Technical discussions on Emissions and Atmospheric Modeling (TEAM)

Gregory Frost (NOAA), Barron Henderson (EPA), Barry Lefer (NASA)

A new informal activity, Technical discussions on Emissions and Atmospheric Modeling (TEAM), aims to improve the scientific understanding of emissions and atmospheric processes by leveraging resources through coordination, communication and collaboration between scientists in the Nation's environmental agencies. TEAM seeks to close information gaps that may be limiting emission inventory development and atmospheric modeling and to help identify related research areas that could benefit from additional coordination efforts. TEAM is designed around webinars and in-person meetings on particular topics that are intended to facilitate active and sustained informal communications between technical staff at different agencies. The first series of TEAM webinars focuses on emissions of nitrogen oxides, a criteria pollutant impacting human and ecosystem health and a key precursor of ozone and particulate matter.

Retrospective analysis of the air quality health benefits of phasing out coal-fired power plants in Ontario

Yasar Burak Oztaner, Marjan Soltanzadeh, Shunliu Zhao, Amanda Pappin, 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 (USEPA); Rob W. Pinder; 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); Daewon Byun.

Based on several studies which have examined the health and environmental impacts of Coal-Fired Power Plants (CFPPs) in the US and Canada, the Ontario government established a policy to phase out its CFPPs by the end of 2013. This incremental phase out began in 2005 and the required energy has been replaced with natural gas and renewable energy sources. We aim to carry out a retrospective analysis of Ontario's decision to phase out coal.

Studies conducted previously investigating the CFPP phase-out were scenario-based and suggested reduction of pollutant concentrations and mortality rates. In this study, we present a backward/adjoint analysis to provide source-specific analysis to quantify the marginal health benefits of phasing out these power plants in Ontario for the years 2005 and 2011, while considering the impact of trans-border emission from the US. To this end, we also compare the cost/benefit ratio of the coal phase-out with the other policy options for improving air quality.

We apply U.S. EPA's (CMAQ) and its adjoint to quantify the health benefits of emission reduction of NOX, and PM2.5. Meteorological inputs are from the Weather Research and Forecasting (WRF) model, and emissions for Canada and the US are taken from National Pollutant Release Inventory (NPRI) and National Emission Inventory (NEI) for the years 2005 and 2011, respectively. Subsequently, these emissions are processed in (SMOKE) model to get hourly emissions. The simulation is carried out at a 12 km resolution for these years. Marginal benefits due to the reduced NOx (i.e., the impact on NO2, as well as an O3 and PM2.5) and the subsequent impact on long-term mortality are calculated and compared to those found previously in scenario-based studies. Our preliminary results show that in 2011, health benefits for specific plants in Ontario were as high as 3.5 $/KWh, i.e. greater than the price of electricity, and more significant than previous estimates. We expect to see higher total damage estimations in 2005 when emissions were higher. The findings of marginal health benefit analysis and an evaluation matrix for control policy options for Ontario will be discussed.

Evaluation of NOx Emissions and Modeling

Barron Henderson; Heather Simon; Brian Timin; Pat Dolwick; Chris Owen; Alison Eyth; Kristen Foley; Claudia Toro; Kirk Baker

Studies focusing on ambient measurements of NOy have concluded that NOx emissions are overestimated and some have attributed the error to the onroad mobile sector. We investigate this conclusion to identify the cause of observed bias.

First, we compare DISCOVER-AQ Baltimore ambient measurements to fine-scale modeling with NOy tagged by sector. Sector-based relationships with bias are present, but these are sensitive to simulated vertical mixing. This is evident both in sensitivity to mixing parameterization and the seasonal patterns of bias. We also evaluate observation-based indicators, like CO:NOy ratios, that are commonly used to diagnose emissions inventories.

Second, we examine the sensitivity of predicted NOx and NOy to temporal allocation of emissions. We investigate alternative temporal allocations for EGUs without CEMS, on-road mobile, and several non-road categories. These results show some location-specific sensitivity and will lead to some improved temporal allocations.

Third, near-road studies have inherently fewer confounding variables, and have been examined for more direct evaluation of emissions and dispersion models. From 2008-2011, the EPA and FHWA conducted near-road studies in Las Vegas and Detroit. These measurements are used to more directly evaluate the emissions and dispersion using site-specific traffic data. In addition, the site-specific emissions are being compared to the emissions used in larger-scale photochemical modeling to identify key discrepancies.

These efforts are part of a larger coordinated effort by EPA scientist to ensure the highest quality in emissions and model processes. We look forward to sharing the state of these analyses and expected updates.

Developing Strategies to Attain Air Quality Standards in the South Coast Air Basin of California and Evaluating Precursor Emission Trading Ratios for PM2.5 Formation

Marc Carreras-Sospedra, Scott Epstein, Xinqiu Zhang, Sang-Mi Lee

The South Coast Air Basin of California is in "Serious" Non-Attainment status for the 24-hour and annual PM_2.5 standards, and in "Extreme" Non-Attainment status for ozone. The recently approved 2016 Air Quality Management Plan (AQMP) sets the path to reductions in smog precursor emissions, in order to attain the National Ambient Air Quality Standards by the assigned deadlines. The AQMP relies heavily on nitrogen oxides (NO_X) emission reductions in order to attain the ozone goals, and as a co-benefit, the NO_X reductions are estimated to lead to the attainment of the PM_2.5 standards. This work analyzes the contribution of criteria pollutant emissions to the formation of PM_2.5, and the sensitivity of PM_2.5 design values to changes in PM_2.5 precursor emissions. We conducted a series of annual air quality simulations for Southern California using CMAQ, with parametric changes in emissions of NO_X, SO_X, VOC, NH₃, and direct PM. Results allow us to determine location-specific emission trading ratios for PM precursors. We conducted these series of simulations for the years 2012 and 2025, to determine the sensitivity of trading ratios to baseline emissions. In addition, we used CMAQ-DDM to further analyze the model sensitivity with respect to PM formation.

Reconciling modeled and observed upper tropospheric NO2 for the interpretation of satellite measurements

Rachel F. Silvern, Daniel J. Jacob, Katherine R. Travis, Eloise A. Marais, and Ronald C. Cohen

Satellite observations of NO₂ have been used as a top-down constraint to document and quantify the reduction in US NO_x emissions over the past decade. Recent observations from the SEAC⁴RS aircraft campaign in August-September 2013 over the Southeast US showed much lower NO/NO₂ ratios than predicted by the GEOS-Chem chemical transport model with implications for interpreting satellite observations of NO₂. We show that this discrepancy in modeled NO/NO₂ partitioning can largely be accounted for by an underestimate of the NO + O₃ reaction rate constant (k₁) at low temperatures and the remaining model overestimate can be attributed to insufficient peroxy radicals and halogens in the free and upper troposphere. Lowering the activation energy of k₁ at low temperatures improves the modeled NO/NO₂ ratio in upper troposphere to 2.1 0.7 mol mol^-1 compared to 1.5 1.3 mol mol^-1 observed. Increasing k₁ has implications for global tropospheric chemistry by decreasing the global mean tropospheric OH concentration by 6% and the global mean tropospheric O₃ burden by 5% for 2013. Using the revised NO₂ profiles from GEOS-Chem we can largely correct GEOS-Chem for its low bias in the upper troposphere and find the upper troposphere above 8km contributes 35% to the total column. Correcting for the bias in the upper troposphere will allow satellite observations of NO₂ to be used to diagnose uncertainties in surface NO_x emissions including uncertainties in NEI anthropogenic emissions in the Southeast US and soil NO_x emissions in the South-central US.

Hybrid Plume/Grid Modeling for the Allegheny County, Pennsylvania Annual PM2.5 State Implementation Plan

Emery, C., M. Zatko, B. Brashers, R. Morris, Ramboll Environ
J. Maranche, T. Sadar, Allegheny County, PA

Allegheny County is designated as a nonattainment area for the 2012 annual PM_2.5 National Ambient Air Quality Standard (NAAQS), with a 2012-2014 Design Value of 13.0 g/m³ at the South Allegheny High School (Liberty) monitoring site. This site resides in complex river-valley terrain that is approximately 3 miles wide and 5 miles long. PM_2.5 exceedances at the Liberty monitoring site are caused by a combination of long range transport and local sources, including both primary and secondary PM_2.5. Addressing these complexities for the Allegheny County (AC) State Implementation Plan (SIP) requires a flexible modeling system that can simultaneously resolve dispersion and chemistry on scales ranging from plume (~10-100 m) to local (~1-10 km) to regional (100+ km). Additionally, an established and well-vetted source apportionment capability is needed to assess contributions from local vs. regional sources. Here we describe a regulatory application of the Comprehensive Air quality Model with extensions (CAMx) to support the AC PM_2.5 SIP. CAMx was run for the year 2011 on a nested 4-grid domain: a continental US grid at 36 km resolution, a northeast US grid at 12 km resolution, a Pennsylvania grid at 4 km resolution, and an Allegheny County grid at 1.33 km resolution. Meteorological modeling for 2011 was conducted using the Weather Research and Forecasting (WRF) model on similar grids. The 2011 National Emissions Inventory (NEI) was processed to generate gridded, hourly, speciated emissions. Local emission sources were updated and point sources were modeled using the CAMx Plume-in-Grid (PiG) model. CAMx was run for the 2011 base year and 2021 future year with CONUS boundary conditions derived from the GEOS-Chem global chemistry model. A model performance evaluation was conducted with particular focus on PM_2.5 performance at the Liberty monitoring site. Annual and 24-hour PM_2.5 Design Values were projected from 2011 to 2021 and compared against the NAAQS. The CAMx Particulate Source Apportionment Technology (PSAT) was used to assess sector-specific contributions from local and regional sources.

Evaluation of Emissions of Nitrogen Oxides in Houston, Texas Using Three-Dimensional Aircraft Observations during the DISCOVER-AQ 2013 Mission

Uarporn Nopmongcol, Jim Smith, Greg Yarwood, Zhen Liu, Jeremiah Johnson, and Wei Chun

Several recent analyses have inferred that EPA's MOVES model over-estimates mobile NOx emissions by as much as a factor of two based on various comparisons between modeled and observed concentrations, including aircraft and satellite observations. The modeling used in these comparisons has typically used top-down emissions allocated to counties from national emission inventories. The Texas Commission on Environmental Quality uses a customized version of the Emissions Processing System instead of MOVES to produce very highly temporally- and spatially-resolved link-based on-road mobile emissions for Texas ozone nonattainment areas and well-resolved emissions for the remainder of Texas. Photochemical modeling for these areas has generally shown relatively good agreement between modeled and observed NOx concentrations at surface sites, in contrast to MOVES-based top-down applications in other areas. To see whether this agreement holds through the mixed layer, additional modeling was performed with the goal of extending the comparison beyond surface concentrations to those aloft using measurements from the 2013 DISCOVER-AQ mission in the Houston area. Analysis of the model results shows reasonably good agreement with both surface and aloft concentrations of NOx and NOy and finds no strong evidence that the TCEQ's on-road NOx emissions are overstated.

Modeled Impacts of Wood Burning Controls on 24-hour PM2.5 Attainment in Fairbanks, Alaska

Mark E. Hixson, Wenxian Zhang, Thomas R. Carlson

Fairbanks, Alaska was designated as nonattainment for the 24-hour PM_2.5 standard in 2009. The area suffers from uniquely severe winter inversions and cold temperatures. During these stagnation events the temperatures at the surface can frequently drop below and remain at -20 F for several days reaching extremes of -40 F. The combined impact of these meteorological conditions leads to high concentrations of PM_2.5 due to the increased emissions of home heating sources and reduced ventilation near the ground level. Using source contribution analyses solid fuel burning appliances were identified as the primary source driving high PM_2.5 concentrations.

An emissions inventory was developed to cover all source sectors in the nonattainment area. Special emphasis was placed on the development of home heating sources and mobile sources. A customized home heating energy model was developed to respond to daily temperature changes with a local device and fuel mix based on a multi-year survey. Local studies of onroad vehicle emissions and the use of plugin block heaters were accounted for in customizations to EPA's Mobile Source Emissions (MOVES). Modeling of meteorology, emissions, and air quality was developed on a 1.33 x 1.33 km grid resolution with a very highly resolved vertical structure. Meteorological data was generated by the Weather Research Forecasting (WRF) model covering two severe winter episodes for 2008. Emissions inventory inputs were processed through the Sparse Matrix Operator Kernel Emissions (SMOKE) model and SMOKE-MOVES models with some customizations to preserve hourly gridded home heating inventories.

Air quality modeling was performed with the Community Multiscale Air Quality (CMAQ) model for a baseline 2008, 2015 projected, 2015 control, 2019 projected and 2019 control inventories. Control scenarios focused on reducing PM_2.5 emissions from wood burning in the State's Moderate SIP. The control strategy developed for the SIP included a wood stove change out program, dry wood, device retrofits, and voluntary measures. Model performance tests for the 2008 baseline episodes showed overall good total PM_2.5 performance with an underperformance in sulfate and ammonium. Analysis of the control strategies demonstrated that attainment was impracticable by 2015 in Fairbanks though significant reductions would be achieved.

Real world emissions of NOx and other pollutants in the Ft. McHenry tunnel

A. Khlystov and D. Campbell

We will present results of measurements of NOx and various other air pollutants in Ft. McHenry tunnel during two one-week-long campaigns, one in winter (February 2015) and one in summer (August 2015). The tunnel is located in Baltimore, MD, on US I-95, the main highway on the East Coast of the U.S. The measurements were used to determine emission factors for light duty and heavy duty vehicles. We will also discuss how the measured emission factors compare to predictions using the MOVES2014a model during both seasons.

Source apportionment of reactive nitrogen deposition in the Greater Yellowstone Area using CAMx-PSAT

Rui Zhang, Colorado State University

Michael Barna, National Park Service

Tammy Thomson, American Association for the Advancement of Science

Bret Schichtel, National Park Service

Excess reactive nitrogen (N_r) deposition in sensitive ecosystems of the Greater Yellowstone Area (GYA) has passed critical thresholds adversely affecting these ecosystems. The Grand Teton Reactive Nitrogen Deposition Study (GrandTReNDS) field campaign was conducted in 2011 to characterize the spatiotemporal variations of the composition of N_r and its origin in Grand Teton National Park which is in the GYA. As part of this study, the CAMx chemical transport model using Western Air Quality Study (WAQS) emission and meteorology inputs was used to simulate N_r deposition in GYA. The CAMx Particle Source Apportionment Technology (PSAT) tools were used to estimate the contributions from agriculture, oil and gas and other source types from 27 source regions to the simulated N_r. The evaluation showed that the model consistently underestimated NH₃ and overestimated HNO₃ and contributions of ammonium sulfate and nitrate from the boundary conditions were likely significantly overestimated. Sensitivity analyses showed large variability between uni- and bi-directional NH₃ deposition processes, but neither mechanism provided superior in the model evaluation and both produced similar apportionment results. Source apportionment results suggest that the Snake River Valley is the largest source of N_r which could account for more than 80% of the N_r on the west side of GYA. Emissions from Northern Utah, western Wyoming, California, and boundaries were also important contributors throughout the GYA. Oil and gas activity had little impact on most lands except in the southern Wind River Mountain Range in winter.

MOVES-Based NOx Analyses for Urban Case Studies in Texas

Song Bai¹, Yuan Du¹, Annie Seagram¹, Kenneth Craig¹

¹Sonoma Technology, Inc., Petaluma, CA

Emissions inventories are an important component of air quality planning and a key input to photochemical grid models that support air quality assessments. Findings from recent studies suggest that emissions of nitrogen oxides (NO_x) may be overestimated in the U.S. Environmental Protection Agency's (EPA) National Emissions Inventory, perhaps by as much as a factor of two. This overestimate has generally been attributed to the mobile source sector, for which emission estimates are prepared using EPA's MOVES model. A number of potential issues have been identified with MOVES, including reliance on the model's default input data rather than more representative local inputs.

This study, supported by the Texas Air Quality Research Program, builds on previous work by examining MOVES emission estimates at the local scale using near-road monitoring data. Specifically, the project team is comparing MOVES emission results with ambient monitoring data, using well-established emissions reconciliation techniques. These analyses are being performed for case studies in three Texas metropolitan areas: Dallas-Fort Worth, Houston, and El Paso. In addition, sensitivity analyses featuring MOVES emission results from default vs. local data are being used to identify which input parameters have the greatest influence on NO_x emission estimates. To support the analyses, the project team is collecting local MOVES input data from planning agencies such as the North Central Texas Council of Governments (for the Dallas-Ft. Worth area).

The results of this work will support emissions inventory development and air quality management efforts in Texas by providing information on the accuracy of MOVES NO_x emission estimates and insights on the key input parameters for which local data are critical in MOVES.

Representative Meteorological Data for AERMOD: A Case Study of WRF-Extracted Data Versus Nearby Airport Data

Brian Holland, CCM, Senior Consultant, Trinity Consultants

Dr. Qiguo Jing, Senior Consultant, Trinity Consultants

Tiffany Stefanescu, Senior Consultant, Trinity Consultants

Dr. Weiping Dai, Director, Trinity Consultants

Historically, meteorological data for near-field air dispersion modeling (such as with AERMOD) has come from either the closest airport station to the facility being modeled, or from purpose-built "on-site" stations located at or near the facility. In areas where nearby observational data is not available or where meteorological conditions change rapidly with distance, these typical data sources become less representative of the actual facility location, introducing substantial error.

Recent changes to U.S. EPA's Appendix W air dispersion modeling document have opened the possibility of increased use of mesoscale meteorological model data (WRF or MM5) as an alternative source of meteorological data for near-field air dispersion modeling. Site-specific mesoscale model data is promising in that it has the potential to eliminate most of the distance-based representativeness error described above. However, this comes at the cost of introducing forecast error from the mesoscale model, which will typically be larger than the observation error of a perfectly-placed surface meteorological station. Weighing the representativeness error of a somewhat distant airport meteorological station against that of an imperfect mesoscale meteorological model is a necessary but potentially difficult task in deciding which meteorological data source is most representative of a given location.

This study examines the relative magnitude of the errors in these two meteorological data sources in two case studies: one using a facility located in relatively flat terrain, and another using a facility located in complex terrain. In both cases, an on-site meteorological station is used as "truth". Meteorological data taken from a moderately distant airport station and from the closest grid cell of a WRF model run are compared to the on-site station's observations to quantify the relative error of each. AERMOD model runs are then carried out using each data source (site-specific "truth", off-site airport, and mesoscale model) to quantify the extent to which error in each meteorological source translates into dispersion model result error.

Modeling Ozone in the Eastern U.S. using a Fuel-Based Mobile Source Emissions Inventory

Brian McDonald, Stuart McKeen, Yuyan Cui, Ravan Ahmadov, Si-Wan Kim, Gregory Frost, Michael Trainer

A fuel-based mobile source emissions inventory of nitrogen oxides (NOx) and carbon monoxide (CO) is developed for the continental US. Emissions are mapped for the year 2013, including emissions from on-road gasoline and diesel vehicles, and off-road engines. We find that mobile source emissions in the National Emissions Inventory 2011 (NEI11) are ~50% higher and ~100% higher for NOx and CO, respectively, than results from this study. We model chemistry and transport of emissions from the NEI11 and our fuel-based inventory during the Southeast Nexus (SENEX) Study period in the summer of 2013, using the Weather Research and Forecasting with Chemistry (WRF-Chem) model. In the Eastern US, there is a consistent over-prediction of tropospheric ozone (O3) levels when simulating emissions from the NEI11, with the largest biases located in the Southeastern US. Using our fuel-based inventory, we test O3 sensitivity to lower NOx emissions. Lowering NOx emissions improves model biases of 8-hour O3 by up to 4 ppb, and up to 6 exceedance days above the 70 ppb standard during summer 2013. Model results of NOy, CO, and O3 are also compared with aircraft measurements.

Application of ABaCAS (Air Benefit and Cost and Attainment Assessment System)-TX for Ozone Non-attainment in Southeast Texas

Tingting Cao, Wenwei Yang, Shengze Huang and C. Jerry Lin

Ground level ozone is one of the six criteria pollutants regulated by the United States Environmental Protection Agency (USEPA) due to its adverse effects on environment and public health. Houston-Galveston-Brazoria (HGB) in southeast Texas has been classified as an ozone non-attainment area by the National Ambient Air Quality Standards (NAQQS) since 2004, due to excessive emission of ozone precursors and unfavorable meteorological conditions (i.e., high pressure stagnant air brings dry and sunny weather). Policy makers are seeking for assessment tools for cost-effective emission control strategies to lower ground level ozone concentrations in the HGB non-attainment area, but few policy-oriented software are available for such assessments. In this study, a next generation software package. The Air Benefit and Cost and Attainment Assessment System (ABaCAS) is for the first time applied in Southeast Texas, aiming at assessing control strategies for ozone attainment, and estimating public health and economic benefit for emission reduction efficiently. A response surface modeling (RSM) is first developed, where ozone precursors of NOx from point and mobile sources and VOCs from area and point sources are controlled during ozone season of year 2015, to examine the effectiveness of emission control on Maximum 8-hour ozone concentration. Three emission control scenario cases are successfully demonstrated in this study. The results suggest that NOx limits ozone production during the selected high ozone day; emission of NOx from point source has the highest impact on ozone formation, followed by NOx emission from mobile source. Design values of control cases can improve from 80 ppbv to 75, 66, 68 ppbv respectively when emissions are adjusted to certain levels (i.e., NOx and VOC from point source increased to 137% and 102% of the base case emissions respectively, while VOC from area source and NOx from mobile source reduced to 91% and 68% of base case emissions respectively), indicating attainment status of 2008 NAQQS for ozone. Improvement of air quality brings health benefit end points (prevention of mortality, hospital admissions, emergency room visits and school loss day) with health benefit value to cost ratio of 2.52, 2.12 and 4.98, respectively.

Comparison of light-duty gasoline NOx emission rates estimated from MOVES with real-world measurements

Darrell Sonntag, David Choi, James Warila

Recent studies have shown differences between air quality model estimates and monitored values for nitrogen oxides. Several studies have suggested that the discrepancy between monitor and modeled values is due to an overestimation of NOx from mobile sources in EPA's emission inventory, particularly for light-duty gasoline vehicles.

Studies that directly measure vehicle emissions provide useful data for evaluating MOVES. In this paper, we present comparisons of MOVES2014 to thousands of real-world NOx emissions measurements from individual gasoline vehicles. The comparison studies include in-use vehicle emissions tests conducted on chassis dynamometer tests in support of Denver, Colorado's Vehicle Inspection & Maintenance Program and remote sensing data collected by road-side instruments in multiple locations and calendar years in the United States. In addition, we conduct comparisons of fleet-wide emissions measured from the Caldecott Tunnel near Oakland, California and the Van Nuys tunnel near Los Angeles in 2010. We will discuss how this work fits in the context of larger-scale efforts to evaluate and improve the MOVES model.

Source Apportionment Modeling to Investigate Background, Regional, and Local Contributions to Ozone Concentrations in Denver, Phoenix, Detroit, and Atlanta

Kenneth Craig¹, Garnet Erdakos¹, Shih Ying Chang¹, Naresh Kumar²

¹Sonoma Technology, Inc., Petaluma, CA
²Electric Power Research Institute, Palo Alto, CA

Background ozone (i.e., ozone produced by natural sources and international transport) combines with regionally transported and local emissions to impact ozone concentrations in urban areas. We are conducting detailed source apportionment modeling analysis at 12-km grid resolution with CAMx for the 2011 ozone season to investigate the extent of these impacts on ozone concentrations in Denver, Phoenix, Detroit, and Atlanta. These cities face complex air quality challenges and are expected to be designated non-attainment for the 2015 ozone National Ambient Air Quality Standards. Using the U.S. Environmental Protection Agency's (EPA) 2011 regulatory modeling platform as a starting point, this modeling study incorporates Version 2 of EPA's 2011 emissions platform and includes lightning NO_x emissions (0.52 Tg N in 2011), estimated based on the CMAQ approach using modeled convection and observed lightning flash rates. Simulations are being conducted with CAMx version 6.3, which includes recent updates to the ozone source apportionment technology (OSAT3) that improves estimates of local vs. non-local ozone partitioning. A source tagging strategy was developed to track international ozone transport, ozone produced by natural sources, ozone transported from adjacent and distant regions, and ozone produced by local emissions. The modeling tracks ozone contributions from 21 source regions, boundary conditions, and four emissions source groups: 1) natural sources, defined here as wildland and prescribed fires, lightning, and biogenic emissions; 2) on-road sources; 3) electrical generating unit (EGU) point sources; and 4) other anthropogenic sources (e.g., nonroad and non-EGU point sources).

The presentation will show results from the modeling analysis, highlighting local vs. non-local and anthropogenic vs. natural ozone contributions on high ozone days in the cities of interest. The presentation will also show results from 1) a "zero out" sensitivity modeling analysis of North American background ozone and 2) a sensitivity modeling analysis with a 50% reduction in on-road NO_x emissions to understand how the partitioning of ozone between local and non-local contributions may be influenced by on-road NO_x emission uncertainties.

Updates on Production of NOx by Lightning

Kenneth Pickering, Dale Allen, Eric Bucsela, Kristin Cummings

Allen and Pickering (2012) developed and tested an algorithm that calculates NO_x production by lightning for use in the CMAQ model. The algorithm allows the user options for specifying flash rates, NO_x production values per flash for cloud-to-ground and intracloud flashes, and the distribution of lightning NO_x emissions in the vertical. Since the initial development of the CMAQ algorithm, several research activities have yielded new information on the NO_x production efficiency by lightning. The joint NSF and NASA Deep Convective Clouds and Chemistry (DC3) field program has provided estimates of production per flash for several observed storms in Colorado and Oklahoma based on aircraft data taken during thunderstorm anvil transects. A production estimate based on cloud-resolved model simulations constrained by aircraft data has also been performed. Another approach at making lightning NO_x production estimates is the application of a specialized algorithm, which uses retrievals of NO₂ from the OMI instrument on NASA's Aura satellite and flash rates from the World Wide Lightning Location Network (WWLLN). An initial study covered the Gulf of Mexico region, yielding a mean of 80 45 moles per flash. Subsequent analyses are being conducted for the mid-latitude continental regions as a whole and for individual tropical regions. An important finding from both the DC3 and OMI research is that NO_x production per flash is likely inversely proportional to flash rate. The implication of these results for the current NO_x production algorithm in CMAQ will be discussed along with the steps needed to implement a revised CMAQ algorithm in the future.

Predicting Future-Year Ozone Concentrations: Integrated Observational-Modeling Approach for Probabilistic Evaluation of the Efficacy of Emission Control strategies

M. Astitha¹, H. Luo¹, C. Hogrefe², R. Mathur², and S.T. Rao^1,3
1University of Connecticut, ²US Environmental Protection Agency, ³North Carolina State University

In current regulatory applications, regional air quality models are applied for a base year and a future year with reduced emissions using the same base year meteorological conditions. The base year design value is multiplied by the ratio of the average of the top 10 ozone concentrations for the base and future years to assess whether the estimated future year design value meets the ozone standard. The SIP process requires that the adopted emissions control strategy would help meet and maintain the ozone standard in the future but does not explicitly account for differences in meteorological conditions between the base year and future year. Because the same meteorological conditions would never prevail in future years and observations for the future year are not available, the current attainment demonstration methodology can never be evaluated in the real world. We present a new method for applying regional ozone air quality models in the regulatory setting, using the information extracted from the short-term (synoptic) and long-term (baseline) forcings embedded in ozone observations during the 30+ year period from 1981 to 2014. This new method, named the baseline projection method, provides the confidence limits for the design value and 4th highest MDA8 ozone for each given emission loading scenario along with the probability of exceeding any given threshold (e.g. 70 ppb) at each monitoring site. We also evaluate the ability of the baseline projection methods in capturing the observed ozone design values at all monitoring stations in the future year. Results indicate that combining the change in the ozone baseline predicted by the air quality model with the meteorological forcing embedded in the historical ozone observations would enable us to evaluate the efficacy of the selected emissions control strategy in meeting the ozone standard in future years while also providing confidence limits for such predictions.

Influence of different canopy reduction functions on biogenic NO emission patterns in northern Europe

Jan A. Arndt, Volker Matthias, Armin Aulinger, Johannes Bieser

Effects of Relative Reduction Factor Grid Cell Sampling Approaches on Model Attainment Demonstrations

Byeong-Uk Kim and James Boylan

State air quality agencies must submit attainment State Implementation Plans (SIPs) for ozone nonattainment areas (NAAs) and infrastructure SIPs to demonstrate that their state does not contribute to ozone nonattainment in a downwind state. These SIPs include modeling to project future design values at ambient monitoring sites for comparison to the National Ambient Air Quality Standards (NAAQS). The future design value projections are the product of the base year design values (based on three years of ambient measurements) and the relative reduction factor (YYF), where the YYF is defined as the ratio of future modeled ozone concentrations to base year modeled ozone concentrations. The U.S. EPA's modeling guidance recommends selecting the grid cell with the highest modeled ozone concentration in the base year from a matrix of 3x3 grid cells surrounding the monitoring site (3x3 Method). However, this value can be significantly different than the modeled ozone concentration in the grid cell containing the monitor and/or can be significantly different than the ambient ozone measurements at the monitor. An alternative approach is to select the single grid cell containing the monitor location (1x1 Method). This allows for the direct comparison of modeled ozone concentrations with ambient measurements. In this study, we examine the spatial variations of estimated future design values due to the choice of grid cell sampling approaches used in the YYF calculations. We utilized the most recent 2011/2023 modeling runs performed by EPA. In general, the future design values using the 1x1 Method are higher than using the 3x3 Method. The choice of 3x3 Method vs. 1x1 Method can have an impact of up to 10 ppb on future design values and can significantly impact the list of monitors that are projected to be attainment or maintenance in the future. The largest impacts are typically at land-water interfaces.

