Factors Influencing The Concentrations of A Traffic Related Pollutant In The
Vicinity Of A Complex Urban Junction.
A.S. Tomlin*
1
, R.J.Smalley
1
, J.E. Tate
2
, S.J. Arnold
3,4
, A. Dobre
4
. J.F. Barlow
4
and S.E.
Belcher
4
.
1 Energy and Resources Research Institute (ERRI), 2 Institute for Transport Studies,
1,2
University of Leeds, Leeds, LS2 9JT, UK.
3
Department of Environmental Science and Technology, Imperial College London SW7
2AZ, UK .
4
Department of Meteorology, University of Reading, RG6 6BB, UK.
* Corresponding author, tel: +44113 3432500, fax:+44113 2467310, E-mail fueast@leeds.ac.uk
ABSTRACT
The study focuses on the dispersion of a traffic related pollutant within an area close to a busy intersection between two street canyons in Central London. Depending on the rooftop wind direction, both flow channelling and recirculation regimes were identified within the canyon and intersection. At the intersection, merging of channelled flows from the canyons increased the flow complexity and turbulence intensity. The intersection recorded the highest measured carbon monoxide concentrations ([CO]), independent of roof-top wind direction, which can be attributed to nearby queuing traffic. Within the main streetcanyon, a helical flow regime was found for oblique roof-top flows, leading to increased
[CO] on the canyon leeward side – influence of wind speed – signifying advection? – background wind direction of 90-180 at high wind speeds gives highest concentrations which results in in-street flows of 270-360 i.e. advection from junction at an oblique angle towards sensor along street. For all locations, small changes in the background wind direction caused large changes in the in-street mean wind angle and local turbulence intensity under certain background conditions, implying that the dispersion of pollutants within the streets could be highly sensitive to small changes in above roof wind direction.
During the period of the study the predominant wind direction led to significantly higher diurnal average [CO] on the canyon leeward side than the windward side. During peak periods, concentrations within parallel side-streets were approximately four times lower than within the main canyon and intersection.
Overall diurnal average is higher at the intersection – but the highest 15 minute concentrations are found in canyon under recirculating flow creating high leeward pollution.
[CO] at Intersection less dependant on wind direction and wind speed. – close to an area source?, increased local turbulence intensities.
1. INTRODUCTION
Urban areas are susceptible to elevated concentrations of pollutants due to the high density of traffic and also the dense arrangement of buildings that inhibits the ventilation of pollutants emitted at street level. The accurate prediction of concentrations, necessary for air quality management, requires knowledge of the modified wind flow between buildings as well as the magnitude and location of traffic emissions. Many previous studies of dispersion within urban areas have concentrated on the idealised case of the 2D street canyon (Pavageau & Schatzmann, 1999, Sini et al., 1996, Louka et al., 2000). The skimming flow regime that may occur during near perpendicular winds for canyons with height to width ratios between 0.33 and 1 (Oke, 1987, Leonardi et al, 2003) has received much attention, since pollutants released by vehicles at street level are then transported by a single across-canyon vortex leading to elevated concentrations on the canyon leeward side relative to the windward side. The presence of such a flow regime may also affect the extent of vertical mixing and thus the flux of pollutants out of the canyon. Such recirculating flows have been observed in canyon studies in the field by for example
DePaul and Sheih (1986), Boddy et al. (2005), Longley et al., (2004) and in wind tunnel studies such as Hoydysh and Dabberdt (1988). Most cities however, consist of complex road networks incorporating street canyons, intersections and side streets. In addition, an understanding of the influence of the whole range of background wind directions should be gained in order to assess the influences on potential exposure within real streets. For example, in a numerical study into the effects of background wind direction on flow and dispersion in short street canyons (Kim and Baik, 2004), three in canyon flow patterns were identified with considerably different dispersion characteristics depending on the incidence wind angle. The study showed that as the incidence angle became oblique to the street axis more pollutants escaped from the street canyon.
A range of field studies are therefore necessary in order to asses whether features such as cross canyon vortices, present in unbroken canyons under near parallel background winds, are relevant to real city environments, and as to whether additional features may result from the presence of side streets and intersections (Boddy et. al., 2005, Dobre, et al.,
2005). The influence of such features on the dispersion of traffic related pollutants is of importance in determining the key factors that may affect road side concentrations and therefore exposure within urban environments. In addition, despite the relatively good coverage of monitoring stations within UK cities, questions remain as to whether they accurately reflect the potentially broad range of pollution levels experienced within urban streets where concentrations can vary greatly within a very small distance depending on local meteorology, traffic density and vehicle speeds. Wind tunnel measurements (Robins et al. 2002) suggest that intersections play an important role in dispersing passive scalars, and that even small asymmetries in geometry or wind direction lead to very different dispersion patterns. In addition, personal exposure to pollutants is heavily dependant on the in-street location, and it is therefore important to determine the causes of variability in personal exposure to air pollution. To explore the impact of such complexities, this study focuses on a busy intersection between two street canyons (Marylebone Road and
Gloucester Place) in Central London (NW1) and forms part of the DAPPLE (Dispersion of
Air Pollution and Penetration into the Local Environment) field campaign (Arnold et al.,
2004).
