Measurement of microenvironmental ozone concentrations in Durham, North Carolina, using a 2B Technologies 205 Federal Equivalent Method monitor and an interference-free 2B Technologies 211 monitor

Abstract

During August and September of 2012, researchers conducted a microenvironmental (ME) monitoring study in Durham, North Carolina, using two 2B Technologies O₃ monitors: a dual-beam model 205 Federal Equivalent Method (FEM) 254 nm photometer and a newly developed model 211 interference-free dual-beam photometer. The two monitors were mounted in a wheeled, fan-cooled suitcase together with a battery, a disposable N₂O cartridge for the model 211 monitor, and filtered sample lines. A scripted technician made paired O₃ measurements in a variety of MEs within 2 miles of a fixed-site FEM O₃ photometer at the Durham National Guard Armory. The ratio of the 211 to Armory O₃ concentrations tended to be lowest (<0.3) for 45 indoor MEs and highest (>0.8) for 104 outdoor MEs. The mean values of the ratio for in-vehicle MEs tended to fall between 0.2 and 0.7—the mean for all 27 in-car tests was 0.3. The ratio values for indoor MEs tended to be higher when the enclosure was well ventilated. The outdoor ratios tended to be lower when the measurement was made downwind of nearby roadways, likely due to exhaust NO. The in-vehicle ratios tended to be larger with windows open than closed; the smallest occurred with closed windows, active air conditioning, and vent recirculation. The 205 − 211 measurement differences were generally small, with 94% of the 176 sample differences below 5 ppb. Five differences were above 10 ppb with the largest values (173.9 and 63.6 ppb) occurring inside a violin repair shop. Roadway proximity tended to increase the differences for outdoor locations. The largest in-vehicle difference (6 ppb) occurred at a convenience store service station. As addressed in regulatory models, such differences may reduce estimated population O₃ exposure by 30–50% in indoor and in-vehicle MEs where individuals spend more than 80% of their time.

Implications: Computer models used to estimate exposures of human populations—such as the Air Pollution Exposure Model (APEX) developed by the U.S. Environmental Protection Agency—can be improved by use of direct microenvironmental (ME) measurement comparisons to nearby fixed-site monitors used for determining regulatory compliance. Simultaneous measurements made by model 211 and model 205 ozone monitors in a variety of MEs indicated that Federal Equivalent Method photometers similar to the model 205 may read high in the presence of various interferences associated with indoor sources and motor vehicles, increasing modeled exposures in such environments by 20–100%.

Introduction

The U.S. Environmental Protection Agency (EPA) has developed a version of the Air Pollution Exposure Model (APEX) applicable to ozone (APEX-Ozone). APEX-Ozone estimates population exposures to ozone by simulating typical movements of people through representative microenvironments (MEs). The exposure estimates are calculated using ambient monitoring data, diary-derived activity patterns, activity-specific breathing rate estimates, and appropriate ambient-data adjustment factors specific to each microenvironment (EPA, 2011). The exposure estimates obtained from APEX-Ozone can be validated through personal monitoring that directly measures microenvironmental ozone concentrations (Stock et al., 1985; Chang et al., 2000; EPA, 2013).

In 2005, TRJ Environmental, Inc. (TRJ), identified the 2B Technologies miniaturized, battery-powered, continuously data-logging model 202 and model 205 254 nm ozone photometers (Boulder, CO) as promising instruments for use in personal ozone monitoring studies and evaluated their ruggedness in a pilot test of these monitors in Raleigh, North Carolina. A 205 monitor was carried in a backpack while following a prepared script through a sequence of microenvironments representative of children's activities, high-energy activities, and elevated ozone concentrations. Measurements made by the 205 monitor were compared with ambient ozone concentrations measured simultaneously by a fixed-site monitor located at Millbrook High School where a 202 monitor was co-located.

This study indicated strong microenvironmental effects on ozone concentrations, consistent with expectations. Ozone concentrations were lower within indoor and in-vehicle microenvironments with closed windows. Opening windows increased ozone concentrations but not to the level of outdoor values. The effect of being upwind or downwind of roadways while walking and cycling was less noticeable. The 2B monitors appeared to offer a promising method for collecting personal ozone monitoring data at high time resolution. A summary of this work appears in an International Society of Exposure Science (ISES) conference poster by Long et al. (2008).

Researchers conducted a follow-up study during the summer of 2012 using paired monitors: the model 205 monitor used in the 2005 Raleigh study and a model 211—a new “interference-free” monitor developed by 2B Technologies. A subcontractor (Zedek Corporation, Durham, NC) mounted the two monitors in a wheeled suitcase together with a battery, a disposable N₂O cartridge for the model 211 monitor, and filtered sample lines. Following a prepared script, a technician used the two monitors to make simultaneous ozone measurements in a variety of microenvironments in Durham, North Carolina, within 2 miles of the fixed-site monitor located at the Durham National Guard Armory.

Analyses of the ozone measurements made by the paired monitors in various microenvironments determined a systematic bias in the measurements made by the 205 relative to the 211, particularly within indoor environments. The data provided by the 2012 Durham study provide a means of better evaluating exposure estimates obtained from the current version of APEX as applied to ozone.