Top-down Estimation of North Korean Emissions and Their Impacts on South Korean Air Quality

Minah Bae¹, Hyun Cheol Kim^2,3, Byeong-Uk Kim⁴ Yong-Jae Lim⁵, Jae-Bum Lee⁵, Chul Yoo⁶ and Soontae Kim*¹

North Korean emissions were constrained using space-borne observations from the Ozone Monitoring Instrument (OMI) and the Global Ozone Monitoring Experiment-2 (GOME-2) instruments. Being located downwind side of strong anthropogenic emission sources, South Korea has been affected by regional scale transport of pollutants and their precursors. While it is evident that northern China is one of dominant emission sources in the region, North Korea is also in the middle of major pollutant transport pathways into South Korea. Updating recent changes of emissions from North Korea and quantifying its impact is necessary to identify regional sources' impact but also very challenging due to limited socioeconomic information from North Korea. We utilized remote sensing observations of NO2, SO2 and HCHO from OMI and GOME-2 to adjust emission fluxes from current emissions inventories. Base CMAQ simulations were conducted over 27-km (East Asia) and 9-km (North and South Korea) during 2014, using CREATE 2015 and CAPSS 2013 emissions inventories for Asia and South Korean emissions, respectively. Adjusted emissions were estimated by comparison of modeled and observed column densities after vertical and spatial data processing (e.g. averaging kernel and downscaling techniques). We also demonstrate long-term changes of two major point sources in North Korea (e.g. Pyeongyang and Bukchang power plants) to discuss recent changes of power generation in North Korea.

Improvements to Emission Projection Methods for non-EGU Point and Nonpoint Sources

Caroline M. Farkas, Lee Tooly, Elineth Torres, Brian Keaveny, Alison Eyth, Jeffrey Vukovich 

Future year emission estimates are used in regulatory analysis and air quality modeling to predict the influence of emission control programs and economic factors on the attainment of the National Ambient Air Quality Standards (NAAQS). The EPA develops the projection inventories as part of its emissions modeling platform. The EPA is actively reviewing the existing methods used for projecting emissions for the non-Electric Generating Unit (non-EGU) point and nonpoint sources. Review includes analyses of growth and control factors that are applied to estimate future year emissions. The 2014v1 National Emissions Inventory (NEI) is initially used to help prioritize emission sources and focus the methods review. Federal control programs that have been more recently promulgated and not yet accounted for in existing methods are identified and five industry-specific regulatory programs are investigated for potential control factor updates. Several priority emission sectors are reviewed for potential updates to existing growth assumptions including estimating ammonia emission from agricultural livestock. This on-going review process is expected to incrementally improve and update growth and control assumptions for a subset of the emissions sources that may be applied to the next emissions modeling platforms, e.g., the 2014 NEI-based modeling platform.

A unified prescribed fire database for the Southern United States

Fernando Garcia Menendez, Sadia Afrin and Talat Odman

Prescribed fire is used extensively in the Unites States to reduce hazardous wildfire risk and pursue different ecological objectives. However, prescribed fires are also one of the largest sources of air emissions in the country and can adversely impact air quality. Each year the South experiences the largest number of wildland fires in the United States. The vast majority of these are prescribed fires. In this region, a large population lives within a 167 million acre urban-wildland interface, neighboring or intermixed with undeveloped fire-prone land. Therefore, it is important that prescribed burning programs in the South consider their potential air quality impacts. As a part of an ongoing study investigating the effects of prescribed fire on air quality and health, we are developing a unified prescribed fire database for the Southern U.S. based on burn permit records from agencies across Southern states. We have systematized available fire records with varying data attributes into a single database. Additionally, we show discrepancies with satellite-derived fire estimates and how remote sensing inventories may fail to include low-intensity fires. Finally, we compare statistical correlations between fire records, satellite fire detections, and observed PM2.5 concentration at monitoring stations in close proximity to prescribed fire activity. Ultimately, the unified database will improve representations of prescribed fire sources in emissions inventories used to drive air quality modeling.

Improvement of Speciated Anthropogenic VOC Emissions using KORUS-AQ/MAPS-Seoul Aircraft Field Campaign Data

Jinseok Kim, Jung-Hun Woo*, Younha Kim, Chanjong Bu, Yungu Lee, Young-Kee Jang, Isobel J. Simpson

The KORUS-AQ/MAPS-Seoul are the international aircraft field campaigns to investigate air pollution form the internal and external source (e.g. China) over the Korea region. The CREATE (Comprehensive Regional Emissions inventory for Atmospheric Environment) emissions inventory and SMOKE-Asia emission processing system were used to support chemical transport modeling and to serve as a priori emission information for evaluation. Initial results of inter-comparison analysis showed large discrepancies in VOC species over South Korea, especially over urban regions. Several aromatic VOC species ( e.g. toluene and xylene ) were observed high near megacities and petro-chemical plants but under-predicted by Chemical Transport Models (CTMs) - possibly due to relatively low emissions or incorrect selection of chemical speciation profiles. The chemical speciation profiles and emissions inventory for each emission sources, therefore, have to be re-visited to improve emissions information. In this study, we have; 1) re-examined our emissions inventory and emission speciation processes, 2) and tried to find possible missing sources and alternative chemical speciation profiles, to improve our modeling emissions inventory. Initial review of the inventory sector composition and speciation profile mapping revealed that the painting and surface coating sources were linked with the SPECIATE chemical speciation profiles with relatively low aromatic species contribution. In Korea, however, the solvent use and industrial processes are dominant sectors for anthropogenic VOCs emission whereas residential sector occupies highly in China. These are the inventory sectors which needs careful selection of chemical speciation profiles. We, therefore, reviewed more recent chemical speciation profiles using the literatures from Korea and China, rather than using those from U.S EPA. As a results, ARO1(e.g. toluene, benzene) and ARO2(e.g. xylene) emissions were increased by 4 and 1.4 times, respectively. ARO1 shows a strong increase over the eastern China and South Korea region and ARO2 tends to decrease over China and North Korea, but East China and South Korea region shows weak increase. Amount of SOA from FAC method showed the highest in the southeastern part of China where Shanghai is located, and for Korea is the highest region was the Gyeonggi province which is a part of Seoul Metropolitan Area. Also, the simulated ozone concentration showed higher agreements with measurement data when the updated VOCs emission were used. More in-depth analysis of major urban and industrial areas of the country will be presented at site.

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[22173080800] " and "This work is financially supported by Korea Ministry of Environment(MOE) as Graduate School specialized in Climate Change ."

Assessment of important SPECIATE profiles in EPAs emissions modeling platform and current data gaps

Casey D. Bray¹, Madeleine Strum², Heather Simon², Lee Riddick³, Mike Kosusko⁴, Venkatesh Rao²
1 Contractor to the Air, Climate and Energy National Program, US EPA, RTP, NC; Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Raleigh, NC
2 Office of Air Quality Planning and Standards, US EPA, RTP, NC
3 National Exposure Research Laboratory, US EPA, RTP, NC
4 National Risk Management Research Laboratory, US EPA, RTP, NC

The US Environmental Protection Agency (EPA)'s SPECIATE database contains speciation profiles for both particulate matter (PM) and volatile organic compounds (VOCs) that are key inputs for creating speciated emission inventories for air quality modeling. The objective of this work is to identify the most influential profiles based on mass and reactivity for various regions of the US. These profiles will be further investigated to characterize the profile quality and determine whether current matching between profiles and source types appropriately captures source type and regional variability in speciation. In cases where this analysis identified either low quality or poorly matched profiles, an in-depth review of the SPECIATE database and the literature will be conducted to identify currently available suitable replacements. In cases where no suitable replacement profiles are found, this analysis will identify important gaps in the current literature which may be used to prioritize future speciation source testing. Through this process we aim to identify critical research needs, improve the SPECIATE database and improve a critical input for photochemical modeling efforts.

PolEmiCa model for local air quality assessment in airports
Kateryna Synylo*, Oleksandr Zaporozhets
National Aviation University, Kosmonavta Komarova 1, 03058 Kyiv, Ukraine

Main purpose of the PolEmiCa is to provide the dispersion (Pollution) and inventory (Emission) calculations for the aircraft engine emission during the LTO-cycle of the aircraft movement inside airport area. It includes the aircraft emission from Start-up procedures, APU and GSE also. Current version of the PolEmiCa combines the calculation for the main stationary sources (power plants, fuel farms) of the emission and road vehicles inside airport area with character toxic compounds for aircraft engine emission: CO, HC, NOx, SOx, PM10, PM2.5 and fuel vapors (HC).

Assessment and Improvement of the National Emissions Inventory 2013 of South Korea for Ozone Forecasts

Seunghee You, Byeong-Uk Kim, Hyun Cheol Kim, Yongjae Lim, Insuk Suh, Jaebum Lee, and Soontae Kim

In recent years, high concentration ozone is frequently observed in South Korea, and the government is issuing ozone forecasts to provide precise air quality information to the public. Since ozone is a secondary product, a model that can predict its complicated generation process and accurate model input data are needed. In this study, air quality simulation was performed using Clean Air Policy Support System (CAPSS) 2013, an anthropogenic emission inventory of South Korea, and the concentration difference was quantitatively confirmed by comparing the simulated concentration and observed concentration of nitrogen oxide (NOx) and volatile organic compounds (VOCs) by area. In the case of VOC, the emission amount of it was revised by applying the ratio of the simulated concentration and observed concentration for the areas where the photochemical monitoring network is located; and using the revised emission amounts, air quality simulation was carried out to assess ozone warnings and determine its improvement level depending on the changes in emission amounts. The study period was from May to August in 2016, which is summer season when high concentration ozone is observed. Air quality simulation was conducted using a 3-km horizontal resolution grid for South Korea by refining the model resolution. For meteorological inputs, Weather Research and Forecast model was simulated using the FNL data. To generate emission amounts, CAPSS 2013 was used for Korea, and Comprehensive Regional Emissions inventory for Atmospheric Transport Experiment 2015 for overseas areas. VOC measurements from 18 Photochemical Air Monitoring Stations were collected and divided into six regions for the comparison with the modeled values in the study. For the emission revision, a ratio of modeled and observed monthly mean concentrations was applied to the VOC emissions in the target areas. The results showed that the daily maximum ozone concentration increased by more than 8 ppb depending on area.

Characterizing Space Heating Emissions for Fairbanks PM2.5 Modeling

Wenxian Zhang, Mark E. Hixson, Thomas R. Carlson

Fairbanks, surrounded by the high ridges in Central Alaska, frequently experiences high PM_2.5 concentrations in winter due to extreme air stagnation and a changing mix of air emissions tied to cost-driven shifts in space heating fuels. A PM_2.5 State Implementation Plan (SIP) has been developed and submitted to EPA to address the attainment of the 24-hr PM_2.5National Ambient Air Quality Standards (NAAQS).During episodic wintertime conditions, residential space heating represents the single largest source sector, comprising over half of all direct PM_2.5 emissions.The CMAQ modeling system was used to evaluate control measures and demonstrate attainment. To predict reliable PM_2.5 levels and design effective control strategies, it is essential to provide an accurate representation of the local emissions accounting for spatial and temporal variations.This study presents the method adopted for developing a refined PM_2.5 emissions inventory from residential space heating source sector and developing gridded hourly resolved "CMAQ-ready" emissions inventory using SMOKE.

To develop the residential space heating emission inventory; (1) an instrumentation study with 30 households was conducted to quantify hourly energy usage of different types of space heating device; (2) home heating surveys with more than 700 local households for successive six years were conducted to estimate the regional fuel and device usage; (3) laboratory tests were performed to assess the emission factors of different types of wood with various moisture content; and (4) a regression model was developed to estimate the regional space heating energy use, based on the results of the instrumentation study and the home heating surveys.Since SMOKE is typically applied using annual averaged emission inventory for residential area sources such as space heating emissions, this study modified programs within SMOKE to preserve the spatially gridded and hourly resolved space heating emission inventories. The modified SMOKE programs were used to develop the "CMAQ-ready" hourly gridded PM_2.5 emissions inventory for residential space heating source sector.

Sensitivity of MOVES emissions specifications on modeled air quality using traffic data and near-road ambient measurements from the Las Vegas and Detroit field studies

R. Chris Owen, Michelle Snyder, Heather Simon, Sue Kimbrough

Near-road measurement campaigns were conducted along interstates in Las Vegas, NV, and Detroit, MI to more fully characterize the impacts of highway vehicle emissions in a near-source environment. Continuous upwind and downwind gas-phase air pollutant concentrations were measured at multiple distances from the interstates. Traffic count data was recorded simultaneously with the gas-phase data. The traffic data also included vehicle speed and vehicle length by length bin. This work uses the traffic data to evaluate several aspects of traffic modeling typically conducted for regulatory purposes. Specifically, we focus on the emissions estimates from EPA's Motor Vehicle Emission Simulator (MOVES) model using county-specific versus default fleet-mixes, age distributions, alternative fuel specifications, and alternative drive cycles. We evaluate emissions and air quality estimates by conducting a model-to-monitor comparison between the modeled concentrations from several dispersion models and the on-site ambient measurements. We expand on previous analysis, which simulated concentrations at monitors as due solely to emissions from the roadway. Here we add the air pollution increment from roadway emissions to urban background concentrations that were characterized using data from an upwind monitor

Unconventional Constraints on Nitrogen Chemistry using DC3 Observations and Trajectory-based Chemical Modeling

Chemical transport models underestimate nitrogen dioxide observations in the upper troposphere (UT). Previous research in the UT succeeded in combining model predictions with field campaign measurements to demonstrate that the nitric acid formation rate (HO + NO2 HNO3 (R1)) is overestimated by 22% (Henderson et al., 2012).  A subsequent publication (Seltzer et al., 2015) demonstrated that single chemical constraint alters ozone and aerosol formation/composition. This work attempts to replicate previous chemical constraints with newer observations and a different modeling framework. 

We apply the previously successful constraint framework to Deep Convection Clouds and Chemistry (DC3). DC3 is a more recent field campaign where simulated nitrogen imbalances still exist. Freshly convected air parcels, identified in the DC3 dataset, as initial coordinates to initiate Lagrangian trajectories. Along each trajectory, we simulate the air parcel chemical state. Samples along the trajectories will form ensembles that represent possible realizations of UT air parcels. We then apply Bayesian inference to constrain nitrogen chemistry and compare results to the existing literature. 

Our anticipated results will confirm overestimation of HNO3 formation rate in previous work and provide further constraints on other nitrogen reaction rate coefficients that affect terminal products from NOx. We will particularly focus on organic nitrate chemistry that laboratory literature has yet to fully address. The results will provide useful insights into nitrogen chemistry that affects climate and human health.

Evaluating CO:NOx in a near-road environment using ambient data from Las Vegas

Heather Simon, R. Chris Owen, Richard Baldauf, Megan Beardsley, James Crawford, Pat Dolwick, Kristen Foley, Barron Henderson, Sue Kimbrough, Norm Possiel, Darrell Sonntag, Claudia Toro, Lukas Valin

Previous studies have used CO: NOx (the increment above background of CO relative to the increment above background of NOx) to identify major pollutant source contributions or to evaluate emissions inventories. Some studies have isolated mobile source emissions using near-road or morning measurements while others have relied on ambient measurements in urban and rural areas. Recent work shows the limitations of interpreting this ratio when it is derived from data representing the ambient urban environment. Near-road measurements taken in Las Vegas, NV provide a unique opportunity to evaluate the utility of using the CO: NOx ratio to characterize onroad mobile source emissions in a near-road environment and to evaluate emissions estimates of this ratio. The Las Vegas measurements were taken at 4 near-road sites including one site 100 m upwind of a major highway and 3 sites at varying distances downwind (20 m, 100 m, and 300 m) over the period from Dec 2008 to Jan 2010. In this work, we determine the CO: NOx across the I-15 freeway from 5-minute data using two methods: 1) calculating relative enhancement of downwind monitor concentrations relative to the upwind monitor and 2) using a linear regression of downwind CO versus NOx concentrations. We first compare the two methods to understand whether the regression method reliably predicts the enhancement across the roadway and evaluate how this relationship changes temporally, with downwind distance and with other explanatory variables such as light-duty and heavy-duty traffic and wind speed. Then, we compare the enhancement ratio to emitted CO:NOx ratios derived from simulations using the Motor Vehicle Emission Simulator (MOVES) model.

Title: Sensitivity of MOVES-estimated vehicle emissions to inputs when comparing to real-world measurements
Darrell Sonntag (EPA, OTAQ), David Choi (EPA, OTAQ), Megan Beardsley (EPA, OTAQ), Claudia Toro (ORISE Participant Hosted by EPA)

Comparing MOVES emission rates to real-world measurements is a key component of evaluating MOVES emission inventories. One of the challenges to properly compare MOVES estimates to real-world measurements is to develop MOVES inputs which accurately capture the conditions of the study-location. This study demonstrates the sensitivity of MOVES emission rates (e.g. grams NOx/kg-fuel) to model inputs, based on the Caldecott Tunnel near Oakland, California during the summer of 20101. For inputs where local data were not collected during the period of measurements, such as vehicle age distribution, driving behavior, and fuel properties, we developed ranges of plausible inputs, and demonstrate the sensitivity on the MOVES results. We demonstrate the importance of using local inputs, instead of MOVES national default inputs, when comparing MOVES to real-world measurements.

CAMx Model Sensitivity Analysis of Emissions Temporal Profiles; Impacts on 2011 Modeled NOx/NOy Concentrations

Brian Timin, United States Environmental Protection Agency, Research Triangle Park, NC.
Patrick Dolwick, United States Environmental Protection Agency, Research Triangle Park, NC.
Norm Possiel, United States Environmental Protection Agency, Research Triangle Park, NC.
Heather Simon, United States Environmental Protection Agency, Research Triangle Park, NC.
Alison Eyth, United States Environmental Protection Agency, Research Triangle Park, NC.
Jeffrey Vukovich, United States Environmental Protection Agency, Research Triangle Park, NC.
Sarah Roberts, United States Environmental Protection Agency, Ann Arbor, MI.
Claudia Toro, United States Environmental Protection Agency, Ann Arbor, MI.

Ozone is formed through chemical reactions of precursor (NOx and VOC) emissions during conducive meteorological conditions.
In order to accurately model ozone concentrations with a chemical transport model (CTM), it is important to accurately estimate
the precursor emissions from each type of emissions source (e.g. power plants, on-road mobile sources, off-road mobile sources, etc.).
Some studies have attempted to evaluate emissions inventories by examining measured vs. modeled precursor concentrations and ratios.
These studies have produced evidence that mobile source inventories may in some cases over-predict NOx emissions.
In addition to the absolute number of annual tons of emissions, an important aspect of modeling is the temporalization of annual,
seasonal, and or monthly inventory information down to the hourly input level which is needed by CTMs.
This study looks at important sources of NOx emissions and potential improvements in the temporalization of certain major point sources,
on-road, and non-road emissions sources. We use CAMx modeling of a 2011 case and measurements of NOx, NOy, and ozone to examine the precursor
and ozone impacts of these changes to emissions temporalization.

Investigating modeling platform emissions for grid cells associated with a near-road study site during a field campaign in Las Vegas

C. Toro¹, R.C. Owen², J. Vukovich², D. Sonntag³

¹ORISE participant, Assessment & Standards Division, Office of Transportation and Air Quality, US Environmental Protection Agency, Ann Arbor, Michigan

²Air Quality Assessment Division, Office of Air Quality Planning and Standards, US Environmental Protection Agency, RTP, North Carolina

³Assessment & Standards Division, Office of Transportation and Air Quality, US Environmental Protection Agency, Ann Arbor, Michigan

A number of studies have suggested that onroad mobile emissions are responsible for discrepancies between modeled and monitored NOx emissions. However, a variety of factors within the air quality modeling platform could potentially play a role in the reported discrepancies, including activity inputs such as fleet composition and traffic volumes. The work presented will focus on contrasting inputs to the modeling platform with information obtained during a near-road field campaign in Las Vegas, NV. We focus on understanding differences in aspects such as diurnal patterns of emissions and vehicle fleet mix. This investigation aims to contribute to the diagnostic analysis of factors that could potentially influence the modeling platform output resulting in the observed NOx discrepancies. 

Exploring differences in nitrogen oxides overestimation at the seasonal and day-of-week levels to understand potential relationships with mobile source emission inventories.

Claudia Toro¹, Heather Simon², Kristen Foley³, Pat Dolwick², Megan Beardsley⁴, Barron Henderson², Norm Possiel²

¹ORISE participant, Assessment & Standards Division, Office of Transportation and Air Quality, US Environmental Protection Agency, Ann Arbor, Michigan

²Air Quality Assessment Division, Office of Air Quality Planning and Standards, US Environmental Protection Agency, RTP, North Carolina

³Computational Exposure Division, National Exposure Research Laboratory, US Environmental Protection Agency, RTP, North Carolina

⁴Assessment & Standards Division, Office of Transportation and Air Quality, US Environmental Protection Agency, Ann Arbor, Michigan

Previous studies have suggested that onroad mobile emission inventories are the drivers behind discrepancies observed between modeled and measured nitrogen oxides (NOx and NOy). These studies have mostly focused on top-down approaches using summer ambient measurements to assess emission inventories. In this work, we explore seasonal differences (particularly between summer and winter) in NOx or NOy biases and mobile emissions for different sites across the US. We also examine weekday-weekend differences in NOx (NOy) bias and their relationship with weekday-weekend differences in mobile emissions by sector, by season and by time of day. The results of this diagnostic investigation challenge the likelihood that a single factor results in the observed NOx (NOy) overestimation and provide insight into other processes that could potentially play a role in the reported discrepancies. 

Development of the Simple Indoor Air Chemistry Simulator (SIACS)

Serena H. Chung, Vito Ilacqua, Anna Freitheim, Sharon Wildberger, Jordan Zambrana, Randolph Chapman

Exposure to air pollutants occurs mostly indoors, where building characteristics, consumer products, and near sources all affect air composition, along with ambient air quality. Yet, the risk paradigm for ambient air pollution largely omits these factors, and epidemiological analyses leave these sources of heterogeneity uncharacterized as noise. In fact, the U.S. EPA does not have a comprehensive indoor air chemistry model to help develop public health guidelines that take into account both ambient air quality and the indoor sources in a particular environment (e.g., should people ventilate their houses while cooking if they live near dense traffic ). In a changing climate, which will modify indoor-outdoor air exchange, the options about how to best manage indoor environments to minimize risk become even less clear. A new Simple Indoor Air Chemistry Simulator (SIACS) is being developed with the objective of using it as a screening-level indoor chemistry model to identify research questions to be explored with a more complex model. SIACS considers gas-phase chemistry of volatile organic compounds, nitrogen oxides, and ozone and accounts for emissions, deposition, ventilation, and filtration. Sensitivities of predicted pollutant concentrations to chemical mechanisms, ambient concentrations, ambient conditions will be investigated. The long-term goal is to develop a tool that can be used to evaluate the impacts of a changing environment, building use or modification, and behaviors on indoor air quality and develop guidelines to minimize total exposure and risk.

Regional impacts of extending inorganic and organic cloud chemistry with AQCHEM-KMT

Fahey, K., Carlton, A., Sareen, N., Hutzell, W., Luecken, D.

Starting with CMAQ version 5.1, AQCHEM-KMT has been offered as a readily expandable option for cloud chemistry via application of the Kinetic PreProcessor (KPP). AQCHEM-KMT treats kinetic mass transfer between the gas and aqueous phases, ionization, chemical kinetics, droplet scavenging of interstitial aerosol, and wet deposition occurring within clouds. Here we investigate the impacts of expanding the AQCHEM-KMT mechanism with additional inorganic and organic chemistry with regional CMAQ applications for summer and winter periods.

The cloud chemistry added here includes the formation of organic acids from glyoxal, methylglyoxal, glycolaldehyde, and acetic acid (replacing the in-cloud SOA parameterization of CMAQ's standard AQCHEM), sulfate formation from NO2 and HNO4, as well as several additional reactions of radicals and other inorganic and organic species. Preliminary summer simulations indicate an increase in in-cloud SOA concentrations as well as a changing spatial distribution, in part due to the addition of new precursors to in-cloud SOA, which may originate from other sources than those precursors in the base model. Initial winter simulations show greater (and variable) impacts on inorganic species, with sulfate concentrations increasing and nitrate decreasing in much of the eastern United States.

Because the addition of new species and reactions to the cloud chemical mechanism will lead to an even greater model computational burden, we also revisit using equilibrium assumptions to dictate the distribution of species between phases and ions. An attempt is made to identify those species/conditions for which such simplifying assumptions can be made to improve model efficiency, while minimizing adverse impacts on accuracy.

Implementing Subgrid-Scale Cloudiness into the Model for Prediction Across Scales - Atmosphere (MPAS-A) for Next Generation Global Air Quality Modeling

Jerold Herwehe, Robert Gilliam, Russell Bullock, Jonathan Pleim, and Hosein Foroutan

A next generation air quality modeling system is being developed at the U.S. EPA to enable seamless modeling of air quality from global to regional to (eventually) local scales. State of the science chemistry and aerosol modules from the Community Multiscale Air Quality (CMAQ) model will be online-coupled to the Model for Prediction Across Scales -- Atmosphere (MPAS-A), a global meteorological model developed at the National Center for Atmospheric Research (NCAR). To prepare the next generation air quality model for conducting retrospective simulations, several additional preferred physics schemes and options from the Weather Research and Forecasting (WRF) model have been implemented by our team into MPAS-A: the Pleim surface layer (PSL), the Pleim-Xiu (PX) land surface model with fractional land use for a 40-class National Land Cover Database (NLCD40), the Asymmetric Convective Model 2 (ACM2) planetary boundary layer scheme, and analysis nudging four-dimensional data assimilation (FDDA). As released, MPAS-A includes the 2004 version of the Kain-Fritsch (KF) convective parameterization and only allows the resolved, grid scale clouds to affect the radiation budget. This presentation discusses results from updating the KF scheme in MPAS-A to the latest version which adds subgrid-scale cumulus cloud feedback to the radiation schemes, multiple convection triggers, and a scale-aware convective time scale. Test simulations of a Northern Hemisphere summer month (July 2013) were conducted on a global variable resolution mesh with the higher resolution cells centered over the contiguous United States. Initial conditions and driving fields for the FDDA and soil nudging were provided by NOAA/NCEP's GDAS/FNL, GFS, and RUC analyses. Results from the MPAS-A simulations utilizing these added physics schemes and subgrid-scale cloud-radiation interactions were evaluated against observational data [such as those available from NCEP's Meteorological Assimilation Data Ingest System (MADIS)] to ascertain the impact of these MPAS-A enhancements on air quality-relevant meteorological parameters.

Comparing CMAQ Forecasts with a Neural Network Forecast Model for PM2.5 in New York

Samuel Lightstone, Barry Gross, Fred Moshary

Human health is strongly affected by the concentration of fine particulate matter (PM). The need to forecast unhealthy conditions has driven the development of Chemical Air Quality Transport Models such as CMAQ. These models attempt to simulate the complex dynamics of chemical transport by combined meteorology, radiation, 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. To be useful, the release of this data must be coordinated with the state agencies that are directed to provide these forecasts. In the simplest utilization, any forecast should focus on next day pollution levels, and should be provided by the end of the business day (5PM local). 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 prior PM2.5 concentrations together with both prior and forecast meteorology. We find that this approach generally provides a more accurate prediction of future pollution levels on a local level, except under conditions where pollution transported events, such as smoke plumes, are encountered. Methods to improve the NN for transported cases, as well as improving the use of CMAQ for local forecasts, are also explored.

New Developments in the Eulerian and Lagrangian Modeling of the Chemistry of Biomass-Burning Plumes
C.R.Lonsdale, C. M. Brodowski, M. J. Alvarado, J. Hegarty, J. M. Henderson, J. R. Pierce, E. Ramnarine, A. Kochanski, J.C. Lin

We present a newly developed plume-scale process model SAM-ASP (System for Atmospheric Modeling integrated with AER's Aerosol Simulation Program) used to develop a parameterization of the near-source biomass-burning plume evolution for use in the CAMx air quality model, as well as other Eulerian regional and global air quality models. This parameterization improves upon the ASP-based parameterization of Lonsdale et al. (2015) by including the complex processes of dispersion, advection, and deposition on plume chemistry. We will present the new parameterization, compare it to observed rates of O3 formation in biomass burning plumes, and discuss its implementation in air quality models. We also present the Lagrangian chemical transport model tool STILT-ASP (Stochastic Time Inverted Lagrangian Transport model with ASP) to determine the contribution of fires to O3 and PM2.5 measured at a specific receptor. We will present results of a recent evaluation of STILT-ASP v2.0, with updated emissions inventories.

