Start Submission

Reading: Changes in summertime ozone in Colorado during 2000–2015


A- A+
Alt. Display

Research Article

Changes in summertime ozone in Colorado during 2000–2015


Tanja Bien,

Institute of Arctic and Alpine Research (INSTAAR), University of Colorado, Boulder, Colorado, US
X close

Detlev Helmig

Institute of Arctic and Alpine Research (INSTAAR), University of Colorado, Boulder, Colorado, US
X close


In 2016, the Denver Metro Area (DMA)/Northern Colorado Front Range (NCFR) was reclassified from a Marginal to a Moderate O3 Non-Attainment Area due to the prevalence of high summer ozone (O3) occurrences. Hourly surface O3 data collected during 2000–2015 from a total of 80 monitoring sites in the State of Colorado were investigated for geographical features in O3 behavior and O3 changes over time. We particularly focus on summer O3 (June, July, August), which is the time when most exceedances of the O3 National Ambient Air Quality Standard (NAAQS) have been recorded. Variables investigated include the statistical (5th, 50th (median), and 95th percentile) distribution of O3 mixing ratios, diurnal amplitudes, and their trends. Trend analyses were conducted for 20 site records that had at least ten years of data. The majority of Colorado ozone monitoring sites show an increase of the 5th (16 total; 11 of these are statistically significant (p-value ≤ 0.05) trends) and 50th (15 total; 4 statistically significant trends) percentile values. Changes for the 95th percentile values were smaller and less consistent. One site showed a statistically significant declining trend, and one site an increasing trend; the majority of other sites had slightly negative, albeit not statistically significant declining O3. Ozone changes at the two highest elevations sites (>2500 m asl) are all negative, contrasting increasing O3 at U.S. West Coast sites. NCFR urban sites do not show the rate of decreasing higher percentile O3 as seen for the majority of urban areas across the U.S. during the past 1–2 decades. The amplitudes of diurnal O3 cycles were studied as a proxy for nitrogen oxides (NOx) emissions and the diurnal O3 production chemistry. The majority of sites show a decrease in the median summer O3 diurnal amplitude (19 total/10 statistically significant). This is mostly driven by the increase in nighttime O3 minima, which is most likely a sign for a declining rate of nighttime O3 loss from titration with nitric oxide (NO), indicating a change in O3 behavior from declining NOx emissions. Since median and upper percentile surface O3 values in the DMA have not declined at the rates seen in other western U.S. regions, thus far the reduction in NOx has had a more pronounced effect on the lower percentile O3 distribution than on high O3 occurrences that primarily determine air quality. An assessment of the influence of oil and gas emissions on Colorado, and in particular DMA O3, is hampered by the sparsity of monitoring within oil and gas basins. Continuous, long-term, high quality, and co-located O3, NOx, and VOC monitoring are recommended for elucidating the geographical heterogeneity of O3 precursors, their changing emissions, and for evaluation of the effectiveness of O3 air quality regulations.
Knowledge Domain: Atmospheric Science
How to Cite: Bien, T. and Helmig, D., 2018. Changes in summertime ozone in Colorado during 2000–2015. Elem Sci Anth, 6(1), p.55. DOI:
 Published on 27 Jul 2018
 Accepted on 12 May 2018            Submitted on 07 Jun 2017
Managing Editor-in-Chief: Michael E. Chang; Brook Byers Institute for Sustainable Systems, Georgia Institute of Technology, US
Guest Editor: Schwietzke Stefan; University of Colorado Boulder, US

1. Introduction

Surface ozone (O3), first identified in the 1940s and 1950s as an air pollutant that adversely impacts vegetation, human health, and crop yield (National Research Council, 1991; Lefohn et al., 1997; The Royal Society, 2008; Jerrett et al., 2009), is a widely recognized air quality problem in many regions around the world (Dentener et al., 2010). The globally averaged O3 tropospheric lifetime is approximately 23 days (Young et al., 2013), with the O3 lifetime being shorter near the surface and in urban areas because of deposition and chemical destruction. The resulting global tropospheric O3 distribution is highly variable by season, location, and altitude (Cooper et al., 2014). Elevated surface O3 and exceedances of the O3 National Ambient Air Quality Standard (NAAQS) are primarily determined by transport of O3 from outside the region, and local and regional production from photochemistry in the lower troposphere. Emissions and resulting concentrations of the O3 precursor gases nitrogen oxids (NOx) and volatile organic compounds (VOC), play a key role in local O3 production, as well as temperature and sunlight (Haagen-Smit, 1952; McKeen et al., 1991; Ryerson et al., 2001).

In the U.S., compliance with the O3 NAAQS is determined by the ‘Design Value’, which is calculated as the mean of three consecutive years of the 4th highest annual value for the maximum daily average 8-hour O3 value (MDA8) (CFR, 2017). The O3 NAAQS has been progressively lowered in recent years in consideration of the increasingly recognized health effects of prolonged O3 exposure. It was lowered from a value of 84 ppbv to 75 ppbv in 2007, and then further to 70 ppbv in 2015, which is the current standard.

Following an abundance of high O3 observations and exceedance of the then 84 ppbv NAAQS, the Northern Colorado Front Range (NCFR), including the Denver Metropolitan Area (DMA), the State of Colorado, and several regional agencies entered into an Early Action Compact (EAC) with the U.S. Environmental Protection Agency (EPA) in 2002. This agreement was voluntary and gave the opportunity to establish emission inventories, control strategies, and achieve compliance with the NAAQS (CDPHE, 2008a). The DMA and parts of Larimer and Weld County continued exceeding the NAAQS through 2007. In 2012, the DMA/NCFR, including the seven counties of Adams, Arapahoe, Boulder, Broomfield, Denver, Douglas, and Denver, as well as portions of Weld and Larimer counties, was designated as a ‘Marginal’ Non-Attainment Area (NAA) for the 2008 NAAQS. The area has remained in non-attainment since. With the lowering of the NAAQS from 75 to 70 ppb in 2015, the NCFR faced an even greater challenge to comply with the standard. A final rule, published by the EPA in 2016, reclassified the DMA/NCFR area from a ‘Marginal’ to a ‘Moderate’ O3 NAA (CDPHE, 2016).

Ozone chemistry, implementation of O3 control strategies, and compliance with the O3 NAAQS, are complicated and extraordinarily challenging in the NCFR for a number of reasons:

  1. The region has experienced remarkable growth, with an estimated 30% population increase for the NCFR from 2000 to 2015 (Supplemental Materials SM_Table_1). This population growth has been associated with urban sprawl, increase in transportation, power generation, and associated industrial stationary and mobile emission sources.
  2. The NCFR is subject to naturally elevated O3 and downfolding transport events that bring elevated O3 from the mid to upper troposphere to near the surface due to its elevation of 1500–1600 m asl and location on the eastern flanks of the Rocky Mountains. Occasions when such events led to significant enhancements in surface O3 and high MDA8 values have been reported (Langford et al., 2009). Musselman and Korfmacher (2014) showed that even remote high elevation areas of the southern Rocky Mountains can experience exceedance of the NAAQS value, in particular during springtime, when stratospheric intrusions contribute to elevated O3 at these elevations.
  3. Emissions from oil and natural gas (O&NG) industries are another complicating influence. In 2017, there were over 53,000 active O&NG wells in Colorado. The O&NG development is concentrated in a number of lower elevation basins (Figure 1),with 22,000 wells located in Weld County in the Denver Julesburg Basin (DJB), and thereby in the NAA (COGCC, 2014). Several recent studies have investigated the influence of O&NG emissions on Colorado O3. Building on the speciated measurements of O&NG-associated VOC, both Swarthout et al. (2013) and Gilman et al. (2013) found that emissions from the O&NG industry alone contribute to more than half of the local O3 production potential at the Boulder Atmospheric Observatory (BAO) in northern Denver. McDuffie et al. (2016), applying a box model, found that O&NG operations contribute approximately 20% to regional photochemical O3 production in the NCFR. These inferred influences have been confirmed by O3 monitoring results: A study conducted by the CDPHE in 2008, using O3 data from four sites along the NCFR, showed that for elevated O3 events during mid-May to mid-August 2006, air transport from the DJB was associated with the highest O3 values, whereas transport from surrounding areas, including the DMA, brought in air with lower O3 levels (CDPHE, 2008b). Using a correlation analysis of ambient O3 observations and winds, Evans and Helmig (2017) found that for the BAO and a site south of Boulder, during 2009–2012, an average of 65% of elevated O3 events were associated with transport from O&NG production regions. Based on case study comparisons near Greeley during the 2014 summer, Cheadle et al. (2017) estimated that O&NG emissions contribute up to ~20 ppbv to O3 production on high O3 days. Applying modeling of the same year’s data for the wider region, Pfister et al. (2017) concluded that, on average, O&NG emissions contribute to 30–40% of the O3 produced on high O3 days.
  4. During recent years, the region has been subjected to an unusual frequency and extent of nearby wildfires and transported wildfire plumes from outside of the state. These wildfire emissions are contributing to occurrences of elevated O3 production. Given the irregular and sporadic occurrences, thus far there is no quantitative assessment of the wildfire contribution to Colorado O3.
Figure 1 

Location and abundance of oil and natural gas wells in the State of Colorado in 01/01/2017 (data from the Colorado Oil & Gas Conservation Commission, interactive map). Brown dots depict individual well locations. Also shown, by the red lines, are the borders of the DMA/NCFR ozone NAA. Red circles are used as indicators for Mesa Verde National Park, Golden – National Renewable Energy Labs (NREL), and Welby, which are the sites chosen for the data examples shown in this paper. DOI:

We investigated 2000–2015 O3 data from across the State of Colorado for the spatial distribution of O3, elevated O3 occurrences, O3 trends, and if and how measures that have been implemented for curbing O3 pollution are reflected in observational records. Besides investigating absolute O3 mixing ratios and their changes over time, we used two variables to examine changes in the O3 production efficiency. 1. We studied the low distribution of the O3 spectrum as an indicator of the degree of nighttime O3 destruction, which in urban areas is largely determined by emission and the abundance of nitric oxide (NO). Direct and long-term measurements of NOx are sparse, we therefore aim to yield insight into changes in NOx from the nighttime O3 behavior. 2. We determined diurnal amplitudes of the surface layer ambient O3 mixing ratio. The diurnal O3 change is driven by multiple processes, including nighttime O3 destruction by NO, diurnal boundary layer dynamics, and daytime rate of O3 production from photochemistry. We examine this variable to investigate changes over time in the chemical processes determining daytime O3 production and nighttime O3 destruction.

