Atmospheric and Climate Sciences
Vol.03 No.01(2013), Article ID:27551,12 pages
10.4236/acs.2013.31004

Understanding of the Fate of Atmospheric Pollutants Using a Process Analysis Tool in a 3-D Regional Air Quality Model at a Fine Grid Scale

Yang Zhang1*, Shiang-Yuh Wu2

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

2Department of Air Quality and Environmental Management, Las Vegas , USA

Email: *yang_zhang@ncsu.edu

ABSTRACT

Received October 16, 2012; revised November 18, 2012; accepted November 26, 2012

The process analysis is performed for August and December, 2002 using the process analysis tool embedded in the Community Multiscale Air Quality (CMAQ) modeling system at a fine horizontal grid resolution of 4-km over an area in the southeastern US that is centered at North Carolina. The objectives are to qunatify the contributions of major at- mospheric processes to the formation of major air pollutants and provide the insights into photochemistry that governs the fate of these pollutants at a fine grid scale. The results show that emissions provide a dominant source for gases in- cluding ammonia (NH3), nitric oxide (NO), nitrogen dioxide (NO2), and sulfur dioxide (SO2) and Particulate Matter (PM) species including fine PM (PM2.5) and its composition such as sulfate, elemental carbon, primary organic aerosol, and other inorganic fine PM in both months. While transport acts as a major sink for NH3, NO, and SO2 at most sites and PM2.5 and most of PM2.5 composition at urban sites, it provides a major source for nitric acid (HNO3) and ozone (O3) at most sites in both months, and secondary PM species in August and most PM species in December at rural and remote sites. Gas-phase chemistry serves as a source for NO2 and HNO3 but a sink for O3 at urban and suburban sites and for NO and SO2 at all sites. PM processes contribute to the formation of PM2.5 and nitrate () at the urban and suburban sites and secondary organic aerosol (SOA) at most sites in December and ammonium () in both months. They reduce formation at most sites in August and at rural and remote sites in December and the formation of PM2.5 and SOA at most sites in August. Dry deposition is an important sink for all these species in both months. The total odd oxygen (Ox) production and the total hydroxyl radical (OH) reacted are much higher at urban and suburban sites than at rural sites. Significant amounts of OH are consumed by biogenic volatile organic compounds (BVOCs) in the rural and remote areas and a combination of anthropogenic VOCs (AVOCs) and BVOCs in urban and subareas areas in August and mainly by AVOCs in December. The amount of NO2 produced by the reactions of hydroperoxy radical (HO2) is similar to that of organic peroxy radical (RO2) at all sites in August but higher than that by the reactions of RO 2 in De- cember. The production rate of HNO3 due to the reaction of OH with NO2 dominates in both months. The ratio of the production rates of hydrogen peroxide (H2O2) and HNO3 (PH2O2/PHNO3) is a more robust photochemical indicator than the ratios of their mixing ratios (H2O2/HNO3) and the afternoon mixing ratios of NOy in both months, and it is highly sensitive to the horizontal grid resolution in August. The use of PH2O2/PHNO3 simulated at 4-km indicates a VOC-limited O3 chemistry in urban and suburban areas in August that was not captured in previous model simulations at a coarser grid resolution.

Keywords: Air Pollutants; Process Analysis; Photochemical Indicator; MM5; CMAQ

1. Introduction

*Corresponding author.

Process Analysis (PA) is a useful tool embedded in a 3-D air quality model that calculates the Integrated Process Rates (IPR) for major atmospheric processes such as emissions, chemical reactions, horizontal and vertical transport, and removal processes and the Integrated Re- action Rates (IRR) for all gas-phase chemical reactions in all model grid cells. The results from IPR provide the relative contributions of individual physical and chemical processes to the formation of gas and Particulate Matters (PM) species. These processes include emissions, verti- cal and horizontal transport, gas-phase chemistry, PM pro- cesses, aqueous-phase processes (or cloud processes), and dry deposition. The results from IRR provide individual

gas-phase reaction rates that can be used to identify key chemical pathways for ozone (O3) and its precursors, the chemical regimes of O3, as well as gaseous precursors of secondary PM with aerodynamic diameter less than and equal to 2.5 m (PM2.5) [1-3]. For example, the net pro- duction and loss of total odd oxygen (Ox) represent the total oxidation capacity that affects the formation effi- ciency of O3 and secondary PM. The list of typical IRR products can be found in Zhang et al. [3]. PA has been conducted in several studies to quantify the contributions of atmospheric processes and chemical reactions to the formation of O3 and PM2.5 [e.g., 3-9]. All those studies focused only criteria pollutants such as O3 and PM2.5 and used a horizontal grid resolution of 36-km or coarser. Very few studies include PA for agriculturally-emitted pollutants such as ammonia (NH3) and ammonium () and are performed at a horizontal grid spacing of 4 - 12 km .

