Introduction
Atmospheric aerosols are of interest due to their impacts on human health,
visibility, and climate radiative forcing through scattering and absorption
of solar radiation (Monks et al., 2009; Stocker et
al., 2013). Notably, numerous studies have shown that aerosols contribute
to respiratory and cardiac disease, leading to an increase in morbidity and
mortality in humans (Dockery et al., 1993; Dockery and Schwartz,
1995; Pope et al., 1995, 2002, 2009; Pope III et al., 1995;
Bascom et al., 1996; Pöschl, 2005; Valavanidis et al., 2008). Moreover,
ecological changes in lakes and national forests from nitrogen deposition
are a driving concern for the sustainability of the ecosystem (Wilson and
Spengler, 1996; Baron et al., 2000; Williams and Tonnessen, 2000;
Blett et al., 2004; Burns, 2004; Seinfeld and Pandis, 2012).
Urban air is comprised of a highly complex mixture of gaseous and
particulate pollutants, including volatile organic compounds (VOCs),
nitrogen oxides (NO and NO2), sulfur dioxide (SO2),
ozone (O3), and fine particulate matter (PM2.5) and is detrimental to
the environment and to the well-being of the public. A significant amount of
submicron aerosol mass in the troposphere is comprised of organic aerosols
(OAs), but direct sources, composition, and formation processes of OA are
still not fully understood (Pandis et al., 1992; Turpin and
Huntzicker, 1995; Odum et al., 1996; Schell et al., 2001;
Claeys et al., 2004; Kroll et al., 2006; Volkamer
et al., 2006; Kroll and Seinfeld, 2008; Hallquist et al., 2009;
Jimenez et al., 2009; Zhang et al., 2011). Generally, OAs
are comprised of primary particles emitted into the atmosphere (i.e.,
primary organic aerosols; POAs) and products formed from multiphase chemical
reactions as secondary organic aerosols (SOAs). Several important factors
including aerosol composition and size determine the extent to which
aerosols affect the environment and health.
The Colorado Front Range continues to face challenges attributed to air
quality. In 2007, the Northern Front Range (NFR) and the Denver metropolitan
area (DM) were designated as federal non-attainment areas for the federal
8 h ozone standard (75 ppbv), averaged over 3 years (EPA, 2008).
Since May 2016, this area has been classified as a “moderate” non-attainment
regions for failure to attain the federal 8 h ozone standard of 75 ppbv
(averaged over 3 years) (EPA, 2016). Furthermore,
under the Clean Air Act, the US EPA Regional Haze Rule mandates the
reduction in anthropogenic emissions to achieve visibility improvement in
wilderness areas, including Colorado's Rocky Mountain National Park.
Additionally, the State of Colorado has implemented a visibility standard
based on optical extinction of 76 M m-1, averaged within a 4 h period
when relative humidity (RH) is less than 70 %. This measure of total
optical extinction is provided by an Optec LRT-2 long-range transmissometer
at 550 nm between east Denver and downtown (39∘44′8.52′′ N,
104∘57′29.50′′ W) from 08:00 to 16:00 (MST in winter and MDT in
summer). The establishment of the Denver visibility standard-setting is
covered in detail by Ely et al. (1993).
The complex topography of the Colorado Front Range leads to terrain-induced
flows and mesoscale circulations that have a significant impact on air
quality. These include cycles of daytime thermally driven upslope from the
plains into the mountains and decoupled, downslope nighttime drainage and
slope flows which can transport and pool particulates and precursors of
secondary aerosols into the wider Platte River valley between Denver and
Greeley, Colorado. Thermally driven upslope flows or cool moist
northeasterly upslope flows can lead to secondary aerosol formation and poor
visibility (Neff, 1997). Many of these upslope flows can be caused by
low-pressure formation in southern Colorado, a lee trough or line of lower
pressure along the foothills, and the Denver Cyclone. The Denver Cyclone
(Wilczak and Glendening, 1988; Wilczak and Christian, 1990; Szoke, 1991;
Szoke et al., 2006) is a mesoscale cyclonic gyre which can form
when there are southeasterly flows across the Palmer Divide (an east-to-west
feature of higher terrain to the south of Denver) and a layer of high
stability above the surface mixed layer and below 700 hPa (Szoke and
Augustine, 1990; Reddy and Pfister, 2016). Reddy et al. (1995)
have shown that the Denver Cyclone plays a key role in the degradation of
visibility and exceedances of the state visibility standard of 76 M m-1
during the winter, but our study is the first to examine the summertime
impacts of the Denver Cyclone during an intensive air quality study with a
detailed suite of aircraft and surface measurements.
Air pollution in the Northern Front Range is impacted by vehicular emissions
from growing urbanization in the Denver metropolitan area, local
powerplants, agriculture (e.g., concentrated animal feeding operations – CAFOs),
and extensive oil and gas (O & G) explorations. Recent studies
have shown O & G emissions of non-methane hydrocarbons (NMHC) such as
short-chain alkanes (C1–C4) and alkenes act as precursors to ozone
(Pétron et al., 2012, 2014; Edwards et al., 2013;
Gilman et al., 2013; Karion et al., 2013), but the potential for these emissions to
contribute to primary and secondary OA in the region has not been
investigated. Additionally, agricultural practices and powerplant
operations in the greater Colorado region contribute to visibility
impairment and ecosystem degradation through formation of secondary nitrate
and sulfate-containing compounds (Williams and Tonnessen, 2000;
Nanus et al., 2003; Blett et al., 2004; Burns, 2004;
Boy et al., 2008; Malm et al., 2013; Mast and Ely, 2013; T. M. Thompson et al., 2015).
Emission sources and meteorological conditions affecting air quality in the
greater Front Range have been previously studied in the region. The 1973
Denver Air Pollution Study (Russell, 1976), focused on episodes of
winter pollution in Denver, described occurrences of rapid dispersal of
pollutants to the north-northeast of Denver due to strong winds and
recurring reversal of winds, bringing aged pollutants back to the urban
center. Additionally, the Denver Haze Study conducted in the winter of
1978–1979 and the 1987–1988 Metro Denver Brown Cloud study provided objective
apportionment to the observed brown cloud pollution over Denver. The
occurrence of the wintertime inversion layer and emissions from the local
gas and coal-burning powerplants had a profound effect on air quality and
visibility degradation. Among the measured aerosol species, elemental
carbon, ammonium sulfate, and ammonium nitrate contributed to the majority
of optical extinction, decreasing visibility in the visible range by about 38,
20, and 17 %, respectively (Countess et al., 1980;
Groblicki et al., 1981; Wolff et al., 1981; Watson
et al., 1988; Neff, 1989).