Methodology

The Durham, North Carolina study area

The study area was defined as a circle with radius of approximately 2 miles centered on the ozone fixed-site monitor located at the Durham National Guard Armory (latitude = 36.0330°, longitude = 78.9043°). It included the residence of the monitoring technician, enabling the use of this home for ozone measurements during typical indoor tasks within the indoors—residential microenvironment. The study area–which included representative microenvironments of interest, residential areas, schools, parks, arterial roads, and a major shopping mall—was close to the offices of Zedek Corporation, permitting easy access to their calibration and zero-span services.

Historically, ozone levels reported by the Durham National Guard Armory monitor (EPA ID = 37-063-0015) have been similar to those reported by the Raleigh Millbrook monitor used in the 2005 study. Figure 1 shows the Armory air pollution monitoring station viewed from the east side. The project report by Johnson and Capel (2012) provides maps and satellite views of the area surrounding this monitor. Ozone concentration at the Armory site is measured by a Thermo model 49C ozone analyzer (Franklin, MA). The ozone calibration standard for the Armory site is a Thermo model 49C-PS monitor. The Armory ozone data used in the analyses presented here are considered preliminary by the North Carolina Division of Air Quality and are subject to change.

Figure 1. The air pollution monitoring station (37-063-0015) at the Durham National Guard Armory.

Selection of potential monitoring locations and preparation of scripts

A survey of the designated study area provided a list of potential geographic locations that included microenvironments likely to be visited by the general public. Each of the surveyed locations included a variety of outdoor microenvironments (e.g., entrance ways, picnic areas, parking lots, sidewalks near roads). In most cases, the location included at least one usable indoor microenvironment.

Five scripts were developed. Each script specified the monitoring activities for one calendar day, typically within a 10-hr period (10 a.m. to 8 p.m.) with elevated ambient ozone concentrations. Four scripts specified a distinct geographic location for each hour and up to four microenvironments at each location. The technician visited each of the four microenvironments specified for a particular clock hour and took an ozone measurement over about 10 min. This schedule left 20 min during each hour to move between monitoring locations and set up equipment. Ideally, the technician would complete about 40 ozone measurements per scripted day.

A fifth script defined ozone measurements made in the technician's car (a Toyota Prius was chosen to minimize vehicle self-pollution and provide a better measure of roadway exposure). Monitoring conditions were varied to test the effects of changing ventilation states such as windows open, windows closed, different road types, traffic volumes, and vehicle speeds.

Setup, calibration, and zero span of ozone monitors

Two ozone 254 nm photometers manufactured by 2B Technologies were employed during the study: a model 205 and a model 211.

The model 205 ozone monitor is a dual-beam Federal Equivalent Method (FEM) instrument that uses two detection cells to simultaneously measure ultraviolet (UV, 254 nm) light transmitted through Hopcalite-scrubbed, deozonized reference air and unscrubbed sample air. Detailed technical specifications for the unit can be found at http://www.twobtech.com/products.htm.

The model 211 scrubberless ozone monitor is a similar dual-beamed 254 nm photometer that uses the reaction between ozone and nitric oxide (NO)—photolytically generated in situ from N₂O—to more selectively remove ozone from the sample than a Hopcalite scrubber. Because the NO reaction with ozone is orders of magnitude faster than with any other ambient gas, it more efficiently removes ozone without removing compounds that also absorb 254 nm light. The UV light intensity is measured simultaneously in the reference and sample modes, and the concentration of ozone is calculated from Beer's law. According to the manufacturer (http://www.twobtech.com/products.htm), the instrument is ideally suited for measurements of ozone where interference is likely from particles, mercury, or 254-nm-absorbing volatile organic compounds (VOCs) (Birks et al., 2013).

Zedek Corporation staff mounted the two monitors in a wheeled suitcase together with a battery, a disposable N₂O cartridge for the model 211 monitor, and filtered sample lines. The cart was typically operated in an upright (vertical) position resting on all four wheels, as shown in Figure 2. Zedek calibrated both monitors and performed a zero-span check of each unit after each day of monitoring using a National Institute of Standards and Technology (NIST)-traceable certified Dasibi 1008-PC transfer standard (Glendale, CA). The calibration and zero-span results were used to adjust the ozone concentrations to account for instrument calibration and daily zero/span (100 ppb) drift (typically <2 ppb).

Figure 2. Cart in operating mode.

Ozone measurements in typical indoor-outdoor microenvironments

Four of the scripts that listed a distinct geographic location for each hour of the monitoring period, specifying up to four different microenvironments at each location, were implemented on August 24 and September 7, 14, and 21, 2012. Specific dates were selected a day or two in advance based on ozone air quality forecasts provided by the North Carolina Division of Air Quality and local weather forecasts to maximize anticipated ambient ozone levels.

Ozone measurements in a motor vehicle

The fifth script implemented on September 27, 2012, specified ozone measurements mostly made in the technician's hybrid car (a 2010 Toyota Prius) that included commuting on four road types (Interstate, arterial, commercial, and residential); parking in residential driveways, parking lots, parking decks, roadside shoulder locations, and service stations; driving under a variety of ventilation conditions including windows and vent open, windows closed (vent open), and windows closed (air conditioning on, vent recirculating); and using drive-through lanes at fast-food restaurants. The fifth script also included a few indoor and outdoor locations, including a daily sampling visit to the fixed-site monitor near the Durham National Guard Armory.