In this study, 3-D model simulations and PA are con- ducted at a horizontal grid spacing of 4-km to simulate O3, PM, and their precursors. The objective of this study is to identify the governing atmospheric processes of major air pollutants including both creteria and non- creteria air pollutants and associated seasonalities at a fine grid resolution. An area in the southeastern US that centers over North Carolina (NC) is selected for this study. This area feasures with very high emissions of NH3 from agricultural livestock, which account for about 91% (i.e., 482.9 tons∙day−1) in August and 81% (i.e., 253.4 tons∙day−1) in December of total NH3 emissions [10]. PA over this area allows an understanding of the fate of non-creteria pollutants such as NH3, , nitric acid (HNO3), and reduced nitrogen (NHx = NH3 +), in addition to that of creteria air pollutants such as nitric oxide (NO), nitrogen dioxide (NO2), sulfur dioxide (SO2), O3, and PM2.5.

2. Modeling Domain and Simulation Setup

The modeling system consists of the Pennsylvania State University (PSU)/National Center for Atmospheric Re- search (NCAR) Mesoscale Modeling System Generation 5 (MM5) version 3.7 [11], the Sparse Matrix Operator Kernel Emissions (SMOKE) Modeling System version 2.1 [12], and the Community Multiscale Air Quality (CMAQ) modeling system version 4.4 [13]. The PA tool embedded in CMAQ is used to quantify the contributions of major atmospheric processes of major air pollutants. The 3-D model simulations are conducted for August and December of 2002 at a 4-km horizontal grid spacing over a domain that covers nearly the entire state of NC, and a portion of several adjacent states including South Caro- lina (SC), Georgia (GA), Tennessee (TN), West Virginia (WV), and Virginia (VA), as shown in Figure 1. This area consists of three well-developed physiographic divi- sions from east to west: the Coastal Plain, the Pidemont, and the Mountains. The complex topography and weather patterns as well as a combination of industrial, agricul- tural, traffic, and biogenic emissions make this area one of the most complex and representative airsheds in the US.

The model input files for initial and boundary condi- tions (ICs and BCs) and meteorology at a 4-km horizon- tal grid spacing are developed based on the MM5/CMAQ model simulations at a 12-km horizontal grid spacing ob- tained from the Visibility Improvement State and Tribal Association of the Southeast’s (VISTAS) 2002 modeling program (http://www.vista-sesarm.org.asp). For consis- tency, the model configurations and options for physics and chemistry for the MM5/CMAQ simulations at 4-km in this work are set to be the same as those used in the 2002 base year VISTAS Phase II modeling study at 12-km. The vertical resolution includes 19 layers from surface to the tropopause (~ 15 km ) with ~ 38 m for the first layer height. The emission inventories for gaseous and PM species are based on the VISTAS 2002 emissions. As described in the work of Wu et al. [10], the model evaluation showed that MM5/CMAQ gave an overall good performance for meteorological variables and O3 mixing ratios and a reasonably good performance for PM2.5. A more detailed description of the model con- figurations, ICs and BCs, the databases used for the op- erational evaluation for meteorological and chemical predictions, and the model performance evaluation for both MM5 and CMAQ can be found in Wu et al. [10].

A detailed PA analysis is performed at 17 sites from three surface networks: seven sites from the Speciation Trends Network (STN): Kinston, Asheville, Hickory, Fayetteville, Winston-Salem, Charlotte, and Raleigh; four sites from the Interagency Monitoring of Protected Vis- ual Environments (IMPROVE): GRSM1, LIGO1, SHRO1, and SWAN1; and six sites from the Clean Air Status

Figure 1. Simulation domain and locations of observational sites selected for process analysis, including seven STN sites (red): Kinston , Asheville , Hickory , Fayetteville , Winston-Sa- lem, Charlotte, and Raleigh; four IMPROVE sites (green): GRSM1, LIGO1, SHRO1, and SWAN1; six CASTNET sites (blue): BFT142, CND125, COW137, PNF126, SPD111, and VPI120.

Trends Network (CASTNET): BFT142, CND125, COW- 137, PNF126, SPD111, and VPI120. Their locations are shown in Figure 1. Among the STN sites, Kinston and Fayetteville are located in the Coastal Plain region, Ash- ville is located in the Mountains, and Hickory , Winston- Salem, Charlotte , and Raleigh are located in the Pide- mont region. Among the IMPROVE sites, GRSM, LI- GO1, and SHRO1 are located in the Mountains and SWAN1 is in the Coastal Plain region. Among the CASTNET sites, BFT142 is a Coastal Plain site, CND125 is a Pidemont site, and COW137, PNF126, SPD111, and VPI120 are all located in the Mountains.

