Aircraft study of the impact of lake-breeze circulations on trace gases and particles during BAQS-Met 2007

K. L. Hayden, D. M. L. Sills, J. R. Brook, S.-M. Li, P. A. Makar, M. Z. Markovic, P. Liu, K. G. Anlauf, J. M. O’Brien, Q. Li, and R. McLaren Air Quality Research Division, Environment Canada, Toronto, ON, Canada Department of Chemistry, University of Toronto, Toronto, ON, Canada Cloud Physics and Severe Weather Research Section, Environment Canada, Toronto, ON, Canada Meteorological Service of Canada Operations-Ontario, Environment Canada, Toronto, ON, Canada Centre for Atmospheric Chemistry, York University, North York, ON, Canada retired


Introduction
The temporal and spatial structure of the boundary layer and lower troposphere near coastlines can be complex due to radiative property differences between land and water. This can be an important feature affecting the transport and transformation of air pollutants in such regions. Temperature gradients between cooler air over water and 5 warmer air over land result in a pressure gradient that can initiate and sustain a lake (sea) breeze during the day and a land breeze at night. During the day, air over a lake moves inland in a shallow inflow layer (typically < 500 m) and air aloft over land moves offshore in a return flow to replace air from over the lake. As the cooler lake air moves over the warmer land surface, a thermal internal boundary layer (TIBL) is 10 created which grows in height with an increase in inland distance (Lyons and Olsson, 1973;Garratt, 1990). At the leading edge of the lake breeze, air is forced upwards at the convergence zone (lake-breeze front) that separates the cooler lake air from the warmer air inland. The passage of a lake-breeze front is often characterized by increased upward motion, enhanced moisture and wind shear, decreased temperatures 15 and changes in wind direction and speed (Lyons, 1972). Due to the upward motion of air at the lake-breeze front, a line of cumulus clouds may form along the frontal zone. At night, the temperature gradient is reversed and a land breeze is formed.
Numerous studies have identified the importance of lake (sea) breeze circulations to air quality in coastal urban areas (e.g. McElroy and Smith, 1986;Lu and Turco, 1994 mode. Typically, fragments of mass to charge ratio (m/z) < 300 in the MS mode are used to determine mass concentrations of sulphate, nitrate, ammonium and total organics (Allan et al., 2003;Jimenez et al., 2003). During the study, continuous power was supplied to the instrument so that the pumps could provide constant vacuum and maintain a low background signal. On many occasions, the C-AMS signals were af-5 fected by low-level electronic noise, likely from cell phone towers in the area. The noise was most prevalent at mass fragments of m/z > 100 amu, masking signals that are normally due to organics, if present. During periods of no cell phone interference, the organic mass determined with fragments m/z < 300 amu was only 2.4 ± 1.3% higher than the organic mass determined using a limited range of fragments, m/z < 100 amu. 10 Therefore, since the organic mass including fragments m/z > 100 amu was small, it was ignored in the organic mass determination. Five ionization efficiency calibrations were performed prior to and during the field campaign and showed very low variation (<9%). Detection limits for 30-s sampling duration were determined to be 0.270, 0.024, 0.014 and 0.073 µg m −3 for organics, SO −2 4 , NO − 3 and NH + 4 , respectively.

Aircraft sampling inlets and particle collection efficiency
Trace gases were sampled through a rear-facing 0.25 inch O.D. Teflon tube. Ozone, SO 2 and NO/NO 2 were sampled through a pressure-controlled inlet that maintained a set-point pressure of ∼1 atm during flight to avoid pressure variations in these systems. The Aerolaser CO instrument did not sample through the pressure-controlled inlet. 20 Particles were sampled through a forward-facing near isokinetic stainless steel diffuser particle inlet. The inlet was mounted on the roof of the Twin Otter toward the front of the aircraft. Theoretical calculations taking into account the inlet dimensions, volume flow and velocity indicated 95% transmission efficiency for PM 1 through the inlet. Air was drawn through a 0.5 inch O.D. stainless steel tube using controlled rapid air move-25 ment (RAM) from the aircraft at ∼70 LPM. The PSAP, CNC and C-AMS subsampled from this manifold. The residence time from the inlet tip to the AMS, CNC, and PSAP was <1 s. Sub-micron particles (0.120-2 µm) were measured with the PCASP that was mounted in an under-wing pod (Liu et al., 1992 The particle collection efficiency (CE) of the C-AMS for ambient measurements is typically determined through comparisons with other particle chemical measurements such as with the Particle Into Liquid Sampler-Ion Chromatography instrument (PILS-IC) and/or with mass estimated from instruments such as Scanning Mobility Particle Sizers (SMPS). The CE is a function of the particle transmission through the aerodynamic 5 lens, the efficiency of particles being focussed by the lens and directed onto the vaporizer, and the extent to which particles bounce off the vaporizer. Many researchers have determined a CE ∼ 0.5 for ambient particles (e.g. Dunlea et al., 2009;Kleinman et al., 2007;DeCarlo et al., 2008;Drewnick et al., 2004;Allan et al., 2004). However, higher CE values between 0.5 and 1 were found for particles dominated by ammonium nitrate 10 and for acidic particles (Quinn et al., 2006;Kleinman et al., 2007). During BAQSMet, the aerosol neutralization ratio (ANR) defined as the molar ratio of ammonium to (sulphate + nitrate) indicated that the particles were neutralized most of the time with the exception of data taken in high concentration sulphate plumes, where the particles had not yet been neutralized by NH 3 . A comparison of mass concentrations from the C- 15 AMS with those estimated from the PCASP indicated a CE of 0.5 for neutralized particles (ANR ∼ 1.0), but approached 1.0 for acidic particles (ANR < 0.5). Therefore, the C-AMS data in this study were adjusted using CE = 0.5 for neutralized particles transitioning linearly to a CE = 1.0 for acidic particles. A pressure controlled inlet (PCI) was used in front of the AMS to remove variations in particle sizing and transmission due to 20 pressure changes in the aerodynamic lens of the AMS (Jayne et al., 2000;Hayden et al., 2008;Bahreini et al., 2008). In the PCI, a low pressure region, between a 200 µm orifice upstream of the AMS inlet and a 130 µm orifice in the AMS inlet (replacing the standard 100 µm orifice) was maintained at a set point of 470 torr. This low pressure region was variably pumped so that the inlet pressure of the AMS was maintained at 25 1.3 torr. Under this configuration, no corrections to particle sizing were required for altitudes up to 3000 m a.g.l. during this study. Transmission efficiency experiments performed during and after the study indicated that particles were transmitted through the PCI with 100% efficiency.

