Scholarly article on topic 'Impacts of interstate transport of pollutants on high ozone events over the Mid-Atlantic United States'

Impacts of interstate transport of pollutants on high ozone events over the Mid-Atlantic United States Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Atmospheric Environment
Keywords
{Ozone / "Air quality" / "Interstate transport" / "Decoupled direct method"}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Kuo-Jen Liao, Xiangting Hou, Debra Ratterman Baker

Abstract The impacts of interstate transport of anthropogenic nitrogen oxides (NOx) and volatile organic compound (VOC) emissions on peak ozone formation in four nonattainment areas (i.e., Baltimore, Philadelphia-Wilmington-Atlantic City, Pittsburgh-Beaver Valley and Washington, DC) in the Mid-Atlantic U.S. were quantified in this study. Regional air quality and sensitivities of ground-level ozone to emissions from four regions in the eastern U.S. were simulated for three summer months (June, July and August) in 2007 using the U.S. EPA's Community Multiscale Air Quality model with the decoupled direct method 3D. The emissions inventory used in this study was the 2007 Mid-Atlantic Regional Air Management Association Level 2 inventory, developed for State Implementation Plan screening modeling for the Ozone Transport Commission region. The modeling results show that responses of peak ozone levels at specific locations to emissions from EGU (i.e., electric generating unit) and non-EGU sources could be different. Therefore, emissions from EGU and non-EGU sources should be considered as two different control categories when developing regional air pollution mitigation strategies. Based on the emission inventories used in this study, reductions in anthropogenic NOx emissions (including those from EGU and non-EGU sources) from the Great Lake region as well as northeastern and southeastern U.S. would be effective for decreasing area-mean peak ozone concentrations during the summer of 2007 in the Mid-Atlantic ozone air quality nonattainment areas. The results also show that reductions in anthropogenic VOC emissions from the northeastern U.S. would also be effective for decreasing area-mean peak ozone concentrations over the Mid-Atlantic U.S. In some cases, reductions in anthropogenic NOx emissions from the Great Lake and northeastern U.S. could slightly increase area-mean peak ozone concentrations at some ozone monitors in the Pittsburgh-Beaver Valley and Washington, DC areas. However, the disbenefit of the slight increase in ozone concentrations attributed to the NOx emission controls was far outweighed by the overall ozone air quality benefits over the Mid-Atlantic region.

Academic research paper on topic "Impacts of interstate transport of pollutants on high ozone events over the Mid-Atlantic United States"

Contents lists available at ScienceDirect

Atmospheric Environment

journal homepage: www.elsevier.com/locate/atmosenv

Impacts of interstate transport of pollutants on high ozone events over the Mid-Atlantic United States

CrossMark

Kuo-Jen Liaoa *, Xiangting Houa, Debra Ratterman Baker b

a Department of Environmental Engineering, Texas A&M University-Kingsville, 700 University Blvd., MSC213, Kingsville, TX 78363, United States b Department of Atmospheric and Oceanic Sciences, University of Maryland-College Park, College Park, MD, United States

HIGHLIGHTS

• Impacts of pollutant transport on peak ozone concentrations were quantified.

• Responses of peak ozone formation to EGU and non-EGU emissions could be different.

• Contributions of anthropogenic NOx emissions to peak ozone levels were significant.

• Controls of anthropogenic VOC emissions would decrease peak ozone levels.

ARTICLE INFO

ABSTRACT

Article history: Received 26 March 2013 Received in revised form 23 October 2013 Accepted 28 October 2013

Keywords:

Air quality

Interstate transport

Decoupled direct method

The impacts of interstate transport of anthropogenic nitrogen oxides (NOx) and volatile organic compound (VOC) emissions on peak ozone formation in four nonattainment areas (i.e., Baltimore, Philadelphia-Wilmington-Atlantic City, Pittsburgh-Beaver Valley and Washington, DC) in the Mid-Atlantic U.S. were quantified in this study. Regional air quality and sensitivities of ground-level ozone to emissions from four regions in the eastern U.S. were simulated for three summer months (June, July and August) in 2007 using the U.S. EPA's Community Multiscale Air Quality model with the decoupled direct method 3D. The emissions inventory used in this study was the 2007 Mid-Atlantic Regional Air Management Association Level 2 inventory, developed for State Implementation Plan screening modeling for the Ozone Transport Commission region. The modeling results show that responses of peak ozone levels at specific locations to emissions from EGU (i.e., electric generating unit) and non-EGU sources could be different. Therefore, emissions from EGU and non-EGU sources should be considered as two different control categories when developing regional air pollution mitigation strategies. Based on the emission inventories used in this study, reductions in anthropogenic NOx emissions (including those from EGU and non-EGU sources) from the Great Lake region as well as northeastern and southeastern U.S. would be effective for decreasing area-mean peak ozone concentrations during the summer of 2007 in the Mid-Atlantic ozone air quality nonattainment areas. The results also show that reductions in anthropogenic VOC emissions from the northeastern U.S. would also be effective for decreasing area-mean peak ozone concentrations over the Mid-Atlantic U.S. In some cases, reductions in anthropogenic NOx emissions from the Great Lake and northeastern U.S. could slightly increase area-mean peak ozone concentrations at some ozone monitors in the Pittsburgh-Beaver Valley and Washington, DC areas. However, the disbenefit of the slight increase in ozone concentrations attributed to the NOx emission controls was far outweighed by the overall ozone air quality benefits over the Mid-Atlantic region.

Published by Elsevier Ltd.

1. Introduction

Near-surface ozone is formed in the atmosphere through photochemical reactions and is associated with adverse human

* Corresponding author. Tel.: +1 361 593 3898; fax: +1 361 593 2069. E-mail address: kuo-jen.liao@tamuk.edu (K.-J. Liao).

1352-2310/$ — see front matter Published by Elsevier Ltd. http://dx.doi.org/10.1016lj.atmosenv.2013.10.062

health effects (Bell et al., 2005; Devlin et al., 2012) and crop production loss (Hollaway et al., 2012). One important indicator of regional air quality is the daily maximum 8-h average ozone concentration, which is regulated by the U.S. Environmental Protection Agency (EPA) under the U.S. Clean Air Act (42 U.S.C. §7401 et seq., 1970). Ozone concentrations are typically measured in parts per million (ppm) or parts per billion (ppb). The National Ambient Air Quality Standards (NAAQS) for 8-h ozone was set at 0.08 ppm

(=0.84 ppb) in 1997 then lowered to 0.075 ppm (=75 ppb) in 2008 (73 FR 16436-16514, March 27, 2008). Regions that have an 8-h ozone "design value" (the three-year average of the annual fourth-highest daily maximum 8-h ozone concentration for a single monitor) above the NAAQS when it is implemented are designated "nonattainment". The Clean Air Act requires states that have non-attainment areas to submit State Implementation Plans (SIPs), which show how a particular region will reduce pollutant concentrations to the point where the region meets the NAAQS within a specified time period depending on the severity of their ozone problem.

Observations show that high summertime ozone concentrations have caused nonattainment of the NAAQS over many areas in the Mid-Atlantic region in the U.S. It is widely recognized that longrange transport of air pollutants is a contributing factor in ozone pollution issues in downwind areas of emission sources (Hains et al., 2008; Taubman et al., 2004). Nitrogen oxides (NOx = NO + NO2) and volatile organic compounds (VOCs) create ambient ozone in the presence of sunlight. Emissions of both ozone precursors come from anthropogenic and biogenic sources. However, NOx is primarily anthropogenic, with some from biogenic sources like the soil and lightning (Allen et al., 2012). VOCs are emitted by both anthropogenic and biogenic sources although anthropogenic sources are dominant in most urban areas (Baker et al., 2008). The chemistry of ozone is nonlinear in that an area can switch between two chemical "regimes": a NOx-limited regime and a VOC-limited regime (Finlayson-Pitts and Pitts, 1997). Most areas, particularly outside urban areas, generally fall into a NOx-limited regime, so NOx reductions are generally an effective ozone control measures. However, in some VOC-limited regimes, a reduction in NOx emissions can have an air quality "disbenefit" (i.e., it can increase ozone). The type of regime varies not only by location but also by time of day and season of the year.

To identify the cause of high ozone events in the summer (i.e., June, July and August) of 2007 in the Mid-Atlantic region, the goal of this study is to investigate how interstate transport of anthropogenic NOx and VOC affected daily maximum 8-h average ozone concentrations in Mid-Atlantic nonattainment areas. We chose the 2007 summer episode for this study since observations showed high daily maximum 8-h average ozone concentrations during the episode in the Mid-Atlantic region (Fig. 3). Bergin et al. (2007) and Hakami et al. (2006) used the sensitivity analysis approach to investigate the contribution of interstate emissions to air pollutant levels in the East and Continental U.S., respectively. This study is focused on peak ozone events over Mid-Atlantic ozone nonattain-ment areas. The results of this study are expected to provide useful information for improving ground-level ozone air quality over the Mid-Atlantic U.S.