3. Process Analysis

3.1. Integrated Process Rates

Figure 2 shows the monthly-mean contributions of indi- vidual processes to the mixing ratios of gaseous species in the surface layer at the 17 locations in August and December 2002. Emissions provide a dominant source of NH3 at all STN sites except for Asheville in August and at all STN sites in December. Among the 17 sites, the largest emissions occur at Kinston and Fayetteville where NH3 emissions from agricultural livestock are high. Trans- port reduces the mixing ratios of NH3 at most sites (ex- cept for CND 125 in August), particularly at Kinston in August and at Kinstron, Fayetteville, and Charlotte in December. Dry deposition also acts as a sink for NH3 at all sites, particularly at Kinstron, Fayetteville, and CND 125 in August and at Kinston in December. PM processes such as gas-to-particle mass transfer convert NH3 to at most sites in both months. Emissions of NO are high in both months at all STN sites expect for Kinston, providing the main source of nitrogen oxides (NOx = NO + NO2) at these sites. Major loss processes of NO in both months include gas-phase chemistry (i.e., its titration reactions with O3) and horizontal and vertical transport. The same titration reaction of NO with O3 produces NO2, which is the most important source of NO2 at the STN sites in both months. Emissions of NO2 provide addi- tional sources at several STN sites including Hickory, Fayetteville, Winston-Salem, and Charlotte. The major loss processes of NO 2 in both months include transport at most sites, particularly at the STN sites, and dry deposi- tion at all sites. Comparing to the STN sites, the process contributions to NH3 and NOx at the IMPROVE and CASTNET sites are relatively small, due to a lack of pollutant sources in the Costal Plain and mountain re- gions. While transport contributes to the accumulation of HNO3 at all 17 sites in August and at all sites except for Kinston, Hickory, and CND 125 in December, dry depo- sition is a major sink of HNO3 at these sites. Cloud and PM processes also contribute to its sink in December. In August, emissions at all STN sites except for Kinston provide a major source of SO2, and transport is the main process to accumulate SO2 at the IMPROVE and CAST- NET sites except for SWAN1 and COW137. While both dry deposition and transport are important sinks at the STN sites, dry deposition dominates the loss of SO2 at the IMPROVE and CASTNET sites. In December, emis- sions provide a main source of SO2 at all STN sites ex- cept for Kinston, Asheville, and Raleigh where transport is either a dominant source or equally important to its emissions. Transport also helps accumulation of SO2 at all IMPROVE and CASTNET sites. Dry deposition, gas- phase chemistry, and cloud processes including aqueous- phase chemistry and wet scavenging contribute to the loss of SO2 at all these sites. In August, O3 comes pri- marily from transport and it is lost due mainly to gas- phase chemistry at Hickory , Fayetteville , and Winston- Salem, both gas-phase chemistry and dry deposition at Charlotte , and dry deposition at all remaining sites. In December, transport accumulates O3 at all STN sites ex- cept for Kinston, two IMRPOVE sites (LIGO1 and SHRO1) and one CASTNET site (PNF126). Gas-phase chemistry provides a major sink at all sites, in particular at all STN sites except for Kinston.

Figure 3 shows the monthly-mean contributions of in- dividual processes to the mass concentrations of PM2.5 and its major composition including sulafte (), , nitrate (), elemental carbon (EC), primary organic aerosol (POA), secondary organic aerosol (SOA), and other inorganic fine PM (OIN) in the surface layer at the 17 sites. For PM 2.5 in August, emissions provide a do- minant source at all STN sites except for Kinston and transport helps its accumulation at three IMPROVE sites (i.e., GRSM, LIGO1, and SHRO1) and two CASTNET sites (i.e., PNF126 and VPI120). Transport is a major sink process at most STN sites and dry deposition con- tributes the most to the loss of PM2.5 at LIGO1, SHRO1, and PNF126. In December, both emissions and PM pro- cesses are important sources of PM2.5 at most STN sites, and transport helps its accumulation at GRSM, LIGO1, SHRO1, and PNF126. Transport plays a similar role to that in August, depleting PM2.5 at the STN sites. Dry deposition is a major sink of PM2.5 at several sites in- cluding Ashville, LIGO1, SHRO1, and PNF126. Cloud processes also contribute to the loss of PM2.5 at all sites in both months. For, PM processes provide a do- minant source at all sites in August and most sites in De- cember. A more significant conversion of NH3 to occurs at Kinston in December than in August (40% - 60% vs <20% - 30%) due to the formation of NH4NO3 under the favorable weather condition as shown in Wu et al. [10], leading to much higher contributions of PM pro- cesses to at this site in December than in August. Transport causes the loss of at the STN sites and several CASTNET sites in both months, but contributes

Figure 2. The monthly-mean process contributions to the surface mixing ratios of NH3, NO, NO2, HNO3, SO2, and O 3 in ppb∙hr1 during August and December 2002.

Figure 2. The monthly-mean process contributions to the surface mixing ratios of NH3, NO, NO2, HNO3, SO2, and O 3 in ppb∙hr1 during August and December 2002.

Figure 3. The monthly-mean process contributions to the surface concentrations of PM2.5, , , , EC, POA, SOA, and OIN in μg∙m−3∙hr1 during August and December 2002.

Figure 3. The monthly-mean process contributions to the surface concentrations of PM2.5, , , , EC, POA, SOA, and OIN in μg∙m−3∙hr1 during August and December 2002.

to its accumulation at several other sites including GRSM, CND125, and PNF 126 in August and GRSM, LIGO1, SHRO1, SWAN1, and PNF 126 in December. For, the emissions provide a dominant source at all STN sites except for Ashville where transport pro- vides the source in both months and Winston-Salem where PM processes such as homogeneous nucleation generate PM in August. Transport and dry deposition are major sink processes for at most of these sites in both months, and cloud processes also contribute to its loss at all sites, particularly at Fayetteville in August, and at Hickory in December. At most IMPROVE and CASTNET sites, transport and dry deposition provide a dominant source and sink, respectively, for in both months. Mass balance adjustment contributes to some losses of at GRSM1, LIGO1, and VPI 120 in August and at LIGO1, SHRO1, and PNF 126 in De- cember; it also contributes to some gains of at PNF126 and COW 137 in August and at VPI120. All of these sites are located in the Appalachian Mountains re- gion. The relatively larger contributions of mass bal- ance adjustment indicate the difficulty of MM 5 in simu- lating advection and vertical mixing processes over com- plex terrains, which propagates into chemical predic- tions of CMAQ. at all sites comes primarily from transport in August, and it is removed mainly through PM processes such as evaporation back to the gas-phase, aqueous processes such as aqueous-phase chemistry and wet scanvenging, and dry deposition. For comparison, in December, at all STN sites is produced by PM processes such as the condensation of HNO3 but by transport at all other sites. is reduced by transport at all STN sites and additionally by dry deposition at