Model description
The Unified Regional Air-quality Modelling System (AURAMS) air quality model was used to support the BAQSMet field study. Details of the model are provided in Makar et al. (2010), but a brief description is presented here. The model has three main components: (1) a prognostic meteorological model GEM (Global Environmental Multiscale 5 model: Côté et al., 1998); (2) an emissions processing system, (Sparse Matrix Operator Kernel Emissions: Houyoux et al., 2000;CEP, 2003); and (3) an off-line regional chemical transport model, the AURAMS Chemical Transport Model (CTM: cf. Cho et al., 2009;Makar et al., 2009;Gong et al., 2006 analysis of pollutant behaviour and processing during a lake-breeze event. In addition to the AURAMS simulations, high resolution forward and backward trajectories were calculated using the Canadian Meteorological Centre (CMC) trajectory model (D'Amours et al., 2001). This model made use of the 2.5 km resolution, 2 min timestep GEM wind fields that were also used to drive AURAMS at its highest reso-25 lution. Back trajectory endpoints and forward trajectory starting points were chosen to help elucidate the flow patterns associated with the driving meteorology, as will be discussed in more detail.

Results and discussion
A brief summary of meteorological sampling conditions on aircraft flight days is provided in Table 1. A detailed analysis of the meteorology and lake breezes observed throughout the study has been described elsewhere (e.g. OME, 2008;Sills et al., 2011;Slowik et al., 2011). The synoptic wind flow was determined from the high resolution 5 back trajectories computed for air parcels arriving into the study region at 500 m a.g.l.
Large scale synoptic flow varied considerably, and lake breezes were identified on each day (Sills et al., 2011). The extent to which the lake-breeze circulations were deformed by the synoptic wind was used to classify lake breezes as Low Deformation (LD), Moderate Deformation (MD) or High Deformation (HD) (Sills et al., 2011). to moderate southwesterly synoptic flow that contributed to a poor air quality episode between 24-27 June. On 25 June, lake breezes developed in association with each of the lakes in the region and aircraft flights (Flights 4 and 5) were designed to sample 15 across an entire LSC lake-breeze circulation at multiple altitudes. Lake-breeze fronts are shown as magenta lines in Fig. 1. The lake-breeze circulations on this day were typical of LD situations that occurred several times throughout the study period (Sills et al., 2011). In this paper, a complete lake-breeze circulation on 25 June is analyzed. Hourly surface mesoscale analyses (see Sills et al., 2011) are used as a guide to 20 identify signatures of lake-breeze fronts in the aircraft meteorological data and compared with spatial and vertical changes in pollutant concentrations. Cross-sectional plots constructed from the aircraft data are analyzed in conjunction with model simulations to interpret pollutant transport and processing within the lake-breeze circulations. Conceptual models of the lake-breeze circulation are proposed followed by estimates Introduction   Figure 2 also shows that the back trajectories are sensitive to perturbations in air mass flow due to lake breezes. For example, over the Harrow site, the upwind flow was westerly, with transport of emissions from the west end of Lake Erie, but after the onset 25 of the lake breezes (11:00 LT), the back trajectories switched to southerly closer to the site. This is consistent with surface measurements at Harrow that showed a change in wind direction from westerly to southerly at this time. end of LSC also shows a change in direction from westerly to northeasterly in response to the LSC onshore lake breeze.

Lake-breeze identification
Mesoscale analyses for each hour (Sills et al., 2011) and the inland progression of the lake breezes from 12:00 to 17:00 LT p.m. on 25 June 2007 is presented in Fig. 1. The 5 positions of the lake-breeze fronts are shown as magenta lines. Lake breezes on this day were first detected at 11:00 LT and persisted until 21:00 LT. At 12:00 LT (Fig. 1a), lake-breeze fronts were detected along the south and north shores of LE, as well as almost all the way around LSC. The north shore LE front progressed northward toward LSC until the LE and LSC fronts merged along the south shore of LSC at 15:00 LT ( Fig. 1d). The merged fronts then remained quasi-stationary until after 17:00 LT. On 25 June 2007, two of the three aircraft flights sampled across the LSC lakebreeze circulation. Aircraft tracks for Flights 4 and 5 are shown in Fig. 1a and f respectively, for the corresponding time interval in which they occurred. Flight 4 included horizontal transects between the north shore of LE (near the Harrow supersite) and near 15 Lambton (north of the Sombra site) at 1560 m a.g.l. (free troposphere), 800 m a.g.l. and 300 m a.g.l. Flight 5 included a similar flight pattern with multiple transects at altitudes of 2600 m a.g.l., 1560 m a.g.l., 800 m a.g.l., 460 m a.g.l. and 300 m a.g.l. At the beginning of Flight 5, the aircraft performed a vertical spiral over the LSC buoy extending from 200-2600 m a.g.l. Using the analyzed surface lake-breeze front locations as a guide, 20 the aircraft meteorological and chemical data were examined to identify signatures due to lake breezes. On 23 June, low deformation lake breezes also occurred with even stronger lake-breeze convergence zones (Sills et al., 2011). Although, pollutant concentrations were low during this period of northwesterly synoptic flow (Table 2), aircraft data on this day were similarly analyzed for lake-breeze boundaries. It was found that 25 on June 23 sharp changes in the dewpoint temperature and sustained upward gust velocities were good indicators of lake-breeze front crossings. This provided confidence that using these parameters to identify lake-breeze fronts on 25 June was applicable. This approach was also consistent with an independent detailed analysis of all flight data.