2. Material and methods

2.1. Regional meteorology modeling

The 2007 meteorology used as input to air quality modeling in this study was prepared by the University of Maryland, College Park for modeling in support of state SIPs by the Maryland Department of the Environment (MDE) and the Ozone Transport Commission (OTC). 2007 was chosen as the base year because it reflected a variety of meteorological conditions that led to several exceedance days for ozone, PM2.5, and haze in the Northeast. The mesoscale meteorological model used was the Weather Research and Forecasting (WRF) model v.3.1 (Skamarock et al., 2008). The meteorology domain consisted of a 36 km domain (165 x 129) encompassing the continental United States and a nested 12 km domain (250 x 250) of the eastern United States (Fig. A1 in

Appendix). Using the 12 km horizontal resolution is sufficiently high to simulate mesoscale features and is also in the appropriate range to use both WRF microphysics and cumulus convection schemes. The vertical grid structure consisted of 34 layers with a higher number of levels in the lowest 1 km. The first level was set at ~20 m so the layer average could be compared directly with routine meteorology measurements, which were taken at 10 m.

The initial and boundary conditions for the WRF model were obtained from the National Centers for Environmental Prediction (NCEP) North American Mesoscale (NAM) analysis datasets. To minimize the accumulation of model errors and retain as much mesoscale circulation as possible, the Objective Analysis (OBSGRID) and Four-Dimensional Data Assimilation (FDDA) techniques were used to include observations of the surface winds and upper-level meteorological information (Gilliam et al., 2009). Physics configurations chosen in this study were tested by Baker et al. (2010) and presented in Appendix. The New York State Department of Environmental Conservation (NYSDEC) evaluated the WRF results, including comparing the modeling output to CASTNet (Clean Air Status and Trends Network) observations of wind speed and temperature. The results for wind speed and temperature evaluation are in Appendix (Table A1) and are "in fair agreement with the measured data" (Sistla et al., 2011).

2.2. Regional emissions inventory and modeling

The emissions inventory used for this study was the 2007 Mid-Atlantic Regional Air Management Association (MARAMA) Level 2 inventory, developed for OTC SIP screening modeling and released in 2010 (Vickers et al., 2011). The inventory includes criteria air pollutants (CAPs) and ammonia (NH3) for the Ozone Transport Region (OTR) plus Virginia (MACTEC, 2011). This is a "baseline" and "base case" inventory under EPA definitions and includes the following sources: point, area, on-road mobile, non-road mobile, and biogenic.

In OTR and Virginia, real-time 2007 Continuous Emissions Monitoring Systems (CEMS) data from Clean Air Markets Division (CAMD) was used for large electric generating unit (EGU) and non-EGU point sources (EPA, 2013a). Other point sources were based on the preliminary data from the 2008 National Emissions Inventory (NEI) with specific updates by the states (EPA, 2010). For area sources, the 2008 NEI data was supplemented by the states. For the other regions in the domain, the 2008 NEI data was used. The U.S. EPA released three different versions of the 2008 NEI, and the version 1 NEI was used in this emission modeling work.

The National Mobile Inventory Model (NMIM) model was used to estimate monthly non-road mobile sources while new inventories for marine vessels, airport, and rail were added (EPA, 2013b). The on-road mobile source emissions were based on updated 2007 vehicle, road, and trip length data from the states. A preliminary version of the Motor Vehicle Emissions Simulator (MOVES) model that has been identified as MOVES 2007 version 1 by MARAMA was used to produce on-road mobile emissions (MACTEC, 2011). The biogenic emissions were processed through the Model of Emissions of Gases and Aerosols from Nature (MEGAN) which uses landcover, meteorology, and air chemistry inputs to calculate the emissions of gases and aerosols from terrestrial environments (Guenther et al., 2006).

The Sparse Matrix Operator Kernel Emissions (SMOKE) model, v. 2.7, was used for the speciation, temporal, and spatial allocation of the emissions to the chemical mechanism, time period, and grid (Houyoux et al., 2000). SMOKE produced emission fields for different source categories (i.e., area, mobile, non-road and point sources) and merged emissions from all anthropogenic and

biogenic sources used as input to the regional air quality model in this study.

2.3. Regional air quality modeling

Atmospheric chemical transport models (CTMs) are widely used to improve the understanding of atmospheric physical and chemical processes and to develop air pollution mitigation strategies. In this study we used the U.S. EPA's Community Multiscale Air Quality (CMAQ) Model (Byun and Schere, 2006), version 4.7.1, released in June 2010, to simulate three-dimensional gridded concentrations of ambient ozone and other air pollutants. A uniform grid of 12 by 12 km horizontal cells with 34 vertical layers was employed in the CMAQ simulations. The depth of each vertical layer is not identical. The layers near the surface have smaller depth (the depth of the lowest vertical layer is ~20 m) to get more detailed air pollutant concentration information near ground level.

We used the Carbon Bond 05 (CB-05) gas-phase chemical mechanism (Yarwood et al., 2005) and AERO5 aerosol module. The Backward Iterative (EBI) homogenous chemical solver for CB05 was selected. The Meteorology-Chemistry Interface Processor (MCIP) v.3.5.beta was used to converts the WRF files for use as input files for CMAQ (Otte and Pleim, 2010).

Photolysis rates in CMAQ were calculated from look-up tables by the photolysis rate (J) preprocessor (JPROC). Horizontal diffusion was handled by the Multiscale Eddy Diffusivity Scheme (Byun and Schere, 2006). Vertical diffusion was represented by an eddy diffusivity approach which is a local mixing scheme and the only available module in CMAQ decoupled direct method 3D (DDM-3D) which was used in this study. Cloud convective mixing in CMAQ was handled with the asymmetric convective model (ACM) (Pleim, 2007). The Yamartino scheme was chosen for mass conservation and advection (Yamartino, 1993).

The simulations were conducted using initial and boundary conditions provided by CMAQ's initial conditions preprocessor (ICON) and boundary conditions preprocessor (BCON), respectively, using predefined concentration profiles. A seven-day spin-up time was run and discarded to minimize the effects of uncertainties in the initial conditions on modeling results. The boundary conditions prepared using BCON were used since boundary conditions from coarser grid resolutions (e.g., 36 km-by-36 km) were not available in this study. However, the areas of interest were significantly far away from the boundaries so that the effects of boundary conditions on modeling results were expected to be small. To investigate the impacts of interstate transport of air pollutants on ozone air quality in the Mid-Atlantic region, the CMAQ modeling domain

covers most part of the eastern U.S. and was divided into four regions that correspond with the Central Regional Air Planning Association (CENRAP), Lake Michigan Air Directors Consortium (LADCO), Ozone Transport Region and Virginia (OTR + VA) and Southeastern States Air Resource Managers and West Virginia (SESARM + WV) regional organizations (Fig. 1). Daily maximum 8-h average ozone concentrations, similar to the form of the NAAQS, in four Mid-Atlantic nonattainment areas: Baltimore, MD, Philadelphia-Wilmington-Atlantic City (PWA), PA-NJ-MD-DE, Pittsburgh-Beaver Valley, PA, and Washington DC-MD-VA (based on the 2008 standard) were specifically investigated in this study (counties of the four nonattainment areas are presented in Table A2 in Appendix).

