Vertical Circulation of Atmospheric Pollutants near Mountains during a Southern California Ozone Episode

This study investigates the air pollutant interactions and emission source contributions to ozone (O3) formation within a complex terrain. Air quality simulations using the Community Multiscale Air Quality (CMAQ) Model focused on vertical distributions of O3 for the July 14–18, 2005 episode in the South Coast Air Basin (SoCAB). The Zero-Out method was applied in sensitivity tests for seven emission source categories. Elevated O3 concentrations were found near the top of the planetary boundary layer (PBL, ~1200 m) and in the free troposphere (~3500 m) over the eastern SoCAB. Low O3 concentrations were found near the surface at the center of the basin due to nitrogen oxide (NO) titration by fresh vehicle exhaust. Sea and land breezes, enhanced by up-slope flows (the “mountain chimney effect”) transported O3 upward. Formation of O3 is sensitive to the H2O2/HNO3 ratio, depending on fresh vs. aged pollutant mixtures. Biogenic emissions were important contributors to O3 formation, both in the SoCAB and at the top of the PBL. In contrast, the highest vehicle contributions to O3 were found far from urban areas and in the lower free troposphere. Vertical cross-sectional analysis provided some insights into the O3 formation and mixing processes present in the SoCAB.


INTRODUCTION
The South Coast Air Basin (SoCAB) in California, including the city of Los Angeles (LA), is located in complex terrain surrounded by mountains with ridges of more than 2000 m above mean sea level (asl).Anthropogenic air pollutants from urban areas, transported by sea breezes, contribute to high ozone (O 3 ) concentrations on the mountain slopes of the eastern SoCAB during summer.
Insufficient ventilation and accumulation of photochemical air pollutants in the SoCAB have been studied since Haagen-Smit (1952).The area has been known to report the highest O 3 concentrations in the U.S. Hourly O 3 concentrations, which exceeded 500 ppb during the 1960s, have been reduced to < 120 ppb for the maximum 8-hr average in recent years.Fujita et al. (2013) showed the decreasing O 3 trend in SoCAB, and pointed to changes in the photochemical reaction due to variations in volatile organic compound (VOC)/oxides of nitrogen (NO x ) ratios during the past three decades.
Updrafts generated by solar radiation on mountain slopes result in vertical transport of air pollutants, called the "mountain chimney effect", (e.g., Lu and Turco, 1996;Langford et al., 2010).Three-dimensional (3-D) observations using aircraft and remote sensors, such as NASA's Discover-AQ (http://www.nasa.gov/mission_pages/discover-aq/news/DAQ-20130226.html)have improved understanding of the vertical structure of these events.To understand air pollutant evolution in complex terrain, it is important to consider O 3 formation through atmospheric transport and photochemical reactions of pollutants.
Since three-dimensional observations have temporal and spatial limitations, a comprehensive chemistry-transport model can be a useful complement to understand the influence of transport and chemical mechanisms.Most air quality modelling efforts have estimated present and future ground-level concentrations, with little attention given to O 3 that accumulates aloft, especially near mountainous terrain.Only a few of these studies examined the sourcespecific contributions to high O 3 levels.Zhang et al. (2014) examined emission source contributions to fine particles and O 3 concentrations for eight large cities in the southeastern U.S. With respect to engine exhaust contributions to O 3 formation, NO x titration yielded negative effects within the urban centers, overall contributions were low for nearby suburban ground-based concentrations, but exhaust contributions to O 3 increased 30-50 km far from the city centers, especially in mountainous regions.Similar results were observed by Collet et al. (2012), with SoCAB O 3 concentrations downwind of the urban center being influenced by engine exhaust.However, it is difficult to determine which types of emissions have the greatest impact on O 3 formation given the complex interactions among emissions, meteorology, and photochemistry in complex terrain.
In this study, a high SoCAB O 3 episode near the San Bernardino Mountains in the eastern part of the basin is investigated with the Community Multiscale Air Quality (CMAQ) Modeling System (Byun and Schere, 2006).Instead of examining the horizontal distribution for ground-based concentration, the approach taken here details vertical transects while systematically eliminating precursor emissions from major SoCAB source categories.Ratios of precursor, intermediate and end-product concentrations, which are also calculated by CMAQ, are used to better understand the atmospheric chemistry.One-hour average O 3 concentration for 96 hours measured at 43 monitoring sites of the U.S. EPA in SoCAB were compared with CMAQ simulation results of the relevant time period and location in Table 2; the mean normalized bias (MNB), the mean normalized gross error (MNGE), and the peak accuracy ratio of 9.5, 15.4, and 16.3%, respectively, show reasonably good agreement.Large discrepancies were found at three monitoring sites located near the northwest computational domain near Bakersfield.However, obtained MNB and MNGE result in significantly less than 15 and 30%, respectively, which is the former simulation guideline of U.S. EPS.In addition, correlation between O 3 concentration measured at the San Bernardino-4 th Street site and the Pomona site close to the point shown Fig. 1(b), were both 0.86, and further discussion with this accuracy is considered to be possible.