The influence pre-existing organics on secondary organic aerosol formation from reactive uptake of isoprene epoxydiols in a regional scale model

Mutian Ma¹, Havala Pye², Yue Zhang¹, Yuzhi Chen1, Chitsan Wang¹, Jason D. Surratt¹, William Vizuete¹

¹University of North Carolina, Gillings School of Global Public Health, Department Environmental Science and Engineering, Chapel Hill, NC, USA

²Environmental Protection Agency at Research Triangle Park, RTP, NC, USA

Isoprene oxidation products formed under low-nitric oxide (NO) conditions are important precursors for the formation of particulate matter (PM). Specifically, isoprene epoxydiols (IEPOX), an abundant isoprene-derived oxidation product, forms secondary organic aerosol (SOA) via acid-catalyzed multiphase (heterogeneous) reactions. Recent experimental studies at UNC-Chapel Hill, using synthetically-derived IEPOX, have quantified the reactive uptake of IEPOX to determine how much gas-phase IEPOX could form SOA. An accurate description of this process is critical in regulatory modeling to predict the formation of SOA. These experiments, however, consisted only of pure sulfate aerosol of varying acidity without a pre-existing organic coating. Under some atmospheric conditions, organic layers can exist on pre-existing aerosols potentially changing both the physical and chemical properties of IEPOX-derived SOA. Prior experiments, using a-pinene as a surrogate for an organic coating on an ammonium bisulfate seed, have found that IEPOX uptake ceased when the mass of a-pinene derived coating exceeds 85% of the total aerosol mass. Ongoing experiments at UNC are quantifying how the thickness and type of organic coating at varying levels of relative humidity will influence IEPOX diffusivity within the aerosol phase. Leveraging this new data, we will estimate the impact of organic coating on regional model predictions of IEPOX-derived SOA using the Community Multiscale Air Quality (CMAQ) v5.2 system. This version of CMAQ was developed by Environmental Protection Agency (EPA) and includes IEPOX and aqueous-phase chemical reaction pathways leading to IEPOX-derived SOA formation. These modeling algorithms, however, have yet to include the influence of pre-existing organic coatings. Using the new UNC experimental data as a guide, we will provide sensitivity runs changing the reactive uptake parameter and predicting the change in IEPOX-derived SOA. This modeling episode also coincided with the Southern Oxidant and Aerosol Study (SOAS), allowing for comparisons with measured IEPOX-derived SOA. 

Quantifying Primary and Secondary Ultrafine Particle Sources in the United States with CMAQ-NPF

Benjamin N. Murphy1 and Francis Binkowski2

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

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

We have updated the Community Multiscale Air Quality (CMAQ) model v5.2 with new algorithms and data to represent the emission, secondary formation, and growth of ultrafine particles. This enhanced model, CMAQ-NPF, is evaluated with measurements from several urban and rural sites throughout the US. We then analyze CMAQ results at the continental US level to determine the impacts of primary and secondary particle sources on ultrafine particle exposure.

Emerging evidence shows that organic compounds and amines are capable of participating with sulfuric acid to form new particles or, when sulfuric acid concentrations are low, generating particles on their own. Moreover, because they are so abundant, organic vapors drive the growth of particles to larger sizes in many environments. The new aerosol processing module in CMAQ is designed for robust prediction of particle number concentrations, sources and sinks, while accounting for the participation of inorganic and organic compounds. The new module leverages the speed and flexibility of modal aerosol techniques with state-of-the-art schemes for treating new particle formation (via results from recent laboratory observations and the highly detailed Atmospheric Cluster Dynamics Code), coagulation, and intermodal transference.

We apply the new model to observations made throughout the US including California (the CalNex and CARES 2010 campaigns), the southeast (Duke Forest, NC site) and the northeast (Queens College and Pinnacle State Park Air Quality System sites) and evaluate model performance against observed number concentrations and size distributions. This combination of sites allows us to constrain the model in urban, suburban, and rural locations, which is critical since the drivers of ultrafine particle concentrations are expected to change dramatically among these receptor types. Characterization of ultrafine particle pollution with CMAQ-NPF improves our understanding of the most effective ways to mitigate the highest concentrations in the US.

Bromine and iodine chemistry update in CMAQ
Golam Sarwar1, Kathleen Fahey1, Kristen Foley1, Hosein Foroutan1,2, Rohit Mathur1, Jia Xing3, Tom s Sherwen4, Alfonso Saiz-Lopez5

1 National Exposure Research Laboratory, Environmental Protection Agency, RTP, NC
2 Virginia Tech, Blacksburg, VA
3 School of Environment, Tsinghua University, Beijing, China
4 University of York, York, UK
5 Institute of Physical Chemistry, Madrid, Spain

We previously incorporated bromine and iodine chemistry into the Community Multiscale Air Quality (CMAQ) model. Here, we further update the bromine and iodine chemistry in CMAQ. We update rate constants for the existing reactions and also include the photolysis of higher iodine oxides. We speciate sea-salt emissions into seven species, including bromide, and revise the model to simulate particulate bromide. In the updated chemistry, we incorporate several heterogeneous reactions of bromine and iodine species. On atmospheric particles, we incorporate the heterogeneous reactions of gas-phase hypobromous acid (HOBr), hypoiodous acid (HOI), iodine nitrate (INO3), and nitryl iodate (INO2) with particulate chloride and bromide. On atmospheric ice crystals, we incorporate heterogeneous reactions of HOBr and hydrobromic acid (HBr), HOBr and hydrochloric acid (HCl), and hypochlorous acid (HOCl) and HBr. We update the halocarbon and inorganic iodine emissions estimates but the inorganic bromine emissions estimate remains unchanged. We use anthropogenic emissions from the Emissions Database for Global Atmospheric Research and biogenic emissions from Global Emissions InitiAtive. We performed simulations without and with the bromine and iodine chemistry over the Northern Hemisphere for summer months using meteorological fields obtained from the Weather Research and Forecasting model using the Morrison microphysics scheme. The bromine and iodine chemistry effectively reduces ozone over both marine and land environments. However, it is more effective in reducing ozone over marine environments. While the largest impact is seen near the surface, it also reduces ozone throughout the troposphere. The updated chemistry enhances bromine oxide not only in the lower atmosphere but also aloft due to the heterogeneous reactions on ice crystals. In the poster, we will present a detailed analysis of its impact on model results and present a comparison of predicted bromine oxide, iodine oxide, and ozone with available observed data.

Machine Learning Approaches for Air Quality Modeling Applications

Satish Vutukuru, Alexander Cohen, Fernando Garcia Menendez

Traditional air quality modeling is based on mechanistic approaches that aim to simulate the underlying chemical and physical phenomena occurring in the ambient atmosphere. In recent years, machine learning and statistical modeling approaches have undergone rapid advancement. Although these approaches rely on the availability of large training datasets and are computationally intensive, they are fast becoming an important modeling tool in many scientific domains. Therefore, machine learning approaches show immense potential as an alternative for modeling ambient air quality. However, there have been only few attempts to apply the latest generation of machine learning techniques to predict ambient air quality.

In this study, we present an application of neural networks based deep learning architectures to forecast air quality in the Atlanta area using the TensorFlow machine learning library. A training dataset was constructed from output of the CMAQ model, emission estimates from SMOKE, and meteorological parameters using WRF. This machine learning model is then used to predict future air quality and evaluated using out of sample monitoring data. We show that machine learning methodologies can complement mechanistic modeling tools to better understand and predict the ambient air quality and aid public policy efforts.

FEST-C 1.3 & 2.0 for CMAQ Bi-directional NH3, Crop Production, and SWAT Modeling
Dongmei Yang1, Ellen Cooter2, Limei Ran2, Verel Benson3, Jared Bowden1, Kevin Talgo1, and Adel Hanna1
1 Institute for the Environment, University of North Carolina, Chapel Hill
2 ORD NERL/USEPA, Research Triangle Park, NC;
3 Benson Consulting, Columbia, MO

The Fertilizer Emission Scenario Tool for CMAQ (FEST-C) is developed in a Linux environment, a festc JAVA interface that integrates 14 tools and scenario management options facilitating land use/crop data processing for the Community Multiscale Air Quality (CMAQ) modeling system and Soil and Water Assessment Tool (SWAT), and visualization. FEST-C version 2.0, EPIC1102 with an improved treatment of N modeling will replace version FEST-C 1.3, EPIC0509. Extracted and summarized yearly/daily data can be used for CMAQ Bi-directional NH3 and SWAT modeling, agricultural production, and environmental impact analysis.

It is always challenging to accurately estimate ammonia emissions in space and time since fertilizer applications vary in time and quantity, across crop types and locations. In addition, soil chemical processes differ according to soil properties and management practices. A coupled atmospheric (CMAQ) and soil biogeochemical agricultural crop modeling system enables more temporally and spatially resolved soil NH3 emission estimates than estimates based on county-level fertilizer sales data and state-level climatology.

A CMAQ adjoint with aerosol capabilities

Shunliu Zhao, Matthew Russell, Amanda Pappin, and Amir Hakami (Carleton University); Matt D. Turner, and Daven K. Henze (University of Colorado); Shannon Capps (Drexel Uiversity); Peter B. Percell (University of Houston); Jaroslav Resler (ICS Prague); Jesse O. Bash, Sergey L. Napelenok, Kathleen Fahey (USEPA); Rob W. Pinder; Armistead G. Russell and 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); Daewon Byun.

We have developed a multiphase adjoint for CMAQ, which has been under rigorous testing and is now ready for release. An adjoint air quality model provides location- and time-specific gradients of an air quality metric to model inputs and lends itself to backward sensitivity analysis, source attribution, optimal pollution control, data assimilation and inverse modeling for scientific and policy applications. A gas-phase adjoint model for CMAQ was previously developed (Hakami et. al, 2007) and has been used in various applications related to ozone. The current work updates the scientific routines in the previous adjoint and extends the adjoint capabilities to involve aerosols, which have a significant impact on human health and climate. 

The adjoint model development is assisted with Automatic Differentiation (AD) tools. Code pre-processing is required for AD and one example is to address the problem associated with the bisection procedure in the aerosol thermodynamics module, ISOYYOPIA, and the secondary organic aerosol module. The adjoint code generated from AD is evaluated against the brute force Finite Difference Method (FDM) and the Complex Variable Method (CVM). Unlike the FDM, the CVM is not subject to subtraction errors and helps in situations when the FDM fails to produce accurate sensitivities.

The process-by-process validation has been insightful and has revealed a number of subtleties and/or challenges in adjoint code development or the underlying model. We have also encountered two major obstacles when testing with the full model: the exploding of gradient values over a period of run and the oscillating patterns in sensitivity field. The gradient exploding problem is addressed by treating certain intermediate variables as constants. The issue of oscillating patterns is resolved by replacing the discrete adjoint of advection with a continuous version (Hakamie et al., 2007; Henze et al., 2007). We will present the details of adjoint model validation, treatments of these issues and the implications. Finally, we will discuss possible applications of the developed CMAQ adjoint model and provide example applications to address policy and public health questions.

Modeling of reactive ammonia uptake by secondary organic aerosol in CMAQ: application to continental US
Shupeng Zhu, Jeremy R. Horne, Julia Montoya, Mallory L. Hinks, Sergey A. Nizkorodov, Donald Dabdu

Ammonium containing aerosol makes up an important fraction of total PM_2.5 mass, such as ammonium nitrate and ammonium sulfate. Those aerosols are responsible for both adverse health effects and visibility reduction. In the US, the largest ammonia emissions sources are agricultural activities, such as the intensive farming in California central valley area and industrialized hog farms in central North Carolina. The ammonia rich plumes from those areas drive most of the nitric acid into the particle phase, resulting in high PM_2.5 concentrations in those regions. While the conversion of inorganic gases into particulate phase sulfate, nitrate, and ammonium is now fairly well understood, there is considerable uncertainty over interactions between gas phase ammonia and secondary organic aerosols (SOA). Observations have confirmed that ammonia can react with carbonyl compounds in SOA. In order to investigate the importance of this process on air quality modeling, in this study, a first order loss rate for ammonia due to SOA is implemented into the CMAQ model based on the ammonia uptake coefficients reported by laboratory measurements. Simulations over the continental US are performed with a range of uptake coefficients to evaluate the sensitivity of ammonia removal due to the magnitude of the uptake coefficient. Simulation results indicate a significant reduction of gas-phase ammonia is possible due to the SOA uptake, thereby indirectly affecting the amount of ammonium sulfate and ammonium nitrate in particulate matter.

Highlights from Multiple Recent Studies of Modeling Single Source O3 and Secondary PM2.5 Impacts

Kirk R. Baker, U.S. Environmental Protection Agency, Research Triangle Park.

The U.S. Environmental Protection Agency (EPA) provides guidance on the procedures to estimate using models the potential impacts from proposed new emission sources on air quality (AQ) and air quality related values (AQRV; visibility and surface deposition) under the Prevention of Significant Deterioration (PSD) program. Some sources may need to estimate ozone and secondarily formed PM2.5 as part of the permit application process. Photochemical grid models provide a realistic chemical and physical environment for assessing changes in air quality resulting from changes in emissions. When using these tools for single source impact assessments, it is important to differentiate a single source impact from other emissions sources and to understand how well contemporary grid model applications capture near-source transport and chemistry. This poster is intended to provide highlights from recent work (Baker et al., 2014; Baker and Kelly, 2014; Baker et al., 2016; Baker and Woody, 2017; Kelly et al., 2015) focused on understanding single source impacts and where possible evaluating model skill in replicating single source impacts.

Baker, K.R., Hawkins, A., Kelly, J.T., 2014. Photochemical grid model performance with varying horizontal grid resolution and sub-grid plume treatment for the Martins Creek near-field SO2 study. Atmospheric Environment 99, 148-158.

Baker, K.R., Kelly, J.T., 2014. Single source impacts estimated with photochemical model source sensitivity and apportionment approaches. Atmospheric Environment 96, 266-274.

Baker, K.R., Kotchenruther, R.A., Hudman, R.C., 2016. Estimating ozone and secondary PM2.5 impacts from hypothetical single source emissions in the central and eastern United States. Atmospheric Pollution Research 7, 122-133.

Baker, K.R., Woody, M.C., 2017. Assessing Model Characterization of Single Source Secondary Pollutant Impacts Using 2013 SENEX Field Study Measurements. Environmental Science & Technology 51, 3833-3842.

Kelly, J.T., Baker, K.R., Napelenok, S.L., Roselle, S.J., 2015. Examining single-source secondary impacts estimated from brute-force, decoupled direct method, and advanced plume treatment approaches. Atmospheric Environment 111, 10-19.

What have we learned A review of novel data assimilation techniques for source apportionment
Cesunica E. Ivey1, Haofei Yu2, Josephine Bates2, Xinxin Zhai2, Sivaraman Balachandran3, Heather A. Holmes1, Yongtao Hu2, James A. Mulholland2, Armistead G. Russell2

1Atmsopheric Sciences Program, Department of Physics, University of Nevada Reno, Reno, NV; 2Department of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA; 3Department of Biomedical, Chemical and Environmental Engineering, University of Cincinnati, Cincinnati, OH

In recent years, the need for spatially and temporally complete source apportionment data has increased. In particular, daily spatial fields of source impact estimates are especially important for health analyses that estimate associations between human exposure to air pollutants from targeted sources and adverse health risks. Here, a review is presented for several source apportionment techniques and applications that make use of statistical data assimilation. Methods include a hybrid chemical transport-receptor modeling approach, a method for correcting biases in modeled impacts on secondary PM2.5 species, a Bayesian-based ensemble chemical mass balance approach, an integrated mobile source indicator approach, and a dispersion-chemical transport modeling technique. The methods and studies presented are critiqued for their usefulness and performance in generating spatial fields of air quality data that reflect known knowledge. Also highlighted are the new benefits that are provided by these methods, including high source specificity of spatial and temporal data, finer scale estimates of mobile source exposures, a reduction in computation time, and improved agreement with field observations. We propose recommendations for best practices on use of these novel source apportionment methods for air quality applications and for use as exposure metrics in human health studies.

The sensitivity of surface ozone concentration to geographically-distributed VOC and NOx emissions over Kao-Ping air basin in Taiwan

Ciao-Kai Liang¹, J. Jason West¹, Joshua S. Fu², Hsin-Chih Lai³, Der-Min Tsai⁴, Li-Wei Lai³

¹Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, North Carolina 27599, USA, ²Department of Civil and Environmental Engineering, The University of Tennessee, Knoxville, Tennessee 37996, USA, ³Department of Engineering & Management of Advanced Technology, Chang Jung Christian University, Tainan 71101, Taiwan, ⁴Department of Information & Communication, Kun Shan University, Tainan 71003, Taiwan

For the purpose of air pollution management, Taiwan is divided geographically into seven air basins. Kao-Ping air basin (KPAB), located in the southwestern region of Taiwan, has been in non-attainment of the ozone standard for years, mainly due to its dense population, intensive industrial output, and heavy traffic volume. In addition, its location downwind of other source regions and poor atmospheric mixing, particularly in fall, contribute to poor ozone air quality in KPAB. Tropospheric O3 is formed through complex photochemistry involving its precursors of nitrogen oxides (NOx) and volatile organic compounds (VOCs). Controlling ozone to meet the ozone air quality goal requires reductions in NOx and VOCs emissions from local and upwind sources. Here we study the effect of emissions controls on the formation of ozone in the Kao-Ping air basin under the typical stagnation conditions that existed on 1-5 October, 2010. The community multiscale air quality model (CMAQv5.0.2) with its high-order decoupled direct method (HDDM) is applied to analyze the sensitivity of ozone concentrations to changes in precursor emissions (i.e. NOx and VOCs) from local and upwind source regions and calculate the contribution from local and upwind emission source regions to ozone concentration in KPAB. CMAQ is driven by meteorological fields generated by the Weather Research and Forecasting (WRF) version 3.4.1 model. Initial and boundary conditions were derived from the global chemical transport model GEOS-Chem. Emissions from the Taiwan Emission Data System (TEDS 8.1) are processed by the Sparse Matrix Operator Kernel Emissions (SMOKE) version 3.7 for input to CMAQ. CMAQ-HDDM calculates the first and second order ozone sensitivity coefficients with respect to its precursor emissions from local (i.e. KPAB) and 4 upwind regions (i.e. North and Chu-Miao Air Basin (NCMAB), Central Air Basin (CAB), Yun-Chia-Nan Air Basin (YCNAB), and Yi-Lan and Hua-Dong Air Basin (YLHDAB), and initial and boundary conditions. A backward trajectory analysis was also employed to identify the pathway for the accumulated ozone. The results are expected to determine which emission sources are responsible for ozone formation in KPAB quantitatively and to characterize ozone sensitivities, to support hypothetical emission reduction management strategies.

Elevated Ozone Along the Coastline of Lake Michigan

Jenny Liljegren, Kirk Baker

This presentation will include information regarding elevated ozone observed along the coastline of Lake Michigan relative to areas farther inland, provide an overview of the meteorology and atmospheric chemistry contributing to elevated ozone along the shoreline, and describe the history and regulatory framework associated with controlling ozone precursors in this region. A photochemical model simulation will be compared to ground-based monitor data as well as aircraft observations over the lake.

Speciated Source Apportionment of PM2.5 using Positive Matrix Factorization
J. Hegarty

The EPA Positive Matrix Factorization (PMF) tool is used to identify the major species source contributions to PM2.5 concentrations in BIBE. The PMF analysis identified six source contribution factors. Results are consistent with earlier results conducted as part of the Causes of Haze Assessment study (COHA). However, an important difference is that the proportional contribution distribution from the earlier period showed no sensitivity to the PM2.5 concentrations. We also investigate potential source regions are investigated using footprints generated with the STILT model.

Modified BART Analysis Methodology for the Regional Haze Rule

Neelesh Sule, PhD, PE, Senior Consultant, Trinity Consultants

Christine Chambers, Principal Consultant, Trinity Consultants

Jeremy Jewell, Principal Consultant, Trinity Consultants

The Regional Haze Rule (RHR) established a comprehensive visibility protection program for Class I areas and requires States to set reasonable progress goals (RPGs) towards achieving natural visibility conditions for Class I areas within their borders. States must evaluate emissions controls through Best Available Retrofit Technology (BART) determinations and Reasonable Progress (RP) analyses for stationary sources emitting air pollutants that cause or contribute to visibility impairment in a Class I area. RPGs are intended to be used for comparison between a state's long-term strategy and the uniform rate of progress (URP) curve, while providing a target by which a state's Regional Haze State Implementation Plan (SIP) can be assessed for adequacy. For RP and recent BART analyses, the Comprehensive Air Quality Model with Extensions (CAMx) modeling system can be used to assess Class I area visibility impacts from individual sources coupled with the cumulative visibility impact of all other emissions sources including naturally occurring events and international emissions. CAMx is a rigorous modeling system that can reliably predict visibility impacts, and that prediction reliability is improved through correlating the modeled data with available monitored data. Currently, EPA's post-processing methodology for BART analyses is focused on finding the absolute highest modeled impact without tethering the modeled results with monitored data. With Trinity's proposed post-processing methodology, the CAMx modeled output is tethered with monitored data to provide robust results that compensate for the potential model bias issues that the current BART methodology fails to address. This study will discuss the BART analysis methodologies and results for post processing CAMx outputs for sources in EPA Region 6.

Characterization of CMAQv5.2: Science Processes, Performance, and Development

Ben Murphy, Bill Hutzell, David Wong, Kristen Foley, Havala Pye, Hosein Foroutan, Wyat Appel, Christian Hogrefe, Deborah Lueken, Kathleen Fahey, Jesse Bash, Donna Schwede, Matthew Woody, Tanya Spero, Sergey Napelenok, Chris Nolte, Limei Ran, Shawn Roselle and Jon Pleim

National Exposure Research Laboratory, Office of Research and Development, US EPA

CMAQv5.2 was officially released June 30, 2017, after an unofficial "beta" release in October 2016. Several important improvements were implemented in this recent major version, including updates to the treatment of gas and aerosol processing, meteorology, and atmosphere-surface interactions. We will summarize the motivation and impact of many of these improvements including new algorithms for treating organic aerosol production, wind-blown dust emission, and lightning-NO _x generation among others. We will also demonstrate the capabilities for ozone and hazardous air pollutant predictions with the new Carbon Bond 6 gas-phase chemical mechanism (CB6r3). In addition to science updates, recent model developments have yielded reductions in computational burden (which we quantify) and improved resources documenting both the theory and structure of the model. Finally, the CMAQ development teams at EPA and UNC IE are now routinely leveraging GitHub and its interface to support ongoing model development - we will summarize the most important recommendations for users wanting to interact with the CMAS community through this framework and make contributions to the model system.

Development of a Year 2016 Emissions Modeling Platform

Alison Eyth, United States Environmental Protection Agency, Research Triangle Park, NC.
Jeff Vukovich, United States Environmental Protection Agency, Research Triangle Park, NC.
George Pouliot, United States Environmental Protection Agency, Research Triangle Park, NC.
Madeleine Strum, United States Environmental Protection Agency, Research Triangle Park, NC.
Chris Allen, CSRA, Research Triangle Park, NC.
James Beidler, CSRA, Research Triangle Park, NC.

The US EPA has developed a year 2016 emissions modeling platform to support modeling studies of that year. The starting point for the platform was version 1 of the 2014 National Emissions Inventory (2014NEIv1), but specific updates were made to better represent the year 2016. Fire emissions for 2016 were derived from SMARTFIRE in the U.S. and the FIre INventory from NCAR (FINN) data set in Canada and Mexico. Day-specific agricultural fire emissions for 2016 are included. Electric generating unit (EGU) emissions for the year 2016 were incorporated based on continuous emissions monitoring system (CEMS) data. Locomotive and agricultural ammonia emissions were updated from those used in 2014NEIv1. Onroad mobile source activity data were updated and oil and gas emissions were adjusted to better represent the year 2016. This platform includes both 36km and 12km grids and new spatial surrogates were developed for these grids for use in this and other year 2014-2016 cases. Updates were implemented related to the spatial allocation and plume rise for commercial marine vessel emissions. Ancillary data files for speciation and temporal allocation were updated for several source categories. Canadian emissions were updated to represent the year 2013 from the previously used year 2010 data.

Enabling sensitivity analysis in CMAQ with the complex-step approach

Isaiah Sauvageau1, Bryan Berman1, Shunliu Zhao2, Amir Hakami2, Daven Henze3, and Shannon Capps1

1 Drexel University

2 Carleton University

3 University of Colorado Boulder

The application of the decoupled direct method (DDM) in CMAQ has proven to be one of the most widely-used augmentations of the model for regulatory applications and scientific investigations. DDM enables simultaneous evaluation of the influence of a few parameters on all modeled concentrations at a relatively low computational cost, which is similar to a tangent linear model. These sensitivities are instrumental in evaluating the effectiveness of emissions controls. They can also expedite the process of discovering how the model can be improved by mechanistic modifications. More recently, data assimilation approaches that leverage tangent linear models have shown promise.

The wide adoption of this auxiliary component of the model necessitates that new processes added to CMAQ be included in DDM. The computational advantages of DDM arise from development of the analytical derivative of each module or development in CMAQ, which can be very time consuming and technically difficult depending on the structure of the algorithm. This work introduces an alternative approach to calculating the sensitivity of a select model parameter with respect to all modeled concentrations. The complex-step method (Squire and Trapp, 1998) employs the imaginary component of complex numbers as a space for propagating sensitivity information. When CMAQ is equipped to use complex variables, the work of incorporating new modules into the sensitivity analysis framework is greatly simplified. This work will demonstrate the complex-step approach in CMAQ and discuss its utility.

Computational Enhancement Approach in Mobile Emissions Simulation: Parameterization of MOVES Emission Factors Lookup Tables
B.H. Baek, Alejandro Valencia and Michelle Snyder
University of North at Chapel Hill, Chapel Hill, NC

Soontae Kim and Changhan, Bae
Ajou University, South Korea

Recent versions of the Community Multiscale Air Quality (CMAQ) model have been integrating its chemistry and transport processes with meteorologically-driven emission processes which are biogenic emissions, bi-directional NH3 from fertilizer applications, and point source plume rise calculation over the years. Simulating emissions inline in CMAQ is especially 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 mobiles, vegetation, fertilizer applications, and wildfires. Especially, the onroad mobile sector, which is one of the largest emitters of NOx, PM2.5, and CO, has significant temperature and humidity dependencies, particularly for NOx, VOC, and PM2.5 emissions. We propose the plan to enhance SMOKE and CMAQ models to process mobile emissions efficiently inline using the state-of-the-science Motor Vehicle Emissions Simulator (MOVES) emission factors for a better representation of the meteorological influences on emissions without any computational bottlenecks.

Satellite Cloud Assimilation - The Impact of Improved Cloud Fields on 2013 Air Quality Simulations

Arastoo Pour Biazar, Maudood Khan, Andrew White, Richard T. McNider, Bright Dornblaser, Yuling Wu, Peiyang Cheng

Despite many efforts to improve cloud simulation, numerical weather prediction (NWP) models still have difficulty in creating clouds in the right place and at the right time compared to observations. This in turn constitutes a major source of uncertainty for air quality simulations. Clouds impact biogenic hydrocarbon emissions, photolysis rates, boundary-layer development, vertical mixing, and induce aqueous phase chemistry. Thus, poor representation of clouds impacts the photochemical model's ability to adequately simulate the air quality.

The utilization of satellite cloud observation in NWP models can potentially improve model cloud simulation in retrospective studies. Our group at UAH has developed a technique to assimilate Geostationary Operational Environmental Satellite (GOES) derived cloud fields within Weather Research and Forecasting (WRF) model to improve simulated clouds. Based on GOES observations, the assimilation technique dynamically supports cloud formation/dissipation within WRF. The technique had previously been implemented and tested successfully in WRF for a month-long simulation during August 2006. In the current study, Cloud Assimilation System (CAS) was used in an air quality simulation over the period of August-September 2013 (NASA's Discover-AQ field campaign). The cloud assimilation on the average improved model cloud simulation by 15%. The cloud correction not only improved the spatial and temporal distribution of clouds, it also improved boundary layer temperature, humidity, and wind speed.

The improved meteorological fields were used for air quality simulations, using WRF/SMOKE/CMAQ modeling system. The improvements in meteorological fields directly impacted the air quality simulations and altered trace gas concentrations. For the summer of 2013 (July-Sept), in general WRF model underestimated cloud cover over the East/Southeastern United States. Due to cloud under-prediction, the control simulation over-estimated isoprene emissions. The simulation with cloud assimilation partially corrected this error and reduced isoprene emissions by about 30%. Over the period of study, by slowing down the photochemistry and decreasing BVOC emissions, cloud correction also reduced O3 bias over the S.E. United States by about 60%. The photochemical correction was episodic and more pronounced where the dominant model error was due to errors in cloud placement and timing and in areas with large emission source. Preliminary results from this study will be presented.

The Load of Lightning-induced Nitrogen Oxides and Its Impact on the Ground-level Ozone during Summertime over the Mountain West States

Daiwen Kang, Rohit Mathur, George Pouliot, and David Wong

Lightning-induced nitrogen oxides (LNO_X), in the presence of sunlight, volatile organic compounds and water, can be a relatively large but uncertain source for ozone (O₃) and hydroxyl radical (OH) in the atmosphere. Using lightning flash data from the National Lightning Detection Network (NLDN) with an updated LNO_X emission estimation algorithm in the Community Multiscale Air Quality (CMAQ) model, we estimate the hourly variations in LNO_X emissions for the summer of 2011, compare with anthropogenic and soil sources, and simulate its impact on distributions of tropospheric O₃ across the continental United States. We find that typical summer-time lightning activity across the Western U.S. injects NO_x emissions comparable to that from anthropogenic sources into the troposphere over the region. Comparison of two model simulation cases with and without LNO_X emissions show that significant amount of ground-level O₃ in the Western U.S. during the summer can be attributed to the lightning NO_X emissions. The model simulated surface-level O₃ in the case with lightning NO_X emissions matched the observed values much closer than the model case without lightning NO_X emissions. Periods of significant reduction in bias in simulated O₃ between these two cases strongly correlates with the periods when lightning activity occurred in the region. The inclusion of LNO_X increased daily maximum 8-hour O₃ by up to 17 ppb and improved model performance relative to measured surface O₃ mixing ratios in the Western U.S. region. The magnitude of LNO_X emissions estimated for other summers is comparable to the 2011 estimates suggesting that summertime surface-level O₃ levels in the Western U.S. region could be significantly influenced by lightning NO_x and needs to be accurately characterized in assessments of O₃ background values in the region.