2.4. Sensitivity analysis of impacts of interstate transport of pollutants on ozone air quality

To investigate the impact of interstate transport of air pollutants on ozone air quality in the Mid-Atlantic region, responses of daily maximum 8-h ozone concentrations to emission perturbations (i.e., sensitivities) were quantified. The CMAQ v. 4.7.1 includes the decoupled direct method 3D (DDM-3D) (Dunker et al., 2002b; Yang et al., 1997) for investigating sensitivities of air pollutant concentrations to precursor emission changes. DDM-3D calculates linearized (i.e., first-order derivative) local sensitivities (Sij) of species i concentration (Ci) to perturbations in emission ej, scaled by its nominal value Ej. Mathematically, the first-order sensitivity was calculated as:

J ~ Ej dej '

Due to the nonlinear chemistry involved in ozone formation, a major limitation of first-order sensitivities is that they may not adequately describe the ozone response if the magnitude of the emission change is large (Dunker et al., 2002a). In this study, the sensitivities are presented as responses of daily maximum 8-h ozone concentrations to a 10% change in emissions and can be extrapolated to 50% changes according to previous sensitivity studies (Cohan et al., 2005; Dunker et al., 2002b; Hakami et al., 2003). Another limitation of DDM is to assume a uniform change in emissions in source regions. The assumption may not be realistic in policy objectives. However, dividing the sources into four regions would, to some extent, address this issue. Here, positive sensitivities indicate that reductions in precursor emissions decrease pollutant levels. On the other hand, reductions in precursor emissions increase air pollutant concentrations if sensitivities are

Fig. 1. Air quality modeling domain (color regions), four emission regions (i.e., CENRAP (green), LADCO (blue), OTR + VA (yellow) and SESARM + WV (brown)) and four Mid-Atlantic ozone air quality nonattainment areas.

negative. The impact of anthropogenic NOx and VOC emissions from the four regions on daily maximum 8-h average ozone concentrations in the four Mid-Atlantic nonattainment areas was quantified based on the results of the sensitivity analysis. The sensitivity analysis approach can also be used to investigate responses of other air pollutants (e.g., particulate matters) to emission changes (Bergin et al., 2007). This study is focused on peak ozone events due to the severe ozone air pollution issue in the Mid-Atlantic U.S.

3. Results

3.1. Anthropogenic air pollutant emissions from EGU and non-EGU sources

Since NOx and VOC emissions cause ozone air pollution, the knowledge of amount of anthropogenic NOx and VOC emissions from the different regions and source categories is required to select effective management strategies for improving ozone air quality in the Mid-Atlantic U.S. Anthropogenic NOx and VOC are emitted from diverse sources. In air quality management, anthropogenic sources are usually divided into four categories: area, on-road mobile, non-road mobile and point sources. Area sources are small sources of air pollution whose emissions are too small to be measured individually. Examples of area sources include commercial and consumer products (such as hairspray and paints), gasoline refueling stations and printing facilities. Point sources are primarily manufacturing businesses that produce emissions equal to or greater than 10 tons per year (tpy) of VOCs or 25 tpy of NOx. Large industrial plants such as EGUs and chemical manufacturers are examples of point sources. Sources of air pollutant emissions that are not stationary are referred to as mobile sources and are broken down into two categories: on-road and non-road mobile sources. The former include cars, vans, trucks and buses (i.e., vehicles that operate on highways and other roadways). Non-road mobile sources include lawn and garden equipment, construction equipment, marine, airports, and locomotives. Emission sensitivities are broken down into EGU emissions and non-EGU emissions, where non-EGU emissions include non-EGU point sources, area sources, on-road mobile and non-road mobile emissions.

Fig. 2 shows emission rates calculated from the SMOKE outputs. It is noted that the emission rates presented in Fig. 2 may not be identical to emission inventories originally used by MARAMA since SMOKE split total emissions to different species (e.g., NOx to NO and NO2) and spatially and temporally allocated them to modeling grid cells. In this study, the total SMOKE-modeled daily anthropogenic NOx emissions from the four regions ranged from ~ 1800 to 4600 tons day-1 (Fig. 2a). Fig. 2a shows high anthropogenic NOx emissions from the LADCO, OTR + VA and SESARM + WV regions were from mobile sources. Similar to LADCO, OTR + VA and

SESARM + WV, mobile sources (including on-road and non-road) were significant anthropogenic NOx emission sources for the CENRAP region. EGU-point sources also had significant contributions (~ 700—1000 tons day-1) to total anthropogenic emissions for the LADCO, OTR + VA and SESARM + WV regions.

Similar to the anthropogenic NOx emissions, the SMOKE-modeled results show that OTR + VA (~6900 tons day-1), SESARM + WV (~4500 tons day-1) and LADCO (~4200 tons day-1) had high VOC emissions from anthropogenic sources (Fig. 2b). Anthropogenic VOC emissions from point sources were small for the four regions. Generally speaking, area sources were the most significant sources of anthropogenic VOC emissions among all anthropogenic emission sectors. On-road and non-road mobile sources also emitted significant amount of VOC in the LADCO, OTR + VA and SESARM + WV regions.

3.2. Model evaluation results

Modeled daily maximum 8-h average ozone concentrations were compared against observed ozone air quality data to investigate the capability of CMAQ estimating air pollutant concentrations during the episode simulated. Observations of ozone concentrations were retrieved from the U.S. EPA's Air Quality System (AQS; http://www.epa.gov/ttn/airs/airsaqs/, last accessed: 9/ 27/2013) database, which contains hourly air quality data, including ground-level ozone concentrations, collected by federal, state, and local agencies at thousands of locations nationwide. In the 2007 summer episode, there were 51 monitoring sites with hourly ozone concentration data in the four Mid-Atlantic non-attainment areas without missing data. The Baltimore area contained 7 ozone AQS monitoring sites, the Philadelphia-Wilmington-Atlantic City (PWA) area contained 14 ozone monitoring sites, the Pittsburgh-Beaver Valley area contained 13 ozone monitoring sites, and the Washington, DC area had 17 ozone monitoring sites. Following the EPA's guideline, air quality modeling results were evaluated using both qualitative (e.g., graphical) and quantitative (e.g., statistical) approaches (USEPA, 1991, 2005). The EPA's guidelines of using time series plots and statistics for evaluate modeling results were developed for single monitors. Since there were at least 7 monitors in each of the nonattainment areas, in this study, observed and modeled ozone concentrations were averaged over each of the areas for simplicity. Time series plots show that, for the 2007 summer episode, CMAQ successfully simulated high ozone events for all the nonattainment areas although daily maximum 8-h ozone concentrations were overestimated in Baltimore and Washington, DC (Fig. 3). Statistical measures were also used for the CMAQ performance evaluation. Statistical metrics used in this evaluation were the Mean Normalized Bias (MNB) and Mean Normalized Gross Error (MNGE) which are EPA-recommended metrics for air quality model performance evaluation at single

Fig. 2. SMOKE-modeled daily average anthropogenic (a) NOx and (b) VOC emissions for different emission categories for the four regions.

Fig. 3. Time series for modeled and observed daily maximum 8-h average ozone concentrations during the 2007 summer for (a) Baltimore, (b) Philadelphia-Wilmington-Atlantic City, (c) Pittsburgh-Beaver Valley and (d) Washington, DC. Note: the observed and modeled ozone concentrations were averaged across all monitors in each of the nonattainment areas.

monitors (USEPA, 1991, 2007). The largest MNB and MNGE were 11.8% and 15.3%, respectively, across all monitors in each of the four nonattainment areas (Table A3 in Appendix). The CMAQ simulation results satisfy the performance criteria of MNB < 15% and MNGE < 35%, which are suggested by EPA (USEPA, 1991), and are sufficiently accurate for this study.

3.3. Effects of interstate transport of pollutants on peak ozone events in the Mid-Atlantic U.S.

Since the simulation results include 92 days of 8-h ozone concentrations in the summer of 2007, we have focused the results on peak ozone events (i.e., days with modeled daily maximum 8-

h average ozone concentrations higher than 75 ppb). Given that ozone air quality management is usually considered as a regional issue, we averaged all modeled 8-h maximum concentrations in each of the nonattainment areas and then compared the average to the 75 ppb threshold. In this study, we refer to the average as the "area-mean" ozone concentrations.

3.3.1. Baltimore, MD

The Baltimore area has suffered from high ozone concentrations during the summertime. A long-term statistical study shows that ozone air pollution events in Baltimore were related to sunny and hot conditions as well as high wind speeds from the west and northwest (Walsh et al., 2008). The sunny and hot conditions

represent local meteorological effects, while the high wind speeds from the west and northwest prove the evidence of interstate transport of pollutants.

The results of sensitivity analysis show that anthropogenic NOx emissions from non-EGU sources in the OTR + VA region were the most important contributor to ozone formation during peak ozone days in the summer of 2007 in the Baltimore area (Fig. 4a). A 10% reduction in non-EGU anthropogenic NOx emissions from the OTR + VA region could decrease area-mean daily maximum 8-h average ozone in Baltimore by 1.2—2.5 ppb on peak ozone days (OTRV_ANOx in Fig. 4a). The main sources of anthropogenic non-EGU NOx emissions in the OTR + VA region were mobile vehicles, and they accounted for more than half of the total anthropogenic NOx emissions from the OTR + VA region (Fig. 2a). A 10% reduction in EGU NOx emissions from the OTR + VA region (OTRV_EGU_A-NOx) could also significantly decrease peak ozone levels (up to about 0.6 ppb) in the Baltimore area. The responses of high ozone concentrations to anthropogenic NOx emissions from EGUs and non-EGUs were different. It was likely that the differences in the simulated ozone responses were due to emissions being injected into different chemical regimes as a result of differences in spatial distribution of these sources (both horizontal and vertical).