METHODS
Several methods have been used to evaluate source contributions to O 3 , such as sensitivity analysis using the decoupled direct method (DDM) (Dunker, 1984), the source oriented tagged methods: OSAT (Ozone Source Apportionment Technology; Yarwood et al., 1997), or the Integrated Source Apportionment Method (ISAM; Release notes for CMAQ Version 5.0.2).The brute force method (BFM), a sensitivity analysis tool, is applied for this study with 100% emission reductions (i.e., Zero-Out method) for seven major source types, including: on-road gasoline engine exhaust, on-road diesel engine exhaust, non-road gasoline engine exhaust, non-road diesel engine exhaust, industrial (including point sources, i.e., power plant, boilers, steel industry, and solvent utilization), residential, and biogenic emitters.Area source emissions from small factories, offices, and homes are included in the residential category, whereas wildfires are included with biogenic emissions.

Parameter Equation Results
Mean Normalized Bias (MNB) aM i = Model concentration for species i. O i = Observed concentration for species i.
Vertical wind velocities were estimated from the convergence and divergence of horizontal wind velocities in CMAQ-ready wind data for hydrostatic balance.The basic equations and methods follow those of Kato and Saito (1995).To explore a refined 3-D distribution of air pollutants and air flows, data visualization with Cartesian coordinates instead of spatial plots with σP coordinates (the original system in the CMAQ Model) are used.(The σP coordinate system is a common coordinate system used in computational models, the details numerical handling is described in Zavisa Janjic et al., 2010.)

RESULTS
Fig. 1 shows an example of vertical O 3 and wind distributions.For better resolution, finer vertical coordinate grids of CMAQ are used at the altitude close to the surface according to the vertical coordinate parameter (VGLVLS).H 2 O 2 to HNO 3 ratios have been used as chemical indicators for O 3 formation (Sillman, 1995;Tonnesen and Dennis, 2000a, b;Sillman and He, 2002).Hydroxyl radicals (OH), which are photolysis-generated radicals from O 3 to olefin reaction, and aldehyde produce H 2 O 2 via hydroperoxide (HO 2 ).HO 2 radicals and OH also produce HNO 3 from NO via NO 2 .Since HNO 3 is to be generated earlier than H 2 O 2 in the photolysis process and shows a tendency to decrease at night, H 2 O 2 /HNO 3 ratios also serve as an indicator of aging.The H 2 O 2 /HNO 3 ratio shows  increasing tendency with time (in other words aging) after when HNO 3 concentration shows decreasing trend in early afternoon.Grossmann et al. (2003) reported diurnal variations of H 2 O 2 and HNO 3 concentrations in Berlin, Germany, and observed typical diurnal trend of both species.Fig. 4 shows the diurnal variation of H 2 O 2 and HNO 3 and H 2 O 2 /HNO 3 ratios at Riverside (Fig. 4(i)) and on the western mountain slopes (Fig. 4(ii)).The time difference of HNO 3 peak at the western mountain slope which locates in the downstream of 52 km distance from Riverside, showed 4 hours.However, H 2 O 2 /HNO 3 ratio increased almost same time at 1600 PST.The maximum H 2 O 2 /HNO 3 ratio near Riverside was 1.57, somewhat higher than the 0.5 ratio reported by Grossmann et al. (2003) in Berlin.This ratio increased to 5 at the San Bernardino Mountains during the night, and to ~10 in the morning, as HNO 3 concentrations decreased to nearly zero in the aged air mass.The H 2 O 2 /HNO 3 peaked near the mountain range summit.
The non-linear relationship between O 3 contributions and H 2 O 2 /HNO 3 ratios is examined in Fig. 5.In order to understand source contributions to O 3 concentrations and O 3 precursors, the Zero-Out method for the seven emission source categories was used.A rapid increase in the vehicle exhaust contribution to O 3 formation was found for H 2 O 2 /HNO 3 ratios from 0.1 to ~1.6.As air mass ages, vehicle contribution decreases with increasing H 2 O 2 /HNO 3 ratios from ~1.6 to ~3.Biogenic contributions to O 3 seem to be more sensitive to the H 2 O 2 /HNO 3 ratio with the highest contribution found at the lowest ratios.(1995).Contributions to O 3 formation levels off for both sources as H 2 O 2 /HNO 3 ratios exceeded 3.However, the result of Silman showed lower value (0.3-0.5).area promoted by the mountain chimney effect during the simulation period.These areas were the accumulation area of air pollutants after transportation of the updraft flow.Since the production mechanism of each hot spot is different, a base case O 3 concentration and emission source contributions at time of the localized O 3 maxima are shown in Fig. 6.The hot spot pollutant concentrations and H 2 O 2 /HNO 3 ratios are summarized in Table 3.At 2100 PST on July 17, Location No. 1 (elevation ~990 m) in Fig. 6 represents transport from the northwest, with the highest H 2 O 2 (4.9 ppb) and lowest HNO 3 (1.1 ppb) concentrations, characteristics of an aged air mass.Source contributions to O 3 were dominated by residential emissions, accounting for 36.5% of O 3 , followed by biogenic sources (26.1%).On-road gasoline and diesel exhaust contributions were 4.8% and 4.1%, respectively.