Database preparation

Each data set associated with these five scripts consisted of a series of 32–38 test events usually performed between 10 a.m. and 8 p.m. on a single day. The data for each test event included the start time, the sampling duration (about 10 min), the average ozone concentrations measured by the 205 and 211 monitors during the event, and all site characteristics data recorded on the field data sheet for the event. The data for each test event also included the hourly-average ozone concentration measured by the Thermo model 49C monitor at the Armory site for the hour containing the test interval, selected location (latitude, longitude, altitude), and meteorological (temperature, relative humidity, wind speed) variables collected by the technician on Garmin GPS II (Kansas City, KS) and Kestrel 3000 (Birmingham, MA) instruments at outdoor MEs, and weather parameters reported by Station No. WBAN 13722 at the Raleigh-Durham airport. The combined database includes 176 test events: 45 indoor events + 104 outdoor events + 27 in-vehicle events.

Although this paper focuses on the average ozone concentrations measured during each test event, we also examined the minute-by-minute concentrations recorded by the 205, 211, and Armory (49C) monitors during each 10-hr monitoring period, including the periods between tests when the cart was “in-transit” between test microenvironments. Figure 3 illustrates these data for September 14, 2012.

Figure 3. Time-series plot of 1-min average Durham Armory Thermo 49C, 2B Techologies 205, and 2B Technologies 211 microenvironmental ozone concentrations (ppb) illustrating sequential visits to a car repair garage office (10:00 a.m.), an urban park restroom facility (11:00 a.m.), an outdoor location next to the outdoor Armory fixed-site monitor (11:15 a.m.), a fast-food restaurant during lunch (12:15 p.m.), a guitar and violin repair shop (1:15 p.m.), an art and frame shop (2:00 p.m.), a book store (3:15 p.m.), a record shop (4:00 p.m.), a pizzeria during dinner (5:00 p.m.), an ice cream shop (6:15 p.m.), a residence, indoors (7:15 p.m.), and a residence, outdoors (7:30 p.m.).

Special tests of possible indoor interferences

Two days of special tests were conducted in the residence and parked car of the technician to test sources of monitor interference likely to be present in the residential environment. The technician simulated 14 household activities (cooking, laundering, burning candles, cleaning, etc.), each of which was expected to emit potential interferences (e.g., VOCs, particulate matter [PM]) into the air sampled by the two ozone monitors (Spicer et al., 2010). Each of these emission tests consisted of a background period (˜5 min of monitoring before the activity began), an emission period (10–40 min of monitoring while the source was emitting), and a postemission period (˜10 min of monitoring after the emissions had ceased). The script also included three “no-emission” test periods to provide an indication of the general background levels in the residence.

Although the results summarized later in this paper are based on the average ozone concentrations measured during the background, emission, and postemission periods, we also examined the minute-by-minute concentrations measured by the 205 and 211 monitors during each monitoring day, including periods between the emission tests. Figure 4 provides a plot of household activity data for October 11, 2012, together with the available 1-hr average ozone concentrations reported by the Armory (49C) monitor.

Figure 4. Illustrative time-series plots of 1-hr average Durham Armory Thermo 49C, 1-min average 2B Tech 205, and 1-min 211 ozone concentrations (ppb) measured during household activities that included toasting and grinding spices during lunch preparation in kitchen (10:45 a.m.), cooking lunch on the kitchen gas stove (11:45 a.m.), briefly sampling outdoor air passing to (12:00 p.m.) and from (12:55 p.m.) test vehicle cabin outside the residence, sampling closed test vehicle cabin before and after installing an air freshener (12:30 and 12:45 p.m.), washing lunch cooking and dining utensils in the kitchen dish washer (1:15 p.m.), shampooing hair during a bathroom shower (1:40 p.m.), mopping the bathroom floor with a disinfectant cleaner (3:45 p.m.), and briefly sampling the ambient air outside the residence (5:05 p.m.).

Results and Discussion

We calculated the following values for each test event associated with the five scripts:

1

2

R _211/Armory relates the relatively interference-free concentration measured by the 211 monitor in a particular microenvironment location to the ambient (outdoor) concentration measured at the same time by the fixed-site monitor at the Durham National Guard Armory. D _205−211 is the concentration difference between the readings of the 205 and 211 monitors when sampling the same air within a particular microenvironment. A positive value of D _205−211 suggests that the 205 monitor reading is biased high due to interference effects associated with the microenvironment.

Mean values by microenvironment

Table 1 lists event means for R _211/Armory and D _205−211 for 26 indoor microenvironments, including residences, offices, restaurants, stores, a museum, and a hospital. With respect to the 26 mean values for R _211/Armory, 11 of the ME means are below 0.10, 12 range from 0.10 to 0.30, and 3 are above 0.42. The latter three means are associated with a garden supply store with open doors (0.4250), a car repair shop with open garage doors (0.6857), and three unheated park restrooms (0.7846) open to outdoor air. These locations likely had relatively large air exchange rates during the tests due to open doors, windows, or vents. The mean value of R _211/Armory for the 45 indoor events is 0.1695.