In addition to emissions from OTR + VA, reductions in emissions from the other regions could also decrease peak ozone levels in the Baltimore area. A 10% reduction in anthropogenic NOx emissions from non-EGU sources in LADCO (LADC_ANOx) and SESARM + WV (SESA_ANOx) could decrease peak ozone levels by up to about 0.6 ppb and 0.9 ppb, respectively, in the Baltimore area (Fig. 2a). For VOC emissions, only anthropogenic VOC emissions from OTR + VA had a significant contribution to area-mean peak ozone formation, and the VOC contribution was similar to that of anthropogenic NOx from non-EGU sources in LADCO and SESARM (Fig. 4a). Anthropogenic VOC emissions were mainly from area, on-road mobile and non-road mobile sources for the OTR + VA region (Fig. 2b).

3.3.2. Philadelphia-Wilmington-Atlantic City, PA—NJ—MD-DE (PWA)

For all peak ozone days during the modeling episodes, reductions in anthropogenic NOx emissions from LADCO, OTR + VA and SESARM + WV were expected to be effective for decreasing the high O3 concentrations in the PWA area (Fig. 4b). A10% reduction in anthropogenic NOx emissions from EGU and non-EGU sources in OTR + VA was expected to decrease area-mean daily maximum 8-h ozone concentrations by up to about 0.3 and 2 ppb (on August 2), respectively. Reductions in anthropogenic VOC emissions from the OTR + VA region could also significantly decrease peak ozone concentrations in the PWA area (up to about 0.5 ppb for a 10% reduction). It was found that reductions in anthropogenic emissions from regions where PWA is not located in could also decrease peak ozone levels in PWA. A 10% reduction in anthropogenic non-EGU NOx emissions from SESARM + WV (SESA_ANOx in Fig. 4b) was expected to decrease the peak ozone level by up to about 1 ppb (on June 8) in PWA. Contributions of non-EGU anthropogenic NOx emissions from LADCO to peak ozone levels were significant since their reductions could effectively decrease peak ozone levels in PWA (up to about 0.8 ppb for a 10% reduction). Overall, reductions in anthropogenic NOx emissions from non-EGU sources in OTR + VA, LADCO and SESARM + WV as well as anthropogenic VOC emissions from OTR + VA would be effective control strategies for decreasing the area-mean peak ozone levels in PWA.

3.3.3. Pittsburgh-Beaver Valley, PA

The results of sensitivity analysis show that changes in daily maximum 8-h ozone levels in the Pittsburgh-Beaver Valley area were affected by precursors emitted from various anthropogenic

sources in the emission regions, except CENRAP (Fig. 4c). Reductions in anthropogenic NOx from both EGU and non-EGU sources in OTR + VA and LADCO were effective for decreasing ozone concentrations during most peak ozone days over the Pittsburgh-Beaver Valley area. A 10% reduction in anthropogenic NOx emissions from EGU and non-EGU sources in OTR + VA was expected to decrease ozone concentrations in the Pittsburgh-Beaver Valley area by up to about 0.5 (on July 9) and 1.1 (on August 2) ppb, respectively. A 10% reduction in anthropogenic NOx emissions from EGU and non-EGU sources in LADCO was expected to decrease ozone concentrations in the Pittsburgh-Beaver Valley area by up to about 0.5 (on July 9) and 1.0 (on August 3) ppb, respectively. In addition to emissions from OTR + VA and LADCO, reductions in anthropogenic NOx emissions from both EGU non-EGU sources in SESARM + WV could decrease peak ozone concentrations significantly in the Pittsburgh-Beaver Valley area. On July 9, 2007, a 10% reduction in both EGU and non-EGU sources in SESARM + WV could decrease the area-mean daily maximum 8-h ozone levels by about 1.0 ppb in the Pittsburgh-Beaver Valley area. The majority of the SESARM + WV impacts were likely coming from West Virginia due to the proximity to the Pittsburgh area.

3.3.4. Washington, DC-MD-VA

The Washington, D.C. Metropolitan Area results show that anthropogenic NOx emissions from EGU and non-EGU sources in the OTR + VA region had the most significant effects on peak ozone formation in Washington, DC (Fig. 4d). A 10% reduction in anthropogenic NOx emissions from non-EGU sources in the OTR + VA region could decrease peak ozone concentrations by up to 2.6 ppb (on August 2) in Washington, DC. In addition to anthropogenic NOx emissions from non-EGU sources, reductions in EGU NOx emissions from the OTR + VA region could also decrease peak ozone concentrations over the Washington, DC area (up to 0.7 ppb for a 10% reduction). The results of sensitivity analysis show that daily maximum 8-h ozone concentrations in Washington, DC were also affected by anthropogenic precursor emissions from the LADCO and SESARM + WV regions on peak ozone days during the modeling episode. Fig. 4d shows that a 10% reduction in anthropogenic NOx emissions from non-EGU sources in LADCO and SESARM + WV could decrease area-mean daily maximum 8-h average ozone concentrations by up to 2.7 and 1.2 ppb, respectively, in Washington, DC. On the other hand, the negative ozone sensitivity for July 10, 2007 implied that reductions in EGU NOx emissions from LADCO would slightly increase daily maximum 8-h average ozone concentrations (~0.6 ppb for a 10% emission reduction) in the Washington, DC area.

4. Discussion: regional ozone air quality management for the Mid-Atlantic U.S.

Since ozone air pollution is considered as a regional issue, the development of ozone air quality management strategies should be considered for all counties in the Mid-Atlantic nonattainment areas simultaneously. Observations show that daily maximum 8-h ozone concentrations were above the ozone NAAQS of 75 ppb on August 2 and 4 in all the Mid-Atlantic nonattainment areas (Fig. 3), and we considered developing ozone air quality management strategies for decreasing peak ozone levels on those two days as a case study in this discussion. Fig. 5a shows that, on August 2, reductions in anthropogenic NOx emissions from non-EGU sources in OTR + VA would decrease area-mean ozone concentrations in Baltimore, PWA and Washington, DC. On the other hand, the negative ozone sensitivity for the Pittsburgh-Beaver Valley area implied that such reductions would slightly increase ozone concentrations at some monitors in the area. However, the monitors with negative

Fig. 4. Sensitivities of area-mean daily maximum 8-h average ozone to 10% emission reductions (vertical axis on the left) as well as modeled (red circles) and observed (black diamonds) daily maximum 8-h average concentrations (vertical axis on the right) for (a) Baltimore, (b) Philadelphia-Wilmington-Atlantic City, (c) Pittsburgh-Beaver Valley, and (d) Washington, DC. Note: sensitivities of ozone to NOx emissions from EGUs in CENRAP as well as VOC emissions from all EGUs are = 0 and not shown in the figure. In all figures, AVG represents the average of ozone concentrations and sensitivities over all the peak ozone days.

sensitivities are not the highest monitors in the area (Table A4), so the disbenefit of the small increase in ozone concentrations at these monitors would be outweighed by the benefit of ozone air quality at the monitors with higher ozone concentrations in Pittsburgh and the other nonattainment areas.

On August 4, reductions in anthropogenic NOx emissions from non-EGU sources in OTR + VA would decrease ozone concentrations at all monitors in Baltimore and PWA (Fig. 5b). However,

ozone concentrations at some monitors in Pittsburgh and Washington, DC, would slightly increase since sensitivities of ozone concentrations at those monitors to the anthropogenic NOx emission (OTRV_ANOx) reductions were negative. The results also show that reductions in anthropogenic NOx emissions from non-EGU sources in LADCO as well as EGU NOx emissions from all the emission regions were effective for improving ozone air quality over the Mid-Atlantic region. Overall, the results show that,

~ K 0.5

August 2, 2007

I 1 1 1 i 1

f i i t 1 1

, I L i TT I

Î . 1 i I 1 T i,

LADC_ANOx ¡SESA EGU NOx

OTRV_ANOx LADC AVOC

60 I S

x <■>

40 2 =

I SESA_ANOx OTRV AVOC

LADC_EGU_NOx SESA AVOC

August 4, 2007

Baltimore PWA Pi

OTRV EGU NOx

Fig. 5. Sensitivities of daily maximum 8-h ozone to 10% emission reductions (vertical axis on the left) as well as modeled (red circles) and observed (black diamonds) daily maximum 8-h ozone concentrations (vertical axis on the right) for (a) August 2 and (b) August 4. Note: ozone sensitivities to emissions from CENRAP = 0 and not shown in the figure. Error bars represent standard deviations of ozone concentrations and sensitivities over each of the nonattainment areas.

reductions in anthropogenic NOx emissions from LADCO and OTR + VA as well as EGUs in the four emission regions would be effective for improving ozone air quality in the Mid-Atlantic region although ozone concentrations could slightly increase at some monitors with lower ozone concentrations in Pittsburgh and Washington, DC (Table A5). Finally, it should be noted that control strategies for attaining ozone NAAQS in each area would be determined by looking at the starting design value for each monitor over the Mid-Atlantic and applying relative response factors (RRFs) based on the sensitivity results (USEPA, 2007).