During 1996–1997, measurements of aerosol composition and inorganic aerosol
precursors were carried out in winter and summer months at several urban and
rural sites during the Northern Front Range Air Quality Study (NFRAQS).
Summertime 24 h PM2.5 mass concentrations at different sites ranged
from 4 to 26 µg m-3, while winter measurements indicated variable
PM2.5 mass in the range of 1–51 µg m-3, depending on the
sampling location and year (Watson et al., 1998). During summer
1996 and at an urban site northeast of downtown Denver, OA was the most
dominant component of PM2.5 mass, contributing 46 % of the mass
with an average organic carbon mass of 4.2 µg m-3
(Watson et al., 1998). During this time, secondary inorganic
aerosol contributed to 18 % of PM2.5 mass, about 50 % lower than
the wintertime observations, with average sulfate and nitrate concentrations
of ∼ 1.4–1.5 and 0.9–1.2 µg m-3,
respectively (Watson et al., 1998). On average, crustal
components of PM2.5 were low in concentration (less than 0.5 µg m-3)
during summer 1997 (Watson et al., 1998). Since the
late 1990s, emissions in the Front Range have likely changed due to changes
in the vehicular fleet, urbanization, and growth in O & G-related
activities. Despite these changes, recent comprehensive characterization of
summertime air quality in the Colorado Front Range has been lacking. More
importantly, limited studies have evaluated the summertime air quality
implications of the Denver Cyclone that result in transport of pollutants
from the Northern Front Range to the urban center.
During the summer of 2014, two major field campaigns, the Front Range Air
Pollution and Photochemistry Experiment (FRAPPÉ) cosponsored by
NSF/NCAR and the Colorado Department of Public Health and Environment (CDPHE),
and the fourth deployment of the NASA DISCOVER-AQ were carried
out to study summertime atmospheric pollution in the Northern Colorado Front
Range. In this paper, we focus our analysis on the data obtained during
FRAPPÉ to assess the impact of the Denver Cyclone on the region's air
quality (Flocke, 2015).
Measurements
Field campaign
Airborne measurements were made during the Front Range Air Pollution and
Photochemistry Experiment (FRAPPÉ) from 16 July through 18 August 2014.
Fifteen research flights were conducted over the northern Colorado
plains, foothills, and west of the Continental Divide to sample air masses
under the influence of diverse sources and meteorological patterns that
impact the overall air quality in the region. The C-130 flight tracks,
overlaid on a map including the location of active oil and gas
wells in the region, are shown in the Supplement of
this paper (Fig. S1) (COGCC, 2016). In this analysis,
measurements made in the geographical area of the greater Denver metropolitan area (latitudes of 39∘27′00′′–40∘15′36′′ N
and longitudes of 104∘17′24′′–105∘19′48′′ W) and northern Colorado counties in NFR (latitudes of 40∘15′58′′–41∘00′00′′ N and longitudes of
104∘45′00′′–105∘19′48′′ W) during days when the Denver Cyclone
was strongly developed (27–28 July) are contrasted with measurements made
during days without the presence of a Denver Cyclone (26 July, 2–3 August).
Airborne data presented in this analysis are limited to
measurements in the boundary layer (i.e., altitudes below 2300 m east of the
foothills as further discussed in Sect. 2.3) to capture air masses
impacted by various local sources.
Instrumentation
In situ size-resolved composition measurements of non-refractory
submicron aerosols (NR-PM1); organic OA; nitrate – NO3-; sulfate – SO42-; ammonium – NH4+; and chloride – Cl-)
were made with an aerosol mass spectrometer equipped with a
compact time-of-flight detector (mAMS, Aerodyne Inc.). Principle details of
the instrument are described in depth elsewhere (Jayne et al.,
2000; Drewnick et al., 2005; Canagaratna et al., 2007). In
short, aerosols form a narrow particle beam by passing through an
aerodynamic lens system (Liu et al., 1995a, b). After
traveling through the high-vacuum particle time-of-flight chamber and
impacting an inverted-cone tungsten vaporizer at approximately
600 ∘C, non-refractory components of aerosols are evaporated and
ionized by electron impact ionization. The data are acquired at 15 s
intervals in two distinct acquisition modes (Jimenez et
al., 2003). In the particle time-of-flight (PToF) mode, the particle beam
is modulated by a multi-slit chopper system, allowing for particle sizing.
In the mass spectrometry mode (MS), the chopper completely blocks or opens
the particle beam, allowing the determination of the ensemble mass spectra
of aerosol species.
The mass response of the mAMS was calibrated regularly by sampling
size-selected, dry, monodisperse NH4NO3 particles with the
procedure and calculations described in previous literature to determine the ionization efficiency (IE) of nitrate and ammonium (Jimenez et
al., 2003; Zhang et al., 2005). The average ratio of the nitrate
ionization efficiency ratio to the air beam signal was (2.57 ± 0.26) × 10-13
from five calibrations performed during the study,
indicating stability of the instrument throughout the project.
Composition-dependent collection efficiency was applied to all the data in
this study (Middlebrook et al., 2012). mAMS data analysis was
carried out using the standard SQUIRREL analysis software (v1.56,
Sueper, 2014) with Igor Pro 6.37 (WaveMetrics, Lake Oswego, OR).
Ambient aerosols were sampled through a secondary diffuser inside a forward-facing NCAR High-performance Instrumented Airborne Platform for
Environmental Research (HIAPER) modular inlet (HIMIL) (Rogers, 2011),
mounted under the aircraft, with a total residence time of 0.5 s between the
HIMIL inlet and the mAMS. Assuming the sample flow reached the same
temperature as the cabin air within this time, relative humidity of the
sample flow was estimated to be less than 40 % for the data presented
here. For the ambient conditions in the boundary layer (i.e., 20 ∘C
and 70 kPa), the secondary diffuser inlet was estimated to be a PM2
inlet, i.e., with 50 % transmission efficiency of 2 µm spherical
particles (density of 1500 kg m-3). A pressure-controlled inlet (PCI)
(Bahreini et al., 2008) was used to maintain a constant
pressure of 350 Torr in the mAMS inlet to eliminate fluctuations in particle
size and transmission efficiency with ambient pressure variations.