Kinston and Asheville , but it is lost via PM processes such as the evaporation and coagulation at all IMPROVE and CASTNET sites (except for SPD111) and addition- ally by dry deposition at three mountain sites (i.e., LIGO1, SHRO1, and PNF126). The gain and loss of in December show a strong correlation with those of at all STN sites, indicating the formation of NH4NO3 under the favorable weather and chemical con- ditions at these sites. For EC, POA, and OIN, the main production and loss are emissions and transport, respec- tively, at most STN sites. OIN may also be produced by emissions at other sites such as SWAN1 and BFT 142 in both months or LIGO1, SHRO1, and PNF 126 in De- cember. Different from POA, transport is a dominant source for SOA at most sites in August and December. Gas/particle mass transport is a major contributor to SOA formation at Hickory in August and SWAN1 and BFT 142 in December.

Figure 4 shows the monthly-mean contributions of in- dividual processes to the mass concentrations of reduced nitrogen (NHx = NH3 +) and total nitrate (TNO3 = HNO3 +) in the surface layer. The fate of NHx is dominated by that of NH3, with a major gain from emis- sions and a major loss by transport and deposition at most STN sites in both months. The fate of TNO3 is dominated by that of HNO3, with a major gain from transport and a major loss by deposition at all sites in both months.

3.2. Integrated Reaction Rates (IRR)

Figure 5 shows the spatial distributions of the monthly- mean production and loss of Ox, the total hydroxyl radical

Figure 4. The monthly-mean process contributions to the surface concentrations of NHx in ppb∙hr1 and TNO 3 in μg∙m−3∙hr1 during August and December 2002.

Figure 4. The monthly-mean process contributions to the surface concentrations of NHx in ppb∙hr1 and TNO 3 in μg∙m−3∙hr1 during August and December 2002.

Figure 5. The spatial distribution of monthly-mean production and loss rates of Ox, total OH reacted, and total OH reacted with VOCs, AVOCs, and BVOCs in the surface layer in August and December 2002.

Figure 5. The spatial distribution of monthly-mean production and loss rates of Ox, total OH reacted, and total OH reacted with VOCs, AVOCs, and BVOCs in the surface layer in August and December 2002.

(OH) reacted, the total OH reacted with volatile organic compounds (VOCs), anthropogenic VOCs (AVOCs), and biogenic VOCs (BVOCs) at the surface layer. A much higher total Ox production indicates a much higher

oxidation capacity in August than in December, leading to higher monthly-mean O3 mixing ratios as shown in Wu et al. [10]. The highest Ox production concentrates in urban and suburban areas along the main highways such as highways I-40, I-95, I-85, I-77, and I -26 in the Pied- mont and Mountains regions in August and over the southern portion of the domain in December. The loss of Ox occurs in the Piedmont and Mountains regions where the emissions of NOx are high in August but spreads out the whole domain due to prevailing westerlies with rela- tively high wind speeds as reported in Wu et al. [10] that transport and mix the precursors of Ox more uniformly in December. Similar to the spatial distributions of total Ox production, the amount of OH reacted with major gases such as VOCs, NOx, SO2, and carbon monoxide (CO) is much higher over urban and suburban areas along the main highways in the Piedmont and Mountains regions in August and over the southern portion of the domain, particularly in the southwestern NC and northwestern SC in December. Significant amounts of OH are consumed by VOCs in both months, with BVOCs in the rural and re- mote areas and a combination of AVOCs and BVOCs in urban and subareas areas.

Figure 6 shows the production and loss rates of Ox, OH reacted with AVOCs and BVOCs (referred to as OH- AVOCs and OH-BVOCs hereafter), the production of NO2 from the reactions involving hydroperoxy radicals (HO2) and organic peroxy radicals (RO2), and the pro- ductions of HNO3 due to the reaction of OH with NO2 and that of VOCs with nitrate radical (NO3) at the 17 sites in both months. At all locations, Ox production rates are higher than its loss rates by factors of 2.7 - 14.1 in August and 4.9 - 10.7 in December. The production rate of Ox is much higher at most STN sites (except for Kinston ,

Figure 6. The monthly-mean production and loss rates of Ox, the rate of OH reacted with AVOCs and BVOCs, the NO2 pro- duction rates by the reaction of HO2 and RO2, and the HNO3 production rates by the reactions of OH + NO2 and NO3 + VOCs at seven STN sites: Kinston, Asheville, Hickory, Fayetteville, Winston-Salem, Charlotte, and Raleigh; four IMPROVE sites: GRSM1, LIGO1, SHRO1, and SWAN1; and six CASTNET sites: BFT142, CND125, COW137, PNF126, SPD111, and VPI 120 in August and December 2002.