5. Conclusions

The impact of interstate transport of anthropogenic NOx and VOC on peak ozone concentrations in the four nonattainment areas in the Mid-Atlantic U.S. were quantified in this study. The modeling results show that the responses of peak ozone levels at specific locations to emissions from EGU and non-EGU sources in the four emission regions could be different. Therefore, reductions in emissions from EGU and non-EGU sources should be considered as two different control categories when developing regional air pollution mitigation strategies. Based on the emission inventories used in this study, reductions in anthropogenic NOx emissions from OTR + VA (i.e., the northeastern U.S.), SESARM (i.e., the southeastern U.S.) and LADCO (i.e., the Great Lake region) would be effective for decreasing area-mean peak ozone concentrations during the summer of 2007 in Mid-Atlantic U.S. although the reductions in anthropogenic NOx emissions from OTR + VA and LADCO could slightly increase ozone concentrations at some monitors in the Pittsburgh-Beaver Valley and Washington, DC areas. However, the disbenefit of the slight increase in ozone concentrations was far outweighed by the overall ozone air quality benefits over the Mid-Atlantic U.S. In addition to anthropogenic NOx emission reductions, controls of anthropogenic VOC emissions from area, on-road mobile and non-road mobile sources in OTR + VA would be effective for decreasing ozone concentrations during peak ozone days in the summer of 2007 over the Mid-Atlantic U.S.

Acknowledgments

We thank the U.S. EPA for providing funding under the STAR grant R835218. The authors also thank to Mike Ku, Winston Hao, Christian Hogrefe, and Eric Zalewsky of the New York Department

of Environmental Conservation for creating the CMAQ input files. Thanks to Prof. Da-Lin Zhang of the University of Maryland, College Park, for the WRF meteorology output. Thanks to Julie McDill and Susan Wierman of MARAMA for providing the emissions inventory and technical support. The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the U.S. EPA or other agencies.

Appendix

Physical configurations used in the WRF simulations:

PBL: Modified Blackadar (Zhang and Anthes, 1982; Zhang and Zheng, 2004);

Surface Layer: Pleim-Xiu (Pleim, 2007); Land Surface Model: Pleim-Xiu (Xiu and Pleim, 2001); Microphysics: WRF Single Moment 6 (Hong and Lim, 2006); Cumulus Convection: Modified Kain-Fritsch (Kain and Fritsch, 1990); and

Shortwave Radiation: Dudhia (Dudhia, 1989)

135°W 120°W 105°W 90°W 75°W 60°W

120°W 110°W 100°W 90°W 80°W

Fig. A1. Meteorology model outer and inner domains (Baker et al., 2010).

Table A1

Evaluation statistics for 2007 WRF output (Sistla et al., 2011). Summer represents June, July and August.

Variable

Wind speed (m s Temperature (K)

Annual Summer Annual Summer

Mean observed 3.09 2.41 284.4 295.5

Mean modeled 3.07 2.24 284.3 294.8

Mean bias 0.51 0.17 0.40 0.15

Root mean square error (RMSE) 1.62 1.25 2.42 2.29

Normalized mean bias (%) 21% 10% 0.2% 0.1%

Fractional mean bias (%) 11% -5% 0.2% 0.1%

Mean error 1.25 0.98 1.83 1.73

Normalized mean error (%) 53% 57% 0.6% 0.6%

Fractionalized mean error (%) 62% 71% 0.6% 0.6%

Correlation coefficient 0.63 0.53 0.94 0.92

Table A2

Nonattainment areas for 8-h ozone air quality (based on 2008 ozone NAAQS) in the Mid-Atlantic Region (http://www.epa.gov/airquality/greenbook/gnca.html, last accessed: 7/9/2013).

Baltimore. MD

Maryland: Anne Arundel Co, Baltimore City, Baltimore Co, Carroll Co and Harford Co

Philadelphia-Wilmington-Atlantic City. PA-MD-DE Delaware: Kent Co, New Castle Co and Sussex Co Maryland: Cecil Co

Pennsylvania: Chester Co, Delaware Co, Montgomery Co and Philadelphia Co Pittsburgh-Beaver Valley, PA

Pennsylvania: Allegheny Co, Armstrong Co, Beaver Co, Butler Co, Fayette Co, Washington Co and Westmoreland Co

Washington. DC-MD-VA District of Columbia: Entire District

Maryland: Calvert Co, Charles Co, Frederick Co, Montgomery Co and Prince George's Co

Virginia: Alexandria, Arlington Co, Fairfax, Fairfax Co, Falls Church, Loudoun Co, Manassas, Manassas Park and Prince William Co

Table A3

Statistical evaluations for the four Mid-Atlantic nonattainment areas (60 ppb was a used a threshold in performance evaluation as recommended by U.S. EPA, 1991. Guideline for Regulatory Application of the Urban Airshed Model, Research Triangle Park, JVC).

Baltimore PWA

Pittsburgh Washington, DC

Total number of 8-h ozone observation > 60 ppb in the summer of 2007

427 294

Mean observation 72.62 70.57 69.84 70.18

Std. observation 10.20 9.60 7.71 8.45

Mean model 77.03 73.37 71.90 78.13

Std. model 13.71 13.48 14.32 12.74

Mean normalized bias (MNB) 6.89 4.32 3.06 11.75

Mean normalized gross error 14.33 12.51 13.45 15.34

(MNGE)

MNB^EPM

x 100%,

— '55

mcNrorocNmroro oooooooo oooooooo oooooooo 00000000

rO"^r-.r-.TPmi-iOTP TPLnOlOltnmTPCim

MOOOlflrHNrHio

OOOOOTHOOO

000000000

CO CN CN

o o o o o o o o o odd

O O CO

m m CN CN

o o o o o o o o o o o o dodo

o o o o o o

o o o o o o

o o o o o o

0 0 0 0 0 0

MOrOCTl^COCOTOOrN

oromcNroooooooi-<o

LnttOOOIOOlNLnOO OOOCNOCNCNCNtOTP 0000000000

o o o o

o o o o

o o o o

o o o o

0 0 0 0

o o o o

o o o o

o o o o

o o o o

0 0 0 0

o o o o

o o o o

o o o o

o o o o

0 0 0 0

o o o o

o o o o

o o o o

o o o o

0 0 0 0

o o o o o o

o o o o o o

o o o o o o

o o o o o o

0 0 0 0 0 0

o o o o o o

o o o o o o

o o o o o o

o o o o o o

0 0 0 0 0 0

m m o o w m ^ •

MrsCDtDtNOOtlÛOO

CN r- to to r-. co o o

. CN CN CO

ro to to in 0 0 0 0 0

to CN TP m

Oi—'oot—^ci ci m ro r~» cN^ootor^r^tooto

^(¿otooocorHrHjM

tOTPrOrOCNCNCNCNrO 000000000

M31rHrHC3iN«3Ln

rocNrorocNromro oooooooo oooooooo 00000000

OlCNinmcNrOrOrOOl

GioorotOTprocirom tomtototomtor-»"^ ooooooooo ooooooooo 000000000

o o o o o o o o o o o o o o o

o o o ro o o ■

o o o o o o

o o o o o o o o dodo

o o o o o o

> o o o o o

o o o o o o

o o o o o o o o dodo

o o o o rn o

o o o o o o

o o o o o o

o o o o o o

0 0 0 0 0 0

0*5 tp m m

01 in H n

tp oi r-. r-.

