Observations of Lake-Breeze Events During the Toronto 2015 Pan-American Games

Transcription

1 Boundary-Layer Meteorol DOI /s RESEARCH ARTICLE Observations of Lake-Breeze Events During the Toronto 2015 Pan-American Games Zen Mariani 1 Armin Dehghan 1 Paul Joe 1 David Sills 1 Received: 20 March 2017 / Accepted: 1 August 2017 Springer Science+Business Media B.V Abstract Enhanced meteorological observations were made during the 2015 Pan and Parapan American Games in Toronto in order to measure the vertical and horizontal structure of lake-breeze events. Two scanning Doppler lidars (one fixed and one mobile), a C-band radar, and a network including 53 surface meteorological stations (mesonet) provided pressure, temperature, humidity, and wind speed and direction measurements over Lake Ontario and urban areas. These observations captured the full evolution (prior, during, and after) of 27 lake-breeze events (73% of observation days) in order to characterize the convective and dynamic processes driving lake breezes at the local scale and mesoscale. The dominant signal of a passing lake-breeze front (LBF) was an increase in dew-point temperature of 2.3 ± 0.3 C, coinciding with a 180 shift in wind direction and a decrease in air temperature of 2.1 ± 0.2 C. Doppler lidar observations over the lake detected lake breezes 1 hour (on average) before detection by radar and mesonet. On days with the synoptic flow in the offshore direction, the lidars observed wedge-shaped LBFs with shallow depths, which inhibited the radar s ability to detect the lake breeze. The LBF s ground speed and inland penetration distance were found to be well-correlated (r = 0.78), with larger inland penetration distances occurring on days with non-opposing (non-offshore) synoptic flow. The observed enhanced vertical motion (>1 ms 1 ) at the LBF, observed by the lidar on 54% of lake-breeze days, was greater (at times >2.5 ms 1 ) than that observed in previous studies and longer-lasting over the lake than over land. The weaker and less pronounced lake-breeze structure over land B Zen Mariani zen.mariani@canada.ca Armin Dehghan armin.dehghan@canada.ca Paul Joe paul.joe@canada.ca David Sills david.sills@canada.ca 1 Cloud Physics and Severe Weather Research Section, Environment and Climate Change Canada, Toronto, Canada

2 Z. Mariani et al. is illustrated in two case studies highlighting the lifetime of the lake-breeze circulation and the impact of propagation distance on lake-breeze intensity. Keywords Convection Doppler lidar Doppler radar Lake breeze Mesonet PanAm games 1 Introduction Environment and Climate Change Canada (ECCC) conducted enhanced weather monitoring in the Toronto urban area in the summer of 2015 that coincided with the 2015 Pan and Parapan American Games. This included the deployment of new meteorological instruments for nowcasting the weather and enhanced air quality monitoring and forecasting throughout southern Ontario (Joe et al. 2017). During the Games (10 July to 15 August 2015), 53 surface meteorological weather stations (mesonet), a 14-station three-dimensional lightning mapping array, four new air-quality stations, a mobile air-quality laboratory, two wavespectra buoys, two meteorological supersites, and four mobile weather stations were deployed throughout the Toronto area. In addition, two scanning lidars and an existing C-band radar provided Doppler observations in near-real time. These observations were used to identify and characterize the vertical and horizontal wind structure of lake-breeze events over Lake Ontario and the urban environment. The dynamics of lake breezes, which are the lake counterpart of sea breezes, have been studied previously using theoretical, empirical, and modelling methods (e.g., Lyons 1972; Atkinson 1981; Pielke 1974, 1984; Simpson 1994; Sills et al. 2011). The lake breeze is a thermally-direct circulation that arises from the differential heating between the lake and surrounding land, producing a mesoscale horizontal pressure gradient in the lower troposphere. Figure 2 ofsills et al. (2011) provides a detailed illustration of a typical lake breeze. A prominent diurnal cycle is characteristic of the lake breeze, where the surface onshore flow moves inland during the daytime until the land breeze forms at night. The lake-breeze front (LBF) is the leading edge of the onshore flow and can have an impact on local severe weather (e.g., thunderstorms) and air quality. The narrow region of convergence and lift at the LBF can initiate thunderstorms as well as influence air-pollution transport, sometimes rapidly reducing air quality along the LBF, resulting in enhanced smog formation (Lyons 1972; Sills 1998; Hastie et al. 1999; Clappier et al. 2000; King et al. 2003; Brook et al. 2013; Wentworth et al. 2015). A passing LBF is typically associated with a decrease in air temperature, increase in dew-point temperature, and onshore flow near the surface. The changes to air temperature and dew-point temperature often become more subtle as the LBF progresses inland due to the modification of the lakebreeze air mass over land (Lyons 1972). Depending on its strength, the lower tropospheric synoptic-scale flow can alter the lake breeze s location, shape, vertical structure, and/or modify the inland penetration distance. For instance, a light onshore synoptic-scale flow increases inland penetration, while a stronger offshore synoptic-scale flow prevents a LBF from reaching the shore at all (Sills et al. 2011). Several studies have analyzed lake-breeze events in the Great Lakes region using radar and mesonet observations. For instance, during the Effects of Lake-breezes on Weather (ELBOW) 1997 (King etal. 1998) and ELBOW 2001 (Sills et al. 2002) campaigns, several case studies of the LBF affecting thunderstorm development in southern Ontario were analyzed (Sills 1998). During the 2007 Border Air Quality and Meteorological Study (BAQS-Met) cam-

