. The coexistence of biomass burning and dust alters the aerosol radiative effect efficiency by ~10%. This study is the first investigation into the coexistence of biomass burning and dust over East Asia with the WRF/CMAQ modeling system.

Impacts of Changing Land Use and Land Cover on Regional Climate in Sub-Saharan Africa

Timothy Glotfelty¹, Diana Ramirez², Adrian Ghilardⁱ², Jared Bowden³, Robert Bailis⁴, and J. Jason West¹

¹Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, 135 Dauer Drive, Chapel Hill, NC 27516, USA

²Centre for Research in Environmental Geography, Universidad Nacional Autonoma de Mexico, Mexico City, Mexico

³Department of Applied Ecology, North Carolina State University, 100 Brooks Ave, Raleigh, NC 27695, USA

⁴Stockholm Environment Institute - US Center, 11 Curtis Ave, Somerville, MA, 02144

Sub-Saharan Africa is a dynamic system in which human activity, ecosystems, and the atmosphere are tightly coupled. Changes in land use and land cover (LULC) driven by human activities are expected to continue across the continent over the coming decades. Rates of change vary depending on local biophysical conditions, demand for forest products like timber and woodfuels, as well as expansion of crop cultivation and grazing land, driven by growing domestic, regional, and global markets. Whereas past changes and future pathways of LULC have been the subject of multiple studies, few have focused on understanding land-atmosphere interactions. The tight coupling of land surfaces to the atmosphere in Sub-Saharan Africa may result in regional changes in climate that could be enhanced or negated by global climate change, and ultimately affect populations and ecosystems in climate-sensitive regions. Some evidence suggests that severe droughts in this region, which threaten livelihoods and are associated with increased conflict, may be linked to LULC changes, and thus understanding the impact of LULC changes on the regional climate is of great importance to local policy makers.

In this work the Weather Research and Forecasting model (WRF) is used to explore the impact that recent changes in LULC have had on the current climate of Sub-Saharan Africa. Two WRF simulations are forced by yearly-varying global reanalysis fields - one using LULC specific for each year and one using static LULC representative of the year 2001 to see the influence of LULC changes. LULC changes are processed/simulated using the Dinamica EGO framework. Within Dinamica EGO, observed LULC changes from the Moderate Resolution Spectroradiometer (MODIS) land cover MCD12Q1 product between 2001 and 2007 are compared by means of conditional probabilities (i.e. weights of evidence) against various spatial proximal explanatory variables such as population density, distance to the road network, elevation, among others. The resulting map depicts the probability that a grid cell within the domain will shift from one LULC class to another. By integrating the amount of change from the calibration period with the probability map, new simulated LULC maps are produced, and compared against observed maps for validation purposes.

Preliminary results indicate that changes in LULC between the years 2001 and 2013 can significantly impact the atmosphere's energy balance and hydrologic cycle. Additionally, utilizing LULC representative of 2013 improves the overall WRF model performance. Here we extend on these preliminary simulations to understand the climate response to evolving LULC from 2001-2017 by comparing the traditionally static LULC applied in WRF, MODIS, to the evolving LULC using Dinamica EGO

The role of Weather conditions conducive to severe haze and regional transport of air pollutants of three heavy polluted episodes in Henan Province, China during 2015-2016

Haijiang Kong

Yang Zhang

Zhenhua Dong

Han Li

The frequency of Henan winter severe haze episodes has increased substantially over the past decades, and it is commonly attributed to increased pollutant emissions from China's rapid economic development. Henan Province is one of the most polluted areas in the Beijing-Tianjin-Hebei and its outskirts. Due to the prevailing surface wind and the Taihang Mountains terrain in the west part of Beijing-Tianjin-Hebei, a pollutant transport passage has been formed. By using the WRF-Chem simulation outputs and ERA-Interim reanalysis data, the pollutant transport characteristics of three episodes was investigated in Henan Province during 2015-2016. The results show that weather conditions are conducive to the outbreaks of heavy pollution episodes. In particular, stagnation (weak wind, drier above 700hPa while more moisture in the boundary layer, inversion, weak wind vertical shear, etc.) often occurs over Henan. Stagnation Weather Index (SWI) has been used for forecasting the weather conditions in Henan, Numerical simulation results show that contributions from neighboring provinces varied because of accumulation of different pollutions, wind direction, and pollution episodes. The regional transport characteristics of the pollutants have also been analyzed.

Air Pollution and Health Co-Benefits of the Paris Agreement on Climate Change

Kathleen M. Mulvaney, Steven J. Smith, Haewon C. McJeon, J. Jason West

Under the Paris Agreement on climate change, member countries submit Nationally Determined Contributions (NDCs) outlining plans to mitigate greenhouse gas emissions (GHGs). Such efforts to mitigate GHGs will reduce consumption of fossil fuels and can achieve significant reductions in other air pollutants that impact human health, such as ozone and particulate matter smaller than 2.5 μm (PM2.5). Air pollution reductions and improved health outcomes resulting from policies that target GHG mitigation are considered co-benefits. In this study, we use an interdisciplinary set of modeling tools to estimate the air pollution and health co-benefits of the Paris Agreement NDCs. We project 2015 emissions (sulfur dioxide, nitrogen oxides, carbon monoxide, organic carbon, black carbon, ammonia, and non-methane volatile organic compounds) from the Community Emissions Data System (CEDS) to future years using the Global Change Assessment Model (GCAM). GCAM is an integrated assessment model that simulates economic activity, energy, land use, and emissions by world region and sector. We use results from GCAM scenarios that simulate GHG reduction pathways based on the submitted Paris Agreement NDCs. The NDCs specify national GHG reductions in 2025 or 2030. We therefore analyze the air pollutant and health co-benefits of the NDCs in 2030, relative to a reference scenario, and in scenarios that extend mitigation to 2100. We use projected precursor emissions and present-day MEYYA2 meteorology as input to CAM-Chem, the atmospheric chemistry component of the Community Earth System Model, which simulates global ozone and PM2.5 concentrations in 2030, 2050, and 2100 under the policy scenarios. Here we present our projected precursor emissions results as well as ozone and PM2.5 concentrations from preliminary CAM-Chem simulations. We also discuss our methodology for calculating health impacts using the CAM-Chem simulation results.

The air quality and health impacts of projected long-haul trucks and rail freight transportation in the United States in 2050

Shuai Pan, Anirban Roy, Yunsoo Choi, H. Oliver Gao

Diesel emissions from freight are a key threat to public health. By considering fleet turnover, climate policy, and technology evolvement, this study linked the projected freight emissions in 2050 to air quality and public health over the continental United States. Using a WRF-SMOKE-CMAQ-BenMAP framework, we quantified the impacts of diesel fine particulate matter (PM2.5) emissions change on air quality, health, and economic benefits. With a projected business-as-usual socioeconomic growth and fleet turnover, simulated PM2.5 concentrations have widespread reductions, between 1-1.5 g m-3. Health impact results indicate reduction in emissions would prevent premature deaths of 10.3 (95% CI: 6.9 - 13.6) thousands, causing economic benefits of $107 (95% CI: $10 - $291) billion dollars. The carbon pricing climate policy can obtain ~6% more health benefits nationally, however, it also affects the air quality and hence health impacts regionally due to transition of demand from truck to rail. Further technology improvements to eliminate high emitting conditions in the truck fleet provide substantial additional benefits. These results prove that a combination of continuous adoption of stringent emission standards and strong improvements in technology and fuels are required to meet the sustainable freight and community heath goals. States with high population density, such as California and Texas, could take necessary actions to achieve air quality and health benefits earlier.

Development of Python-based Mobile Emissions Inventory Model

B. H. Baek, Rizzieri Pedruzzi, and Jung-Hun Woo

Air pollution is a severe problem in major urban areas due to increasing numbers of vehicles, reduced road capacity and few investments in public transportation, which make the populations vulnerable to air pollution-induced health risks. Due to the increase in the number of vehicles over these metropolitan areas, there is the increase in traffic and vehicle operating time which brings the increase of pollutants emissions over these urban areas.

For estimating the emissions, there are, in general, the top-down and bottom-up methodologies. The top-down approach uses the aggregated activity data, average emissions factors and national/regional combined data and then the emissions are split into the area of interest. The bottom-up methodology relies on the use of specific activity data and emissions factors and it focuses more on link-level, which means higher resolution. Both approaches are interesting to estimate the emissions, but the selections of best methodologies rely on the goal of the study and mainly on the available data Developing the inventory using the bottom-up method is quite demanding because it requires to incorporate various types of activity data and emissions factors to build local/regional emissions inventory. Currently, there are some tools which do that, like MOVES (Motor Vehicle Emissions Simulator) developed by U.S. Environmental Protection Agency, and the European model COPERT (COputer Programmed to calculate Emissions from Road Transport). All these models can estimate high quality mobile emissions. However, it has been challenging for other countries to implement them due to due to the lack of input datasets required to run these models as well as the usage of inappropriate vehicle-specific emission factors. It requires a great deal of expertise to modify and/or update the existing emission factors with their own.

This study aims to develop a Python-based mobile emissions inventory model to not only build a fast on-road and non-road mobile emissions inventory model that can easily utilize the simple CSV-formatted input data files, such as local-based traffic related datasets like vehicle specific emissions factors, and vehicle activities including hourly regional averaged and/or link-level activities, but also can be used to reflect the changes in policies and control strategies in regional and national scales.

EPAs SPECIATE 4.5 Database: Development and Uses

Frank Divita, Jonathan Dorn and Ying Hsu, Abt Associates, Durham, NC

Marc Menetrez and Madeleine Strum, U.S. Environmental Protection Agency, OAQPS, RTP, NC

SPECIATE is the U.S. Environmental Protection Agency's (EPA) repository of volatile organic gas and particulate matter (PM) speciation profiles of air pollution sources. Some of the many uses of these source profiles include: (1) creating speciated emissions inventories for regional haze, PM, greenhouse gas (GHG), and photochemical air quality modeling; (2) estimating hazardous and toxic air pollutant emissions from PM and organic gas primary emissions; (3) providing input to chemical mass balance (CMB) receptor models; and, (4) verifying profiles derived from ambient measurements by multivariate receptor models (e.g., factor analysis and positive matrix factorization).

Abt Associates, Inc. developed SPECIATE 4.5 through a collaboration involving EPA's Office of Research and Development (ORD) and Office of Air Quality Planning and Standards (OAQPS) in Research Triangle Park, NC, and Office of Transportation and Air Quality (OTAQ) in Ann Arbor, MI. EPA released SPECIATE 4.5 in September 2016 and, in total, the SPECIATE 4.5 database includes 6,206 PM, volatile organic compound (VOC), total organic gases (TOG), and Other Gases profiles. New to SPECIATE 4.5 are the incorporation of individual and composite volatile organic gas and PM profiles from the oil and natural gas sector, motor vehicle exhaust, biomass combustion, waste incineration, and tire and break wear emissions. This paper describes the SPECIATE 4.5 database and upcoming SPECIATE 5.0 database, provides specific examples of its use, and makes recommendations for future improvements.

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

Jonathan Dorn 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, and N₂O) 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.

In February 2018, EPA released eGRID2016. This paper will: 1) discuss the procedures for developing eGRID and the recent improvements found in eGRID2016; 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.

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

Jonathan Dorn 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.

Emission Modeling over India: Developing CMAQ Model-Ready Emissions using the HTAP Inventory and Intercomparison with the ECLIPSE-Based Emissions

Syuichi Itahashi (North Carolina State University, USA; Central Research Institute of Electric Power Industry, Japan)

Prakash Doraiswamy (RTI International, USA)*

Yang Zhang (North Carolina State University, USA)

Alexandra Karambelas (Columbia University, USA)

*corresponding author

Air pollution in India has gained global attention due to its poor air quality, ranked among the regions with the worst air quality in the world. Evaluation of India's air quality is limited by a paucity of surface observations and low temporal resolution of satellite observations, and three-dimensional air quality modeling is an essential component of an air quality management framework to tackle the ambient air pollution problem. Several global inventories exist that can be used as inputs for air quality modeling over India. Recently, CMAQ simulation using the Evaluating the Climate and Air Quality Impacts of Short-Lived Pollutants (ECLIPSE) version 5a emission inventory over India suggested the overestimation over urban areas and underestimation over rural areas from NO₂ emissions for each season of 2010, implying high emission uncertainty for India. It is important to assess the other global inventories to understand such uncertainties. The objective of this work is two-fold: a) to develop CMAQ-ready emissions using the Task Force Hemispheric Transport of Air Pollution (HTAP) inventory; and b) to compare the HTAP-based CMAQ-ready emissions with those based on ECLIPSE. To achieve our goals, we first develop the CMAQ-ready emissions based on HTAP and then analyze the differences of the two sets of model-ready emissions and associated ECLIPSE and HTAP emission inventories. We will identify the differences in the two emission inventories and attempt to understand the causes for such differences. The resulting model-ready emissions based on HTAP will be used as inputs for CMAQ to simulate the air quality over India in future work. 

BVOC flux from a tropical montane ecosystem estimated using the BIGA biogenic emission model

Jade Alexandra Li Ramirez¹

Jeannette del Carmen Zambrano Nejera¹

Carlos Mario Gonzales Duque¹

Alex Guenther²

Beatriz Helena Aristizabal¹

¹Hydraulic Engineering and Environmental Research Group, Universidad Nacional de Colombia, Sede Manizales

²University of California Irvine (UCI)

A model to estimate emissions of biogenic volatile organic compounds (BVOC), called BIGA (by its initials in Spanish, estima las emisiones BIogenicas teniendo en cuenta el Gradiente Altitudinal), was developed. The first version was created using Matlab and ArcGis and it has been updated to Python and QGis in order to provide an open source version. The model was applied to estimate the emissions of BVOC in a high montane tropical ecosystem with different thermal floors and different altitudinal gradients (500 to 5200 m.a.s.l), a region characterized by many distinct climactic zones that range from dry forest valleys to tropical Andean forests, cloud forest, and paramo shrubland. The basic algorithm proposed by Guenther et al. (1993) was used for the model formulation. Input information to the model includes meteorological surface observations (temperature and solar radiation), a digital elevation model (DEM), a land use and a cover map. The BVOC emissions can be calculated at any spatial resolution (this study used 90 x 90 meter). Furthermore, the model is able to estimate each minute, hourly, monthly or annual emissions depending on input information, delivering a BVOC emission inventory disaggregated in space and time in ASCII format.

