along the western shore of Lake Michigan found that high concentrations of ozone observed at ground-level were associated with southeasterly or onshor

Transcription

1 Chapter 5 Lake Breeze and Oxidants Over South-Eastern Ontario Due to the thermal inertia of large bodies of water, lakes (or oceans) warm and cool considerably slower than the adjacent land surfaces On days with limited cloud cover, solar radiation warms land surfaces more rapidly than the lake surface and a temperature dierential between the land and lake surfaces develops The dierence in temperature between the land and the lake surfaces results in a dierence in the temperature of the atmosphere above, which results in a slight perturbation to the pressure eld The response of the wind eld to the temperature-induced perturbations in the pressure eld gives rise to a ow from the lake towards the land at the lowest levels of the atmosphere, and a corresponding return ow, from the land towards the lake, aloft At a certain distance inland, the onshore ow in the lowest levels of the atmosphere rises and return towards the lake as part of the oshore ow aloft, forming the lake breeze front (Simpson, 1977) Interest in the inuence of lake breeze circulations on air quality in the Great Lakes region has existed for more than 20 years Observations in the early 1970's 189

2 along the western shore of Lake Michigan found that high concentrations of ozone observed at ground-level were associated with southeasterly or onshore ow as a result of a lake breeze circulation (Lyons and Cole, ) Lake breeze or sea breeze circulations have also been found to have signicant eects on air quality along the shore of Lake Erie in southwestern Ontario (Mukammal et al, 1982), in Los Angeles, California (Lu and Turco, 1996), Athens, Greece (Lalas et al, 1983) and along the Mediterranean coast of Spain (Martin et al, 1991 Millan et al, 1996) The eects of lake and sea breeze circulations on air quality have beeninvestigated for a variety oflocations with diering topography and meteorology For example, a modelling study by Lu and Turco (1996) of the eects of a sea breeze circulation on air quality in Los Angeles has shown that the sea breeze can bring relatively unpolluted air inland The onshore ow of the sea breeze becomes progressively more polluted as the airmass moves over regions with large anthropogenic emissions of hydrocarbons and NO x, while vertical motions at the sea breeze front act to lift ozone and other pollutants and result in layers with higher concentrations of oxidants aloft These elevated layers of high concentrations of oxidants may bemixed down to the surface the following day For the case of the interaction of emissions from Toronto with a lake breeze over Lake Ontario, the most applicable study is that performed by Lyons and Cole (1976) They studied the eects of a lake breeze circulation over Lake Michigan on emissions from Chicago and surrounding regions Observations of the ground-level ozone concentration along the western shore of Lake Michigan showed that under conditions of onshore ow during a lake breeze, ozone concentrations were highest from 1 { 8 kilometres inland from the lakeshore Ozone concentrations were lower within one 190

3 kilometre of the lakeshore and also decreased at distances greater than 10 km inland from the lakeshore To explain the observations which showed the highest ozone concentrations occurring several kilometres inland from the lakeshore, Lyons and Cole suggested that the bulk of the ozone, while over the lake, resides above the conduction inversion As the airmass is advected onshore, locations near the shoreline are exposed to the less polluted air which was found within the lowest layers of the atmosphere over the lake As the airmass moves further inland, the thermal internal boundary layer (TIBL) rapidly deepens and the ozone aloft is eventually mixed down to the ground resulting in the higher ground-level concentrations observed just inland of the shoreline Still further inland, ground-level ozone concentrations decrease due to the continued deepening of the TIBL and loss processes which act on ozone such as dry deposition and titration by freshly emitted NO Lyons and Cole (1976) proposed that the generation of elevated layers of high ozone concentrations over the lake were the result of the advection of VOCs and NO x out over the lake by the gradient ow or a land-breeze circulation during the early morning The ozone precursors, released as the nocturnal inversion is breaking down but before the boundary layer has reached its full depth, will ow up and over the conduction inversion present over the lake, leaving the air within the lowest 100 { 150 m of the atmosphere less polluted With limited vertical mixing over the lake, the VOCs and NO x will remain trapped aloft and during the afternoon ozone will be photochemically produced Since the ozone is physically separated from the surface, dry deposition will not act to reduce ozone concentrations appreciably With light gradient winds, a lake breeze develops later in the day and the ozone which had 191

4 formed over the lake is advected onshore and mixed down to the surface as described above From an analysis of tetroon ights, Lyons and Cole suggested that recirculation of pollutants in the lake breeze may be a possibility Pollution advected inland by the onshore ow of the lake breeze could move inland to the lake breeze front, rise, and return towards the lake in the return circulation aloft Eventually, the pollutants could reenter the onshore ow of the lake breeze and return to land In this manner, pollutants could be advected along the western shore of Lake Michigan in a broad helical trajectory by a combination of the lake breeze circulation and the gradient wind A modelling study using the Regional Atmospheric Modeling System (RAMS) in conjunction with a Lagrangian particle dispersion model (Lyons et al, 1995) has reproduced the recirculation phenomenon described by Lyons and Cole (1976), though only approximately 30% of the particles were predicted to reenter the onshore ow and complete a full recirculation Sillman et al (1993) have shown that the suppression of vertical mixing and the extremely slow rate of dry deposition of ozone and NO x species over a large body of water is sucient to allow very high concentrations of ozone to be photochemically generated over Lake Michigan The model meteorology used by Sillman et al did not include a representation of the lake breeze, though advection of the Chicago urban plume over Lake Michigan, particularly early in the morning before turbulent mixing over land had distributed the precursors over a deeper layer, and the combination of slow rates of vertical mixing and dry deposition were sucient for the model to calculate high concentrations of ozone in a shallow layer over the lake Therefore it appears that a lake breeze circulation need not necessarily be present for high 192