Measurements of gas phase tracers used in this analysis include carbon
monoxide (CO), measured by vacuum UV resonance fluorescence
(Gerbig et al., 1999; Holloway et al., 2000;
Takegawa et al., 2001) on the C130 and by Differential Absorption
Carbon Monoxide Measurement (DACOM) instrument with an
in situ diode laser spectrometer system (Choi
et al., 2008; Warner et al., 2010) on the NASA DISCOVER-AQ P-3
aircraft. NOx (NO and NO2), were measured by chemiluminescence
(Ridley et al., 2004). Mixing ratios of NOy (total
reactive oxidized nitrogen species) were estimated as the sum of NOx,
aerosol nitrate (NO3-), nitric acid (HNO3)
(Huey et al., 1998; Huey, 2007), peroxyacetyl nitrate (PAN),
and peroxypropionyl nitrate (PPN), measured by chemical ionization mass
spectrometry (CIMS) (Slusher et al., 2004), and alkyl
nitrates (ANs), measured using thermal dissociation-laser induced
fluorescence (TD-LIF) (Thornton et al., 2000; Day
et al., 2002). A compact quantum cascade tunable infrared laser
differential absorption spectrometer (QC-TILDAS) was used for ammonia (NH3)
measurements (Ellis et al., 2010), while
C2H6 and CH2O were measured by mid-infrared spectrometry
using the Compact Atmospheric Multi-species Spectrometer (CAMS)
(Weibring et al., 2006, 2007;
Richter et al., 2015). Volatile organic compounds (VOCs), including
C6–C9 aromatics, i-pentane, and n-pentane were
measured by online proton-transfer mass spectrometry (PTR-MS) and the Trace
Organic Gas Analyzer (TOGA), respectively (Lindinger et al.,
1998; de Gouw and Warneke, 2007; Apel et al., 2015).
C-130 flight tracks in the Colorado Front Range for (a) non-cyclone
days (26 July, 2–3 August 2014) and (b) cyclone days (27–28 July 2014);
red marked boundaries represent three different study regions: In-Flow, Northern
Front Range (NFR), and Denver metropolitan area (DM). Peach colored markers
represent active oil and natural gas wells in the region with data available
from the Colorado Oil and Gas Conservative Commission.
Data processing
Reported data are a subset of the FRAPPÉ 2014 data collected aboard the
NSF/NCAR C-130 aircraft. All data presented here were limited to air masses
sampled below ∼ 2300 m a.s.l. and values for aerosol
concentrations are reported at standard temperature and pressure (STP; 1013 hPa and 273 K, µg s m-3).
Additionally, data were chosen from days before (26 July),
during (27–28 July), and after (2–3 August) the Denver Cyclone period,
with the strongest features of the cyclone being observed on 27 July. To
evaluate the impact of the Denver Cyclone in different regions of the Front
Range, measurements were analyzed in three regions, labeled as In-Flow, NFR, and DM, based on
cluster analysis of wind patterns and aerosol and gas phase tracer
concentrations observed on the day with the strongest Denver Cyclone,
27 July. Flight tracks and outlines of the latitude and longitudinal boxes for
these regions are shown in Fig. 1.
To assess the extent of boundary layer mixing and dilution, potential
temperature profiles measured by the Pennsylvania State University NATIVE
integrated ozonesonde (A. M. Thompson et al., 2015), launched near
Platteville (40∘10′53′′ N, 104∘43′36′′ W) during NASA
DISCOVER-AQ, were examined. Except for 26 July when at 12:00 MST the
boundary layer (BL) height was observed to be at 2200 m a.s.l., midday BL
heights on other days were consistently at ∼ 3400–3600 m a.s.l.
Additionally, except for the high, constant-altitude legs, the sampling
altitude on 26 July and the other flights was lower than
∼ 2000 and ∼ 2300 m a.s.l., respectively. Therefore, the data
discussed here represent mainly the altitudes of the boundary layer air masses.
Variability in the extent of boundary layer dilution due to differences in
daytime flight hours (takeoff times of 08:30–14:00 MST) showed some effects
on the observations; however, as further discussed in Sect. 3.3.1,
dilution differences were not the main driving factor in the observed trends
of absolute concentrations of gaseous and aerosol species.
ISORROPIA II modeling
An aerosol thermodynamics model, ISORROPIA II (Nenes, 2013) was used
to predict the phase and composition of the major inorganic aerosol
components. Detailed equilibrium relations and thermodynamic parameters used
in ISORROPIA II are outlined in Fountoukis and Nenes (2007). The model
was initiated with the average measured values of temperature (T), relative
humidity (RH), and total concentrations of ammonium (NH3(g) + NH4+),
sulfate (SO42-), and nitrate (HNO3(g) + NO3-).
Assuming chemical equilibrium and the presence of metastable
aerosols, the model predicted concentrations of sulfate, nitrate, and
ammonium present in the gas and aerosol phase, allowing the estimation of the
aerosol nitrate fraction (fNO3 = NO3-/(HNO3(g) + NO3-)) at equilibrium.
Results and discussion
Meteorology
Meteorological measurements presented in Table 1 show average ambient
temperature (T), RH, and wind speed (WS) during selected flights for each of
the three regions of interest on non-cyclone and cyclone days.
During non-cyclone days, T, RH, and WS were similar in all regions with an
average of 23 ± 1.6 ∘C, 35 ± 6.0 %, and 3.4 ± 1.5 m s-1,
respectively. During the cyclonic episode, the average T
across all three regions was 22 ± 1.6 ∘C and lower by
2–8 % in NFR and DM areas compared to the In-Flow region.
Additionally, average RH was higher in NFR and DM (64–70 %) compared to
the In-Flow region (37 %) during this mesoscale event. We further address
the importance of the contrast in RH between the events for aerosol nitrate
partitioning in Section 3.5. Average wind speed showed a 65 % increase in
the In-Flow region (6.3 ± 1.9 m s-1) during the cyclone event,
with a gradual decrease in the average wind speeds across NFR and DM.
Average temperature (T, ∘C), relative humidity (RH, %), and
wind speed (WS, m s-1) for measurements separated into In-Flow, NFR, and DM regions during the non-cyclone and cyclone episodes.
Region
T (∘C)
RH (%)
WS (m s-1)
Non-cyclone (26 July, 2–3 August)
In-Flow
22.8 ± 1.7
33.3 ± 4.6
3.8 ± 1.4
NFR
22.5 ± 1.7
38.4 ± 6.9
3.5 ± 1.6
DM
23.7 ± 1.4
34.0 ± 6.6
3.0 ± 1.3
Cyclone (27–28 July)
In-Flow
22.4 ± 1.4
37.0 ± 5.5
6.3 ± 1.9
NFR
21.8 ± 1.3
70.4 ± 7.2
4.1 ± 1.4
DM
20.6 ± 2.0
64.5 ± 7.7
3.2 ± 1.4
RAP model analysis runs at 13 km resolution for (a, b) 26 July 2014
(12:00 UTC (05:00 MST), 21:00 UTC (14:00 MST), respectively) and
(c, d) 2 August 2014 (12:00, 21:00 UTC, respectively). Arrows show
surface wind vectors, while the color scale represents surface RH.