In this study, a representative day for two typical seasons in tropical climates: dry (January 6, 2013) and wet (November 29, 2013), was chosen for estimating isoprene, monoterpenes and other BVOCs emissions. Preliminary results obtained revealed hotspots of BVOCs in lower elevation zones. The highest emission occurred in the dry period, with differences between dry and wet days of around 60% for isoprene and 20% for monoterpene species. The highest fluxes occurred at approximately 15:00 h (LST), with mean fluxes of 5.44 and 1.43 ton h ^-1 for isoprene and monoterpenes respectively. A significant contribution of these VOC was emitted by biogenic sources in the study area, and significant changes between different types of BVOCs were observed. Isoprene emissions were also calculated with the WRF-CHEM model that has MEGAN 2.1 model (Guenther et al., 2012) by a simulation (Grell et al., 2005) in the study area with a resolution of 1km x 1km in the inner most domain. Comparison between MEGAN and BIGA isoprene estimates suggested that both models estimated similar fluxes (0-11600 μg m ^-2 h ^-1) in equivalent regions and times of the day (maximum emissions at around midday). However, the BIGA model was able to estimate higher resolution fluxes that may be important in zones dominated by forests and high temperatures.

References

Guenther, A. et al. (1993) Isoprene and monoterpene emission rate variability: Model evaluations and sensitivity analyses, Journal of Geophysical Research, 98(D7), p. 12609. doi: 10.1029/93JD00527.

Guenther, A. et al. (2012) The model of emissions of gases and aerosols from nature version 2.1 (MEGAN2.1): An extended and updated framework for modeling biogenic emissions, Geoscientific Model Development, 5(6), pp. 1471-1492. doi: 10.5194/gmd-5-1471-2012.

Grell, G.A. et al. (2005). Fully coupled online chemistry within the WRF model. Atmos. Environ. 39, 6957-6975. doi:10.1016/j.atmosenv.2005.04.027.

The 2017 National Emission Inventory for Crop Residue Burning

George Pouliot, Jeff Vukovich, Venkatesh Rao

Biomass burning from grasslands, rangelands, and crop residue is an important contributor to the degradation of air quality because of its impact on ozone and particulate matter. This poster will summarize improvements to the 2017 National Emission Inventory (NEI) for grasslands, rangelands, and crop residue emission sources. In addition, some of these improvements will be incorporated into the 2016 modeling platform currently being developed by EPA and the states. For the 2017 NEI, prescribed burning in the tall grass prairies of the Kansas Flint Hills region (which extends into Oklahoma) will have a unique source classification code. It will use measured emission factors based on recent field work, and the area burned will be based on a collaboration with the Kansas Department of Health & Environment. In addition, for the 2017 NEI, emission factors for hazardous air pollutants (HAPs) will be harmonized with volatile organic compound (VOC) speciation to ensure that the criteria air pollutants (CAP) and HAP emission inventories provide a consistent estimate of all VOCs. Finally, we will use state specific information to improve the classification of fires within the fire inventory.

Filling the gaps: Estimating roadway emissions using inconsistent traffic measurements in Las Vegas, Nevada near-road field study.

Michelle G. Snyder, R. Chris Owen, R. Cleary

The accuracy of air-quality modeling is reliant on the validity of model input characterization. For modeling projects that include emissions from mobile sources, real-time traffic measurements of volume and speed are needed to determine emissions associated with a roadway. Traffic measurements, like all measurements, can have uncertainties, inaccuracies, and errors. However, unlike many other measurements, there are fewer standards for calibration and QA. Thus, when there are unexpected changes in traffic, such as lane shifts or instrument error resulting in missing data, they are often undiagnosed. As a result, inaccurate traffic measurements misrepresent true roadway conditions and hinder the ability to produce reasonable emissions estimates for an air-quality model. In the Las Vegas near-road field study, missing traffic measurements and measurements with suspiciously low volumes and speeds were identified for months at a time over the yearlong study period. In this work, we leverage near-by and on-site traffic measurements to fill the gaps when traffic measurements are suspicious or missing. We examine traffic volume and speed patterns across multiple measurement locations and develop a methodology to determine the most accurate estimates of traffic volume and speeds throughout the Las Vegas near-road field study when traffic measurements are missing or inconsistent. The adjusted traffic estimates will be used in a refined modeling exercise, which will be compared with air-quality measurements to determine the importance of accurate traffic estimates in the air-quality modeling process.

A Case Study for Hawaii Volcano Eruption in 2018

Youhua Tang^1,2, Daniel Tong^1,2,3, Pius Lee¹, Barry Baker^1,2, Jeff McQueen⁴, Li Pan^4,5, Jianping Huang^4,5 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

The existing National Air Quality Forecasting Capability (NAQFC) covers Hawaii Islands but does not include volcano emissions. During the sustained eruption of Mt. Kilauea, May and June 2018, significant SO2 and PM2.5 enhancements were observed. Based on the monitoring data by U. S. Geological Survey's Hawaiian Volcano Observatory (HVO), we estimated the SO2 emission during this period. Besides the wind direction and speed, the plume rising height was one of key factors affecting the SO2 surface concentrations. We estimated the lava heat flux according to lava outflow volume, temperature and cooling speeds, which were used to calculate the vertical distribution of the plume rise. Since the Kilauea volcano's lava outflow was not explosive, and the initial upward momentum of volcano pollutants, or vog, was relatively low, its plume rise was mainly driven by the hot air's buoyancy. We employed the plume rise scheme for wildfire to calculate the vog's vertical distribution. The CMAQ 5.0.2 driven by the meteorology of North America Mesoscale model (NAM) was used to study the events, and the results were verified with in-situ measurements and satellite retrievals. We also performed the sensitivity study for the influence of volcano heat flux on the vog plume rise.

Extension of Version 4 of the Biogenic Emissions Landuse Database (BELD4) to Canada

Junhua Zhang, Michael D. Moran, and Zhuanshi He

Volatile organic compounds (VOCs) released from terrestrial vegetation are an important source of VOC emissions to the atmosphere. Globally, natural sources of VOC emissions have been estimated to be much larger than anthropogenic sources and to account for 80-90% of total global VOC emissions. The accurate estimation of biogenic VOC emissions requires input of landuse data with detailed vegetation types at relatively high resolution. The Biogenic Emissions Landuse Database, Version 3 (BELD3), which contains 230 vegetation classes at 1-km resolution, was compiled for most of North America (Pierce et al., 2000) and has been used widely by many regional air quality models for the last two decades to estimate biogenic emissions. However, the BELD3 vegetation fields were based on satellite imagery from the early 1990's and are now outdated for many parts of North America, particularly for areas undergoing rapid, large-scale development. Recently, the U.S. EPA updated BELD from Version 3 to Version 4 with 286 landuse categories. However, these updates were mostly for the contiguous United States. For Canada, only 17 broad landuse types based on MODIS satellite retrievals are contained in BELD4. In this presentation, a U.S.-equivalent extension of the BELD4 dataset to Canada will be described. This extended BELD4 dataset is based on a recent version of the Canadian National Forest Inventory (NFI) with 109 tree species at 250-meter resolution (Beaudoin et al., 2014) and the 2016 Canadian Annual Crop Inventory (ACI) with 71 vegetation and other fields at 30-meter resolution. Spatial distributions of various vegetation species from BELD3 and the extended version of BELD4 will be compared within Canada and differences for selected species between the extended and EPA versions of BELD4 will also be examined along the international border.

Beaudoin, A., et al., Mapping attributes of Canada's forests at moderate resolution through kNN and MODIS imagery. Can. J. For. Res., 44, 521-532, dx.doi.org/10.1139/cjfr-2013-0401, 2014.

Pierce Jr., T. E., Kinnee, E. J., and Geron, C. D.: Development of a 1-km vegetation database for modeling biogenic fluxes of hydrocarbons and nitric oxide. Sixth International Conference on Air Surface Exchange of Gases and Particles, July 3-7, Edinburgh, Scotland, https://www.epa.gov/sites/production/files/2015-08/beld3_web.ppsx, 2000.

Analysis of the Influence of the Metropolitan Area of Belo Horizonte Brazil on Local Meteorology Using WUDAPT - Urban WRF Model
Amanda Noronha Moreira de Carvalho1, Fabio Soares dos Santos1, Lizandro Gemiacki2, Taciana Toledo de Almeida Albuquerque1,3
1Department of Sanitary and Environmental Engineering, Universidade Federal de Minas Gerais, Brazil.
2 Instituto Nacional de Meterologia, Brazil. Post-Graduation Program in Environmental Engineering, Universidade Federal do Espirito Santo, Brazil.
corresponding author taciana@desa.ufmg.br

Population growth has intensified land use change by the expansion of urban areas. Variations in surface features are capable to interfere in the energy balance causing local meteorological changes. Currently in Brazil nearly 86% of the total population is estimated to be urban (WORLD BANK, 2017). The Metropolitan Area of Belo Horizonte (MABH) is the third most populated metropolis in Brazil with 5.3 million inhabitants distributed throughout a surface area of 9,500 km2 (IBGE, 2018). Characterized by a tropical climate with wet (October to March) and dry (April to September) seasons, MABH's air quality are significantly influenced by mesoscale circulation due to the complex topography and the emission from the vehicular fleet (Assis, 2010; Santos, 2018). Land use changes, as the expansion of MABH's urban area, can also influence on local circulation patterns. Thus, given the importance of the meteorological conditions to the environmental quality of the MABH, this paper investigates the impact of its urban cover growth on local meteorology, using the Weather Research and Forecasting (WRF) model coupled with the urban module Building Effect Parameterization (BEP) (Martilli et al., 2002). Firstly, model was validated and then, a scenario of urban sprawl was tested. The numerical simulations were performed for a period of ten days in September of 2017 (dry season) by WRF model version 3.9.1. Three nested domains, centered at the same coordinate, have been set up: a larger domain on 25 km spatial resolution, the second domain on 5 km resolution and the innermost domain on 1 km resolution. To take full advantage of the urban parameterizations, a vertical resolution of 44 eta levels was used, with the first 18 levels located below 1.5 km height. The initial and boundary conditions were provided by the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) final analysis data on a 0,25 x 0,25 degree horizontal resolution. A warm-up of 24 h was applied in the simulations to minimize the impacts of trading conditions. It was considered two urban surface parameterizations in the simulations: default and multi-layer Building Effect Parameterization. For the finer domain of WRF-BEP, urban land use data were updated to the ten built types of Local Climate Zone classification (Stewart & Oke, 2012), obtained according to WUDAPT level 0 methodology. Selected training areas on the Google Earth along with Landsat 8 images (acquisition date: 18 Sep. 2016) were imported to SAGA software for supervised classification. Natural classes from Moderate Resolution Imaging Spectroradiometer (MODIS) were maintained. Urban canopy parameters used for BEP's initialization were based on the studies of Stewart et al. (2014). Initially, a set of physical parameterizations were tested for both default and urban WRF versions. Simulation results were then analyzed as suggested by Emery et al. (2001), comparing modeled data to observed data through statistical indexes calculation. For the model validation were used data from four meteorological stations located at MABH. Parameters considered were air temperature, wind speed, wind direction and specific humidity. After model validation, was noted that the inclusion of more accurate land surface data and sub-grid calculation can improve simulation. Urban WRF was capable of, considerably, reducing wind speed mean bias. Near-surface temperature was also better predicted, with lower mean absolute gross error. For specific humidity and wind direction, however, BEP's did not presented any further gain. For simulation of the urban sprawl was used a scenario of population growth expected for the year of 2030. Land use and land cover were reclassified and it was considered that vegetation would be replaced for the LCZ 3 - compact low-rise. Results showed that population increase expected for the next 10 years is not significantly important to meteorological conditions alteration.

Updates in the Atmospheric Model Evaluation Tool (AMET) Version 1.4

K. Wyat Appel, Robert C. Gilliam, Christian Hogrefe and Kristen M. Foley

A new version of the Atmospheric Model Evaluation Tool (AMET) will be released in the summer of 2018. This new version (AMETv1.4) contains a large number of updates from the previous of version of AMET (v1.3) released in 2017. On the air quality side of AMET, an option has been added to read the paired model-observation files directly, bypassing the MySQL database. While this will limit some of the querying capabilities of AMET, it will simplify the installation of the software. In addition, several new analysis scripts have been developed that leverage third-party software to create interactive plots. One example is a new spatial plot that uses street-level mapping to allow zooming to better assess the location of observation sites. Similarly, a new interactive time-series plot has been added. On the meteorology side of AMET, additional analysis scripts have been added to perform upper-air analyses, along with several other scripts. In addition to the updates described above, several bug fixes and other minor updates are also included in the new version of AMET. This presentation will summarize the major updates in AMETv1.4. Also discussed will be updates to several of the other model evaluation tools that the Computational Exposure Division of EPA develops and releases to the public.