3 Observations of Lake-Breeze Events paign, Great Lakes lake-breeze events in south-western Ontario were found to occur more frequently than previously reported, penetrating up to 200 km inland (Sills et al. 2011). The same observational techniques have been used to study lake breezes in other regions, such as lake breezes in Manitoba (e.g., Curry et al. 2015; Kehler et al. 2016). Previous studies of the Lake Ontario lake breezes in the Greater Toronto Area (GTA), however, are few (e.g., Comer and McKendry 1993; Wentworth et al. 2015). Characteristics of the LBF, such as its horizontal and vertical wind structure over the water, are highly variable and difficult to measure accurately (Darby et al. 2002). Here, we utilize Doppler lidar (hereafter referred to as lidar) technology in conjunction with radar and mesonet observations to characterize the LBF in an urban environment. Lidars have been used to study the characteristics of lake/sea breezes elsewhere. The life cycle of the land-breeze and sea-breeze systems at Monterey Bay, California, was analyzed using lidar measurements and Regional Atmospheric Modelling System simulations (Darby et al. 2002). The role of sea-breeze dynamics on pollution transport in the Marseille area, France, was investigated using meteorological surface stations and lidars (Bastin et al. 2005). A mobile lidar was applied to the measurement of sea breezes during the Qingdao International Regatta, China, in 2006, to map and characterize winds during the sailboat racing events (Sheng et al. 2007). The multilayered structure of larger, regional-scale sea breezes covering the Tokyo metropolitan area, Japan, were analyzed using a ground-based coherent lidar (Tsunematsu et al. 2009). Most recently, a lidar was used to study LBF shape and vertical velocities in southern Manitoba (Curry et al. 2017). This is the first study to utilize observations from a mesonet, satellite, weather radar, and two lidars to identify, track, and characterize the full evolution of the lake breeze (prior, during, and after) over Lake Ontario and over land. Section 2 describes the instrumentation used to measure the atmospheric state and wind field during lake-breeze events, while Sect. 3 describes the sampling strategy for observing the LBF and subsequent analysis using the integrated observations. Section 4 presents a summary of all lake-breeze observations during the observational period including two case studies of lake-breeze observations over Lake Ontario and over land. The impact of lake-breeze propagation on lake-breeze intensity is also discussed. A discussion of results and conclusions are provided in Sect Instrumentation 2.1 Doppler Lidars Scanning Doppler lidar is a relatively new technology that has only recently been made more affordable. The instrument emits a pulsed laser and measures the radial velocity component of the wind towards/away from the lidar (assuming the target aerosols are following the wind) using the Doppler shift of the backscatter. Real-time observations during clear, cloudy, and light precipitation conditions at any elevation or azimuth are possible. The lidar measurement technique is similar to radar, except with the added benefit that measurements can be taken at low elevation angles along the surface since the lidar s narrow beam removes the issue of ground clutter. This makes lidar measurements useful in complex terrain (e.g., Banta et al. 1997, 1999; Darby et al. 1999; Fast and Darby 2003). Notable disadvantages compared to weather radars include the lidar s lower scan speed (time required to perform a horizontal/vertical beam sweep) and decreased measurement range by a factor of 5 60, depending on the system s power.

4 Z. Mariani et al. Table 1 Technical specifications for the HAN and MOB lidars Technical specification HAN lidar MOB lidar Manufactured date Deployment site Hanlan s Point, Toronto Mobile truck at several sites Centre Island Laser pulse energy 80 µj 15µJ Maximum range 10 km a 7.5 km a Minimum measureable signal 30.5 db 26 db Scan cycle 11 min 14 min Laser wavelength 1.5 µm Laser pulse rate 10 khz Beamwidth 56 mm at aperture Beam divergence 50 µrad Detector type PIN diode Nyquist velocity ±19.4 ms 1 Doppler velocity resolution Range resolution ± 0.04 m s 1 Customizable; operated with 3 m overlapping range gates a Note the typical maximum range strongly depends on atmospheric conditions Two scanning lidars, designated HAN (for Hanlan s Point) and MOB (for mobile), were deployed during the observational period and conducted continuous, automated measurements. The lidar technical specifications are listed in Table 1; both lidars are HALO-Photonics Stream Line models. The HAN lidar had 65 µj additional pulse energy, improved thermal stability, and was smaller and lighter compared to the MOB lidar. Both lidars use fibre optics to prevent alignment issues with temperature fluctuations and their measurements are range-corrected and atmospheric-absorption corrected. The lidars were subjected to quality control procedures based on the signal-to-noise ratio (SNR) within each range gate. While there are a number of quality-control studies (e.g., Frehlich and Yadlowsky 1994; Dabas 1999), Päschke et al. (2015) demonstrated that the manufacturer s threshold is too conservative, limiting data availability, using the same HALO lidar model. Their analysis determined an SNR + 1 threshold value provides almost entirely good velocity measurements and this was the SNR + 1 threshold value used for both lidars. All lidar scans were conducted using overlapping 60-m range gates, providing a range resolution of 3 m. It should be noted that since aerosols are the primary scattering targets for lidars, the backscatter signal strength (and thus lidar maximum range) primarily depends on the aerosol concentration and relative humidity, along with additional aerosol characteristics (Fast and Darby 2003). The typical lidar maximum range varied from 2 to 5 km during the observational period. The HAN lidar was deployed at Hanlan s Point, which is located on Toronto Centre Island (43.6 N, 79.4 W), and was installed on an elevated platform within an Ontario Ministry of the Environment and Climate Change air-quality monitoring station and remained at this location throughout the observational period, as shown in Fig. 1. The MOB lidar was installed in the back of a pickup truck enabling mobility (Fig. 1), with the truck equipped with a generator, cell phone modem, and a Vaisala WXT520 instrument (with temperature, T, pressure, P, and relative humidity, RH, measurements). No significant difference in vertical velocity was