in r-s 1-1 rH

tp oo oi m o

m CN O CN oo

oi oo r-. to tp

r-. to r-. ro

CNCNfOfOCNCNCNCNO

ooicioooi^mroromoooo ^P'-iTPfOtO'T^i-ii-iTptOOO tocNr-oomorororommo

TpinqtotqcNTpoooqcNi-jro dtNNNNOrirlriridd

o o o o o o o o o o o o o o o o o o o o o o o

o o o o o o o o o o o o o o o o o o o o o o o

o o o o o o o o o o o o o o o o o o o o o o o

o o o o o o o o o o o o o o o o o o o o o o o

o o o o o o o o o o o o o o o o o o o o o o o

^p 01 ro ro 01 00 31 o to ro r-. ro r-. to 01 m ro r-. r-. ^p o

01 CN ro ro CN 01 X 00 ro o o 00 01 r-> TP o CN 00 00 ro 01

in o oi oi o o CN oi K ro ro oi CN ro o K o oi oi oi o o oi

r-. 01 00 00 01 o 01 00 r-. 01 00 00 r-s 01 00 00 r-s 00 00 00 01 00

o o o o o o o ro 00 o o o o o o o o o o o o o o

o o o o o o o ^P 00 o o o o o o o o o o o o o o

to CN TP to LO ,-H iri 00 oi o to od to ro to iri ,-H iri < oi ,-H TP oi

r-. r-> to r-> 01 00 00 r-s r-. r-. 00 r-> to r-. 00 00 00 01 r-. 00 o 00

V o o o o o o o o o o o y h m o h oi

!5,m m M m m OO O 1 IN N

_ ^ o_ o_ o_ o_ o_ ' n 'o"o"o"o"o' es u u u u u

o o o o

o o o o o ■

o o o o o •

oooooooo

01 o o o o

-3 <u <U

«r O O < £ £

o _ V V

S O ,P ,P

Ç *« *« *« <3,2124:

<CQCQ03UIIS

^ifoforommmoim

OOOOOOi-CNTP

'o"o"o"o"o"o"o"o"o"o"o"o"o"o' uuuuuuuuuuuuuu

ï ¡5 ¡5 2 ¡5

P. "03 *ô3 *ô3 *ôî

td T3 TD

ro ro n3 fB

a g g g S S :

u 11 U 5) 3 3

«ZZZmm

Oh OH OH OH

Pittsburgh

Washington, DC

Mean 81.71 85.00 0.0000 1.5121 0.0000

Std 9.96 6.67 0.0000 1.0610 0.0001

Allegheny Co(003-0008) 81.00 64.07 0.1479 -1.8603 0.0538

Allegheny Co(003-0010) 80.00 64.07 0.1479 -1.8603 0.0538

Allegheny Co(003-0067) 72.00 78.70 0.2480 0.5275 0.1100

Allegheny Co(003-1005) 99.00 71.19 0.0804 0.6780 0.0281

Armstrong Co(005-0001) 100.00 74.17 0.3230 1.8534 0.0365

Beaver Co(007-0002) 72.00 79.49 0.9879 0.3280 0.4474

Beaver Co(007-0005) 72.00 86.26 1.0180 -0.0878 0.2728

Beaver Co(007-0014) 79.00 86.96 1.0212 0.1936 0.2337

Washington Co(125-0005) 74.00 86.10 0.0002 0.5049 0.6563

Washington Co( 125-0200) 69.00 64.33 0.0142 0.9553 0.2389

Washington Co(125-5001) 69.00 77.77 0.6106 0.4277 0.4196

Westmoreland Co( 129-0006) 81.00 66.22 0.0315 -0.6323 0.1665

Westmoreland Co(129-0008) 75.00 67.74 0.0009 -0.2192 0.4184

Mean 78.69 74.39 0.3563 0.0622 0.2412

Std 10.15 8.79 0.4077 1.0406 0.1957

Entire District(001-0025) 78.00 100.87 0.0000 2.1157 0.0016

Entire District(001-0041) 75.00 103.66 0.0000 2.7013 0.0021

Entire District(001-0043) 75.00 103.66 0.0000 2.7013 0.0021

Calvert Co (009-0011) 71.00 90.82 0.0000 2.8224 0.0004

Charles Co (017-0010) 72.00 83.07 0.0000 2.3219 0.0045

Frederick Co(021-0037) 88.00 92.15 0.0000 2.5489 0.0024

Montgomery Co(031-3001) 103.00 101.17 0.0000 3.6032 0.0029

Prince George's Co(033-0030) 84.00 99.55 0.0000 3.0994 0.0019

Prince George's Co(033-8003) 76.00 84.13 0.0000 2.3943 0.0026

Arlington Co(013-0020) 74.00 92.38 0.0000 1.0908 0.0017

Fairfax Co(059-0005) 78.00 101.92 0.0000 3.3124 0.0044

Fairfax Co(059-0018) 72.00 86.48 0.0000 2.6353 0.0031

Fairfax Co(059-0030) 72.00 91.72 0.0000 2.8649 0.0023

Fairfax Co(059-1005) 84.00 91.72 0.0000 2.8649 0.0023

Fairfax Co(059-5001) 89.00 91.72 0.0000 2.8649 0.0023

Loudoun Co(107-1005) 91.00 101.92 0.0000 3.3124 0.0044

Prince William Co( 153-0009) 73.00 89.16 0.0000 2.9953 0.0048

Mean 79.71 94.48 0.0000 2.7205 0.0027

Std 8.87 6.92 0.0000 0.5644 0.0012

0.0045 0.3476 0.0000

0.0025 0.1655 0.0000

0.2476 0.1331 0.0052

0.2476 0.1331 0.0052

0.5847 0.1282 0.0158

0.0892 0.3477 0.0047

0.2367 0.5101 0.0004

1.0642 0.0446 0.0426

0.7804 -0.0384 0.0219

0.9436 0.5125 0.0093

0.0025 1.1599 0.3712

0.0339 0.3657 0.2466

0.8847 0.1005 0.0610

0.0405 0.7083 0.1166

0.0032 1.0131 0.2425

0.3968 0.3937 0.0879

0.3978 0.3765 0.1214

0.0039 0.5138 0.0001

0.0037 0.5088 0.0004

0.0037 0.5088 0.0004

0.0033 0.5699 0.0000

0.0054 0.8289 0.0019

0.0025 0.3855 0.0000

0.0027 0.4013 0.0000

0.0037 0.4301 0.0000

0.0051 0.7249 0.0001

0.0046 0.6348 0.0005

0.0030 0.3541 0.0001

0.0056 0.6480 0.0014

0.0050 0.5477 0.0008

0.0050 0.5477 0.0008

0.0050 0.5477 0.0008

0.0030 0.3541 0.0001

0.0034 0.3820 0.0003

0.0040 0.5228 0.0005

0.0010 0.1338 0.0005

0.0000 0.1570 0.0002

0.0000 0.2271 0.0001

0.0000 0.3949 0.0047

0.0000 0.3949 0.0047

0.0000 0.1123 0.0005

0.0000 0.2516 0.0051

0.0000 0.0635 0.0018

0.0000 0.0011 0.0016

0.0000 0.0292 0.0228

0.0000 0.0270 0.0078

0.0000 0.0107 0.0077

0.0000 0.0113 -0.0040

0.0000 -0.0009 -0.0003

0.0000 0.1404 0.0018

0.0000 0.0403 0.0013

0.0000 0.1136 0.0043

0.0000 0.1436 0.0064

0.0000 0.1477 0.0005

0.0000 0.0886 0.0003

0.0000 0.0886 0.0003 i.

0.0000 -0.0434 0.0002 а

0.0000 -0.0497 0.0002 §

0.0000 -0.0408 0.0001 jj

0.0000 -0.0367 0.0002 ^

0.0000 0.0024 0.0003 ^

0.0000 -0.0612 0.0003 I

0.0000 0.1993 0.0005 •§.

0.0000 -0.0358 0.0001 3.

0.0000 -0.0430 0.0002

0.0000 -0.0158 0.0002 1.