CONTRIBUTION OF WILDLAND AND AGRICULTURAL FIRES FROM NORTHERN SOUTH AMERICA TO PM2.5 and PM10 IN COLOMBIAN CITIES

Karen Ballesteros, Juan Felipe Mendez, Maria Paula Perez, Ricardo Morales

Open biomass burning is an important source of atmospheric aerosol particles, contributing a significant fraction to the global emissions of black carbon, brown carbon, and other species. The region of Amazonia in South America has long been identified as one of the largest contributors to short lived pollutants globally. However, massive natural wildfires and agricultural burns also occur every year in the grassland plains of Northern South America (NSA) during the dry season (November to April). Occurrence of drought events during La Nina years, combined with anthropogenic land-use changes, have been associated with the increase of wildland fires events in northern South America, deteriorating air quality at local and regional scales. The purpose in this research is to assess the seasonal contribution of biomass burning emissions from natural and agricultural fires in NSA to particulate matter (PM) levels in the central region of Colombia using a regional chemical transport model (WRF-Chem). Modeling domain covers the northern half of South America at a horizontal resolution of 27 km, with 121 x 121 grid cells and 41 vertical levels, centered in Colombia (lat. 5.194, long. -73.522). MOZART chemical mechanism coupled with MOSAIC aerosol scheme was used to perform simulations. Two 30-days simulation periods were selected based on number of fires reported by the Fire Information for Resource Management System (FIRMS) and PM concentrations reported at local air quality monitoring network (AQMN). The modeling periods were selected with anomalously high and low number of FIRMS detected fires respectively. Five experiments were developed per scenario, in which the sensitivity of PM2.5 and PM10 concentrations relative to fire emissions in main biogeographic regions of NSA (Amazonia, Orinoco grasslands, Caribbean plains, Pacific tropical rain forest, and Andean regions). The evaluation was performed comparing the concentrations and composition of PM2.5 and PM10 between simulation scenarios and observed data at AQMN stations, showing significant contribution of wildland fires to the total mass of PM in the city of Bogota. Simulation results identified fires from Venezuelan Amazon and Venezuelan Orinoco as main regions of contribution of PM.

Roadway dispersion modeling using AERMOD and R-LINE: an investigation into length, width, and dispersion model methodology for the Las Vegas near-road field study

Ryan Cleary, Michelle Snyder, R. Chris Owen

Model selection and characteristics of a modeled roadway, such as length and width, impact estimates of near-road concentrations. As modelers seek to maximize simulation efficiency without sacrificing the quality of model results, we explore three models to estimate dispersion from a roadway: AERMOD's LINE source, AERMOD's VOLUME source, and R-LINE, and compare their accuracy and runtime, a balance of efficiency. In this work we examine the methodology of each model and the impact on estimated concentrations in parallel, perpendicular, and upwind conditions. We explore the length necessary to effectively model impacts of a roadway based on the distance between the roadway and the receptor. Impacts vary depending on the dispersion model methodologies, such as meander and line versus point integration methods. We explore the ratio between the receptor distance and the length of the line to in various meteorological conditions. We present recommendations on the necessary roadway length when using each of the three models and the proper set-up for each model to achieve the most accurate and efficient results for project-level analyses.

CMAQ 5.2 AND 5.2.1 PARALLEL PERFORMANCE WITH MPI AND OPENMP

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

This presentation reports on hybrid MPI and thread parallel performance results for CMAQ 5.2 and 5.2.1.Attention is focused on the GEAR solver in the Chemistry Transport Model (CTM), for both FSparse [1], and the legacy JSparse [2] algorithms. The former implements OpenMP thread parallelism for which the GEAR solver is well suited. Portland and Intel FORTRAN compilers are compared for execution performance with a 24 hour scenario included with the CMAQ download. Both the original (EPA) JSPARSE version and the FSPARSE thread parallel version are compared. Results will be presented for both MPI and thread scaling on a heterogeneous cluster of 10 nodes with a total of 128 cores.

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

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

Integrating Speciated Particulate Matter Data to Improve Model Performance in Bogota

James East, Juan Sebastian Montealegre, Johan Sebastian Vanegas, Fernando Garcia Menendez, Jorge E. Pachon

Bogota, Colombia frequently exceeds the coarse and fine particulate matter (PM10 and PM2.5) daily standards. Air quality modeling has been conducted for the past three years to understand the fate and transport of pollutants and evaluate emission reduction scenarios. However, lack of chemically speciated PM measurements has been a barrier to PM model evaluation. Recently, a network of ground-based PM measurements was set-up and now provide information on the concentrations of PM and the chemical species from four unique sites across the city. This work uses speciated PM10 and PM2.5 measurements to improve CMAQ model performance and gain insight into local air quality in Bogota. CMAQ v5.2 was used to simulate air quality using 2014 meteorology and a local emissions inventory. Modeling efforts performed since speciated data have become available resulted in poor predictions of elemental and organic carbon fractions in particulate matter (PM) when model output was compared to observations. This poor performance was likely due to emission profiles not adequately representing local sources. The chemically speciated data were used to adjust the source profiles for modeled PM components and reduce model shortcomings to better reflect actual air quality conditions. The improved model and emissions inventory will be useful in future air quality health and cost-benefit analyses for Bogota. In addition, this study highlights strategies to best integrate speciated PM data in air quality simulations over regions where chemical composition information has not been previously available.

Evaluation of WRF Simulations for Two High Ozone Episodes in El Paso, Texas

Jennifer Hegarty¹, Amy McVey¹, and John Henderson¹

¹Atmospheric and Environmental Research, Lexington Massachusetts

Two recent high ozone episodes in El Paso, TX on June 21, 2015 and June 23, 2016 produced maximum daily average 8 hour (MDA8) ozone values exceeding 70 ppbv. These episodes highlight the need for photochemical modeling to better understand the causes of high ozone in the region, which requires high resolution meteorological inputs generated by a mesoscale numerical weather prediction (NWP) model. In this study, the Advanced Research Weather Research and Forecasting model (WRF) was used to simulate the meteorology of these episodes to evaluate the model's ability to reproduce the fine-scale meteorological features that influence the build-up of ozone. The WRF grid configuration included an inner horizontal domain with 1 km grid size to represent the influence of the complex terrain of the Rio Grande Valley. Several physics options including three planetary boundary layer (PBL) parameterizations, Mellor-Yamada-Janjic (MYJ), Mellor-Yamada-Nakanishi-Niino (MYNN) and Shin-Hong (SH) were tested. In addition the feasibility of improving the simulations by observational nudging was investigated. For both cases synoptic conditions featured generally light winds throughout the lower half of the troposphere. The simulations depicted wind channeling through the valley with general upslope winds during the day and downslope winds at night; however, winds at most surface locations including the El Paso airport were light (< 2 m/s) and did not show evidence of this diurnal cycle of wind shifts. Modeled winds at these locations were also light but increased for short periods of time due to transient convective cells that were difficult to verify given the available observations. Only minor differences were attributable to the different physics options but the simulations with the SH PBL scheme were slightly better. Observational nudging was not effective in improving the simulations possibly due to the complex terrain, which limits the spatial relevance of individual observations. These results point out the challenges of simulating flow through complex terrain and evaluating the simulations in low wind conditions using standard model observation comparisons at individual locations and times. Future work will examine the transport patterns over one or several days with Lagrangian models and how these patterns are affected by WRF inputs generated with different physics options.

Development and Application of a Simple Model Assessment R Tool (SMART)

Daiwen Kang

As an environment for statistical computing and graphics, the programming language R is widely used and freely available. The EPA's Atmospheric Model Evaluation Tool (AMET) is developed using R to produce statistical analysis and generate various graphics. Complementary to AMET, SMART provides an alternative tool to perform model evaluation for simple projects independent of databases and other languages for pre-processing. With the design of unique data structure and common interfaces to process model outputs and observations, SMART can easily pair any model outputs with observations from common observation networks and special field studies. The large memory requirement and slow processing speed associated with the R platform (the language and running environment) when dealing with large datasets are resolved by using SQLDF packages and parallel computing packages. Saved R objects serve as pseudo database tables to perform statistical analysis and generate graphics for various model scenarios. With functionality to subset and merge for the designed data structure, it is easy and fast to generate statistics and graphics for any subsets of the model-observation pairs in time and space. Since R is the only language used for site-compare and statistical analysis, it is completely portable to any computing environment with R installed. To run SMART, except for the shell script that is used to define all the environmental variables (such as input and output directories and variable names), users are not required to have any knowledge of R.

Practical Application of Python for Air Quality Data Analysis and Modeling

Byeong-Uk Kim, Marcus Trail, Tao Zeng, Di Tian, and James Boylan

Python libraries can be used by state air quality agencies to perform various mission-critical tasks. Several examples for analyzing and reviewing ambient air quality data, emission inventory data, and air quality modeling inputs/outputs will be presented. Some examples focus on data processing and task controls while others demonstrate effective and efficient data visualization. Each example consists of a problem statement, challenges with traditional approaches, and the Python libraries employed to solve the problem. Generic purpose libraries (e.g., Pandas, Matplotlib, python-netCDF4, and ArcPy) as well as air quality specific libraries (e.g., pysplit and PseudoNetCDF), will be discussed. Finally, we will share our perspective on emerging libraries that may replace currently used libraries due to their support limitations in the future or official migration plans.

Updating resuspended particulate matter (RPM) emissions of Bogota, based on a response analysis of simulated PM10 concentrations and their species using CMAQ.

Juan Sebastian Montealegre, James East, Johan Sebastian Vanegas, Fernando Garcia Menendez, Jorge E. Pachon

Air pollution is a large concern in Bogota, Colombia where concentrations of PM₁₀ and PM_2.5 frequently do not meet air quality standards. Local emission inventories show resuspended particle matter (RPM) as the largest fraction of PM emissions, greater than combustion from mobile or point sources. However, CMAQ modeling conducted in Bogota reflects an overestimation of simulated PM concentrations, likely due to the magnitude of RPM emissions. In this work, we summarize our efforts to create an updated resuspended particulate matter (RPM) emissions inventory. This update includes the implementation of local emissions factors (EFs) and chemical profiles for urban road dust in CMAQ simulations. Additionally, the use of chemically speciated PM₁₀ and PM_2.5 information enables model evaluation by PM species focused on coarse fractions typically associated to RPM emissions. Improved model performance will benefit health and cost-benefit analyses in Bogota. This methodology could be of interest in regions where RPM are significant sources of ambient PM and poorly simulated by chemical transport models.

Improving Ozone Simulations in the Great Lakes Region: Sensitivity to Emissions and Chemistry

Momei Qin¹, M. Talat Odman¹, Haofei Yu^1,2, Yongtao Hu¹, Armistead G. Russell¹, Arastoo Pour-Biazar³, Richard T. McNider³, and Eladio Knipping⁴

1. Georgia Institute of Technology

2. University of Central Florida

3. University of Alabama at Huntsville

4. Electric Power Research Institute

The Great Lakes Region still experiences exceedances of the ozone (O₃) standards in spite of years of emissions controls. This is, in part, due to complex interactions between the meteorology influenced by the lakes and emissions from surrounding large cities, which pose a challenge to fully capturing the O₃ dynamics in the region. A series of sensitivity tests were conducted using the Community Multiscale Air Quality (CMAQ) model for high O₃ periods in July 2011 to identify the roles of emissions biases (e.g., biogenic VOCs and anthropogenic NO_x) and chemical mechanism choice. The base model/emissions configuration used the Carbon Bond mechanism CB05 and biogenic emissions from the Biogenic Emission Inventory System (BEIS), along with anthropogenic emissions from the 2011 National Emissions Inventory (NEI). This led to a high bias in the daily maximum 8-hour average (MDA8) O₃ across the domain, particularly at coastal sites. Using CB6 chemical mechanism or 50% reduction of NO_x emissions from mobile sources led to substantial domain-wide decreases in O₃ from the base case, and improvements in model performance particularly along the Lake Michigan shoreline. Using the Model of Emissions of Gases and Aerosols from Nature (MEGAN) instead of BEIS to estimate biogenic emissions had minor impacts on simulated O₃, while combining MEGAN with CB6 resulted in improved elevated (i.e., greater than 60 ppb) MDA8 O₃ simulation along the western coast of Lake Michigan. Using CB6 combined with MEGAN and a 30% reduction of mobile NO_x emissions led to the best performance. Together with improvements in meteorological modeling, the recommended configuration could serve as a starting point for future O₃ modeling in the region.

Multiyear Model Evaluation over Lake Michigan Region
Arastoo Pour Biazar, Richard T. McNider, Kevin Doty, Andrew White, Yuling Wu, Momei Qin, Yongtao Hu, Talat Odman, Bright Dornblaser, Patricia Cleary, Eladio Knipping, Stuart McKeen, Pius Lee

Previous field studies over Lake Michigan region have indicated that the transport of ozone precursors from urban centers to areas over the lake allows ozone production and accumulation within the shallow boundary layer over water, and a return flow due to Lake Breeze bring the elevated ozone back onshore and over the population centers. However, some studies suggest that the air quality models overestimate ozone concentrations over cooler bodies of water. Combined with inaccuracies in the timing and extent of the onshore flow during the day that transports the high ozone and aged precursors back over land, levels of ozone at the shorelines cannot be simulated accurately. This may be partly due to the representation of the stable boundary layer in the model and/or due to errors in lake and land temperatures that define the strength of lake/land breezes.

In the current study, the average summertime behavior of flow field and chemistry over the lake is analyzed. The results from air quality simulations over summers of 2009, 2011, and 2013 are used to ascertain the key factors impacting ozone production and accumulation over the lake. The results indicate that while there are pronounced similarities in the average flow field and spatial variation of ozone over the Lake Michigan, the strength of the predominant westerlies, land/water temperature contrast, and the curvature of the shoreline determine the onset of land breeze and the location of a divergence zone at the lake surface that is prone to high ozone production. Furthermore, the role of vertical mixing over the lake impacting higher surface ozone will be discussed.

Fusing Observational Data and Chemical Transport Model Simulations to Create Spatiotemporally Resolved Ambient Air Pollution Fields for Health Analysis

Niru Senthilkumar, Francesca Metcalf, Mariel Friberg, Armistead Russell, James Mulholland

Using Data Fusion (DF) we have addressed the need for spatially and temporally complete air quality data fields across the contiguous United States. This method uses statistical regression approaches to combine ground-based observational data to the results of CMAQ at a 12km resolution over a ten-year period. The ground-based observational data is taken from the EPA's AQS database which contains ambient air pollution data collected by the EPA, state, and local air pollution control agencies across the US. This DF technique is applied for 12 species (CO, NO₂, NO_x, O₃, SO₂, PM₁₀, and five speciated PM_2.5 species). This method involves first developing regression parameters to correct the modeled CMAQ results according to the available observational data. Next, we create a ratio between the corrected CMAQ and observational data which are interpolated through inverse distance weighting across the contiguous US to create a smooth field of concentrations. Cross-validation through 10% data withholding has shown a range in performance across the species, with the model predicting Ozone and PM_2.5 the best and SO₂ the worst. The lack of performance in SO₂ is theorized to stem from CMAQ's limitations in capturing SO₂ plumes at a 12km resolution. Overall the fusion approach is able to provide daily concentration fields that are able to capture the uncertainties and variability of the observational data while maintaining the spatial patterns of the simulated CMAQ results.