Advanced Land Surface Processes in the Coupled WRF/CMAQ with MODIS Input

Limei Ran¹, Robert Gilliam¹, David Wong¹, Hosein Foroutan², Jonathan Pleim¹, George Pouliot¹, Wyat Appel¹, Daiwen Kang¹, Shawn Roselle¹, Brian Eder1, Ellen Cooter¹

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

²Department of Biomedical Engineering and Mechanics, Virginia Tech, Blacksburg, VA, USA

Land surface modeling (LSM) is important in WRF/CMAQ for simulating the exchange of heat, moisture, momentum, trace atmospheric chemicals, and windblown dust between the land surface and the atmosphere. Vegetation and soil treatments are crucial in LSM for surface energy budgets and water and carbon cycles. Vegetation is a source and sink of many atmospheric pollutants and precursor chemicals such as O₃ and volatile organic compounds while soil properties directly influence windblown dust emissions. There are recent improvements to the vegetation and soil processes and ozone dry deposition on the bare soil in the offline WRF/CMAQ modeling system through the use of Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation product. This presentation highlights the use of MODIS vegetation, improved soil treatment, and a simple irrigation scheme in a coupled-WRF/CMAQ which is updated with the recent improvements.

We will demonstrate the improvements in surface meteorology, O₃, and PM_2.5 simulations using the updated coupled-WRF/CMAQ. Results from simulations covering continental U.S. with 12 km grids for April and August 2016 and 2017 will be presented. In addition, simulation results for May and June 2010 covering California with 4 km grids during the CalNex campaign period will be demonstrated with the irrigation option. Distinct improvements in dust emissions estimation are observed with reduced PM_2.5 high bias during dust outbreaks. Surface O₃ estimation is reduced in areas with dominant bare land, and surface O₃ simulation is improved in areas with better vegetation representation from MODIS during the green up season. The simple irrigation scheme improves 2-m temperature and mixing ratio results in irrigation-dominant crop lands. The influence on air quality will be evaluated against the CalNex campaign observations.

Effect of top-down emission adjustment using OMI NO2 and HCHO column densities in South Korean air quality simulation

Changhan Bae¹, Hyun Cheol Kim^2,3, Byeong-Uk Kim⁴, Chul Yoo⁵, Yongjae Lim⁶, Jaebum Lee⁶ and Soontae Kim¹

The accuracy of input data seriously affects the performance of air quality simulation. Emissions data, one of the main input data for air quality simulation, are usually constructed using a bottom-up approach, which uses numerous basic socio-economic information (i.e. population and activities) or emission factors. Although the bottom-up emission estimation provides more complete information for the detailed features in anthropogenic and/or natural emission release, it is also seriously limited due to inherent uncertainties in basic information or lack of timely update.

In this study, we try to supplement the existing bottom-up emissions inventory by combining factors estimated from satellite-based measurements. First, base air quality simulations were conducted using CAPSS 2013 and CREATE 2013 emissions inventories for South Korea and other Asian countries, respectively. Second, Ozone Monitoring Instrument NO₂ and HCHO column density observations were compared to the base simulation. The ratio between the base simulation and satellite observations were used to adjust current bottom-up emissions inventory (e.g. adjusted run). MEGAN was used for base biogenic emission estimation, and only isoprene emissions were adjusted using the modeled-observed HCHO column ratios. WRF version 3.6.1 and CMAQ version 4.7.1 were used for meteorological and chemistry simulation over 27-km East Asia domain and 9-km South Korea domain during May-Aug 2015. Using the top-down emission adjustment, the simulated period-mean surface NOx concentration in South Korea increased by 8.8 ppb from 29.7 ppb 38.5 ppb, and the model bias was significantly improved from 7.8 ppb to -0.9 ppb.

Mechanistic representation of Soil N in CMAQ v5.1

Quazi Ziaur Rasool, Jesse Bash, Ellen Cooter and Daniel S. Cohan

Recent global nitrogen (N) budgets estimate that soil reactive N emissions (predominantly from biochemical transformations in soil) have increased by a factor of 2-3 from pre-industrial levels. These increases are especially pronounced in agricultural regions. The reactive N emissions from biogeochemical transformations can be in reduced (NH₃) or oxidized (NO, HONO, N₂O) form, depending on complex biogeochemical transformations of soil N reservoirs. Air quality models like CMAQ typically neglect soil emissions of HONO and N₂O. Even our previous update of the soil NO scheme in CMAQ estimated emissions parametrically, in a manner inconsistent with soil NH₃ emissions. Thus, there is a need to more mechanistically and consistently represent the soil N processes that lead to emissions to the atmosphere. Our updated approach estimates soil NO, HONO and N₂O emissions by incorporating detailed agricultural fertilizer inputs from EPIC and CMAQ modeled N deposition into the soil N pool. EPIC addresses the nitrification, denitrification and volatilization rates along with soil N pools for agricultural soils only. The dynamic soil properties and nutrient (C and N) data for non-agricultural areas are hence used from global soil property and nutrient database available in literature. The EPIC-CMAQ framework utilizes bi-directional exchange only for NH₃, which will be extended to other species. The NO and N₂O emissions from nitrification and denitrification are computed mechanistically using the N sub-model of DAYCENT. These mechanistic definitions use soil water content, temperature, NH₄⁺ and NO3- concentrations, gas diffusivity, and labile C availability as dependent parameters at various soil layers. Soil HONO emissions will be estimated from the total nitrification NO_x emission estimated from DAYCENT N sub-model. The model sets the ratio of HONO vs NO emissions based on soil moisture, pH and land use definitions modeled at sub-grid scales. We implement the scheme in CMAQ for the continental US. Comparison of the model estimated emission rates from the new mechanistic scheme with other existing schemes and measurements in the literature will be discussed.

Impact on Recent North American AQ Forecasts of Replacing a Retrospective U.S. Emissions Inventory with a Projected Inventory

Michael D. Moran, Qiong Zheng, Junhua Zhang, Radenko Pavlovic, and Mourad Sassi

To make AQ forecasts we generally have to use emissions inventories that are retrospective, that is, representative of a past base year. When emissions are changing, however, retrospective inventories often do not reflect current emission levels very well, as is the case in North America given the continuing decrease of criteria-air-pollutant emissions in both Canada and the U.S. since 1990. One way to deal with this mismatch is to use a projected, future-year emissions inventory that has been generated by incorporating assumptions about expected changes to the economy, to population and housing, to the on-road and off-road vehicle fleets, etc. along with any emissions changes expected to result from the phased implementation of existing AQ control legislation.

Since 2015, in order to make operational North American AQ forecasts with the Environment and Climate Change Canada regional AQ prediction system, we have been using emissions files based on version 1 of the 2010 Canadian Air Pollutant Emission Inventory (APEI), version 1 of the 2011 U.S. National Emission Inventory (NEI), and version 1 of the 1999 Mexican inventory. Recently, however, we tested the impact of adopting new emissions files based on the 2013 Canadian APEI, a projected 2017 U.S. inventory based on version 3 of the 2011 U.S. NEI, and the 2008 Mexican inventory. For the continental U.S. the change in inventories reduced NOx, SO2, VOC, and CO annual anthropogenic emissions by 33%, 65%, 11%, and 24%, respectively. The use of these new inventories has resulted in an overall improvement in our AQ forecasts for Canada and the U.S., including ozone forecasts over the eastern U.S. More details will be shown in this presentation.

Advances the National Air Quality Forecast Capability predictions

Ivanka Stajner¹, Jeff McQueen², Pius Lee³, Jianping Huang^2,5, Li Pan^3,6, Ho-Chun Huang^2,5, Daniel Tong^3,6, Ariel Stein³, James Wilczak⁴, Irina Djalalova^4,8, Dave Allured^4,8, Phil Dickerson⁷, Sikchya Upadhayay^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

NOAA provides operational air quality 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 at http://airquality.weather.gov. The National Air Quality Forecast Capability (NAQFC) providing these predictions was most recently updated in June 2017. Ozone and PM2.5 predictions are produced using a system operationally 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. In addition, smoke and dust predictions are separately produced by NOAA's HYSPLIT model.

An overview of recent updates and evaluations of air quality predictions will be presented. Recent updates to NAQFC predictions have focused on updating CMAQ to version 5.0.2. Both HYSPLIT and CMAQ model are now driven by the new version of NAM 4.0, and include wildfire smoke emissions from a newer version of the USFS BlueSky system. The new configuration of the CMAQ system is initialized by rerunning it over the previous 24 hours to include wildfire emissions at the time when they were observed from the satellites, to better represent wildfire emissions prior to beginning 48 hour predictions. Post processing to reduce the bias in PM2.5 predictions has been further refined using the Kalman filter analog (KFAN) technique to better capture day-to-day variability. Dust related aerosol species at the CMAQ domain lateral boundaries are updated to use the NEMS Global Aerosol Component (NGAC) v2 forecasts.

Testing is ongoing to extend CMAQ predictions to 72 hours. Fire emissions data from Environment and Climate Change Canada (ECCC) also are being tested. Finally, the KFAN technique is being modified and tested to improve ozone predictions.

Impact of a Major Update to VOC Emissions Processing on Regional Air Quality Model Predictions

Junhua Zhang, Michael D. Moran, Paul A. Makar, and Qiong Zheng

Volatile organic compounds (VOCs) are primary precursors to the formation of both ozone and particulate matter. Emissions inventories usually report bulk VOC emissions whereas a more detailed description of VOC species is required by air quality (AQ) models. Process-specific VOC speciation profiles are typically used to disaggregate bulk VOC emissions into multiple model VOC species whose definition depend on the AQ model's gas-phase chemistry mechanism. The Global Environmental Multiscale - Modelling Air-quality and CHemistry (GEM-MACH) AQ modelling system developed by Environment and Climate Change Canada employs the ADOM-2 chemistry mechanism. Recently, the ADOM-2-specific VOC speciation profile library for GEM-MACH was updated from an old version based on the U.S. EPA's SPECIATE V3.2 library to a new version based on SPECIATE V4.5. New versions of the speciation-profile cross-reference file that links individual VOC profiles to emissions source types by SCC code and sometimes country, state, and county codes and the VOC-to-TOG conversion factor file have also been adopted based on the recent EPA air emissions modeling 2011 Version 6.3 Platform (https://www.epa.gov/air-emissions-modeling/2011-version-63-platform). With these updates, the total number of VOC profiles used for processing Canadian, U.S., and Mexican VOC emissions was increased from 287 to 335, of which 86 are newer versions of existing VOC profiles or new profiles altogether, including profiles for oil and gas fracking. This major update had a significant impact on emissions of the ADOM-2 model VOC species over the North America continental model domain, including a 90% increase in propane emissions and a 34% and 8% decrease in toluene and higher-alkene emissions, respectively. In this presentation, the impact of using these updated VOC emissions on AQ predictions by the GEM-MACH model will be shown for a summer period. Results from sensitivity studies for key sectors with large VOC emissions, such as the oil and gas industry, will be discussed as well.

AQcast: A Scalable, Automated System for Photochemical Modeling on the Amazon Cloud

Matthew J. Alvarado, Erik J. Fanny, Ethan H. Fahy, Chantelle R. Lonsdale, and Elizabeth S. Bettencourt

Cloud computing allows for an arbitrarily large number of air quality simulations to be performed simultaneously, thus reducing the time and cost of completing air quality modeling studies. However, taking full advantage of cloud computing requires the development and use of new software tools to run the models, access and store the input and output datasets, and to evaluate the results. In addition, running the variety of models and preprocessors necessary to prepare the inputs for photochemical modeling can be complex and labor intensive. Here we present a scalable, cloud-based system that automates most of the needed steps for photochemical modeling: collecting the needed inputs from publically available sources or in-house archives; running the model; and performing some automatic analysis and quality checks on the output. This system, named AQcast, builds upon AER's successful nCast system for the Weather Research and Forecasting (WRF) model. AQcast v1.0, which can be run for any region in the continental US, consists of a meteorological component based on nCast WRF, an emissions component based on the US EPA National Emission Inventory (NEI) modeling platform, and a chemical transport model component based on the Community Multiscale Air Quality (CMAQ) model and the associated pre-processors. The input options for all components are read from a single XML file generated by a web-based interface, thereby allowing both novice and expert users to submit their own photochemical modeling jobs to be run on the system. We will present modeling results from AQcast v1.0 for NH3 during the NOAA SENEX campaign in the southeast US and describe our plans to expand the system to perform photochemical modeling over any region of the globe.

The leaf enclosure measurement for emission factors of Marijuana

Chi-Tsan Wang, Christine Wiedinmyer, Kirsti Ashworth, Peter Harley, John Ortega, William Vizuete*

In 2014, Colorado became the first US state to legalize the industrial-scale cultivation of marijuana plants. There are now more than 700 marijuana cultivation facilities (MCFs) in operation in the greater Denver area. High concentrations of biogenic volatile organic compounds (VOCs), predominantly monoterpenes (C₁₀H₁₆) such as alpha-pinene, myrcene, and limonene have been observed in the grow rooms of MCFs, suggesting MCFs have the potential to release a significant amount of reactive VOCs into the atmosphere. Further, many MCFs are located in the urban core, where other urban emission sources are concentrated, resulting in interactions which can lead to the formation of ozone, impacting air quality. The little research done on marijuana has focused on indoor air quality and occupational exposure, or identification of the compounds associated with the characteristic smell of marijuana. No previous studies have identified or quantified the monoterpene emission rates from marijuana. Here, we collected air samples from leaf enclosures around different marijuana clones at different growth stages onto sorbent cartridges. These samples were analyzed using GC-MS/-FID to identify and quantify the VOCs emitted by growing marijuana plants. Base emission factors at standard conditions were back-calculated for each clone at different life periods for use in biogenic emissions and air quality models.

Development of the Next Generation Air Quality Modeling System

Jonathan E. Pleim¹, Robert C. Gilliam¹, David Wong¹, Hosein Foroutan², Jerold A. Herwehe¹, O. Russell Bullock Jr.¹, George Pouliot¹, Christian Hogrefe¹, and Limei Ran¹

¹National Exposure Research Laboratory, U.S. Environmental Protection Agency,

Research Triangle Park, North Carolina, USA

²Department of Biomedical Engineering and Mechanics, Virginia Tech,

Blacksburg, Virginia, USA

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 (eventually) local scales. We envision that 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 although we may also use a regional version of MPAS that has recently been developed at NCAR.

In the presentation we will describe our modifications to MPAS to improve its suitability for retrospective air quality applications and show evaluations of global and regional meterological simulations. Our modifications include addition of physics schemes that we developed for WRF that are particularly designed for air quality applications: the Pleim surface layer (PSL), the Pleim-Xiu (PX) land surface model with fractional land use for a 40-class National Land Cover Database (NLCD40), and the Asymmetric Convective Model 2 (ACM2) planetary boundary layer scheme. We also added analysis nudging four-dimensional data assimilation (FDDA) to control error growth for long term retrospective simulations. In addition, we updated the KF scheme in MPAS to the latest version which adds subgrid-scale cumulus cloud feedback to the radiation schemes, multiple convective triggers, and a scale-aware convective time scale (Jerry Herwehe will present this work). We will also show preliminary MPAS-AQ results where we have incorporated CMAQ modules for atmospheric chemistry and deposition in MPAS.

Modeling crop residue burning experiments and assessing the fire impacts on air quality

Luxi Zhou1, Kirk R. Baker1, Sergey Napelenok1, George Pouliot1, Robert Elleman2, Susan O'Neil3, Shawn Urbanski4, David C. Wong1

1. U.S. Environmental Protection Agency, Research Triangle Park, Research Triangle Park, NC 27711, USA.

2. U.S. Environmental Protection Agency, Region 10, Seattle, WA 98101, USA.

3. U.S. Forest Service, Pacific Northwest Research Station, Seattle, WA 98103, USA.

4. U.S. Forest Service, Rocky Mountain Research Station, MT 59808, USA.

Crop residue burning is a common land management practice that results in ambient emissions of a variety of primary and secondary pollutants with negative health impacts. The chemical transport model CMAQ (Community Multiscale Air Quality) is used to simulate crop residue burning experiments in southeast Washington state and northern Idaho from the summer of 2013 with 2 km grid spacing. Ground and airborne measurements from the field experiment are used to evaluate the model performance in capturing surface and aloft impacts from the burning events. The results suggest that improvements to the current parameterizations are needed in order for CMAQ to reliably reproduce smoke plumes from burning. In addition, there is enough variability in the smoke emissions, stemming from variable field-specific information such as field size, that attempts to model crop residue burning should use field-specific information whenever possible.

Aerosol's Direct Feedback Effects in the U.S. using a Two-Way Coupled WRF-CMAQ Model driven by MEYYA - a global reanalysis dataset
Chowdhury G. Moniruzzaman, Jared Bowden and Saravanan Arunachalam
Institute for the Environment, The University of North Carolina at Chapel Hill

Aerosol or particulate matter (PM) in air scatters and absorbs radiation from sun which blocks incoming short-wave radiation reaching earth surface, known as aerosol direct effects (ADEs), thus affecting ozone (O3) photolysis reaction and surface temperature which in turn affects PM concentration. The changed PM concentration further affects chemistry and meteorology by ADEs through a feedback cycle. The aerosol direct feedback effects (ADFEs) are neglected in traditional air quality models (where meteorology is used as input and not affected by chemistry). In this study, a coupled Weather Research and Forecasting - Community Multiscale Air Quality (WRF-CMAQ) modeling system was used to determine the ADFEs on surface O3 and PM2.5 change as well as change of meteorological variables such as short-wave radiation (SWR) at surface, temperature at 2 m (T2) and planetary boundary layer (PBL) height in continental USA in a 36-km grid domain for the year 2005. The novelty of this new application is that the WRF fields were driven using the global-scale NASA s Modern Era Reanalysis for Research and Applications (MEYYA) meteorological data. We found that ADFEs cause annual perturbation of domain average of O3 and PM2.5 by -0.4 ppb (daily domain minima and maxima from -23 to +22 ppb) and +0.28 micro-g/m3 (daily domain minima and maxima from -32 to +25 micro-g/m3) respectively in 2005 in continental USA. ADFEs perturb the meteorological variables: SWR, T2, and PBL (domain-wide annual average) by -7.37 W/m2, -0.47 K and -21 m respectively. We also found that as high as 60% of PM2.5 increases by the ADFEs during summer were caused by secondary organic aerosol increase. We extended this application to assess the incremental air quality impacts due to a single source sector aircraft emissions during landing and takeoff (LTO) cycles and found that ADFEs in this coupled WRF-CMAQ application decrease aircraft-attributable surface O3 and PM2.5 change by 21% and 23% respectively.

Implications of emission inventory choice for modeling fire-related pollution in the U.S.

Shannon N. Koplitz and Christopher G. Nolte

Fires are a major source of fine particulate matter (PM_2.5), one of the most harmful ambient pollutants for human health globally. Within the U.S., fire emissions can account for more than 30% of total PM_2.5 emissions annually. In order to represent the influence of fire emissions on atmospheric composition, regional and global chemical transport models (CTMs) rely on fire emission inventories developed from estimates of burned area (i.e. fire size and location). Burned area can be estimated using a range of top-down and bottom-up approaches, including satellite-derived remote sensing and on-the-ground incident reports. While burned area estimates agree with each other reasonably well in the western U.S. (within 20-30% for most years during 2002-2014), estimates for the southern U.S. vary by more than a factor of 3. Differences in burned area estimation methods lead to significant variability in the spatial and temporal allocation of emissions across fire emission inventory platforms. In this work, we implement fire emission estimates for 2011 from three different fire emission products - the USEPA National Emission Inventory (NEI), the Fire INventory of NCAR (FINN), and the Global Fire Emission Database (GFED4s) - into the Community Multiscale Air Quality (CMAQ) model to quantify and characterize differences in simulated fire-related PM_2.5 and ozone concentrations across the contiguous U.S. due solely to the emission inventory used. Understanding the sensitivity of modeling fire-related PM_2.5 and ozone in the U.S. to fire emission inventory choice will inform future efforts to assess the implications of present and future fire activity for air quality and human health at national and global scales.

Development and validation of a rapid urban scale dispersion modelling platform

Dr Scott Hamilton, Dr Nicola Masey (both Ricardo Energy and Environment, UK)

We present the development of an air quality model (RapidAir) that can estimate air pollution emissions and concentrations at fine spatial resolution over large geographical areas with fast run times. The road traffic emissions model in RapidAir uses open source python libraries (e.g. numpy, pandas) to provide emissions for many tens of thousands of road links in seconds. RapidAir utilises the USEPA AERMOD code controlled by custom python modules to provide dispersion kernels, thus enabling efficient computation of results across large domains at high resolution. RapidAir was evaluated by estimating NO_x and NO₂ at 86 continuous monitoring sites in London, UK. Concentrations were modelled at sub 5m spatial resolution over an area of ~3,500 km² in < 10 minutes per run (including emissions calculation). We tested the effect of including discrete canyon models or alternatively using geospatial surrogates (sky view factor, hill shading and wind effect) to improve the accuracy of model predictions at kerbside locations. Geospatial surrogates provide alternatives to discrete street canyon models where it is not practical to run canyon models for hundreds to thousands of separate streets within a large city dispersion model (with advantages including: ease of operation; faster run times; and more complete / transparent treatment of building effects with no edge effects). Other examples will be presented involving domains up to 10,000 km2 with higher spatial resolution and up to 600 million discrete spatial predictions. Finally we present the development of an operational version of the platform with chemical boundary conditions provided by the CMAQ model coupled to RapidAir. We anticipate that the coupled model can be applied in air quality forecasting applications in cities where road traffic remains a key driver of air pollution exposures.

Emission Control Benefits of Lowering Ambient PM Exposure in Canada and the U.S.: Application of CMAQ-Adjoint for Aerosols

Amanda Pappin, Masoud Nasari, Mieczyslaw Szyszkowicz, and Rick Burnett (Health Canada), Shunliu Zhao, Burak Oztaner, Marjan Soltanzadeh, Matthew Russell, Amir Hakami (Carleton University), Matt Turner and Daven Henze (University of Colorado), Shannon Capps (Drexel University), Peter Percell (University of Houston), Jaroslav Resler (ICS Prague), Ted Russell and Athanasios Nenes (Georgia Tech), Sergey Napelenok, Jesse Bash, and Kathleen Fahey (EPA), Jaemeen Back, Greg Carmichael, and Charlie Stanier (University of Iowa), Adrian Sandu (Virginia Tech), Tianfeng Chai (University of Maryland)

The health benefits of emissions control depend both on atmospheric conditions along the pathway from source to receptors and concentration-response functions (CRFs) drawn from epidemiological studies. Recently, new forms of non-linear CRFs have been suggested for mortality and long-term PM2.5 exposure in a population-based Canadian cohort. Such a shape of the CRF indicates a heightened sensitivity of populations to changes in exposure at lower ambient concentrations. We estimate the benefits-per-ton of primary particles and precursor emissions control for reducing the public health burden of PM2.5 exposure in Canada and the U.S. We employ various forms of CRFs (single/multi-pollutant and linear/log-linear models) to capture the toxicity of the ambient pollutant mixture. We estimate benefits-per-ton using adjoint sensitivity analysis in CMAQ-adjoint (version 5.0) for aerosol processes. We estimate cause-specific mortality due to long-term PM2.5 (as well as NO2 and O3) exposure in Canada and the U.S., attributed to 1 ton of primary PM and precursor species emitted in 2013 at a 12-km resolution. Our initial results indicate significant and widespread long-term benefits of PM control, particularly in major urban areas of Canada. In our preliminary results we estimate more than 9,400 annual deaths that are attributable to sources in Canada and the U.S., to which primary elemental and organic carbon are the major contributors. We find benefits-per-ton upwards of $600,000/ton and $800,000/ton for Toronto, Ontario through primary organic and elemental carbon emissions control. We will further explore how the choice of CRF impacts health benefit estimates. We argue that unlike the traditional view of benefits-per-ton, which dictates a focus on emitters within urban areas and immediately upwind, a supralinear CRF also draws attention to emissions located in moderately populous rural areas.

Air quality impacts of projections of natural gas-fired distributed generation

Jeremy R. Horne, Marc Carreras-Sospedra, Donald Dabdub, Paul Lemar, Uarporn Nopmongcol, Tejas Shah, Greg Yarwood, David Young, Stephanie L. Shaw, Eladio M. Knipping, K. Max Zhang, Bo Yang

A variety of electricity industry drivers, including the availability of low-cost natural gas, are creating a renewed interest in natural gas fired distributed generation and utilities are interested in re-benchmarking the cost, performance, and potential air quality impacts of distributed generation options. This study assesses the potential impacts on emissions and air quality from the increased adoption of natural gas-fired distributed generation of electricity (DG) both at the local level for single source case studies and at the regional level for a widespread penetration scenario, including displacement of power from central power generation, in the contiguous United States. The single source case studies simulate the impacts on local air quality of (1) a simple cycle combustion turbine with and without heat recovery using a combination of the AERMOD air dispersion model and a computational fluid dynamics (CFD) model, and (2) a distributed combined heat and power (CHP) facility using on-site emission measurement and the CFD model. The regional study includes four major tasks: (1) modeling of distributed generation market penetration; (2) modeling of central power generation systems; (3) modeling of spatially and temporally resolved emissions; and (4) photochemical grid modeling to evaluate the potential air quality impacts of increased DG penetration, which includes both power-only DG and combined heat and power (CHP) units, for 2030. Overall, regional air quality impacts from DG vary greatly based on meteorological conditions, proximity to emissions sources, and the number and type of DG installations.

Multi-Year (2013-2016) North American Population Exposure to PM2.5 Emissions from Wildfires Estimated from Operational Air Quality Forecasts

Rodrigo Munoz-Alpizar, Radenko Pavlovic, Michael D. Moran, Jack Chen, Sylvie Gravel, Sarah B. Henderson, Sylvain Menard, Jacinthe Racine, Annie Duhamel, Samuel Gilbert, Paul-Andre Beaulieu, Hugo Landry, Didier Davignon, and Sophie Cousineau

FireWork is an on-line meteorology-chemistry model with near-real-time wildfire emissions. It was developed by Environment and Climate Change Canada to deliver operational real-time forecasts of biomass-burning pollutants, in particular fine particulate matter (PM_2.5), over North America. Such forecasts provide guidance for early air quality alerts that could reduce air pollution exposure and protect human health. A multi-year (2013-2016) retrospective analysis has been conducted of FireWork forecasts during the North American "fire season", the 5-month period from May to September. An archive of FireWork forecasts was used to quantify wildfire contributions to hourly total PM_2.5 surface concentrations across North America. Different concentration thresholds, ranging from 0.2 to 28 g m^-3, were considered. It was found that on average over the fire season, 76% of Canadians and 69% of Americans were affected by seasonal wildfire-related PM_2.5 levels above 0.2 g m^-3. In much of the western U.S. and northwestern Canada, wildfire emissions contributed more than 1 g m^-3 of daily average PM_2.5 concentrations on 30% or more fire-season days. And in July and August, the peak months for wildfires, significant numbers of Canadians and Americans were also exposed to average monthly wildfire-related PM_2.5 concentrations above 28 g m^-3.

Assessing the Combustion Performance, Emissions, and Air Quality (AQ) Impacts of Renewable Fuel Blending in the Natural Gas System in California

Michael MacKinnon, Vince McDonell, Andres Colorado, Donald Dabdub, Jack Brouwer, G.S. Samuelsen

The injection of renewable fuels into the existing natural gas system, including renewably-sourced hydrogen and biomass derived fuels (i.e., biogas) represents a key pathway towards meeting California (CA) greenhouse gas (GHG) and renewable energy goals. However, the combustion performance, emissions, and air quality (AQ) implications of using such fuel blends in gas-consuming combustion devices (e.g., gas turbines, reciprocating engines, industrial boilers, household appliances) are not well understood. End-use devices are optimized for operation on unadulterated natural gas, and changing fuel composition may alter the performance of legacy combustion devices including residential appliances, commercial and industrial equipment, and electricity generators. Emission changes could impact regional AQ with considerable importance given the scale of deployment, i.e., system- and economy-wide. Current research has focused primarily on the feasibility and safety of injection and blending of fuels, and little is known regarding the potential emissions and AQ impacts.

Therefore, research is needed to support the development of holistic combustion device and burner deployment strategies targeting minimal emissions of criteria pollutants and maximum AQ and human health benefits. This study will address these needs by 1) leveraging existing tools for modeling combustion burner performance and emissions operating on pure natural gas, hydrogen/natural gas blend, and CO₂/natural gas blend simulating biogas injection, 2) identify and characterize emissions for a range of gas-consuming end-use devices to create economy-wide scenarios representative of renewable fuel blending, and 3) characterize pollutant emission and AQ impacts in California via simulations of atmospheric chemistry and transport - including consideration of those to under-served communities.