2. Experimental study

2.1. Methods

2.1.1. Location of study and data sources

We were able to retrieve surface O3 data from a total of 79 sites in Colorado for 2000–2015. Data for 70 sites were downloaded from the EPA Air Quality System (AQS) archive in September 2016. AQS includes Colorado O3 data from the Colorado Department of Public Health and Environment (CDPHE), the US Forest Service (USFS), Boulder County Public Health (BCPH), Garfield County, the US National Park Service (NPS), the Southern Ute Indian Tribe Reservation (SUIT), Meteorological Solutions Inc. (MSI), the Colorado Bureau of Land Management (CBLM), the City of Aspen, CH2M Hill, as well as from monitoring by the EPA. Additional data from three sites were obtained from the National Oceanic and Atmospheric Administration (NOAA) Global Monitoring Division (GMD). Data from six sites operated during the summer 2014 Front Range Air Pollution and Photochemistry Experiment (FRAPPE), and from the NSF-funded Long-Term Ecological Research (LTER) Niwot Ridge program, were collected by the Institute of Arctic and Alpine Research (INSTAAR) at the University of Colorado, Boulder. Data for Dinosaur National Monument (DNM) were received from the NPS. NOx data for 12 sites were downloaded from AQS.

Site information is summarized in Table 1, and Figure 2 depicts a map with all site locations. AQS sites are represented by yellow markers, NOAA sites by green markers, INSTAAR measurements by blue markers, and data obtained directly from the NPS in red. The NPS Dinosaur National Monument (DNM) site coordinates given in the data record placed the site within Colorado, but we later discovered that the actual measurement location is ~22 km west in the State of Utah. Nonetheless, we retained the DNM record for our analyses. The location setting column in Table 1 characterizes sites as rural, suburban, and urban. The amount of data varied widely among sites, from ~2 months to a maximum of 15 years; the fraction of data that were available for each site and year is summarized in SM_Table_2.

Table 1

Summary and information on the location, operating agency, and classification of ozone monitoring sites considered in this study. DOI:

Site No Data Source County City Site Name Agency/Program Latitude Longitude Elevation (masl) Location Setting

1 AQS Adams Welby Welby CDPHE 39.84 –104.95 1554 urban
2 AQS Arapahoe Southglenn Highland Reservoir CDPHE 39.57 –104.96 1747 suburban
3 AQS Arapahoe Aurora Aurora East CDPHE 39.64 –104.57 1799 rural
4 AQS Boulder not in a City Eldora Ski Area USFS 39.94 –105.61 3201 rural
5 AQS Boulder Boulder South Boulder Creek CDPHE 39.96 –105.24 1669 suburban
6 AQS Boulder Boulder Boulder Fire Station BCCHD 40.01 –105.25 1662 rural
7 AQS Boulder not in a City Longmont BCCHD 40.15 –105.06 1499 suburban
8 AQS Chaffee not in a City Trout Creek Pass USFS 38.91 –106.00 2920 rural
9 AQS Clear Creek not in a City Geneva Basin USFS 39.58 –105.73 3474 rural
10 AQS Clear Creek not in a City Goliath Peak USFS 39.64 –105.59 3518 rural
11 AQS Clear Creek not in a City Mount Evans USFS 39.59 –105.64 4300 rural
12 AQS Clear Creek not in a City Mines Peak CDPHE 39.79 –105.76 3806 rural
13 AQS Denver Denver Denver – Camp CDPHE 39.75 –104.99 1593 urban
14 AQS Denver Denver Denver – Carriage CDPHE 39.75 –105.03 1621 suburban
15 AQS Denver Denver Denver – Animal Shelter CDPHE 39.70 –105.00 1594 urban
16 AQS Denver not in a City La Casa CDPHE 39.78 –105.01 1602 urban
17 AQS Douglas not in a City Chatfield Reservoir CDPHE 39.54 –105.07 1678 rural
18 AQS Douglas not in a City Chatfield State Park CDPHE 39.53 –105.07 1676 rural
19 AQS El Paso not in a City U.S. Air Force Acad. CDPHE 38.96 –104.82 1971 rural
20 AQS El Paso Manitou Springs Manitou Springs CDPHE 38.85 –104.90 1955 suburban
21 AQS Garfield Rifle Rifle – Health Dept. CDPHE 39.54 –107.78 1640 suburban
22 AQS Garfield not in a City Bell Ranch USFS 39.49 –107.66 1785 rural
23 AQS Garfield not in a City Flattops #3 USFS 39.80 –107.62 2904 rural
24 AQS Garfield not in a City Ripple Creek Pass USFS 40.09 –107.31 2930 rural
25 AQS Garfield not in a City Sunlight Mountain USFS 39.43 –107.38 3224 rural
26 AQS Garfield not in a City Wilson USFS 39.49 –107.17 2358 rural
27 AQS Garfield Battlement Mesa Battlement Mesa Garfield Cty 39.44 –108.03 1690 suburban
28 AQS Garfield Glenwood Springs Glenwood Springs Garfield Cty 39.55 –107.33 1756 suburban
29 AQS Garfield not in a City Carbondale Garfield Cty 39.41 –107.23 1867 suburban
30 AQS Gunnison not in a City McClure Pass USFS 39.09 –107.23 2930 rural
31 AQS Gunnison not in a City Gothic USEPA 38.96 –106.99 2926 rural
32 AQS Jackson not in a City Walden – Chandler Ranch NPS 40.88 –106.31 2417 rural
33 AQS Jefferson Arvada Arvada CDPHE 39.80 –105.10 1640 suburban
34 AQS Jefferson Lakewood Welch CDPHE 39.64 –105.14 1742 suburban
35 AQS Jefferson not in a City Rocky Flats – N CDPHE 39.91 –105.19 1802 rural
36 AQS Jefferson Applewood Golden – NREL CDPHE 39.74 –105.18 1832 suburban
37 AQS Jefferson Golden Lookout Mountain Rd CDPHE 39.73 –105.25 2278 suburban
38 AQS Jefferson Aspen Aspen Park CDPHE 39.54 –105.30 2473 rural
39 AQS La Plata not in a City Shamrock Station USFS 37.30 –107.48 2367 rural
40 AQS La Plata not in a City Ignacio SUIT 37.14 –107.63 1983 rural
41 AQS La Plata not in a City Bondad SUIT 37.10 –107.87 1920 rural
42 AQS Larimer not in a City RMNP – Long’s Peak NPS 40.28 –105.55 2742 rural
43 AQS Larimer Fort Collins Fort Collins – West CDPHE 40.59 –105.14 1571 rural
44 AQS Larimer not in a City Rist Canyon CDPHE 40.64 –105.28 2058 rural
45 AQS Larimer Fort Collins Fort Collins – CSU CDPHE 40.58 –105.08 1524 suburban
46 AQS Larimer not in a City RMNP USEPA 40.28 –105.55 2743 rural
47 AQS Mesa Palisade Palisade CDPHE 39.13 –108.31 1521 rural
48 AQS Mesa not in a City Grand Mesa USFS 38.93 –108.23 3040 rural
49 AQS Mesa not in a City Silt – Collbran USFS 39.34 –107.71 2485 rural
50 AQS Mesa not in a City CO Nat. Mon. NPS 39.11 –108.74 1740 rural
51 AQS Moffat not in a City Lay Peak CDPHE 40.51 –107.89 1902 rural
52 AQS Moffat not in a City Elk Springs CDPHE 40.33 –108.49 1902 rural
53 AQS Montezuma Cortez Cortez – Health Dept. CDPHE 37.35 –108.59 1890 urban
54 AQS Montezuma not in a City Mesa Verde NP NPS 37.20 –108.49 2170 rural
55 AQS Morgan not in a City LS Power – High Plains Energy MSI 40.39 –103.50 1260 rural
56 AQS Park not in a City Kenosha Pass USFS 39.42 –105.76 3098 rural
57 AQS Park not in a City Fairplay BLM 39.24 –105.98 3027 rural
58 AQS Pitkin not in a City Ajax Mountain USFS 39.15 –106.82 3415 rural
59 AQS Pitkin Aspen Aspen City of Aspen 39.20 –106.84 2415 suburban
60 AQS Pueblo Pueblo West Black Hills CH2M Hill 38.35 –104.69 1525 rural
61 AQS Rio Blanco not in a City CO Plant Science Bldg. NPS 40.04 –107.85 1994 rural
62 AQS Rio Blanco Rangely Rangely – Golf Course NPS 40.09 –108.76 1655 rural
63 AQS San Miguel not in a City Miramonte Reservoir USFS 38.01 –108.41 2357 rural
64 AQS San Miguel Norwood Norwood USFS 38.13 –108.29 2137 rural
65 AQS Teller not in a City Manitou Exp. Forest USFS 39.10 –105.09 2360 rural
66 AQS Weld Greeley Greeley – 15th St. CDPHE 40.42 –104.69 1420 urban
67 AQS Weld Greeley Greeley – Weld Cty Twr CDPHE 40.39 –104.74 1484 suburban
68 AQS Weld not in a City Briggsdale USFS 40.65 –104.33 1483 rural
69 AQS Weld not in a City Pawnee Buttes USFS 40.81 –104.04 1661 rural
70 INSTAAR Boulder Boulder Boulder – INSTAAR FRAPPE 40.01 –105.25 1607 urban
71 INSTAAR Boulder not in a City Sugar Loaf Fire Dept. FRAPPE 40.02 –105.36 1982 rural
72 INSTAAR Boulder not in a City Coughlin Meadows FRAPPE 40.00 –105.48 2527 rural
73 INSTAAR Boulder Lyons Lyons FRAPPE 40.20 –105.25 1615 suburban
74 INSTAAR Boulder Lafayette Dawson School FRAPPE 40.06 –105.11 1561 suburban
75 INSTAAR Boulder not in a City Lost Angels Fire Dept. FRAPPE 40.02 –105.40 2393 rural
76 NOAA Boulder Erie Boulder Atmos. Observatory NOAA 40.05 –105.00 1584 rural
77 NOAA Boulder not in a City Niwot Ridge – Tundra NOAA 40.05 –105.59 3538 rural
78 NOAA Boulder not in a City Niwot Ridge – C1 NOAA 40.04 –105.54 3035 rural
79 INSTAAR Boulder not in a City Niwot Ridge – Soddie INSTAAR 40.04 –105.57 3340 rural
80 NPS Moffat not in a City Dinosaur Nat. Monument1 NPS 40.29 –108.94 1463 rural

1 Actual monitor location is in Utah, 22 km west of the state border.

Figure 2 

Location of ozone sites within the State of Colorado that were considered in this study. Site numbers refer to the listing in Table 1. The color of the dots represent the data-source: Yellow – AQS, blue – INSTAAR, green – NOAA, and red – NPS. Some site numbers were shifted slightly to avoid marker overlap. Red circles show sites which are used for the case study in this paper. DOI:

2.1.2. Data processing

Data were downloaded as zip files, converted to .xlsx files, and then imported into Matlab. All data are reported as a molar ratio in air, i.e. units of parts per billion by volume (ppbv [nmol mol–1]). All data were converted to Mountain Standard Time. A few stray data values <–5 ppbv were removed from records. During the review of this manuscript, deviations in the O3 data for the southwestern Colorado sites Bondad and Ignacio were pointed out; after further investigation, the 2000–2004 data for these two sites were excluded from further analyses (SM_Text_1). Data were plotted as time series, and box-whisker plots were prepared when at least 80% of data were available for a given year. For box-whisker plots, the center line represents the data median, and the upper and lower horizontal bars of the box the 25 and 75 percentiles of the data, the whiskers are drawn to a length of 1.5 times the interquartile range away from the top and bottom of the boxes, and values outside the whisker range are marked as red crosses. As we particularly focus on summer O3 (June, July, August), when most exceedances of the O3 NAAQS occur, hourly O3 values from 1st June (12:00 am) until 31st August (11:00 pm) were extracted and used to compute the summer 5th, 50th, and 95th O3 percentile values for each year. When at least 80% of summer data were available and a site had at least 10 years of data, trends were calculated for the 5th, 50th, and 95th percentiles with a linear regression fit. The statistical significance of the linear regression result was evaluated from the p-value, tested against the null hypothesis, R2 = 0, which indicates no linear relationship by using the standard F-statistic. The null hypothesis was rejected with a confidence level ≥95% when the probability P associated with the F statistic p-value was ≤0.05. For the MDA8 and Design Value calculation, the Code of Federal Regulations (2017) protocol was used, with a linear regression line fit through the calculated annual data points for each year. In the maps showing the geographical distribution of trend analyses, slope values are given by the color scale, and the significance of each trend result is represented by the size of the marker: Statistically significant trend results are shown in 4 times larger size than statistically non-significant values.