Evaluation of the Weather Research and Forecasting (WRF) Model Performance for Surface Energy Balance Terms during Persistent Cold Air Pool Events

Xia Sun, Heather A. Holmes

Atmospheric Sciences Program, Department of Physics, University of Nevada, Reno, NV, USA

Land-atmosphere interactions play a crucial role in the boundary layer development by governing the exchange of heat, moisture and momentum through surface turbulent fluxes. The boundary layer structures during persistent cold air pools (PCAPs) are featured with stable stratifications with limited mixing. Realistic simulation of the land surface processes is important for the prediction of the PCAPs evolution. The surface energy balance terms include net radiation (Rn), surface sensible heat flux (H), surface latent heat flux (LE), and ground heat flux (G). The relative importance of the surface turbulent fluxes is determined by the partitioning of surface available energy (Rn-G), which is impacted by the land surface characteristics, such as soil moisture.

The Weather Research and Forecasting (WRF) model simulations of the SEB components during PCAPs were evaluated with observation datasets from the Persistent Cold Air Pool Study conducted in Salt Lake Valley, Utah. Four WRF simulations were conducted using the Asymmetric Convective Model v2 (ACM2), Yonsei University (YSU), Mellor-Yamada-Janjic (MYJ), and Mellor-Yamada Nakanishi & Niino (MYNN) PBL schemes with corresponding recommended LSMs to study the land surface processes sensitivity to WRF PBL parameterizations. Simulated radiation components, including downward shortwave radiation (DSR), upward shortwave radiation (USR), downward longwave radiation (DLR), and upward longwave radiation (ULR), were also compared with observations.

Modeling Spatiotemporal Variations of Energy- and Health-Related Pollutants over Continental U.S.: Multi-Model Intercomparison and Ensemble

Yang Zhang, Kai Wang, and Chinmay Jena

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

Three-dimensional (3-D) air quality models (AQMs) are powerful tools for simulation of current and future air quality. Despite significant progress in the past decades, current AQMs have limited capabilities in reproducing observed spatial and temporal variations of health-related pollutants (e.g., O3, PM2.5, PM composition), which lead to models' incapability in representing regional heterogeneity in observed exposures and adverse human health effects. As part of the U.S. EPA program on Air, Climate, and Energy centers, the center of Solutions for Energy, AiR, Climate, and Health (SEARCH), four state-of-science online-coupled climate and air quality models are applied to generate the best possible predictions of ozone (O3) and fine particulate matter (PM2.5) concentrations in order to enhance public health effect assessment under various energy transition scenarios. These models include WRF/Chem, WRF/Chem-ROMS, WRF-CAM5, and WRF-CMAQ. Following initial application and evaluation of these models for a five-year time period (2008-2012) over continental U.S. (CONUS), model biases are first identified and then reduced through improving model formulations, constraining model inputs, and generating multi-model ensembles. A number of month-long sensitivity simulations using WRF/Chem are conducted with improved model inputs and schemes to evaluate their impacts on model performance and identify optimal model configurations, which are then applied to all models for the 5-year simulations over the CONUS. Two ensemble methods are used to further reduce the model biases and errors including a simple ensemble mean based on an average of ensemble members, and a weighted ensemble mean based on the multi-linear regression. The skills of individual models and their ensembles are evaluated using available data from surface networks and satellite retrievals. Sensitivity simulations show that updated boundary conditions (BCONs) downscaled from the new global simulations, constrained BCONs based on satellite retrievals, and modified dust emission schemes lead to overall better spatiotemporal variation for major air pollutants such as O3, carbon monoxide (CO), PM2.5, and PM10. Four individual models perform generally well for major meteorological variables such as 2-m temperature and relative humidity, 10-m wind speed, and precipitation and for major chemical species such as O3 and PM2.5. Moderate-to-large model biases and errors exist for some PM2.5 species and column variables, due to uncertainties associated with model inputs such as biogenic and wildfire emissions, model treatments such as online dust emission modules, and satellite retrievals. Compared to individual models and the simple ensemble mean, the weighted ensemble predictions based on the multi-linear regression outperform in terms of normalized mean error, root mean square error, and correlation coefficient for most variables. These results illustrate the current capability of the online-coupled air quality models and the potential of weighted ensemble in generating the best possible estimates of concentrations of multi-pollutants for the assessment of their impacts on air quality and public health. Improved model results will be used to identify pollution hot spots and provide baseline for air quality and human health impact studies under a variety of energy transition and climate change scenarios.

Understanding Source-Receptor Relationship of Local Air Pollution in Taiwan using Response Surface Modeling Method

Qingzhao Zhu, Xinyi Dong and Joshua Fu

As air quality has a great impact on human health, governments exert huge efforts on emission control strategies to reduce sources of pollution from various sectors such as industry, power plant, transportation, and so on. It is important to assess how the air pollution status responds to emission reduction scenarios so that the policy- and decision-makers could enact an effective strategy to reduce the negative effect of poor air quality. In this study, we use the Response Surface Modeling (RSM) method with WRF/CMAQ modeling system to investigate the source-receptor relationship between emissions and concentrations in different areas in Taiwan. Emission control scenarios based on different sectors (i.e., PP, ST, OP, AR, TR, PC, OT) and regions (i.e., TC, CH, NT, YL, TN, TS) in four typical months (i.e., January, April, July, and October) from each season are designed. 451 simulations will be done for each month in order to generate accurate RSM fitting results. Although most of the industrials and energy generating units located at the north and south poles in Taiwan, the air pollutants and the precursors from primary emission sources may transport across the whole area of Taiwan due to the complicated sub-tropical climate condition. In this study, we will evaluate the effectiveness of emission control efforts from each source region on the improvement of air quality in both local and other downwind areas. Different emission scenarios of tracer species from different emission sectors and regions will be investigated to explore the response of air quality such as the concentration of PM2.5 and ozone.

Development and Assessment of Predictive Algorithms for Fine Particulate Matter in New York State

Samuel Lightstone, Barry Gross, Fred Moshary

Human health is strongly affected by the concentration of fine particulate matter (PM_2.5). The need to forecast unhealthy conditions has driven the development of Chemical Transport Models such as CMAQ. These models attempt to simulate the complex dynamics of chemical transport by combined meteorology, emission inventories (EI's), and gas/particle chemistry and dynamics. Ultimately, the goal is to establish useful forecasts that could provide vulnerable members of the population with warnings. This paper explores the potential of different approaches in providing these forecasts. First, we assess the potential of CMAQ forecasts at the single grid cell level (12km), and show that significant variability not encountered in the field measurements occurs. This observation motivates the exploration of other data driven approaches, in particular, a neural network (NN) approach. This approach makes use of meteorology and PM_2.5observations as model predictors. Furthermore, utilizing a combination of spatial Kriging from PM_2.5ground stations with satellite measured Aerosol Optical Depth, we employ our data driven approach to create continuous forecasts maps. Most importantly, these maps provide high resolution forecast data to regions that have insufficient coverage due to the sparse and limited number of ground sampling stations. We find that the data driven approach generally results in a more accurate prediction of future pollution levels at the 12km spatial resolution scale of CMAQ. Additionally, we find that the NN is able to adjust to the sharp transitions encountered in pollution transported events, such as smoke plumes from forest fires, more accurately than CMAQ.

Splitting the Combined Prescribed Burning Impact from CMAQ-DDM into Individual Burn Impacts by using HYSPLIT Dispersion Model

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

Prescribed burning (PB) is a land management practice used to improve native vegetation and wildlife habitat, control insects and disease, and reduce wildfire risk in the US. Its emissions remain one of the largest sources of PM2.5 in the southeastern US. A PB impact forecasting system has been developed to facilitate the dynamic management of air quality by forecasting PB emissions. However, the burn impact calculated by the forecasting system is the combined impact of all the burns. In order to split the combined burn impact and find the number of acres that could burn in each county without causing any air quality issues, a dispersion model (HYSPLIT) will be used to partition the impact to individual burns upwind. PM2.5 concentrations in a case study will be simulated using the Community Multiscale Air Quality (CMAQ) model and the Decoupled Direct Method (DDM), a sensitivity analysis technique for computing sensitivity coefficients simultaneously while air pollutant concentrations are being computed, will be used to provide the combined impact of burns in Georgia. BlueSky framework will be used to estimate emissions from the burns. Then, HYSPLIT will be used to generate single burn dispersion fields. The combined burn impact from CMAQ-DDM will be split and rescaled according to dispersion fields from HYSPLIT. The final fields will be evaluated by comparing them with the single burn impacts calculated by CMAQ-DDM. Single burn impacts obtained by splitting the combined burn impact obtained from CMAQ-DDM could help local fire and air quality managers to decide which burns should be allowed based on their individual impact on people or areas of concern.

An Optimization Model for National Ambient Air Quality Standards (NAAQS) Benefit-Cost Analysis

Alexander Macpherson, Heather Simon, Ali Kamal, Jenny Thomas, Neal Fann

An Optimization Model for National Ambient Air Quality Standards (NAAQS) Benefit-Cost Analysis

Alexander Macpherson, Heather Simon, Ali Kamal, Jenny Thomas, Neal Fann

Office of Air Quality Planning and Standards, Office of Air and Radiation, United States Environmental Protection Agency, 109 T.W. Alexander Drive, Research Triangle Park, NC 27711 USA.

Macpherson et al. (2017) presented a mathematical programming model that can help identify potential minimum-cost emissions control strategies that employ regional strategies to meet an air quality standard. That model combined detailed emission control information and a reduced-form air quality model to rapidly evaluate alternative attainment planning scenarios, test regional strategies, and identify monitors that potentially have significant influence on attainment strategies. This work extends the optimization framework to incorporate an improved characterization of air quality and adds a health impacts analysis module. The improved reduced-form air quality model draws upon source apportionment modeling to characterize the impact of emissions reductions within states on ozone design values at monitors and seasonal mean 8-hour daily maximum ozone levels in 12km2 grid cells across the lower 48 states. The health impacts analysis module uses the changes in seasonal mean 8-hour daily maximum ozone levels resulting from the application of emission control strategies to evaluate changes in the incidence of health impacts and generates monetized benefits estimates. These improvements enable the examination of different strategies to attain ozone air quality standards, including: achieving a standard via a least cost approach, maximizing net benefits, or optimizing on distributional aspects of health impacts. This talk presents preliminary results comparing and contrasting national control strategies under alternative management objectives.

Disclaimer: This work represents the views of the authors and does not reflect official Environmental Protection Agency policy.

Cloud Application of a Photochemical Grid Model with Reduced Resolution: An Alternative to Reduced Form Models

Christopher Emery, Pradeepa Vennam, Greg Yarwood, Eladio Knipping, Naresh Kumar

Reduced form models (RFM) are used as screening tools that simplify calculations relating pollutant emissions to downstream consequences: air concentrations; human and agricultural exposure; and health, welfare and monetized risk. These models enable policy makers to quickly screen multitudes of emission scenarios and resulting risks and costs, but with a loss of fidelity to varying degrees. RFMs are limited to specific chemical species (ozone and/or PM), and pre-defined spatial (county or state) and temporal (seasonal or annual) scales and metrics. Hyper-space response surface models, developed with state-of-the-science photochemical grid models (PGMs), involve extensive resources to build and update, and require that scenarios of interest fit within the pre-defined matrix of sensitivities. Here we describe an alternative approach that leverages the power of a scalable cloud computing environment to run a PGM for an entire year on a US-wide domain with moderate spatial resolution (36 km, 13 layers) well within a single day. This approach offers advantages in speed, cost, and flexibility. Emission scenarios can be derived with an online front-end application that applies factors to criteria air pollutant emissions by major source category/region. A comprehensive range of gridded output metrics including air concentrations and deposition loadings can be generated for any species carried by the PGM at the spatial scales of the grid and at temporal scales ranging from hourly to annual. Specific outcome metrics such as visibility, exposure, and risk can be post-processed off-line or with a back-end application. We present analyses of impacts to PGM performance against measurements, and to responses from deep NOx, SOx, and VOC emission reductions, relative to EPA's full-resolution 2011 US Modeling Platform. Speed and costs are presented for a particular cloud platform and configuration.

COMPILERS AND PROBLEMS WITH CMAQ 5.2 AND 5.2.1

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

This presentation reports on build and runtime problems encountered with Portland and Intel FORTRAN compilers for CMAQ 5.2 and 5.2.1. Attention is focused on the GEAR solver in the Chemistry Transport Model (CTM), for both FSparse [1], and the legacy JSparse [2] algorithms, Specific compiler problems have been reported to the vendors but others remain unresolved while work-arounds have been used. These are often related to different releases of a compiler, and include failure in a CMAQ build due to differences between them. Runtime failures include NaNs with some versions of the Intel compiler (ignored at runtime), and/or the Portland compiler (correctly causing execution termination). These issues are due in part to the rapid transformations in the CMAQ releases from 5.1, to 5.2, and 5.2.1. A summary will itemize some key observations in CMAQ builds over the last year with recommendations for the CMAQ user community.

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

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

Preparing MPAS-A for Global Retrospective Air Quality modeling: An evaluation of MPAS with comparisons to WRF

Robert Gilliam, Jerold Herwehe, Russell Bullock, David Wong and Jonathan Pleim

The US EPA has a plan to leverage recent advances in meteorological modeling to develop a "Next-Generation" air quality modeling system that will allow consistent modeling of atmospheric dynamics and chemistry from global to local scale. The meteorological model of choice is the Model for Prediction Across Scales-Atmosphere (MPAS-A) that has been developed by the National Center for Atmospheric Research in recent years. While CMAQ developers have been working on a method to couple CMAQ components to MPAS-A for full global chemical transport modeling, a team of weather modelers has been preparing MPAS-A for accurate meteorological simulations. This includes four dimensional data assimilation that allows for long simulations of past weather with no growth in error. We have also added the Pleim-Xiu land-surface model (P-X LSM), Asymmetric Convective Model 2 (ACM2) and Pleim surface layer, which are key components in the current meteorological model WRF.