Regional Air Quality Modeling of Wildfires for Health Assessments over the Continental United States
Cesunica E. Ivey1, Cong Liu2, Yang Liu3, Howard Chang4, Matt Strickland5, and Heather A. Holmes1

1Atmsopheric Sciences Program, Department of Physics, University of Nevada Reno, Reno, NV; 2School of Energy and Environment, Southeast University, Nanjing, China; 3Department of Environmental Health, Emory University, Atlanta, GA; 4Department of Biostatistics and Bioinformatics, Emory University, Atlanta, GA; 5School of Community Health Sciences, University of Nevada Reno, Reno, NV;

Human exposure to wildfire smoke is difficult to estimate both at the individual and regional levels due to uncertainties in wildfire smoke modeling and the spatial heterogeneity of fuel composition leading to uncertainties in fire emissions classification. Hence, in this work we incorporated wildfire emissions from the Fire Inventory of NCAR (FINN) into the CMAQ modeling framework, with the goal of improving modeled estimates of the chemical composition of the wildfire emissions and the transport of the related primary and secondary pollutants. FINN emissions have a higher spatial and temporal correlation with observed fire activity, due to the use of MODIS active fire products in the inventory development. We present a multi-year evaluation of CMAQ simulations over the continental U.S., for which FINN emissions replace the wildfire emissions in the National Emissions Inventory. Simulated concentrations of bulk fine particulate matter (PM2.5), PM2.5 components (e.g., organic carbon, elemental carbon, and trace metals), and ozone are evaluated by comparing the estimates with observations from regulatory monitoring networks. Regional estimates of PM2.5, PM2.5 components, and ozone provide surrogate metrics for human exposure to wildfire smoke around and down-wind of active fires.

Estimates of the Marginal Cost of NOX and PM2.5 Emissions Reductions from U.S. Manufacturing from 1990 to 2014 with Projections to 2025

Alexander Macpherson, U.S. EPA

Robin Langdon, U.S. EPA

Jenny Thomas, U.S. EPA

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

Air pollution reductions in the United States result from a range of technology choices, such as production-level decisions, end-of-pipe emissions controls, energy efficiency measures, fuel switching, input or process changes, and other emission reduction strategies. Characterizing emission reductions and costs from many of these measures can be quite challenging. Aiken and Pasurka (2003) drew on an analytical approach from the productivity analysis literature to provide insights into the marginal cost (or shadow price) of reducing emissions of SO2 and PM10 from 1970 to 1996, for 19 manufacturing sectors in the United States. They applied a distance function-based productivity model that jointly models the conversion of inputs (e.g., capital, labor, energy, material, and services) into desirable outputs, such as market goods and services, and undesirable outputs, such as emissions that have detrimental effects on ambient air quality. The model has numerous advantages, including that the analyst does not need data on specific pollution control choices and costs to estimate the shadow price of reducing emissions. In the first stage of this study, we update the Aiken and Pasurka analysis to estimate the marginal cost of reducing NOX and PM2.5 emissions over the 1990 to 2014 timeframe for 17 three-digit NAICS manufacturing sectors in the United States. As a validation measure, these results are compared to other marginal cost estimates in the literature and in government reports. In the second stage of the analysis, we estimate the relationship between the marginal cost of reducing NOX and PM2.5 and emission intensities (in terms of tons of emissions per million dollars in output). Parameter estimates are then combined with projected 2017 and 2025 NOX and PM2.5 emission intensities to project the marginal cost of emissions reductions in these years.

Significantly reduced health burden from ambient air pollution in the United States under emission reductions from 1990 to 2010

Yuqiang Zhang¹, J. Jason West², Rohit Mathur¹, Jia Xing³, Christian Hogrefe¹, Shawn J. Roselle¹, Jesse O. Bash¹, Jonathan E. Pleim¹, Chuen-Meei Gan¹, David C. Wong¹

¹The US Environmental Protection Agency, Research Triangle Park, NC 27711, USA

²Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA

³State Key Joint Laboratory of Environmental Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China

The 2015 Global Burden of Disease (GBD) study has listed air pollution as the fourth-ranking global mortality risk factor. Many studies have estimated the global or national burden of disease attributed to air pollution. However, little effort has been focused on understanding how these burdens change through time, especially in the United States (US). Here we aim to estimate air pollution-related mortality in the continental US for each year from 1990 to 2010, to understand the trend over this time period. We also analyze the relative contributions of changes in air pollutant concentrations, population, and baseline mortality to the overall trend and to the interannual variability in mortality estimates. To achieve this goal, we use a 21-year model simulation of PM_2.5 and O₃ concentrations from 1990 to 2010, with grid resolution of 36km 36km. We also use annual county-level baseline mortality rates and population data archived by the US Centers for Disease Control (CDC).

We find that the PM_2.5-related mortality burden from ischemic heart disease, chronic obstructive pulmonary disease, lung cancer, and stroke, has steadily decreased, with a reduction of 54% from 1990 to 2010. The PM_2.5 -related mortality burden would have decreased only by 26% if the PM_2.5 concentrations had not decreased from 1990, due to decreases in baseline mortality rates for major diseases affected by PM_2.5. The mortality burden associated with O₃ for chronic respiratory disease has larger inter-annual variations than the PM_2.5-related burden, primarily due to inter-annual variations in O₃ mixing ratios. The O₃-related mortality burden has increased by 28% from 1990 to 2010, despite ozone decreases, mainly due to increases in the baseline mortality rates and population. The O₃-related mortality burden would have increased by 59% if the O₃ concentration was kept constant at the 1990 level. We concluded that the air quality changes have significantly accelerated (or decelerated) the decreasing (or increasing) trends for PM_2.5 -(or O₃-) related mortality burden change for the past 2 decades.

Projecting state-level air pollutant emissions using an integrated assessment model: GCAM-USA

Wenjing Shi^1,2, Yang Ou^1,2,3, Dan Loughlin², Chris Nolte², Steve Smith⁴, Catherine Ledna⁴

¹ Oak Ridge Institute for Science and Education

² Office of Research and Development, U.S. Environmental Protection Agency, RTP, NC

³ Environmental Sciences and Engineering, University of North Carolina at Chapel Hill

⁴ Joint Global Change Research Institute, Pacific Northwest National Laboratory, College Park, MD

The Global Change Assessment Model (GCAM) is an integrated assessment model that links representations of the economy, energy sector, land use, and climate within an integrated modeling environment. GCAM-USA, which is an extension of GCAM, provides U.S. state-level resolution within a regionalized global modeling framework. The overall goal of our research is to explore how GCAM-USA can be used to address air, climate, and energy system goals simultaneously and cost-effectively.

Here we incorporate characterizations of U.S. air pollutant emission factors and air pollutant controls as well as state-level air pollutant and energy policies into GCAM-USA. Specifically, base-year and future U.S. emission factors for NO_X, SO₂ and PM_2.5 now represent on-the-books regulations and are harmonized with emission factors from EPA regulatory modeling efforts; state-level electric-sector NO_X and SO₂ emission caps based upon national air pollutant regulations are added; characterizations of NO_X and SO₂ control technologies for coal-fired electric utilities are implemented; and light-duty vehicle fuel efficiency requirements are applied.

GCAM-USA air pollutant emission outputs are compared to a base-year inventory and future projections that were previously used in U.S. Environmental Protection Agency (EPA) regulatory activities. GCAM-USA projections are also compared with those of the base GCAM model in which the U.S. is represented as a single region. A Quality Metric (QM) is applied in this comparison. The QM also is used to quantify GCAM-USA performance for each pollutant at the sectoral and state levels. This information provides insights into the types of applications for which GCAM-USA is currently well suited and highlights where additional refinement may be warranted.

HiRes-X: Scientific and Geographic Extension of an Operational, High Resolution Air Quality Modeling System Providing Smoke Impact Forecasts for Health Protection, Ecosystem Management and Economic Development

Yongtao Hu¹, M. Talat Odman¹ and Armistead G. Russell¹, Ha Hang Ai², Ambarish Vaidyanathan³, Scott Goodrick⁴

¹School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA, 30332

²School of Computational Science and Engineering, Georgia Institute of Technology, Atlanta, GA, 30332

³Center for Disease Control and Prevention

⁴US Forest Service, Southern Research Station

In US, 2.3 million acres of forests are subject to prescribed fires, approximately 2/3 of which are in the Southeast, along with 3 million acres of agricultural burning. In the Southeast, prescribed burning is the largest source of PM2.5 emissions (20% or 210 Gg). The use of prescribed burning for forest and crop health has economical and ecosystem-related benefits but must be weighed and managed with respect to the potential health and welfare impacts, and concerns over those activities leading to air quality non-attainment. We have developed an advanced prescribed fire impact forecasting system -the HiRes II forecasting system, to provide daily forecasts of potential prescribed fire impacts based on meteorological and forest characteristics. The HiRes II forecasting products are currently used by the Georgia Department of Natural Resources (GaDNR) to provide official forecasts of air quality for public health protection, and are used by the Georgia Forest Commission (GaFC) in its 6 fire districts for prescribed burning management as well.

In this presentation, we will report our efforts to extend HiRes II to a newer version called HiRes-X (for extended and expanded). HiRes-X will have extended capabilities and expanded usage compared to HiRes II. The extension will aim to enhance the HiRes-X burn forecasting system using additional (and upcoming) Earth observations, as well as additional monitoring using inexpensive air quality sensors. The expansion will focus on expanding the burn forecasts to states outside of Georgia, particularly Florida, South Carolina, and Alabama. The expansion will also include providing forecasts of forest and crop prescribed burning-related air quality impacts to more local and state public health agencies in our forecast areas and to the CDC and its state partners on a real-time basis. We will also perform source apportionment analysis for CDC to conduct an epidemiologic assessment using retrospective data.

Forecasts will be provided publically via our website, and alerts of particularly high particulate matter or ozone levels will be sent out, specifically to health agencies. For this, we are redesigning our website to disseminate the HiRes-X forecasting products in an interactive way. The new website will use webgis technologies to display the forecasting products, real time and historical measurements, and earth observation datasets. It will overlap these datasets with geographical information, which can virtually show forecasted air pollution impacts on sensitive places. This will enhance the use of our products by fire and air quality managers in the Southeast.

Source-specific estimation of air quality co-benefits of CO2 reductions: a multi-country analysis

Reducing combustion-based CO2 emissions often entails significant ancillary benefits for public health by reducing emissions of criteria co-pollutants and precursors. In previous work, we used the adjoint of CMAQ to estimate location-specific co-benefits for mobile sectors in Canada and the United States. Location and sector dependencies, captured in adjoint sensitivity analysis, are important factors that can change the overall source and sector neutrality of co-benefits, and provide a measure of uncertainty/variability associated to their magnitudes. This study aims to present a multi-continental co-benefit assessment to evaluate how the co-benefits of reducing criteria co-pollutants vary spatially and by sector in different countries across the world.

The adjoint of USEPA'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 in on-road mobile, Electric Generation Units (EGUs), and residential combustion, a location-by-location basis across the US, Canada, Europe as well as parts of 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 2011 NEI at the county level, and point source data from EPA's Air Markets Program Data and National Pollutant Release Inventory (NPRI) for Canada and Emissions Database for Global Atmospheric Research (EDGAR) for Asia and Europe. The meteorology data and emissions are processed using WRF and SMOKE.v4.0 respectively. To quantify how biased a coarse resolution (36km x 36km) analysis is for calculating the health co-benefits, we further investigated a nested domain of (12km x 12km) over a major metropolitan area.

Our preliminary results 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 in Canada and $500/ton-CO2 in US for mobile on-road gasoline light duty (GLD) sector. For EGUs, co-benefits are estimated at up to $5000/ton-CO2 and $300/ton-CO2 for the US and Canada, respectively. Results also show that more refined domains (e.g.,12km x12km) produce higher benefits of up to 30% in monetized health co-benefits in Greater Toronto Area (GTA) for mobile GLD. Implications of co-benefit spatial variability in devising control policy options that effectively address both climate and air quality objectives will be discussed.

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

S. Marcela Lor a-Salazar

Cesunica E. Ivey

Heather A. Holmes

Howard H. Chang

Multiple statistical data fusion models have been used to estimate spatially resolved surface PM2.5 concentrations using PM2.5 observations and satellite retrievals of aerosol optical depth (AOD). Studies based on satellite-derived PM2.5 estimates show encouraging results worldwide. However, previous investigations in the western U.S. reported that atmospheric physics (e.g. planetary boundary layer, temperature inversions), Trans-Pacific aerosol pollution, low aerosol concentrations, complex terrain, high surface reflectance, and heterogeneous chemical composition diminish the ability to estimate near-surface PM2.5 concentrations using columnar AOD from satellite retrievals. This investigation aims to obtain daily, 12-km resolution surface PM2.5 concentrations for 2013 and 2014 in the semi-arid western U.S. using a data fusion model with AOD satellite retrievals from the Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) instruments. 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, and surface reflectance and fire radiative power from MODIS 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. In addition to physics parameters, land surface and air pollution emissions that impact the strength of surface PM2.5 concentrations were also implemented as covariates in the data fusion model. It is expected that by selecting high-resolution, temporally resolved physics and emissions variables as covariates it will improve the surface PM2.5 concentration estimates in the data fusion model for the western U.S.

Estimating Environmental Co-benefits of U.S. CO2 Reduction Pathways Using the GCAM-USA Integrated Assessment Model

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

Various technological pathways can lead to reduced CO₂ emissions. However, different pathways can have substantially different impacts on other environmental endpoints, such as air quality and energy-related water demand. The Global Change Assessment Model (GCAM) is a high resolution integrated assessment model developed to examine scenarios of the evolution of the U.S. and global economy, energy system, buildings, transportation, land use, and climate system. GCAM-USA is an extension of GCAM in which U.S. energy supply and demand markets are disaggregated to state-level resolution. In this study we use GCAM-USA to compare two stylized low-carbon pathways focused in the electric sector, one emphasizing nuclear energy and carbon capture and sequestration (CCS) for both coal and natural gas generation (NUC/CCS) and one emphasizing renewable energy (RE). These are compared with a baseline scenario (BASE) in which all technologies are available. Air pollutant emissions, mortality costs attributable to fine particulate matter (PM_2.5), and energy-related water demands are evaluated for 50% and 80% GHG mitigation targets in the U.S. in 2050, compared to a 2005 baseline. These two 2050 CO₂ targets are chosen to be consistent with the Energy Modeling Forum (EMF) 24 study, a multi-model comparison to assess alternative pathways for meeting climate targets. The EMF 24 studies focused on economic costs but did not explicitly evaluate other environmental implications, thus our study expands upon EMF 24 by exploring these tradeoffs, focusing on health impacts related to PM_2.5 exposure and energy-related water demand, both at the national and state levels.

Modeling results with GCAM-USA show that deployment of low-carbon power generation technologies and large-scale use of biofuels are responsible for the majority of CO₂ emissions mitigation, followed by electrification of the end-use sectors. The RE pathway requires lower water withdrawal and consumption than NUC/CCS due to the large cooling demands of nuclear power and CCS technologies. Both alternative pathways result in less PM health benefits under CO₂ reduction targets compared with BASE, as fewer low-carbon options are available in the electric sector. Environmental co-benefits differ among states as a result of their respective energy system composition, technology shares in end-use sectors, and resource availability.

Characterizing spatial variability of air pollution from vehicle traffic around the Houston Ship Channel area and potential associations with obesity in a Mexican-American Cohort

Xueying Zhang¹, Hua Zhao², Wong-ho Chow², Elena Craft³, Kai Zhang^1,4.

1. Department of Epidemiology, Human Genetics and Environmental Sciences, The University of Texas Health Science Center at Houston School of Public Health, Houston, TX 77030, USA. 2. Department of Epidemiology, The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA. 3. Environmental Defense Fund, 301 Congress Avenue, Austin, TX 78701, USA. 4. Southwest Center for Occupational and Environmental Health, The University of Texas Health Science Center at Houston School of Public Health, Houston, TX 77030, USA

The Houston Ship Channel area is the home of a large number of diesel powered vehicles emitting PM_2.5 and NO_x. However, the spatial variability of traffic-related air pollutants in that area has rarely been investigated. We calculated vehicle emission rates using the MOVES2014a based on the traffic data obtained from Texas Department of Transportation Road Inventory. The emission data and meteorological data were entered into Research LINE-source Dispersion Model (RLINE) to simulate the mass concentrations of traffic-related fine particulate matter (PM_2.5; 2.5 mm in aerodynamic diameter) and nitrogen oxides (NO_x) at 50 m resolution within 300 m to major roads and at 150 m resolution in the rest of the area. Our findings indicate that traffic-related PM_2.5 was mainly emitted from trucks, while traffic-related NO_x was emitted from both trucks and cars. The traffic contributed 0.90 mg/m³ PM_2.5 and 29.23 mg/m³ NO_x to the annual average mass concentrations of on-road air pollution. The concentrations of the two pollutants decreased by nearly 40% within 500 m distance to major roads. The seasonally varied near-road gradients illustrate that proximities to major roads are not an accurate surrogate of traffic-related air pollution. This analysis will be used as a basis for exploring the association between exposure to traffic-related air pollution and adult obesity in Mexican Americans. Study participants include the MD Anderson Cancer Center Mano a Mano Cohort, the largest on-going Mexican American cohort study in the United States, This part of presentation will consist of a brief introduction of the study cohort, background, exposure assessment, statistics methods, and preliminary results.

Understanding How Aircraft Emissions Lead to the Formation of Air Pollutants

Calvin Arter, Sarav Arunachalam

We present higher order sensitivity coefficient calculations for O3 and PM2.5 formation with respect to aircraft emissions in the continental United States, showing the importance of including second order sensitivity coefficients when utilizing sensitivity analyses methods for understanding the impacts of aviation emissions. We designated NOX and VOC emissions as sensitivity parameters for O3 formation; and NOX, SO2, VOC, EC, OC, and SO4 emissions as sensitivity parameters for PM2.5 formation. The Community Multiscale Air Quality Model (CMAQ) was used to estimate the concentrations and sensitivities of air pollutants in 36 36 km grid cells across the continental United States. We present an analysis utilizing these sensitivity coefficients to estimate the emission reduction needed in five airport grid cells to bring hypothetical regions in PM2.5 nonattainment into attainment.

Secondary Organic Aerosol Formation from Commercial Fuels and Single Component Aromatics: Comparison of Results from Chamber Experiments with Box Model Predictions under Varying Environmental Conditions

Jeff K. Bean, Terry L. Lathem, John Gingerich

Secondary organic aerosol (SOA) is an important contributor to ambient concentrations of particulate matter. SOA is formed from oxidation of anthropogenic and biogenic volatile organic compounds (VOCs) in the atmosphere. Chemical transport models (CTMs), such as the Community Multi-scale Air Quality Model (CMAQ) and the Comprehensive Air Quality Model with Extensions (CAMx), provide estimates of SOA formation based on data acquired from environmental chamber experiments and ambient measurements. The model parameters used to calculate SOA must be continuously evaluated and updated as research on SOA advances. It is important to evaluate model performance for multiple VOC precursors and across a variety of environmental conditions as this can highlight areas where further improvements can be made to increase the accuracy of these models.

The present study compares SOA formation in an environmental chamber with Statewide Air Pollution Research Center (SAPRC) box model predictions, both for evaporative emissions of commercial fuels and single component aromatics under a variety of environmental conditions. Experiments were conducted using UV lights to induce photo-oxidation of VOCs in a 7.5 m³ environmental chamber. Experiments were modeled using the SAPRC box model with the SAPRC07 chemical mechanism and the AERO6 SOA parameters, similar to the SAPRC07TC_AE6_AQ mechanism used by CMAQ version 5.1. Comparison of model and experimental results in the temperature range of 0 C to 40 C will be presented. The effects of NOx concentrations were probed by adjusting initial VOC:NOx ratios to 10:1, 1:1, and 1:10 and the results will be presented.

Contribution of Long-range Transport to Winter Haze in Northern China

Xinyi Dong^a, Joshua S. Fu^a*, Jiani Tan^a, Kan Huang^a

Winter and early spring haze has been severely affecting the densely populated areas in Northern China during recent years. While many of the pilot studies have been devoted to investigate the contributions from local anthropogenic emission, natural dust, and biomass burning, very few attentions have been paid to the impact of long-range transport. Europe and Russia have significant amount of anthropogenic emissions that may be carried downwind by the westerlies and deposited into China hence increase the background concentration of aerosols. In addition, due to the cooler climate conditions, aerosols precursors such as NOx, SO₂, and NH₃ have slower atmospheric chemical reaction rates while emitted from higher latitude in Europe and Russia. But these precursors may play an important role in generating secondary inorganic aerosols when they are carried over China. In this study, we used the ensemble modeling results from the Task Force on Hemispheric Transport of Air Pollution (TF HTAP) Phase II to investigate the long-range transport impact of Europe and Russia's anthropogenic emission on haze formation in Northern China. Simulations were conducted for January 2010. This study is the first investigation into the contribution of long-range transport to local haze in China with multiple datasets from different models.

Health benefits of decreases in PM2.5 and Ozone in the United States from 1998 to 2016

Omar Nawaz, Yuqiang Zhang, Daniel Q. Tong, Randall V. Martin, J. Jason West, Aaron van Donkelaar

Between 1990 and 2015 the US average concentration of PM2.5 decreased by 37% and ozone decreased by 22%, driven mainly by environmental regulations. These decreases in pollutants are expected to have brought substantial benefits for public health in the US. Here we assess the effects of this decrease by estimating the total burden of PM2.5 and ozone on premature mortality each year between 1998-2016. Two data sets of ambient concentration were implemented in the health impact analysis that included different subsets of the time period. First, we use data from the North American Chemical Reanalysis project, which uses OMI NO2 and MODIS AOD observations for data assimilation to constrain ozone and PM2.5 between 2007-2016. Second we use satellite-derived estimates of ground level PM2.5 using AOD retrievals from NASA satellites combined with the GEOS-Chem chemical transport model between 1998-2015; these estimates are calibrated to ground-based observations using a geographically weighted regression. Using yearly data on population and baseline mortality provided by the Centers for Disease Control (CDC), we assess how air pollution-related mortality has changed annually within the continental United States, and analyze trends to determine how effectively the health burden has been reduced by environmental regulations and other factors.

Modeling the Impact of Cookstove Use on Ambient Aerosol in Rural India

Brigitte Rooney, Kirk Smith, John Seinfeld, Ajay Pillarisetti, Rufus Edwards, Lauren Fleming, Sergey Nizkorodov, Tami Bond, Nicholas Lam, Sumit Sharma, Seema Kundu, Shaocai Yu, Pengfei Li, Kelvin Bates, Ran Zhao

In India, particulate matter (PM) pollution is a significant cause of many health problems and increased mortality. However, specific source contributions to PM have remained largely understudied, as work that has been done has mostly focused on indoor air quality. Within the residential sector, a majority of Indian households still rely on solid biofuels, mainly wood and dung, for cooking. Combustion of solid fuels is known to produce more aerosol than liquid fuels, and thus cookstove use has been identified as a potential significant source of PM pollution.

To better understand the production of ambient aerosol by the residential sector in rural India, we have undertaken a series of modeling experiments using the Community Multiscale Air Quality (CMAQ) atmospheric chemistry transport model. This work incorporates updated emissions inventories and time varying boundary conditions, and represents some of the highest resolution (1km and 4km grid cells) modeling of its kind. By varying such parameters as fuel type and emission factors, we are able to determine the fraction of PM attributable to cookstove use. Our expectation is that the results will further incentivize the transition from household use of solid fuels to more modern liquid fuels in order to reduce the health risks associated with aerosol pollution.

Evaluating Source Attributions using the CMAQ-DDM method for ozone during the T-COPS eastern Washington field campaign

Etesamifard, M., Vaughan, J., Jobson, T., Lamb, B.

Localized high ozone episodes affecting the Tri-Cities area in eastern Washington, initially detected via the AIRPACT5 air quality forecast system for the Pacific Northwest region by AIRPACT5, provided the motivation for the Tri-Cities Ozone Precursor Study (T-COPS) to investigate the temporal and spatial patterns of ozone precursors in the area.  In T-COPS, measurements were conducted during a three-week period (July 27 to August 18, 2016) when elevated ozone was expected in the Tri-Cities.  The observations were compared to AIRPACT5 outputs to evaluate the forecast system performance. The Community Multiscale Air Quality (CMAQ) Decoupled Direct Method in 3 Dimensions (DDM-3D) capability was applied to calculate ozone sensitivities with respect to emissions of ozone precursors and to initial and boundary concentrations.  Calculating the sensitivity of ozone to the emissions of its precursors determines the spatial variation of   NOx/VOC-sensitivity regimes to help guide potential control strategies. 

Synchronous evaluation of source apportionments for ambient concentration and deposition of sulfate in East Asia

Syuichi Itahashi

The source apportionments of ambient concentration, dry and wet deposition of sulfate aerosol (SO₄^2-) were synchronously evaluated in East Asia, where one of the main sources of anthropogenic sulfur dioxide (SO₂) emissions, based on the Comprehensive Air-quality Model with Extentions (CAMx). Owing to the difficulty of measuring deposition velocity directly, sensitivity simulations on the two different dry deposition schemes were conducted. In addition, sensitivity simulations on the different emission inventory, which is the largest uncertainty source for the air quality model, were also conducted. The model reproducibility of ambient concentration and wet deposition were verified with the ground-based observation network over China, Korea, and Japan. The modeling system configured by four experiments captured the observations features. The synchronous evaluation of source apportionments for ambient concentration, dry and wet deposition showed the dominated contribution from anthropogenic emissions of China to downwind regions of Korea and Japan. The differences in the dry deposition scheme and emission inventory, and the comparison to other studies were discussed in the presentation.

Variation and Sensitivity of Regional Emissions Contributions to Particulate Matter Concentration in the Seoul Metropolitan Area, Korea
Eunhye Kim, Hyun Cheol Kim, Byeong-Uk Kim, Jeonghoon Cho and Soontae Kim

Located on the downwind side of strong sources of anthropogenic emissions, South Korean air quality has been affected by the regional transport of pollutants and their precursors. The impact of regional emissions (e.g., domestic and international) on surface particulate matter (PM) concentrations in the Seoul Metropolitan Area (SMA), South Korea and its sensitivities to meteorology and emissions inventories are quantitatively estimated for 2014 using regional air quality modeling systems. Several sensitivity scenarios were established for meteorology (2 cases), emissions inventory combinations (4 cases) and contribution estimation methods (3 cases). We utilized two regional air quality simulation systems: (1) a Weather Research and Forecasting and Community Multi-Scale Air Quality (CMAQ) system; and (2) a United Kingdom Met Office Unified Model and CMAQ system. Four combinations of emission inventories are used using the Intercontinental Chemical Transport Experiment-Phase B, Inter-comparison Study for Asia 2010, and the National Institute of Environment Research Clean Air Policy Support System. Partial contributions of domestic and foreign emissions are estimated using a brute force approach, adjusting South Korean emissions to 50% or 100%, or adjusting foreign emission to 50%. Impacts from each source categories (industrial, residential, power plants and transportation sectors in China) are also estimated. Results show that foreign emissions contributed ~60 % of SMA surface PM concentration in 2014. Estimated contributions display clear seasonal variation, with foreign emissions having a higher impact during the cold season (Fall to Spring), reaching ~70 % in March, and making lower contributions in the summer, ~45 % in September. We also found that simulated surface PM concentration is sensitive to meteorology, but estimated contributions are mostly consistent. Regional contributions are also found to be sensitive to the choice of emissions inventories.

Dynamical downscaling of meteorology from a global model by WRF towards resolving US PM2.5 distributions for the mid 21st century
Surendra B. Kunwar, Jared H. Bowden, George Milly, Michael Previdi, Arlene M. Fiore, J. Jason West

In the coming decades, anthropogenically induced climate change will likely impact PM_2.5 through both changing meteorology and feedback in natural emissions. A major goal of our project is to assess changes in PM_2.5 levels over the continental US due to climate variability and change for the period 2005-2065. We will achieve this by using regional models to dynamically downscale coarse resolution (2^o X 2^o) meteorology and air chemistry from a global model to finer spatial resolution (12 km), improving air quality projections for regions and subregions of the US (NE, SE, SW, NW, Midwest, Intermountain West). We downscale from GFDL CM3 simulations of the RCP8.5 scenario for the years 2006-2100 with aerosol and ozone precursors fixed at 2005 levels. We will carefully select model years from the global simulations that sample the range of PM_2.5 distributions for different US regions at mid 21^st century (2050-2065). Here we will show results for the meteorological downscaling (using WRF version 3.8.1) for this project, including a performance evaluation for meteorological variables with respect to the global model. In the future, the downscaled meteorology presented here will be used to drive air quality downscaling in CMAQ (version 5.2). Analysis of the resulting PM_2.5 statistics for US regions, as well as the drivers for PM_2.5 changes, will be important in supporting informed policies for air quality (also health and visibility) planning for different US regions for the next five decades.

Modeling of the Foehn effect in the north coast off Colombia.South America.

Jose Luis Rodriguez Castilla ¹, Luis Carlos Angulo Argote ²

Gloria Maria Restrepo Vasquez 3,

Universidad Popular del Cesar - UPC, Grupo GEAB-CIDTEC ^{1, 2}

Universidad de Antioquia 3 - Medellin, Colombia.

joselrodriguez@unicesar.edu.co ¹ lcangulo@unicesar.edu.co 2

Valledupar, Cesar, Colombia.

The presence of variations in the topography affetc the near-surface wind components; in some cases the wind is forced to ascend over the topographic and surpass it causing what is known as Foehn Effect. The WRF model calculates the winds components that are conditioned by the height of the terrain and the atmospheric stability. In the north off the Colombia country are located a branch of the Andes Mountains (Perija mountain chain) with a maximum height of 2400 meters together with the Mountain formation of Santa Marta that has a maximum height of 4850 meters.