0.0000 -0.0158 0.0002 §

0.0000 -0.0158 0.0002 |

0.0000 -0.0358 0.0001 S

0.0000 -0.0352 0.0000 S

0.0000 0.0057 0.0002 "g

0.0000 0.0767 0.0001 К

Table A5

Sensitivities (unit: ppb) of daily maximum 8-h ozone to 10% emission reductions on August 4. Area County Monitor Obs Model LADC_ANOx OTRV_ANOx

Baltimore

Pittsburgh

Washington, DC

Anne Arundel Co(003-0014) 118.00 124.05 0.3124 3.3628

Baltimore (City)(510-0054) 59.00 70.45 0.6664 1.0667

Baltimore Co(005-1007) 65.00 74.64 0.6854 1.1009

Baltimore Co(005-3001) 82.00 74.64 0.6854 1.1009

Carroll Co(013-0001) 77.00 70.45 0.6664 1.0667

Harford Co(025-1001) 87.00 80.38 0.3541 2.0729

Harford Co(025-9001) 82.00 78.92 0.3766 1.9210

Mean 81.43 81.93 0.5353 1.6703

Std 18.98 18.95 0.1766 0.8628

Kent Co(001-0002) 78.00 75.35 0.0068 0.0142

New Castle Co(003-1007) 78.00 74.77 0.0041 0.0189

New Castle Co(003-1010) 77.00 80.13 0.0053 0.0118

New Castle Co(003-1013) 77.00 73.75 0.0043 0.0177

Sussex Co(005-1002) 81.00 72.03 0.0055 0.0142

Sussex Co(005-1003) 71.00 105.59 0.0021 0.0315

Cecil Co(015-003) 79.00 73.02 0.0048 0.0170

Chester Co(029-0100) 74.00 78.51 0.0051 0.0123

Delaware Co(045-0002) 79.00 79.97 0.0040 0.0200

Montgomery Co(091-0013) 75.00 79.53 0.0050 0.0162

Philadelphia Co(101-0004) 64.00 79.53 0.0050 0.0162

Philadelphia Co(101-0014) 71.00 75.99 0.0048 0.0153

Philadelphia Co(101-0024) 80.00 78.93 0.0041 0.0102

Philadelphia Co(101-0136) 82.00 80.97 0.0050 0.0069

Mean 76.14 79.15 0.4698 1.5895

Std 4.83 8.16 0.1030 0.5701

Allegheny Co(003-0008) 82.00 90.38 0.0084 -0.0237

Allegheny Co(003-0010) 82.00 90.38 0.0084 -0.0237

Allegheny Co(003-0067) 77.00 92.75 0.0087 -0.0001

Allegheny Co(003-1005) 82.00 79.39 0.0088 0.0102

Armstrong Co(005-0001) 80.00 61.48 0.0091 0.0054

Beaver Co(007-0002) 71.00 83.85 0.0107 0.0009

Beaver Co(007-0005) 69.00 82.55 0.0103 -0.0031

Beaver Co(007-0014) 72.00 93.78 0.0088 0.0078

Washington Co(125-0005) 83.00 68.26 0.0051 0.0097

Washington Co( 125-0200) 77.00 67.13 0.0067 0.0035

Washington Co( 125-5001) 70.00 75.46 0.0125 0.0010

Westmoreland Co( 129-0006) 82.00 102.13 0.0060 0.0083

Westmoreland Co(129-0008) 82.00 92.19 0.0053 0.0116

Mean 77.62 83.06 0.8356 0.0610

Std 5.32 12.16 0.2135 1.1648

Entire District(001-0025) 79.00 79.06 0.0065 -0.0201

Entire District(001-0041) 89.00 91.52 0.0077 0.0117

Entire District(001-0043) 87.00 91.52 0.0077 0.0117

Calvert Co (009-0011) 89.00 82.81 0.0053 0.0146

Charles Co (017-0010) 92.00 111.62 0.0052 0.0323

Frederick Co(021-0037) 79.00 69.82 0.0067 0.0082

Montgomery Co(031-3001) 82.00 75.94 0.0070 0.0125

Prince George's Co(033-0030) 82.00 80.50 0.0068 0.0123

Prince George's Co(033-8003) 110.00 115.27 0.0049 0.0315

Arlington Co(013-0020) 88.00 79.06 0.0065 -0.0201

Fairfax Co(059-0005) 69.00 85.37 0.0076 0.0100

Fairfax Co(059-0018) 88.00 118.63 0.0073 0.0332

Fairfax Co(059-0030) 88.00 114.59 0.0072 0.0241

SESA_ANOx LADC_EGU_Nox OTRV_EGU_NOx SESA_EGU_NOx CENR_AVOC OTRV_AVOC SESA_AVOC

0.2620 0.0768 0.1423 0.1423 0.0768 0.0251 0.0167 0.1060 0.0848

0.0002 0.0003 0.0002 0.0002 0.0012 0.0013 0.0002 0.0002 0.0002 0.0004 0.0004 0.0002 0.0004 0.0004 0.0417 0.0358

0.0006 0.0006 0.0011 0.0000 0.0000 0.0023 0.0010 0.0002 0.0008 0.0014 0.0028 0.0000 0.0000 0.0830 0.0891

0.0032 0.0025 0.0025 0.0002 0.0055 0.0006 0.0020 0.0022 0.0025 0.0032 0.0032 0.0037 0.0033

0.1183 0.1293 0.1503 0.1503 0.1293 0.0864 0.0836 0.1211 0.0273

0.0012 0.0010 0.0009 0.0009 0.0014 0.0006 0.0010 0.0010 0.0008 0.0008 0.0008 0.0008 0.0007 0.0008 0.0912 0.0208

0.0041 0.0041 0.0055 0.0013 0.0018 0.0084 0.0026 0.0020 0.0009 0.0037 0.0052 0.0008 0.0008 0.3166 0.2284

0.0024 0.0021 0.0021 0.0011 0.0028 0.0012 0.0017 0.0017 0.0017 0.0024 0.0018 0.0028 0.0025

0.6953 0.1241 0.2422 0.2422 0.1241 0.4464 0.4420 0.3309 0.2081

0.0029 0.0031 0.0011 0.0033 0.0034 0.0035 0.0030 0.0022 0.0032 0.0012 0.0012 0.0032 0.0018 0.0016 0.2489 0.0915

0.0023 0.0023 0.0025 0.0009 0.0006 0.0004 -0.0097 0.0039 0.0023 0.0005 0.0012 0.0024 0.0056 0.1172 0.3579

0.0022 0.0028 0.0028 0.0059 0.0069 0.0012 0.0016 0.0018 0.0057 0.0022 0.0039 0.0044 0.0035

0.1612 0.0503 0.0934 0.0934 0.0503 0.0227 0.0164 0.0697 0.0505

0.0002 0.0003 0.0001 0.0002 0.0008 0.0004 0.0002 0.0002 0.0002 0.0001 0.0001 0.0001 0.0002 0.0002 0.0237 0.0184

0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0002 0.0006 0.0000 0.0000 0.0000 0.0066 0.0181

0.0020 0.0013 0.0013 0.0002 0.0033 0.0004 0.0012 0.0014 0.0016 0.0020 0.0009 0.0016 0.0016

0.0004 0.0151 0.0121 0.0121 0.0151 0.0090 0.0105 0.0106 0.0050

0.0001 0.0001 0.0001 0.0001 0.0000 0.0000 0.0001 0.0001 0.0000 0.0001 0.0001 0.0001 0.0001 0.0001 0.0074 0.0038

0.0015 0.0015 0.0011 0.0010 0.0008 0.0009 0.0027 0.0013 0.0001 0.0001 0.0003 0.0014 0.0006 0.1014 0.0704

0.0002 0.0001 0.0001 0.0002 0.0000 0.0001 0.0001 0.0001 0.0000 0.0002 0.0001 0.0000 0.0001

-0.1861 -0.0103 -0.0080 -0.0080 -0.0103 -0.0427 -0.0430 -0.0440 0.0647

-0.0006 -0.0004 0.0009 -0.0002 -0.0005 -0.0008 -0.0004 0.0004 0.0001 0.0006 0.0006 0.0003 0.0017 0.0020 0.0262 0.0857

0.0069 0.0069 0.0016 0.0004 -0.0001 0.0002 0.0013 0.0006 0.0001 0.0002 0.0000 0.0040 0.0006 0.1744 0.2513

0.0049 0.0011 0.0011 0.0001 -0.0010 0.0000 0.0000 0.0003 -0.0014 0.0049 0.0002 -0.0002 0.0010

-0.0108 -0.0014 -0.0031 -0.0031 -0.0014 -0.0012 -0.0007 -0.0031 0.0035

0.0000 0.0000 0.0000 0.0000 -0.0001 -0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 -0.0016 0.0031

0.0001 0.0001 0.0001 0.0000 0.0000 0.0002 0.0003 0.0000 0.0000 0.0000 0.0001 0.0000 0.0000 0.0065 0.0098

0.0000 0.0000 0.0000 0.0000 -0.0002 0.0000 0.0000 0.0000 -0.0001 0.0000 0.0000 -0.0001 0.0000

References

O O fN

fN m ^ CD ^T <.0

in in ffl CI ffl M

m m m m ^

o o o o rn rn

o o o o m ^

CD CD CD C) ci C^

in in M tN H

fN fN ^ fN 00

CD CD CD CD O

O O O O fN

0 0 0 0 0

fN fN UD O m

r> r> h- rn oo o

O O O O oo CD

CD CD CD CD UD O

OD OD O ^ '—1

m m m oo cd oo

^r ^r ui UD on UD

^ ^ to M ^

o o o o

Allen, D.J., Pickering, K.E., Pinder, R.W., Henderson, B.H., Appel, K.W., Prados, A., 2012. Impact of lightning-NO on eastern United States photochemistry during the summer of 2006 as determined using the CMAQ model. Atmos. Chem. Phys. 12, 1737-1758.