The evaluation of MPAS-A and comparison to WRF includes a full annual 2016 simulation using both a 92-25 km and 46-12 km global mesh with finer meshes centered over the CONUS. Evaluation includes comparisons of the model runs to surface meteorology, PRISM precipitation, radiation and upper air rawinsonde measurements. MPAS-A generally compares well with WRF for most of 2016. Error levels in surface meteorology are at or below WRF levels during the cooler parts of the year and slightly worse during the summer. MPAS and WRF both simulate annual precipitation patterns that are highly correlated with PRISM observations. MPAS has slightly lower mean absolute error than WRF with better performance in the western and southern parts of the US. WRF is slightly better in the Midwest. Solar radiation in MPAS-A is higher than WRF and measurements, indicating a possible lack of clouds in MPAS-A versus WRF. Twice daily rawinsonde measurements indicate both models simulate precise temperature and wind in the free troposphere. MPAS simulates many instances of greater than 95% relative humidity in the upper troposphere unlike WRF, and this characteristic of MPAS is not supported by observations. Finally, a comparison of the 92-25 km and 46-12 km meshes shows clear improvements in surface meteorology as grid scale is reduced. The comparisons indicate the meteorology of MPAS-A is sufficient for retrospective global air quality modeling.

Evaluation of CMAQ adjoint and Future Steps

Shunliu Zhao, Amir Hakami (Carleton University) Matt D. Turner, Daven K. Henze (University of Colorado); Shannon L. Capps (Drexel University); Peter B. Percell (University of Houston); Jaroslav Resler (ICS Prague); Jesse O. Bash, Sergey L. Napelenok, Kathleen Fahey, Rob W. Pinder (USEPA); Armistead G. Russell,Athanasios Nenes (Georgia Tech); Jaemeen Baek, Greg R. Carmichael, and Charlie O. Stanier (University of Iowa); Adrian Sandu (Virginia Tech); Tianfeng Chai (University of Maryland).

A cross-institutional team of researchers has worked for the past few years on the development of a full adjoint for CMAQ (including aerosol, clouds and aqueous chemistry, gas-phase chemistry, and transport). CMAQ-ADJ development is now complete. This poster presents final evaluations of the adjoint model, a retrospective view of challenges faced, limitations, and future steps in adjoint model development.

PseudoNetCDF v3
Barron Henderson

PseudoNetCDF has provided fundamental data interaction for binary formats used by air quality models. This poster will cover major developments in PseudoNetCDF including new data formats, easier access to many formats, and better plotting options. In addition, PseudoNetCDF has a recent integration into the xarray platform. That means that any formats that PseudoNetCDF can read get access to the rich features available in xarray. This poster will highlight these new features through use case diagrams, example scripts and visualized outputs.

Top-down NOx Emissions Estimation using a Perfect Model Assumption and Deep Neural Networks

Hyun Cheol Kim, 1,2 Barry Baker1,2, Tianfeng Chai1,2, Changhan Bae3, Seunghee You3, Byeong-Uk Kim4 and Soontae Kim3

1 Air Resources Laboratory, National Oceanic and Atmospheric Administration, College Park, MD

2 Cooperative Institute for Climate and Satellites, University of Maryland, College Park, MD

3Ajou University, Dept. of Environmental Engineering, Suwon, Korea 4 Georgia Environmental Protection Division, Atlanta, GA

Top-down constraint of NOx emissions using observations from satellites and monitoring sites has been a popular way to complement bottom-up emissions estimation to provide accurate modeling inputs for regional air quality studies. However, its actual application is somehow limited due to the nonlinear characteristics of the observed responses to the change of emissions. To consider these chemical nonlinearity and physical uncertainties (e.g. transport pattern changes due to wind fields), we developed a Python-based emission inverse modeling framework using a perfect model assumption and a newly developed machine learning technique. The perfect model assumes that the modeled world is true. Based on this assumption, we build pseudo- observations and pseudo- satellite that mimic true observations of real world. Pseudo observations are used to train the system to learn the emission-to-concentration relation using the deep neutral networks method in the Google TensorFlow library. Three sets of twelve years CMAQ simulations over East Asia (27-km domain) and South Korea (9-km domain), using MICS-Asia, INTEX-B, CREATE (for Asia) and CAPSS (for South Korea) emission inventories, are used for the training. Results demonstrate that, after training, the system is able to regenerate NOx emissions of independent period with accuracy. We further discuss the uncertainty and efficiency in using pseudo- observations, and limitations in actual application to real world observations.

Orders-of-magnitude speedup in atmospheric chemistry modeling through neural network-based emulation

Makoto M. Kelp, Christopher W. Tessum, and Julian D. Marshall

Chemical transport models (CTMs) are computationally expensive, largely owing to the computational intensity of the atmospheric chemistry solver. Here we investigate the potential for using machine learning to reproduce the behavior of a chemical mechanism with reduced computational expense. We train the network to match chemical mechanism predictions of changes concentrations of 77 chemical species after one hour with RMSE of 1.97 ppb or less (median RMSE of 0.02 ppb) while achieving a ~230 times computational speedup using the same hardware. An additional two orders of magnitude speedup is possible when running the neural network on a graphics-processing unit (GPU). The neural network is able to reproduce the emergent behavior of the chemical system over diurnal cycles using Euler integration, but additional work is necessary to constrain the propagation of error as simulation time progresses.

Global Particulate Matter Source Apportionment

Lamancusa, Carmen, Wagstrom, Kristina

Source apportionment at regional scales has provided researchers and regulators with a powerful tool to quantify and estimate the impacts of a variety of sources at regional scales. We have developed a new source apportionment tool for the GEOS-Chem global atmospheric chemical transport model to expand these capabilities to global models. The Global Particulate Matter Source Apportionment Technology (GPSAT) module represents the first full incorporation of source apportionment into a global model based on the same principles as the Particulate Matter Source Apportionment Technology (PSAT) in CAMx. We built GPSAT to handle not only primary particulate matter but also secondary particulate matter and its precursors. We have developed GPSAT within GEOS-Chem so that a wide variety of users will have access to GPSAT. By building GPSAT into the core of GEOS-Chem we have made it possible to track different species within GEOS-Chem without needing to create customized tracers or additional emissions files. This approach minimally impacts GEOS-Chem's runtimes as it does not add tracers into the computationally expensive chemistry and advection calculations. We have completed several tests to quantify the computational impact of running GEOS-Chem with and without GPSAT. By removing the need to create custom tracer and emissions files and remaining conscious of the computational cost, we have designed GPSAT to lower the barrier to new and exciting research. Using GPSAT, we have quantified the relative impacts of the twenty largest megacities on global nitrate, sulfate, and total particulate matter concentrations. As part of this study we have calculated the average distance traveled by the particulate matter from these megacities. We have identified the regions of the globe most disproportionately impacted by their collective and individual emissions.

A Dynamic Evaluation of Aviations Contribution to Fine Particulate Matter in the United States

Calvin Arter, Institute for the Environment, The University of North Carolina at Chapel Hill, NC, USA

Kevin Lane, Department of Environmental Health, Boston University School of Public Health, Boston, MA, USA

Jonathan Levy, Department of Environmental Health, Boston University School of Public Health, Boston, MA, USA

Sarav Arunachalam, Institute for the Environment, The University of North Carolina at Chapel Hill, NC, USA

In this study, we present a dynamic evaluation of the WRF-SMOKE-AEDT-CMAQ modelling system to assess the air quality and health impacts of U.S.-wide aircraft emissions during landing and takeoff (LTO) cycles. We focus on ambient concentrations of fine particulate matter (PM2.5) in the U.S. during the years 2005, 2011, 2015 and 2025. Specifically, we take a previously completed assessment of the aviation-attributable health impacts for a base year 2005 and future projected year 2025, and compare these with more recently explicitly modelled years for 2011 and 2015. The estimates from the explicitly modeled years will be compared with interpolated estimates from the previous 2005 to 2025 trajectory to dynamically assess the efficacy of the modeling system for accurately characterizing air quality and health impacts during contemporary periods. The modeled air quality contributions will be used to estimate the aviation-related health impacts by combining the concentrations with concentration-response functions for PM2.5 mortality and population data. Our assessment will be novel in presenting a dynamic evaluation of the entire modelling system, but focusing on the aviation emissions sector, which is small compared to other source sectors but of growing significance due to projected growth in global aviation activity.

Development of TracMyAir Smart Phone App for Modeling Individual Exposures to Ambient PM2.5 and Ozone

Michael Breen,Yadong Xu, Catherine Seppanen, Sarav Arunachalam, Timothy Barzyk, Janet Burke, David Diaz-Sanchez, Peter Egeghy, Stephen Graham, Vito Ilacqua, Vlad Isakov, Vasu Kilaru, Laura Kolb, John Langstaff, Thomas Long, Steven Prince, Jennifer Richmond-Bryant, Ronald Williams

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 have several limitations, including need for substantial technical exposure modeling expertise. Near real-time predictions are not possible since large and diverse input data must be collected, organized, and processed by complex exposure models. To address these limitations, we are developing TracMyAir, an iPhone application that automatically estimates exposures to ambient PM_2.5 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 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 PM_2.5 and ozone when indoors. The app will be an efficient and cost-effective tool for researchers to apply exposure predictions for epidemiology studies and other situations where direct air quality monitoring cannot be performed. TracMyAir can also be used for developing public health strategies to help susceptible people (e.g., those with asthma) reduce their exposure to air pollution. Future versions of TracMyAir can be integrated with air quality models when nearby monitoring data is not available.

An Environmental Data Web Service Based on Near-Road Dispersion Modeling to Support the Los Angeles Pediatric Research Integrating Sensor Monitoring Systems (PRISMS) Informatics Center

Kenneth Craig¹, Shih Ying Chang¹, Annie Seagram¹, Fred Lurmann¹, Alex Bui², Anahita Hosseini², Majid Sarrafzadeh², Rose Rocchio², Rima Habre^3
1Sonoma Technology, Inc., Petaluma, CA
²University of California Los Angeles (UCLA), Los Angeles, CA
³University of Southern California, Los Angeles, CA

The Biomedical REAl-Time Health Evaluation (BREATHE) platform is being developed by the Medical Imaging Informatics Group at the UCLA David Geffen School of Medicine and other project collaborators to provide a comprehensive framework for pediatric asthma that standardizes and organizes sensor data, local and regional environmental data, clinical data, and patient-reported activities and outcomes. The BREATHE platform is being used to develop a contextualized database of observations to support pediatric asthma research and to support smartwatch/smartphone applications that will provide real-time context-based information and alerts to help children avoid asthmatic incidents. An environmental data web service has been developed that provides real-time weather, air quality, and traffic-related air pollution data streams to support the platform. These data are accessible through simple web service calls that will be available to PRISMS-enabled smartphone device as participants conduct their daily activities. The web service returns JavaScript Object Notation (JSON) objects that are parsed and integrated into the BREATHE platform.

A substantial proportion of childhood asthma morbidity is a consequence of near-roadway pollution, and traffic density and near-roadway proximity are strong predictors of local-scale variations in near-road pollution. The centerpiece of the web service is an integrated modeling system based on the RLINE dispersion model that uses real-time local weather data from the Real-time Mesoscale Analysis (RTMA) and a comprehensive database of roadways, annual traffic volumes, and vehicle NOx and PM2.5 emission factors for southern California to estimate ambient pollutant concentrations contributed by on-road mobile source emissions. A unique dispersion model simulation is carried out for each web service request based on estimated emissions from freeways and major roads within 2 km of the participant's location. RLINE results are computed based on a trimmed mean of a receptor matrix centered at the participant's location. Near-road NO2 concentrations are estimated from RLINE NOx predictions based on NO2/NOx ratios derived from an inverse distance weighting of real-time observations. The near-road pollution estimate is combined with regional real-time air quality data from EPA's AirNow system and the South Coast Air Quality Management District to provide a comprehensive snapshot of a participants' local exposure to pollution. After accounting for background concentrations, RLINE overpredicted near-road NO2 concentrations at the four near-road monitors in southern California, particularly in stable and low-wind speeds conditions. This bias was reduced by adjusting the minimum lateral turbulent wind component, which controls the weighting between the plume and meander components in RLINE. The web service also provides real-time weather information at the participant's location based on the spatially-resolved RTMA gridded analyses, as specific weather conditions can also be asthma triggers.

Evaluation of uncertainties in estimating air pollution and health impacts of a large-scale wildfire event

Megan M. Johnson and Fernando Garcia Menendez

Elevated concentrations of fine particulate matter (PM_2.5) from wildfire events can be harmful to human health by causing respiratory problems, exacerbating existing health issues, and even causing premature death. However, there is considerable uncertainty in estimating wildfire-attributable concentration fields and the associated health impacts. We use the 2016 Southern Appalachian wildfires as a case study to evaluate the uncertainties in estimating wildfire-attributable air pollution and health outcomes for the population of North Carolina. These fires burned more than 150,000 acres across five states in the Southeastern US over the course of several months, with the largest PM_2.5 concentrations observed in November. Through multiple spatially- and temporally-resolved concentration fields, this work compares the uncertainty associated with fire-attributable PM_2.5 estimation methods to the uncertainty within and among concentration-response functions from different epidemiological studies. Several data sources and methods are used to explore the range of fire-attributable PM_2.5 concentration fields, including ground-based monitor data (central monitor method, spatial interpolation), Hazard Mapping System (HMS) smoke product, NOAA HYSPLIT smoke forecast product, aerosol optical depth (AOD), satellite imagery and fire activity information from InciWeb. Uncertainty in health impacts due to increased PM_2.5 is quantified in BenMAP-CE using various concentration-response functions for multiple health endpoints. Significant impacts from this fire event were found for some PM_2.5-related endpoints but not for others. By evaluating and contrasting uncertainty associated with representations of exposure to fire-related PM_2.5 and that associated with health effects modeling related to air pollution, this work helps identify the largest research needs to achieve improved estimates of wildfire-attributable impacts on public health.