The orography is a factor that influences considerably the speeds that the air takes in its displacement. Mountain chains, such as the Andes mountains, in this case the Perija mountain chain, which opposes the flow of the Northeast winds, constitute a natural physical barrier that alters the flow of air currents across the mountain ranges and according to their orientation or physiographic accidents, may lead to the strengthening or weakening of the winds. For the modeling of the meteorology in the Cesar River Valley, which is the principal affected and where is located one of the principal opencast mining complex of the Colombia country was necessary knowing how the Foehn effect influence in the behavior of the winds and in the dispersion of particulate matter produced by opencast mining activities.

In this study was modeling with WRF in the year of 2014 for the resolutions de 3 y 9 km. The Foehn effect was most intense in the months of January, February and March, when the Alisios winds in the northern hemisphere blow with greater intensity.

Ozone results of the Model Intercomparison Study in the Asia (MICS-Asia) Phase III
Jie Li1, Zifa Wang1, Joshua S. Fu2, Gregory R. Carmichael3, Tatsuya Nagashima4, Baozhu Ge1, Jun-ichi Kurokawa5

1 Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
2 Department of Civil and Environmental Engineering, University of Tennessee, Knoxville, TN, USA

3 Center for Global and Regional Environmental Research, University of Iowa, Iowa City, IA, USA
4 Center for Regional Environmental Research, National Institute for Environmental Studies, Tsukuba, Japan
5 Asian Center for Air Pollution Research, Niigata, Japan

The MICS-Asia Phase III project is a continuing activity after the Phase I and Phase II. This is a cooperative multi-model effort by brining together modelers from China, Japan, Korean, Southeast Asia and the United States. The MICS-Asia III is an independent project supported by Institute of Atmospheric Physics (IAP) and Asian Center for Air Pollution (ACAP) through Joint International Center on Air Quality Modeling Studies (JICAM) between IAP and ACAP. The model inter-comparison activities in Phase III have been conducted to be carried out using common meteorological fields, emission data, boundary conditions, etc. in order to allow the discussion of the cause of disagreement among participating models rather than just showing the variability of model output and uncertainties. The participating models including, CMAQ (different versions), WRF-Chem, NAQPMS, NHM-Chem, GEOS-Chen, etc. It is aimed to evaluate strengths and weaknesses of current air quality models for air quality prediction, and provide techniques to reduce uncertainty and improve performance in Asia. Models captured the general seasonal cycle of surface ozone in China, and low latitudes over eastern Asia, but failed in Japan. Major discrepancies between model results and observations over East Asia are the significant overestimation of ozone levels over polluted regions in summer. Comparison with observations shows that the improvement on NO2 is very important to O3 simulation.

This activity not only provides scientific assessment of current status of air quality and regional climate in East Asia but also can provide sound recommendation for the efficient abatement of air pollutant emissions.

Modeling the impacts of green infrastructure land use changes on air quality and meteorologycase study and sensitivity analysis in Kansas City

Yuqiang Zhang¹, Jesse Bash¹, Shawn Roselle¹, Angie Shatas², Christian Hogrefe¹, Rohit Mathur¹, Andrea Repinsky³, Tom Jacobs³

¹Office of Research and Development, US Environmental Protection Agency, NC-27719

²Office of Air and Radiation, US Environmental Protection Agency, NC-27719

³Mid-America Regional Council, MO-64105

Changes in vegetation cover associated with urban planning efforts may affect regional meteorology and air quality. Here we use a comprehensive coupled meteorology-air quality model (WRF-CMAQ) to simulate the influence of planned land use changes from green infrastructure implementation in Kansas City (KC) on regional meteorology and air quality. Current and a plausible green infrastructure land use scenarios were provided by the Mid-America Regional Council for 2012 and a scenario with land use changes due to green infrastructure implementation only. These land use datasets were incorporated into the WRF-CMAQ modeling system allowing the modeling system to propagate the changes in vegetation and impervious surface coverage on meteorology and air quality. The WRF-CMAQ model was run for the continental US using a 12km by 12km horizontal grid spacing and nested simulations for the greater Kansas City area were performed with a finer grid spacing of 4km by 4km. The simulations were performed for one year to study seasonal variations of the air quality changes stemming from the land use change.

Preliminary model findings indicate that the adoption of green infrastructure reduced urban temperatures in the day and night. The change in pollutant concentrations is a balance between increased deposition to the increase in vegetation and the concentration of pollutants in a lower planetary boundary layer due to the cooling of urban areas. Primary emitted pollutants with a relatively slow deposition velocity, e.g. PM_2.5, increased in these initial simulations. The initial simulations appear to be sensitive to the parameterization of the meteorological models land surface model. To further investigate this, several sensitivity simulations will be presented using different WRF configurations to further investigate the response of both meteorology and air quality to the planned land use changes.

Drought Impacts on Secondary Organic Aerosol, a Case Study in the Southeast United States

Zijian Zhao1, Yuxuan Wang1,2, Momei Qin3, Yongtao Hu3, Armistead G. Russell3

1Department of Earth System Sciences, Tsinghua University, Beijing, China 2Department of Earth and Atmospheric Sciences, University of Houston, Houston, Texas, USA 3School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA

Drought has a widespread influence in the US and has been shown to be associated with higher surface pollution levels. A previous observational and modeling analysis indicated a large enhancement of organic aerosol in the southeastern US during the 2011 severe summer drought, but the causes for this enhancement are not well understood. This study uses the Community Multiscale Air Quality (CMAQ) modeling system to understand the complex effects of drought on emissions, formation, and deposition of air pollutants in the southeast US, with a focus on secondary organic aerosol (SOA). Preliminary analysis suggests that 78% of the change in total PM2.5 concentrations at the surface between June 2011 (drought year) and June 2013 (wet year) come from organic aerosols (OA), and 94% of this OA enhancement is caused by biogenic SOA formation. Part of the biogenic SOA increase is due to higher emissions of biogenic volatile organic compounds (BVOCs), which show an increase of 19%-31% during the drought month through the Biogenic Emission Inventory System (BEIS3) in-line calculation in the model. Differences in wet deposition fluxes also impact the resulting SOA levels. Different model sensitivity simulations are carried out to calculate the SOA budget and quantify the perturbation of drought on deposition, emissions, and chemistry of SOA.

Overview of the Model and ObservatioN Evaluation Tool (MONET) version 1.0 for evaluating chemical transport models

Barry Baker^1,2, Li Pan^1,2, Pius Lee¹, Youhua Tang^1,2, Daniel Tong^1,2,3

1 NOAA Air Resources Laboratory, College Park, MD

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

3. Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA 22030

Development and initial applications of the Model and ObservatioN Evaluation Tool (MONET) v1.0. MONET was developed to evaluate the Community Multiscale Air Quality Model (CMAQ) for the NOAA National Air Quality Forecast Capability (NAQFC) modeling system. MONET is designed to be a modularized Python package for 1) pairing model output to observational data in space and time, 2) leveraging the pandas Python package for easy searching and grouping, and 3) analyzing and visualizing data. A convenient method for evaluating model output is introduced through this process. Data processed by MONET is easily searchable and can be grouped using meta-data found within the observational datasets. Included in the package are common statistical metrics (e.g. bias, correlation, and skill scores), plotting routines such as scatter plots, timeseries, spatial plots, and more. MONET is well modularized and effortlessly able to add further observational datasets and different models.

Using California Field Studies as a Platform for Model Evaluation

Kirk R. Baker, United States Environmental Protection Agency, Research Triangle Park, NC.
James T. Kelly, United States Environmental Protection Agency, Research Triangle Park, NC.
Matt Woody, United States Environmental Protection Agency, Research Triangle Park, NC.

Ozone and PM2.5 formation in the atmosphere is the result of precursor emissions and conducive meteorological conditions. California is particularly challenging to represent as elevated air quality episodes are typically associated with a variety of physical processes including terrain blocking of air masses, complex ocean-land circulation, and a broad range of stationary and mobile emissions sources. Ground and aircraft measurements made during the California Research at the Nexus of Air Quality and Climate Change (CalNex) 2010, Carbonaceous Aerosols and Radiative Effects Study (CARES) 2010, and Biosphere Effects on Aerosols and Photochemistry Experiment (BEARPEX) 2009 field studies provide a unique opportunity to evaluate meteorology, chemistry, and emissions in northern, central, and southern California. The Community Multiscale Air Quality (CMAQ) model was applied at 4 km coincident with these field studies. This poster is intended to provide highlights from recent work focused on understanding meteorology, inorganic chemistry, organic chemistry, and biogenic emissions in California (Baker et al., 2015; Baker et al., 2013; Bash et al., 2016; Jathar et al., 2017; Kelly et al., 2014; Woody et al., 2016).

Baker, K., Carlton, A., Kleindienst, T., Offenberg, J., Beaver, M., Gentner, D., Goldstein, A., Hayes, P., Jimenez, J., Gilman, J., 2015. Gas and aerosol carbon in California: comparison of measurements and model predictions in Pasadena and Bakersfield. Atmospheric Chemistry and Physics 15, 5243-5258.

Baker, K.R., Misenis, C., Obland, M.D., Ferrare, R.A., Scarino, A.J., Kelly, J.T., 2013. Evaluation of surface and upper air fine scale WRF meteorological modeling of the May and June 2010 CalNex period in California. Atmospheric Environment 80, 299-309.

Bash, J.O., Baker, K.R., Beaver, M.R., 2016. Evaluation of improved land use and canopy representation in BEIS v3. 61 with biogenic VOC measurements in California. Geoscientific Model Development 9, 2191.

Jathar, S.H., Woody, M., Pye, H.O., Baker, K.R., Robinson, A.L., 2017. Chemical transport model simulations of organic aerosol in southern California: model evaluation and gasoline and diesel source contributions. Atmospheric Chemistry and Physics 17, 4305-4318.

Kelly, J.T., Baker, K.R., Nowak, J.B., Murphy, J.G., Markovic, M.Z., VandenBoer, T.C., Ellis, R.A., Neuman, J.A., Weber, R.J., Roberts, J.M., 2014. Finescale simulation of ammonium and nitrate over the South Coast Air Basin and San Joaquin Valley of California during CalNex2010. Journal of Geophysical Research: Atmospheres 119, 3600-3614.

Woody, M.C., Baker, K.R., Hayes, P.L., Jimenez, J.L., Koo, B., Pye, H.O., 2016. Understanding sources of organic aerosol during CalNex-2010 using the CMAQ-VBS. Atmospheric Chemistry and Physics 16, 4081-4100.

CMAQ PERFORMANCE WITH PORTLAND AND INTEL COMPILERS

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

This presentation reports on performance of CMAQ 5.1 with Portland and Intel FORTRAN compilers. Both the new Chemistry Transport Model (CTM) sparse solver, FSparse [1], and the legacy, JSparse [2], algorithms are studied in this comparison. The focus is on the Rosenbrock and SMV Gear algorithms in the CTM because these are well suited for an OpenMP thread-parallel implementation. It is noteworthy that both compilers offer extensions to FORTRAN for off-load of parallel code segments to PCI attached accelerators: the Intel PHI co-processor (Intel compiler) and AMD Firepro (Portland compiler). Results for the former were previously reported [1], and future work will investigate the feasibility of the latter. The FSparse algorithm has been completely revised to present a more modular code to the compiler. This has required a laborious debugging effort, but seems to avoid some problems that were previously encountered with compilers. Specific compiler problems have been reported to the vendors but others remain unresolved and work-arounds have been used. It is anticipated that results of thread scaling will be presented for both host CPU and attached accelerators.
.
[1] G. Delic, Annual CMAS conference, 2012, 2013, 2016.
[2] M. Jacobson and R.P. Turco (1994), Atmos. Environ. 28, 273-284

Meteorological factor analyses of NOAA NAQFC surface ozone predictions' biases and improvement with a bias correction approach
Jianping, Huang1,2*, Jeff McQueen2, James Wilczak3, Irina V. Djalalova4, Dave Allured4, Ho-Chu, Huang1,2, Pius Lee5, Li Pan5,6, Daniel Tong5,6, Youhua Tang5,6, Sikchya Upadhayay7,8, and Ivanka Stajner8

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 Cooperative Institute for Research in the Environmental Sciences (CIRES), University of Colorado, Boulder, CO
4 NOAA, Earth Systems Research Laboratory (ESRL), Boulder, CO
5 University of Maryland, College Park, MD
6 NOAA Air Resources Laboratory, Silver Spring, MD
7 Syneren Technologies Corporation, Arlington, VA 22201
8 NOAA, Office of Science and Technology Integration, Silver Spring, MD

The NOAA National Air Quality Forecasting Capability (NAQFC) has been producing operational predictions of surface ozone (O3) for the Continental United States (CONUS) since 2007. Recently, the NAQFC was upgraded with the Non-hydrostatic Multi-scale Model on the Arakawa staggered B-grid (NMMB) v4.0 and the EPA Community Multiscale Air Quality (CMAQ) model v5.0.2. However, the NAQFC still exhibits over-predictions of surface O3 during daytime in summer as well as nighttime for all seasons, especially over the Eastern U.S. In this study, a comprehensive analysis is performed first to understand the meteorological factors that cause the surface O3 prediction biases. A bias correction method, the Kalman-Filtering combined with historical forecast ANalogs (KFAN), then, is incorporated into the NAQFC to reduce surface O3 prediction biases. The performance of bias-corrected predictions is examined with varying training periods and number of predictors that are used to determine O3 prediction analogs. Both raw and bias-corrected predictions are evaluated with AIRNow measurement by calculating a series of statistical parameters (e.g. mean bias, root mean squared errors) and forecast skill scores (e.g., critical successful index, fraction correct) of the threshold of 70.0 ppbv for daily 8-hr average maximum O3 for the CONUS domain and different sub-regions. Finally, a discussion is presented to illustrate the limitation of the KFAN bias correction method on improving surface O3 predictions for several extreme O3 exceedance cases.

Estimation of primary and secondary PM2.5 in CONUS using standard CMAQ with post-processing tools
Jiaoyan Huang1 and Sarav Arunachalam1

1Institute for the Environment, The University of North Carolina at Chapel Hill
100 Europa Drive, Suite 490, Chapel Hill, NC 27517

Currently, ambient fine particulate matter (PM2.5) causes 4.2M premature deaths globally, and there are 92 % of people living in the areas where air quality is worse than the WHO health-based standard. Understanding of temporal and spatial variations of primary versus secondary PM2.5 ratio helps estimating ambient PM2.5 contribution from individual source sectors. The fraction of primary and secondary PM2.5 ratios to the total PM2.5 varies depending on locations and seasons. In the current standard CMAQ configuration, primary and secondary organic PM2.5 are estimated. However, primary and secondary inorganic species, such as SO4, NO3, and NH4, are lumped together. Although additional CMAQ tools such as the Direct Decoupled Method (DDM), Integrated Source Apportionment Method (ISAM), and Process Analysis (PA), can help separate primary and secondary inorganic PM2.5, these advanced CMAQ simulations are computationally expensive and require additional efforts. We have proposed an approach to create new post-processing tools to quantify total primary and secondary PM2.5 from standard CMAQ simulations. As part of this approach, we offer three different methods: a) based on previous field source apportionment model results, b) CMAQ PA analysis, and c) Chemical species Mass Balance approach that use existing CMAQ output from a standard CMAQv5.0.2 simulation for the continental U.S. After estimating primary/secondary PM2.5 in different regions, the relative importance of primary anthropogenic emissions and atmospheric chemistry can be estimated. This information benefits air pollution scientists and policy-makers with critical information for developing mitigation options for anthropogenic emissions and for reducing overall public health risk due to ambient PM2.5.

Evaluation of differences in ozone concentration between chemical mechanisms of CMAQ in Japan

Kyo Kitayama¹, Yu Morino¹, Kazuyo Yamaji², Satoru Chatani
¹National Institute for Environmental Studies, Tsukuba, Japan
²Kobe University, Kobe, Japan

Differences in ozone concentration between chemical mechanisms were evaluated for understanding errors caused by the selection of chemical mechanisms in the CMAQ model. Compared chemical mechanisms were SAPRC07tc, CB05tucl, RACM2 and SAPRC99 in CMAQv5.0.2. Ozone concentration was calculated in 15 km grid domain covering all over Japan and 5 km grid domain covering the Kanto urban area for July 22 to August 10, 2013 by using each mechanism. In 15 km grid domain, average ozone concentration of CB05tucl was lower than the concentration of SAPRC07tc in the all target region. For RACM2, average ozone concentration was lower than the concentration of SAPRC07tc on the sea and in the comparable level in land. SAPRC99 shows higher concentration in the urban area and lower concentration in the other area compared to SAPRC07tc. The differences of theses concentrations ranged about 10 ppb in the all target area and about 5 ppb in land. In 5 km grid domain, the model concentrations were compared with observations at the sites in the Kanto area. The average concentration of SAPRC07tc at the observational points was about 20 ppb higher than the observed average. For the other chemical mechanisms, the average concentrations at the points were lower than the average of SAPRC07tc. The differences were about 4 ppb for CB05tucl, about 2 ppb for RACM2 and under 1 ppb for SAPRC99. These differences suggested that the selection of chemical mechanisms influenced resulted ozone concentrations in some extent. The causes of these differences will be investigated in relation to chemical equations in the chemical mechanisms.

Dynamic evaluation of two decades of WRF-CMAQ ozone simulations over the contiguous United States

H. Luo¹, M. Astitha¹, S.T. Rao^1,2, C. Hogrefe³, R. Mathur³, and N. Kumar^4
1 University of Connecticut, ² North Carolina State University, ³ US Environmental Protection Agency, ⁴ Electric Power Research Institute

Dynamic evaluation of the fully coupled Weather Research and Forecasting (WRF) - Community Multiscale Air Quality (CMAQ) model ozone simulations over the contiguous United States (CONUS) using two decades of simulations covering the period from 1990 to 2010 is conducted to assess how well the changes in observed ozone air quality are simulated by the model. The changes induced by variations in meteorology and/or emissions are also evaluated during the same timeframe using spectral decomposition of observed and modeled ozone time series with the aim of identifying the underlying forcing mechanisms that control ozone exceedances and making informed recommendations for the optimal use of regional-scale air quality models. The evaluation is focused on the warm season's (i.e., May - September) daily maximum 8-hr (DM8HR) ozone concentrations, the 4th highest (4th) and average of top 10 DM8HR ozone values (top10), as well as the spectrally-decomposed components of the DM8HR ozone time series using the Kolmogorov-Zurbenko (KZ) filter. Results of the dynamic evaluation are presented for six regions in the U.S., consistent with the National Oceanic and Atmospheric Administration (NOAA) climatic regions. During the earlier 11-yr period (1990-2000), the simulated and observed regional average trends are not statistically significant. During the more recent 2000-2010 period, all observed trends are statistically significant and WRF-CMAQ captures the observed downward trend in the Southwest and Midwest but under-predicts the downward trends in observations for the other regions. Observational analysis reveals that it is the magnitude of the long-term forcing that dictates the maximum ozone exceedance potential; there is a strong linear relationship between the long-term forcing and the 4th highest or the average of the top10 ozone concentrations in both observations and model output. This finding indicates that improving the model's ability to reproduce the long-term component will also enable better simulation of ozone extreme values that are of interest to regulatory agencies.

Evaluating Isoprene Oxidation Mechanisms in Regional Models Using In Situ Observations of Formaldehyde

Margaret R. Marvin, Glenn M. Wolfe, Ross J. Salawitch, Timothy P. Canty, Allison Ring, Daniel L. Goldberg, Alan Fried, Thomas F. Hanisco, Jennifer Kaiser, Frank N. Keutsch

The OH oxidation of isoprene is a complex chemical process that can significantly influence atmospheric composition in isoprene-rich regions, such as the summertime Eastern US. Formaldehyde (HCHO), a high-yield product of isoprene oxidation, is a precursor to ozone and provides insight into consequences for air quality. Most regional models employ chemical mechanisms that account for isoprene oxidation; however, the representation of isoprene chemistry and associated production of HCHO varies widely between mechanisms. We use the CAMx v6.40 regional model, driven by meteorology and emissions for 2011, to simulate atmospheric composition in the summertime Eastern US. Simulations are performed using Carbon Bond mechanisms CB05 and CB6r2, which are currently available for implementation in CAMx. We also test the recently proposed mechanism CB6r2-UMD, which has been shown to improve HCHO production from isoprene oxidation relative to CB6r2 (Marvin et al., 2017). The isoprene schemes of all considered mechanisms are evaluated with respect to in situ observations of HCHO from the DISCOVER-AQ and SENEX aircraft campaigns, which took place in the Eastern US during the summers of 2011 and 2013, respectively. Accurate simulation of ozone precursors such as HCHO in regional air quality models is critical to the development of effective ozone management strategies.

Contribution of on-road mobile sources to SOA formation in Bogota: A sensitivity analysis coupling WRF-Chem and the traffic model VISUM

Perez-Pena, Maria Paula¹mp.perez@uniandes.edu.co

Morales B., Ricardo¹r.moralesb@uniandes.edu.co

Contreras B., Yadert¹ yd.contreras@uniandes.edu.co

Affiliations:

¹Universidad de los Andes. Department of Civil and Environmental Engineering. Research group CIIA

Secondary organic aerosols (SOA) have been found to contribute a significant fraction of submicrometer particle mass in many different environments, including remote and urban atmospheres. In the city of Bogota, few studies have investigated the contribution of organics to PM2.5. In this project we performed a series of sensitivity analysis to determine the potential contribution of SOA from mobile sources, to the total submicrometer particle mass in the city. For this, the Weather Research and Forecasting coupled with Chemistry model (WRF-Chem) is used. WRF-Chem has been implemented in a domain of 100 x 100 km, with a resolution of 27km, centered in Bogota, Colombia. In order to assess the contribution of mobile sources to SOA, emissions were estimated by means of the traffic analysis software VISUM. This traffic software simulates all roads and their interactions (PTV, 2017). VISUM is fed with road characteristics and estimation of origin-destination matrices, and outputs vehicular counts disaggregated by type of vehicle. In this way, the spatially disaggregated activity factors by each fleet are mapped into the WRF-Chem grid. The temporal disaggregation of emissions is based on scaling according to available vehicle counts in many parts of the city. Activity patterns are then used to compute emission rates. Three different approaches are used in the study, ranging from using constant emission factors from literature, to utilizing emissions models such as IVE and MOVES. Both emission models have previously been used in the city in different studies (Carmona et al., 2015; Giraldo & Behrentz, 2005; Zarate, Belalcazar, Clappier, Manzi, & Van den Bergh, 2007) . Point sources emissions inventories constructed by local environmental agencies are used. Both inventories were speciated according to the requirements of the selected gas-phase chemical mechanism. The emissions generated through the output of the traffic model VISUM are in good agreement with previous emission inventories performed in the city. WRF-Chem was run for a period of one week using the global database EDGAR-HTAP 2010 for anthropogenic emissions, on-line biogenic emissions calculation from land use, initial meteorological conditions from FNL, chemical initial and boundary conditions from MOZART-GEOS5. Selected physical parameterizations for meteorology were chosen based on previous work developed in Bogota (Arango & Ruiz, 2011). RADM2 and RACM gas-phase chemical mechanisms have been tested, both coupled with MADE-SORGAM aerosol scheme. The preliminary results indicate that regional contribution from biogenic sources is only minor compared to the local emissions generated by mobile sources in the city. RADM2 was found to overestimate PM2.5 concentrations compared to RACM.

Regional Differences in 2012 CONUS Model Performance: An AMET Case Study

Thuy Phi and TCEQ Photocheminal Modeling Team Staff

The Texas Commission on Environmental Quality (TCEQ) reconfigured its 2012 modeling platform to include a 12-km continental US domain.  The Atmospheric Model Evaluation Tool (AMET) has played an important role in characterizing model performance over the large domain.  Specifically, several meteorological model configurations were evaluated, including different land surface models and the use of observational nudging.  This poster will showcase the regional differences in CAMx model performance in 2012 using AMET v1.2.

PM2.5 concentrations observed and modeled for the 2016 southern Appalachian wildfire event
Ian McDowell1,2, Thomas Pierce2*, Brian Eder2, Kristen Foley2, Robert Gilliam2, George Pouliot2, Joseph Wilkins2

*Corresponding author: pierce.tom@epa.gov

1 Oak Ridge Association of Universities (ORAU), Research Triangle Park, NC
2 Computational Exposure Division, National Exposure Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC

During November 2016, wildfires in the southern Appalachian region of the United States burned over 50,000 ha leading to a widespread outbreak of elevated levels of fine particulate matter (PM2.5). Daily average PM2.5 concentrations above the current National Ambient Air Quality Standard (NAAQS) of 35 g m-3 were measured across much of eastern Tennesse, northern Georgia, and western North Carolina during a two-week period, with numerous locations reporting greater than 100 g m-3. As the prevalence of wildfires is projected to increase in future years -- due to changes in climate and the growing abundance of fire-prone forest biomass, models like the Community Multiscale Air Quality (CMAQ) modeling system may be a useful tool for identifying harmful levels of PM2.5. A near-real-time version of CMAQ has been used to simulate the impact of the wildfire event during November 12-30, 2016. Simulations with and without wildland fire emission estimates have been compared with monitoring data, and the CMAQ modeling system demonstrates some utility at estimating the impact of smoke emissions during the 2016 southern Appalachian wildfire event.

Multiple-Year Application and Evaluation of Two-Way Coupled WRF-CMAQ with Aerosol Direct and Indirect Effects over the Continental U.S.

Kai Wang¹, Yang Zhang¹, Shaocai Yu², David Wong³, Jonathan Pleim³, and Rohit Mathur^3
1Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC, 27695
²College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, Zhejiang, P.R. China, 310058
³Computational Exposure Division, National Exposure Research Laboratory, U.S. EPA, Research Triangle Park, NC, 27711

The Community Multiscale Air Quality (CMAQ) modeling system developed by the U.S. Environmental Protection Agency (EPA) has been widely used by the research community to study various air quality issues such as tropospheric ozone, regional haze, and acid deposition and to provide scientific guidance to policymakers for air quality regulations. Traditionally the CMAQ model has been driven by 3-D meteorological fields generated offline by other meteorological models such as the Weather Research Forecasting (WRF) model. With the rapid development of computational resources and better understanding of meteorology-chemistry interactions, a two-way coupled WRF-CMAQ has been developed to simulate the complex chemistry-aerosol-cloud-radiation feedbacks. Compared with the offline-coupled CMAQ, the two-way coupled WRF-CMAQ provides greater fidelity in coupling atmospheric dynamics and chemistry calculations and consistency in prediction of interactions between radiation, cloud, precipitation, and air quality.

In this study, we present a multiple-year (2008-2012) application and evaluation of the two-way coupled WRF-CMAQ with both aerosol direct and indirect effects over the continental U.S. (CONUS) using emissions based on the U.S. National Emission Inventories (NEI) and chemical initial and boundary conditions derived from an advanced Earth system model (i.e., the Community Earth System Model). Our objectives are to 1) perform comprehensive model evaluation for key surface meteorological, radiation, cloud, and chemical variables; 2) improve the model's capability in reproducing spatio-temporal variations of air pollutants for accurate health effect assessments. The preliminary results show that WRF-CMAQ performs well for meteorological variables such as temperature and relative humidity at 2-m, wind speed at 10-m, and precipitation and for radiation variables such as short- and long-wave radiation. The model also moderately underpredicts cloud radiation forcing which is estimated by difference of shortwave (or longwave) at the top of atmosphere between the clear-sky and all-sky conditions. WRF-CMAQ also predicts ozone and fine particulate matter (PM2.5) very well in terms of both temporal and spatial variations despite some relatively high biases for PM2.5 components. The overall model performance is quite promising when compared with other online-coupled models and also helps identify some issues such as moderate underpredictions for sulfate for further investigation. Overall, this work will provide valuable information on the future model further development and improvement of two-way coupled WRF-CMAQ model.

Estimating Oxidative Potential of PM2.5 across the Eastern United States Using an Advanced CMAQ-DDM/CMB Hybrid Method

Josephine T. Bates, Rodney J. Weber, Joseph Abrams, Vishal Verma, Ting Fang, Cesunica Ivey, Mitchel Klein, Matthew J. Strickland, Stefanie E. Sarnat, Howard H. Chang, James A. Mulholland, Paige E. Tolbert, Armistead G. Russell

Oxidative stress is a potential mechanism of adverse PM2.5 health impacts. Previous research has linked PM2.5 oxidative potential to emergency department visits in Atlanta, GA related to cardiorespiratory outcomes such as asthma and congestive heart failure. However, due to lack of measurements, the spatial domain of ambient PM2.5 oxidative potential estimates has been limited. This work attempts to resolve this problem by using the relationship between oxidative potential measurements and CMAQ-DDM sources to simulate oxidative potential across the Eastern United States. Oxidative potential of water-soluble PM2.5 was measured using an acellular dithiothreitol (DTT) assay at five locations in the Southeastern United States, three urban and two rural, from June 2012-July 2013. Sixteen CMAQ-DDM source impacts were simulated during the same time period at a 12km resolution across the Eastern United States. These sources were adjusted at speciated monitoring locations and temporally and spatially interpolated using a CMAQ-DDM/CMB hybrid method. Sources of secondary pollutants (organic carbon, nitrate, sulfate, and ammonium) were further adjusted using pollutant measurements and an advanced correction algorithm to obtain best estimates. Final source impact estimates were related to oxidative potential measurements at all sites using least squares regression with backward selection. This model was applied to daily source impacts across the Eastern United States to obtain a large spatial map of estimated PM2.5 oxidative potential. Furthermore, the regression and CMAQ-DDM/CMB source impacts were compared to a regression obtained using CMB source impacts at just one oxidative potential measurement site in Atlanta, GA on which previous work was based.