Figure 6. The monthly-mean production and loss rates of Ox, the rate of OH reacted with AVOCs and BVOCs, the NO2 pro- duction rates by the reaction of HO2 and RO2, and the HNO3 production rates by the reactions of OH + NO2 and NO3 + VOCs at seven STN sites: Kinston, Asheville, Hickory, Fayetteville, Winston-Salem, Charlotte, and Raleigh; four IMPROVE sites: GRSM1, LIGO1, SHRO1, and SWAN1; and six CASTNET sites: BFT142, CND125, COW137, PNF126, SPD111, and VPI 120 in August and December 2002.

which is an agricultural site with very high NH3 emis- sions located in Coastal Plain region) in August than other sites. In December, the production rates of Ox at all sites are overall similar, with higher values at several mountain sites such as COW137, Asheville, and Hickory. The rates of OH-AVOCs and OH-BVOCs are much higher at all sites in August than in December. In August, the rates of OH-BVOCs are higher than those of OH- AVOCs at all rural sites including Kinston, Asheville, GRSM1, LIGO1, SHRO1, SWAN1, BFT142, CND125, COW137, PNF126, SPD111, and VPI120, because of higher BVOCs emissions at these sites. In December, the rates of OH-AVOCs are higher than those of OH- BVOCs at all sites. O3 is produced through the photolysis of NO2 followed by the reaction of atomic oxygen (O) with molecular oxygen (O2). Most NO2 come from the conversion of NO by HO2 and RO2 radicals. As shown in Figure 6, the amount of NO2 produced by the reactions involving HO2 is similar to that involving RO2 at all sites in August, with slightly higher production rates from the NO + RO2 reaction at the rural sites than at the urban sites. It is, however, higher than that by the reactions involving RO 2 in December. This indicates that VOCs contribute to O3 formation similarly to NOx in August, but less than that of NOx due to a VOC-limited O3 chem- istry in December. The production rate of HNO3 due to the reaction of OH with NO2 dominates over that due to the nighttime reactions of VOCs with NO3 radicals in both months, with much higher reaction rates of OH + NO2 (by up to a factor of 42.6) at urban sites than other sites in August and more uniform reaction rates of OH + NO2 (within a factor of 3) at all sites in December.

The ratio of the production rates of H2O2 and HNO3 (PH2O2/PHNO3) is a useful indicator for O3 photochem- istry that is calculated in IRRs. The threshold value of PH2O2/PHNO3 is 0.2, values below which indicate a VOC-limited O3 chemistry and at or above which indi- cate a NOx-limited chemistry [14,15]. The ratio of the mixing ratios of H2O2 and HNO3 (H2O2/HNO3) and NOy mixing ratios in the afternoon have also been frequently used as photochemical indicators, with a range of thresh- old values suggested by several studies accounting for differences in meteorological and chemical conditions for measurements or model configurations such as hori- zontal grid resolutions and airsheds used in modeling studies. For example, the threshold value proposed for H2O2/HNO3 was 0.2 by Sillman et al. [16], Tonnesen and Dennis [17], and Hammer et al. [18], 0.4 by Sillman [14], 0.8 - 1.2 by Lu and Chang [19], and 2.4 by Zhang et al. [3]. The values of H2O2/HNO3 below these threshold values indicate a VOC-limited O3 chemistry, otherwise a NOx-limited O3 chemistry. The threshold value proposed for NOy was 3 - 5 ppb by Lu and Chang [19], 5 ppb by Zhang et al. [3], 10 - 25 ppb by Milford et al. [20] and 20 ppb by Sillman [14]. The values of NOy larger than these threshold values indicate a VOC-limited O3 chemistry, otherwise a NOx-limited O3 chemistry. Among these three photochemical indicators, PH2O2/PHNO3 has been the most robust one in both summer and winter months [3]. It is therefore selected as a benchmark to determine the robustness of H2O2/HNO3 and NOy as a photochemi- cal indicator for O3 chemistry in this work.

Figure 7 shows simulated spatial distributions of three photochemical chemical indicators: PH2O2/PHNO3, H2O2/ HNO3, and NOy mixing ratios in the afternoon (noon- time-6 pm) in both months. In August, the values of PH2O2/PHNO3 over most areas are above 0.2, indicating a NOx-limited O3 chemistry. Those over urban and sub- urban areas are below 0.2, indicating a VOC-limited O3 chemistry. Zhang et al. [3] performed a 1-year process analysis in 2001 using the PA tool in CMAQ over the continental US at a horizontal grid resolution of 36-km and showed values of PH2O2/PHNO3 of 0.4 - 2.4 over urban and suburban areas and higher values over the re- maining areas in the simulation domain used in this study. Despite a different year (i.e., 2002) and emissions, the use of a much higher horizontal grid resolution of 4-km in this work shows a VOC-limited chemistry in urban and suburban areas that is not shown in the simulation at 36-km in Zhang et al. [3], demonstrating the benefit of the fine-scale modeling. The values PH2O2/PHNO3 are below 0.2 in December in nearly the whole domain, in- dicating a VOC-limited O3 chemistry, which is consistent with the O3 chemical regime in December 2001 obtained by Zhang et al. [3]. The comparison of this work and Zhang et al. [3] indicates that the predictions of PH2O2/ PHNO3 are highly sensitive to the horizontal grid resolu- tion in summer but insensitive to it in winter. For H2O2/HNO3, all values are above 0.2 and nearly all val- ues are above 2.4 in August, indicating a NOx-limited chemistry that is consistent with that based on PH2O2/ PHNO3. In December, using a threshold value of 2.4 will indicate a VOC-limited O3 chemistry in most of the do- main except for an area in the eastern NC in the Coastal Plain region, which is consistent with that based on PH2O2/PHNO3. For NOy, in August, the threshold value of 10 ppb gives similar VOC-limited O3 chemistry over urban and suburban areas and NOx-limited O3 chemistry over remaining areas as compared to that indicated by PH2O2/PHNO3. In December, the lower the threshold value is, the more consistency can be obtained for areas with the VOC-limited O3 chemistry indicated by PH2O2/ PHNO3. The use of 5 ppb as a threshold value indicates a VOC-limited chemistry over nearly the whole domain that is more consistent with the VOC-limited chemistry regime based on PH2O2/PHNO3 than the use of 10 ppb. These results indicate that H2O2/HNO3 and NOy are more robust indicators for O3 chemistry in summer than in