The contribution of wildland fire emissions to nutrient deposition in the contiguous U.S.: Implications for vulnerable ecosystems

Shannon Koplitz, Chris Nolte, Robert Sabo, Chris Clark

To sustain growth and healthy function, many species and ecosystems require access to certain key nutrients, particularly nitrogen (N) and phosphorus (P). However, excessive deposition of these compounds can also damage ecosystems through acidification and nutrient imbalances, leading to significant reductions in productivity and harmful shifts in fundamental ecosystem structure. While wildland fires are a known source of atmospheric N, P, and other ecologically relevant compounds like sulfur (S), relatively little has been done to examine the implications of wildland fire nutrient deposition for vulnerable ecosystems. In this work we combine wildland fire emission estimates, atmospheric chemistry modeling, and forest inventory data to 1) quantify the contribution of wildland fire emissions to nutrient deposition across the contiguous U.S., and 2) assess the subsequent impacts on tree growth and survival rates. We find that wildland fires contributed an estimated 0.2 kg N/ha and 0.04 kg S/ha annually on average across the contiguous U.S. during the 2008-2012 study period, with maxima up to 1.4 kg N/ha and 0.6 kg S/ha. Preliminary results suggest that impacts of wildland fire nutrient deposition are most pronounced in the Pacific Northwest and Rocky Mountain regions, where wildland fires contributed over 20% of total N and S deposition in some areas. Ongoing work will quantify the impacts of simulated N and S deposition on tree growth and survival rates, as well as the estimated contribution of wildland fires to atmospheric P deposition. Understanding the broader environmental impacts of wildland fires in the U.S. will inform future decision making related to both fire management and ecosystem services conservation.

Environmental benefits from a sustainable BRT system in Colombia

Jorge E. Pachon

Sebastian Montealegre

Johan Vanegas

Beatriz Ortiz

Alejandro Parra

Transportation is the largest source of emissions to the atmosphere in Colombia, both greenhouse gases (GHG) and criteria pollutants. Bogota, Colombia's capital, implemented a Bus Rapid Transit (BRT) system in 2000, and since then the fleet has been growing and expading throughout the city. Nowadays, Transmilenio (TM), Bogota's BRT, comprises 2,000 buses operated with diesel fuel. Although Transmilenio represents an advance in terms of mobility and pollution with respect to the old system, the fleet has not been renovated as it was planned, and currently pollution levels of GHG and criteria pollutans from TM have increased. Additionally, exposure studies have observed large impacts on TM users. The city aims to renovate the old buses and replace them by low or zero emissions buses, i.e. electric fleet or alternative fuels, such as natural gas. In this work, we estimated emission reduction from a sustainable TM system in Bogota and conducted air quality modeling to assess environmental benefits. This approach is of potential interest of cities where sustainable transportation systems are desirable, especially towards a low carbon economy.

Global distribution and trends of ozone relevant to human health from the Tropospheric Ozone Assessment Report

Joseph Pinto, Zoe Fleming, Ruth Doherty, Erica von Schneidermesser, Christopher Malley, Owen Cooper, Augustin Colette, Xiaobin Xu, David Simpson, Martin Schultz, Allen Lefohn, Samara Hamad, Raessa Moolla, Sverre Solberg, Zhaozhong Feng

This study represents the first systematic effort to characterize worldwide surface data for  ozone. Results for several metrics relating to short- and long-term human exposures will be presented. Highest MDA8 values are found in North America, Southern Europe and East Asia and are generally similar. The data show negative trends in MDA8 at most North American and European sites, but positive trends are found at many sites in South Korea and Hong Kong with mixed results In Japan. The data also show that ozone levels on a daily basis still exceed those at which health effects have been observed to occur at several locations. 

Health Impacts and Cost-Benefit Analyses of Surface O3 and PM2.5 over the U.S. under Future Climate and Emission Scenarios

Peilin Yang¹, Kai Wang¹, Yang Zhang¹, Prakash Doraiswamy², and Seung-Hyun Cho²

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

²Air Quality and Exposure Center, RTI International, 3040 E Cornwallis Rd, Durham, NC 27709, USA

Adverse health effects of ozone (O3) and fine particulate matter (PM2.5) have been studied for decades. Most focused on past pollution episodes, fewer studies have been performed to assess their health effects and analyze cost-benefit under future climate and emission conditions. The Environmental Benefits Mapping and Analysis Program - Community Edition (BenMAP-CE) has been recently applied by the authors' group to estimate the health impacts of O3 and PM2.5 in terms of mortality and associated economic values in the U.S. under the Representative Concentration Pathways (RCP) 4.5 and 8.5 scenarios. The concentrations of O3 and PM2.5 used in BenMAP-CE are based on annual and seasonal mean predictions from an advanced online-coupled climate and air quality model, the Weather Research and Forecast Model with Chemistry (WRF/Chem) for current (i.e., 2001-2010) and future (i.e., 2046-2055) decades. Our results showed that the major health and economic benefits from future O3 reduction in O3 season (May - Sept.) occur in California and eastern part of continental U.S. under both RCP 4.5 and 8.5, leading to several thousands of avoided all-cause mortality and several tens of billion U.S. dollars avoided costs on an annual basis. By contrast, in responses to the projected increases in future annual-mean O3 concentrations under RCP8.5, BENMAP-CE estimates several thousands additional all-cause premature mortalities and about twenty billion U.S. dollars additional costs per year. As a result of the reduction in future PM2.5 concentrations in summer and winter under both RCP scenarios, the total annual avoided all-cause deaths are estimated to be about several tens thousands over the entire U.S., with the annual economic benefits around several hundreds billion U.S. dollars. The ischemic heart disease is the main reason causing premature death due to PM2.5 pollution, contributing about 67% to total mortality, followed by lung cancer with ~15%. In this work, the BenMAP-CE modeling of O3 and PM2.5 mortality and cost-benefit will be extended to other seasons and to the cases considering future population growth. The health impact results of both pollutants will be further examined at state and county levels to better understand their spatial distributions. Multiple health functions will be explored using an ensemble modeling approach to provide a range of the health effect estimates. In addition to air pollution-related premature deaths, several morbidity categories such as respiratory hospital admissions, respiratory emergency room visits, and school loss days will be included. The results from this work will have important implications in the development of the integrated emission control strategies to reduce air pollution, improve human health, and maximize economic benefits.

Updates on Using Multi-media Modeling to Investigate Conditions Leading to Harmful Algal Blooms

Christina Feng Chang¹, Catherine Nowakowski², Marina Astitha¹, Valerie Garcia³, Ellen Cooter³, Chunling Tang³, Penny Vlahos⁴, David Wanik¹

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

² University of Rhode Island, RI 02881, USA

³ National Exposure Research Laboratory, Office of Research and Development, US Environmental Protection Agency, Research Triangle Park, NC 27711, USA

⁴ Marine Sciences, University of Connecticut, Groton, CT, USA

Lake Erie is surrounded by seventeen metropolitan areas and provides drinking water to over 11 million people in the United States and Canada. A combination of urban areas, industries, and agricultural activities have contributed to an increase amount of pollutants in the watershed. Particularly, agricultural runoff drains excess nutrients into the lake, such as phosphorus, which has historically led to harmful algal blooms (HABs) events in Lake Erie since the late 1960s. Dramatic implementation of phosphorus reduction controls have led to the recovery of the lake by 1990 but the reemergence of HAB events since the mid-1990s has once again raised concerns, with 2011 being a record-breaking event. In this study, we use in-situ measurements of chlorophyll-a concentrations, dissolved oxygen, total phosphorus and total nitrogen in Lake Erie from 2002-2012 measured by the Lake Erie Committee Forage Task Group (LEC) and the Great Lakes National Program Office (GLNPO) to serve as proxies to identify algal blooms. We demonstrate the relationship between observed response variables and a broad range of environmental model variables (explanatory variables) from linked and coupled physical models. We use meteorological weather variables from the WRF model, hydrological variables from the VIC model, atmospheric air quality variables from the CMAQ model, and agricultural management practices variables from the EPIC model. Through statistical regression techniques and machine learning models, we determine the significance of these simulated variables on response variables that are linked to algal blooms within Lake Erie.

The Understanding of Pollutant Transport in the Atmosphere: Lagrangian Coherent Structures

Peter Nolan, Hosein Foroutan, Shane Ross

Eulerian chemical transport models (e.g., CMAQ) are powerful tools to simulate the transport and fate of air pollutants in various scales from local to global. Informed by meteorological drivers (e.g., WRF), these models calculate the advection of gases and aerosols in the atmosphere and provide information about how pollutants are transported from one location to another. Identifying atmospheric transport pathways and barriers is important to understand pollutants effects on weather, climate, and human health. The atmospheric wind field is variable in space and time and contains complex patterns due to turbulence mixing. In such a highly unsteady flow field, streamlines (Eulerian entities showing the local direction of the velocity at any given instant) do not provide useful information regarding the transport over a finite time. Therefore, particle trajectories are often used to study how pollutants move in the atmosphere. Nevertheless, individual trajectories are sensitive to their initial conditions, cannot fully resolve the complexity of the flow, and often result in a chaotic and tangled field difficult to analyze. Therefore, a more systematic and robust approach is needed. Lagrangian coherent structures (LCSs) are evolving material surfaces with maximum attraction or repulsion rates in a fluid medium. They form the template of transport and mixing phenomena for passively (and some actively) advected chemical species in a fluid. Several such critical material surfaces are present in any given geophysical flow and may aid in understanding Lagrangian transport patterns. In the atmosphere, they persist on a time-scale of a few minutes to a few days, depending on the scale of interest. Here, we present the results of LCS computations for a few atmospheric and air quality cases over various scales. Using Finite-Time Lyapunov Exponents (FTLE), we extract LCSs from WRF-CMAQ and MPAS-A simulations and investigate atmospheric transport of species governed by these structures on a regional as well as global scales.

Modeling Analysis of East Asian Dust Event during the Spring Season over Taiwan: An Implementation of the New Windblown Dust Module into CMAQ

Steven Soon Kai Kong^1,2, Ming-Tung Chuang², Xin-Yi Dong³, Maggie Chel Gee Ooi^1,2, Wei-Syun Huang¹, Neng-Huei Lin^1*

¹Department of Atmospheric Sciences, National Central University, Taoyuan, Taiwan.

²Graduate Institute of Energy Engineering, National Central University, Taoyuan, Taiwan.

³Department of Civil and Environmental Engineering, the University of Tennessee Knoxville, TN37996, USA

East Asian Dust (EAD) is responsible in deteriorated air quality and poor visibility over East Asia (EA) over the past decade. The sources of the mineral dust particle, which originates from the deserts in China and Mongolia, bring adverse impact towards the downwind areas including China, Japan, Korea and Southeast Asia region. The present study analyzes the long-range transport of EAD over EA during the spring seasons 2017 and 2018. The CMAQ(v5.2) model is applied with the implementation of the new windblown dust emission scheme modulated by Foroutan et al. (2017). Overall, the simulation underestimated the particulate matter (PM), probably due to the underestimation of modeled wind speed and overestimation of dry deposition. With the new dust module, the CMAQ simulation shows better performance than without the module incorporation in simulating both PM₁₀ and PM_2.5, with normalized mean bias (NMB) of -32.0% and -25.5% improved from -25.6% and -19.0% respectively. NMB for the new simulation is shown to increase around 6.5% for both PMs as compared to the default simulation. The analysis of a dust storm episode during the period of 6th-8th and 15th-17th April 2018 suggested that the CMAQ model is capable of capturing the dust aerosol concentration. The occurrence of the dust storm can be due to the Mongolian cyclonic which initiated the strong northwesterly wind over deserts regions. As for implication, the accuracy of CMAQ simulation can be enhanced by incorporating the new dust module in projecting the long-range transport of dust particle.

Mitigation of severe urban haze pollution by a precision air pollution control approach

Shaocai Yu^1,2, Pengfei Li¹, Liqiang Wang¹, Yujie Wu¹, Si Wang¹, Kai Liu², Tong Zhu³, Yuanhang Zhang³, Min Hu³, Liming Zeng³, Xiaoye Zhang⁴, Junji Cao⁵, Kiran Alapaty⁶, David C. Wong⁷, Jon Pleim⁷, Rohit Mathur⁷, Daniel Rosenfeld⁸, and John H. Seinfeld²

¹Research Center for Air Pollution and Health; Key Laboratory of Environmental Remediation and Ecological Health, Ministry of Education, College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, Zhejiang 310058, P.R. China.

²Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA 91125, USA

³College of Environmental Sciences and Engineering, Peking University, Beijing 100871, P.R. China

⁴Key Laboratory for Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, 46 Zhong Guan Cun S. Ave., Beijing 100081, China

⁵Key Lab of Aerosol Chemistry and Physics, SKLLQG, Institute of Earth Environment, Chinese Academy of Sciences, Xixan, China

⁶Systems Exposure Division, National Exposure Research Laboratory, U.S. Environmental Protection Agency (EPA), Research Triangle Park, NC 27711, USA.

⁷Computational Exposure Division, National Exposure Research Laboratory, U.S. Environmental Protection Agency (EPA), Research Triangle Park, NC 27711, USA.

⁸Institute of Earth Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel.

Severe and persistent haze pollution involving fine particulate matter (PM2.5) concentrations reaching unprecedentedly high levels across many cities in China poses a serious threat to human health. Although mandatory temporary cessation of most urban and surrounding emission sources is an effective, but costly, short-term measure to abate air pollution, development of long-term crisis response measures remains a challenge, especially for curbing severe urban haze events on a regular basis. Here we introduce and evaluate a novel precision air pollution control approach (PAPCA) to mitigate severe urban haze events. The approach involves combining predictions of high PM2.5 concentrations, with a hybrid trajectory-receptor model and a comprehensive 3-D atmospheric model, to pinpoint the origins of emissions leading to such events and to optimize emission controls. Results of the PAPCA application to five severe haze episodes in major urban areas in China suggest that this strategy has the potential to significantly mitigate severe urban haze by decreasing PM2.5 peak concentrations by more than 60% from above 300 g m-3 to below 100 g m-3, while requiring ~30% to 70% less emission controls as compared to complete emission reductions. The PAPCA strategy has the potential to tackle effectively severe urban haze pollution events with economic efficiency.