DISCUSSION
The maximum O 3 concentration of 110 ppb was found at Location No. 2, situated ~990 m above an urban area between the cities of LA and Riverside at 1700 PST.In contrast to Location No.1, No. 2 showed the lowest level H 2 O 2 (3.0 ppb) and highest HNO 3 (11.3ppb) concentrations among the six points, characteristic of a fresh air mass.Biogenic Fig. 6.Time averaged (1300 to 2300 PST on July 17, 2005 ) hot spots (O 3 > 75 ppb, indicated by black dots) with emission source contributions determined by the zero-out cases for: 1) largest residential emission contribution (36.5%); 2) largest biogenic emission contribution (21.6%); 3) largest on-road gasoline engine emission contribution (13.9%); 4) largest industrial emission contribution (14.6%); 5) lowest biogenic contribution (12.7%); and 6) largest VOC and NO 2 concentrations, and 2 nd highest biogenic contribution (21.7%) to O 3 .The highest engine exhaust contribution, accounting for 29.6% of O 3 (including 13.9% from on-road gasoline vehicle exhaust) was found at Location No.3 (elevation ~3417 m), over northeastern LA and Riverside, at 1800 PST.This was accompanied by a 19% contribution from biogenic and 17.5% from residential emissions.The second and third highest engine exhaust contributions were found at Locations No. 4 and No. 5 in Fig. 6, both of which are above the PBL and in the eastern part of the SoCAB.This suggests that engine exhaust emissions from the urban center began to oxidize, mix with biogenic emissions, and gradually contribute to O 3 formation during transport.The highest source contribution to O 3 at Location No. 4 was industry (14.6%).At Location No. 5, away from anthropogenic emission sources, the biogenic contribution to O 3 formation was the lowest (12.7%) among the hot spots.The highest VOC and NO 2 concentrations along with the second highest biogenic emission contribution (21.7%) to O 3 were found in Location No. 6, whereas the lowest VOC and NO 2 concentrations were found at the elevated Locations No. 4 and No. 5 toward the mountain and desert areas.
The mountain chimney effect is further demonstrated in Fig. 7 for the same time period (1300 to 2300 PST) on July 14.As seen in Fig. 6  highest contributions (13.9%) found above the PBL (~3417 m) outside of the LA urban areas.In contrast, the highest biogenic contribution (21.6%) was found near the urban area between the cities of LA and Riverside.These simulations emphasize the importance of understanding pollutant aging, evolution, and transport.
Fig. 1 shows an example of vertical O 3 and wind distributions.For better resolution, finer vertical coordinate grids of CMAQ are used at the altitude close to the surface according to the vertical coordinate parameter (VGLVLS).Fig. 1(a) includes the direct CMAQ output.The vertical spacing differs for sea and mountain because the simulation is