Table 1. Descriptive statistics for ozone concentrations associated with indoor microenvironments

ME No.	Indoor Microenvironment	Number of Test Events	Mean Ratio of Monitor Concentrations (211/Armory)	Mean Difference between Monitor Concentrations (205 − 211), ppb
101	Art gallery and frame shop	1	0.0063	11.6
102	Beauty salon	1	0.0000	3.7
103	Bicycle shop	1	0.1281	2.8
104	Bookstore	2	0.0115	4.1
105	Cafeteria	1	0.2561	1.1
106	Car repair garage	1	0.6857	−0.5
107	Car repair office	1	0.0619	0.6
108	Clothes and novelties store	1	0.0241	4.5
109	Dry cleaner	1	0.1978	2.5
110	Family restaurant	1	0.3000	2.5
111	Fast-food restaurant	5	0.1385	2.6
112	Garden supply store	1	0.4250	1.3
113	Smoking supplies shop	1	0.1294	3.6
114	Hospital	1	0.1618	0
115	Ice cream restaurant	2	0.0845	4.7
116	Laundromat	1	0.2098	1.8
117	Museum	1	0.0045	4.0
118	Office building	4	0.0684	2.7
119	Park restroom	3	0.7846	0.3
120	Pizza restaurant	2	0.0281	4.1
121	Recreation center	1	0.2186	2.7
122	Residence	4	0.1184	5.7
123	Shopping mall food court	3	0.1637	2.0
124	Shopping mall pedestrian	2	0.1176	1.9
125	Toy store	1	−0.0121	12.3
126	Violin shop	2	0.0320	118.8
	All	45	0.1695	8.3
	All (omitting violin shop)	43	0.1759	3.2

The 26 microenvironmental means for D _205−211 in Table 1 vary from −0.5 to 118.8 ppb. (Where adjustment for average daily monitor drift results in a small negative value for near-zero readings, the calculated D _205−211 subtracts this negative number, slightly enlarging the 205 − 211 difference). Twenty-two of the 26 indoor means are less than 5 ppb. The other three means are associated with an art gallery/frame shop (11.6 ppb), a toy store (12.3 ppb), and a violin sales and repair shop (118.8 ppb). The violin shop was visited on two different days, producing values for D _205−211 of 173.9 and 63.6 ppb. During these visits, the violin repair staff was using various solvents and fixatives, which may have produced interference effects in the 205 monitor. The mean D _205−211 value for the 45 indoor samples is 8.3 ppb, although the mean value drops to 3.2 ppb (range: −0.5–14.1 ppb) if we omit the two violin shop samples.

Table 2 lists results for 18 outdoor microenvironments, including parks, playgrounds, residential yards, commercial properties, parking facilities, and roadside locations. The 18 microenvirionmental means for R _211/Armory vary from 0.6806 to 1.2275. The mean value of R _211/Armory for the 104 outdoor events is 0.9288.

Table 2. Descriptive statistics for ozone concentrations associated with outdoor microenvironments

ME No.	Outdoor Microenvironment	Number of Test Events	Mean Ratio of Monitor Concentrations (211/Armory)	Mean Difference between Monitor Concentrations (205 − 211), ppb
201	Away from road (not listed below)	14	0.9697	1.0
202	Basketball court in park	1	0.9667	0.6
203	Bus stop	2	0.8009	1.0
204	Car storage yard	1	1.1095	−1.0
205	Driveway	2	0.9142	0.8
206	Near parking lot	1	0.8867	1.9
207	Near road	46	0.9047	1.0
208	Near road and driveway	2	0.6806	1.7
209	Park	5	0.9177	0.5
210	Parking deck	1	0.7804	0.1
211	Parking lot	14	0.9226	0.9
212	Parking lot near fast-food drive-through	4	0.9227	0.6
213	Playground	1	0.8405	0.8
214	Residence	4	1.2275	−0.3
215	Residence near pool	1	0.9650	−3.3
216	Shopping mall	1	1.1200	0
217	Shopping mall near parking	1	1.0410	−0.1
218	Sidewalk near parking	4	0.8691	1.0
	All	104	0.9288	0.7

The 18 outdoor microenvironmental means for D _205−211 in Table 2 range from −3.3 ppb (residence yard near a swimming pool) to 1.9 ppb (near parking lot). The mean D _205−211 for the 104 outdoor events is 0.7 ppb, with a range of −3.3 to 3.4 ppb.

Table 3 provides results for 14 vehicle (in-car) microenvironments. The means for R _211/Armory range from 0.1799 (driving—arterial roads) to 0.6780 (refueling—convenience store). The mean R _211/Armory for all 27 vehicle events is 0.3250.