Measurement and Modeling of Air Quality for Bangkok, Thailand
Pornpan Uttamang1, Patrick Campbell2, Viney Aneja1 and Adel Hanna1,3

1 Marine Earth and Atmospheric sciences, North Carolina State University
2 National Academies/National Research Council Fellowship Participant at National Exposure Research Laboratory, U.S. Environmental Protection Agency
3 Institute for the Environment, University of North Carolina, Chapel Hill

Thailand has experienced rapid industrialization and urbanization in the past three decades which has led to an impact on urban air quality. 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, BMR has begun to experience air quality degradation, in particular, enhanced ozone (O3) owing to elevated ozone precursor emissions, strong solar radiation (peak density of direct radiation ~1,350 kWh m-2 yr-1), high temperature (yearly average ~29 C), and high humidity (yearly average ~64%). 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), origin of the air masses, and local and regional contributions of Ox (O3 + NO2). Interconversion between O3, NO and NO2 indicates crossover points between the species occur when [NOx] (NO + NO2) is ~60 ppb. O3 dominates when [NOx] < 60 ppb, but NO dominates when [NOx] > 60 ppb. To better quantify the local and regional contributions to increased [O3] in 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 local vs. regional processes that contribute to elevated O3 levels in BMR, in particular, the role of precursor transport from China on O3 over 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 double-nested domain (36- and 12-km horizontal resolutions), where the outer domain covers a majority of China and Thailand, and the inner domain is focused on Thailand and the BMR region specifically. Simulations including a baseline and a sensitivity (10% NOx emission reduction over China) simulations are performed. The meteorological ICONs/BCONs are from NCEP FNL and anthropogenic emissions are based on EDGAR-HTAP.

An Exceptional Event Screening Tool to Estimate Natural Impacts on Ozone Exceedances

Chantelle R. Lonsdale, Matt Alvarado, Richard Pernak, John Henderson

The EPA has put forth an Exceptional Events Rule that provides a mechanism by which air quality data can be excluded from regulatory decisions if an exceptional event causes an exceedance of the NAAQS ozone standard. These exceptional events can be caused by natural phenomena such as wildfires and stratospheric intrusions. The current EPA guidance for an exceptional event demonstration requires detailed modeling and analysis work to be done for each event, but states, such as Texas, have limited resources to perform extensive investigations of every ozone exceedance. We have developed an exceptional events screening tool that efficiently determines which ozone exceedance days were significantly impacted by stratospheric ozone intrusions upwind of the reporting monitor. This allows states to identify which events should be further investigated with a full photochemical model study. The tool includes a back-trajectory component, based on the Stochastic Time Inverted Lagrangian Transport (STILT) model, that determines the multiple source regions of air transported to the reporting monitor. A second component uses satellite observations of ozone to estimate the impact of a stratospheric intrusion at the monitor. The tool provides a metric that indicates the likelihood that an exceptional event has occurred, thus allowing air quality managers to allocate modeling resources more efficiently and ultimately help in determining the contribution of natural sources to local ozone pollution. In this presentation, we will demonstrate the capabilities of the tool and present results for case studies in Texas.

Multi-scale modeling of a short term NO2 air quality action plan in Madrid (Spain)

Rafael Borge¹, Jose Luis Santiago², David de la Paz¹, Fernando Martin², Jessica Domingo¹, Cristina Valdes³, Mi Teresa Rozas³, Beatriz Sanchez², Sonia Lazaro¹, Javier Perez¹, Esther Rivas², Christina Quaassdorff¹, Alvaro Fernendez³, Juan Manuel de Andres¹, Carlos Yagce⁴, Julio Lumbreras¹

1 Laboratory of Environmental Modelling. Chemical and Environmental Engineering Department. Universidad Politecnica de Madrid (UPM)
2 Department of Environment, CIEMAT, Madrid
3 Madrid Municipal Transport Company (EMT)
4 Department of Earth Physics and Astrophysics, University Complutense of Madrid, Faculty of Physical Sciences

In this contribution we present a multi-scale modeling exercise to understand the effect of the application of a short term air quality in Madrid (Spain) during a high pollution episode that took place in December 2016. The stage 3 of the so-called NO₂ protocol, that involves drastic limitations on road traffic, was trigger for the first time during this episode. We assessed the effect of the measures taken by comparing the actual scenario with a reference one that reflects a hypothetical no-action situation. We relied on the WRF-SMOKE-CMAQ modeling system to assess the impact of the protocol at city scale (1 km² resolution). In addition we applied the Star-CCM+ (RANS CFD model) to investigate the effect at street level in a microscale domain in the city center. Despite several limitations in our methodology, we obtained rather consistent results from both simulations that are also in agreement with the main conclusions derived from the analysis of observational data. We believe that this study is relevant to improve short term action plans in urban areas and to optimize our strategies to minimize population exposure.

Simulation study for influences of consumption in major countries on air quality and human health in Asia through global supply chains

Satoru Chatani, Midori Kurogi, Yuta Fujii, Susumu Tohno, Keiichiro Kanemoto, and Keisuke Nansai

Economic activities such as productions of goods and services cause emissions of ambient pollutants. They are induced by consumption not only within one's own country, but also in any other countries which could influence the economic activities in one's country through global supply chains. This issue is particularly important in Asia. Developing countries in Asia are serving as the world's factories to export various products to countries in all over the world. On the other hand, they are suffering from heavy air pollution. Therefore, heavy air pollution in Asia can be partially attributed to consumption in other countries.

We conducted a simulation study to investigate such a relationship through global supply chains. The global anthropogenic emission inventory, EDGAR version 4.3.1, was combined with the global supply chain database, EORA, which consists of a multi-region input-output table, to derive anthropogenic pollutant emissions which are induced by consumption in each country. CMAQ was applied to evaluate contributions of emissions induced by consumption in United States, Japan, Germany, United Kingdom, China, and Rest Of the World (ROW) to ambient PM2.5 concentrations and premature deaths caused by PM2.5 in Asia with the integrated exposure response model for 5 major diseases.

Whereas significant fractions of PM2.5 concentrations and premature deaths were attributed to emissions induced by consumption in China (890,000 deaths), it was estimated that consumption-based emissions of United States, Japan, Germany, United Kingdom, and ROW induced 64,000, 38,000, 14,000, 13,000 and 1,090,000 of premature deaths in Asia in 2010, respectively. The results suggested that imported premature mortality by Japan from China mainly through transboundary transport is comparable to exported one mainly through the international trade in terms of bilateral relations. It is important to understand such a relationship for sustainable development in Asia to be realized by all the countries concerned.

Using air quality model to support Canadian policy making: two applied cases
Patrick M. Manseau, Jacinthe Racine, Mourad Sassi, Annie Duhamel, Sophie Cousineau, David Niemi, Rabab Mashayekhi, Jessica Miville, Cassandra Bolduc, Radenko Pavlovic
Air Quality Policy-Issue Response Section (REQA), Meteorological Service of Canada, Environment and Climate Change Canada, 2121 Trans-Canada Highway, Dorval, Quebec, H9P 1J3, Canada

The main mandate of the Environment and Climate Change Canada (ECCC) Air Quality Policy-Issue Response Section (REQA) is to support the development of Canadian regulations and standards related to different air pollutants. Air quality model scenarios analysis is the most common tool used to estimate the impact of emission changes on atmospheric pollutant concentrations. In partnership with other groups from ECCC and Health Canada (HC), air quality modelling results are also used as inputs to produce cost-benefit analyses.

This poster will give an overview of the modelling platform and methodologies used by REQA for its air quality scenario analyses. Two applied cases will be presented to demonstrate the usefulness of these analyses.

The first case is the determination of the Canadian Ambient Air Quality Standards (CAAQS) for nitrogen dioxide (NO2). CAAQS are established as the objectives for the Air Quality Management System, under the Canadian Environmental Protection Act of 1999. Achieving the CAAQS will drive continuous improvements in air quality across Canada.

The second case is the Amendments to the Reduction of Carbon Dioxide Emissions from Coal-fired Generation of Electricity Regulations. Coal-fired electricity generating units are the highest emitting stationary sources of GHGs and air pollutants in Canada. The Amendments require all coal-fired electricity generating units to comply with an emissions performance standard of 420 tonnes of carbon dioxide per gigawatt hour of electricity produced by 2030. Of the 36 units operating in 2017, 20 are expected to shut down before 2030, as they will reach their end of life under the Regulations.

Inter-state Transport Modeling for the 2015 Ozone Standard in the Midwest and the Northeast

Tsengel Nergui, Zac Adelman, Mark Janssen, and Donna Kenski

The Lake Michigan Air Directors Consortium

In support of "Good Neighbor" State Implementation Plans for meeting the 2015 ozone (O3) National Ambient Air Quality Standard (NAAQS), the Lake Michigan Air Directors Consortium (LADCO) simulated the Comprehensive Air Quality Model with Extensions (CAMx version 6.4) model to assess the impacts of interstate transport on O3 concentrations in the Midwest and Northeast U.S. Three aspects of this modeling work will be discussed: 1) the effects of different projections for electricity generating units (EGU) emissions; 2) state contributions to maintenance and nonattainment areas in the region; and 3) the impacts of model performance for future design value (FDV) calculation. LADCO's modeling assessment suggests that if EGU emissions reductions take place as projected by the Eastern Regional Technical Advisory Committee EGU Forecasting Tool (ERTAC version 2.7) along with the EPA's projections for other emissions sources, the Midwest and the Northeast could attain the 2015 O3 NAAQS by 2023 under meteorological condition of 2011. A few sites by the Lake Michigan and Connecticut shorelines are projected to be at or near maintenance status of the standard. However, we believe that estimated future DVs in the study region to be biased low. Some reasons for this underestimated FDVs will be discussed. Excluding water cells for attainment test calculation resulted in higher FDVs for the lakeshore monitors in the LADCO region, but lower FDVs in Connecticut shorelines. This finding indicates stagnation or recirculation of polluted air play important role in air quality status near lakeshores than in ocean shores when assuming both the base year emissions and its projected reductions in these regions were in similar range. Initial results from a test using bias-filtered model outputs for FDV calculation will be presented.

Modeling and Data Analysis to Understand the Causes of High Ozone Events in Atlanta, Georgia

Tao Zeng, Byeong Kim, Marcus Trail, Di Tian, James Boylan

The metro Atlanta area is currently violating the 2015 ozone National Ambient Air Quality Standard (70 ppb). Elevated ozone concentrations have large spatial variations among the ground ozone monitors in the metro Atlanta area. To guide future air quality management practices to improve air quality, various types of modeling and data analysis were performed to better understand the causes of ozone exceedances in the metro Atlanta area in 2016 and 2017. Multi-linear regression analysis and Classification and Regression Tree Analysis (CART) analysis were used to understand the relationship between ozone and environmental variables. Low relative humidity, high temperature, and low cloud coverage are three major factors contributing to the ozone exceedances in Atlanta. Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) back trajectory analysis was used to determine the origin of air masses and establish source-receptor relationships on ozone exceedance days. Animations of 5-minute ozone observations and 4-km Weather Research and Forecasting (WRF) model wind vectors were used to better understand ozone formation and transport. The transport of ozone and precursor emissions from the Atlanta urban core has large impacts on the downwind ozone monitors. Finally, 2005-2016 National Aeronautics and Space Administration (NASA) Ozone Monitoring Instrument (OMI) satellite observations for NO₂ and HCHO and 2016-2017 ground observations for NOx and VOC species were analyzed to better understand the impacts of ozone precursors.

Web-based Interactive Tools to Evaluate Air Quality Modeling Source Apportionment Results

Weining Zhao and Doug Boyer

The Texas Commission on Environmental Quality has developed interactive and intuitive web-based model performance evaluation tools that help visualize and assess the air quality model's source apportionment results. Interactive charts and maps from the TCEQ's 2012 CAMx modeling platform will be presented through a live demonstration. Interactive features include organizing source contribution elements via drag-and-drop of web objects. The client and server-side programming technologies used to develop these tools will also be discussed.

Photochemical model estimated fire impacts on O3 and PM2.5 evaluated with field studies and routine data sources

Kirk R Baker, Luxi Zhou, Matt Woody

Highly instrumented field studies provide a unique opportunity to evaluate multiple aspects of photochemical grid model representation of fire emissions, dispersion, and chemical evolution. Fuel information and burn area for a specific fire coupled with near-fire and downwind chemical measurements provides information needed to constrain model predicted fire plume transport and chemical evolution of important pollutants such as particulate matter (PM2.5) that have deleterious health effects. Most local to regional scale field campaigns to date have made relatively few transects through plumes from fires with well characterized fuel type and consumption. While more comprehensive field studies are being planned for 2018 and beyond (WE-CAN and FIREX), existing measurement data from multiple field campaigns including 2013 SEAC4RS, satellite data, and routine surface networks are used to assess how a regulatory modeling system captures fire impacts on local to regional scale O3 and PM2.5. The Community Multiscale Air Quality (CMAQ) model does well at capturing local (Zhou et al., 2018) to regional (Baker et al., 2018) scale plume height and transport when provided realistic inputs about fire size and meteorology. The modeling system systematically over-estimates surface level O3 from wildland fire (Baker et al., 2016; Baker et al., 2018). A comparison model estimated speciated PM2.5 from specific fires with routine PM2.5 organic carbon surface measurements at rural locations in proximity to the 2013 Rim fire, 2011 Wallow fire, and 2011 Flint Hills fires show overprediction downwind from the 2011 Flint Hills prescribed fires while results were mixed at sites downwind of the 2013 Rim fire and 2011 Wallow fire (Baker et al., 2016; Baker et al., 2018). These results suggest differences in fuel characterization (e.g., emission factors, emissions speciation, burn period, etc.) between these areas may contribute to differences in PM2.5 model prediction.