For calculation of daily amplitudes, the hourly minimum O3 value occurring in the first 12 hours of a day was subtracted from the following daily maximum. Amplitudes were only calculated when at least 20 hourly data points were available for a given day. The seasonal change of O3 diurnal amplitude during the year was calculated from the 50th percentile amplitude for every month when at least 80% of data were available. For each site with at least 10 years of data, box-whisker plots show the year-to-year variability and change of diurnal amplitudes. As for the O3 mixing ratio, we focused again on the summer months’ O3 amplitudes, and computed the 5th, 50th, and 95th percentile values for each summer when at least 80% of data were available. For sites with at least 10 years of data, a linear regression fit was calculated to evaluate trends in the summer amplitude.

Additionally, all summer data that were available for the 5-year period of 2011–2015 (plus three 2007–2008 Boulder County site data sets) were used to calculate the 50th and 95th percentile summer O3 mixing ratio, and the median summer daily amplitude. In the maps displaying these results, the marker area is proportional to the square of the amount of available data.

3. Results

3.1. O3 mixing ratio

3.1.1. O3 summer mixing ratios

The full time series records for all 80 sites are available in SM_Figure_1. 69 data sets have data for 2011–2015. The geographic distribution of the 50th and 95th percentile summer O3 mixing ratios for this time window is visualized in Figure 3. The amount of data considered for each point is reflected by the marker size; the smallest marker, i.e. site No. 52 (Elk Springs) represents 735 hourly data points, and the largest marker, site No. 47 (Palisade), had 10977 hourly data points. Overall, summer median O3 results span a range from 32–62 ppb across these Colorado sites (Table 2). Within the DMA, relatively low median mixing ratios of ~35 ppbv were found, whereas relatively higher values were calculated for suburban Denver (43–47 ppbv). The highest summer median O3 mixing ratios were seen at the high elevation mountain sites, with the highest elevation site, Mount Evans (4300 m), having the highest value of 62 ppbv, ~30 ppbv above those for the DMA sites.

Figure 3 

95th (top) and 50th percentile (bottom) hourly summer ozone of all data available for 2011–2015 (please note that for Longmont (site 7) and Lyons (site 73) data are summer 2007 values). The marker size is a function of the amount of data which was used for the plot (see SM_Table_2). Please note the different color bar scales in the two graphs. DOI:

Table 2

Overview of the median and 95th percentile summer O3, and the daily median and 95th percentile summer O3 amplitude of every site with data available for 2011–2015. DOI:

Site No Site Name Median Summer O3 [ppbv] 95th Percentile Summer O3 [ppbv] Median Daily Summer Amplitude [ppbv] 95th Percentile Daily Summer Amplitude [ppbv]

1 Welby 37 69 60 79
2 Highland Reservoir 49 73 40 65
3 Aurora East 49 68 30 47
4 Eldora Ski Area 58 76 14 36
5 South Boulder Creek 43 71 42 67
6 Boulder Fire Station 32 57 42 62
7 Longmont 36 72 61 81
8 Trout Creek Pass 50 65 18 33
10 Goliath Peak 54 70 16 38
11 Mount Evans 62 77 13 32
12 Mines Peak 49 65 11 29
13 Denver – Camp 35 62 48 68
14 Denver – Carriage 41 72 59 81
15 Denver – Animal Shelter 38 67 52 76
16 La Casa 35 66 54 73
18 Chatfield Reservoir 46 75 45 72
19 U.S. Air Force Acad. 45 68 50 68
20 Manitou Springs 47 68 29 47
21 Rifle – Health Dept. 35 59 42 55
23 Flattops #3 52 64 11 22
24 Ripple Creek Pass 51 64 10 23
25 Sunlight Mountain 58 73 12 25
26 Wilson 50 65 22 38
27 Battlement Mesa 43 62 34 44
28 Glenwood Springs 33 55 36 48
29 Carbondale 32 54 40 52
30 McClure Pass 48 60 14 24
31 Gothic 41 59 28 41
32 Walden – Chandler Ranch 37 57 41 54
33 Arvada 40 74 61 81
34 Welch 44 71 42 64
35 Rocky Flats 50 76 35 60
36 Golden – NREL 48 75 38 67
38 Aspen Park 46 67 31 53
39 Shamrock Station 48 65 29 44
40 Ignacio 37 64 42 57
41 Bondad 39 64 41 56
42 RMNP – Longs Peak 51 70 23 48
43 Fort Collins – West 45 73 42 64
44 Rist Canyon 46 68 32 53
45 Fort Collins – CSU 37 66 46 70
46 RMNP – Collocated 49 68 23 47
47 Palisade 45 62 30 44
48 Grand Mesa 53 64 11 24
49 Silt – Collbran 50 64 17 29
50 CO Nat. Mon. 49 64 26 36
51 Lay Peak 44 61 32 44
52 Elk Springs 41 56 20 33
53 Cortez 40 61 37 51
54 Mesa Verde NP 50 64 16 29
56 Kenosha Pass 49 63 15 33
57 Fairplay 43 60 21 36
58 Ajax Mountain 54 66 8 21
59 Aspen 40 58 28 45
61 CO Plant Science Bldg. 41 58 27 39
62 Rangely 44 61 29 43
64 Norwood 44 62 29 44
67 Greeley – Weld Cty Twr 43 71 51 75
68 Briggsdale 41 64 46 64
69 Pawnee Buttes 47 65 27 43
70 Boulder – INSTAAR 39 65 41 65
71 Sugar Loaf Fire Dept. 40 65 35 57
72 Coughlin Meadows 45 64 27 51
73 Lyons 45 71 15 39
74 Dawson School 40 67 42 60
75 Lost Angels Fire Dept. 46 66 26 48
76 Boulder Atmos. Obs. 41 68 43 65
77 Niwot Ridge – Tundra 59 74 14 31
78 Niwot Ridge – C1 48 66 23 49
79 Niwot Ridge – Soddie 47 62 19 38
80 Dinosaur Nat. Mon. 45 64 39 55

A different spatial pattern was found for the 95th percentile summer O3 mixing ratios, where the the high elevation locations on the eastern slopes and the greater DMA sites exhibited the highest values. Most of the DMA sites in the blue enlargement box have values greater than 67 ppbv. 95th percentile values were on the order of 20 ppbv or less above the 50th percentile vales for many of the rural low elevation, and for high elevation sites. In contrast, a more dynamic O3 range was found for sites located in the NCFR. These sites all show a wider distribution, with 95th percentile values exceeding the median by ~30 ppbv or more. The relationship between the 50th and 95th percentile mixing ratios and the site altitude is shown in Figure 4. There is a statistically significant increase of the summer O3 50th percentile value with altitude of 6.5 ± 1.6 ppbv km–1 (p-value < 0.05). In contrast, the 95th percentile O3 values are much more variable and do not show a statistically significant dependence on elevation.

Figure 4 

Dependence and correlation of the 2011–2015 50th and 95th percentile summer ozone on site altitude. The insets shows the linear regression fit results to the data (slope values are given in ppbv km–1), and the dotted lines show the 95% confidence bounds of the regression fit. DOI:

3.1.2. O3 hourly time series data

Three sites with different characteristics were chosen for illustrating the data records and emphasizing main features; their results are displayed in Figure 5: 1. Mesa Verde National Park (MVNP), in the southwestern part of Colorado, representing a rural reference site, 2. Welby (WEL), a suburban site located northeast of Denver, at the southern edge of the DJB, and 3. the Golden – National Renewable Energy Laboratory (NREL) site, close to and west of Denver, in a densely populated area within the NCFR O3 NAA.

Figure 5 

Time series examples for the full record of 1-hour ozone at the rural background site Mesa Verde National Park (top), the urban Golden NREL (middle), and Welby (bottom). Statistical analyses of these data with box-whisker plots (see data processing section for box-whisker format explanation), representing each year of data, are shown to the right. DOI:

Of the sites shown here, MVNP is the one with the lowest total O3 variability over the year, indicated by the lowest spread of the data in the summary box-whisker plots included as the right series of graphs in Figure 5. Ozone mixing ratios at MVNP oscillate around a value of 40 ppbv in winter, and ~55 ppbv during summer; summer maxima are ~70 ppbv. A strong seasonal variability can be seen in the NREL data. Here, winter values are between 0–40 ppbv. NREL has the highest summer values of these three example sites; 1-hour O3 maxima in the summer exceeded 100 ppbv on a total of 69 occasions. WEL O3 values are not quite as high as for NREL, with all 1-hour values remaining below 100 ppbv. WEL is characterized by a high abundance of low winter values, when O3 frequently drops to near 0 ppbv. WEL shows the most evident increasing trend in the data, most notable in the first part of the record.

3.1.3. Summer O3 trends

Summer O3 trends were investigated by conducting a linear regression fit to the entire summer data record. Results for the three sample sites are shown in Figure 6; graphs for all other sites are available in SM_Figure_2.