Aircraft observations of trace gases, particles and meteorology
The wind direction, dewpoint temperature, vertical gust velocity, CO, SO 2 , particulate SO 2− 4 , and particulate organics (OA) are shown for Flights 4 and 5 in Figs. 3 and 4, 5 respectively. The bottom axis is the distance between Lake Erie (LE) and Lambton (just past Sombra) (see Fig. 2 for locations), the blue horizontal bar represents LSC and the grey arrow indicates the aircraft flight direction. The blue and green arrows indicate the position of the lake-breeze fronts as identified from the aircraft data. The light blue boxes are selected time slices that are discussed in Sect. 3.7. Although the 10 fronts above the surface were likely not in the same horizontal location as at the surface because of sloped frontal zones, the difference was likely small, i.e. <1 km. Thus, the mesoscale surface analyses were used to guide the front locations in the aircraft data. An additional complication considered in the analysis was the movement of the fronts over time i.e. between the 800 and 300 m a.g.l. aircraft passes, the LE front migrated 15 northward approximately 10 km. In Flight 4, the 1560 m a.g.l. track was in the free troposphere above an estimated 1200 m a.g.l. synoptic inversion height (Fig. 3a1, a2). CO mixing ratios were <140 ppbv and SO 2 was below detectable levels. Particle mass concentrations of OA and SO 2− 4 ranged between 0.5-3 µg m −3 . These observations are indicative of homogeneous regional background air. In the 800 m a.g.l. transect, 20 travelling from LE to Lambton (Fig. 3b1), the LE lake-breeze front can be clearly identified by a 2 • C increase in the dewpoint from 18 • C to 20 • C (blue arrow at 11:56 LT), which coincides with the position of the lake-breeze front based on analyses of surface meteorological data. The increase in the dewpoint is due to updrafts at the front bringing up moister air from the surface. There was also a sudden increase in the aircraft vertical gusts (Fig. 3b1) and a visible line of clouds. At this altitude the aircraft was below the synoptic inversion, but above a shallower LE lake-breeze inflow layer (on Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | the lake side of the front), and thus, the relatively low chemical levels ( Fig. 3b2) reflect "residual" air or remnants of the previous day's convective boundary layer (Sills et al., 2011). The aircraft then crossed the LE front and measurements showed an increase in primary and secondary pollutants consistent with a polluted convective boundary layer. Elevated pollutant concentrations were observed over land from the LE lake-5 breeze front to the south shore of LSC. Evidence of the south shore LSC lake-breeze front at 800 m a.g.l. was more difficult to identify based on meteorology alone. Since CO is a relatively long-lived species in the troposphere, these measurements were examined for perturbations related to lake-breeze boundaries. Guided by the location of the surface lake-breeze front, a brief, sharp change in CO of 24 ppbv (green solid 10 arrow at 12:01 LT), was observed ( Fig. 3b2). This was coincident with in increase in the vertical gusts ( Fig. 3b1). On the LSC side of the LSC front, pollutants remained elevated followed by a sharp decrease over the south shore of LSC. The identification of the lake-breeze front north of LSC (green dashed arrow) is based on a change in the dewpoint from 21 • to 19 • C (12:12 LT); there were no discernible features suggesting 15 the presence of a surface-based front. At Lambton, the aircraft turned around and travelled southward toward Harrow at 300 m a.g.l. A LSC front north of LSC was not detectable in the aircraft data (Fig. 3c1,c2). As the aircraft passed over the south shore of LSC at 12:35 LT, the wind direction changed from southerly to northeasterly (Fig. 3c1) Fig. 1 for site locations). This is due to the onshore LSC lake breeze and demonstrates that the aircraft was successful in penetrating and sampling in the shallow lake-breeze inflow layer at 300 m a.g.l. Highest pollutant concentrations were be- Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | flight path. Concentrations were also observed to decrease sharply over the south end of LSC. Sharp changes in the dewpoint of ∼1 • C signified the presence of the LE (12:39 LT) and LSC (12:37 LT) lake-breeze fronts, as supported by the surface mesoanalyses. The LSC front was also identified in the aircraft data by the onshore wind direction. Between Harrow and the LE lake-breeze front, plumes of SO 2 and SO 2− 4 5 were observed having been advected from over LE. During the time between Flight 4 and Flight 5, the LE lake-breeze front pushed north ∼35 km and merged with the LSC front along the south shore of LSC by 15:00 (Fig. 1d). As a result, the CBL air between LE and LSC was replaced with lake air leaving the study region being largely influenced by lake-breeze circulations and lake-modified air 10 masses. The first two transects in Flight 5 at 2600 m a.g.l. and 1560 m a.g.l. were in free tropospheric air. Figure 4a1, a2 shows the measurements from the transect at 1560 m a.g.l. indicating that the predominant flow was still from the southwest. As in Flight 4, low pollutant concentrations at the higher altitudes are believed to reflect regional background air. The measurements in the remaining transects ( Fig. 4b-d) 15 reflect residual air influenced or modified by lake-breeze return flow. Multiple, positive peaks in the updraft velocity at the three altitudes within the residual layer (16:36 LT (800 m a.g.l.), 16:43 LT (460 m a.g.l.), 17:20 LT (300 m a.g.l.)) are interpreted as updrafts along the merged LE/LSC lake-breeze front. Pollutant concentrations in this later afternoon flight were generally lower and less variable compared to Flight 4, but 20 elevated pollutant concentrations were observed north of LSC, possibly from the CBL between LE and LSC, or pollutants from over LSC earlier that were advected northward and then mixed upward. North of LSC, SO 2− 4 was >15 µg m −3 and CO, in particular, was variable ranging between 180-390 ppbv indicating a non-homogenous air mass. As the aircraft went further north just past Sombra, the measurements were likely influenced by power plant emissions from the Lambton area that were advected northeast in the southwesterly flow.