Baker, A.K., Beyersdorf, A.J., Doezema, L.A., Katzenstein, A., Meinardi, S., Simpson, I.J., Blake, D.R., Rowland, F.S., 2008. Measurements of nonmethane hydrocarbons in 28 United States cities. Atmos. Environ. 42,170-182.

Baker, D., Downs, T., Ku, M., Hao, W., Sistla, G., Kiss, M., Johnson, M., Brown, D., 2010. WRF Protocol: Sensitivity Testing of WRF Physics Parameterizations for Meteorological Modeling in Support of Regional SIP Air Quality Modeling in the Ozone Transport Region.

Bell, M.L., Dominici, F., Samet, J.M., 2005. A meta-analysis of time-series studies of ozone and mortality with comparison to the national morbidity, mortality, and air pollution study. Epidemiology 16, 436-445.

Bergin, M.S., Shih, J.S., Krupnick, A.J., Boylan, J.W., Wilkinson, J.G., Odman, M.T., Russell, A.G., 2007. Regional air quality: local and interstate impacts of NOx and SO2 emissions on ozone and fine particulate matter in the Eastern United States. Environ. Sci. Technol. 41, 4677-4689.

Byun, D.W., Schere, K.L., 2006. Review of the governing equations, computational algorithms, and other components of the models-3 community multiscale air quality (CMAQ) modeling system. Appl. Mech. Rev., 51-77.

Cohan, D.S., Hakami, A., Hu, Y.T., Russell, A.G., 2005. Nonlinear response of ozone to emissions: source apportionment and sensitivity analysis. Environ. Sci. Technol. 39, 6739-6748.

Devlin, R.B., Duncan, K.E., Jardim, M., Schmitt, M.T., Rappold, A.G., Diaz-Sanchez, D., 2012. Controlled exposure of healthy young volunteers to ozone causes cardiovascular effects. Circulation 126,104-111.

Dunker, A.M., Yarwood, G., Ortmann, J.P., Wilson, G.M., 2002a. Comparison of source apportionment and source sensitivity of ozone in a three-dimensional air quality model. Environ. Sci. Technol. 36, 2953-2964.

Dunker, A.M., Yarwood, G., Ortmann, J.P., Wilson, G.M., 2002b. The decoupled direct method for sensitivity analysis in a three-dimensional air quality model - implementation, accuracy, and efficiency. Environ. Sci. Technol. 36, 2965-2976.

EPA, 2010. 2008 National Emissions Inventory Preliminary Data. Available from: http://www.epa.gov/ttn/chief/net/2008inventory.html (accessed 26.03.13).

EPA, 2013a. Continuous Emissions Monitoring Fact Sheet.

EPA, 2013b. National Mobile Inventory Model.

Finlayson-Pitts, B.J., Pitts, J.N., 1997. Tropospheric air pollution: ozone, airborne toxics, polycyclic aromatic hydrocarbons, and particles. Science 276, 10451052.

Gilliam, R., Pleim, J., Otte, T., 2009. Multiscale meteorological modeling for air quality applications. In: EPA Atmospheric and Modeling Analysis Division Peer Review Conference, Research Triangle Park, NC.

Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P.I., Geron, C., 2006. Estimates of global terrestrial isoprene emissions using MEGAN (model of emissions of gases and aerosols from nature). Atmos. Chem. Phys. 6, 31813210.

Hains, J.C., Taubman, B.F., Thompson, A.M., Stehr, J.W., Marufu, L.T., Doddridge, B.G., Dickerson, R.R., 2008. Origins of chemical pollution derived from Mid-Atlantic aircraft profiles using a clustering technique. Atmos. Environ. 42, 1727-1741.

Hakami, A., Odman, M.T., Russell, A.G., 2003. High-order, direct sensitivity analysis of multidimensional air quality models. Environ. Sci. Technol. 37, 2442-2452.

Hakami, A., Seinfeld, J.H., Chai, T.F., Tang, Y.H., Carmichael, G.R., Sandu, A., 2006. Adjoint sensitivity analysis of ozone nonattainment over the continental United States. Environ. Sci. Technol. 40, 3855-3864.

Hollaway, M.J., Arnold, S.R., Challinor, A.J., Emberson, L.D., 2012. Intercontinental trans-boundary contributions to ozone-induced crop yield losses in the Northern Hemisphere. Biogeosciences 9, 271 -292.

Houyoux, M.R., Vukovich, J.M., Coats, C.J., Wheeler, N.J.M., Kasibhatla, P.S., 2000. Emission inventory development and processing for the seasonal model for regional air quality (SMRAQ) project. J. Geophys. Res.-Atmos. 105, 9079-9090.

MACTEC, 2011. Technical Support Document for the Development of the 2007 Emission Inventory for Regional Air Quality Modeling in the Northeast/Mid-atlantic Region.

Otte, T.L., Pleim, J.E., 2010. The meteorology-chemistry interface processor (MCIP) for the CMAQ modeling system: updates through MClPv3.4.1. Geosci. Model Dev. 3, 243-256.

Pleim, J.E., 2007. A combined local and nonlocal closure model for the atmospheric boundary layer. Part l: model description and testing. J. Appl. Meteorol. Cli-matol. 46,1383-1395.

Sistla, G., Ku, M., Hao, W., 2011. Assessment of 2007 WRF Meteorological Modeling in Support of Regional Air Quality Modeling in the Ozone Transport Region for 2007. Technical Support Document, August 2010.

Skamarock, W.C., Klemp, J.B., Dudhia, J., Gill, D.O., Barker, D.M., Duda, M.G., Huang, X.Y., Wang, W., Powers, J.G., 2008. A Description of the Advanced Research WRF Version 3.

Taubman, B.F., Marufu, L.T., Piety, C.A., Doddridge, B.G., Stehr, J.W., Dickerson, R.R., 2004. Airborne characterization of the chemical, optical, and meteorological

properties, and origins of a combined ozone-haze episode over the eastern United States. J. Atmos. Sci. 61,1781-1793.

USEPA, 1991. In: Agency, U.S.E.P. (Ed.), Guideline for Regulatory Application of the Urban Airshed Model. Research Triangle Park, NC.

USEPA, 2005. CMAQ Model Performance Evaluation for 2001: Updated March 2005.

USEPA, 2007. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5, and Regional Haze, Research Triangle Park, North Carolina.

Vickers, A., McDill, J., Davis, P., 2011. MARAMA 2007/2017/2020 Modeling Emissions lnventory Version 2 Preliminary Trends Analysis.

Walsh, K.J., Milligan, M., Woodman, M., Sherwell, J., 2008. Data mining to characterize ozone behavior in Baltimore and Washington, DC. Atmos. Environ. 42, 4280-4292.

Yamartino, R.J., 1993. Nonnegative, conserved scalar transport using grid-cell-centered, spectrally constrained blackman cubics for applications on a variable-thickness mesh. Mon. Weather Rev. 121, 753-763.

Yang, Y.J., Wilkinson, J.G., Russell, A.G., 1997. Fast, direct sensitivity analysis of multidimensional photochemical models. Environ. Sci. Technol. 31, 2859-2868.

Yarwood, G., Rao, S., Yocke, M., Whitten, G.Z., 2005. Updates to the Carbon Bond Chemical Mechanism: CB05. Final Report to the U.S. EPA, RT-0400675.

References for the WRF physical configurations

Dudhia, J., 1989. Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci. 46, 3077-3107.

Hong, S.-Y., Lim, J.-O., 2006. The WRF single-moment six-class microphysics scheme (WSM6). J. Kor. Meteorol. Soc. 42,129-151.

Kain, J.S., Fritsch, J.M., 1990. A one-dimensional entraining detraining plume model and its application in convective parameterization. J. Atmos. Sci. 47,2784-2802.

Xiu, A.J., Pleim, J.E., 2001. Development of a land surface model. Part I: application in a mesoscale meteorological model. J. Appl. Meteorol. 40,192-209.

Zhang, D., Anthes, R.A., 1982. A high-resolution model of the planetary boundary-layer - sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteorol. 21, 1594-1609.

Zhang, D.L., Zheng, W.Z., 2004. Diurnal cycles of surface winds and temperatures as simulated by five boundary layer parameterizations. J. Appl. Meteorol. 43,157169.