5 concentrations of ozone to be generated over large bodies of water, though lake breeze circulations will be importantincontrolling the dispersion of pollutants over the lake The motivation for a modelling study of the lake breeze eects on oxidants over Lake Ontario was provided by observations made as part of the SONTOS study during the summers of 1992 and 1993 Figure 51 shows the concentration of ozone and NO y measured at the Hastings site on August 26, 1993 (D R Hastie, private communication) With the exception of a brief period at approximately 1830 GMT, the concentration of ozone was observed to slowly increase during the afternoon Between 2040 GMT and 2120 GMT (16:40 and 17:20 EDT) ozone concentrations increased from 46 ppb to 74 ppb and NO y concentrations increased from 22 ppb to 40 ppb Concurrent with the increase in concentration of ozone and NO y, the concentration of NO x,co,so 2,PAN and HCHO were also observed to increase Figure 51: The concentration of (a) ozone and (b) NO y as observed at the Hastings measurement site from 0600 GMT, August 26 to 0600 GMT, August 27,

6 Meteorological parameters measured at the Hastings site showed that a decrease in the dry bulb temperature of 25 C and an increase in the specic humidity by 20% was associated with the change in trace species concentrations noted above The change in meteorological parameters shows that the polluted airmass was cooler and more moist than the less polluted airmass it replaced, suggesting that the more polluted airmass had been modied through contact with a large body of water It was also possible to identify portions of the lake breeze front on visible satellite imagery for August 26, as it appeared to be marked by a line of broken cumulus clouds An analysis of the visibly satellite imagery shows that the lake breeze front passed over the Hastings measurement site between 2000 and 2200 GMT (Hastie et al, 1998) Eppley radiometer measurements showed a brief decrease in solar radiation as the chemical and meteorological measruments recorded the change in airmass, presumably as a result of clouds associated with the lake breeze front Taken together, the meteorological and chemical measurements made at the Hastings site, as well as the position and inland progress of the lake breezefront deduced from satellite imagery, suggest that a lake breeze circulation brought a more polluted airmass north from Lake Ontario to the Hastings measurement site Given that a westerly to southwesterly gradient ow existed across southern Ontario on August 26, it seems probable that the polluted airmass observed at Hastings originated from Toronto and surrounding urbanized regions around the western end of Lake Ontario It is hypothesized that, in a situation similar to that described by Lyons and Cole (1976) for the western shoreline of Lake Michigan, VOCs and NO x emitted from Toronto during the early morning are advected over Lake Ontario by the existing gradient ow, possibly aided by a land breeze circulation During the day, with little 194

7 vertical mixing and small rates of dry deposition, the photochemical production of ozone gives rise to high concentrations of ozone over the lake which are advected inland by the lake breeze circulation later in the afternoon Behaviour of chemical and meteorological variables similar to that seen on August 26 was observed on three additional days during the six week intensive measurement campaign in 1993 Given that Hastings is 40 km north of the Lake Ontario shoreline, it appears that lake breeze circulations can signicantly, and regularly, aect air quality along the northern shore of Lake Ontario for a considerable distance inland Further, since lake breeze circulations over the Great Lakes have been found to occur on 25 { 40% of days during the summer (Comer and McKendry, 1993 Chermack, 1986 Lyons et al, 1972), it seems likely that the inuence of lake breeze circulations on air quality for sites nearer the Lake Ontario shoreline is more frequent than that observed at Hastings The MC2-online model had been used to test the hypothesis that high concentrations of ozone are generated over Lake Ontario and advected inland by the onshore ow of a lake breeze The model has been run for two cases in 1993 for which a rapid increase in the concentration of ozone and related trace species were observed to occur at the Hastings site late in the afternoon The objectives of the modelling study are twofold: 1) to see whether the model qualitatively agrees with the hypothesized mechanism by which high concentrations of oxidants are produced over Lake Ontario and consequently advected northward by the lake breeze and 2) to assess to what degree the model quantitatively agrees with the observed meteorological and chemical elds 195

8 51 Method The two cases which have been selected for the modelling study are August 8 and 26, 1993 For both study cases an initial run of MC2 without chemistry was performed at a horizontal resolution of 42 km using CMC objective analysis to provide the meteorological boundary conditions Unfortunately, regional objective analyses in sigma coordinates were not available for these days and it was necessary to use a coarser resolution global analysis in pressure coordinates The global analysis has a horizontal resolution of 15 by 15 The 42 km resolution run was made on the standard 42 km resolution domain discussed in Chapter 2, and used nudging to limit the growth of errors in the synoptic scale features The run was three days and three hours long, begun at 0000 GMT, two days before the study day and ending at 0300 GMT on the day after the study day A 21 km horizontal resolution run of MC2 with chemistry was made using meteorological boundary conditions interpolated from the previously completed 42 km resolution run Initial conditions and time-varying boundary conditions for the advected chemical species were taken from the global CTM as described in Chapter 2 The 21 km resolution run began and ended at the same times as the 42 km resolution run The nal run of the MC2-online model was made at a horizontal resolution of 5292 km on an grid point domain centered over Lake Ontario The run was 27 hours long, beginning at 0000 GMT on the study day, and used initial and boundary conditions for the meteorological and chemical variables interpolated from the 21 km horizontal resolution run Meteorological and chemical boundary conditions 196