5 Observations of Lake-Breeze Events Fig. 1 HAN lidar (upper-left background) and MOB lidar (centre-right foreground) during installation and side-by-side comparison testing on June , at Hanlan s Point observed between the two lidars during side-by-side vertical stare comparisons (Fig. 1); the two lidar vertical velocities agreed to within <±0.5 ms 1 up to 1.8 km above ground level (a.g.l.) with root-mean-square velocity errors <0.3 ms 1. This agreement is similar to previous studies comparing in-situ wind observations and a lidar (Durran et al. 2002) and radiosonde/wind profiler radar/lidar measurements (Päschke et al. 2015). 2.2 Mesonet Data The mesonet consisted of 53 new ground-based compact meteorological stations that complemented existing weather stations in the region. Each station was equipped to provide fully automated continuous monitoring of surface temperature, pressure, relative humidity, dew-point temperature, and wind speed and direction at a temporal resolution of 1 min. The majority of these stations comprised compact stations and some were installed on rooftops of existing buildings (Joe et al. 2017). Mesonet observations were quality controlled in real time using standardized quality-control procedures to filter outliers based on a 95% confidence interval. Nine mesonet stations made measurements at standard heights of 1.5 m (T, P, RH) and 10 m (wind speed and direction). Emphasis was placed on observations from six of these nine stations, numbered 1 6 (south to north) in Fig. 2, which were located along an analysis transect defined for this study (black line in Fig. 2) to obtain precise meteorological data as the LBF moved inland. This analysis transect was selected over alternatives since it passed through the largest number of mesonet stations while being perpendicular to the shoreline. 2.3 King City Radar The King City weather radar station is located just north of the city of Toronto in King City, Ontario, as indicated in Fig. 2. The dual-polarization Doppler radar is an operational radar that is also used for radar research purposes; it has a frequency of 5625 MHz, wavelength of 53.3 mm, antenna diameter of 6.1 m, and beamwidth of Volume scan data (Doppler

6 Z. Mariani et al. Fig. 2 Satellite image of southern Ontario (left), including the location of the King City C-band radar. The dashed box outlines the zoomed-in area of the GTA (right). The magenta circle indicates the HAN lidar observation site on Toronto Centre Island, yellow circles indicate MOB lidar observation sites (red outlines indicate primary observation sites), numbered green markers indicate the six primary mesonet surface stations used in this study, and the black line indicates the LBF analysis transect Terrametrics, 2016 Google and dual-polarization) are collected to a range of 250 km (Doppler range 113 km) in 10- min intervals (Hudak et al. 2006; Boodoo et al. 2010). Radar fine lines caused by insects along narrow regions of lift were used together with GOES-13 visible satellite imagery and mesonet data to track the position of LBFs (Wilson et al. 1994; Russell and Wilson 1997; Sills et al. 2011). 3 Data Analysis 3.1 Lake-Breeze Identification A list of LBF criteria for satellite, radar, and surface station observations are provided in Sills et al. (2011) and were used in this study to identify the presence of the lake breeze. LBF analyses were undertaken following the approach described in Sills et al. (2011); as shown in Wentworth et al. (2015), this method is highly reliable and made easier via the network of observations installed during the Games (Joe et al. 2017). These datasets were combined and visually inspected at hourly and 5-min intervals using the Aurora workstation (Greaves et al. 2001). The positions of LBFs and other low-level boundaries such as synoptic fronts and thunderstorm outflow boundaries were manually tracked within <±5 km(or <±1 km if a radar fine line existed) throughout each lake-breeze day to provide a detailed mapping of their evolution. Uncertainties are due to cloud-line displacement relative to the front and parallax due to the satellite viewing angle. Note that while these analyses were applied over the entire southern Ontario region and included lake-breeze detections from various lakes, only days with a LBF passing through the centre of the GTA (approximately outlined in Fig. 2, right) were analyzed. Lidar-based lake-breeze identification criteria were developed using the approach developed by Sills et al. (2011),asshowninTable2. This enabled lidar observations to be used to precisely identify and time the location of the LBF. The lidar can detect the LBF signal

7 Observations of Lake-Breeze Events Table 2 Lake-breeze identification criteria for the different types of lidar wind observations Lidar observation Positive factors Negative factors Ambiguous Horizontal winds (near-surface PPI scan) Vertical winds (RHI and/or stare scans) Roughly 180 rotation in wind direction from offshore to onshore flow Radial velocities decrease to <4 ms 1 during LBF passage Radial velocities increase after LBF passage, typically doubling Enhancement of vertical velocity (>1ms 1 ) near the LBF Onshore flow depth <900 m a.g.l. Offshore (return) flow above onshore flow Radial velocities increase after LBF passage, typically doubling Offshore wind direction Offshore wind direction near surface Downdraft near the LBF Rotation of wind direction <100 Synoptic-scale flow causing the onshore flow Variability in onshore flow direction Low radial velocities <2 ms 1 Vertical velocity >1 ms 1 not near the LBF No clear separation between onshore/offshore flow and measure distinct features of the LBF s vertical and horizontal structure at high resolution (e.g., Banta et al. 1993; Darby et al. 2002; Bastin et al. 2005; Tsunematsu et al. 2009). Note that the lidar lake-breeze identifications were independent of the mesonet lake-breeze identifications. The MOB truck s observations of the surface temperature, pressure, and relative humidity could also independently verify the presence and passage of the LBF to the MOB lidar operators in real-time. 3.2 Lidar Observations of the Lake Breeze Plan position indicator (PPI) scans of constant elevation (5 elevation, 3 azimuth) were performed for a full 360 horizontal sweep with the MOB lidar. The HAN lidar conducted scans (1 elevation, 1 azimuth) only in the southern sector (over Lake Ontario) to avoid blockage by nearby trees. Range-height indicator (RHI) scans of constant azimuth were performed by both lidars in the north-south direction at every 2 in elevation, with 3-beam Doppler beam swinging (DBS) and 8-beam velocity azimuth display (VAD) scans performed to collect wind profiles above the lidar. The scan sequence repeated every 10 and 14 min, respectively, for the HAN and MOB lidars. When the MOB lidar s truck parked at a pre-determined observation site, it conducted vertical, PPI, RHI, DBS, and VAD scans. Several observation sites were pre-selected prior to the Games based on their geographic location and sightlines, as shown in Fig. 2. The MOB lidar was parked at one of the three southern sites each morning to measure wind speed and direction over the lake prior to the LBF s formation. If no LBF was detected, or if the LBF remained stationary south of the shoreline, the MOB lidar remained stationary. If the LBF moved inland, the MOB lidar relocated (while continuing to conduct vertical stare measurements in-transit) to a northern site such as Downsview (Fig. 2, right) where it