Figure 7. Simulated spatial distributions of monthly-mean PH2O2/PHNO3, H2O2/HNO3, and afternoon (noon-6 pm) NOy mixing ratios in August and December 2002.

Figure 7. Simulated spatial distributions of monthly-mean PH2O2/PHNO3, H2O2/HNO3, and afternoon (noon-6 pm) NOy mixing ratios in August and December 2002.

winter, and a greater adjustment (e.g., adjusting the thresh- old value of H2O2/HNO3 from 0.2 to 11 (instead of sug- gested 2.4 by Zhang et al. [3]) and that of NOy from 20 ppb to 3 ppb (instead of suggested 5 ppb by Zhang et al. [3]) are needed make them more robust in indicating O3 chemistry regimes in winter.

Figure 8 shows the temporal variations of observed and simulated afternoon NOy mixing ratios and simulated PH2O2/PHNO3 at four sites in NC in August, 2002. Two threshold values (10 and 20 ppb) and one threshold value (0.2 ppb) are also plotted for NOy and PH2O2/PHNO3, respectively, to help determine the O3 chemistry regimes. The simulated values of PH2O2/PHNO3 are above 0.2 at Kinston, indicating a NOx-limited chemistry at this site. The simulated values of PH2O2/PHNO3 are mostly below 0.2 at Winston-Salem , Raleigh , and Charlotte , indicating a VOC-limited O3 chemistry. The observed and simu- lated afternoon NOy mixing ratios at Kinston agree well, and they are below 10 ppb, indicating a NOx-limited chemistry at this site, consistent with the O3 chemistry regime results using PH2O2/PHNO3. About 50% of ob- served NOy values are below 10 ppb and 99% of them are below 20 ppb. While the use of 20 ppb indicates a NOx-limited chemistry at these sites, the use of a NOy

value of 10 ppb or lower gives VOC-limited O3 chemis- try regime that is more consistent with the O3 chemistry regime results obtained using PH2O2/PHNO3 during some afternoon hours. Compared with the observed NOy, the simulated NOy values are higher during most hours at Winston-Salem, Raleigh, and Charlotte. About 25% of simulated NOy values are below 10 ppb and 60% of them are below 20 ppb. This indicates that simulated NOy may not be as robust as PH2O2/PHNO3 to indicate the O3 chemistry regime, because of inaccurate model predic- tions of NOy.

4. Summary

The process analysis is performed at a horizontal grid spacing of 4-km over an area in the southeastern US that is centered over NC for August and December, 2002. Emissions provide a dominant source for primary pol- lutants such as NH3, NO, and SO2, and an important source for some secondary pollutants such as NO2 at all sites in August and December. While transport acts as a major sink for these pollutants, it provides a major source for HNO3 and O3 at most sites. Dry deposition is an im- portant sink for all these species, in particular, HNO3, SO2, and O3. Gas-phase chemistry may serve as a source

Figure 8. The temporal variations of simulated ratios of hourly production rates of H2O2 and HNO3 (PH2O2/PHNO3) and observed and simulated afternoon (noon-6 pm) NOy mixing ratios at four sites in NC in August 2002. The solid and dash lines indicate the original and adjusted threshold values, respectively.

Figure 8. The temporal variations of simulated ratios of hourly production rates of H2O2 and HNO3 (PH2O2/PHNO3) and observed and simulated afternoon (noon-6 pm) NOy mixing ratios at four sites in NC in August 2002. The solid and dash lines indicate the original and adjusted threshold values, respectively.

for some species such as NO2 and HNO3 but a sink for other species such as O3 at urban and suburban sites and NO and SO2 at all sites. The roles of these processes in August and December are overall similar for gaseous pollutants, except that transport and dry deposition make larger contributions to O3 formation in August than in December. Emissions provide a dominant source for PM2.5, , EC, POA, and OIN, particularly at urban and suburban sites in both months. PM processes con- tribute to the formation of PM2.5 and at the STN sites and SOA at most sites in December and the forma- tion of at all sites in August and most sites in December. They reduce at most sites in August and at the IMPROVE and CASTNET sites in December as well as PM2.5 and SOA at most sites in August. In August, transport provides a dominant sink for PM2.5 and most of its composition except for and SOA at most STN sites, and it acts as a source for secondary in- organic PM such as and, and SOA at the IMPROVE and CASTNET sites. In December, transport provides a sink for all PM species at most STN sites but a source for most PM species at the IMPROVE and CASTNET sites. Dry deposition is an important sink for all PM species in both months. The fate of NHx is domi- nated by that of NH3, whereas the fate of TNO3 is domi- nated by that of HNO3.