Table 3. Descriptive statistics for ozone concentrations associated with vehicle (in-car) microenvironments

ME No.	Vehicle Microenvironments	Number of Test Events	Mean Ratio of Monitor Concentrations (211/Armory)	Mean Difference between Monitor Concentrations (205 − 211), ppb
301	Driving—arterial roads	3	0.1799	2.0
302	Driving—commercial district	3	0.2833	1.0
303	Driving—fast-food drive-through	1	0.6561	1.8
304	Driving—Interstate 85	3	0.2500	1.2
305	Driving—residential streets	3	0.2172	−0.3
306	Parked—lot, convenience store	2	0.3356	3.9
307	Parked—lot, city park	2	0.4180	1.5
308	Parked—lot, near roadway	1	0.0063	0.8
309	Parked—lot, office building	2	0.2483	1.7
310	Parked—shopping mall deck	3	0.2536	0.6
311	Parked—residence	1	0.6966	0.2
312	Refueling—convenience store	1	0.6780	1.4
313	Roadside—arterial road	1	0.6170	1.8
314	Roadside—Interstate 85	1	0.5655	0.6
	All	27	0.3250	1.3

The 18 means for D _205−211 in Table 3 range from −0.3 ppb (ME no. 305: driving—residential streets) to 3.9 ppb (ME no. 306: parked—convenience store lot). The 3.9 ppb value was measured in the same convenience store parking lot where the car had been recently refueled (ME no. 312). The mean value of D _205−211 for the 27 vehicle events is 1.3 ppb.

Mean values for open and closed indoor microenvironments

Table 4 is an expanded version of Table 1 in which the indoor microenvironments are further characterized by window and door status (open/closed). The table lists 28 microenvironmental categories—21 “closed” categories containing 35 test events and seven “open” categories containing 10 test events. The mean values of R _211/Armory for the 21 closed categories range from −0.0121 to 0.3000; the mean for the 35 test events in these 21 categories is 0.0932. The mean values of R_211/Armory for the seven open categories range from 0.0978 to 0.7846; the mean for the associated 10 test events is 0.4369.

Table 4. Comparison of descriptive statistics for ozone concentrations associated with open and closed indoor microenvironments

ME No.	Indoor Microenvironment	Open Door or Window?	Number of Test Events	Mean Ratio of Monitor Concentrations (211/Armory)	Mean Difference between Monitor Concentrations (205 − 211), ppb
101	Art gallery and frame shop	No	1	0.0063	11.6
102	Beauty salon	No	1	0.0000	3.7
103	Bicycle shop	Yes	1	0.1281	2.8
104	Bookstore	No	2	0.0115	4.1
105	Cafeteria	No	1	0.2561	1.1
106	Car repair garage	Yes	1	0.6857	−0.5
107	Car repair office	No	1	0.0619	0.6
108	Clothes and novelties store	No	1	0.0241	4.5
109	Dry cleaner	No	1	0.1978	2.5
110	Family restaurant	No	1	0.3000	2.5
111	Fast-food restaurant	No	5	0.1385	2.6
112	Garden supply store	Yes	1	0.4250	1.3
113	Smoking supplies shop	No	1	0.1294	3.6
114	Hospital	No	1	0.1618	0
115	Ice cream restaurant	No	2	0.0845	4.7
116	Laundromat	Yes	1	0.2098	1.8
117	Museum	No	1	0.0045	4.0
118	Office building	Yes	1	0.0978	4.9
		No	3	0.0586	1.9
119	Park restroom	Yes	3	0.7846	0.3
120	Pizza restaurant	No	2	0.0281	4.1
121	Recreation center	No	1	0.2186	2.7
122	Residence	Yes	2	0.2342	3.5
		No	2	0.0025	7.9
123	Shopping mall food court	No	3	0.1637	2.0
124	Shopping mall pedestrian	No	2	0.1176	1.9
125	Toy store	No	1	−0.0121	12.3
126	Violin shop	No	2	0.0320	118.8
—	All	Yes	10	0.4369	1.8
		No	35	0.0932	10.2
		—	45	0.1695	8.3
—	All (omitting violin shop)	Yes	10	0.4369	1.8
		No	33	0.0969	3.6
		—	43	0.1759	3.2

The D _205−211 mean for the (closed) violin shop is 118.8 ppb. The mean values of D _205−211 for the other 20 closed categories range from 0 to 12.3 ppb. The mean for the 35 closed events in these 21 categories is 10.2 ppb. Omitting the two violin shop events, we obtain 3.6 ppb as the mean D _205−211 for the remaining 33 closed events. The mean values of D _205−211 for the seven open categories range from −0.5 to 4.9 ppb; the mean D _205−211 for the associated 10 test events is 1.8 ppb.

Mean values by road orientation for outdoor microenvironments

Table 5 lists six “near-road” outdoor microenvironments for which the technician reported his orientation to the nearest roadway (downwind, upwind, or calm). The resulting nine microenvironmental categories include 49 of the 104 outdoor test events.

Table 5. Descriptive statistics for ozone concentrations associated with outdoor microenvironments with respect to road orientation

ME No.	Outdoor Microenvironment	Orientation of Monitor to Nearest Roadway	Number of Test Events	Mean Ratio of Monitor Concentrations (211/Armory)	Mean Difference between Monitor Concentrations (205 − 211), ppb
203	Bus stop	Downwind	2	0.8009	1.0
205	Driveway	Upwind	1	0.9442	1.2
207	Near road	Calm	13	0.9724	1.2
		Downwind	22	0.8760	0.9
		Upwind	7	0.8931	0.4
		Upwind ^a	1	1.0526	1.4
		—	43	0.9120	0.9
208	Near road and driveway	Calm	1	0.6806	1.7
212	Parking lot near fast-food drive-through	Downwind	1	0.9816	0.3
217	Shopping mall near parking	Downwind	1	1.0410	−0.1
—	All	Calm	14	0.9516	1.2
		Downwind	26	0.8806	0.9
		Upwind	8	0.8994	0.5
		Upwind ^a	1	1.0526	1.4
		—	49	0.9075	0.9
Note: ^aWind direction almost parallel to roadway.