Figure 6 

Box-whisker plots of the hourly summer ozone distribution. Linear regression fits trough the 5th, 50th and 95th percentile of the data for each year show the trend behavior, with slope results in the legend in ppbv yr–1. DOI:

MVNP shows signs for decreasing O3 at all percentile values, with the 95th percentile value of –0.35 ± 0.34 ppbv yr–1 being a statistically significant negative trend. The O3 mixing ratios at NREL show a steeply increasing trend of 0.55 ± 0.44 ppbv yr–1 for the 5th percentile values. Changes in the median values (increasing) and 95th percentile values (decreasing) are not statistically significant. The WEL results are characterized by statistically significant increasing trends for the 5th, 50th, and 95th percentiles. Details of the trend analyses results for sites that had long enough records to fulfill the selection criteria are listed in Table 3. Maps showing the geographical distribution of trend results for the 5th, 50th, and 95th values are given in Figure 7. In Table 3 we also include trends for the annual 4th highest MDA8 and Design Value, (time series graphs for the Design Value and MDA8 trend determinations are shown in SM_Figure_3) for comparison of those regulatory metrics with the 95th percentile results that are presented throughout this paper. Given the longer data coverage needed for calculation of the Design Value (i.e. a three-year running mean through continuous data), this calculation was only possible for a subset of the site records.

Table 3

Linear regression results for the trend analyses of summer O3 for sites with at least 10 years and 80% of data for a given summer, categorized by the 5th, 50th, and 95th percentile levels. Statistically significant trends (p-value ≤ 0.05) are highlighted in bold font. Additionally, the linear regression results for the trend analyses of the 4th highest MDA8 and the Design Value are listed. DOI:

Site No Site Name 5th Percentile 50th Percentile 95th Percentile 4th highest MDA8 Design Value

Slope ± 95% confidence limit, ppbv yr–1 Slope ± 95% confidence limit, ppbv yr–1 Slope ± 95% confidence limit, ppbv yr–1 Slope ± 95% confidence limit, ppbv yr–1 Slope ± 95% confidence limit, ppbv yr–1

1 Welby 0.20 ± 0.13 0.99 ± 0.46 0.55 ± 0.48 0.61 ± 0.46 0.68 ± 0.51
2 Highland Reservoir 0.88 ± 0.45 0.29 ± 0.45 –0.18 ± 0.57 –0.20 ± 0.95
5 South Boulder Creek 0.39 ± 0.31 0.16 ± 0.44 0.00 ± 0.50 –0.04 ± 0.57 –0.10 ± 0.69
14 Denver – Carriage 0.66 ± 0.23 1.01 ± 0.61 –0.01 ± 0.72 –0.09 ± 0.91 –0.44 ± 0.85
18 Chatfield State Park 0.24 ± 0.59 0.05 ± 0.79 –0.21 ± 0.87 –0.34 ± 1.03
19 U.S. Air Force Acad. 0.60 ± 0.25 0.35 ± 0.47 –0.30 ± 0.49 –0.30 ± 0.52 –0.27 ± 0.67
20 Manitou Springs 0.87 ± 0.80 0.37 ± 0.93 –0.16 ± 0.98 –0.81 ± 0.94
33 Arvada 0.78 ± 0.30 0.85 ± 0.69 0.26 ± 0.85 0.08 ± 0.99 –0.02 ± 0.94
34 Welch 0.94 ± 0.49 0.59 ± 0.56 0.55 ± 0.72 0.55 ± 0.70 0.62 ± 0.70
35 Rocky Flats – N 0.44 ± 0.41 0.07 ± 0.42 –0.31 ± 0.54 –0.31 ± 0.65 –0.33 ± 0.66
36 Golden – NREL 0.56 ± 0.44 0.06 ± 0.38 –0.28 ± 0.51 –0.30 ± 0.71 –0.33 ± 0.83
39 Shamrock Station –0.10 ± 0.74 –0.31 ± 0.70 –0.39 ± 0.82 –0.55 ± 0.75
41 Bondad 0.04 ± 0.42 0.00 ± 0.99 –0.21 ± 0.45 0.03 ± 0.49
42 RMNP – Long’s Peak 0.01 ± 0.43 –0.13 ± 0.57 –0.40 ± 0.64 –0.50 ± 0.59 –0.56 ± 0.56
43 Fort Collins – West –0.36 ± 0.64 –0.56 ± 0.80 –0.64 ± 0.83
45 Fort Collins – CSU 0.49 ± 0.22 0.33 ± 0.37 –0.13 ± 0.42 0.01 ± 0.54 –0.03 ± 0.74
54 Mesa Verde NP –0.16 ± 0.33 –0.31 ± 0.32 –0.35 ± 0.34 –0.23 ± 0.35 –0.19 ± 0.51
67 Greeley – Weld Cty Twr 0.12 ± 0.26 0.01 ± 0.36 –0.37 ± 0.50 –0.41 ± 0.80 –0.30 ± 0.89
78 Niwot Ridge – C1 –0.56 ± 0.53 –0.53 ± 0.59 –0.55 ± 0.79 –0.58 ± 0.68
80 Dinosaur Nat. Mon. 0.47 ± 0.68 0.01 ± 0.52 0.03 ± 0.83
Figure 7 

Ozone changes (in ppbv yr–1) for the 95th (top), 50th (middle) and 5th percentile (bottom) values of summer ozone. All sites with at least 10 years of summer data are shown. The colors of the markers represent the value of the slope. Statistically significant trend values (p-value ≤ 0.05) are plotted four times larger than statistically non-significant results. DOI:

For the 5th percentile values, most sites (16 out of 20) show increasing O3 values, with 11 of these passing the statistical significance test. Five of these results are >0.63 ppbv yr–1, equivalent to a remarkable >10 ppbv increase in the 5th percentile O3 value over the 16-year record. For the 50th percentile, 15 sites show increasing O3 results, 4 being statistically significant. For the 95th percentile values most (15) sites show a decrease in O3, with one of these being statistically significant; four rates of change are positive, with one site (WEL) being a statistically significant positive trend. The increasing trend results in the 95th percentile value at WEL are consistent with trend results for the annual 4th highest MDA8 and Design Value. WEL is the only site that showed statistically significant increasing O3 trends at all (5th/50th/95th) percentile levels. The geographical distribution of these trend analyses (Figure 7) shows that the increase in the low percentile O3 values is prominent and consistent at the urban and NCFR sites. Increases in median O3 are most pronounced in the DMA.

3.2. O3 Daily summer amplitudes

3.2.1. Seasonal cycle of O3 diurnal amplitudes

Time series results for diurnal O3 amplitudes for the full data records are presented in SM_Figure_4. The full record of O3 diurnal amplitudes for the case study sites is shown in Figure 8. The same data, binned by month for illustrating the mean seasonal cycle, are shown in Figure 9 (see SM_Figure_5 for results for other sites). At all sites, diurnal amplitudes are highest during the summer months. Absolute values differ largely between sites, with MVNP having the lowest values, ranging from (median) i.e. ~10 ppbv in winter to ~20 ppbv in summer. These are on the order of 30–50% of those observed at NREL. WEL has an even larger seasonal cycle than NREL, and interestingly, the seasonal cycle shows a wider spread summer maximum.

Figure 8 

Left column: Ozone time series of the daily amplitude. Right column: Box-whisker plots of the statistical distribution of the summer amplitude data for every year that had at least 80% of data available for a given year. Linear regression fits trough the 5th, 50th and 95th percentile of the data for each year show the trend behaviour. Regression analysis slope results (in ppbv yr–1) are given in the legends. DOI:

Figure 9 

Seasonal change of the O3 diurnal amplitude at Mesa Verde National Park, Golden-NREL, and Welby. DOI:

3.2.2. Daily summer amplitude for 2011–2015

The same sites for which the 50th and 95th percentile summer O3 mixing ratios were determined (Figure 3) were subjected to a statistical analysis of the summer diurnal O3 amplitude: Tabulated results were included in Table 2, and the geographical distribution is shown in Figure 10. Urban areas consistently show higher daily amplitudes than rural areas. Median DMA amplitude values are on the order of 55 ppbv. Values in suburban Denver are somewhat lower, on the order of 35–45 ppbv. Within the inner city, on high O3 production summer days, the diurnal O3 increase is on the order of 60–80 ppbv. Rural and high elevation sites show significantly lower diurnal O3 changes, with median daily amplitude values of 10–15 ppbv for the most remote locations. There are likely two contributing processes determining the larger urban amplitudes. Some portion is due to the low nighttime values from titration with NO, and mixing of air from aloft with higher O3 during the breakup of the nocturnal surface layer. The remainder is due to daytime photochemical production. The linear regression analyses between amplitude and site elevation yielded statistically significant negative slope relationships for both the 50th and 95th percentiles, as shown in Figure 11a, b. The O3 diurnal amplitude is inversely correlated with median O3 (Figure 11c).

Figure 10 

95th (top) and 50th percentile (bottom) summer O3 daily amplitude of all data available for 2011–2015 (please note that for Longmont (site 7) and Lyons (site 73) data are summer 2007 values). The marker size is a function of the amount of data which was used for the plot (see SM_Table_2). Please note the different color bar scales in the two graphs. DOI:

Figure 11 

Dependence and correlation of the 2011–2015 95th(a) and 50th(b) percentile daily amplitude of summer O3 on the site altitude. The insets show the linear regression fit results to the data (slope values are given in units of ppbv km–1), and the dotted lines show the 95% confidence bounds of the regression fit. (c) Dependence of the summer median daily amplitude on the median summer O3 mixing ratio. The inset shows the linear regression fit results to the data, and the dotted lines show the 95% confidence bounds of the regression fit. DOI:

3.2.3. Summer daily amplitude trend analyses

Trend analyses on the diurnal amplitude data were prepared to investigate changes in the summertime dynamic O3 behavior. Results for the case study sites are shown in the right side set of graphs in Figure 8, results for all other sites are presented in SM_Figure_6. At MVNP, summer diurnal amplitudes have been moving towards slightly lower values over the study period. While slope values are small, the median and 95th percentile results (–0.16 ± 0.14/–0.37 ± 0.25 ppbv yr–1) are statistical significant changes. At NREL, there has been a downward trend of dirnal amplitudes, with statistical significant decreases of the median and 95th percentile slope values of –0.67 ± 0.37 and 0.94 ± 0.54 ppbv yr–1, respectively. Results from WEL contrast the other two sites. Here, amplitudes have been increasing at all chosen percentile values; but none of these changes are statistically significant.

The results for all datasets are listed in Table 4. The geographical distribution of the slopes is shown in Figure 12. Notably, the majority of sites exhibit declining amplitudes, with those being reflected at all percentile values. Ten of the negative median slope values passed the statistical significance test. Negative values were determined for most of the sites in the DMA and NCFR. WEL is the only DMA site that shows positive slopes at all percentile values.