The total Ox production and loss, and the total OH re- acted are much higher in August than in December, par- ticularly at most STN sites, indicating a higher oxidation capacity in August than in December and at urban and

suburban sites than at rural sites. The highest Ox produc- tion and loss occur in urban and suburban areas in the Piedmont and Mountains regions but more uniformly throughout the simulation domain in December. The amount of OH reacted with major gases is much higher over urban and suburban areas in August and over the southern portion of the domain in December. Significant amounts of OH are consumed by BVOCs in the rural and remote areas and a combination of AVOCs and BVOCs in urban and subareas areas in both months. The rates of OH-BVOCs are higher than those of OH-AVOCs at all rural sites in August because of higher BVOCs emissions but the opposite occurs in December. The amount of NO2 produced by the reactions involving HO2 is similar to that involving RO2 at all sites in August but higher than that by the reactions involving RO 2 in December. The production rate of HNO3 due to the reaction of OH with NO2 dominates over that due to the nighttime reactions of VOCs with NO3 radicals in both months.

The values of PH2O2/PHNO3 indicate a NOx-limited O3 chemistry over most areas in August and a VOC-lim- ited O3 chemistry over all areas in December, which is consistent with previous studies [e.g., 3,9]. They also indicate a VOC-limited O3 chemistry in urban and sub- urban areas in the simulation domain in August that is not found in previous model simulations at a coarser grid resolution. The values of PH2O2/PHNO3 are highly sen- sitive to the horizontal grid resolution in summer but insensitive to it in winter. H2O2/HNO3 with a threshold value of 2.4 can indicate O3 chemistry regimes that are overall consistent with those based on PH2O2/PHNO3 over most of areas in both months. Simulated afternoon NOy with a threshold value of 10 ppb indicates a similar O3 chemistry regime to that indicated by PH2O2/PHNO 3 in August. Its threshold value in December may need to be adjusted to be below 5 ppb to make it a more robust photochemical indicator. The O3 chemistry regimes in- dicated by PH2O2/PHNO3 at several sites are consistent with those indicated by observed afternoon NOy values at these sites when a threshold value of 10 ppb or lower is used in August. When the simulated NOy values deviate significantly from observed NOy values, they may not be as robust as PH2O2/PHNO3 to indicate the O3 chemistry regime.

5. Acknowledgements

This work is supported by the National Research Initia- tive Competitive Grant no. 2008-35112-18758 from the USDA Cooperative State Research, Education, and Ex- tension Service Air Quality Program. Thanks are due to Mike Abraczinskas, George Bridgers, Wayne Cornelius, and Karen Harris of NCDENR for providing VISTAS’s emissions and CMAQ modeling results with a 12-km grid spacing and Don Olerud, BAMS, for providing VISTAS’s MM5 simulation results.