The mean values of R _211/Armory for the four downwind microenvironmental categories range from 0.8009 to 1.0410; the mean for the associated 26 test events is 0.8806. The mean values of R _211/Armory for the three upwind categories range from 0.8931 to 1.0526; the mean for the associated nine test events is 0.9164.

One of the nine upwind events is characterized as having wind direction almost parallel to the road. The R _211/Armory value for this event is 1.0526. The mean value of R _211/Armory for the other eight events is 0.8994.

The two categories characterized as “calm” (thus no wind orientation) have mean R _211/Armory values of 0.6806 and 0.9724. The mean of the 14 associated calm test events is 0.9516.

The D _205−211 means for the nine microenvironmental categories range from −0.1 to 1.7 ppb. The mean value for the 49 test events in these nine categories is 0.9 ppb.

Mean values by ventilation and engine status for vehicle microenvironments

Table 6 lists mean values of R _211/Armory and D _205−211 for 27 in-car categories defined by microenvironment and seven combinations of window status, vent status, air conditioning (AC) status, and engine status. The means for the test events belonging to each of the seven window-vent-AC-engine combinations are listed at the bottom of the table.

Table 6. Descriptive statistics by ventilation and engine status for ozone concentrations associated with vehicle microenvironments

ME No.	Vehicle (In-Car) Microenvironment	Window Status	Vent Status ^a	AC Status	Engine Status	Number of Test Events	Mean Ratio of Monitor Concentrations (211/Armory)	Mean Difference between Monitor Concentrations (205 − 211), ppb
301	Driving—arterial roads	Closed	Open	Off	Driving	1	0.1528	1.6
		Closed	Recirc	On	Driving	1	0.0038	1.7
		Open	Open	Off	Driving	1	0.3830	2.6
		—	—	—	—	3	0.1799	2.0
302	Driving—commercial district	Closed	Open	Off	Driving	1	0.2313	1.1
		Closed	Recirc	On	Driving	1	−0.0125	0.9
		Open	Open	Off	Driving	1	0.6312	0.9
		—	—	—	—	3	0.2833	1.0
303	Driving—fast-food drive-through	Open	X	Off	Driving	1	0.6561	1.8
304	Driving—Interstate 85	Closed	Open	Off	Driving	1	0.2310	1.1
		Closed	Recirc	On	Driving	1	0.0224	1.2
		Open	Open	Off	Driving	1	0.4966	1.2
		—	—	—	—	3	0.2500	1.2
305	Driving—residential streets	Closed	Open	Off	Driving	1	0.0414	0.0
		Closed	Recirc	On	Driving	1	0.0310	0.1
		Open	Open	Off	Driving	1	0.5793	−1.0
		—	—	—	—	3	0.2107	−0.3
306	Parked—lot, convenience store	Closed	Recirc	On	Idling	1	0.0068	1.7
		Open	P	Off	Off	1	0.6644	6.1
		—	—	—	—	2	0.3356	3.9
307	Parked—lot, city park	Closed	Open	Off	Off	1	0.0721	2.9
		Open	P	Off	Off	1	0.7639	0.1
		—	—	—	—	2	0.4180	1.5
308	Parked—lot, near roadway	Open	P	Off	Off	1	0.0063	0.8
309	Parked—lot, office building	Closed	Recirc	On	Idling	1	0.0103	1.9
		Open	P	Off	Off	1	0.4863	1.4
		—	—	—	—	2	0.2483	1.7
310	Parked—shopping mall deck	Closed	Open	Off	Off	1	0.0821	0.8
		Closed	Recirc	On	Idling	1	0.0143	1.0
		Open	P	Off	Off	1	0.6643	0.0
		—	—	—	—	3	0.2536	0.6
311	Parked—residence	Open	P	Off	Off	1	0.6966	0.2
312	Refueling—convenience store	Open	P	Off	Off	1	0.6780	1.4
313	Roadside—arterial road	Open	P	Off	Off	1	0.6170	1.8
314	Roadside—Interstate 85	Open	P	Off	Off	1	0.5655	0.6
—	All	Closed	Open	Off	Off	2	0.0771	1.9
		Closed	Open	Off	Driving	4	0.1641	1.0
		Closed	Recirc	On	Driving	4	0.0112	1.0
		Closed	Recirc	On	Idling	3	0.0105	1.5
		Open	Open	Off	Driving	4	0.5225	1.0
		Open	P	Off	Off	9	0.5713	1.4
		Open	X	Off	Driving	1	0.6561	1.8
		—	—	—	—	27	0.3250	1.3
Notes: ^arecirc = recirculation; P = not reported, car parked with windows open; X = not reported, car driving in fast-food drive-through with windows open.