9 were updated every three hours, with linear interpolation used to derive boundary conditions at intermediate times 52 Case Study: August 8, 1993 The rst case chosen for study occurred on August 8, 1993 The concentrations of ozone and NO y as observed at the Hastings eld measurement site are given in Figure 52 The observations show that within 15 minutes, beginning at 2221 GMT (1821 EDT), the concentration of ozone increased from 47 to 59 ppb and the concentration of NO y increased from 26 to 43 ppb While the increase in ozone on August 8was rather modest, the time of day atwhich the increase occurred and the rapidity of the change were similar to other cases observed at Hastings Figure 52: The concentration of (a) ozone and (b) NO y as observed at the Hastings measurement site from 0600 GMT, August 8to0600GMT, August 9,

10 Figure 53 shows the synoptic conditions for 0000 GMT, August 8 and 0000 GMT, August 9, as taken from the CMC objective analysis A large area of high pressure covered southern Ontario and much of the northeastern United States on August 8 Winds across southern Ontario were light and from the west Winds recorded at synoptic observing stations in southern Ontario, though away from Lake Ontario, were generally less than 5 km hr ;1 and had variable direction during the afternoon Daytime high temperatures were between 23 and 25 C Figure 53: Sea-level pressure and 850 mb winds taken from the CMC objective analysis for 0000 GMT August 8, 1993 (panel a) and 0000 GMT August 9, 1993 (panel b) Figure 54 shows the model calculated wind eld and CO concentration over the 53 km resolution domain at 1100 GMT, or 0700 EDT The horizontal cross-section shows the CO concentration in the lowest model level, approximately 10 m thick, and the wind at 10 m above the surface The near-surface winds over the western end of Lake Ontario were generally from the north with speeds between 7 and 12 m s ;1 The 198

11 CO concentration eld, used as a long-lived tracer for anthropogenic emissions, shows that emissions from Toronto were being advected to the south across the western end of Lake Ontario The vertical cross-section of the CO concentrations shows that the plume of CO from Toronto was conned to within 200 m above the surface of the lake As shown by the 10 m winds in Figure 54, a weak cyclonic circulation was calculated by the model to be centred over the southeastern portion of Lake Ontario The cyclonic circulation is associated with a weak low pressure system that is, erroneously, calculated by the model to be present over eastern Lake Ontario at this time Observations show thataweak low pressure system was centered over Lake Michigan at 1200 GMT, August 6, though by 1200 GMT, August 7 the low had substantially lled in The 42 km resolution run of MC2 more or less correctly captures the behaviour of this centre of low pressure However, the 21 km resolution run allows the low pressure centre to persist, and as can be seen from the 53 km resolution wind elds, the low is placed over the eastern end of Lake Ontario by the model during the early morning of August 8 The increased pressure gradient across southern Ontario due to the low pressure centre over eastern Lake Ontario, leads to the strong northerly gradient ow predicted by the model for the early hours of August 8 Figure 55 presents the observed 10 m winds from hourly synoptic observations and the Hastings SONTOS site over southern Ontario Winds at the surface are generally from the west and windspeeds are low: typically less than 15 ms ;1 Though the observations are scarce, there is some evidence of a land breeze circulation with winds owing towards the lake at St Catherines, Toronto Island and Kingston Compared with observations, the model calculated wind eld is too strong and has limited the 199

12 Figure 54: Horizontal and vertical cross-sections of the CO concentration and wind eld at 1100 GMT, August 8, 1993 calculated by the model at a horizontal resolution of 53 km The horizontal cross-section shows the CO concentration in the lowest 10 m of the model and winds at 10 m above the surface The position of the vertical cross-section is given by the thick line drawn across the horizontal cross-section panel 200

13 formation of a land breeze circulation Figure 55: The observed wind at 10 m above the ground for 1100 GMT (0700 EDT) on August 8, 1993 The station location is given by the solid dot at the end of each barb The direction of the wind is given by the direction of the barb and an observation of calm conditions is denoted by an unlled circle Wind speed is given by the tick marks attached to each barb Windspeeds less than 13 m s ;1 are given by a barb with no ticks Windspeeds between 13 and 38 ms ;1 are denoted by a barb with a half tick mark and windspeeds between 38 and 64 m s ;1 are shown by abarb with a full tick mark The stations plotted are: 1) Toronto International Airport 2) Toronto Island 3) Peterborough 4) Hastings 5) Waterloo 6) Hamilton 7) St Catherines 8) Kingston and 9) Ottawa The wind eld and CO concentration predicted by the model at 1600 GMT on August 8 is given in Figure 56 By 1600 GMT the model has begun to produce onshore ow associated with a lake breeze in the lowest layers of the model (panel a) The strong northerly ow across the western end of Lake Ontario has been replaced by light winds which, though not well represented in Figure 56, show several poorly dened centres of divergent ow During the initial development of the lake breeze the gradient ow over the lake weakens for levels less than 200 m above the surface The eect on precursor emissions 201