RAP model analysis runs at 13 km resolution for the Denver Cyclone
on Sunday, 27 July 2014, at (a) 10:00 UTC (03:00 MST), (b) 12:00 UTC
(05:00 MST), (c) 15:00 UTC (08:00 MST), and (d) 18:00 UTC
(11:00 MST). The blue line represents a convergence zone or front associated
with the cyclone. Arrows show surface wind vectors, while the color scale
represents surface RH.
We used analysis runs of the National Centers for Environmental Prediction (NCEP)
13 km resolution Rapid Refresh (RAP) model for the periods of
interest. These analysis runs reflect extensive assimilation of
observational data. Plots were generated and analyzed with surface wind
vectors, RH, and specific humidity for days with and without the influence
of the cyclone. Surface wind direction and speed for both case scenarios are
shown in Figs. 2–3 and S2–S3. As previously described by
Toth and Johnson (1985), cyclic terrain-driven circulations
in this region are common during the summer when synoptic-scale influences
are weak. When synoptic-scale flows are weak and the Denver Cyclone is not
active, nighttime and early morning slope and drainage flows are formed as
radiative cooling in the higher terrain to the north, west, and south of DM causes denser, cooler air to flow downhill, with a general westerly
component along the valleys over Denver (Figs. 2a, c and S2a, c). The
surrounding terrain channels this drainage flow to the northeast through
Denver. This flow can carry emissions away from the urban center. During the
day, typical thermally driven flows reverse these winds, and transport is
generally towards the higher terrain. This daytime regime can also interact
with synoptic-scale winds leading to a hybrid pattern. Such a pattern is
apparent for the daytime winds plotted in Figs. 2b, d and S2b, d, where
thermally driven upslope flow was more apparent over the higher terrain to
the west and synoptic-scale flows had a greater influence over the plains.
Short-range return flows which can be formed by various mesoscale phenomena
(Reddy et al., 1995), including the Denver Cyclone, can occur
any time of the day and lead to a shift in direction of the winds with an
easterly component. These can draw the Platte Valley air masses uphill and
back over the greater Denver metropolitan area, enhancing the mixing of
older and new emissions (Neff, 1989).
Pronounced and fully developed surface mesoscale circulations of the Denver
Cyclone were observed on Sunday, 27 July 2014. Surface wind patterns and RH
in Fig. 3a–d display the development of the Denver Cyclone between 10:00 and
18:00 UTC (03:00 and 11:00 MST, correspondingly) on 27 July. Figure 3a
depicts the early stages of the cyclone with converging flows and the
beginnings of a counterclockwise circulation pattern centered to the
northeast of Denver. As seen in Fig. 3b, by 12:00 UTC (05:00 MST on 27 July), RH
was beginning to peak on the western or return flow side of the cyclone
center, which was still to the northeast of Denver. This northeasterly–northerly–northwesterly return flow on the western side of the cyclone transported
cool and moist air masses from the Platte Valley north of Denver towards the
urban core as the cyclone matured. As shown in Fig. S3b–d, air masses with
higher water content were advected westward by easterly winds, ahead of the
intensifying low-pressure system that was developed by 18:00 UTC (11:00 MST
on 27 July). A well-organized, well-defined cyclone circulation continued
with its center in the same location at 18:00 UTC (11:00 MST on 27 July)
with a warm, dry inflow to the east of the center and convergence line and a
cool, humid wraparound flow on the west side of the Denver Cyclone (Fig. 3d).
Spatial distribution maps along the C130 flight track of ethane (C2H6),
ammonia (NH3), and carbon monoxide (CO) in the Colorado Front Range during
non-cyclone (a–c) and cyclone episodes (d–f). Major highways
are shown with black lines, and grey markers in (a) represent the location
of active oil and gas wells in the region with data available from the Colorado
Oil and Gas Conservative Commission.
Various tracers were considered in the Weather Research and Forecasting
Model (WRF) to predict the distribution of emitted pollutants in the Front
Range at a horizontal resolution of 3 km × 3 km. The model was initialized
with the Global Forecast System (GFS) at 0.5∘ × 0.5∘
resolution and at 00:00 UTC (17:00 MST, on previous day) and 12:00 UTC
(05:00 MST) to produce 48 h forecasts. Figure S4a–f of the supplementary material
presents the distribution of the O & G tracer on 27 July. These
forecasting results represent the cyclone development on 27 July
well, with the surface winds reflecting the counterclockwise circulations
(NE to SW) though the cyclone core was predicted to be further northeast of
the Denver urban area. In this case, the model was able to predict the
cyclone episode and transport of emission tracers 24 h in advance, driving
the motivation for carrying out aircraft measurements during this event. In
the next sections, the impacts of this synoptic-scale recirculation flow on
pollutant distribution in the region are discussed.
Spatial distribution of trace gases and aerosols
The meteorological conditions described above are critical when considering
atmospheric aerosol formation, evolution, and spatial distribution.
Figure 4a–f show the spatial distribution of ammonia (NH3),
ethane (C2H6), and carbon monoxide (CO), i.e., tracers for agricultural
and CAFOs, O & G, and combustion and vehicular emissions on non-cyclone and cyclone days. Additionally, spatial
representations of nitrogen oxides (NOx = NO + NO2), secondary
gaseous pollutants (O3 and PAN), and major aerosol components (OA,
NO3-, and SO42-) during non-cyclone and cyclone days are
shown in Figs. 5 and 6.
Consistent with the meteorological conditions presented above, there is a
contrast in the spatial distribution and separation of pollutants during the
non-cyclone and cyclone situations. Westward transport of emissions was seen
on the non-cyclone (26 July, 2–3 August) days with the separation of
pollutants in the northern and southern latitudes as depicted in Fig. 4a and b
for C2H6 and NH3. Ethane observations indicate that emissions
from O & G, which are concentrated northeast of Denver, were mostly
localized downwind and to the west of the sources during the non-cyclone
periods. NH3 point sources are predominantly concentrated in areas near
Fort Collins and Greeley where a significantly large number of animal and
livestock feeding operations reside. Nitrate production has both an urban
and agricultural component due to oxidation of NOx to HNO3,
the subsequent reaction of HNO3 with gas phase NH3, and the partitioning
of ammonium nitrate into the aerosol phase. These interactions will be
explored further with the ISORROPIA II model in Sect. 3.5. The cyclonic
circulation on 27–28 July transported emissions from point sources in NFR
down to DM (e.g., C2H6 and NH3 in Fig. 4d and e) and concentrated
secondary photochemical products (e.g., O3, PAN, OA, and NO3-
in Figs. 5e, f and 6d, e) in and around the Denver/Boulder metropolitan area compared to the northern counties (Fig. 4d and e). Regional trends in trace gas
and aerosol concentrations during cyclone and non-cyclone periods are
discussed in Sect. 3.3.