8 Z. Mariani et al. repeated its scan operations. This ensured that the MOB lidar sampled the evolution of the lake breeze at both the shoreline as well as inland throughout the day. 3.3 Measuring Lake-Breeze Characteristics Once the presence of a LBF was identified by the lidar using the criteria in Table 2, seven main characteristics of the lake breeze s horizontal and vertical structure were measured to quantify the lake-breeze structure. These were: 1. Change in surface temperature: T max T min temperature when the LBF passed overhead (within a 40-min interval centred on the time of LBF passage) as measured by mesonet surface station 1 close to the shoreline, 2. Change in dew-point temperature (using the same methodology as for surface temperature), 3. Enhanced vertical velocity near the LBF: a yes/no criterion was assigned based on whether enhanced vertical velocity near the LBF existed (>1 ms 1 during the time the LBF passes the lidar) as measured by either lidar s vertical scan, 4. Lake-breeze flow depth: the onshore flow height as measured by the HAN lidar RHI scan 1-h after the LBF passed overhead, 5. Lake-breeze radial velocity: the average radial velocity inside the lake breeze as measured by the HAN lidar RHI or PPI scan near the surface 1 h after the LBF passed overhead, 6. Lake-breeze ground speed: the average velocity at which the LBF progressed inland as measured by selecting the initial and final LBF positions, D, over a time period, t,when the LBF was moving continuously inland (typically over several hours) and computing its velocity, v(v= D final D initial /t), and, 7. LBF inland penetration distance: the maximum distance the LBF travelled inland (from the shoreline). The LBF ground speed and lake-breeze inland penetration distance were calculated along the LBF analyses transect (black line in Fig. 2) using the position and time of the LBF as measured by the satellite, radar, and mesonet analysis. 4 Results 4.1 Summary of Lake-Breeze Observations A complete summary of all lake-breeze days and lake-breeze characteristics is provided in Table 3. This analysis of lake-breeze characteristics includes observations from the lidar data and the mesonet analysis. Averages and standard errors (σ / N) of all lake-breeze characteristics are also provided. The Lake Ontario lake breeze was observed for 27/37 days (73%) within the study domain (central Greater Toronto Area) during the observational period. The lidars were unable to identify the presence of three lake-breeze events (22 July, 1 and 11 August) that occurred outside of the lidar range; only the mesonet analysis data were used for these days. Note that there were five Lake Ontario lake breezes that did not travel through the GTA during the observational period; these were not included in this analysis. Thus the lake-breeze occurrence rate for the entire Lake Ontario region (not just for LBFs passing through Toronto) during the observational period was higher: 87% as measured by the mesonet analysis.

9 Observations of Lake-Breeze Events Table 3 Lake-breeze characteristics for each Lake Ontario lidar-detected lake-breeze day between 10 July and 15 August 2015 Date (2015) Temperature change ( C) Dew-point temp. change ( C) Enhanced (>1ms 1 ) vertical velocity at LBF Lake-breeze flow depth (m) Lake-breeze radial velocity (averaged) [m s 1 ] LBF ground speed [m s 1 ] LBF inland penetration distance (km) July N July N July N July Y July Y July Y July Y July N/A N/A N/A July Y July Y July Y July Y July Y July N July N July Y

10 Z. Mariani et al. Table 3 continued Date (2015) Temperature change ( C) Dew-point temp. change ( C) Enhanced (>1ms 1 ) vertical velocity at LBF Lake-breeze flow depth (m) Lake-breeze radial velocity (averaged) [m s 1 ] LBF ground speed [m s 1 ] LBF inland penetration distance (km) Aug N/A N/A N/A Aug N Aug Y Aug N Aug N Aug N Aug N Aug N/A N/A N/A Aug Y Aug Y Aug N AVG 2.1 ± 0.2 C +2.3 ± 0.3 C 54% Y 560 ± 50 m 5.6 ± 0.4 ms ± 0.2 ms 1 22 ± 4km Descriptions of each lake-breeze characteristic are provided in Sect Averages and standard error (σ/ N) are included at the bottom. Case study days are bolded