14 Figure 56: Horizontal and vertical cross-sections of the model calculated CO concentration and wind eld at 1600 GMT, August 8, 1993 The horizontal cross-sections show these elds at 10 m (panel a) and 435 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections 202

15 can be seen in the distribution of CO near the surface (panel a) With the weakening of the gradient ow over the lake, precursors emitted from Toronto have become trapped over the lake and have formed a `pool' of precursors over the surface of the lake and below approximately 200 m With the strengthening of the onshore ow of the lake breeze, the precursors are advected both towards the shore around the western end of Lake Ontario, and in an easterly direction along the long axis of the lake The onshore ow component of the lake breeze is initially quite shallow and deepens with time As can be seen in the east{west vertical cross-section of Figure 56 (panel a), the advection of precursors towards the east only occurs for levels below approximately 100 m above the surface The initially shallow onshore components of the lake breeze circulation results in a vertically thin layer of precursors advected towards the east At greater heights above the surface, shown in panel (b) of Figure 56, the direction of the gradient wind has been largely unaected by the lake breeze, and emissions from Toronto are still advected towards the south The north{south vertical cross-section shown in panel (b) of Figure 56 shows that the depth of the Toronto plume decreases towards the south Portions of the Toronto plume near the southern shore of Lake Ontario had been emitted from Toronto earlier in the morning and were trapped by a shallow boundary layer before being advected over the lake As the boundary layer increased in depth during the morning, emissions from Toronto were mixed over a deeper and deeper layer before being advected over the lake The changing vertical depth of the plume over the lake is a reection of the boundary layer depth over Toronto at the time the emissions occurred The observed 10 m winds at 1600 GMT are shown in Figure 57 Away from 203

16 Lake Ontario the windspeeds were generally less than 2 ms ;1 and the winds were from a westerly to northwesterly direction Formation of a lake breeze over Lake Ontario is evidenced by the onshore ow observed at St Catherines, Toronto Island and Kingston The winds observed at Kingston, for example, were from the southwest at 4 ms ;1 Figure 57: The observed wind at 10 m above the ground for 1600 GMT (1200 EDT) on August 8, 1993 A comparison of the model wind elds with observations shows that the model predicts winds inland from Lake Ontario to be from a northerly to northeasterly direction, while observations show winds were from a more westerly direction Errors in the direction of the gradient wind calculated by the model appear to be largely the result of the erroneous centre of low pressure, positioned just to the southeast of Lake Ontario at 1600 GMT Eects of the low pressure centre on the model solution decrease with time as the low moves further to the east and the central pressure is calculated by the model to increase Windspeeds calculated by the model north of Lake Ontario are between2and3ms ;1 and agree reasonably well with the observed 204

17 windspeeds Winds observed at Toronto Island were calm or light from the north until 1300 GMT, 0900 EDT Beginning at 1300 GMT the winds were from the south to southwest at speeds from 1 to 2 m s ;1 Observations at St Catherines and Kingston show onshore ow beginning an hour later, at 1400 GMT The MC2 model does not show the onset of an onshore ow until 1600 GMT, and appears to be late in developing a lake breeze circulation on this day Figure 58 shows the model CO concentration and wind elds at 2100 GMT (1700 EDT) The 10 m winds, given in the horizontal cross-section of panel (a), show the development of a mesohigh over the extreme southwestern corner of Lake Ontario A broad band of southwesterly winds, with speeds between 4 and 6 ms ;1, exist over much of the lake From the vertical cross-sections shown, the onshore ow of the lake breeze is approximately m deep at 2100 GMT The winds at 435 mabove the surface, shown in the horizontal cross-section of panel (b), are, therefore, near the top or slightly above the onshore ow of the lake breeze circulation The `pool' of CO which was present over the western end of Lake Ontario at 1600 GMT, just as the lake breeze was beginning to form, has been dispersed by the strengthening lake breeze circulation A signicant fraction has been advected inland around the western end of Lake Ontario, while a second portion of the CO has been advected towards the northeast over the lake The plume of CO advected towards the northeast can be seen in the horizontal cross-section of panel (a) to the southeast of Toronto Vertically, this plume is less than 100 m thick and is not separated from the surface of the lake As shown by the north-south vertical cross-section of panel (b), the plume of CO 205

18 Figure 58: Horizontal and vertical cross-sections of the model calculated CO concentration and wind eld at 2100 GMT, August 8, 1993 The horizontal cross-sections show these elds at 10 m (panel a) and 435 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections 206