Trends in trace gas and aerosol concentrations
Variations in the spatial distribution of pollutants during the cyclone and
non-cyclone events highlight the impacts of numerous sources and meteorology
on air quality and aerosol formation within the Front Range. Here, we
evaluate measurements of several auxiliary gases and aerosol chemical
composition to gain insights on the influence of atmospheric dynamics on
aerosol formation in the three regions of interest in the Front Range.
Spatial distribution maps along the C130 flight track of NOx
[NO + NO2], ozone (O3), and peroxyacetylnitrate (PAN) in
the Colorado Front Range during the non-cyclone (a–c) and cyclone
episodes (d–f). Major highways are shown with black lines, and grey
markers in (a) represent the location of active oil and gas wells in
the region with data available from the Colorado Oil and Gas Conservative Commission.
Spatial distribution maps along the C130 flight track of aerosol
species (OA, NO3-, and SO42-) in the Colorado Front
Range during the non-cyclone (a–c) and cyclone episodes (d–f).
Major highways are shown with black lines, and grey markers in (a)
represent the location of active oil and gas wells in the region with data
available from the Colorado Oil and Gas Conservative Commission.
Gas phase tracers
As discussed in Sect. 3.2., depending on the presence or absence of the
cyclone, trace gases were transported and dispersed differently in the
region. In Fig. 7, the statistical distribution of several gas phase tracers,
namely NH3, C2H6, the sum of C6–C9 aromatics, and CO
measured in the In-Flow, NFR, and DM during the non-cyclone and cyclone
periods are shown. Volatile organic compounds (VOCs) play important roles as
atmospheric precursors to ground-level ozone and SOA (Turpin and
Huntzicker, 1995; Song et al., 2005; Volkamer et al.,
2006; Kroll and Seinfeld, 2008; Hallquist et al., 2009; von
Stackelberg et al., 2013; Riva et al., 2015). The
aromatics highlighted in Fig. 7 represent a subset of the measured VOCs,
typically found in O & G and vehicular emissions, that are known to form
SOA (Ng et al., 2007; Gentner et al., 2012).
During the non-cyclone periods, the mean mixing ratio of NH3
(Fig. 7a) in In-Flow and NFR areas was 13 ± 11 ppbv, while a
significantly lower mean mixing ratio (3.8 ± 3.6 ppbv) was observed in
DM, owing to the high concentration of major ammonia point sources in the
northeastern parts of the Front Range. Additionally, the mean mixing ratio
of C2H6 (Fig. 7b) was higher by a factor of 2–2.6 in NFR
(11.9 ± 8.0 ppbv) compared to the In-Flow (4.6 ± 4.1 ppbv) and DM (6.0 ± 7.8 ppbv), due to substantial density of O & G exploration
activities in NFR. For ∑ C6–C9 aromatics (Fig. 7c), mixing
ratios were higher over DM (∼ 0.4–0.5 ppbv) compared to NFR
(∼ 0.15–0.3 ppbv) during both cyclone and non-cyclone events.
This is in contrast to the pattern observed for C2H6, suggesting
that the emission sources of C6–C9 aromatics are more concentrated
in DM. Similar to ∑ C6–C9 aromatics and consistent with
combustion processes being the dominant source of aromatics and CO, mean
mixing ratios of CO (Fig. 7d) were highest over DM during non-cyclone and
cyclone periods.
Mean mixing ratios of CO over DM during the cyclone were 144 ± 23 ppbv
compared to 110 ± 8.7 ppbv in In-Flow and 114 ± 12 ppbv in NFR.
Additionally, mean values of CO and C2H6 in DM increased during
the cyclone events compared to non-cyclone days (Fig. 7b and d). Since vehicular
sources of CO are concentrated in DM, the slight increase in CO over DM
during the cyclone was likely due to changes in the background CO in the
region and a shallower morning boundary layer on 27–28 July. However, the
increase in C2H6 could be due to the release of emissions into a
shallower morning boundary layer on cyclone days, the cyclonic mixing of air
masses from northern latitudes with higher emissions of C2H6
from O & G operations, or a combination of these two phenomena. The
observed increase in the mean C2H6 mixing ratio in DM during the
cyclone compared to the non-cyclone days was 10.2 ± 6.2 ppbv
vs. 6.0 ± 7.8 ppbv, respectively. To better understand the influence of
O & G operations over DM during the cyclone, we examined the ratio of
i-pentane to n-pentane since O & G emissions show a
characteristic ratio in the range of 0.8–1.2 (Gilman et al.,
2013; Swarthout et al., 2013; Thompson et al., 2014;
Halliday et al., 2016) in contrast to urban sources predominately
impacted by vehicular emissions, which typically have a higher ratio between 2 and 3
(Broderick and Marnane, 2002; Baker et al., 2008). Figure 7e
represents the statistical analysis of the i-pentane to
n-pentane ratio in the threes study regions. Non-cyclone days show
a significant urban source of pentanes in DM compared to NFR. During the
cyclone, a minor decrease in the ratio was observed in NFR, whereas the
ratio decreased substantially in DM to values close to those in NFR. These
observations suggest that the significant increase in the C2H6 mixing
ratio observed over DM during the cyclone cannot be solely explained
by BL height differences but is rather driven by transport of O & G-impacted
and C2H6-rich air masses from NFR into DM. Similarly, cyclonic
transport of NH3 from NFR to DM resulted in a 30 % increase in
average NH3 mixing ratios over DM, from 3.8 ± 2.8 to 8.8 ± 3.9 ppbv, while the mixing ratios in In-Flow and NFR did not change significantly.
Statistical representation of the distribution of gas tracers (NH3,
C2H6, ∑C6–C9 aromatics, CO) and i-pentane to
n-pentane ratios within the three study regions. The box and whiskers
indicate 10th, 25th, 75th, and 90th percentiles, while the solid lines and
circles mark the median and mean values, respectively.
NR-aerosol composition
Average boundary layer values of non-refractory submicron aerosol (NR-PM1)
composition in the Front Range in both non-cyclone and cyclone
episodes are shown in Fig. 8, with the exclusion of Cl- due to average
mass loadings that were below its average detection limit of 0.19 µg s m-3.