Table 4

Linear regression results of the trend analyses of the daily summer O3 amplitudes derived from the yearly summer data, broken up the 5th, 50th, and 95th percentile levels. Statistically significant trends (p-value ≤ 0.05) are highlighted in bold font. DOI:

Site No Site Name 5th Percentile 50th Percentile 95th Percentile

Slope ± 95% confidence limit, ppbv yr–1 Slope ± 95% confidence limit, ppbv yr–1 Slope ± 95% confidence limit, ppbv yr–1

1 Welby 0.55 ± 0.83 0.30 ± 0.50 0.38 ± 0.62
2 Highland Reservoir –0.27 ± 0.73 –0.70 ± 0.35 –0.66 ± 0.59
5 South Boulder Creek 0.09 ± 0.53 –0.19 ± 0.38 –0.06 ± 0.62
14 Denver – Carriage –0.16 ± 1.27 –0.68 ± 0.66 –0.71 ± 1.07
18 Chatfield State Park 0.21 ± 0.74 –0.37 ± 0.62 –0.23 ± 0.58
19 U.S. Air Force Acad. –0.72 ± 0.69 –0.86 ± 0.35 –0.76 ± 0.47
20 Manitou Springs –0.34 ± 0.48 –0.68 ± 0.34 –0.91 ± 0.77
33 Arvada 0.19 ± 1.35 –0.31 ± 0.79 –1.05 ± 1.23
34 Welch 0.27 ± 0.55 –0.21 ± 0.40 –0.30 ± 0.69
35 Rocky Flats – N 0.01 ± 0.37 –0.42 ± 0.31 –0.69 ± 0.48
36 Golden – NREL –0.03 ± 0.41 –0.67 ± 0.37 –0.94 ± 0.54
39 Shamrock Station –0.22 ± 0.58 –0.49 ± 0.38 –0.14 ± 0.38
41 Bondad –0.33 ± 0.58 –0.30 ± 0.23 –0.43 ± 0.76
42 RMNP – Long’s Peak –0.09 ± 0.29 –0.24 ± 0.33 –0.07 ± 0.48
43 Fort Collins – West 0.11 ± 0.55 –0.20 ± 0.62 0.02 ± 1.27
45 Fort Collins – CSU 0.13 ± 0.68 –0.41 ± 0.33 –0.44 ± 0.67
54 Mesa Verde NP –0.05 ± 0.20 –0.16 ± 0.14 –0.37 ± 0.25
67 Greeley – Weld Cty Twr 0.23 ± 0.69 –0.56 ± 0.53 –0.69 ± 0.89
78 Niwot Ridge – C1 –0.10 ± 0.34 –0.25 ± 0.39 0.12 ± 0.69
80 Dinosaur Nat. Mon.a –0.65 ± 0.73 –0.51 ± 1.23 –0.40 ± 1.84

a Dinosaur National Monument has just 9 years ≥80% of summer amplitude data available.

Figure 12 

Changes in the summer O3 daily amplitude for the 95th percentile (top), 50th percentile (middle), and 5th percentile (bottom). Shown are all sites with at least 10 years of summer data. The colors of the marker represent the value of the slope. Statistically significant values (p-value ≤ 0.05) are plotted four times larger than statistically not significant values. DOI:

4. Discussion

4.1. Year-to-year variability

Surface O3 is dependent on numerous influences, causing significant year-to-year variability (Strode et al., 2015). The range (spread), representing the year-to-year variability, of the median O3 summer values within individual records was between 8–18 ppbv for these Colorado sites. Study of causal relationships determining O3 have been an active field of research, motivated by the necessity to elucidate the effect of human influences. Processes determining year-to-year changes include prevalence of high pressure weather systems (Reddy and Pfister, 2016), occurrences of stratospheric intrusions (Langford et al., 2009; Langford et al., 2015), wildfires (Jaffe et al., 2008; Jaffe and Wigder, 2012; Jaffe et al., 2013; Baylon et al., 2016; Lu et al., 2016), long-range to trans-continental transport events (Jaffe and Ray, 2007; Zhang et al., 2008; Reidmiller et al., 2009; Zhang et al., 2009), and changes in anthropogenic O3 precursor emissions. In Colorado, elevation and proximity to urban centers are two factors further influencing absolute levels and variability. The individual contribution of these effects on the O3 data considered here is beyond the objective of this study. However, recognition of these multiple influencing factors is of importance for evaluating the trend results presented here. While the 10-year minimum cutoff and 95% p-test confidence interval provide robust trend results, it cannot be completely excluded that trend results are influenced by multi-year extremes of any or a combination of the above listed factors.

4.2. Dependency of ozone on elevation

Highest O3 median values were observed at the high elevation sites. The summer elevation gradient of 6.5 ± 1.6 ppbv km–1 found in the data considered here is significantly lower than the 15 ppb km–1 reported by Brodin et al. (2010) for a ~30 km horizontal/2 km elevation east-west transect west of Boulder. These different results are likely due to the abundance of considered network sites within the NCFR for the study presented here, and the wide spread seen in the low elevation data (Figure 4). While Brodin et al. (2010) referenced their data to a single downtown Boulder site, where O3 values were relatively low, many of the lower elevation sites that went into the calculation here are in suburban areas, where median O3 levels are higher overall because these sites have less nighttime O3 loss from titration by NO. While median O3 increases with elevation, there is not as much of a dynamic diurnal O3 change at the high elevation sites, as a larger fraction of O3 at these sites is due to long-range transport and the increase of O3 with elevation. The lower elevation sites, in contrast, are influenced more by local O3 production, causing larger diurnal O3 amplitudes. Those two dependencies determine the inverse relationship and scatter in the data in Figure 11.

4.3. Ozone trends at rural sites

For rural locations in the U.S., O3 trends have been inconsistent and smaller than in urban areas. The behavior in the western U.S. contrasts that in eastern regions of the country, where more homogenous, downward O3 trends have occurred (Cooper et al., 2012, 2014). Results from the analyses presented here provide further evidence for this conclusion, as most of the rural and higher elevation sites showed neutral to only slightly downward summer O3 for the 50th and 95th percentile values.

4.4. Ozone trends at high mountain sites

Seasonal studies have reported an increase in O3 at western U.S. mountain sites (Parrish et al., 2012; Cooper et al., 2014; Gratz et al., 2015). This has mostly been associated to increasing O3 in transcontinental transport from Asia, with this effect being most influential in spring. Two high mountain sites on the U.S. west coast that have been central in demonstrating O3 increases stemming from outflow from Asia are Lassen NP (1756 m asl), and Mt Bachelor (2763 m) (Jaffe et al., 2003; Jaffe and Ray, 2007; Parrish et al., 2012; Gratz et al., 2015). Incorporating O3 sonde data in their analyses, Cooper et al. (2012) calculated that springtime free tropospheric O3 over western North America increased at a rate of 0.41 ± 0.27 ppbv yr–1 between 1995–2011.

Available trends from two sites in the Colorado network above 2500 m asl, i.e. Niwot Ridge C1 (3035 m), and Rocky Mountain NP Longs Peak (2742 m) offer a comparison with these west coast high elevation sites. Remarkably, the mean (of these two, ± standard deviation) 5th/50th/95th percentile O3 trends are all negative, i.e. –0.23 ± 0.40/–0.33 ± 28/–0.48 ± 0.11 ppbv yr–1, and inconsistent with the observations from the U.S. west coast. This difference is probably due to the high elevation west coast sites being relatively isolated from North American influence, whereas the two Rocky Mountain sites are more responsive to the changes in background ozone in continental North America and the NCFR.

4.5. Ozone regulatory considerations

There are two groups of sites with MDA8 and Design Values over the >70 ppbv threshold (SM_Table_3, and SM_Figure_7 showing the geographical distribution): 1. high elevation mountain sites and 2. suburban NCFR sites. Sunlight Mountain, the 7th highest elevation site at 3225 m, has the highest average number of exceedances per year (32). This site record, however, is relatively short (3 years), which makes these data more susceptible to biases from interannual variability, and places a relatively high uncertainty on this ranking. There are several other high mountain sites where the MDA8 regularly exceeds 70 ppbv, such as Niwot Ridge, and Rocky Mountain NP Longs Peak. These observations are in line with previous elaborations that pointed out the high O3 at these mountain sites, and the significant contribution from long-range continental and upper troposphere/lower stratosphere transport on elevated O3 (Langford et al., 2009; Lin et al., 2012; Cooper et al., 2015). The second group are NCFR sites, and here, in particular, the suburban locations on the western periphery of the City of Denver. This chain of stations along the eastern foothills of the Rocky Mountains shows a consistent behavior of a year after year high number of high O3 days, with an average of 10–27 days per year when the O3 MDA8 exceeds 70 ppbv. A contributing factor appears to be the diurnal air flow pattern along the NCFR, where predominantly upslope flow conditions during mid-day to late afternoon bring air that has been enriched in O3 precursors towards the mountain slopes, causing elevated O3 in the afternoon at the Plains-Rocky Mountain transition zone (Evans and Helmig, 2017). The analyses presented here expand upon the observational study from two sites near Boulder (Evans and Helmig, 2017), and recent modeling work (Pfister et al., 2017), demonstrating this spatial elevated O3 distribution along a ~30 km wide belt extending some 150 km north-south (Fort Collins to Chatfield) along the NCFR foothills.

4.6. Ozone changes in relation to NOx and VOC

Mandated by the Clean Air Act, emission reductions of O3 precursors in NAAs have been remarkably successful in many U.S. urban regions, in particular in the mid-western U.S. (Lurmann et al., 2015; Sather and Cavender, 2016). For instance, reductions for the MDA8 ranging from 18–38 ppbv have been reported for Baton Rouge, Houston, El Paso, Houston, and Dallas-Fort Worth over the past 30 years. The most dramatic improvements have been achieved in Southern California, where on average the O3 MDA8 has decreased by 2.6 ± 0.2% yr–1 between 1973–2010 (Pollack et al., 2013). Nationwide, a 17% decrease has been reported for 2000–2015, with a regional decrease of 10% for the Southwest in 2000–2015, which includes the state of Colorado (EPA, 2016). Averaging the O3 trend results for DMA sites (#1, 2, 5, 14, 18, 33, 34, 35, 36) yields median 50th/95th summer O3 trends of 0.29 ± 0.41/–0.01 ± 0.34 ppbv yr–1. None of these results are indicative of declining O3. A major finding from these comparisons is that, despite the NOx reductions, O3 changes in Colorado do not mirror the O3 improvements (reductions) noted in the above mentioned literature from other western U.S. cities.