Distribution and evolution of trace gases and particles
Data from the aircraft transects, averaged to 10 s, and measurements from eight surface sites were interpolated using an ordinary linear kriging method (GS + ver. 5.1.1, Gamma Design Software) to produce cross-sectional plots of CO, SO 2 and vertical winds (Figs. 5,6). Overlaid on the plots are the interpolated plan-view horizontal wind 5 directions (along with the corresponding orientation of the north arrow) and the size of the arrows represents the magnitude of the wind speed. The vertical wind component is not included in the wind direction, but is shown separately in the third panel of the figures to highlight the measured vertical motions. It was assumed that physical and chemical processes were sufficiently stable to build 2D cross sections from the aircraft 10 transects. CO and SO 2 were chosen because they are both good tracers for different emission sources, both had fast instrument time responses (Table 2) and due to its 1 s time-response, CO appeared to be a good chemical tracer for lake-breeze motions. The surface sites chosen for this analysis were within 12.4 km horizontal distance of the Flight 4 and 5 mean flight path and included Harrow (3.3 km), Cottam (1.3 km), 15 Essex (5.7 km), Woodslee (0 km), Lighthouse Cove (12.4 km), the LSC Buoy (3.8 km), Bear Creek (11.4 km) and Sombra (0.8 km) (Fig. 1a, f). The sites are indicated as black circles at the bottom of Figs. 5 and 6. In Fig. 5, the LE lake-breeze front separates LE air and polluted CBL air. In Fig. 5c, a return flow associated with this front is not present, but an area of reduced wind 20 speed (red circle) suggests that it may have been masked by the larger synoptic flow with a resulting net flow along the axis of the transect (southwesterly). Updrafts associated with convective motion were observed in the CBL, and a significant downdraft was measured on the LSC side of the LSC lake-breeze front. The synoptic inversion is estimated to be ∼1200 m a.g.l. In Fig. 5, CO and SO 2 were observed to be well mixed LE and LSC. A horizontal gradient in CO mixing ratios was observed with highest levels further north compared to SO 2 mixing ratios which were elevated in the entire region between LE and LSC. This is consistent with the back trajectories ( Fig. 2) that show air masses that arrived near Harrow had crossed over SO 2 emission sources at the west end of LE, whereas other trajectories arriving at points further north of Har-5 row also crossed over the Detroit/Windsor urban area. At the Harrow supersite, SO 2 was observed to be "plume-like"; elevated SO 2 was observed between 10:00-11:20 LT with mixing ratios reaching >100 ppb. Although lake-breeze circulations were present around LE with onshore flow near Toledo, in order for plumes of SO 2 to travel northward, they were either lofted over top of the lake-breeze inflow and/or were in transit 10 prior to the formation of the lake breezes. Based on aircraft measurements, the spatial extent of the SO 2 plumes is estimated to range between 7.0-13.5 km horizontally; this indicates that the plumes were remarkably intact suggesting minimal dispersion, and thus mostly transported over the lake. Lyons and Pease (1973) found that plumes from power plants travelled long distances over Lake Michigan with little dilution and 15 attributed this to the absence of convective mixing over the lakes. It might be expected that if the plumes had travelled over top of the lake-breeze inflow, they would have undergone some mixing and dilution. The back trajectories indicate a 4-6 h transit time from the west end of LE to arrive near Harrow (Fig. 2) and since this is prior to the onset of the lake breezes, the likely explanation for the SO 2 was that the plumes were already 20 in transit and not affected by local lake breeze mixing at the western end of LE. In addition, the observations from the later afternoon flight do not show an influence from SO 2 emissions (except north of LSC from other sources, Figs. 4, 6) suggesting that the lake-breeze onshore flow near Toledo, which was still present at this time ( Fig. 1), impedes transport to the study region even though the large scale flow continues to be 25 southwesterly. Elevated mixing ratios of CO and SO 2 were also observed on the LSC side of the LSC lake-breeze front (Figs. 3,5). The sharp decrease of pollutant levels at the LSC shore points toward important differences between lake and land surfaces and Introduction influences from the lake-breeze circulation. It is hypothesized that pollutants in the CBL were lofted upward at the LSC lake-breeze front, transported northward in the synoptic flow, transported in the downdraft on the north side the front, and then confined by the LSC onshore flow along the south shore of LSC. This is plausible since some of the CBL air closest to the LSC front would be expected to be advected in the 5 upward vertical motion at the frontal convergence zone, and the downdraft that was measured behind the front (Fig. 5c) would draw air from above. This would result in the transport, at least partially, of air from the CBL by the downdraft behind the LSC front. The pollutants are then confined along the LSC shoreline by the onshore lake breeze. The development of a thermal internal boundary layer (TIBL) over land behind 10 the lake-breeze front would limit vertical mixing; as the TIBL grows vertically with inland distance it does not get as high as the fully developed CBL in the short distance between the lake shore and the front. Figure 3b2, c2 shows that CO and SO 2 behind the LSC front are less vertically developed than in the CBL supporting this theory. The lake-breeze effect of confining pollutants is also evident at the surface. During 15 the time of Flight 4, significant levels of SO 2 (> 50 ppb) and SO 2− 4 (>35 µg m −3 ) were observed by the CRUISER mobile laboratory from 12:40-13:10 LT at Mitchell's Bay, located on the shoreline at the eastern edge of LSC (see Fig. 1) (Chan et al., 2011). Also along the eastern