Throughout the non-cyclone period, the average mass
concentrations of NR-PM1 aerosols were consistently lower in all
three regions, by a factor of ∼ 2.5. Additionally, the NR
aerosol was dominated by OA (75 %, 3.25 ± 1.45 µg s m-3),
followed by sulfate (13 %, 0.58 ± 0.27 µg s m-3), ammonium
(6 %, 0.28 ± 0.88 µg s m-3), and nitrate (6 %,
0.26 ± 0.27 µg s m-3) (Fig. 8a). During the cyclone events,
OA still dominated NR-PM1 aerosol composition but with a lower
fraction (60 %), while the contribution of nitrate, and correspondingly
ammonium, increased to 16 and 11 %, respectively. It is worth
comparing the current measurements with those made during NFRAQS (summer 1996).
The overall composition of NR aerosols was similar in 1996, with OA as the
dominant species present. However, assuming a conservative organic matter
mass to organic carbon ratio of 1.7 (Turpin and Lim, 2001;
Aiken et al., 2008), OA mass of PM2.5 during 1996 was
estimated to be 7.14 µg m-3, which is more than a factor of 2
higher than the average non-cyclone OA concentration during FRAPPÉ.
Additionally, average concentrations of sulfate and nitrate during the
NFRAQS (summer 1996) were factors of ∼ 2–4 higher than those on
the non-cyclone days of FRAPPÉ. Note that comparison of 1996 vs. 2014
data is not exact due to a higher (PM2.5) size cut of the
1996 measurements. Wintertime measurements during the Metro Denver Brown Cloud Air Pollution Study indicated that aerosol composition was again dominated by OA (68 %), followed by
sulfate (14 %), nitrate (10 %), ammonium (8 %), and chloride (< 1 %).
Shown in Fig. S5 are additional NR-PM1 compositional pie charts for
individual regions (In-Flow, NFR, DM) during the non-cyclone and cyclone
periods of FRAPPÉ. As noted previously, OA was the single dominant
species in all three regions. Relative NR-PM1 composition in In-Flow
was most similar between the non-cyclone and cyclone periods, whereas the relative contribution of NO3- increased during the cyclone period
in NFR and DM at the expense of OA. Represented in Fig. 9a–c are the
observed trends in the NR-PM1 aerosol concentrations (OA,
NO3-, and SO42-) measured in In-Flow, NFR, and DM during
the non-cyclone and cyclone periods. Mass concentrations were consistently
lower in non-cyclone periods for all the measured aerosol species and within
all three regions. On average, there was a 40 % increase in average OA
(Fig. 9a) on cyclone days across all three regions, while the increase during
the cyclone episode was up to ∼ 80 % for DM – an important
consideration for air quality measures. During the non-cyclone days, average
NO3- was slightly higher in NFR (0.43 ± 0.39 µg s m-3)
compared to DM (0.20 ± 0.20 µg s m-3), whereas
during the cyclone episode, average NO3- was a factor of 3.3
higher in DM (2.21 ± 1.44 µg s m-3) compared to NFR
(0.67 ± 0.54 µg s m-3). Overall, average SO42-
(Fig. 9c) mass concentrations also displayed a 2-fold increase across all
regions during the cyclone period. Consistent with the observations for
NH3 and C2H6, significantly larger increases in aerosol mass
concentrations during the cyclone period were observed in DM compared to
NFR, suggesting that mass concentrations during the cyclone were only
slightly impacted by a shallower BL. Instead, transport of precursors and
possibly aerosols from northern latitudes towards DM was the main driver for
the observed increased concentrations in DM. The fact that the highest
aerosol concentrations during the cyclone period were observed in the
greater DM underscores the importance of the impact of local meteorology on
air quality in an area with a large population density.
Photochemical processing
To assess the degree of atmospheric aging in air masses impacted by
combustion, the relationship between primary emitted NOx (sum of nitric
oxide (NO) and nitrogen dioxide (NO2)) and the resulting oxidized
species NOy (sum of NOx + HNO3 + NO3- + ANs + PAN + PPN) was investigated. We utilized the ratio of
NOx to NOy, as a measure of photochemical processing of
NOx-containing air masses. As the ratio approaches 1, the air masses
are considered fresh, while the value for the more aged air masses approaches
0 (Kleinman et al., 2007; DeCarlo et al., 2008; Langridge et al., 2012).
During the non-cyclone and cyclone periods, NOx / NOy ratios were
observed to be highest (0.42 ± 0.25 and 0.26 ± 0.15, respectively)
over DM where freshly emitted plumes from vehicular traffic are dominant
(Fig. 10). Further away from the urban center, NOx / NOy values
decreased with average values of 0.24 ± 0.07 in the In-Flow and NFR
regions. Compared to the non-cyclone periods, during the cyclone events,
NOx / NOy values were similar in NFR, while the average values
decreased by 37 % in the In-Flow and DM regions, indicating further the extent of
photochemical processing of NOx-containing air masses in these regions.
Average chemical composition of mAMS species in all regions during
(a) non-cyclone and (b) cyclone events. Chloride (Cl-),
not shown, was below the instrument detection limit.
Statistical representation of the distribution of the mass concentrations
of aerosol species (OA, NO3-, SO42-) within the three study
regions. The box and whiskers indicate 10th, 25th, 75th, and 90th percentiles, while the solid lines and circles mark the median and mean values, respectively.
One caveat in this analysis may be the impact of lower NOx emissions
during the weekends (26–27 July), resulting in faster photochemistry and
more secondary formation of NOy species and ozone. Several studies in
high-density population areas such as in Los Angeles have investigated the
weekend effect on ambient ozone (Pollack et al., 2012;
Warneke et al., 2013). These studies demonstrate that the higher
ozone mixing ratios observed on weekends compared to weekdays are due to the
significant weekend decrease in NOx emissions from diesel vehicles and
a marginal, if any, decrease in the emissions of non-methane hydrocarbons
from gasoline vehicles, resulting in faster photochemistry, less ozone loss
due to NOx titration, and more rapid ozone production
(Pollack et al., 2012; Warneke et al., 2013). To
examine changes in the weekend NOx emissions in the Front Range, we
utilized the NOy and CO data measured in the boundary layer onboard
the NASA P-3 aircraft during DISCOVER-AQ, which included data from a total
of 8 weekday and 4 weekend flights from 17 July to 10 August. During the
weekends, the NOx to CO enhancement ratio, determined by error-weighted
(5 % for NOx and 2 % for CO) orthogonal-distance regression (ODR)
fits, was lower by a factor of ∼ 1.8 compared to weekdays
(Fig. 11), which is in close agreement with observations made through
aircraft measurements in the Los Angeles basin (Pollack et
al., 2012), indicating a similar decrease in weekend diesel traffic in the
Front Range as in Los Angeles.