A satellite-dispersion modeling system to generate high-resolution downscaled PM2.5 fields

Frank Freedman (San Jose State University)

Mohammad Al-Hamdan (USRA, NASA MSFC)

Akula Venkatram (University of California Riverside)

Sen Chiao (San Jose State University)

A hybrid-modeling tool coupling routine MODIS satellite aerosol optical depth overpasses and AERMOD-based dispersion modeling is being developed to support efficient generation of high resolution spatial fields and time series of fine particulate species over a community-scale area of interest. An envisioned application is public health tracking at community scale, which requires both air quality estimates at sub-kilometer resolution and daily time scale. The system builds on the PM2.5 database at CDC-WONDER (https://wonder.cdc.gov/wonder/help/PM.html), which was developed using MODIS aerosol optical depth to generate PM2.5 estimates nationally at 10-km resolution and daily time scale from 2003 to 2011. The system in development augments the methodology behind this database by enabling efficient and routine updates to near real-time, and ability to downscale to high resolution using AERMOD-based dispersion models. Additional features to enhance efficient, near-real time implementation are a Lagrangian Background Model (LBM) to efficiently estimate secondary PM and 3-km WRF-HYYR meteorological fields to drive the dispersion model. These remove the need for in-house regional grid modeling (e.g. WRF and CMAQ) to capture background secondary PM and potentially spatially complex wind flow affecting the area of interest. Direct emissions of surface PM and secondary precursors are supplied by coupling to a SMOKE-derived CMAQ emission input fields, provided to us by SCAQMD. The poster and extended abstract will present details of the system configuration, emphasizing linkage of components and implementation details, as well as preliminary results comparing to air monitors at routine and near-roadway sites in Southern California. The work comprises a portion of the team's contribution to the NASA H-AQAST project, which promotes and establishes connections to stakeholders to better enable use of satellite air quality data at air quality management and public health agencies.

Back-trajectory modeling of high time-resolution air measurement data to separate nearby sources

Gayle Hagler, Daniel Birkett, Ronald C. Henry, Eben Thoma

Strategies to isolate air pollution contributions from sources is of interest as voluntary or regulatory measures are undertaken to reduce air pollution.  When different sources are located in close proximity to one another and have similar emissions, separating source emissions trends in situ is difficult.  During 2012-2015, the EPA conducted the Region 2 Port-area Investigation of Emissions Reduction (R2PIER) project which collected 1-minute air quality and meteorological data at a site just south of the Port of Newark, New Jersey.  This monitoring site was situated to maximize the ability to separate clustered sources, including the Port, the Newark International Airport, the New Jersey Turnpike, and other surrounding sources.  Over three years, approximately 1.7 million measurements were made for PM_2.5, sulfur dioxide (SO₂), oxides of nitrogen (NO₂, NO), carbon monoxide (CO), black carbon (BC), and local meteorology.  New analytical approaches were developed to separate slowly varying from fast varying components of the time series, isolating the apparent component of the pollution time series attributable to local direct emissions.  This separation indicated a significant component of the time series had a slowly varying characteristic, contributing a significant fraction of sulfur dioxide (43%), nitrogen oxides (56%), carbon monoxide (76%), particulate black carbon (59%), and particulate matter less than 2.5 micrometers in aerodynamic diameter (PM_2.5, 73%).  The direct emissions impacts isolated from the time series were input, along with meteorology, into the Nonparametric Trajectory Analysis (NTA) model.  The model results indicate that the highest or second highest average concentrations of these pollutants were associated with air that came from the Port of Newark. These Port-attributed concentrations decreased by up to ~50% during the study. The notable exception was PM_2.5 which increased during the study period.  

Development of Two Model Fusion Techniques Utilizing CMAQ and R-LINE to Obtain NOx, CO and PM2.5 Concentrations at 250m Resolution over Atlanta, GA

Josephine Bates, Audrey Flak Pennington, Xinxin Zhai, Mariel D. Friberg, Francesca Metcalf, Matthew Strickland, Lyndsey Darrow, James Mulholland, Armistead Russell

Accurately capturing steep spatial gradients in air pollutant concentration at a fine spatial resolution is critical for accurate exposure estimates in epidemiologic studies. Dispersion models capture these gradients but lack chemistry and complete emissions inventories. Chemical transport models resolve complex chemistry and comprehensive emissions from many sources but are limited by coarse spatial resolutions. This work presents the development of two novel, computationally efficient model fusion techniques that simulate pollutant concentrations at a fine spatial resolution, one method for gaseous pollutants and one method for particulate matter. These methods are applied to 12km CMAQ and 250m R-LINE results that were fused with observations a priori to develop daily estimates of 1-hr maximum NOx and CO and 24-hr averaged PM2.5 over Atlanta, GA from 2002-2011 at a 250m resolution. The method for particulate matter places primary roadway PM2.5 in their respective locations inside the 12km CMAQ grid using 250m R-LINE after removing primary roadway PM2.5 from CMAQ to avoid double counting. The method for gaseous pollutants is slightly different due to high biases in R-LINE and rescales the R-LINE results using a linear adjustment factor between CMAQ and R-LINE. These model fusion results clearly illustrate steep gradients near roadways while retaining secondary pollutant formation and impacts from both local and regional sources. Furthermore, the model fused estimates show an improved spatial and temporal R2 compared to the observation-fused CMAQ and R-LINE results, emphasizing the need for fine spatial resolution to limit biases in pollutant exposure assessment when using modeled data. Finally, these methods can be applied to other pollutants using different models at different spatial scales providing extreme flexibility of use.

Influence of ozone urban background on nitrogen dioxide concentration near roadway sources in Barcelona city (Spain)

Jaime Benavides, Michelle G. Snyder, Marc Guevara, Fulvio Amato, Xavier Querol, Oriol Jorba, Albert Soret

High NO2 levels in ambient environments have been associated with adverse health impacts. NO2 levels within urban areas are elevated due to the proximity of the source (i.e. roadways). Primary NO and NO2 emissions from roadways, NO2 from background sources and local NO-O3 chemistry are the main contributors to NO2 concentrations near road sources in urban environments. The oxidation of urban NO by O3 to form NO2 strongly depends on the availability of O3 supplied from outside the city. We combine observations of NO, NO2, and O3 with model results to analyse the influence of O3 urban background concentrations on urban NO2 in the Barcelona city. First, we estimate the contribution of NO-O3 reaction to NO2 concentrations near road sources using pollutant observations from urban background stations and from an experimental campaign conducted in April 2013 in streets of Barcelona city. Then, we apply CALIOPE-Urban (CMAQ-WRF-HERMES-RLINE) to investigate how background O3 levels affect urban NO2 concentration in April 2013.

Investigate the air pollution problem in mountain area of Taiwan using fine scale air quality simulation

Fang-Yi Cheng, Chia-Ying Lin, Chia-Hwa Lin

The complex topography complicates the flow patterns that also affect air pollution dispersions. In Taiwan, the air pollutants are mainly produced in western coastal region. During the day, the air pollutants can be transported to the inland area along with the sea breeze and affect the inland mountainous area such as Puli basin that is located in central mountainous area of Taiwan. The objective of this study is to investigate the mountain/valley wind structures and its interaction with the land-sea breeze circulation and its subsequent impact on the air pollutants dispersions in Puli basin. To achieve this goal, the fine scale air quality model simulation at 1-km resolution is applied in Taiwan to understand the air pollution problem in central mountainous area of Taiwan. The preliminary results indicate that the air pollutants are transferred from the coastal emission source region to the inland Puli Basin through the sea breeze and valley flow during the day. In addition, the air pollutants are accumulated in the near surface layer in Puli Basin due to the subsidence that is caused by the re-circulation flow. The Details of the results will be presented during the workshop.

WUDAPT-based urban-WRF and HYSPLIT-STILT modeling of San Jose, California: Developments to support case study applications for urban air pollution and heat islands.

Frank Freedman (San Jose State University)

Lucrecia Rivera (San Jose State University)

Collaborative work is ongoing to implement WUDAPT Level 0 local climate zone (LCZ) fields for urban morphology into urbanized-WRF (BEP and BEP-BEM), and apply these to better simulate urban air pollution and heat islands in San Jose, California. Various versions of WUDAPT Level 0 fields (~ 100 m resolution) are being developed using methods for determining training areas that combine analysis of LANDSAT images, county-level zoning data, NDVI, NLDAS land cover data, and local lidar measurements of building heights. The LCZ fields are being incorporated into WRF BEP and BEP-BEM for case study evaluations of urban air pollution and heat-islands. The poster will present a sampling of these case study analyses. The goals of these simulations are to better model and understand important urban-scale pollution and heat island phenomena affecting downtown San Jose and the southern San Francisco Bay Area, with the longer-term vision of applying similar methods for other locations in California, and to support the satellite-based hybrid fine-particulate modeling system being developed by the lead presenter as part of his NASA H-AQAST work (see poster proposal submitted for "Modeling to Support Exposure and Health Studies and Community-scale Applications" session).

The air pollution simulations are carried out by coupling the WRF output to HYSPLIT run in STILT-emulation mode, where HYSPLIT is run backwards in time (1-km meteorology and concentration-calculation grids) using a representative receptor in an urban area-of-interest as the particle release point. Modeled HYSPLIT concentration fields then map footprints estimating the concentration contribution per unit surface emission affecting the receptor, which when multiplied by surface emissions and time-space integrated give the modeled concentration at the receptor. Simulations of CO2 and PM2.5 are being carried out using gridded surface emissions fields on a 1-km grid for CO2 and 4-km for direct PM2.5. The method is novel in that it is both physically advanced and efficient for estimating the local background concentration affecting an urban area-of-interest, which can then be coupled to fine-scale dispersion model output for full hybrid air pollution model application.

Adaptation of Meteorology and R-LINE to Street Canyon Micro-Climates: Application in Barcelona City Spain

Michelle G. Snyder, Jaime Benavides, Marc Guevara, Fulvio Amato, Xavier Querol, Albert Soret, Oriol Jorba

Roadway dispersion models, such as R-LINE, are used to estimate air quality and human health impacts at the street level in urban environments. Traditionally, meteorological measurements taken at airport stations are used as input data for modelling the dispersion of traffic-related pollutants within street canyons due to the lack of measurements within the city. However, depending on the airport location, uniformity of the urban landscape, and source location, this meteorology (e.g. wind speed and wind direction) may not be representative of the dynamics observed in the street canyons. Inaccurate meteorology can lead to uncertainties when modelling the dispersion of pollutants, especially in complex urban environments such as street canyons. The use of mesoscale meteorological models to provide gridded meteorological values for each street can be seen as an alternative to the aforementioned methodology. First, we explore the appropriateness of airport and grid-based meteorology to an urban built environment. Next, we explore techniques to adjust grid- and airport-based meteorology to mimic wind fields of an urban street canyon. This meteorology is then input to a roadway dispersion model, R-LINE, to simulate dispersion of roadway pollutants within the city. Effects of micro-climate meteorology on near-source dispersion estimates are evaluated using a measurement campaign in Barcelona city (Spain) for April 2013.

WRF urban-scale modeling with WUDAPT
Jason Ching1 Gerald Mills2, Benjamin Bechtel3 Linda See4, Adel Hanna1

1CEMPD, UNCIE, 2University College of Dublin, 3University of Hamburg, 4IIASA

The WUDAPT (World Urban Database and Access Portal Tools) project is an international initiative designed to generate globally consistent information on urban form and function. Its critical design benefit is to support mesoscale to fine grid scale modeling and urban climate science in urban scale environmental assessments and development of mitigation and adaptation strategies. The WUDAPT database organises urban data to different heirarchical levels of precision and detail; its portals export parameter values and facilitates usage to various modelling schemes including the popular WRF and by extension, WRF driven air quality simulations. The WUDAPT database is generated in a consistent, highly innovative manner using freely available data and crowdsourcing GIS-based and agreed upon typologies for urban layouts and buildings. The coarsest data (Level 0), generates urban landscape maps using the Local Climate Zone (LCZ) categorization system. Each LCZ has an associated range of values of model relevant descriptors (e.g. roughness, impervious surface cover, roof area, building heights, etc.) linked to local climate impacts. LCZ maps are created using Landsat 8 data, guidance by experts with local knowledge of their cities and free geographic, machine learning software; results are currently available for more than 100 of the world s largest cities each with its unique distribution of LCZ, with 100 m resolution, and across all continents. At higher Levels, high precision, finely gridded specific parameter values are generated and added information on building morphology, material composition data and energy usage for running advanced urban modeling applications are acquired. We illustrate a variety of WUDAPT based urban modeling applications, including intraurban heat stress advisories, air quality simulations, and past-to-present meteorology fields with evolving gowth of urban areas.

Examining Changes in Ozone over the Western United States via Assimilation of Satellite Ozone Products in a Chemistry-Transport Model
Gregory Osterman, Jessica Neu, Thomas Walker, Dejian Fu, Susan Kulawik, and Kevin Bowman
Jet Propulsion Laboratory/California Institute of Technology + BAERI for Susan

We will present results from simulations of the GEOS-Chem CTM utilizing assimilation of different satellite ozone data products that we have performed as a preliminary step in a project to try to better characterize and understand changes in background ozone. As in our previous work using 2005-2010 data from the Tropospheric Emission Spectrometer (TES) (Verstraeten et al. [2015]), we focus on changes in ozone over the Western US and attribution of those changes to regional emissions, long-range transport from Asia and stratosphere-troposphere exchange. Here we use a new dataset retrieved from the NASA Atmospheric Infrared Sounder (AIRS) and Ozone Monitoring Instrument (OMI). This combined AIRS/OMI ozone dataset is based on the retrieval methods developed for TES and allows for the analysis to extend beyond 2010, when TES stopped making global measurements. We will show examples of the AIRS/OMI data, comparisons to ozonesondes, and results from assimilating satellite data into GEOS-Chem. We will also evaluate the GEOS-Chem results using surface data to show the effects of the satellite assimilation on surface ozone.

Integrating regional and local modelling to create a high-resolution air quality forecasting system for Hong Kong

Christina Hood, Jenny Stocker, David Carruthers, William Grayson, Jonathan Handley, Jimmy Fung, David Yeung

The air quality in urban areas shows complex variations in both time and space, due to influences on many scales from regional pollutant emissions to individual road emissions and street canyon geometries. Access to street-scale air quality forecasting data allows residents to make informed choices about transport modes and routes in order to minimise their exposure to air pollutants as they travel around the urban area.
Regional photochemical models such as CMAQ can be used to forecast average pollutant concentrations on neighbourhood scales, but this does not give an accurate representation of the local concentrations, which can vary significantly over distances of a few metres from a road source. Local models such as ADMS-Urban are able to capture the small-scale dispersion and chemistry processes which occur close to individual sources but do not take into account the longer-term transport and chemistry processes affecting emissions from further afield.
Coupling a regional and a local model creates a computationally efficient system for calculating high-resolution predictions of pollutant concentrations, although care is required to avoid double-counting emissions. Such a system was described in the 2014 CMAS conference by Stocker et al., with all components running in the Windows operating system.
For the Hong Kong forecasting application the coupled system has been modified to run on a Linux HPC, with further automation to reduce user inputs and improve its suitability for operational air quality forecasting. Control scripts and utility programs process output files from the mesoscale meteorological model WRF and the regional photochemical model CMAQ to create inputs to ADMS-Urban runs and combine all the outputs into maps of air quality at resolutions as fine as 5 m near roads for a three day forecast. Detailed building geometries are used to calculate the street canyon properties affecting dispersion from each road source and to parameterise the local wind speed and turbulence including urban canopy effects.
The presentation will give an introduction to the concept of the coupled modelling system and its implementation as an operational forecasting system. The scientific performance of the system will be assessed against monitoring data and the computational performance will be demonstrated. Intended future development of the system to incorporate real-time traffic monitoring data will briefly be described.

Burn Area Comparisons between Prescribed Burning Permits in Southeastern USA and two Satellite-derived Products

Ran Huang and M. Talat Odman

Prescribed burning is one of the most prominent sources of PM2.5 in the southeastern US. The emissions estimates are based on the burn areas reported by the states, which may be subject to significant uncertainty since not all prescribed burns have reliable records. Satellite-derived products could be used as a substitute tool to provide burn area data. In order to evaluate burn areas from satellite-derived products and assess whether they can be used in prescribed fire burn area estimation, we conducted a comparison between prescribed burning permit records and satellite-derived burn areas for Georgia and Florida on the first four months of 2015 and 2016, which is the most active burn season in those two states, with two satellite-derived products: Blended Polar Geo Biomass Burning Emissions Product (BBEP) and Global Fire Emissions Database (GFED4s). We also conducted a survey to evaluate the level of uncertainty in the permit recorded data provided by local governments. The survey showed that permit recorded data are more reliable than expected. The comparison results indicate that both satellite-derived data underestimate the burn areas compared to permit recorded data. The comparison results of BBEP burn areas and permit recorded data for 2016 are different from those for 2015, with no correlation between them on daily analysis; on the other hand, GFED4s has good correlations with permit record data on monthly analysis for both 2015 and 2016. Satellite-derived products can capture a cluster of fires better than isolated fires, but may misinterpret those small fires together as one big fire, as shown in our specific days' analysis in Florida. Considering the openness and wide area of the sugarcane plantations and the high frequency of the burns, sugarcane burns should be detected more efficiency by the satellites. However, burn area comparisons of BBEP with permit records show only a slight improvement compared to other types of burn. Satellites can be more effective in detecting fires over deciduous forests in winter, as shown by the 2015 Georgia data. Overall, current satellite-derived products have limitations in estimating the burn areas of small fires; therefore, they cannot be used as a reliable alternative in estimating prescribed burning emission.

CALIOPE-urban: coupling R-LINE with CMAQ for urban air quality forecasts over Barcelona

Jaime Benavides, Michelle G. Snyder, Marc Guevara, Fulvio Amato, Xavier Querol, Carlos Perez Garcia-Pando, Albert Soret, Oriol Jorba

CALIOPE is a mesoscale air quality modelling system that provides 48 hour air quality forecasts at 12 km horizontal resolution over Europe, 4 km over Spain and 1 km over urban areas (e.g. Barcelona city). The CALIOPE system integrates the Weather Research and Forecasting meteorological model (WRF), the High-Elective Resolution Modelling Emission System (HERMES), the Community Multiscale Air Quality Modeling System (CMAQ) and the mineral Dust REgional Atmospheric Model (BSC-DREAM8b). Performance of the system has been tested over the last 10 years in several evaluation studies, proving CALIOPE to be capable of reproducing the observed patterns and variability of pollutant levels in urban and rural environments. Nevertheless, in urban environments some pollutants (e.g. NO2) present strong concentration gradients near urban sources (e.g. roadways) that cannot be reproduced by CALIOPE since large concentration variations can occur within a grid cell. In order to overcome this limitation, the combination of regional and urban scale models is presented as a solution to successfully estimate air quality at the street level within the larger grid cell. This work describes a methodology to couple CALIOPE with the Research LINE source dispersion model (R-LINE). R-LINE was developed to predict these strong gradients for roadways, which has been adapted to provide air quality estimates at street level within Barcelona's geometrical conditions (i.e. street canyon pattern). The coupled modeling system is evaluated using ambient street-level pollutant and meteorological measurements collected during a field study in April 2013 and compared with the current mesoscale solution applied to Barcelona.

A Novel Approach for Identifying Dust Storms with Hourly Surface Air Monitors in the Western United States

Barry Baker^1,2, Li Pan^1,2, Pius Lee¹, Youhua Tang^1,2, Daniel Tong^1,2,3

1 NOAA Air Resources Laboratory, College Park, MD

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

3. Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA 22030

A new algorithm is developed to identify windblown dust events with hourly surface air monitors over the continental United States. The method uses the U.S. EPA Air Quality System (AQS) to provide hourly data of PM₁₀, PM_2.5, CO and wind speed. Ganor et al. (2009) used an approach using 30 min PM₁₀ observations in Tel-Aviv. This approach is modified for use within the U.S. Tong et al. (2012) developed a method for use with the IMPROVE network in which the PM_2.5 to PM₁₀ ratio is designated to be less than 0.2 during a local windblown dust event. This additional requirement is added to all collocated PM_2.5 to PM₁₀ monitor sites. Likewise, collocated wind speed and CO measurements are used to identify windblown dust events vs fugitive dust. The algorithm uses all available monitors at each location to get a best guess of windblown dust. Comparisons are shown between the Tong et al. (2012) method and the newly developed dust detection algorithm. A long-term trend analysis is conducted for Arizona. The results show that while anthropogenic emissions and overall PM10 concentrations decreases, the frequency of dust storms and their contribution to aerosol loading have increased considerably in the past decades, implying future challenges of dynamic air quality management in the Southwestern states under a changing climate.

Photochemical Modeling of Industrial Flare Plumes using SCICHEM

Prakash Karamchandani¹, Lynsey Parker¹, Greg Yarwood¹, Ron Thomas², Marvin Jones²

¹Ramboll Environ, Novato, CA; ²Texas Commission on Environmental Quality (TCEQ), Austin, TX

Plumes from industrial flares can promote active photochemistry occurring at spatial scales too fine to be resolved by grid models such as the Comprehensive Air quality Model with extensions (CAMx) and the Community Multi-scale Air Quality (CMAQ) model. SCICHEM is an open-source state-of-the-science puff model with gas, aerosol, and liquid phase chemistry modules comparable in detail to those in CAMx or CMAQ. The latest version, SCICHEM 3.1, includes the Carbon Bond 6 revision 2 (CB6r2) photochemical mechanism, which is used by TCEQ for ozone SIP modeling with CAMx. Previous studies with SCICHEM have shown that it can simulate the HRVOC-NOx-ozone chemistry of the Houston ship channel plume, as confirmed by comparisons with aircraft data for ozone, NOy (total oxidized nitrogen compounds) and chemical tracers of reacted HRVOC. Thus, SCICHEM is a suitable tool for modeling flaring scenarios. The study described in this paper involved the selection of 3 industrial flaring events of HRVOC within Texas that could be linked definitively to elevated ozone events, as well as base case and sensitivity modeling of these flaring events with SCICHEM. TCEQ developed a hierarchical process to search and analyze the State databases for flares that met the necessary criteria. For the SCICHEM modeling, we developed emissions files for the 3 flare scenarios and developed meteorological inputs using available surface and upper air measurements in the vicinity of the flares. We compared ozone formation in the flares with ozone events measured at downwind monitors to determine the flare contributions to these events. For flares with relatively low HRVOC emission rates, these contributions were small (less than 5 ppb). For a wintertime flare with high ethylene emission rates during the night, significant ozone formation (nearly 30 ppb) was predicted during the following day. The study shows that the model is an efficient tool to quantify flare event contributions to ozone formation downwind.

Investigation of the Atmospheric Conditions Governing the Relationship between Aerosol Optical Depth and Surface PM2.5 Concentrations in the Western U.S.

S. Marcela Loria-Salazar

Anna Panorska

W. Patrick Arnott

James C. Barnard

Jayne M. Boehmler

Heather A. Holmes

Data fusion models used to estimate spatially resolved exposure metrics rely on columnar aerosol optical depth (AOD) from satellite retrievals to be used as a spatial surrogate of surface PM2.5 concentrations. Those models show optimistic results in the eastern U.S. because of the strong correlation between AOD and PM2.5. However, models based on purely statistical approaches are not able to represent surface concentrations of aerosol pollution in the western U.S. because they are challenged by complex atmospheric conditions (e.g., planetary boundary layer mixing, regional and long-range transport of aerosol pollution), air pollution composition, and uncertainty of satellite retrievals due to instrument calibration as well as non-ideal model parametrizations and assumptions. Therefore, models based on purely statistical relationships may not be able to capture the physical conditions governing the relationship between AOD and PM2.5 in the western U.S. Thereby requiring an intensive examination of the atmospheric conditions that drive the complicated relationship between AOD and PM2.5 to improve surface estimates of spatial PM2.5. The main goals of this investigation are to: 1) Study the atmospheric processes that impact the relationship between AOD and PM2.5 using ground-based sunphotometry as "ground-truth" and 2) Determine the scenarios under which AOD from satellite retrievals can be used as spatial predictors of surface PM2.5 according to weather and aerosol optical properties measurements. The results show that AOD and PM2.5 are correlated when local sources of pollution or wildfires dominate aerosol pollution under a well-mixed, deep planetary boundary layer (PBL). Moreover, AOD and PM2.5 are anti-correlated when stable PBL conditions, long-range transport, or entrainment from above the PBL are present. Results also show that classifications based solely on seasonality to develop statistical relationships between AOD and PM2.5 may not account for the anti-correlated trends observed between AOD and PM2.5.

Impact of grid resolution on health burdens from air pollution modelled with CMAQ-Adjoint

Melanie Fillingham, Amir Hakami, Burak Oztaner, Amanda Pappin, and Shunliu Zhao (Carleton University); Wayne Boulton, Greg Conley, Martin Gauthier, Jeff Lundgren, Julia Veerman, Carol McClellan (RWDI); Louise Aubin, Eric Tran, Kim McAdam (Regional Municipality of Peel Public Health Department; Matt D. Turner, and Daven K. Henze (University of Colorado); Shannon Capps (Drexel Uiversity); Peter B. Percell (University of Houston), Rest of adjoint development team.

Air quality models (AQM) paired with epidemiological data are widely used to estimate health related burdens from air pollution exposure. One method involves integrating the adjoint sensitivity of the AQM (e.g., CMAQ) with epidemiological data. Previous research using this method has been modeled at coarse resolutions (i.e., 36 km). Coarse grid resolutions may reduce the model accuracy by averaging population, emissions, baseline mortality, and concentrations across a large area, causing dilution within the grid and diminishing the representation of the actual spatial variability. The purpose of this study is to analyze the impact grid resolution has on estimated monetized health benefits due to emission reductions, hereby referred to as marginal benefit (MB). We do so by modeling the monetary health benefits (relating to mortality) from chronic exposure to O3 and NO2 (presented separately and combined) due to reductions of NOx emissions. The results were simulated on a regional scale at progressively refined resolutions (36, 12, 4 and 1 km) to compare grid resolution impacts.

The adjoint of gas-phase CMAQv5.0.2 paired with Health Canada's Air Quality Benefits Assessment Tool (AQBAT) is used to estimate MBs. The simulation includes 2 weeks in both February and July, 2012, over the Greater Toronto Area. Adjoint, a backward sensitivity approach, allows for the spatiotemporal influences of specific emission source to be traced and quantified. The results are presented as MB ($ health benefit over the entire domain / tonne reduction of NOx emissions per grid cell), as well as in monthly total damage over the domain (MB monthly NOx emissions). When presented as MB, for NO2 and O3 exposure combined, preliminary results show the maximum value at the 1 km resolution was up to 10 times larger than the maximum value at the 36 km resolution ($6.3 vs $0.6 million/tonne, respectively), indicating the dilution of localized maxima at coarse resolution. In other words, the fine grid results depict specific locations in which the highest benefits are seen, and therefore those areas in which emission reduction efforts should be focused. Preliminary integrated total damage results across the domain were $2.8B at 1km resolution, compared to $2.0B at 36km. Seasonal influences manifested in the magnitude of the results, with the July MB values being near double (if not higher) than the February values at all resolutions; the trend between resolutions was similar between seasons. Continuation of the project will also include the assessment of PM MB.

Lightning NOx estimates from space-based lightning imagers
William Koshak

The methodology for estimating lightning NOx production from space-based lightning imagers (TRMM/LIS, GLM, and ISS/LIS) will be examined. Preliminary results and trends for TRMM/LIS will be provided.

An air quality impact study for the recent increase in oil and gas activity over the Conterminous U.S.

Pius Lee ¹, Jeff MQueen², Ivanka Stajner³, Daniel Tong^1,4,5, Youhua Tang^1,5, Li Pan^1,5, Jianping Huang^2,6, Hyuncheol Kim^1,5, Barry Baker^1,5, Ho-Chun Huang^2,6, Sikchya Upadhayay^3,7

1. Air Resource Laboratory (ARL), NOAA, College Park, MD

2. Environmental Modeling Center, National Centers for Environmental Prediction, National Weather Service (NWS), NOAA, College Park, MD

3. Science and Technology Integration, NWS, NOAA, Silver Spring, MD

4. Center for Spatial Information science and System, George Mason University, Fairfax, VA

5. Cooperative Institutes for Satellite and Climate, University of Maryland, College Park, MD

6. IMSG, Rockville, MD

7. Syneren Technologies, Arlington, VA

The recent oil and gas activity boom over the Conterminous U.S. is rather steady and wide spread for the past few years. Environmental authorities are paying attention to the possible adverse effects caused by this increase in oil and gas exploration and production. The Marcellus Shale Play represents the fastest expanding oil and gas production region in the country in terms of both its number of wells and production output increase in double digit percentage points since 2007, flattening out only in the past year or so. Therefore the National Air Quality Forecasting Capability (NAQFC) team devised an air quality impact study to estimate a possible upper bound of such activities influencing the surface ozone concentrations in the country. The upper bounds in oil and gas emissions were computed by assuming that the latest increases of oil and gas related activities were as brisk as the most ambitious industrial projections. We consulted both the private sector reports as well as the Energy Information Administration to prescribe the upper bound oil and gas output increases, which are mainly due to gas production by hydraulic fracking. Leakage of air pollutants were estimated as well as resulting process related pollutants. These potentially most adverse estimated conditions were prescribed as input to a NAQFC-beta system in the summer of 2017 for real-time forecasts. Results and verification will be shared and analyzed.