shoreline, at Bear Creek, SO 2 was 15-20 ppb from 11:00-12:30 LT. There was no evidence of any significant pollutant buildup north of LSC, attributed to 20 the weaker offshore lake-breeze circulation compared to the south and east sides of the lake. Mechanisms to transport pollutants within a lake-breeze circulation have been identified in previous studies (Lyons, 1972;Lyons and Cole, 1976;Sills et al., 1998) and are known to be effective in confining pollutants to coastal regions (Simpson, 1994;Lu and Turco, 1995). Dynamics at the lake-breeze front have been studied to understand 25 effects on pollutants. Hastie et al. (1999) noted that the arrival of the lake breeze coincided with maximum trace gas concentrations with rapid decreases behind the front suggesting that vertical mixing generated by the front could result in entrainment of cleaner air from above the inflow layer. Kitada and Kitagawa (1990)  vertical profile of pollutants was strongly impacted by the micro-scale features of the sea breeze such as the transport of pollutants in the downdraft behind the front and suggested that the most aged air mass would be found in the upper part of the circulation behind the front. Since lake and land breezes are quasi-closed circulations and pollutants emitted into them can be recirculated (Lyons, 1972), it might be expected 5 that air in the downdraft of the LSC lake-breeze front would exhibit characteristics of a more aged air mass relative to the air in the CBL. The impact of lake breezes on air mass processing is discussed further in Sect. 3.7. In Fig. 6, a signature of the merged LE/LSC front is not apparent, but surface analyses show that the front was still present just offshore at 17:00 LT and by 19:00 LT 10 the LSC lake breeze moved back onshore, with a shift to onshore winds and a rapid increase in dewpoint. The height of the synoptic inversion, determined from a vertical profile performed over the LSC buoy , was 1300 m a.g.l. It is noted that strong east-southeast winds were measured above the merged front at the 2600 m a.g.l. level which is consistent with 08:00 LT and 20:00 LT DTX soundings show- 15 ing winds with an easterly component between 2 and 3 km. In Flight 5, pollutant concentrations, compared to Flight 4, were more homogeneous across the flight transect. In Fig. 6, with the exception of the north end of the flight track near Lambton, CO and SO 2 levels were lower and less variable compared to Flight 4. The region is largely enveloped in lake-modified air and pollutants in the CBL observed in Flight 4 appear to 20 have been displaced by the merging LE/LSC lake-breeze fronts. Thus, by the late afternoon, only low pollutant concentrations remained between LE and LSC. The region having been being fed by regional lake air from the south is expected to exhibit characteristics of a more aged air mass and is discussed further in Sect. 3.7. It is speculated that pollutants in the CBL between LE and LSC in Flight 4 were advected upward and carried northward by the southerly flow and/or a LSC return flow resulting in increased concentrations of pollutants northeast of LSC. Pollutants would be transported to the surface by land-based convection.

Analysis of model-generated lake-breeze circulation
A more comprehensive view of the lake-breeze circulations is possible through analysis of model data. In Fig. 7, model-predicted SO 2 mixing ratios overlaid with wind direction are shown at 12:00 LT for altitudes of (a) 285 m a.g.l. and (b) 815 m a.g.l. The model altitudes are the closest available to the 300 and 800 m a.g.l. aircraft transects, 5 and the model simulation time corresponds to the middle of Flight 4. After 15:00 LT, the model predicts winds shifting to come from the southeast, while the observations indicate a continuation of southwesterly flow (Figs. 4 and 6). As such, this discussion is reserved to that of the Flight 4 time period because the model in the later afternoon is not simulating the synoptic wind direction adequately. Figure 7a shows two zones of 10 convergence (solid magenta lines) corresponding to the model's predicted locations of the LE and LSC fronts across the Flight 4 aircraft transect (red dashed line). Figure 7b shows a region of divergence (dashed magenta lines) along the north shore of LE in response to the convergence at lower levels. The modelled wind direction along the aircraft transect is quite similar to the measured winds shown in Fig. 3b1 and c1 for 15 the corresponding altitudes. This provides confidence that the model is correctly simulating both the synoptic flow and lake-breeze circulation at this time. Figure 7 shows that high concentrations of SO 2 are located over LE (and over the Lambton area to the north-east of LSC); these pollutants originated from sources on the west end of the lake and were advected into the study region by 12:00 LT, the Flight 4 time period. 20 In Fig. 8 southeastern LSC surface towards Detroit. Then air moves east-northeast back over LSC at ∼800 m a.g.l., moves downward in the downdraft over LSC, moves ashore in the lake breeze and returns to Detroit at ∼300 m a.g.l.; the black arrows in Fig. 8a  to be south of LSC with CO and SO 2 well-mixed up to 1500 m a.g.l., consistent with the observations. However, the model shows elevated CO mixing ratios in a narrow region at the LE lake-breeze front. Also, the model has high levels of CO, SO 2 and SO 2− 4 at altitudes above 600 m a.g.l. over LSC, absent in the observations. In the CBL between the LE and LSC fronts, CO mixing ratios of 160-170 ppbv are lower than the 10 observations (180-340 ppbv). At the LSC front the model simulates a sharp decrease in all three species on the lake side of the LSC front, whereas the observations show higher values extending closer to the shore. In the circulation behind the LSC front, the model shows the downward transport of pollutants from aloft over the south end of LSC and the LSC inflow appears to move the pollutants along the south shore of LSC 15 consistent with the interpretation of the measurements.