Statistical representation of the distribution of the mixing ratios of
NOx / NOy within the three study regions. The box and whiskers
indicate 10th, 25th, 75th, and 90th percentiles, while the solid lines and circles
mark the median and mean values, respectively.
In addition to the weekend change in photochemical processing of NOx,
the meteorological influence of a cyclone may also impact ozone, and
possibly other secondary species, formation. Reddy and Pfister (2016)
indicate that the Denver Cyclone is one of many potential terrain-related
mechanism for limiting area-wide dispersion of O3 and its precursors.
Trace gas spatial distribution maps, provided in Figs. 4 and 5, indeed
indicated strong accumulation of secondary pollutants during the cyclonic
event. Further analysis to investigate the impact of the cyclone on ozone
formation in the Front Range requires chemical box or regional modeling and
is beyond the scope of this paper.
Scatterplot of measured NOy vs. CO using aircraft data from the
DISCOVER-AQ P-3 flights. Weekday (blue dots, 8 combined days) and weekend
(green dots, 4 combined days). Inferred slopes are derived from ODR error-weighted (5 % NOy, 2 % CO) fits.
Scatterplot of OA (µg s m-3) vs. ΔCO (ppbv)
under the most aged air masses (NOx / NOy < 0.5) in DM
for non-cyclone (black) and cyclone (red) days. Slope and intercept values are
based on the ODR error-weighted (30 % OA, 3 % CO) fits, while the correlation
coefficients are based on the linear least-squared regression fits.
Evolution of OA through photochemical aging during the cyclone and
non-cyclone periods was studied in air masses with NOx / NOy < 0.5,
which represent intermediate to strongly processed
NOx-containing plumes. As the plumes age, an increase in the observed
ΔOA / ΔCO ratio suggests SOA production. In this analysis, we
evaluated air masses sampled over DM to determine the extent of
photochemical aging effects on Denver's local air quality. The
error-weighted (30 % uncertainty in OA, 3 % uncertainty in CO) linear
ODR fits to the scatterplots of measured OA against background subtracted
CO were obtained, with the slopes representing the ratios of
ΔOA / ΔCO (Fig. 12). Background CO values (90 and 110 ppbv during
the non-cyclone and cyclone days) were based on the modes of
the Gaussian curves fitted to the frequency distribution plots of CO.
Uncertainties in the slopes represent the propagated uncertainties, i.e.,
the square root of the quadric sum of the relative uncertainties in the ODR
fit, OA concentration, and CO mixing ratio. The average cyclone
ΔOA / ΔCO values were higher (0.060 ± 0.018 µg s m-3 ppbv-1,
r = 0.56) compared to the non-cyclone periods (0.049 ± 0.019 µg s m-3 ppbv-1, r = 0.45), although not
significantly considering the uncertainties associated with the fits.
However, a significantly higher intercept of the fit was obtained on the
cyclone days (5.03 ± 1.52 µg s m-3) compared to the
non-cyclone days (2.05 ± 0.69 µg s m-3), indicating
transport of additional OA relative to CO from the northern latitudes
towards DM during the cyclone events. From an air quality standpoint, such
enhancement in total OA concentration is significant since it is comparable
in magnitude to the average OA over DM during the typical non-cyclone summer days.
Aerosol nitrate production
We assess the regional formation of aerosol nitrate through comparisons of
aerosol nitrate fraction (fNO3 = NO3-/(NO3- + HNO3))
in the In-Flow, NFR, and DM regions with and without the
cyclone influence (Fig. 13a). Low fNO3 values observed in NFR and
DM regions during the non-cyclone days indicate that nitric acid was
predominantly present in the gas phase. In contrast, higher fNO3
values observed during the cyclone suggest increased partitioning of
nitric acid to the condensed phase. As noted earlier, environmental factors
including relative humidity, temperature, and atmospheric dynamics play
important roles in the formation of aerosol nitrate (Stelson et
al., 1979; Stelson and Seinfeld, 1982; Watson, 2002). Slightly lower
temperature and increased RH were observed in NFR and DM during the
cyclone period (Table 1). Higher RH may enhance formation of nitrate
aerosols by promoting aqueous and heterogeneous phase reactions and
increasing the equilibrium partitioning of gas phase NH3 and HNO3
to the condensed particle phase (Stelson et al., 1979; Stelson
and Seinfeld, 1982; Volkamer et al., 2006; Na et al.,
2007; von Hessberg et al., 2009). Moreover, local meteorology during
the cyclone period, facilitating transport of NH3 from the nearby
feedlots in NFR to DM (Sect. 3.3.1, Fig. 4b and e), could have favored
equilibrium partitioning of nitric acid to the aerosol phase due to
abundance of gas phase NH3.
Statistical representation of the distribution of (a) aerosol
nitrate fraction (fNO3 = NO3-/[NO3- + HNO3]
and (b) aerosol optical extinction within the three studied regions
during non-cyclone and cyclone periods. The box and whiskers indicate 10th,
25th, 75th, and 90th percentiles, while the solid lines and circles mark the
median and mean values, respectively. Modeled fNO3 values with actual
inputs of chemical composition and T and RH are shown with green diamonds, while the predicted values with the non-cyclone composition and cyclone T
and RH are shown with blue stars.
To further investigate the role of atmospheric conditions and mixing
patterns in aerosol nitrate formation during the cyclone days, nitrate
partitioning was evaluated by ISORROPIA II (Fountoukis and Nenes, 2007)
model calculations, described in Sect. 2.4. The predicted partitioning
results, summarized in Table S1 in the Supplement and Fig. 13a are in reasonable agreement
with the observed fNO3 values on non-cyclone and cyclone days. Over DM,
the model predicted 24 % more nitrate existing in the aerosol phase
compared to mean values based on the measurements; however, the predicted
fNO3 is still within the limits of variability of the observed
fNO3. To evaluate the influence of RH and T on aerosol nitrate
formation, we considered model input variables based on the non-cyclone
concentrations, while prescribing the higher RH and lower T values
representing conditions of the cyclone period (Table S1). In this case, the
model predicted similar fNO3 values in NFR and significantly lower
fNO3 over DM compared to the measurements, indicating that the observed
higher partitioning of nitrate to the aerosol phase during the cyclone
events was not mainly driven by changes in ambient T and RH, but rather it was due to increased availability of NH3 over DM with the cyclonic
transport from NFR.