11 Observations of Lake-Breeze Events The dominant meteorological signal of the lake breeze was a sharp increase in dew-point temperature (average increase of 2.3 ± 0.4 C), coinciding with a decrease in temperature (average decrease of 2.1 ± 0.2 C) and offshore wind direction. The average change in dewpoint temperature in a recent lake-breeze study of Curry et al. (2017)was2.5 C, which is in good agreement with our results, but the average change in surface temperature was 0.5 C, which is significantly smaller than the present in study. This difference could be due to the shallower LBF and higher Lake Winnipeg temperatures that cause smaller lake land thermal contrasts when compared to Lake Ontario. Overall the results in Table 3, particularly the LBF inland penetration distance and ground speed, agree well with long-term observations of lake breezes from several different lakes in Manitoba (Curry et al. 2015; Kehler et al. 2016). The changes in temperature, dew-point temperature, and relative humidity typically took 1 5 min to occur, providing a clear indication that the LBF passed the mesonet station. On most days with offshore synoptic flow, the LBF would sit along the shoreline above mesonet station 1, producing oscillations and more gradual changes that might last for up to 40 min until the LBF began to move inland. Precise estimates of the width of the LBF are required in order to interpret these oscillations, which is the subject of future work. Enhanced vertical motion near the LBF was observed for 54% of the lake-breeze days; this enhancement was particularly strong when measured along the shoreline (periods of >2.5 ms 1 ). The magnitudes of the vertical velocities at the LBF are in good agreement with recent observations by Curry et al. (2017) of2 3ms 1. The observed range in lakebreeze flow depths of m with mean lake-breeze flow depth of 560 ± 50 m is in good agreement with Lyons (1972), who reported a range of m, and mean of 500 m, in the Lake Michigan area. After the lake breeze moved inland and matured in the afternoon, the lake-breeze radial velocity measured over the lake typically doubled in magnitude over the course of 2 3 h. The largest variability in lake-breeze characteristics relates to the penetration distance. Observed lake-breeze penetration distances from the Great Lakes and several lakes in Manitoba have ranged from near zero to >200 km (Lyons 1972; Comer and McKendry 1993; Sills 1998; King et al. 2003; Sills et al. 2011; Curry et al. 2015). For Lake Ontario, the lakebreeze inland penetration distances ranged from 6.5 to 63 km. The ability to observe lake breezes with minimal penetration distances (<15 km) arises from lidar observations of the LBF close to the shoreline and the increased density of mesonet stations near the shoreline. The observed LBF ground speeds are consistent with previous numerical and observational studies of LBF propagation (Bechtold et al. 1991; Simpson 1994; Bastin et al. 2005; Curry et al. 2015). Figure 3a illustrates the positive correlation between the LBF ground speed and inland penetration distance (r = 0.78, where r is the Pearson product-moment correlation coefficient) found using the mesonet analysis data. A negative correlation was found between the synoptic wind speed, obtained at 10 m a.g.l. by nine WMO-standardized mesonet stations (stations inside the lake breeze were filtered out), and inland penetration distance (r = 0.63) as shown in Fig. 3b. Wind direction is shown as degrees from north to illustrate the impact of non-northerly flow. The presence of even a weak (1 3 m s 1 ) northerly synoptic flow was found to inhibit lake-breeze inland penetration, with stronger (>3 ms 1 ) northerly synoptic flow limiting lake-breeze penetration to 10km inland. The synoptic flowdirection explains outliers in Fig. 3; on the two days with large lake-breeze penetration distances (top-middle of Fig. 3a), a moderate (around 2.5 ms 1 ) synoptic-scale flow was from the south-west (12 July) and south (29 July). The southerly component of the flow on these days aided in transporting the lake-breeze air further inland despite the lower LBF ground speed. No correlation was found between the lake-breeze flow depth, synoptic-scale wind speed, radial

12 Z. Mariani et al. Fig. 3 Correlation between the lake-breeze inland penetration distance and, a LBF groundspeed,b synopticscale wind speed, for 27 lake-breeze days observed during the Games. Data points are coloured based on the synoptic-scale wind direction at the time of lake-breeze formation. The black line indicates the linear fit and r is the Pearson product-moment correlation coefficient. The points circled in red indicate the two case study lake-breeze days velocity, and LBF ground speed. In the next two sub-sections, two case studies are presented to illustrate the extreme differences in observed lake-breeze characteristics: a slow-moving and shallow-penetrating LBF measured along the shoreline on 15 July (red circle, bottom-left of Fig. 3a), and a fast-moving and deeply-penetrating LBF measured well inland on August 9 (red circle, top-right of Fig. 3a) July 2015: Lake-Breeze Observations Over the Lake Measurements of the vertical and horizontal wind structure of a LBF during opposing (offshore) synoptic flow over Lake Ontario were made on 15 July On this day, overnight cumulus clouds gave way to clear skies at sunrise that lasted for the entire day, with the exception of small cumulus clouds forming along the western edge of the LBF. A north/northnorth-east synoptic flow near the surface of around 6 m s 1 persisted throughout much of the southern Ontario region. Maximum inland air temperatures ranged between 13 and 21 C, while the average Lake Ontario temperature measured near Toronto Centre Island was 20 C. The Lake Ontario LBF was detected offshore at 1400 UTC by the HAN lidar and reached the shoreline by It slowly progressed inland, pausing frequently and reaching a maximum inland penetration distance of only 7.5 km by 2000 UTC. The Georgian Bay and Lake Simcoe LBFs were detected at 1900 having travelled to the south, aided by the northerly synoptic flow, but did not interact with the Lake Ontario lake breeze. At 0004 UTC+1, offshore surface-flow observations by the lidar signalled the end of the lake-breeze event Horizontal and Vertical Structure on 15 July The complete evolution of the lake-breeze horizontal structure as measured by the HAN and MOB lidars on 15 July is shown in Fig. 4. The MOB lidar was located at Leslie Spit on this day, slightly north and 3.7 km to the east of the HAN lidar. The LBF first formed offshore and was detected by the HAN lidar at 1400 UTC (Fig. 4b, red arrow), as offshore winds (red) met onshore winds (blue) 300 m south-east of the lidar. Note that the mesonet analysis did not detect the lake breeze for another hour due to the scanning angle of the King