To further investigate the LSC lake-breeze circulation, forward trajectories shown in Fig. 9 were computed starting at several locations along the LSC shoreline within the lake-breeze inflow at 12:00 LT and 225 m a.g.l. The trajectories, derived from the same GEM meteorological data used in the air quality simulations, show a circulation along 20 the LSC lake-breeze front that forms a tighter, nearly complete circuit at the easternmost shoreline departure point (Fig. 9a) with a recirculation time of 2 h (12:00-14:00 LT). Moving toward Detroit (Fig. 9b, c), the trajectories become more helical in nature. For example, the trajectory for the 12:00 LT departure point depicted in Fig. 9b starts at the surface, rapidly climbs to altitudes of 1000 m a.g.l. two hours later, then falls over 25 LSC and continues to Detroit close to the surface. Figure 9d shows a forward trajectory starting at 08:00 LT at 225 m a.g.l. over Windsor, with subsequent motion to the eastnortheast at altitudes of 1000 to 700 m a.g.l., subsidence over LSC (12:00 LT), then subsequent helical recirculation back to the west, along the south shore of the lake. Helical trajectories in lake-breeze circulations have been previously reported (Lyons and Olsson, 1973;Lyons and Cole, 1976;Lyons et al., 1995).

Conceptual models of the lake-breeze circulations
Conceptual models of the lake-breeze circulations on 25 June for Flights 4 and 5 are proposed in Fig. 10. The models are based on an integration of the aircraft, model 5 and trajectory data previously discussed. The aircraft flight path is illustrated as grey dashed arrows showing the altitude and direction of flight; the black arrows indicate the large scale movement and the grey arrows show the lake-breeze motions. In Fig. 10a (Flight 4), the CBL containing a mix of emissions from various sources across LE and Detroit/Windsor is bounded by the LE and LSC lake-breeze fronts over land. The 10 black, squiggly arrow in the CBL depicts the observed convection motion. Regional air containing SO 2 and SO 2− 4 from source areas southwest of LE are transported into the region with the LE lake-breeze front acting as a boundary between LE and CBL air. The CBL contains air composed of a mix of pollutants from the Detroit/Windsor urban area, as well as sources at the west end of LE. The aircraft data clearly shows upward 15 vertical motion at the LE front where air is advected and transported northwestward in the synoptic flow toward the Detroit/Windsor area. The dashed grey arrow (Fig. 10a) represents the region of reduced wind speeds associated with effects from the LE return flow as previously discussed. Model simulations indicate transport of pollutants from the CBL and Detroit/Windsor area to over LSC that are, at least partially, caught 20 in a downdraft over the south end of LSC. A downdraft was observed in the aircraft data behind the LSC lake-breeze front and both observations and model data show that pollutants were advected at low altitudes (<300 m a.g.l.) in the LSC inflow toward the south shore of LSC. Model simulations show that pollutants are lofted upwards and returned back over LSC, with some of the motion along the LSC convergence zone 25 toward Detroit/Windsor. In Fig. 10b (Flight 5) and model results indicate that pollutants in the CBL in Flight 4 were continually lofted at the LE and LSC fronts while the two fronts gradually merged together, transported northward in the synoptic flow, and brought downward through convective mixing north of the lake and observed during Flight 5. Figure 11 shows some of the main features of the Flight 4 lake-breeze circulations in 5 perspective view: the surface position of the lake-breeze fronts are shown as magenta lines, and the circulation flows are shown as arrows coloured according to altitude. Note that in order to demonstrate the lake-breeze flows, the vertical dimension of the arrows, relative to the horizontal dimension, has been greatly exaggerated -in reality, these circulations occur within a very shallow vertical layer of the troposphere. Synoptic 10 flow is from the southwest and below 800 m a.g.l., the lake-breeze circulation creates two regions of uplift at the LE and LSC lake-breeze fronts. There is a subsidence (downdraft) region over the southern and eastern end of LSC with onshore flow associated with the LSC lake-breeze front at <300 m a.g.l. Forward trajectories suggest a helical recirculation moving west along the LSC front (not illustrated).

Impacts of the lake breezes on air mass processing
In Fig. 12, the relationship between two ratios, oxygenated organic aerosol (OOA)/∆CO and SO 2− 4 /(SO 2 +SO 2− 4 ), is shown. Each point represents an average of an air mass type which are defined below. The ratios are indicators of the extent of air mass chemical processing. OOA is ratioed to ∆CO to account for mixing and dilution; ∆CO is 20 determined as (CO -CO background ), where CO background for each flight was taken from an average of CO above the boundary layer and/or during periods where the aircraft was minimally influenced from anthropogenic emissions. OOA was determined by deconvolving OA into two components, OOA and hydrocarbon-like organic aerosol (HOA), using organic mass fragments 44 (CO + 2 ) and 57 (C 4 H + 9 ) as tracers for OOA 25 and HOA respectively (Zhang et al., 2005). Increases in the OOA/∆CO ratio are considered to be indicative of photochemical processes forming secondary organic aerosol (SOA) (DeCarlo et al., 2008;de Gouw et al., 2005;Kleinman et al., 2009) are secondary aerosols resulting from concurrent chemical processing in an air parcel, and as such they are expected to be related in air masses on regional scales (i.e., at locations at least a certain distance downwind (∼>100 km) of the main locations of their precursor emissions). However, the relative magnitudes of OOA and SO 2− 4 formation depend on the mixture and concentrations 5 of SO 2 and precursor hydrocarbons. Figure 12 shows that the relationship between OOA/∆CO and SO 2− 4 /(SO 2 +SO 2− 4 ) for Flight 5 is confined to a range of higher OOA production (200 µg m −3 ppmv −1 % −1 , upper limit), whereas in Flight 4, a range extending to lower OOA production (80 µg m −3 ppmv −1 % −1 ), is observed, relative to SO 2− 4 production. The relationship between OOA/∆CO and SO 2− 4 /(SO 2 +SO 2− 4 ) provides a relatively convenient gauge of how much and how fast OOA is formed since SO 2 conversion in clear air is a better understood process and can be obtained through modelling. The rate of SO 2 conversion in clear air has been reported at 1-3% h −1 (Newman et al., 1981;Davis et al., 1979;Luria and Sievering, 1991) and can be further derived from the model. The modelled SO 2− 4 production rate at 12:00 LT for 285 m a.g.l. 15 for the study domain is shown in Fig. 13 and indicates that the regional background SO 2− 4 production rate (non-plume influenced) is about 1-2% h −1 , though much higher (>20% h −1 ) closer to sources and high concentration plumes. The OOA formation rate for regional background air is determined by multiplying the slope in Fig. 12 (Flight 4, m = 71.39 µg m −3 ppmv −1 · fraction) by the regional background SO 2− 4 formation rates 20 taken from Fig. 13 (1-2% h −1 ). Thus, the regional background OOA formation rate is 0.71-1.4 µg m −3 ppmv −1 h −1 . The regional background formation rates of OOA and SO 2− 4 are used as reference points in the following analysis for determining if the production rates could be enhanced in the lake-breeze recirculation.