The most robust results (according to the statistical significance tests and number of sites) of the trend analyses are the increases in summer low percentile O3 values, and increases of median O3 in the DMA (Table 3, Figure 7). These changes in the distribution of absolute mixing ratios are paralleled by declining diurnal O3 amplitudes (Table 4, Figure 8). Since nighttime O3 minima are largely determined by O3 loss from titration with NO, the changes in nighttime O3 are indicative of gradually declining NOx concentrations in the urban corridor. Similar, albeit weaker, nighttime O3 changes are observed at the rural Colorado locations, suggesting a declining NOx signature on the regional scale as well. In these rural areas, increasing nighttime O3 levels are likely due to 1. a declining nighttime O3 destruction in air transported to the sites, and 2. a decline in the local nighttime O3 destruction. These indirectly inferred changes in NOx are confirmed by surface NOx monitoring data in the state. All five of the 12 records overall (SM_Figure_8) that are long enough for a trend analysis show declining NOx (SM_Figure_9) at all percentiles, with 11 out of these 15 NOx percentile changes being statistically significant downward trends (SM_Table_4). An overall 26.3 ± 5.4% 2005–2013 reduction in tropospheric NOx has been estimated from satellite observations for the Denver area (Lamsal et al., 2015; Lurmann et al., 2015), with indications that the rate of decline has become smaller during the most recent years of the record. These declines in Colorado NOx are consistent with results from other nationwide studies, building on surface monitoring, remote sensing, emission inventories, and modeling (Cooper et al., 2014; Lamsal et al., 2015; Strode et al., 2015; Lin et al., 2016; Sather and Cavender, 2016).

Changes in VOC and VOC reactivity are more difficult to assess than changes in NOx. Tailpipe emissions, constituting a major to dominant source of VOC in urban air, have been decreasing throughout U.S. cities, including in the DMA (Bishop and Stedman, 2008). Despite the growth of the automobile fleet and miles driven per vehicle, the signature of these declining emissions can be seen in declining urban air VOC (Warneke et al., 2012; Rossabi and Helmig, 2018). In the NCFR, VOC emissions from the O&NG industries are the dominant VOC source within and downwind of O&NG activities (Gilman et al., 2013; Swarthout et al., 2013; Thompson et al., 2014). Several studies have demonstrated increasing atmospheric VOC trends linked to O&NG fugitive emissions from the rapid growth of this industry across the U.S. (Schade and Roest, 2015; Vinciguerra et al., 2015; Helmig et al., 2016). Within Colorado, CDPHE has been conducting VOC monitoring at four locations. While some of these records indicate potentially declining VOC, at this time these records are too short, and/or the sampling has been inconsistent, and data are too variable for obtaining robust trend results.

In Figure 13 we plot available paired median summer O3 amplitudes against the summer median NOx values from the records shown in SM_Figure_8 to investigate the O3-NOx/VOC relationship further. Data markers for WEL and Denver Camp (DCP) are color-coded by the year of observation. There are several dependencies: 1. Lowest NOx and diurnal amplitudes were observed at rural and suburban locations; highest values were measured at WEL and DCP. 2. At sites with lower NOx (0–5 ppbv summer median), the median O3 amplitude scales close to linearly with increasing NOx. This behavior then reverses to a slightly negative relationship between 10–25 ppbv of NOx. 3. Despite DCP having the overall highest NOx, at equivalent NOx, O3 amplitudes at DCP are consistently lower (by ~10 ppbv) in comparison to WEL. 4. The color-coding of the WEL and DCP data shows that while for both sites NOx has been decreasing over the available record, O3 amplitudes have changed very little, possibly increasing slightly at WEL.

Figure 13 

Ozone median summer daily amplitude as a function of median summer NOx at 12 sites. Data are color-coded by year according to the scale on the right for Welby and Denver Camp. DOI:

4.7. Ozone at Welby

Data in Figure 13 point towards overall increasing diurnal O3 changes at WEL. Given the decreasing rate of nighttime O3 destruction by NO (as reflected in the statistically significant positive 5th percentile O3 trend), these data argue for an increase in O3 production chemistry at WEL. This conclusion is confirmed by the O3 trend analyses: WEL is the only site with (statistically significant) increasing O3 trends at all (5/50/95) percentiles (Table 3), and the only site where O3 diurnal amplitudes have been increasing at all percentiles. This behavior clearly shows that, thus far, there have been no improvements in O3 at WEL despite the achieved statewide reductions in NOx. Conversely, it appears that local O3 production has increased under declining NOx. Similar conclusions were derived from a recent analysis of trends in weekday/weekend differences in O3 and NOx at selected NCFR sites by Abeleira and Farmer (2017). Their study found that “most sites in the NFRMA were NOx-saturated, but are transitioning to, and in one case may already have reached, the peak P(O3) cross-over point between NOx-saturated and NOx-limited regimes”. Our analyses, relying on further site data and different analytical approaches, confirm the first part of their conclusion.

4.8. Ozone and O&NG development

WEL is situated on the northeastern outskirts of the DMA, within 2 miles of I–76, and with other busy roadways a bit further away. The site is also within the southwestern edge of the DJB (Figure 1). A map showing well pad locations surrounding the site is provided in SM_Figure_10; 20 well pads are located within a 10-km distance, and 1445 wells within a 20-km radius. The association of increasing O3 with the proximity to O&NG operations might suggest that emissions from this growing industry have contributed to increases in O3. Atmospheric VOC have been shown to increase along transects from the periphery towards the center of the DJB (Thompson et al., 2014; Helmig et al., 2015; Rossabi et al., 2017), and VOC have been reported at highly elevated concentrations, largely driven by O&NG emission within the DJB and downwind (Gilman et al., 2013; Swarthout et al., 2013; Thompson et al., 2014). Unfortunately, there are no VOC monitoring data available for WEL. The seasonal cycles of the diurnal amplitudes (Figure 9) show a wider summer maximum than at NREL. This may be linked to differences in VOC/NOx ratios at both sites. There is most likely more NOx, causing lower VOC/NOx ratios at NREL, or higher VOC, causing higher VOC/NOx at WEL. Lower VOC/NOx may inhibit O3 production more during the spring/fall shoulder season when there are fewer photochemically productive hours than during mid-summer (at NREL) in comparison to possibly higher VOC/NOx conditions at WEL.

Greeley – Weld County Tower (site #67) is another site within the DJB to further investigate this dependency. The monitoring location is surrounded by intense O&NG development in all directions. While the site, with a summer 95% percentile value of 71 ppbv, and a median diurnal O3 amplitude of 51 ppbv, is ranked among the top ten of the Colorado sites for highest O3 and diurnal production, trend analyses for the O3 higher percentile values and O3 diurnal amplitudes show signs for decreasing local O3 production (Tables 3 and 4). VOC monitoring in Platteville ~25 km to the southwest has shown highly elevated O&NG associated VOC (Thompson et al., 2014; Halliday et al., 2016). There are, however, indications of potentially declining O&NG related VOC concentrations during recent years at Platteville (Pierce, 2017).

Other recent research building on campaign observations has provided observational and modeling evidence that O&NG emissions significantly contribute to summer O3 production within the DJB and NCFR downwind regions (Gilman et al., 2013; Swarthout et al., 2013; McDuffie et al., 2016; Pfister et al., 2017; Cheadle et al., 2017). This association is difficult to prove with the data and tools that were applied here. Unfortunately, the DJB and its periphery are not well covered by the Colorado surface O3 monitoring network, and regrettably, a valuable site in the SW sector of the DJB, i.e. the Boulder Atmospheric Observatory, was decommissioned in 2017. Further, given the lack of parallel O3, VOC, and NOx monitoring, it is impossible to deconvolute the influence of precursor emissions and their changes over time on the contrasting behavior seen in the data from these DJB sites.

5. Summary and conclusions

Ozone behavior in Colorado is rather diverse, both along the horizontal scale and the ~2000 m elevation change from the Eastern Rocky Mountain Plains to the peaks of the Continental Divide. Elevated O3 is prominent at higher elevation sites, and in the NCFR. However, in the NCFR, highest O3 is not recorded within the central DMA, but instead mostly at suburban sites to the north and west of Denver along the eastern foothills of the Colorado Rocky Mountains.

There is high year-to-year variability in O3 in the NCFR which adds to the uncertainty of surface O3 trend results. Ozone trends at considered non-urban sites, as well as at the highest mountain sites, do not show increasing O3 during the observation period. This contradicts the argument that NCFR elevated O3 can (in part) be explained by increasing O3 transport into the state. Analyses presented here underscore that NCFR O3 trends are primarily determined by local/regional emissions and production.

Our analyses for DMA sites do not show the clear O3 decreases at the higher percentiles as seen for the vast majority of urban areas across the U.S., including the southwestern states during the past 1–2 decades. Only five locations (collocated with O3) in the State of Colorado have NOx data records suitable for trend analyses. Declining trends seen in those NOx data, and the clear signature of decreasing O3 amplitudes, are in agreement with remote sensing and modeling results. Together, these records provide solid evidence for a steady decline of NOx in the region. The lack of the higher percentile O3 response to the NOx decline indicates that O3 in the region has not been sensitive to these NOx reductions thus far.

While we were able to consider data from an impressive 80 O3 monitoring sites for this study, there is only one site that provides a long-term surface O3 record from within the DJB, Colorado’s largest O&NG basin. Sites on the southern and western periphery of the DJB are subject to influence by urban development and changes in emission from non-O&NG emission sectors, which complicates assessing the O&NG influence and how its importance is changing over time. Further O3 monitoring within the DJB and at its northern and eastern boundaries would be beneficial for addressing these questions. With the predominant daytime upslope (east to west) air circulation conditions during summer days, those sites could provide data from upwind of the DJB, which, together with the existing chain of sites along the foothills, would allow assessing daytime O3 production as air travels over the DJB and becomes enriched with O&NG emissions. This would largely enhance capabilities for studying the role of O&NG operations in the NCFR O3 production, and the effectiveness of regulations for curbing O&NG O3 precursor emissions.

The sparseness of direct surface NOx observations poses a severe limitation for understanding O3 responses to emission changes and for directing O3 policy decisions. Continuous, high quality, year-round observations of VOC and NOx, co-located with O3 monitoring, are needed for a more comprehensive evaluation of changing influences and regional differences in the Colorado O3 production chemistry. Here, we explored the use of summer diurnal O3 amplitudes as an indirect indicator for site comparisons and for NOx trend analyses, and found this variable to be a sensitive and valuable tool for this research. The lack of response in O3 amplitudes to declining NOx at two DMA sites indicates that O3 daily production dynamics have been relatively non-responsive to decreasing NOx at those locations, which agrees with the interpretation of the O3 trend analyses for the wider NCFR.