Fig. 3
Fig. 3 shows the vertical distribution of precursor gases that correspond to the vertical O 3 distribution in Fig. 1(b).Since large amounts of nitric oxide (NO) due to human activities are released into the atmosphere, a high concentration was seen in LA as shown in Fig. 3(a).As marine-transported O 3 mixed and reacted with NO in LA, O 3 concentrations decreased, consistent with high NO and nitrogen dioxide (NO 2 ) concentrations in the western part of the basin, and as shown in Fig. 3(b), high NO 2 concentrations in the western and eastern parts of the SoCAB which was transported eastward by the sea breeze.The highest NO 2 concentration was found over the western mountain slopes.A wider and deeper distribution was found for VOCs as shown in Fig. 3(c), as up-slope flows transported VOCs above the PBL.Elevated VOCs over the Pacific Ocean at the top of PBL may reflect VOC transport from the northwest.Fig. 3(d) shows nitric acid (HNO 3 ) concentrations resulting from NO x oxidation.HNO 3 concentrations were highest downwind (eastward) of high NO x concentrations, consistent with elevated NO (i.e., fresh vehicle exhaust emission) in the western part of the SoCAB.Higher HNO 3 concentrations were found above the PBL, possibly resulting from vertical mixing.High hydrogen peroxide (H 2 O 2 ) concentrations at the top of the PBL over the Pacific Ocean (Fig. 3(e)), similar to those found for VOC in Fig. 3(c), were apparently affected by transport from the northwest.H 2 O 2 to HNO 3 ratios have been used as chemical indicators for O 3 formation(Sillman, 1995; Tonnesen and  Dennis, 2000a, b;Sillman and He, 2002).Hydroxyl radicals (OH), which are photolysis-generated radicals from O 3 to olefin reaction, and aldehyde produce H 2 O 2 via hydroperoxide (HO 2 ).HO 2 radicals and OH also produce HNO 3 from NO via NO 2 .Since HNO 3 is to be generated earlier than H 2 O 2 in the photolysis process and shows a tendency to decrease at night, H 2 O 2 /HNO 3 ratios also serve as an indicator of aging.The H 2 O 2 /HNO 3 ratio shows

Fig. 2 .
Fig. 2. Horizontal O 3 concentrations and wind fields at 1800 PST on July 17, 2005 at: (a) 1179 m asl (closest to the top of the planetary boundary layer [PBL]), and (b) 3417 m asl (in the free troposphere).Black areas indicate locations of mountains.The blue bar shows the area corresponding to the latitude cross-section in Fig. 1(b).Pink circles indicate onehour back-trajectory segments from the point of highest O 3 concentration of a red-cross symbol.

Fig. 4 .
Fig. 4. Diurnal variation of H 2 O 2 and HNO 3 and H 2 O 2 /HNO 3 ratios on July 17, 2005 for: (i) at Riverside and (ii) western slope of the San Bernardino mountain range.

Fig. 5 .
Fig. 5. Source contributions to O 3 concentrations as a function of H 2 O 2 /HNO 3 ratios for: (a) engine exhaust, and (b) biogenic emissions, obtained from the cross-sectional area shown in Figure 1b from 1300 to 2300 PST on July 17, 2005.The red line shows average source contributions.The box plot gives the standard deviation of source contributions.The blue line shows maximum and minimum values of source contributions.

Fig. 6
Fig. 6 shows average O 3 concentrations from 1300 to 2300 PST on July 17.A high O 3 concentration (hot spots) was often seen near top of PBL and vertical circulation Fig. 7. Time averaged (1300 to 2300 PST on July 14, 2005) hot spots hot spots (O 3 > 75 ppb, indicated by black dots) with emission source contributions determined by the zero-out cases.(As hot spots were not found on the eastern side of the San Bernardino mountains, only the western side is shown.

Table 1 .
Summary of CMAQ model simulation conditions for the period of July 14-18, 2005 (GMT).

Table 2 .
Comparison between model simulation and ambient measurements.

Table 3 .
Hot spot pollutant concentrations and H 2 O 2 /HNO 3 ratios at different elevations from 1700 to 2200 PST on July 17, 2005.emissionsconstituted the highest source contribution, accounting for 21.6% of O 3 , followed by residential emissions (15.1%) and engine exhaust (12.4%, mostly on-road gasoline emissions [7.9%]).The low H 2 O 2 /HNO 3 ratio of 0.3 suggests the atmosphere was VOC-limited for this part of the SoCAB.Residential and biogenic emissions contributed to 51.7 and 24.1% of total VOCs, respectively.