13 Observations of Lake-Breeze Events Fig. 4 HAN (H) and MOB (M) lidar PPI scans during a passing LBF on 15 July 2015 at, a 1345, b 1400, c 1427, d 1537, e 1633, and f Negative (blue) velocities represent winds towards the lidar; positive (red) velocities represent winds away from the lidar. Black arrows indicate the deduced wind direction from the lidar PPI scan. The red arrow in b indicates the location of the approaching LBF. All times are UTC City Radar (as will be discussed). As the LBF continued moving onshore, it passed directly over the MOB lidar at 1537 UTC (Fig. 4d). At 1633 UTC (Fig. 4e), the radial velocities close to the shore increased from <4 to >6 m s 1 and a complete 180 reversal of the nearsurface flow at Leslie Spit was observed. Radial velocities continued to increase as the lake breeze matured until MOB lidar measurements ended at 2257 (Fig. 4f). The slight clockwise rotation of the lake-breeze wind direction from Fig. 4e, f can be attributed to bulging effects of the lake breeze due to small variations in the shoreline and synoptic-scale flow and, to a

14 Z. Mariani et al. Fig. 5 Mesonet analysis at 2300 UTC (same time as Fig. 4f) on 15 July Meteorological data from individual mesonet stations are displayed on the map (white) overlaid with radar and satellite observations. LBFs, such as the Lake Ontario LBF (along the shoreline), are indicated by purple triangles. Also visible are the Lake Simcoe and Georgian Bay LBFs (centre-left and top-right, respectively). Lidar locations are indicated by the yellow circles lesser extent, the Coriolis force. The LBF remained stationary above mesonet station 2 due to the northerly synoptic-scale flow inhibiting its progression further north as illustrated in the mesonet analysis in Fig. 5. RHI scans provided in Fig. 6 illustrate the evolution of the LBF vertical structure over Lake Ontario as measured by the HAN lidar. In Fig. 6a, the LBF is present during offshore synoptic flow; in Fig. 6b the LBF is detected 80 m offshore from Hanlan s Point with onshore and offshore flows converging; in Fig. 6c, d the LBF had passed inland and strengthened, with the maximum radial velocity occurring near the surface and increasing from <3 to >4 ms 1. The wedge-shaped LBF (approximated by the green outline in Fig. 6b) differs from the idealized plume LBF shape in Fig. 2 of Sills et al. (2011). Low-altitude onshore radial velocities were of similar magnitude to the higher-altitude offshore velocities, while the radial velocities between the two airflows were near zero over a relatively large span (60 m). Strong radar fine lines were observed for the Georgian Bay and Lake Simcoe LBFs while almost no radar fine line was observed for the Lake Ontario LBF (Fig. 5). The lack of a radar fine line along the shoreline is likely due to the King radar overshooting the LBF. At the shoreline, the radar 0.5 PPI scan is roughly 650 m a.g.l. (depending on atmospheric conditions); hence it can only detect LBFs with a lake-breeze flow depth close to this value along the shoreline. While the lake-breeze flow depth increased from 180 to 290 m in almost an hour as the lake breeze matured (Fig. 6c, d, white double arrow), it remained below the radar PPI scans (even below the hourly 0.3 PPI scan), failing to detect enhanced clear-air reflectivity. Note that for lake-breeze days with a lake-breeze flow depth >650 m, radar fine lines were observed when the LBF was near the shoreline.

15 Observations of Lake-Breeze Events Fig. 6 HAN lidar RHI scans southward (180 azimuth) over Lake Ontario during a passing LBF on 15 July 2015: a before the lake breeze at 1357, b first detection of the wedge-shaped (green outline) LBF at 1417, c after the LBF passed overhead at 1437, and d as the LBF matured at Negative (blue/green) velocities represent winds towards the lidar; positive (orange/red) velocities represent winds away from the lidar. Black arrows depict northerly flow and whitearrowsdepict lake-breeze onshore flow. Dotted arrows depict northerly (black) and lake-breeze return flow (white) along the LBF. The green dashed line and white double arrow in c and d depict the lake-breeze flow depth. All times are UTC Figure 7 provides snapshots of the wind profiles before and after the LBF passed the MOB lidar at Leslie Spit (similar results were obtained by the HAN lidar). A relatively stable wind profile with northerly (offshore) winds was observed at 1541 UTC (Fig. 7a), and in Fig. 7b, the wind direction changed near the surface at 1815 UTC, with southerly near-surface flow and northerly flow above the lake-breeze flow (>300 m a.g.l.). The onshore flow s radial velocity was limited to <4.5 ms 1 whereas the higher-altitude offshore winds ranged from 3 to 11 m s 1.FromFig.7a, b, the wind-speed maximum shifted from 300 to 700 m a.g.l., well above the lake-breeze flow. In addition, the offshore flow (from >300 m to approximately 1.1 km) occurred at almost three times the lake-breeze flow depth, which is a larger difference than predicted by Lyons (1972). The return flow from the lake-breeze circulation is likely embedded in the shallowest layer (300 m) of the offshore flow as depicted in Fig. 7b. Within the lake breeze, a large gradient in the wind direction below 300 m and a multi-layered structure of vertical velocity is evident (Fig. 7b), as previously reported (e.g., Darby et al. 2002; Tsunematsu et al. 2009). A strong enhancement in vertical velocity was measured by both lidars at the time the LBF passed overhead. Observations from the HAN lidar vertical stare scan are provided in Fig. 8 from 1000 to 2200 UTC (the LBF reached the HAN lidar at 1425). The lake-breeze flow depth (Fig. 8, black line) increased from 90 to 360 m, varied around 300 m a.g.l. for the majority of the day, then decreased (not shown) as the lake breeze ended. Two unique features persist for this case: (1) a significant decrease in vertical motion for the remainder of the day, particularly within the lake-breeze flow depth. Given that the circulation within the lake breeze is mostly horizontal and thermals are not generated over water, the muted vertical motion is expected; (2) pockets of increased vertical velocity are observed above