19 from Toronto has been undercut by less polluted air within the northward, onshore ow of the lake breeze The less polluted air originated along the southern shore of Lake Ontario, between 200 and 500 m above the surface Entrained into the lake breeze as the circulation strengthened between 1600 and 2000 GMT, the less polluted air was advected towards the north and eventually brought to the surface by the subsidence associated with the lake breeze At this point in time, the bulk of the emissions from Toronto are being forced over top of the onshore ow of the lake-breeze, resulting in the formation of an elevated plume The model CO and wind elds at 0000 GMT, August 9 are illustrated in Figure 59 As above, the CO and wind elds are shown for two horizontal cross-sections, at 10 and 435 m above the surface Note that the scale used to contour the CO concentration has changed from that used above The 10 m wind eld shows that the mesohigh remains in the southwest corner of Lake Ontario, with the near-surface component of the lake breezecirculation consisting of a band of southwesterly to westerly winds which cover most of the lake The onshore ow of the lake breeze is now approximately 500 m deep and the winds at 435 m above the surface (panel b) show a considerable onshore component One of the dominant features of the CO concentration eld at 10 m is the increase in concentration of CO over regions with high emissions The nocturnal inversion has begun to form over land and, with slower rates of vertical mixing, the CO concentration rapidly increases for regions where emissions occur The concentration of CO in the lowest model layer over the western end of Lake Ontario has increased from approximately 140 ppb at 2100 GMT to 180 ppb at 0000 GMT, August 9 The increase is caused by the entrainment of emissions from 207

20 Figure 59: Horizontal and vertical cross-sections of the model calculated CO concentration and wind eld at 0000 GMT, August 9, 1993 The horizontal cross-sections show these elds at 10 m (panel a) and 435 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections 208

21 Toronto into the return ow of the lake breeze As was noted above, during the late afternoon emissions from Toronto, advected to the south by the gradient ow, are forced to rise over the onshore ow of the lake breeze A certain fraction of the emissions that rise over the onshore ow are entrained in the return ow, eventually to enter the onshore ow The increase in the near-surface CO concentration over the western end of Lake Ontario is caused by this phenomenon Panel (a) of Figure 59 shows a north{south vertical cross-section through a shallow layer with higher concentrations of CO This shallow plume, calculated by the model to be less than 100 m thick, is being advected inland along the north shore of Lake Ontario by the lake breeze circulation The plume was originally a portion of the `pool' of CO which formed over the western end of Lake Ontario as the lake breeze circulation was beginning at 1600 GMT The CO distribution at 435 m above the ground, panel (b) of Figure 59, shows the eects of the increasing depth of the lake breeze onshore ow As was discussed above, at 2100 GMT the onshore ow onlyweakly aected the wind at 435 m and CO emitted from Toronto was advected to the south, across the lake, in the gradient ow (for the moment we are ignoring the fraction of the plume entrained within the lake breeze circulation) As the lake breeze onshore ow continues to deepen with time, portions of the plume which had been advected to the south, in the gradient ow, are caught in the onshore ow and advected to the northeast The gradual deepening of the onshore ow appears to result in the entrainment of a signicant amount of ozone precursors into the onshore ow of the lake breeze for situations where the gradient ow advects material from urban areas over the lake The observed 10 m winds at 0000 GMT, August 9 are shown in Figure

22 Winds observed at Toronto Island and St Catherines were calm, suggesting that the lake breeze circulation had weakened considerably by this time In fact, observations at St Catherines show the wind becoming oshore one hour later Figure 510: The observed wind at 10 m above the ground for 0000 GMT, August 9, 1993 (2000 EDT, August 8) While observations suggest a weakening of the lake breeze circulation, the model continues to show strong onshore ow associated with the lake breeze at this time A tendency of models to be too slow in terminating onshore ow was noted by Lyons et al (1995), where it was found that the termination of onshore ow predicted by RAMS, for a case study over Lake Michigan, was as much as 35 hours later than observed The observed winds at Peterborough changed from northwesterly at 2200 GMT to southwesterly at 2300 GMT The change in wind direction observed at Peterborough occurred at approximately the same time as the putative lake breeze front was observed to arrive at Hastings, 30 km to the west of Peterborough The observed surface winds at Peterborough and the observed chemical and meteorological variables 210

23 at Hastings suggest that the lake breeze front had penetrated at least 40 km to the north of the Lake Ontario shoreline by 2300 GMT From Figure 59, it can be seen that the inland penetration of the lake breeze is signicantly underpredicted by the model for this day The model lake breeze never penetrates far enough inland to arrive at the Hastings eld measurement site At its closest, the model lake breeze front is approximately 20 km south of Hastings As was discussed earlier, the incorrect treatment of a low pressure centre contributed to the model calculating a northerly gradient ow, during the early morning hours, that was stronger than observed Later in the period the inuence of the low pressure system on the 53 km resolution results has weakened, however the model continues to predict northwesterly ow from Georgian Bay Typical 10 m windspeeds calculated by the model north of Lake Ontario are between 5 and 7 m s ;1, while observations show variable winds, though generally from a westerly direction, with speeds less than 2 ms ;1 Errors in the gradient ow to the north of Lake Ontario undoubtedly contribute to limiting the inland penetration of the lake breeze front on this day Figure 511 shows the model calculated ozone concentration and wind eld at 2100 GMT for the model levels at 10 m (panel a) and 220 m (panel b) above the surface The ozone concentration in the lowest model level shows concentrations greater than 70 ppb along the north shore of LakeOntario, just to the east of Toronto This area of higher ozone concentrations is associated with the thin layer of CO, and other ozone precursors, advected to the northeast from the `pool' of precursors which formed over the western end of Lake Ontario As can be seen from the vertical crosssection in panel (a), the higher ozone concentrations are found at heights below 250 m, 211