Improving boundary conditions for regional models to address over-prediction of ozone influx from the Gulf of Mexico

Nopmongcol, O., M. Zatko, J. Jung, and G. Yarwood, Ramboll Environ
S.Daniel, B. Exum, and J. Smith, Texas Commission on Environmental Quality

Global models are used to prepare boundary conditions (BCs) for regional-scale air quality modeling, including the Texas Commission on Environmental Quality''s (TCEQ) modeling for the Continental US (CONUS). Regional modeling for the Texas Gulf Coast region is strongly affected by ozone over-predictions in air arriving from the Gulf of Mexico. Recent improvements to the Comprehensive Air Quality Model with extensions (CAMx) by Ramboll Environ have reduced the bias through addition of halogen chemistry to the Carbon Bond version 6 (CB6) chemical mechanism. However, the BCs derived from GEOS-Chem did not include the influence of halogen chemistry and consequently over-predicted concentrations of ozone along the modeling domain boundaries. This study incorporated halogen chemistry into the GEOS-Chem global model with a goal to further reduce bias incident upon the Texas Gulf Coast. This paper describes the effects of halogen chemistry on ozone concentrations and compares predicted halogen species to available observations.

Evaluation of the NOAA NAM-CMAQ predictions for recent air quality episodes

Jeff McQueen, Amanda Sleinkofer, Jianping Huang, Ho-Chun Huang, Perry Shafran, Vijay Tallapragada

Pius Lee, Li Pan, Youhua Tang, Daniel Tong, Ivanka Stajner, Sikchya Upadhayay

Operational air quality predictions for the United States (U. S.) are provided at NOAA by the National Air Quality Forecasting Capability (NAQFC). NAQFC provides nationwide operational predictions of ozone and particulate matter. Predictions are produced twice per day (at 06 and 12 UTC cycles) at 12 km resolution and 1 hour time intervals through 48 hours and distributed at http://airquality.weather.gov. The NOAA National Centers for Environmental Prediction (NCEP) operational North American Mesoscale (NAM) 12 km weather prediction is used to drive the Community Multiscale Air Quality (CMAQ) model. In 2017, the NAM was upgraded in part to reduce a warm 2m temperature bias in Summer (V4). At the same time CMAQ was updated to V5.0.2. The models were run in parallel for several months with the old and current version of NAM and with the two versions of CMAQ. Therefore the impact of improvements from the atmospheric chemistry model versus upgrades with the weather prediction model on ozone and PM2.5 predictions could be assessed for several air quality episodes during the Summer of 2016 and Spring of 2017. Ozone prediction biases were found to be reduced equally by the upgrade to CMAQ and NAM however, the impact on PM2.5 was smaller. The NAM 2 m temperature bias was improved by increasing the opacity of clouds which also resulted in reduced incoming short wave radiation. The reduced shortwave radiation resulted in reduced ozone photolysis thereby reducing the CMAQ ozone over-prediction bias in summer.

Higher resolution operational NWP models have recently been introduced as part of the NCEP model suite. These include the NAM CONUS Nest (3 km horizontal resolution) run four times per day through 60 hours and the High Resolution Rapid Refresh (HYYR, 3 km) run hourly out to 18 hours. In addition, NCEP with other NOAA labs has begun to develop and test the Next Generation Global Prediction System (NGGPS) based on the FV3 global model. This presentation also overviews recent developments with operational numerical weather prediction and evaluates the ability of these models for predicting low level temperatures and capturing boundary layer processes important for driving air quality prediction. The assessed meteorological model errors could help determine the magnitude of possible pollutant errors from CMAQ if used for driving meteorology. The NWP models will be evaluated against standard and mesonet fields averaged for various regions during the summer. An evaluation of meteorological fields important to air quality modeling (eg: near surface winds, temperatures, moisture and boundary layer heights, cloud cover, precipitation) will be reported on.

Comparing Methods for Tracking Large-Scale Ozone Background in Continental-Scale CMAQ Applications

Peng Liu, Christian Hogrefe, Shawn Roselle, Tanya Spero, William Hutzell, and Deborah Luecken

As the National Ambient Air Quality Standards (NAAQS) for ozone become more stringent, there has been growing attention on characterizing the contributions and the uncertainties in ozone from outside the US to the ozone concentrations within the US. Modeling techniques readily available in CMAQ to estimate such contributions or to estimate sensitivity of ozone to boundary conditions include inert tracers, the Integrated Source Apportionment Method (ISAM), and the Decoupled Direct Method in 3 Dimensions (DDM-3D). The computational burden associated with applying these methods ranges from minimal in the case of inert tracer to significant in the case of ISAM and DDM-3D. In this study, we present the implementation of a chemically reactive tracer for ozone from boundary conditions in CMAQ and compare it against prior simulations using chemically inert tracers. The simulations and analyses are performed for the year 2010 and leverage prior analysis performed in the third phase of the Air Quality Model Evaluation International Initiative (AQMEII3) which intercompared chemically inert tracers from different models. Implementation of the reactive tracer has only moderate impacts on CMAQ run time. Simulations are performed for both a base case scenario and a scenario without North American emissions to quantify how the differences between the inert and reactive tracer approaches depend on the chemical environment. Results are discussed in terms of temporal and spatial variations and directions for future work.

What factors contribute to O3 overestimation by the CMAQ Model in the Great Lakes Region

Momei Qin, Yongtao Hu, M. Talat Odman, Armistead G. Russell, Arastoo P. Biazar, Kevin Doty and Richard T. McNider

While meteorological models appear to represent many of the physical features of the lake meteorological systems, air quality models appear to overestimate ozone concentrations over cooler bodies of water, such as the Great Lakes Region. This may be due to a number of reasons. In this study, WRFv3.8.1, EPA 2011 version 6.2 Platform and CMAQv5.1 were used to simulate O3 in the Great Lakes Region throughout July 2011, with 12 and 4 km grid resolutions (one-way nested). Model performance was evaluated against the observed concentrations of ground-level O3 and related gases (i.e. NOx, NOy and NOz) obtained from EPA's AQS dataset. The evaluation shows that CMAQ tends to simulate higher O3 than observed, particularly along Lake Michigan shore, both after midnight and in the afternoon when O3 peaks. Higher positive biases for NOx and NOy occur around 6 a.m. and 8 p.m CST. This may be due to weaker stability caused by the PBL formulation in the models and warmer lake surface temperatures as suggested by MODIS. Several other factors may be responsible for high biases of simulated O3 over this region, including (1) O3 dry deposition velocity over freshwater, (2) biogenic emissions (BEIS vs. MEGAN), (3) possibly overestimated mobile NOx emissions, (4) gas chemistry mechanism (cb05 vs. cb06), (5) plume growth and transport. In this presentation, the effects of each one of these factors on O3 and its related gases will be reviewed in an effort to identify the reasons for the overestimation.

Impact of Chemical Lateral Boundary Conditions (LBC) from GEOS-5 Global Chemical Transport Model Compared to the Operational NGAC LBC

Youhua Tang^1,2, Huisheng Bian^3,4, Zhining Tao^3,5, Luke Oman³, Daniel Tong^1,2,6, Li Pan^1,2, Pius Lee¹, and Barry Baker^1,2 , Sarah Lu⁷, Jun Wang⁸, Jeffery McQueen⁸, Sikchya Upadhayay⁹, Ivanka Stajner⁹

1. NOAA Air Resources Laboratory. 2.CICS, University of Maryland, College Park, MD 20740. 3. NASA Goddard Space Flight Center 4. University of Maryland at Baltimore County, 5. Universities Space Research Association, Columbia, MD 21046 6. George Mason University, Fairfax, VA 22030. 7. Atmospheric Sciences Research Center, State University of New York at Albany 8. NOAA/NCEP/Environmental Modeling Center 9. NOAA/NWS/Office of Science and Technology Integration

The existing National Air Quality Forecasting Capability (NAQFC) operated in NOAA provides operational forecasts of ozone, PM_2.5, fire/smoke, and dust over the contiguous 48 states with Community Multi-scale Air Quality (CMAQ) model. Currently NAQFC is using static chemical lateral boundary conditions (LBCs), which can not capture the stratospheric ozone intrusion events or long-range transported pollutant plumes originated from outside of the model domain. In this study, we tested the chemical LBCs from GEOS-5 global chemical transport model (GCTM) as it is the prototype of future NOAA operational GCTM. To achieve this, we 1) developed the method of mapping GEOS-5 chemical species to CMAQ species; and 2) developed the interface to interpolate GEOS-5 results to CMAQ grid. We tested the GEOS-5's monthly-mean LBC and dynamic LBC for selected months of 2015. During this period, the GEOS-5's LBCs condition increased the NAQFC's domain-wide ozone, leading to a slightly higher ozone biases. However, these GEOS-5 LBCs yielded better correlation and root mean squared error (RMSE) for both ozone and PM_2.5, suggesting that it better captured the spatial variations of external influences, especially during the Canadian fire intrusion event. The dynamic GEOS-5 LBC generally yielded better correlation than the monthly mean GEOS-5 LBC. We also compared GEOS-5's LBC to current operational NEMS GFS Aerosol Component (NGAC) LBC which only contains the aerosols. Both global LBCs improved the NAQFC's PM2.5 prediction, especially over Southeastern States where the Sahara dust intrusion frequently occurred during summertime. Their differences due to their different dynamics and physics etc are also discussed

Measurements and Modeling of Turbulent Fluxes during Persistent Cold Air Pool Events in Salt Lake Valley, Utah

Xia Sun, Heather A. Holmes, Cesunica E. Ivey

Land surface processes are important in meteorology and climate research since they control the partitioning of surface energy and water exchange at the earth's surface. The surface layer is coupled to the planetary boundary layer (PBL) by surface fluxes, which serve as sinks or sources of energy, moisture, momentum, and atmospheric pollutants. Quantifying the surface heat and momentum fluxes at the land-atmosphere interface, especially for different surface land cover types, is important because they can further influence the atmospheric dynamics, vertical mixing, and transport processes that impact local, regional, and global climate.

A cold air pool (CAP) forms when a topographic depression (i.e., valley) fills with cold air, where the air in the stagnant layer is colder than the air aloft. Insufficient surface heating, which is not able to sufficiently erode the temperature inversion that forms during the nighttime stable boundary layer, can lead to the formation of persistent CAPs during wintertime. These persistent CAPs can last for days, or even weeks, and are associated with increased air pollution concentrations. Thus, realistic simulations of the land-atmosphere exchange are meaningful to achieve improved predictions of the accumulation, transport, and dispersion of air pollution concentrations.

The focus of this presentation is on observations and modeling results using turbulence data collected in Salt Lake Valley, Utah during the 2010-2011 wintertime Persistent Cold Air Pool Study (PCAPS). Turbulent fluxes and the surface energy balance over seven land use types are quantified. The urban site has an energy balance ratio (EBR) larger than one (1.276). Negative Bowen ratio (-0.070) is found at the cropland site. In addition to turbulence observations, half-hourly WRF simulated net radiation, latent heat, sensible heat, ground heat fluxes during one persistent CAP event are evaluated using the PCAPS observations. The results show that sensible and latent heat fluxes during the CAP event are overestimated. The sensitivity of WRF results to large-scale forcing datasets, PBL schemes and land surface models (LSMs) are also investigated. The optimal WRF configuration for simulating surface turbulent fluxes and atmospheric mixing during CAP events is determined.

2017 Projections and Interstate Transport of Ozone in Southeastern U.S.

Talat Odman and Yongtao Hu
School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA, 30332-0512, USA

James Boylan
Environmental Protection Division, Georgia Department of Natural Resources, Atlanta, GA, 30354, USA

The "good neighbor" provision of the Clean Air Act (CAA) holds each state responsible for the contribution of its emissions to air pollution in downwind states. The goal of this project was to quantify the contributions of southeastern states to ozone levels across the eastern US in the year 2017, and to determine if there would be any links to nonattainment or interference with maintenance of the ozone National Ambient Air Quality Standards (NAAQS) at downwind monitors. We conducted air quality modeling with two different models: Comprehensive Air quality Model with extensions (CAMx) and Community Multiscale Air Quality Model (CMAQ) using the US Environmental Protection Agency (EPA) 2017 modeling platform, and calculated future year design values (DVFs) for monitors across the eastern US. CB6r2 chemistry mechanism in CAMx resulted in generally larger ozone DVFs compared to the CB05 mechanism in CMAQ. CAMx modeling was accompanied with source apportionment analysis using the Anthropogenic Precursor Culpability Assessment (APCA) tool. CMAQ modeling was followed by brute-force sensitivity analysis where oxides of nitrogen (NOx) emissions from each state were successively zeroed out.

In this paper, total anthropogenic NOx emissions' contributions to ozone calculated by the sensitivity analysis with CMAQ will be compared to those calculated by CAMx-APCA, first by using the maximum Relative Response Factor (YYF) in the 3x3 cell block around each monitor as recommended by EPA in the Cross State Air Pollution Rule (CSAPR), and second by using the monitor cell's YYF. Contributions of EGU NOx emissions and anthropogenic VOC emission to ozone DVFs will also be discussed for the linkages between SESARM states and downwind nonattainment or maintenance sites.

Evaluation and intercomparison of five dry deposition algorithms in North America

Zhiyong Wu^1,*, Donna Schwede², Robert Vet¹, Mike Shaw¹, Ralf Staebler¹, John T Walker², Leiming Zhang¹

¹Environment and Climate Change Canada, Toronto, ON, Canada

²US Environmental Protection Agency, Research Triangle Park, NC, USA

*Now is an ORISE Fellow at US Environmental Protection Agency, Research Triangle Park, NC (wu.zhiyong@epa.gov)

To quantify differences between dry deposition algorithms commonly used in North America, five models were selected to calculate dry deposition velocity (V_d) for O₃ and SO₂ over a temperate mixed forest in southern Ontario, Canada where a five-year flux database had previously been developed. The models performed better in summer than in winter with correlation coefficients for hourly V_d between models and measurements being approximately 0.6 and 0.3, respectively. Differences in mean V_d values between models were as much as a factor of 1.7 in summer and greater than a factor of 2 in winter. Model differences were mainly due to different surface resistance parameterizations for stomatal and non-stomatal uptake pathways, while differences in aerodynamic and quasi-laminar resistances played only a minor role. Modeled V_d in summer was sensitive to the minimum stomatal resistance and, in winter, to the canopy snow cover fraction. Using model forecasted meteorology instead of measured to drive one model in which non-stomatal uptake was parameterized as a function of friction velocity produced 30% higher V_d values for O₃ and SO₂. It is recommended that users of inferential dry deposition models consider an uncertainty factor of 2 or use ensemble modeling results.

The Performance Evaluation of Lightning-NO Algorithms in CMAQ

Daiwen Kang, David Wong, Kristen Foley, George Pouliot, Wyat Appel, and Shawn Roselle

In the Community Multiscale Air Quality (CMAQv5.2) model, we have implemented two algorithms for lightning NO production; one algorithm is based on the hourly observed cloud-to-ground lightning strike data from National Lightning Detection Network (NLDN) to replace the previous monthly NLDN based algorithm and the other is a parameterization scheme based on linear relationship between historically observed NLDN data and model predicted convective precipitation. To evaluate the impact of these algorithms on model performance, four model simulation cases for the time period from April to September in 2011 were conducted. The four cases are: 1) no lightning NO, 2) the monthly NLDN based algorithm, 3) hourly NLDN based algorithm, and 4) the linear regression algorithm. Ground-level O₃ and NO_X from Air Quality System (AQS) and nitrate (NO3) from the National Atmospheric Deposition Program (NADP) are used to assess the model performances in time and space.

Estimating the Tipping Point of Urban NOx Control in Two Major U.S. Cities

Angele Genereux, Amanda Pappin, Amir Hakami (Carleton University)

Recent studies have shown that although nation-wide ozone concentrations in the U.S. have consistently declined over the past decade, increased concentrations of ozone appear in urban and suburban areas. The public health impacts of the increase in ozone concentrations with NOx control in these locations are termed "dis-benefits". These dis-benefits have been quantified in recent publications at current emission levels, suggesting that progressively reducing anthropogenic NOx emissions would eventually turn these dis-benefits into benefits as the region shifts from NOx-inhibited to NOx-limited chemical regime.

We investigate the dynamics of dis-benefits of NOx control through the adaptation of the U.S. EPA's CMAQ model. We use the adjoint of CMAQ to estimate the derivative of an exposure-based cost function, which utilizes non-accidental respiratory mortality rates due to chronic ozone exposure. With this, we estimate both national and local monetized health damages with respect to NOx emissions in the U.S. as benefit-per-ton [BPT]. We investigate how compounding benefits of NOx control compare to dis-benefits in two major urban areas: New York and Los Angeles. We identify the tipping point [TP] of NOx control beyond which negative BPTs become positive along the abatement pathway. We also estimate the break-even point [BEP] where cumulative benefits compensate immediate dis-benefits in urban areas. Furthermore, we examine the impact of applying nationalized vs. localized emission control policies in these two metropolitan areas. For surface-level emissions in NY and LA, we show how the compounding nature of marginal benefits calls for more aggressive emission control policies. For example, abating local emissions in LA results in BPT estimates which start at more than $-500k/ton NOx and increase until the TP is reached at ~50% abatement. The BEP occurs at an abatement level of 86% and BPTs continue to increase, before reaching a maximum of $2100k/ton NOx at 100% abatement.

Effects of interactions between meteorology and ambient pollutants on simulated air quality over metropolitan regions in Japan

Satoru Chatani, Masayuki Takigawa, Kyo Kitayama, and Kazuyo Yamaji

Ambient concentrations of photochemical oxidants and PM_2.5 still exceed the environmental quality standards in Japan. Effective strategies should be implemented to reduce their concentrations. Air quality simulations are useful for considering effective strategies taking nonlinear relationships between ambient pollutant concentrations and precursor emissions into account. We have started a study aiming at establishing reference air quality modeling in Japan through model inter-comparison. Although most of participants are using CMAQ as it is widely used in Japan, some participants are using WRF-Chem. Even if the consistent input emissions are used in CMAQ and WRF-Chem, significant differences appear in simulated ozone and PM_2.5 concentrations over Japan. It is not so easy to find out exact reasons of the differences as two models differ in various aspects.

One of advantages of WRF-Chem is that it can represent interaction between meteorology and ambient pollutants. Meteorological fields simulated by WRF-Chem shows lower solar radiation, lower surface temperature, lower planetary boundary layer height, and higher surface relative humidity around metropolitan regions in Japan than those simulated by WRF. Differences in meteorological fields simulated by WRF-Chem and WRF result in nonnegligible differences in pollutant concentrations simulated by CMAQ. PM_2.5 concentration simulated with the meteorological fields simulated by WRF-Chem is a few ug/m³ higher around metropolitan regions in Japan due to more stable atmosphere as well as conditions to promote partitioning to aerosol phase in thermodynamic phase equilibrium. That should be recognized as one of uncertainties in air quality simulations in which most of modelers in Japan are using CMAQ without interaction between meteorology and ambient pollutants.

Tracking Chemical History of Pollutant Plumes in East Asia using HYSPLIT-CMAQ Tool during the 2015 MAPS-Seoul Campaign

Hyun Cheol Kim ^1,2, MinAh Bae³, Tianfeng Chai^1,2, Fong Ngan^1,2, Ariel Stein^1,2, Changhan Bae³, Eunhye Kim³, 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
3 Ajou University, Dept. of Environmental Engineering, Suwon, Korea
4 Georgia Environmental Protection Division, Atlanta, GA

Regional transport pathways and physical/chemical processes of pollutant plumes and their precursors in East Asia were investigated using Lagrangian and Eulerian analyzing system with in-situ and remote sensing measurements. Continued rapid industrialization in East Asian countries has made this region a very complicated source of anthropogenic emissions, including large industrial complexes and multiple megacities. Air quality in the Seoul Metropolitan Area (SMA), South Korea is manifold because this region's air quality is affected not only by its own emissions sources but also by transported pollutants and their precursors from even stronger emissions sources (e.g. northern China). Understanding formation process of pollutant and quantifying contributions from local and remote emission sources, especially during the transport across the Yellow Sea, are crucial to explain the SMA region's air quality and to set a direction of emission regulation policy. A hybrid tool to efficiently track pollutant plumes' movement and to analyze physical and chemical processes within the plume was developed using NOAA ARL HYSPLIT model and EPA CMAQ chemical transport model. We set the HYSPLIT to launch multi- thousand particles to demonstrate realistic movement and dispersions of pollutant plumes. Each particle movement is linked to a CMAQ diagnostic tool, the Integrated Process Rate Process Analysis (IPR PA), to provide detailed information of the pollutants' formation and removal. For each in-situ measurements during the Megacity Air Pollution Studies - Seoul (MAPS-Seoul) 2015 campaign, the newly developed tool could provide chemical histories of each pollutant plumes based on the CMAQ PA analysis and satellite observations. Preliminary results show that particulate matter formation process during the transboundary transport differs for each component, implying that emission regulation policy in South Korea may need more attention for specific regulation target.

Examining Projected Changes in Climate and Ozone at 2050 and 2090 under Two Climate Scenarios

Chris Nolte, Tanya Spero, Pat Dolwick, Barron Henderson, Rob Pinder

Several published studies have found that climate change leads to higher concentrations of ground-level ozone over polluted continental regions. In those studies, increases in seasonal average ozone levels are often highly correlated with increases in average daily maximum temperatures, but other meteorological factors can also play important roles in determining air quality. In this work, we use the Weather Research and Forecasting model as a regional climate model (RCM) to dynamically downscale the NCAR/DOE Community Earth System Model over North America for decadal periods centered on 2000 as well as 2050 and 2090 for two different Representative Concentration Pathway (RCP) projections. These RCM scenarios are used as meteorological inputs for the Community Multiscale Air Quality modeling system to explore influences of regional climate change on air quality. Though summer average daily maximum temperatures increase 1-5 K throughout the continental U.S. under RCP8.5 at 2090, the ozone signal is more mixed, with increases of up to 10 ppb in summer average maximum daily 8-h ozone levels in the Northern Great Plains and Midwest regions, juxtaposed with decreases of up to 3 ppb projected in the Southeast. In this presentation, we will analyze the meteorological factors influencing ozone levels in these simulations, with a particular focus on the anticorrelated relationship between temperature and ozone changes in the Southeast.

Development of a Standalone Data Fusion Tool for Spatially and Temporally Fused Monitoring and Model Data

Carey Jang, Sharon Phillips, James Kelly and Tyler Fox, USEPA

We have developed a standalone Data Fusion (DF) tool to derive spatial and temporal fields of air quality monitoring and model data by using a suite of data fusion methods commonly used for providing an estimate of air quality impacts on health (e.g., Environmental Benefits Mapping and Analysis Program - BenMAP) and air quality attainment test tool (e.g., Software for Model Attainment Test - SMAT). The objective of this DF tool is to provide scientists and policy makers a user-friendly framework for creating desired spatial and temporal fields from existing air quality monitoring and model data and evaluating the results of fused spatial and temporal fields. This standalone DF tool currently provides three built-in data fusion techniques: eVNA (enhanced Voronoi Neighbor Averaging), VNA and and Downscaler, and has an easy-to-use graphical user interface (UI) to allow users to flexibly create spatial (entire or sub-domain) and temporal (annual, quarterly, monthly, or daily) data fields, conduct cross validation, and provide visualization analysis capabilities, as well as contains a built-in long-term ambient monitoring network database for PM2.5 and its component species and ozone. This new DF tool currently includes five key modules: (1) Data input module, (2) Data fusion technique module, (3) Spatial and temporal option module, (4) Cross validation module, and (5) Data viewer module, to provide the aforementioned functions and capabilities. A series of pilot case studies using national and regional U.S. model simulations (CMAQ and CAMx) are currently undertaken and the results of using various DF techniques on these case studies will be presented.

Projections of atmospheric nutrient deposition to the Chesapeake Bay watershed

Patrick C. Campbell, Jesse O. Bash, Chris Nolte, Tanya Spero, Ellen J. Cooter, Kyle Hinson, Lewis Linker

Atmospheric deposition remains one of the largest loadings of nutrients to the Chesapeake Bay watershed. The interplay between future land use, climate, and emission changes, however, will cause shifts in the future nutrient deposition regime (e.g., oxidized vs. reduced nitrogen). Recent research indicates that the form of nitrogen deposition can have varying effects on ecosystem health, and that the impacts can be habitat or species specific. Furthermore, land use and climate changes are expected to alter key processes in the Chesapeake Bay watershed, and can potentially intensify the impact of excess nutrients. In this work, we modify the Noah land surface model in the Weather Research and Forecast (WRF) model to improve the physical connectivity of WRF/Noah when coupled to the Community Multiscale Air Quality (CMAQ) model. We use the modified WRF/CMAQ model system to explore the relative impacts of emission, land use, and climate changes on atmospheric nutrient deposition to the Chesapeake Bay watershed for a historical (1995 - 2004) and a future period (2045 - 2054). Regional WRF/CMAQ simulations are based on the dynamic downscaling of global Community Earth System Model (CESM) simulations that are used as initial and boundary conditions. The downscaled WRF simulations are also used as inputs to simulate agricultural practices in the agro-economic Environmental Policy Integrated Climate (EPIC) model used in the NH₃ bidirectional exchange module in CMAQ. Initial model simulations show that the modified WRF/CMAQ model system can reproduce the observed deposition and ambient concentrations well, and suggest that reductions in future emissions are the largest factor contributing to a total reduction in atmospheric nitrogen deposition; however, the impacts of land use changes on nutrient deposition have not yet been fully evaluated. As sources of atmospheric reactive nitrogen from the burning of fossil fuel (oxidized nitrogen) decline, the emissions, transport, and fate of atmospheric reactive nitrogen from agriculture (reduced nitrogen) are altered and will likely become the dominant form of atmospheric nitrogen loading to the Chesapeake Bay. Results from this work aid in developing effective policies to protect ecosystems from excess nitrogen deposition in the face of climate change.

Whats new in the Atmospheric Model Evaluation Tool (AMET) version 1.3

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

A new version of the Atmospheric Model Evaluation Tool (AMET) has been released. The new version of AMET, version 1.3 (AMETv1.3), contains a number of updates and changes from the previous of version of AMET (v1.2) released in 2012. First, the Perl scripts used in the previous versions of AMET have been replaced with R scripts, thereby removing the requirement to have Perl and Perl modules installed in order to use AMET. Second, several of the run scripts in AMET have been consolidated and input files removed from the system, resulting in enhanced functionality and greater ease of use. Single run scripts for the meteorological and air quality sides of AMET are now responsible for database and project creation, and populating the project with data. On the meteorological side of AMET, analysis of data from the Model for Prediction Across Scales (MPAS) is now supported. On the air quality side, there is now support for the AMON and FLUXNET networks. In addition, there is now a "batch" analysis option available, which will execute a user defined suite of analysis scripts (from none to all) with limited customization. A number of other updates and bug fixes have been incorporated in AMETv1.3, resulting in a system that is overall easier to install and use. Finally, AMETv1.3 has been ported to the GitHub repository, making code and script updates easier in the future and providing a mechanism for the user community to easily contribute their own codes and scripts to AMET. This presentation will cover all the updates discussed above and provide a brief demonstration of AMETv1.3.

Estimating the Sources of Black Carbon Deposition to the Himalayan Glaciers in Current and Future Climates through Dynamical Downscaling

Matthew J. Alvarado, Ekbordin Winijkul, Rebecca Adams-Selin, Eric Hunt, Christopher Brodowski, Chantelle R. Lonsdale, Drew T. Shindell, Gregory Faluvegi, Gary Kleiman, and Thomas M. Mosier

We used WRF-Chem v3.6.1 and a modified version of the ECLIPSE 5a emission inventory to investigate the sources impacting black carbon (BC) deposition to the Himalaya, Karakoram, and Hindu Kush (HKHK) region by downscaling predictions from the GISS-E2-R general circulation model. This work extends previous studies by simulating deposition to the HKHK region not only under current conditions, but also in the 2040-2050 period under two realistic emission scenarios and in three different phases of ENSO. Under current conditions, sources from outside our South Asian modelling domain have a similar impact on total BC deposition to the HKHK region as South Asian anthropogenic sources, with the outside contribution peaking in July. Most of the in-domain anthropogenic contribution is from industry (primarily brick kilns) and residential solid fuel burning. Under a no further control (NFC) emission scenario for 2040-2050, the relative contributions to BC deposition in the HKHK region are divided between in-domain anthropogenic sources and sources outside the domain. The in-domain anthropogenic BC deposition has significant contributions from industry, solid fuel burning, and diesel fuel. Under the mitigation (MIT) emission scenario, the relative contribution from South Asian anthropogenic sources is significantly reduced, with sources outside the domain becoming relatively more important to the remaining deposition. Furthermore, the relative contribution of on-road diesel fuel use to the remaining in-domain anthropogenic deposition is dramatically reduced, while the relative contribution from industry (including brick kilns) increases. The changes due to phase of ENSO are generally smaller than those between the NFC and MIT cases, and do not seem to follow consistent patterns with ENSO. Future work will use the high-resolution deposition maps developed in this project to determine the impact of different sources of BC on the albedos of glaciers in the HKHK region, and how these albedo changes affect glacier melt and water availability in the region.