The three window-vent-AC-engine combinations that include open windows have the largest mean values for R _211/Armory, ranging from 0.5225 to 0.6561. The two combinations with the smallest mean values of R _211/Armory—0.0105 and 0.0112—are associated with closed windows, AC on, and vent recirculating.

The largest value of D _205−211 for a single in-car test event—6.1 ppb—occurred when the car was parked with open windows at a convenience store near to and upwind of the gas pumps. The value of D _205−211 during refueling of the car (downwind of the pump) at the same store during the same hour was 1.4 ppb. The technician noted that the windows were cracked rather than completely open during the first few minutes of the refueling test event that produced the 1.4 ppb value. Cars were idling next to the parked car during the test event that produced the 6.1 ppb value. The convenience store also blocked the wind while the car was parked at the store.

Outdoor concentrations measured near the armory fixed-site monitor

On four of the sampling days, the technician visited the fixed-site monitoring station near the Durham National Guard Armory and made outdoor ozone measurements within 3 m of the monitor shelter with its inlet 4 m below the elevated fixed-site monitor inlet. Table 7 lists the ozone concentrations measured by the 205 and 211 monitors, together with the available simultaneous 10-min and hourly-average concentrations measured by the fixed-site ozone monitor. There is relatively good agreement across the three monitors given that there is a difference in sample inlet heights (205/211 vs. Armory monitors). The 205 and 211 monitors agree to within 4 ppb with the hourly-average Armory monitor concentration that would be used in APEX model simulations (since subhourly ambient data are often unavailable) and to within 1 ppb of each other. A comparable but more comprehensive 3-month Houston 2B 211- TE 49C monitor comparison is reported by Ollison et al. (2013).

Table 7. Comparison of ozone concentrations measured simultaneously by the model 205 and model 211 monitors and the fixed-site monitor at the Durham National Guard Armory

		Ozone Concentration (ppb)
Date	Start Time for Measurements by 205 and 211 Monitors	10-Minute Average Value Measured by Model 205 ^a	10-Minute Average Value Measured by Model 211 ^a	10-Minute Average Value Measured by Armory Monitor	Hourly-Average Value Reported by Armory Ozone Monitor
September 11, 2012	1:08 p.m.	45.7	45.6	45.0	45
September 14, 2012	11:39 a.m.	33.0	32.5	29.6	29
September 21, 2012	5:14 p.m.	43.2	42.2	42.4	42
September 27, 2012	4:20 p.m.	55.2	55.0	61.3	58
Note: ^aThe monitoring cart was located in the grassy area to the east of the fixed-site monitoring station near the Durham National Guard Armory.

Results of special tests to identify potential sources of monitor interference

During 2 days of supplemental monitoring, the technician simulated 14 household and in-car activities (cooking, laundering, burning candles, cleaning, etc.) that were expected to emit interferences (VOCs, PM) into the air sampled by the two ozone monitors. Each of these emission tests consisted of a background period (˜5 min of monitoring before the activity began), an emission period (10–40 min of monitoring while the source was emitting), and a postemission period (˜10 min of monitoring after the emissions had ceased). The script also included three “no-emission” test periods to provide an indication of the general background levels in the residence and car.

Eight of the 14 tests produced a pattern in the ozone concentrations measured by the 205 monitor that suggested interference effects (Table 8). There is a significant increase in 1-min ozone levels going from the background period to the emission period, followed by elevated ozone levels during the postemission period. We consider these results to be suggestive but not definitive—additional testing is required to determine the range of interference effects that may occur under typical emission conditions (Seaman et al., 2009).

Table 8. Tests showing likely interference effects in the ozone measurements made by the model 205 monitor relative to the model 211 monitor

			Maximum 1-Minute Ozone Concentration (ppb)
			Background		Emissions		Postemissions
Test Group	Emission Source	Location	205	211	205	211	205	211
2	Grinding spices	Kitchen	2.9	0.7	9.7	0.6	7.8	0.7
4	Steaming rice with spices	Kitchen	4.2	0.6	7.1	1.0	6.5	0.9
6	Air freshener	Car (parked)	−0.1	0.6	26.7	3.5	38.1	4.3
7	Dish washer (no detergent ^a )	Kitchen	2.5	0.0	6.9	0.4	8.4	0.7
8	Shower with coal tar shampoo	Back bathroom	3.8	0.6	9.2	0.7	2.4	0.3
11	Mop floor with Pine-Sol cleaner	Front bathroom	4.6	0.5	73.6	2.0	71.4	2.3
16	Heated cooking oil no. 1	Kitchen	4.8	0.8	7.6	0.7	9.9	0.5
17	Heated cooking oil no. 2	Kitchen	6.3	0.7	9.7	0.7	9.9	0.7
Note: ^aDetergent dispenser failed to operate properly (partial release).

Conclusion

Overall, the mean values of R _211/Armory by microenvironment are consistent with our expectations. The values tend to be lowest for indoor microenvironments (typically below 0.3) and highest for outdoor microenvironments (typically above 0.8). The indoor values tend to be higher when the enclosure is well ventilated (e.g., an unheated park restroom open to outdoor air). The outdoor values tend to be slightly lower when the measurement is made downwind from a roadway, likely due to the scavenging effects of NO emitted by motor vehicle exhaust.