24 with even higher ozone concentrations below 100 m The ground-level concentration of ozone decreases rapidly as this plume is advected onshore due to increased vertical mixing over land Figure 511: Horizontal and vertical cross-sections of the model calculated ozone concentration and wind eld at 2100 GMT, August 8, 1993 The horizontal crosssections show these elds at 10 m (panel a) and 240 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections A second region of high ground-level ozone concentrations is predicted at the western end of Lake Ontario This region of high ozone concentrations is associated with the onshore ow of the lake breeze bringing higher ozone concentrations from over the lake Panel (b) of Figure 511 shows the model ozone concentration and wind eld at 240 m above the surface and a north{south vertical cross-section across the western end of Lake Ontario The vertical cross-section shows that over the lake, the 212

25 highest concentrations of ozone are found aloft As was discussed earlier, the ozone precursors emitted from Toronto and advected southward in the gradient ow are forced to rise over the onshore ow of the lake breeze which is bringing less polluted air northward along the surface of the lake The resulting layer of ozone aloft is advected towards land by the lake breeze and mixes down to ground-level due to increased vertical mixing over land 521 Forward Trajectory Analysis For the purpose of visualizing how the lake breeze aects the dispersion of emissions from Toronto, three-dimensional forward trajectories have been generated from the model calculated wind elds Figure 512 shows seven dierent forward trajectories from Toronto The rst forward trajectory was begun at 1000 GMT and a new trajectory was started every hour This set of trajectories were all begun at a height of 20 m above the surface Trajectory (a) left Toronto in the early morning, before the lake breeze had developed, and continued south across the western end of Lake Ontario Trajectory (b) initially followed a similar path as trajectory (a), though became caught in the lake breeze circulation as it developed and was eventually advected towards the north-east within the onshore ow of the lake breeze Several of the subsequent trajectories (trajectories c, d, e and f) became entrained within the lake breeze in a similar manner as trajectory (b), though were initially advected towards the northwest by the onshore ow of the lake breeze This set of trajectories ascended at the lake breeze front, to the west of where they were released, returned towards the center of the lake in the return ow aloft, then re-entered the onshore ow of the lake breeze and were 213

26 Figure 512: Three-dimensional forward trajectories from Toronto for August 8, 1993 The trajectories were begun at one hour intervals between 1000 and 1600 GMT and were released at a height of 20 m above the surface The colour of the line represents the height of the trajectory in metres above ground level advected towards the northeast Trajectory (g) shows an example of emissions from Toronto entrained into the onshore ow of the lake breeze aloft Trajectory (g) initially moved towards the south, then was forced to rise over the lake breeze front before descending while over the lake and eventually becoming entrained in the onshore ow of the lake breeze Note that many of the forward trajectories complete at least one full recirculation within the lake breeze circulation, however trajectories calculated in the manner used here ignore the eects of turbulent mixing on the dispersion of pollutants While many of the trajectories show a tendency to undergo recirculation within the lake breeze, the fraction of precursors that undergo recirculation may be considerably smaller than one (Lyons et al, 1995) 214

27 522 Eects of Dry Deposition The slow rate of dry deposition for many species over water has been suggested as a contributing factor in the production of high concentrations of ozone over lakes (Sillman et al, 1993) The slow rate of dry deposition results from the low solubility of certain species and the high degree of atmospheric stability over water, due to the formation of a conduction inversion during the daytime Model calculated dry deposition rates for ozone over Lake Ontario are not larger than 004 cm s ;1 For a layer of ozone 100 m deep, a deposition velocity of 004 cm s ;1 gives a time constant for the loss of ozone by dry deposition of approximately 70 hours Thus, over the course ofaday, lossofozoneby dry deposition to the lake surface is negligible, even for the case of a shallow layer of higher ozone concentrations in physical contact with the lake surface For elevated layers of ozone, loss by dry deposition will be even smaller Similarly, dry deposition velocities for NO 2 and PAN are smaller over water than over land Typical midday deposition velocities for NO 2 over land are calculated to be between 04 and 06 cm s ;1, while over water the NO 2 deposition velocities are no larger than 0002 cm s ;1 The dierence in dry deposition velocities between land and water surfaces for PAN is not as large as that calculated for NO 2, with PAN deposition velocities typically a factor of three times smaller over water than over land Henry's Law constants for NO 2 and PAN show that NO 2 is considerably less soluble than PAN and therefore NO 2 has a smaller deposition velocity over water surfaces For the more water soluble oxidation products such ashno 3 and H 2 O 2, the lower rates of dry deposition calculated over the lake are the result of greater aerodynamic 215

28 resistance, due to the increased stability over the lake During the daytime, model dry deposition velocities for HNO 3 range from 001 to 04 cm s ;1 over the lake, while over land the range is from 15 to 40 cm s ;1 For H 2 O 2, deposition velocities over water are calculated to vary between 001 and 04 cm s ;1, while over land the range is from 07 to 15 cm s ;1 From the brief analysis of dry deposition velocities over water, it appears that the loss of ozone and NO 2 by dry deposition will be negligible for an airmass over water As suggested by Sillman et al (1993), higher rates of dry deposition would most certainly have resulted in considerably lower concentrations of ozone over the lake This is particularly true for shallow plumes of ozone, predicted by the model to be less than 100 m deep, for which appreciable ozone deposition velocities would have had a signicant eect The model also calculates the generation of elevated plumes of ozone, and for these elevated plumes it can reasonably be expected that higher deposition velocities will have less of an eect since the plumes are physically separated from the surface 53 Case Study: August 26, 1993 The second lake-breeze case study occurred on August 26, 1993 Chemical observations at the Hastings eld site showed that within 10 minutes, beginning at 2039 GMT (1639 EDT), the concentration of ozone increased from 46 to 59 ppb The concentration of ozone continued to increase, though more slowly, until 2144 GMT, at which time a concentration of 79 ppb was observed Likewise, the concentration of NO y increased from 22 to 38 ppb during the rst 10 minutes of the event, then continued to increase to 46 ppb at 2144 GMT See Figure 51 for the observed ozone and NO y 216