Following on the analysis of air mass boundaries and movement in response to the 25 lake breezes (summarized in Fig. 10), time slices of aircraft data were selected that represent different air mass "types" (shown as shaded boxes in Figs. 3 and 4). The selection of air mass types is as follows: RegB represents regional air along flight tracks Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | above the boundary layer; LE represents air originating over LE and characterized by low pollutant concentrations with the exception of SO 2 and SO 2− 4 from power plants across LE; CBL is a section of the boundary layer closest to the LSC lake-breeze front and most influenced by lake-breeze circulations; AF represents air in the downdraft region of the LSC lake-breeze front; and LSC is air least affected by recent land-based 5 emissions and lake-breeze circulations. Changes in the chemical characteristics of OOA and SO 2− 4 in these air masses are expected to reflect the influence of the lakebreeze circulations.
In Fig. 14 ) ratio is more variable in Flight 4 compared to Flight 5. In addition, the OOA/∆CO ratio shows a general increase from Flight 4 to Flight 5 of 10.9-24.2 µg m −3 ppmv −1 . These ratios highlight the spatial heterogeneity in the earlier flight, largely induced by the separation of air mass types 15 due to the lake-breeze boundaries, and the subsequent transition to a comparatively homogeneous air mass during the later afternoon flight. The highest OOA/∆CO ratios in Flight 4 are in RegB, comparable to RegB values in Flight 5, suggesting similarities in the chemical composition of regional background air arriving in the study region throughout the day. In Flight 4 (Fig. 14a), the aircraft measurements in the CBL re-20 flect a mixture of pollutant sources from Detroit/Windsor and from the west end of LE, and as such the low OOA/∆CO value, 20.7 ± 6.0 µg m −3 ppmv −1 indicates limited photochemical processing. This is consistent with reported values for unprocessed urban air. DeCarlo et al. (2008) note that the lowest ratio for relatively "fresh" urban air over Mexico City and surrounding region was ∼35 µg s m −3 ppm −1 . For similarly fresh urban 25 air, Kleinman et al. (2008) showed a ratio of ∼18 µg m −3 ppmv −1 . Figure 14b shows In Flight 4, as most of these air masses are "unconnected" due to the lake-breeze induced boundaries, the variability in the extent of processing is not surprising. However, based on the interpretation of the LSC recirculation inferred in Flight 4 (Figs. 10a, 11), air masses in the CBL appear to be connected to that behind the LSC lake-breeze front (AF). Particles in the CBL are hypothesized to move aloft over the LSC front towards 5 the north into the zone labelled as AF and then to descend and possibly be recirculated through the onshore flow. Thus, the AF and CBL particles can be compared to determine if the lake-breeze recirculation results in changes in particle mass locally. This requires an estimate of the processing times associated with this circulation which can be derived from aircraft observations, model data, and through forward trajectory anal- in the inflow and return flows with a total distance travelled between 14.4-25.6 km (depending on how far over LSC the recirculation pathway might extend) and this distance 20 translates into a total circuit time of 1.6-2.8 h, consistent with that derived based on measurement data. Forward trajectories shown in Fig. 9 indicate that a recirculation time of 2-3 h is reasonable. With these three methods, the recirculation time at the LSC front is somewhere between 1.2-3.0 h. By comparison, using tetroons released into lake-breeze circulations, Lyons and Olsson (1973) showed complete lake-breeze 25 cycles ranging from 1.5-2.0 h. In Fig. 14a (Flight 4), the OOA/∆CO ratio in the AF air mass (behind the LSC lake-breeze front) (28. in the OOA/∆CO ratio is 6.2 µg m −3 ppmv −1 h −1 , but if the recirculation time is closer to 3.0 h, the increase corresponds to 2.5 µg m −3 ppmv −1 h −1 . Thus, the OOA formation rate, relative to excess CO, determined in the lake-breeze recirculation, 2.5-6.2 µg m −3 ppmv −1 h −1 , shows an enhancement over the regional background formation values inferred above (0.71-1.4 µg m −3 ppmv −1 h −1 ). From Fig. 14b, the SO 2 con-5 version rate can be similarly estimated. The average SO 2− 4 /(SO 2 + SO 2− 4 ) ratio in the AF air mass of 0.24 ± 0.03 and in the CBL of 0.24 ± 0.03 results in an estimated SO 2 conversion rate to range between 1.8% h −1 (3 h recirculation) and 4.6% h −1 (1.2 h recirculation). This suggests an enhancement for the shorter estimated recirculation times over the non-plume influenced regional background formation rate (1-2% h −1 ) 10 as predicted by the model. Enhancements in SO 2− 4 and OOA formation rates, relative to regional background formation rates, implies that lake-breeze circulations are an important dynamic in the formation of SO 2− 4 and SOA. One possible explanation for such enhancements is through cloud processing. Cumulus clouds were in fact observed all along the LSC front and in the CBL between 15 the LE and LSC fronts, especially between 12:00 and 14:00 LT. The presence of such clouds can significantly increase the rate of SO 2 oxidation (Luria and Sievering, 1991;Leaitch, 1996;Joos and Baltensperger, 1991) and SOA formation (Lim et al., 2005;Yu et al., 2005;Blando and Turpin, 2000). A second explanation is that the effective confinement of pollutants behind the LSC front may have prevented the dilution of pol-20 lutants from further mixing and led to elevated oxidant concentrations, and therefore enhanced oxidation of primary pollutants including SO 2 and organics. Both processes may be at work in the study region. Regardless of the processes, the lake-breeze front apparently led to locally enhanced SO 2 (and probably organic aerosol precursors) and increased production of SO 2− 4 and SOA. 25 11522 craft flights (Flights 4 and 5 on 25 June 2007) in southwestern Ontario where horizontal transects across the entire lake-breeze circulation at multiple altitudes were performed. Air mass boundaries due to lake-breeze fronts were identified in the aircraft meteorological (dewpoint temperature and vertical gust velocity) and chemical data, which were consistent with the frontal locations determined from surface observations and satel-10 lite and radar images (Sills et al., 2011). The meteorological model was also found to simulate the conditions during the first flight reasonably well.