Baker, K., Woody, M., Tonnesen, G., Hutzell, W., Pye, H., Beaver, M., Pouliot, G., Pierce, T., 2016. Contribution of regional-scale fire events to ozone and PM 2.5 air quality estimated by photochemical modeling approaches. Atmospheric Environment 140, 539-554.

Baker, K., Woody, M., Valin, L., Szykman, J., Yates, E., Iraci, L., Choi, H., Soja, A., Koplitz, S., Zhou, L., 2018. Photochemical model evaluation of 2013 California wild fire air quality impacts using surface, aircraft, and satellite data. Science of The Total Environment 637, 1137-1149.

Zhou, L., Baker, K.R., Napelenok, S.L., Pouliot, G., Elleman, R., O'Neill, S.M., Urbanski, S.P., Wong, D.C., 2018. Modeling crop residue burning experiments to evaluate smoke emissions and plume transport. Science of The Total Environment 627, 523-533.

Estimation of PM2.5 Exposures for an Environmental Justice Community in Denver, CO, Using Various Assessment Approaches

Seung-Hyun Cho1, Lisa Cicutto2, Molly McCullough2, Krysten Crews2, Michelle McCombs1, Nalyn Siripanichgon1, Yang Zhang3, Prakash Doraiswamy1 1. RTI International 2. National Jewish Health 3. North Carolina State University

Methods using surrogate measurements such as central site ambient air quality data are widely used because of impracticality of long-term personal monitoring for epidemiological studies, leading to potential measurement errors. Ambient air monitoring using a low-cost sensor network provides high-resolution exposure estimates for a population affected by multiple sources within their community. Citizen scientists in Globeville, Elyria, and Swansea (GES), a community resides with major highways and industrial sources in north of Denver, conducted air quality measurement pilot studies using a sensor network with the support of researchers. RTI MicroPEM sensors collected real-time PM2.5 data and filter samples within a 16km2 neighborhood scale sensor network, consisting of 17 sites in their community for 3 weeks in August 2017. Those sites included three state-run monitoring sites close to the highways where MicroPEMs were collocated with Grimm reference monitors. MicroPEM 1 hour averaged real-time measurements showed moderate linearity with Grimm reference monitor (R2 >0.6). Measurements were negatively drifted during daytime with high temperatures that reached up to 90F, caused by the optical electronics' temperature sensitivity. After correcting the data for the temperature drift and using filter-based concentration, the linearity improved by 20%. Time-integrated ambient PM2.5 data showed significant spatial variabilities among 17 sites with up to 2.8 times different PM concentrations. Higher PM levels were measured at a couple of sites compared to the reference sites. Overall high PM2.5 temporal correlations were observed among 17 sites (Pearson r=0.67-0.96) with the upwind site at the south of network showing the lowest correlation. PM levels measured by both MicroPEM and Grimm at the reference sites showed relatively small spatial variabilities (2-16%) with very high correlations (r=0.85-0.90). The results show that a sensor network can provide data to better assess the spatial and temporal variability of air pollution for the community of multiple sources. This will help improve the health effects estimate, which could have been severely biased by the exposure metric derived from a central site. This work will be extended to examine the exposure metric from local-scale air quality modeling data of WRF-Chem, and to assess the difference in prediction outcomes of EPA's APEX model using various air quality metrics. EPA Grant Number: RD83618701

Variations in Wintertime PM Among Communities in Sacramento Measured with a Combination of Traditional and Low-Cost Sensor Methods
Steven G. Brown, Nathan Pavlovic, Michael C. McCarthy, Sonoma Technology, Inc., Petaluma, CA
Aleta Kennard, Janice Lam Snyder, Stephen D'Andrea, Sacramento Metropolitan Air Quality Management District, Sacramento, CC

Residential wood smoke is one of the largest contributors to wintertime particulate matter (PM) in Sacramento, CA. To understand how wood smoke and PM varied across Sacramento and between environmental justice (EJ) and non-EJ communities, we conducted measurements during December 2016 and January 2017 of black carbon (BC) with Aethalometers (AE33) at six locations, of PM with low cost AirBeam sensors at 15 locations, of hourly PM with beta attenuation monitors (BAMs) and 24 hour PM via filter measurements at two locations, and of levoglucosan via filter measurements at three locations. Before and after the main study period, 20 AirBeam sensors were collocated with a BAM and a filter PM instrument to determine calibration factors for the sensors. In addition, the AirBeam sensors were collocated with a BAM and a filter PM instrument at two locations throughout the study to assess whether there was drift in the sensor measurements and to determine the comparability of PM measurements among the sensors, BAM, and filter instrument.

The 20 AirBeam sensors had extremely high precision during the pre- and post-study collocations, with AirBeam-to-AirBeam correlation coefficients (r2) greater than 0.95. AirBeams also had very high precision with very little drift at the two locations with collocated AirBeams throughout the study. Since each AirBeam had a consistent PM concentration response relative to the other collocated AirBeams, we developed an AirBeam-specific calibration based on the collocated data, and used the calibrated AirBeam data to assess with high confidence how PM varied across communities. We also determined the extent to which AirBeam data can be corrected to BAM or 24-hr filter measurements using meteorological data.

A novel approach to improving the accuracy of ground-level PM2.5 concentration estimates by application of High Spectral Resolution Lidar

Xinyi Ling¹, Nicholas Meskhidze¹, Kyle Dawson², Matthew S. Johnson³, Barron Henderson⁴, Sharon P. Burton⁵, Chris A. Hostetler⁵, Richard A. Ferrare⁵

¹North Carolina State University, Raleigh, NC; ^2,5Science Systems and Applications, Inc., Hampton VA; ³NASA Ames Research Center, Moffett Field, CA; ⁴US EPA; ⁵NASA Langley Research Center, Hampton, VA

Atmospheric aerosols are ubiquitous and have negative impacts on human health and air quality. Epidemiological studies have correlated atmospheric particulate matter with aerodynamic diameter less than 2.5 m (PM2.5) to the increase of both mortality and adverse human health effects, such as respiratory and cardiovascular diseases. Despite its importance, PM2.5 monitoring stations are sparsely distributed in many areas, particularly in the developing nations. Therefore, it is essential to develop algorithms that can link surface air quality with satellite remote sensing. Aerosol Optical Depth (AOD), an optical measure of the column-integrated aerosol abundance in the atmosphere, has been widely used to connect surface air quality to satellite retrievals. For example, model assimilation is a way of correcting model simulations by adjusting the model to match remotely sensed AOD. However, AOD is based on optical information about the aerosols while the models typically simulate aerosol chemical composition. The AOD-surface PM_2.5 relationship is further complicated by the strong dependence of AOD on relative humidity and the assumption that the majority of the aerosol mass is located near the surface. Moreover, AOD depends on aerosol size distribution and chemical composition; therefore, when model-predicted aerosol mass is scaled to match the satellite AOD, it is implicitly assumed that the model correctly calculates the relative contribution of each aerosol chemical component to the total aerosol mass.

Here we present a novel approach in which we adjust the model chemical composition according to the remote sensing observations, rather than just the total column mass. Remote sensing inferences of aerosol types (i.e., urban, smoke, fresh smoke, dusty mix, and maritime) are used to predict the relative chemical composition (the "makeup") of the particles. The remote sensing component of this study is from High Spectral Resolution Lidar (HSRL) aerosol types retrieved during Discover-AQ (Deriving Information on Surface conditions from COlumn and VERtically Resolved Observations Relevant to Air Quality) Baltimore-Washington Corridor (BWC) campaign. The modeling component is the EPA's CMAQ (Community Multiscale Air Quality) Model. Aerosol types are simulated using the CATCH (Creating Aerosol Types from CHemistry) algorithm developed by Dawson et al. (2017). CATCH calculates a mass fraction of each chemical component for different aerosol types. By combing the CATCH algorithm with the aerosol types provided by HSRL, the relative contribution of aerosol chemical components (i.e., sulfate/nitrate/ammonia, black carbon, organic carbon, sea salt, and mineral dust) can be inferred in every model grid. As CATCH-derived information on aerosol chemical composition is independent of the CMAQ model predictions, we then iteratively adjust the modeled aerosol vertical extinction profiles to match the remotely sensed ones by varying both: the total model-predicted aerosol mass concentration and the aerosol chemical composition. The reliability of this approach is examined by comparing a priori (default) and a posteriori (corrected) model simulation with ground-level measurements of PM_2.5 chemical composition near the study region.

Spatiotemporal estimates of surface PM2.5 concentrations in the western U.S. using NASA MODIS retrievals and data assimilation techniques

S. Marcela Lorea-Salazar1,

Cesunica E. Ivey2,

Heather A. Holmes1,

Howard H. Chang3

1Atmospheric Sciences Program, Department of Physics, University of Nevada Reno, Reno, Nevada, U.S.A.

2Chemical and Environmental Engineering, University of California Riverside, Riverside, California, U.S.A.

3Department of Biostatistics and Bioinformatics, Emory University, Atlanta, Georgia, U.S.A.

Previous investigations have used satellite remote sensing to estimate surface air pollution concentrations. Models developed for the dark-vegetated eastern U.S. are often being used in the semi-arid western U.S. without modifications. These models are not robust in the western U.S. due to specific regional characteristics including 1) irregular topography that leads to complicated boundary layer physics; 2) pollutant mixtures, 3) more heterogeneous vertical profile of aerosol concentrations; and 4) higher surface reflectance. This work uses a data fusion model to estimate surface PM2.5 in the western U.S. with pollutant emissions covariates to account for the physical processes in the atmosphere that impact the complex relationship between AOD and PM2.5. This investigation uses the new collection (6.1) of AOD satellite retrievals from the Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) instruments as a spatial predictor of PM2.5. The data fusion model incorporates high-resolution simulated weather variables from the Weather Research and Forecasting (WRF) model, emissions inventory data from the National Emissions Inventory (NEI), elevation, population density, land cover information, surface reflectance and fire radiative power from MODIS, and plume injection height from the Multi-Angle Implementation of Atmospheric Correction (MAIAC) as covariates that can vary in space and time. Based on our previous research, the covariates were selected to account for complex atmospheric physics and meteorological phenomena that govern the aerosol transport in mountainous regions. It is expected that by selecting high-resolution, temporally resolved physics and emissions variables as covariates, the surface PM2.5 concentration estimates in the data fusion model will be improved for the western U.S. from 2012-2013.

Referecnes 1. Lee, K. & Chung, C. E. Observationally-constrained estimates of global fine-mode AOD. Atmos Chem Phys 13, 2907-2921 (2013).

Application of OMI NO2 retrievals to the evaluation of NOx emissions from on-road mobile sources in the Great Lakes Region

Momei Qin, Yongtao Hu, M. Talat Odman

Having a reliable nitrogen oxides (NO_x, i.e., NO and NO₂) emission inventory is very important because NO₂ is a criteria air pollutant and NO_x participate in the formation of O₃ and particulate matter. The uncertainties in the 2011 US National Emission Inventory (NEI) have been extensively studied, and the on-road mobile sector was reported as a major source of uncertainty, possibly leading to an overestimation of NO_x emissions. In this work, satellite retrievals of NO₂, more specifically Ozone Monitor Instrument (OMI) NO₂, were used to evaluate NO_x emissions over the Great Lakes Region. Slant column NO₂ measurements were converted to vertical column NO₂ using vertical NO₂ profiles adopted from the Community Multiscale Air Quality (CMAQ) model. Then, these recalculated vertical OMI NO₂ columns were compared to the NO₂ columns simulated by CMAQ. It was found that CMAQ tended to overestimate NO₂ column in rural areas while underestimation occurred in urban areas. A preliminary simulation showed that discrepancies between the simulated NO₂ column and satellite retrievals were reconciled with 50% reduction of NO_x emissions from on-road mobile sources. In this presentation, we will present results from a further model-satellite comparison conducted in locations where on-road mobile sources contributed dominantly to NO₂ concentration, as identified by the Decoupled Direct Method (DDM) sensitivity analysis. Ground-based measurements of NO₂ will also be used along with the remote measurements to evaluate NO_x emissions in the NEI, particularly those from mobile sources throughout the Great Lakes Region.

Effects of Total Nitrogen Deposition on the North Pacific and Northwestern Atlantic Ocean

Homaira Sharif, Valerie Garcia, Rohit Mathur

Atmospheric nitrogen deposition is predicted to have measurable impacts on the primary production of biomass in the northwestern Atlantic Ocean and the north Pacific Ocean. Both ocean regions are located downwind of high population densities with elevated sources of airborne nitrogen from agricultural and anthropogenic activities. The rapidly growing human population and modifications of the physical environment -- particularly in east Asia -- have led to increased emissions of nitrogen species. This research uses chlorophyll-a concentrations gathered from Moderate Resolution Imaging Spectroradiometer (MODIS) imagery as a proxy for marine biomass and the Community Multiscale Air Quality (CMAQ) modeling system to examine the impacts of atmospheric nitrogen deposition on biomass production. Simulations with season-specific emissions of total nitrogen deposition were conducted for the years 2003-2010. Linear regression is used to evaluate the relationship between total nitrogen deposition and biomass production. Improved knowledge of this comparison between the two study regions will provide a better understanding of oceanic ecosystems and how they respond to global environmental stressors.

A simple and low-cost mobile sensor system for measuring spatial concentration gradient of PM2.5

Clayton White, Zachary Loeb, Xiaonan Yu, Haofei Yu

Ambient concentrations of air pollutants are known to vary considerably over short distance. Appropriately characterizing such spatial variability are important for exposure assessment and air quality management purposes. However, most instruments designed for mobile sampling are usually large, expensive, and require modifications to carrying vehicles, which restricted their large scale implementations. Here, we demonstrated a simple mobile sensor system based on low-cost optical particle sensors to measure spatial concentration gradient of PM2.5. The system is low-power consumption, small in size and can be easily attached to carrying vehicle without physical alternation to the vehicle. It is also designed to operate even in rainy conditions. The sensor system was tested along a fixed route and preliminary results suggest that the sensor system is able to reasonably capture spatial PM2.5 concentration variations along the test route. Due to its low-cost, the developed mobile sensor system can be utilized in large-scale deployments to characterize spatial concentration variations of PM2.5 in large areas.