Data Accessibility Statement

McClure-Begley, A., Petropavlovskikh, I., Oltmans, S. (2014) NOAA Global Monitoring Surface O3 Network. National Oceanic and Atmospheric Administration, Earth Systems Research Laboratory Global Monitoring Division. Boulder, CO; accessed September 2016; US Environmental Protection Agency (EPA) Air Quality System (AQS) archive; accessed September 2016; Dinosaur National Monument (DNM) data were obtained from the U.S. National Park Service; accessed October 2016;

Supplemental files

The supplemental files for this article can be found as follows:

  • SM_Table_1. Population growth in the Northern Colorado Front Range and La Plata County between 2000–2015. DOI:
  • SM_Table_2. Fraction of data (in %) for each site and year. DOI:
  • SM_Table_3. Average number of days per year with the MDA8 exceeding 70 ppbv and the number of years considered. All sites with at least 75% of data for one year during this period are included. DOI:
  • SM_Table_4. Linear regression results for the trend analysis of the summer NOx of all sites with at least 10 years and 80% of data for a given summer, categorized by the 5th, 50th, and 95th percentile levels. Statistically significant trend results are highlighted in bold font. DOI:
  • SM_Figure_1. Ozone time series results for the full 1-hour record of all sites considered in this study with yearly statistical box-whisker plot analysis for years with more than 80% of data coverage shown to the right of the time series. Sites are sorted according to the order of Table 1 in the article. DOI:
  • SM_Figure_2. Box-whisker plot analyses of the hourly summer ozone data for every year. Shown are all sites that had records with more than 10 years of data. Straight linear regression fits trough the 5th, 50th, and 95th percentile of the data for each year show the trend behaviour, with slope results in the legends in ppbv yr–1. DOI:
  • SM_Figure_3. Ozone Design Value for each year for all sites with at least 10 years of data. The 70 ppbv ozone NAAQS is visualized by a black line. The linear regression fit with the 95% confidence bounds is shown by the red line. The same analysis for the 4th MDA8 is shown in the series of graphs underneath. Trend analyses results are shown in manuscript Table 3. DOI:
  • SM_Figure_4. Amplitude in daily ozone for the full record and for all sites considered in this study. DOI:
  • SM_Figure_5. Statistical box-whisker plot analyses of the monthly change of the ozone diurnal amplitude for all sites. Shown are all sites that have at least 10 years of data with 80% of data coverage for a given year. DOI:
  • SM_Figure_6. Statistical box-whisker plot analyses of the diurnal ozone amplitude during the summer for sites with more than 10 years of data. Straight linear regression fits through the 5th, 50th and 95th percentile of the data for each year show the trend behaviour. Regression analysis slope results (in ppbv yr–1) are given in the legends. DOI:
  • SM_Figure_7. Geographical distribution of the 2000-2015 average number of days per year with the MDA8 exceeding 70 ppbv. All sites with at least 75% of data for one year during this period are included. Please note that the number of years going into the calculation may vary from 1–16. DOI:
  • SM_Figure_8. NOx time series results for the full 1-hour resolution record of those sites that had both NOx and ozone data available. DOI:
  • SM_Figure_9. Box-whisker plot analyses of the hourly summer NOx data. Shown are all sites that had records with more than 10 years of data. Straight linear regression fits trough the 5th, 50th, and 95th percentile of the data for each year show the trend behaviour, with slope results in the legends in ppbv yr–1. DOI:
  • SM_Figure_10. Location of the Welby ozone monitoring site indicated by the red square, with oil and natural gas wells within a 10 km and 20 km radius indicated by blue dots, and further wells outside that radius with red markers. The total well count within the drawn radius is indicated in the lower right corner. DOI:
  • SM_Text_1. Examination of the ozone data for the sites Ignacio and Bondad. DOI:


We thank all station operators and agencies who initiated, conducted, and quality-controlled these measurements and made them available for public use through dissemination in the considered archives. Without the dedicated work of these many colleagues, this study would not have been possible. G. Pierce, CDPHE, provided VOC data from a series of Colorado sites, C. Ellis, Southern Ute Tribe, provided information for evaluation of the Bondad and Ignacio data, and S. Oltmans and O. Cooper, both with NOAA ESRL in Boulder, and G. Pierce gave valuable feedback on earlier versions of the manuscript. The paper also benefited from the insightful comments of four anonymous reviewers.

Funding information

T.B., from the University of Münster, Germany, was sponsored by a grant from the Deutsche Akademischer Austauschdienst (DAAD) while she conducted these data analyses during a student internship at the University of Colorado, Boulder. The O3 monitoring at Niwot Ridge is supported through the US National Science Foundation Long-term Ecological Research award #DEB-1637686. Publication of this chapter was funded by the University of Colorado Boulder Libraries Open Access Fund.

Competing interests

Representatives of the agencies that provided data were contacted for permission to use these data in this publication. While they approved this request, this approval does not imply endorsement of any of the analyses and/or conclusions that are presented in this work. An Elementa EIC is an author on this publication. The anonymous peer-review of the manuscript was handled by the Forum Guest Editor and the Elementa Sustainable Engineering Domain EIC, without any influence of the EIC-author on the peer-review, and with preserving the anonymity of the reviewers.

Author contributions

  • TB: Data retrieval, data quality control, data analyses, manuscript preparation.
  • DH: Direction of the study, data quality control, data analyses, manuscript preparation.


  1. Abeleira, AJ and Farmer, DK. 2017. Summer ozone in the northern Front Range metropolitan area: Weekend-weekday effects, temperature dependences, and the impact of drought. Atmospheric Chemistry and Physics 17(11): 6517–6529. DOI: 10.5194/acp-17-6517-2017

  2. Baylon, PM, Jaffe, DA, Pierce, RB and Gustin, MS. 2016. Interannual variability in baseline ozone and its relationship to surface ozone in the western US. Environmental Science & Technology 50(6): 2994–3001. DOI: 10.1021/acs.est.6b00219

  3. Bishop, G and Stedman, DH. 2008. A decade of on-road emissions measurements. Environmental Science & Technology 45: 1651–1656. DOI: 10.1021/es702413b

  4. Brodin, M, Helmig, D and Oltmans, S. 2010. Seasonal ozone behavior along an elevation gradient in the Colorado Front Range Mountains. Atmospheric Environment 44(39): 5305–5315. DOI: 10.1016/j.atmosenv.2010.06.033

  5. CDPHE. 2008a. Denver Metro Area & North Front Range Ozone Action Plan, 1–76. December 12, 2008 (accessed January 2018).

  6. CDPHE. 2008b. Denver Metropolitan Area and North Front Range 8-Hour Ozone State Implementation Plan, 1–85. Draft Final, October 29, 2008 (accessed January 2018).

  7. CDPHE. 2016. History of Ozone in Colorado. (accessed January 2018).

  8. Cheadle, LC, Oltmans, SJ, Pétron, G, Schnell, RC, Mattson, EJ, Herndon, SCC, Thompson, AM, Blake, DR and McClure-Begley, A. 2017. Surface ozone in the Colorado northern Front Range and the influence of oil and gas development during FRAPPE/DISCOVER-AQ in summer 2014. Elementa – Science of the Anthropocene . DOI: 10.1525/elementa.254

  9. Code of Federal Regulations. 2017. Electronic Code of Federal Regulations. Appendix U.

  10. Cooper, OR, Gao, R-S, Tarasick, D, Leblanc, T and Sweeney, C. 2012. Long-term ozone trends at rural ozone monitoring sites across the United States, 1990–2010. Journal of Geophysical Research: Atmospheres 117: n/a–n/a. DOI: 10.1029/2012jd018261

  11. Cooper, OR, Langford, AO, Parrish, DD and Fahey, DW. 2015. Challenges of a lowered US ozone standard. Science 348: 1096–1097. DOI: 10.1126/science.aaa5748

  12. Cooper, OR, Parrish, DD, Ziemke, J, Balashov, NV, Cupeiro, M, Galbally, IE, Gilge, S, Horowitz, L, Jensen, NR, Lamarque, JF, Naik, V, Oltmans, SJ, Schwab, J, Shindell, DT, Thompson, AM, Thouret, V, Wang, Y and Zbinden, RM. 2014. Global distribution and trends of tropospheric ozone: An observation-based review. Elementa – Science of the Anthropocene . 2(000029).

  13. Dentener, F, Keating, T and Akimoto, H. 2010. Hemispheric Transport of Air Pollution 2010. Air Pollution Studies No. 17. Rep. United Nations , New York and Geneva.

  14. EPA. 2016. Air Quality – National Summary. (accessed November 2016).

  15. Evans, JM and Helmig, D. 2017. Investigation of the influence of transport from oil and natural gas regions on elevated ozone levels in the northern Colorado front range. Journal of the Air & Waste Management Association 67: 196–211. DOI: 10.1080/10962247.2016.1226989

  16. Gilman, JB, Lerner, BM, Kuster, WC and de Gouw, JA. 2013. Source signature of volatile organic compounds from oil and natural gas operations in northeastern Colorado. Environmental Science & Technology 47(3): 1297–1305. DOI: 10.1021/es304119a

  17. Gratz, LE, Jaffe, DA and Hee, JR. 2015. Causes of increasing ozone and decreasing carbon monoxide in springtime at the Mt. Bachelor Observatory from 2004 to 2013. Atmospheric Environment 109: 323–330. DOI: 10.1016/j.atmosenv.2014.05.076

  18. Halliday, HS, Thompson, AM, Wisthaler, A, Blake, DR, Hornbrook, RS, Mikoviny, T, Muller, M, Eichler, P, Apel, EC and Hills, AJ. 2016. Atmospheric benzene observations from oil and gas production in the Denver-Julesburg Basin in July and August 2014. Journal of Geophysical Research-Atmospheres 121: 11055–11074. DOI: 10.1002/2016jd025327

  19. Helmig, D, Hannigan, M, Milford, J and Gordon, J. 2015. Air quality monitoring study to assess exposure to volatile organic compounds and develop cost-efficient monitoring techniques, Final Report . Boulder County, Colorado.

  20. Helmig, D, Rossabi, S, Hueber, J, Tans, P, Montzka, S, Masarie, K, Thoning, K, Plass-Duelmer, C, Claude, A, Carpenter, L, Lewis, AC, Punjabi, S, Reimann, S, Vollmer, M, Steinbrecher, R, Hannigan, J, Emmons, L, Mahieu, E, Franco, B, Smale, D and Pozzer A. 2016. Reversal of global atmospheric ethane and propane trends largely due to US Oil and natural gas production. Nature Geoscience 9: 490–495. DOI: 10.1038/ngeo2721

  21. Jaffe, D, Chand, D, Hafner, W, Westerling, A and Spracklen, D. 2008. Influence of fires on O3 concentrations in the western US. Environmental Science & Technology 42: 5885–5891. DOI: 10.1021/es800084k

  22. Jaffe, D, Price, H, Parrish, D, Goldstein, A and Harris, J. 2003. Increasing background ozone during spring on the west coast of North America. Geophysical Research Letters 30: DOI: 10.1029/2003gl017024

  23. Jaffe, D and Ray, J. 2007. Increase in surface ozone at rural sites in the western US. Atmospheric Environment 41: 5452–5463. DOI: 10.1016/j.atmosenv.2007.02.34

  24. Jaffe, DA, Wigder, N, Downey, N, Pfister, G, Boynard, A and Reid, SB. 2013. Impact of wildfires on ozone exceptional events in the western US. Environmental Science & Technology 47: 11065–11072. DOI: 10.1021/es402164f

  25. Jaffe, DA and Wigder, NL. 2012. Ozone production from wildfires: A critical review. Atmospheric Environment 51: 1–10. DOI: 10.1016/j.atmosenv.2011.11.063