References

  1. J.-C. C. Jang, H. E. Jeffries and S. Tonnesen , “Sensitivity of Ozone to Model Grid Resolution―II. Detailed Process Analysis for Ozone Chemistry,” Atmospheric Environ- ment, Vol. 29, No. 21, 1995, pp. 3101-3114. doi:10.1016/1352-2310(95)00119-J
  2. H. E. Jeffries and S. Tonnesen , “A Comparison of Two Photochemical Reaction Mechanisms Using Mass Bal- ance and Process Analysis,” Atmospheric Environment, Vol. 28, No. 18, 1994, pp. 2991-3003. doi:10.1016/1352-2310(94)90345-X
  3. Y. Zhang, X.-Y. Wen, K. Wang, K. Vijayaraghavan and M. Z. Jacobson, “Probing into Regional O3 and Particu- late Matter Pollution in the United States : 2. An Examina- tion of Formation Mechanisms through a Process Analy- sis Technique and Sensitivity Study,” Journal of Geo- physical Research, Vol. 114, No. D22, 2009, pp. 1-31. doi:10.1029/2009JD011900
  4. Y. Zhang, K. Vijayaraghavan and C. Seigneur, “Evalua- tion of Three Probing Techniques in a Three-Dimensional Air Quality Model,” Journal of Geophysical Research, Vol. 110, No. D2, 2005, pp. 1-21. doi:10.1029/2004JD005248
  5. S. C. Yu, R. Mathur and K. Schere, “Evaluation of Real- Time PM2.5 Forecasts and Process Analysis for PM2.5 Formation over the Eastern United States Using the Eta- CMAQ Forecast Model during the 2004 ICARTT Study,” Journal of Geophysical Research, Vol. 113, No. D6, 2008, pp. 1-20. doi:10.1029/2007JD009226
  6. S. C. Yu, R. Mathur, K. Schere, D. W. Kang and D. Tong, “A Study of the Ozone Formation by Ensemble Back Tra- jectory-Process Analysis Using the Eta-CMAQ Forecast Model over the Northeastern US during the 2004 ICAR- TT Period,” Atmospheric Environment, Vol. 43, No. 2, 2009, pp. 355-363. doi:10.1016/j.atmosenv.2008.09.079
  7. K. Wang, Y. Zhang, C. Jang, S. Phillips and B. Wang, “Modeling Intercontinental Air Pollution Transport over the Trans-Pacific Region in 2001 Using the Community Multiscale Air Quality Modeling System,” Journal of Geophysical Research, Vol. 114, No. D4, 2009, pp. 1-23. doi:10.1029/2008JD010807
  8. X.-H. Liu, Y. Zhang, J. Xing, Q. Zhang, K. Wang, D. G. Streets, C. J. Jang, W.-X. Wang and J.-M. Hao, “Under- standing of Regional Air Pollution over China Using CMAQ: Part II. Process Analysis and Ozone Sensitivity to Precursor Emissions,” Atmospheric Environment, Vol. 44, No. 20, 2010, pp. 3719-3727. doi:10.1016/j.atmosenv.2010.03.036
  9. P. Liu, Y. Zhang, S. C. Yu and K. L. Schere, “Use of a Process Analysis Tool for Diagnostic Study on Fine Par- ticulate Matter Predictions in the US. Part II: Process Analyses and Sensitivity Simulations,” Atmospheric Pol- lution Research, Vol. 2, No. 1, 2010, pp. 61-71. doi:10.5094/APR.2011.008
  10. S.-Y. Wu, S. Krishnan, Y. Zhang and V. Aneja, “Model- ing Atmospheric Transport and Fate of Ammonia in Nor- th Carolina, Part I. Evaluation of Meteorological and Che- mical Predictions,” Atmospheric Environment, Vol. 42, No. 14, 2008, pp. 3419-3436. doi:10.1016/j.atmosenv.2007.04.031
  11. G. A. Grell, J. Dudhia and D. R. Stauffer, “A Description of the Fifth Generation Penn State/NCAR Mesoscale Model (MM5),” National Center for Atmospheric Re- search, Boulder, 1994.
  12. M. R. Houyoux, J. M. Vukovich, C. J. Coats Jr., N. W. Wheeler and P. S. Kasibhatla, “Emission Inventory De- velopment and Processing for the Seasonal Model for Regional Air Quality (SMRAQ) Project,” Journal of Geophysical Research, Vol. 105, No. D7, 2002, pp. 9079- 9090. doi:10.1029/1999JD900975
  13. D. W. Byun and K. L. Schere, “Review of the Governing Equations, Computational Algorithms and Other Compo- nents of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System,” Applied Mechanics Reviews, Vol. 59, No. 2, 2006, pp. 51-77. doi:10.1115/1.2128636
  14. S. Sillman, “The Use of NOy, H2O2, and HNO3 as Indi- cators for Ozone-NOx-Hydrocarbon Sensitivity in Urban Locations,” Journal of Geophysical Research, Vol. 100, No. D7, 1995, pp. 4175-4188. doi:10.1029/94JD02953
  15. G. S. Tonnesen and R. L. Dennis, “Analysis of Radical Propagation Efficiency to Assess Ozone Sensitivity to Hydrocarbons and NOx 1. Local Indicators of Instanta- neous Odd Oxygen Production Sensitivity,” Journal of Geophysical Research, Vol. 105, No. D7, 2000, pp. 9213- 9225. doi:10.1029/1999JD900371
  16. S. Sillman, D. He, C. Cardelino and R. E. Imhoff, “The Use of Photochemical Indicators to Evaluate Ozone-NOx- Hydrocarbon Sensitivity: Case Studies from Atlanta , New York , and Los Angeles ,” Journal of the Air & Waste Management Association, Vol. 47, No. 10, 1997, pp. 642- 652. doi:10.1080/10473289.1997.10464407
  17. G. S. Tonnesen and R. L. Dennis, “Analysis of Radical Propagation Efficiency to Assess Ozone Sensitivity to Hydrocarbons and NOx: 2. Long-lived Species as Indica- tors of Ozone Concentration Sensitivity,” Journal of Geophysical Research, Vol. 105, No. D7, 2000, pp. 9227- 9241. doi:10.1029/1999JD900372
  18. M.-U. Hammer, B. Vogel and H. Vogel, “Findings on H2O2 /HNO3 as an Indicator of Ozone Sensitivity in Ba- den-Württemberg, Berlin-Brandenburg, and the Po Valley Based on Numerical Simulations,” Journal of Geophysi- cal Research, Vol. 107, No. D22, 2002, pp. LOP 3-1- LOP 3-18. doi:10.1029/2000JD000211
  19. C.-H. Lu and J. S. Chang, “On the Indicator-Based Ap- proach to Assess Ozone Sensitivities and Emissions Fea- tures,” Journal of Geophysical Research, Vol. 103, No. D3, 1998, pp. 3453-3462. doi:10.1029/97JD03128
  20. J. B. Milford, D. F. Gao, S. Sillman, P. Blossey and A. G. Russell, “Total Reactive Nitrogen (NOy) as an Indicator of the Sensitivity of Ozone to Reductions in Hydrocarbon and NOx Emissions,” Journal of Geophysical Research, Vol. 99, No. D2, 1994, pp. 3533-3542. doi:10.1029/93JD03224