16 Z. Mariani et al. Fig. 7 MOB lidar wind speed (left panels, red) and wind direction (right panels, blue) profiles before and after a passing LBF on 15 July 2015 at, a 1541 and b All times are UTC (but not inside) the lake breeze, which are indicative of thermals that have been advected over top of the lake breeze. Such thermals form over land but, due to the lake breeze and offshore synoptic-scale wind, may have been shifted over the lake and to higher altitudes. This transport of thermals above the lake breeze is made easier by the wedge-shaped LBF August 2015: Lake-Breeze Observations Over Land Observations of the LBF over land were performed on several days during the observational period; 9 August 2015 will be used as a case study of a deeply penetrating lake breeze. The LBFs observed over land, hours after they moved onshore, had significantly different

17 Observations of Lake-Breeze Events Fig July 2015 HAN lidar vertical velocity measurements before, during (1425 green vertical bar) and after the LBF passed. White vertical gaps indicate periods where the lidar was conducting measurements in other scan modes. The lake-breeze flow depth is also shown (black line). Negative (blue) velocities represent flow towards the lidar (surface); positive (red) velocities represent flow away from the lidar (surface) characteristics than when they were newly formed over Lake Ontario. The presence of an onshore synoptic flow also resulted in several unique differences in LBF structure, as will be discussed below. The 9 August lake breeze occurred during east/east-north-east synoptic flow of around 3 m s 1, which is rare for the southern Ontario region. Clear skies persisted throughout most of this day, with the exception of deepening cumulus clouds along the Lake Ontario LBF in the afternoon as it moved inland. Maximum inland air temperatures ranged between 12 and 23 C, while the average Lake Ontario temperature measured near Toronto Centre Island was 19 C. The Lake Ontario LBF was detected offshore at 1405 UTC by the HAN lidar and reached the shoreline by 1500; at 2000 UTC, the Lake Ontario LBF intersected and continued through the Lake Simcoe LBF south of Lake Simcoe, but this intersection did not result in the development of thunderstorms. The Lake Ontario LBF progressed inland at a propogation speed of 3.2 ms 1, reaching its maximum inland penetration distance of 60.5 km by 2300 UTC. Soon after 2300 UTC, the land breeze was detected via offshore surface-flow observations by the HAN lidar (or, alternatively, the offshore synoptic flow was observed as the lake breeze detached from the lake), signalling the end of the lake-breeze event Horizontal and Vertical Structure on 9 August Lidar observations during the morning of the 9 August lake breeze over Lake Ontario were similar to that for the 15 July case (Fig. 4). As the LBF progressed inland, the MOB lidar relocated to the 400N site 31 km north of the shoreline and conducted PPI scans before, during, and after the LBF reached the site. Unlike the morning lake-side observations, the MOB lidar PPI scans at the 400N site are of a weaker, poorly-defined lake breeze. Surface radial velocities were less over land than over the lake (at the HAN site), and decreased only slightly (from 6 to 5 m s 1 ) as the LBF passed over the MOB lidar at 1824 UTC. The wind direction shifted <100 and no increase in surface radial velocity was observed following

18 Z. Mariani et al. Fig. 9 As in Fig. 5, except for before the LBF reached the 400N site at 1800 UTC on 9 August 2015 LBF passage. The onshore surface flow also had a poorly-defined horizontal velocity gradient compared to observations along the shoreline. The mesonet analysis at 1800 UTC is provided in Fig. 9 to illustrate the progression of the LBF far inland as it approached the 400N site (yellow circle, centre-left). The LBF at the 400N site had less pronounced vertical structure with no observed offshore flow (Fig. 10), unlike 15 July observations. The shallow circulation consists of negative radial velocities inside the lake breeze with a mixture of away/towards velocities above due to thermals being advected above the lake breeze. The lack of offshore flow is due to the synoptic flow in a non-offshore direction and the weaker lake-breeze circulation limiting the return flow. The LBF shape was a rounded wedge, somewhat resembling the idealized plume (roughly approximated by the green outline in Fig. 10) depicted in Sills et al. (2011). In agreement with simulated LBF events in Fig of Simpson (1994), the lake-breeze flow on 9 August was deepest behind the LBF and then tapered back. Only limited enhancement in vertical velocity (around 0.9 ms 1 ) was observed at 1825 UTC (not shown), approximately the time when the LBF passed over the lidar. 5 Discussion and Conclusions Two scanning lidars, a C-band radar, and a 55-station mesonet provided observations of the vertical and horizontal structure of lake breezes during the 2015 Pan and Parapan American Games in Toronto (10 July to 15 August 2015). New lake-breeze identification criteria were created for vertical and horizontal lidar observations and the first measurements of the LBF structure over Lake Ontario were conducted using the lidars. The lidars detected the lake breeze on average 1 h before the mesonet and radar, demonstrating early detection when