29 concentrations at Hastings on August 26 Figure 513 shows the synoptic conditions for 0000 GMT, August 26 and 0000 GMT, August 27, 1993 as taken from the CMC objective analysis A large ridge of high pressure lay just to the west of Lake Ontario at 0000 GMT, August 26 and winds from from the northwest Over the course of the day the ridge moved across southern Ontario to lie to the south and east of Lake Ontario, and the 850 mb winds backed to become southwesterly by 0000 GMT, August 27 Figure 513: Sea-level pressure and 850 mb winds taken from the CMC objective analysis for 0000 GMT August 26, 1993 (panel a) and 0000 GMT August 27, 1993 (panel b) Surface observations, from stations away from Lake Ontario, showed that winds were from a west to southwesterly direction during the afternoon of August 26 with speeds less than 2 m s ;1 Daytime high temperatures from 31 to 33 Cwere recorded across southeastern Ontario Kingston was an exception With winds from the lake for most of the day, the high temperature at Kingston was 26 C 217

30 Figure 514 shows the CO concentration and wind eld in the second model layer, approximately 20 m above the surface, at 1100 GMT The wind eld aloft is quite weak, with model winds at 1 km above ground less than 2 ms ;1 With no strong synoptic forcing the wind eld near the surface is predicted by the model to be quite complex A region of strong convergence is predicted by the model along the southern shore of Lake Ontario A second region of strong convergence is predicted along the northern shore of Lake Erie Northerly ow across Lake Ontario bifurcates along the southern shore with one branchmoving to the southwest and the other branchmoving to the east Emissions from Toronto are advected to the southwest and eventually exit the 53 km resolution domain along the western edge of the model The observed surface winds at 1100 GMT are shown in Figure 515 The model wind eld shows a signicant region of convergence over Lake Ontario as part of a land breeze circulation Observations at Toronto Island and St Catherines seem to support the occurrence of a land breeze with both stations reporting oshore ow with windspeeds of 17 ms ;1 Aside from the probable occurrence of a land breeze circulation, the observations are not suciently numerous to support or refute other features of the model wind eld Figure 516 presents horizontal and vertical cross-sections of the CO concentration and wind eld calculated by the model for 1600 GMT The 10 m winds show acentre of divergence to the south and east of Toronto and onshore ow over all shores of Lake Ontario Inland penetration of the lake breeze appears to be particularly deep to the northwest of Lake Ontario, in the vicinity of Toronto As will be seen in the results for subsequent hours, the deep inland penetration of the onshore ow over this region collapses and a new lake breeze front forms nearer the lakeshore It is unclear 218

31 Figure 514: Horizontal and vertical cross-sections of the CO concentration and wind eld at 1100 GMT, August 26, 1993 The horizontal cross-section shows the CO concentration and wind eld at 20 m above the surface The position of the vertical cross-section is given by the thick line drawn across the horizontal cross-section panel 219

32 Figure 515: The observed wind at 10 m above the ground for 1100 GMT (0700 EDT) on August 26, 1993 whether such behaviour is realistic and the factors which have caused it have not been analyzed One possible cause may be instabilities in the model solution caused by the application of boundary conditions, since the region in which this feature occurred is quite close to the western boundary of the model domain The concentration of CO in the lowest model layer shows a plume from Toronto spreading along the northern shore of the lake A vertical cross-section through the eastern edge of the plume shows weak onshore ow, approximately 100 m deep Over the lake the plume is less than 200 m deep, while over land, due to greater vertical mixing, the plume is approximately 500 m deep Panel (b) of Figure 516 shows the concentrations of CO and the wind eld at 435 mabove the surface The vertical cross-section in panel (b) shows that the boundary layer over Toronto at 1600 GMT is calculated by the model to be approximately 600 m deep Winds at 435 m are from the west and are advecting the Toronto plume along the northern shore of Lake Ontario 220

33 Figure 516: Horizontal and vertical cross-sections of the model calculated CO concentration and wind eld at 1600 GMT, August 26, 1993 The horizontal cross-sections show these elds at 10 m (panel a) and 435 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections 221