Mass extinction efficiency plots of βext against total
NR-PM1 mass for NFR and DM during (a) non-cyclone and (b) cyclone
episodes. Inferred slopes are derived from ODR error-weighted (10 % βext,
30 % mAMS total mass) fits.
We further evaluated the influence of sulfate concentrations and ambient RH
to understand how chemical composition and environmental changes in DM could
impact nitrate partitioning between gas and aerosol phases (Table S2). While
keeping T, RH, gas phase ammonia, and ammonium associated with nitrate at the
same level as in the baseline (i.e., observations on cyclone days over DM),
the absence of aerosol sulfate results in a drastic increase in fNO3,
with almost all of the nitric acid partitioning to the aerosol phase. This
result indicates that background aerosol sulfate concentrations have a
strong effect on equilibrium partitioning of nitric acid. Next, we evaluated the influence of RH, keeping all other variables the same as in the baseline.
Increasing RH from 64 to 85 % resulted in an increase in fNO3
of 0.36 to 0.74, while decreasing RH to 35 % decreased fNO3 by a
factor of 3.6. Taken together, these case scenarios suggest that
meteorological transport patterns, background sulfate concentrations, and RH
all have significant influences on the phase equilibrium of nitric acid and
aerosol nitrate formation. Although the Denver metropolitan are is not typically in
violation of the PM2.5 standard during summer months, higher aerosol
nitrate concentrations may be observed in the presence of a cyclone and with
RH values higher than what was observed during this study.
Impacts on optical extinction
Several studies have discussed the importance of nitrate-containing aerosols
for optical extinction (βext) coefficients, i.e.,
scattering and absorption of light, that impede visibility in affected
regions (Tang, 1996; Watson, 2002; Li et al., 2009;
Langridge et al., 2012; Zhang et al., 2012; Lei and
Wuebbles, 2013). As seen in Fig. 13b, average βext values
measured during FRAPPÉ (λ = 632 nm) were similar in the In-Flow,
NFR, and DM region during non-cyclone days with an average of 10.6 ± 3.5 M m-1,
whereas factors of 1.5–3 increase in the average βext were
observed during the cyclone periods, with the most significant impact observed over DM.
Mass extinction efficiency (MEE) values, defined as the slopes of the
error-weighted (10 % for βext, 30 % for NR-PM1 mass)
linear ODR fits of βext against total NR-PM1 mass, were
compared in Fig. 14. MEE values under the non-cyclone events in NFR and DM
were 1.92 ± 0.62 m2 g-1 (r = 0.71) and 2.72 ± 0.87 m2 g-1
(r = 0.62), respectively, higher by 42 % in the urban
center. During the cyclone events, MEE values were 43 % higher over DM
compared to NFR (2.85 ± 0.90 m2 g-1 (r = 0.84) and
2.00 ± 0.66 m2 g-1 (r = 0.88), respectively) but similar to
the percentage increase observed during the non-cyclone days. On cyclone
days a significant increase in the average mass concentrations of the
aerosol species was noted (Fig. 8). However, the similarity of the MEE percentage
increase in DM during the cyclone and non-cyclone days suggests that the
increase in NR-PM1 mass during the cyclone accompanied a similar
increase in βext and that MEE alone cannot provide detailed
insights on the impact of the cyclone on βext in DM.
Summary table of βext measurements from the Colorado
Department of Public Health and Environment (CDPHE) long-path transmissometer
in downtown Denver for each of the 5 days of interest. On the Visibility Standard
Index Scale, a value of 101 equates to 76 M m-1 standard. Values between 0 and 50
are described as good, those between 51 and 100 as moderate, those between 101 and 200 as poor, and those above 201 as extremely
poor visibility. NA = not available.
Date
Hour
βext
RH
4 h avg.
VSI
Descriptor
(MST)
(M m-1)
(%)
(M m-1)
26 July 2014
11:00
70
33
59
66
Moderate
12:00
50
29
59
66
Moderate
13:00
50
30
57
64
Moderate
14:00
51
32
55
60
Moderate
15:00
45
33
49
49
Good
27 July 2014
11:00
124
66
139
NA
NA
12:00
108
61
125
NA
NA
13:00
95
53
113
NA
NA
14:00
110
49
109
145
Poor
15:00
112
47
106
141
Poor
28 July 2014
11:00
118
56
118
NA
NA
12:00
108
51
112
149
Poor
13:00
84
46
104
138
Poor
14:00
78
43
97
129
Poor
15:00
88
43
90
119
Poor
2 August 2014
11:00
38
33
44
43
Good
12:00
37
31
42
40
Good
13:00
33
23
38
35
Good
14:00
29
21
34
30
Good
15:00
32
21
33
28
Good
3 August 2014
11:00
53
40
62
72
Moderate
12:00
44
34
57
63
Moderate
13:00
41
29
50
50
Good
14:00
37
25
44
42
Good
15:00
37
24
40
37
Good
As mentioned previously, the State of Colorado visibility standard has set a
threshold of 76 M m-1 averaged over a 4 h period when RH < 70 %.
To more directly investigate how the Denver Cyclone impacted
visibility in DM, we refer to the CDPHE LPV-2 long-path transmissometer
measurements of ambient βext at 550 nm in downtown Denver
during 11:00–15:00 MST (Table 2). During the non-cyclone days (26 July,
2–3 August), the 4 h average values βext (550 nm) were
33–62 M m-1, well below the visibility standard. However, during the
cyclone days (27–28 July), 4 h average βext (550 nm) values were
90–139 M m-1, up to a factor of ∼ 2 higher than the
standard, resulting in poor ratings with respect to the visibility
standard index (VSI).
To further understand the role of different aerosol components in driving
the observed increase in airborne measurements of βext
(632 nm), correlations between βext (632 nm) and NO3-, OA,
and SO42- mass under the influence of non-cyclone and cyclone air
masses were examined (Fig. 15). During the non-cyclone events, βext
displayed strong correlations (r = 0.71) with OA and
NO3- in NFR and only with OA (r = 0.70) in DM. βext
was poorly correlated with sulfate aerosols in the region during the
non-cyclone events (r = -0.18, 0.11, for NFR and DM, respectively). During
the cyclone events, all aerosol components equally influenced
βext in NFR (r = 0.88, 0.84, 0.87), while only strong
correlations with NO3- (r = 0.86) were observed in DM.
These results indicate that the Denver cyclone directly influenced
visibility in DM by facilitating transport of an additional aerosol
precursor (i.e., NH3) to the region compared to the non-cyclone events
(detailed analysis in Sect. 3.5).
Correlation coefficients of scatterplots of βext against
individual aerosol species for NFR and DM during (a) non-cyclone and
(b) cyclone episodes.