19 Observations of Lake-Breeze Events Fig. 10 As in Fig. 5, except MOB lidar RHI scan southward (180 azimuth) 31 km inland at the 400N site at 1806 UTC on 9 August The approaching plume-shaped LBF can be seen <60 m from the lidar (roughly approximated by the green curve). Black arrows depict northerly flow being uplifted and whitearrowsdepict lake-breeze onshore flow deployed near the shoreline. The Lake Ontario lake-breeze occurrence frequency is more than double the estimate of 32% provided in Comer and McKendry (1993), likely due to differences in study duration (the Comer and McKendry analysis included time periods outside of the summer months, which experience fewer lake-breeze events) and instrumentation. It is, however, similar to occurrence frequencies of 56 74% provided in Wentworth et al. (2015) using one summer of data from The continuous, autonomous, and maintenance-free operation of the lidars demonstrated their reliability and the applications for meteorological operations. There were several days where two or more lake breezes intersected (typically the Lake Ontario lake breeze moving north and the Georgian Bay lake breeze moving south); despite their intersection, no significant thunderstorm activity was observed, likely due to persistent low convective available potential energy or large convective inhibition environments. Note the observational period encountered significantly less total precipitation in the Greater Toronto Area (22 mm in July 2015), which is almost one-third the climatological normal. The dominant signal of a passing LBF was an average 2.3 ± 0.3 C increase in dewpoint temperature, coinciding with a 180 shift in wind direction and an average decrease in temperature of 2.1 ± 0.2 C. Enhanced (>1 ms 1 ) vertical motion along the LBF was observed by both lidars in 54% of lake-breeze cases. The lake-breeze flow had diminished vertical velocities, while enhanced vertical motion (thermals) was observed above the lakebreeze flow near the shoreline. For lake breezes measured near the shoreline, the largest vertical velocities occurred <600 m a.g.l. around the time of LBF passage and exceed the peak vertical velocity during LBF passage measured in the Monterey Bay area, likely due to the stable lower troposphere at the California coast (Banta et al. 1993). This altitude range of enhanced vertical velocities is below typical convective cloud bases due to the wedge-shaped LBF.

20 Z. Mariani et al. The LBF ground speed and synoptic-scale wind speed correlated well with the lake-breeze maximum inland penetration distance (r = 0.78 andr = 0.63, respectively), with maximum penetration distances occurring on days with non-opposing (non-offshore) synoptic flow. This relationship can be used to forecast LBF location and potential convergence in the afternoon (assuming constant synoptic flow) by measuring the LBF ground speed earlier in the day. While this analysis uses the average LBF ground speed to obtain this relationship, the LBF ground speed changes through the day in a predictable way (Simpson 1994) and can be accounted for when forecasting LBF location. This can have important implications for predicting thunderstorm development and inland pollution transport (Lyons and Olsson 1973; Clappier et al. 2000; Wentworth et al. 2015) and is the focus of future research. As seen in Fig. 4, the HAN lidar observed changes due to the lake breeze about 30 min before the MOB lidar at Leslie Spit. Given that the HAN lidar site is more exposed to Lake Ontario, located south of Leslie Spit, and the variability in the LBF angle of approach, this temporal offset was common throughout the observational period. As measured by both lidars, it typically took around 2 h for a full 180 shift in wind direction to occur. When measured at one of the inland sites, the time for a complete shift in wind direction was typically less (0.5 2 h) due to the LBF s slower progress over the lake compared to over land. Similarly, the pattern of diminishing surface radial velocities as the LBF passed over the lidar (Fig. 4c e), followed by increased surface velocities after the LBF passed (Fig. 4e, f), occurred near the shoreline but not when measured at the inland sites. Similar increases in radial velocities of 1 2 m s 1 within the leading edge of the LBF along the shoreline were observed in Curry et al. (2017). Several differences in lake-breeze characteristics were observed between 15 July and 9 August lake breezes, such as greater LBF ground speed (3.2 ms 1 on 9 August versus 0.5 ms 1 on 15 July) and lake-breeze inland penetration depth (60.5 km on 9 August versus 7.5 km on 15 July). The lake breeze on 15 July had surface radial velocities >6 ms 1, whereas the deeply-penetrating lake breeze on 9 August produced radial velocities <5 ms 1 over land. The greater radial velocities measured over Lake Ontario are likely due to less surface roughness and a shallower boundary-layer height compared to over land (Darby et al. 2002). Almost no radar fine line was observed on 9 August even as the LBF moved within range of the radar. Due to the lack of a northerly synoptic flow, the LBF ground speed was exceptional and convergence at the LBF was limited, producing negligible uplift, likely inhibiting the concentrations of insects to enhance the clear-air reflectivity along the LBF. King radar observations were also found to overshoot the LBF when the lake-breeze flow depth was <650 m a.g.l. near the shoreline, highlighting the significance of the synoptic flow on lake-breeze detection, structure, and characteristics. A wedge-shaped LBF was observed by both lidars when conducting measurements over Lake Ontario for all lake-breeze days with northerly synoptic flow; these are among the first observations of a wedge-shaped LBF using lidar (Curry et al. 2017). Generally speaking, the ability of the LBF to form a wedge shape is dependent on the presence of moderate to strong northerly synoptic flow to tilt the LBF backwards and/or a weak lake-breeze environment (less temperature contrast between the lake and land). Conversely, a light to moderate synoptic flow in a strongly forced lake-breeze environment (larger temperature contrast) produces a more plume-shaped LBF. The implication of a wedge-shaped LBF is the production of a tilted updraft that can have a diminished impact on the upward motion of air parcels to the level of free convection. This inhibits the formation of cumulus clouds along the LBF and limits the LBF s ability to aid in convection initiation. The wedge-shaped LBF also causes a shallower lake-breeze flow near the LBF with increasing lake-breeze flow depth in time, exacerbating the issue of King radar overshooting the LBF. This LBF structure is in agreement with lake-