  26. Jerrett, M, Burnett, RT, Pope, CA, Ito, K, Thurston, G, Krewski, D, Shi, YL, Calle, E and Thun, M. 2009. Long-term ozone exposure and mortality. New England Journal of Medicine 360: 1085–1095. DOI: 10.1056/NEJMoa0803894

  27. Lamsal, LN, Duncan, BN, Yoshida, Y, Krotkov, NA, Pickering, KE, Streets, DG and Lu, Z. 2015. U.S. NO2 trends (2005–2013): EPA Air Quality System (AQS) data versus improved observations from the Ozone Monitoring Instrument (OMI). Atmospheric Environment 110: 130–143. DOI: 10.1016/j.atmosenv.2015.03.055

  28. Langford, AO, Aikin, KC, Eubank, CS and Williams, EJ. 2009. Stratospheric contribution to high surface ozone in Colorado during springtime. Geophysical Research Letters 36. DOI: 10.1029/2009gl038367

  29. Langford, AO, Senff, CJ, Alvarez Ii, RJ, Brioude, J, Cooper, OR, Holloway, JS, Lin, MY, Marchbanks, RD, Pierce, RB, Sandberg, SP, Weickmann, AM and Williams, EJ. 2015. An overview of the 2013 Las Vegas Ozone Study (LVOS): Impact of stratospheric intrusions and long-range transport on surface air quality. Atmospheric Environment 109: 305–322. DOI: 10.1016/j.atmosenv.2014.08.040

  30. Lefohn, AS, Jackson, W, Douglass, SS and Knudsen, HP. 1997. Effect of surface ozone exposures on vegetation grown in the southern Appalachian Mountains: Identification of possible areas of concern. Atmospheric Environment 31: 1695–1708.

  31. Lin, M, Fiore, AM, Cooper, OR, Horowitz, LW, Langford, AO, Levy, H, II, Johnson, BJ, Naik, V, Oltmans, SJ and Senff, CJ. 2012. Springtime high surface ozone events over the western United States: Quantifying the role of stratospheric intrusions. Journal of Geophysical Research-Atmospheres 117. DOI: 10.1029/2012jd018151

  32. Lin, M, Horowitz, LW, Payton, R, Fiore, AM and Tonnesen, G. 2016. US surface ozone trends and extremes from 1980–2014: Quantifying the roles of rising Asian emissions, domestic controls, wildfires, and climate. Atmospheric Chemistry and Physics Discussion 1–56. DOI: 10.5194/acp-2016-1093

  33. Lu, X, Zhang, L, Yue, X, Zhang, J, Jaffe, DA, Stohl, A, Zhao, Y and Shao, J. 2016. Wildfire influences on the variability and trend of summer surface ozone in the mountainous western United States. Atmospheric Chemistry and Physics 16: 14687–14702. DOI: 10.5194/acp-16-14687-2016

  34. Lurmann, F, Avol, E and Gilliland, F. 2015. Emissions reduction policies and recent trends in Southern California’s ambient air quality. Journal of the Air & Waste Management Association 65: 324–335. DOI: 10.1080/10962247.2014.991856

  35. McDuffie, EE, Edwards, PM, Gilman, JB, Lerner, BM, Dubé, WP, Trainer, M, Wolfe, DE, Angevine, WM, deGouw, J, Williams, EJ, Tevlin, AG, Murphy, JG, Fischer, EV, McKeen, S, Ryerson, TB, Peischl, J, Holloway, JS, Aikin, K, Langford, AO, Senff, CJ, Alvarez, RJ, Hall, SR, Ullmann, K, Lantz, KO and Brown, SS. 2016. Influence of oil and gas emissions on summertime ozone in the Colorado Northern Front Range. Journal of Geophysical Research: Atmospheres 121: 8712–8729. DOI: 10.1002/2016jd025265

  36. Musselman, RC and Korfmacher, JL. 2014. Ozone in remote areas of the southern Rocky Mountains. Atmospheric Environment 82: 383–390. DOI: 10.1016/j.atmosenv.2013.10.051

  37. National Research Council. 1991. Rethinking the Ozone Problem in Urban and Regional Air Pollution. Natl. Acad. Press.

  38. Parrish, DD, Law, KS, Staehelin, J, Derwent, R, Cooper, OR, Tanimoto, H, Volz-Thomas, A, Gilge, S, Scheel, HE, Steinbacher, M and Chan, E. 2012. Long-term changes in lower tropospheric baseline ozone concentrations at northern mid-latitudes. Atmospheric Chemistry and Physics 12: 11485–11504. DOI: 10.5194/acp-12-11485-2012

  39. Pfister, G, Flocke, F, Hornbrook, RS, Orlando, J and Lee, S. 2017. Process-based and regional source impact analysis for FRAPPE and DISCOVER-AQ 2014. Final Report to the Colorado Department of Public Health and Environment . Day: 31 Month: July , 2017.

  40. Pierce, G. 2017. Ozone in the North Front Range and policies. Presented at FRAPPE Science Team Meeting. Boulder, CO. May 2–3.

  41. Pollack, IB, Ryerson, TB, Trainer, M, Neuman, JA, Roberts, JM and Parrish, DD. 2013. Trends in ozone, its precursors, and related secondary oxidation products in Los Angeles, California: A synthesis of measurements from 1960 to 2010. Journal of Geophysical Research-Atmospheres 118: 5893–5911. DOI: 10.1002/jgrd.50472

  42. Reddy, PJ and Pfister, GG. 2016. Meteorological factors contributing to the interannual variability of midsummer surface ozone in Colorado, Utah, and other western US states. Journal of Geophysical Research-Atmospheres 121: 2434–2456. DOI: 10.1002/2015jd023840

  43. Reidmiller, DR, Fiore, AM, Jaffe, DA, Bergmann, D, Cuvelier, C, Dentener, FJ, Duncan, BN, Folberth, G, Gauss, M, Gong, S, Hess, P, Jonson, JE, Keating, T, Lupu, A, Marmer, E, Park, R, Schultz, MG, Shindell, DT, Szopa, S, Vivanco, MG, Wild, O and Zuber, A. 2009. The influence of foreign vs. North American emissions on surface ozone in the US. Atmospheric Chemistry and Physics 9: 5027–5042. DOI: 10.5194/acp-9-5027-2009

  44. Rossabi, S, Evans, J, Thompson, C, Hueber, J, Smith, K, Reed, T, Wang, W and Pfister, G. 2017. Surface observations of volatile organic compounds along an elevation gradient in the Northern Colorado Front Range during FRAPPE. Presented at FRAPPE Science Team Meeting. Boulder, CO. May 2–3.

  45. Rossabi, S and Helmig, D. 2018. Changes in atmospheric butanes and pentanes and their isomeric ratios in the continental United States. Journal of Geophysical Research-Atmospheres In Press. DOI: 10.1002/2017JD027709

  46. Sather, ME and Cavender, K. 2016. Trends analyses of 30 years of ambient 8 hour ozone and precursor monitoring data in the South Central US: Progress and challenges. Environmental Science-Processes & Impacts 18: 819–831. DOI: 10.1039/c6em00210b

  47. Schade, GW and Roest, GS. 2015. Is the shale boom reversing progress in curbing ozone pollution? EOS 96. DOI: 10.1029/2015EO028279

  48. Strode, SA, Rodriguez, JM, Logan, JA, Cooper, OR, Witte, JC, Lamsal, LN, Damon, M, Van Aartsen, B, Steenrod, SD and Strahan, SE. 2015. Trends and variability in surface ozone over the United States. Journal of Geophysical Research-Atmospheres 120: 9020–9042. DOI: 10.1002/2014jd022784

  49. Swarthout, RF, Russo, RS, Zhou, Y, Hart, AH and Sive, BC. 2013. Volatile organic compound distributions during the NACHTT campaign at the Boulder Atmospheric Observatory: Influence of urban and natural gas sources. Journal of Geophysical Research-Atmospheres 118: 10614–10637. DOI: 10.1002/jgrd.50722

  50. The Royal Society. 2008. Ground-level ozone in the 21st century: Future trends, impacts and policy implications. Science Policy Report 15/08 , 1–148. (accessed January 2018).

  51. Thompson, CR, Hueber, J and Helmig, D. 2014. Influence of oil and gas emissions on ambient atmospheric non-methane hydrocarbons in residential areas of Northeastern Colorado. Elementa – Science of the Anthropocene 3. DOI: 10.12952/journal.elementa.000035

  52. Vinciguerra, T, Yao, S, Dadzie, J, Chittams, A, Deskins, T, Ehrman, S and Dickerson, RR. 2015. Regional air quality impacts of hydraulic fracturing and shale natural gas activity: Evidence from ambient VOC observations. Atmospheric Environment 110: 144–150. DOI: 10.1016/j.atmosenv.2015.03.056

  53. Warneke, C, de Gouw, JA, Holloway, JS, Peischl, J, Ryerson, TB, Atlas, E, Blake, D, Trainer, M and Parrish, DD. 2012. Multiyear trends in volatile organic compounds in Los Angeles, California: Five decades of decreasing emissions. Journal of Geophysical Research-Atmospheres 117. DOI: 10.1029/2012JD017899

  54. Young, PJ, Archibald, AT, Bowman, KW, Lamarque, JF, Naik, V, Stevenson, DS, Tilmes, S, Voulgarakis, A, Wild, O, Bergmann, D, Cameron-Smith, P, Cionni, I, Collins, WJ, Dalsøren, SB, Doherty, RM, Eyring, V, Faluvegi, G, Horowitz, LW, Josse, B, Lee, YH, MacKenzie, IA, Nagashima, T, Plummer, DA, Righi, M, Rumbold, ST, Skeie, RB, Shindell, DT, Strode, SA, Sudo, K, Szopa, S and Zeng, G. 2013. Pre-industrial to end 21st century projections of tropospheric ozone from the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP). Atmospheric Chemistry and Physics 13(4): 2063–2090. DOI: 10.5194/acp-13-2063-2013

  55. Zhang, L, Zhang, L, Jacob, DJ, Boersma, KF, Jaffe, DA, Olson, JR, Bowman, KW, Worden, JR, Thompson, AM, Avery, MA, Cohen, RC, Dibb, JE, Flock, FM, Fuelberg, HE, Huey, LG, McMillan, WW, Singh, HB and Weinheimer, AJ. 2008. Transpacific transport of ozone pollution and the effect of recent Asian emission increases on air quality in North America: An integrated analysis using satellite, aircraft, ozonesonde, and surface observations. Atmospheric Chemistry and Physics 8: 6117–6136. DOI: 10.5194/acp-8-6117-2008

  56. Zhang, L, Jacob, DJ, Kopacz, M, Henze, DK, Singh, K and Jaffe, DA. 2009. Intercontinental source attribution of ozone pollution at western US sites using an adjoint method. Geophysical Research Letters 36. DOI: 10.1029/2009gl037950

comments powered by Disqus