Cross-sectional plots created from the aircraft horizontal transects indicated that in the early afternoon flight (Flight 4), elevated mixing ratios of CO and SO 2 were wellmixed vertically within a convective boundary layer (CBL) and bounded by the Lake 15 Erie (LE) and Lake St. Clair (LSC) fronts. Back trajectories indicated that the origin of these pollutants was from the west end of LE and the Detroit/Windsor area. Elevated pollutant concentrations were also observed aloft on the lake side of the LSC front and also closer to the surface, which was assumed to be due to subsiding air or downdrafts both observed and modelled. This area of high concentration did not extend far away 20 from the front, thus indicating that the circulation pattern tended to confine pollutants in the lake-modified air relatively close to the front. By the late afternoon (Flight 5), the region was largely enveloped in lake air and pollutants observed in Flight 4 were displaced by the merging LE and LSC fronts. The hypothesized fate of these pollutants was transport aloft toward the north where they were convectively mixed to the surface 25 north of LSC. This is supported by observations and modelling during Flight 4 showing that pollutants in the CBL were lofted in the upward vertical motions at the LE and LSC lake-breeze fronts and transported northwestward in the synoptic flow. Observations and modelling also demonstrate that during Flight 4, a LSC lakebreeze recirculation lofted pollutants from the CBL at the front, carried them in the southwesterly synoptic flow, they were caught in the downdraft over the lake, and then confined by the LSC onshore flow along the south shore of LSC. Modelling suggests that air parcels caught in the LSC recirculation travelled a helical trajectory along the 5 LSC front toward Detroit/Windsor. Based upon measurements and model output, best estimates for recirculation times of the air parcels travelling in the LSC lake-breeze circulation range from 1.2 to 3.0 h. This detailed, integrated approach led to the development of conceptual models that summarizes the complex 3-D circulation patterns and its interaction with the synoptic 10 flow. An attempt was then made at determining local SO 2− 4 and OA production rates in the LSC lake-breeze circulation. Using the relationship between oxygenated organic aerosol (OOA)/∆CO and SO 2− 4 /(SO 2 + SO 2− 4 ) (Fig. 12) as a gauge of how much OOA is formed and a background formation rate for SO 2− 4 ranging between 1-2% h −1 taken from the model, the regional background formation rate for OOA was calculated to be 0.71-1.4 µg m −3 ppmv −1 h −1 . Based on the interpretation of the LSC lake-breeze circulation, air masses in the CBL and behind the LSC front (AF) (Figs. 10a, 11) were compared to determine formation rates of OOA and SO 2− 4 . The maximum and minimum values (3.0 and 1.2 h, respectively) of the best estimates for recirculation time (or processing time) provide upper and lower bounds on the formation rates. The OOA 20 formation rate, relative to excess CO, determined in the lake-breeze recirculation was found to be 2.5-6.2 µg m −3 ppmv −1 h −1 which shows an enhancement over the regional background formation values inferred above (0.71-1.4 µg m −3 ppmv −1 h −1 ). The SO 2− 4 formation rate is estimated to range between 1.8-4.6% h −1 , also enhanced over the regional background formation rate. The enhanced formation rates relative to regional 25 background rates implies that lake-breeze circulations are an important dynamic in the local, near-source formation of SO 2− 4 and SOA. The presence of cumulus clouds associated with the lake-breeze fronts suggests that these enhancements could be due to cloud processes. Additionally, the effective confinement of pollutants along the Introduction  , 10, 19939-19980, doi:10.5194/acpd-10-19939-2010, 10, 19939-19980, doi:10.5194/acpd-10-19939- , 2010 Ozone distribution in coastal central Taiwan under sea-breeze conditions, Atmos. 10 Environ., 36 (21), 3445-3459, 2002. Cho, S., Makar, P. A., Lee, W. S., Herage, T., Liggio, J., Li, S. M., Wiens, B., and Graham, L.: Evaluation of a unified regional air quality modeling system (AURAMS) using PrAIRie2005 field study data: The effects of emissions data accuracy on particle sulphate predictions, Atmos. Environ., 43, 1864-1877.