The mean values of R _211/Armory for the in-car microenvironments tend to fall between 0.2 and 0.7—the mean for all 27 in-car tests is 0.325. The ratio tends to be larger when windows are open. The smallest mean values of R _211/Armory are associated with closed windows, AC on, and vent recirculating.

With respect to D _205−211, the two largest indoor values—173.9 and 63.6 ppb—are associated with repeated visits to a particular violin sales and repair shop. These large values are likely the result of interferences to the 205 monitor caused by solvents and other chemicals being used in the shop. The mean value of D _205−211 for the remaining indoor 43 tests is 3.2 ppb, a value over 4 times the mean value (0.7 ppb) of the 104 outdoor values of D _205−211.

Proximity to roadways tends to increase the value of D _205−211 for outdoor locations. The outdoor tests near roadways have a mean D _205−211 value of 0.9 ppb. The mean value for the other 55 outdoor values is 0.5 ppb.

The mean value of D _205−211 for the 27 vehicle (in-car) events is 1.3 ppb, a value that falls between the means for the indoor and outdoor tests. The largest in-vehicle values (6.1 and 1.4 ppb) are associated with parking and refueling the car at a convenience store.

In general, values of D _205−211 are relatively small, with 166 (94%) of the 176 values falling below 5 ppb. Only five of the 176 D _205−211 values are above 10 ppb. These values are listed below.

a.	173.9 ppb Indoors—violin shop
b.	65.6 ppb Indoors—violin shop
c.	14.1 ppb Indoors—residence
d.	12.3 ppb Indoors—toy store
e.	11.6 ppb Indoors—art gallery and frame shop

Note that all five of these values were measured at indoor locations. In supplemental testing of 14 possible in-home and in-car interference sources, we identified eight that produced high ozone readings on the 205 monitor relative to the 211 monitor (Table 8).

Overall, the results of our study indicate that interferences associated with indoor and mobile sources can bias the conventional 205 photometer high. Consequently, regulatory agencies such as the EPA should be careful about using previous microenvironmental ozone data measured by the 205 (and similar interference-prone conventional photometers) to estimate ME-to-ambient ratios in exposure simulations, particularly if there is reason to suspect the presence of interferences at the sampled locations. In our study, the mean value for the ME-to-Armory ozone ratio based on the 33 closed indoor test events (omitting the violin shop outliers) is 0.1964 for the model 205 and 0.0969 for the model 211. If a point estimate of 0.1964 were used to estimate the ME-to-ambient ratio for closed indoor microenvironments in an APEX-like exposure simulation, the resulting ozone dose estimates in these microenvironments would be more than double the estimates obtained using a point estimate of 0.0969. In a similar manner, the point estimate for the ME-to-ambient ratio for open indoor microenvironments would be 0.5312 based on the model 205 ozone measurements (n = 10), a value 21% higher than the point estimate of 0.4369 obtained from the corresponding model 211 ozone measurements. Likewise, the point estimate for the ME-to-ambient ratio for closed in-vehicle microenvironments would be 0.0968 based on the model 205 measurements (n = 13), approximately 40% higher than the point estimate of 0.0682 obtained from the corresponding model 211 measurements. The newly developed 211 monitor would be a better choice for monitoring such locations in the future, since it seems to be relatively unaffected by the kinds of interferences we encountered in our study.

The 205 monitor may be affected (biased high) by outdoor sources in certain situations that were not included in our study, such as stagnant air conditions often associated with high ozone levels. It should be noted that we monitored towards the end of the ozone season when levels of ambient ozone (and its precursor aromatics) were moderate. The highest hourly-average ozone concentration reported by the Armory monitor during the five monitoring days was 61 ppb (September 27). Hourly-average ozone concentrations did not exceed 48 ppb on the other 4 days.

The data analyses included in this article focus on the relationship between the ozone measured in particular microenvironments and the ozone measured simultaneously by the fixed-site monitor at the Durham National Guard Armory. In future work, we intend to use the Durham data to characterize the relationships between ozone concentrations measured in different microenvironments at the same geographic location at approximately the same time. For example, the data include many grouped measurements where ozone was measured (1) inside a building, (2) outdoors at the front door of the building, (3) outdoors at the edge of the road nearest the building, and (4) outdoors on the other side of the road—all within the same 1-hr period. This grouping provides a means to estimate indoor/outdoor ratios and to evaluate the effects of mobile source emissions (upwind and downwind) on outdoor ozone concentrations.

If the results of our study are used to evaluate APEX and similar population exposure models, researchers should use the 211 measurements to estimate ozone microenvironmental factors rather than the 205 measurements. Ideally, larger microenvironmental studies using the 211 monitor will be performed in the future to increase the database available for these evaluations. It would also be useful to conduct additional studies comparing the ozone concentrations measured simultaneously by the 205 and 211 monitors under a variety of conditions, including the presence of possible test interferences. Such studies may provide correction factors that could be used to adjust ozone measurements measured by the 205 and similar monitors in past microenvironmental studies.

Acknowledgment

The authors would like to thank the North Carolina Division of Air Quality for providing data reported by the ozone monitor located at the Durham National Guard Armory.

Funding

Funding for this study was provided by the American Petroleum Institute.