34 A comparison of the model calculated winds at 10 m above the ground with observations at 1600 GMT, shown in Figure 517, suggest no signicant errors in the model simulation of the surface winds An exception to the general agreement between observations and the model is the observed wind at St Catherines Observations show that the wind at St Catherines at 1600 GMT was from the south-southwest at 4 m s ;1, while the model is calculating light onshore, or northerly, ow associated with the lake breeze The disagreement suggests that the model lake breeze front was too far inland along the southwestern shoreline of Lake Ontario However, the winds observed at St Catherines do become onshore the following hour Figure 517: The observed wind at 10 m above the ground for 1600 GMT (1200 EDT) on August 26, 1993 The CO concentration and wind eld calculated by the model for 2000 GMT are shown in Figure 518 The model shows a well developed lake breeze circulation with strong onshore owover all shores of the lake A mesohigh is centered over the western half of Lake Ontario The model lake breeze front has moved inland approximately 15 to 20 km along the north shore of Lake Ontario in the vicinity of Hastings Analysis 222

35 of the inland penetration of the lake breeze front from visible satellite imagery for this day suggests that the front was approximately 30 km north of the Lake Ontario shoreline by 2000 GMT (Hastie et al, 1998) Figure 518: Horizontal and vertical cross-sections of the model calculated CO concentration and wind eld at 2000 GMT, August 26, 1993 The horizontal cross-sections show these elds at 10 m (panel a) and 435 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections An interesting feature of the spatial distribution of CO in the lowest model level is the lower concentrations present over the lake and the much higher concentrations just inland over the northwestern shore of Lake Ontario Panel (b) of Figure 518 shows the model CO concentration and wind eld at 435 m above the surface The winds at this level are from the west along the northern shore of Lake Ontario and act to advect the Toronto plume to the east Near the surface, in the lake breeze onshore 223

36 ow, the winds are from the southwest and undercut the Toronto plume The onshore circulation of the lake breeze is advecting less polluted air onshore, underneath the eastward progressing plume from Toronto As the onshore ow moves further inland, vertical mixing rapidly mixes the higher concentrations of CO down to ground-level, giving rise to the strong horizontal concentration gradients seen at the surface Though the model resolution is not sucient to resolve such a feature, the onshore advection of less polluted air near the surface, with higher concentrations of pollutants aloft, could be expected to result in a ground-level concentration pattern similar to that found by Lyons and Cole (1976) along the western shore of Lake Michigan: within 1 km of the shoreline concentrations of ozone were found to be lower than those observed further inland The explanation proposed by Lyons and Cole was that an elevated plume of ozone, which existed over the lake, was mixed down to the surface by a deepening TIBL as the onshore ow moved further inland Though there are no observations of a similar phenomenon occurring along the shores of Lake Ontario, the large scale features predicted by the model for this case would result in a similar pattern in the ground-level concentration The nal hour analyzed is 0000 GMT, August 27 (2000 EDT, August 26) The CO and wind elds calculated by the model for this time are shown in Figure 519 The wind elds and the spatial distribution of CO are largely a natural evolution of the patterns found in the model output for 2000 GMT The plume of CO has advected further to the east along the north shore of LakeOntario and continues to be undercut by less polluted air in the onshore ow of the lake breeze In regions where the onshore ow is undercutting the Toronto plume the horizontal concentration gradient at ground-level is not as strong as was seen earlier With slower vertical mixing over 224

37 land, the onshore ow of less polluted air is able to penetrate further inland before the higher concentrations of CO aloft are mixed down Figure 519: Horizontal and vertical cross-sections of the model calculated CO concentration and wind eld at 0000 GMT, August 27, 1993 The horizontal cross-sections show these elds at 10 m (panel a) and 435 m (panel b) above the surface The position of the vertical cross-sections are denoted by the thick solid lines drawn over the horizontal cross-sections Note that the concentration of CO in the plume is decreasing as the plume is advected further to the east, presumably as a result of the entrainment of less polluted air into the solenoidal circulation of the lake-breeze The northern end of the vertical cross-section drawn through panel (a) corresponds to the approximate location of the Hastings observation site As can be seen from the surface wind eld, the model is not able to reproduce the observed inland penetration of the lake breeze front At 0000 GMT, August 27 the model lake breeze front was 225

38 still approximately 15 km south of Hastings Note that the lake breeze front was observed to arrive at Hastings at 2030 GMT The vertical cross-sections presented above suggest signicant horizontal and vertical motions aect the Toronto plume as it is advected towards the east along the north shore of Lake Ontario For example, the vertical cross-section in panel (a) of Figure 519 shows the lake breeze inow layer is approximately 500 m deep Above the onshore ow the return circulation extends to approximately 1000 m above the surface Vertical motions, such as ascent at the lake breeze front and subsidence behind the front, are also evident in the ow elds Figure 520 shows a series of forward trajectories from Toronto, released at one hour intervals beginning at 1200 GMT The forward trajectories suggest a solenoidal circulation as the plume is advected towards the east As rst proposed by Lyons and Cole (1976), with gradient ow roughly parallel to the lakeshore material is advected in the general direction of the gradient ow, though undergoes solenoidal circulations due to the inuence of the lake breeze circulation Figure 521 shows the concentration of ozone calculated for the lowest model level at 2000 GMT, August 26 and 0000 GMT August 27 The spatial distribution and interaction of ozone with the lake breeze is quite similar to that described above for CO: ozone is photochemically generated within the Toronto plume and advected towards the east, along the north shore of Lake Ontario by a combination of the gradient ow and lake breeze circulation The model calculates a ground-level ozone concentration of 84 ppb approximately 15 km to the south of Hastings at 0000 GMT, August 27 (2000 EDT) 226