2. METHODOLOGY
Marylebone Road is a busy 6-laned major arterial and forms the northern boundary of the
London congestion charging zone. Gloucester Place is a one-way, 3-laned street. The streets intersect perpendicularly and could potentially be considered as street canyons away from the intersection. An overview of DAPPLE and a comprehensive description of the experimental site and set-up are presented in Arnold et al . (2004) and at www.dapple.org.uk. Figure 1a presents a schematic view of a subsection of the site showing approximate building heights around the junction and the position of instruments used in the present analysis. Throughout, all angles are from directions (in degrees, increasing clockwise) relative to Marylebone Road where 0° are winds from the westsouth-west.
The in-canyon and intersection velocity field and CO concentrations were measured between 20 April and 11 June 2004. The wind velocity was measured at 6 locations on 4 sites within the street using 3-axis ultrasonic anemometers sampling at 20 Hz. All anemometers were aligned parallel to Marylebone Road and mounted on brackets projecting the instruments 0.4 m from supporting lampposts. Those used in this study were at nominal heights of 4 m or 8 m (lower or upper, respectively) and data from 15 minute averages were used. [CO] were measured at 14 sites using an electrochemical method
(Learian Streetbox sensors) recording 15 minute averages of 1 minute samples. CO was chosen since it is relatively inert on the time-scales considered and therefore provides a useful traffic related tracer. The sensors used in this study were all located at 4.0 m. Colocation of a Streetbox at the ‘Supersite’, one of the London Air Quality Network (LAQN code: MY1) monitoring sites, for the purpose of calibration showed a greater than 90% accuracy for the Streetboxes (Arnold et al., 2004). In addition, two background, or roof-top ultrasonic anemometers were located on a nearby building roof (Westminster Council
House - WCH), 17m above the canyon floor in order to determine the above roof wind direction. A separate study …(Adrian and Sam’s work to be referenced here?)
The traffic flow characteristics (need exact location) were obtained as a by-product of the
SCOOT (Split, Cycle and Offset Optimisation Technique) demand-responsive Urban
Traffic Control system ( Hunt et al., 1991 ) in collaboration with Transport for London.
SCOOT provides a variety of traffic flow measures such as flow and lane occupancy from piezo-electric loop detectors embedded close to the upstream end of a link, or road section.
Average daily traffic totals (ADT) will be used in the following analysis. Manual traffic counts at the site were undertaken as a means to check/calibrate the SCOOT data, in particular, to assess the accuracy of single SCOOT loops used over more than one lane.
Need to say any more than this? James?
3. RESULTS AND DISCUSSION
Air Flow Characteristics
Figure 1b shows the roof-top wind direction (
rt
) frequency recorded on the WCH library roof. The dominant direction was from approximately 120°, resulting in flows oblique to
Marylebone Road. It is possible that structures upwind of the roof-top site for 140° ≤
rt
<
180° may have affected the reference measurement when the background wind originated from these directions. Data from a second nearby roof-top site indicate that these wind directions were infrequent and therefore only the WCH library roof data is used in this study. Reference the other paper and update angles if relevant based on Janet’s re-analysis
– WCC tower in the way?
Depending on the roof-top wind direction, both flow channelling and recirculation regimes were identified within the canyon and intersection. Figure 2a shows that at site 1 lower, located within the intersection, the flow is approximately channelled along Marylebone
Road for 130° ≤
rt
< 240° and 310° ≤
rt
< 50°. Some evidence of channelling along
Gloucester Place (i.e. in-street flows of around 90° or 270°) is also seen for 270° ≤
rt
<
315°. Rapid switching of the in-street flow is seen to occur due to quite small changes in roof-top wind direction under some conditions. For example, relatively large increases in
rt
from 180° do not lead to changes in flow direction at site 1 lower. However, when
rt increases above 240° the flow in the intersection rapidly switches to channelling along
Gloucester place. The scatter in the data around
rt
= 280° suggests that even at 15 minute averages the preferred channelling direction can vary (refer to wind tunnel studies – are they published?). At site 1 upper there is still evidence of flow channelling along
Marylebone Road although channelling along Gloucester Place is not as evident as at the lower site. At the upper site, the in-street flow tends to roughly follow the roof-top wind direction for 240° ≤
rt
< 360°.
At site 2 lower, there is less evidence of channelling along Gloucester Place, perhaps because site 2 is set further back from the intersection. Flow is also not directly channelled along Marylebone Road. For 130° ≤
rt
< 270° the in-street wind angle decreases with increasing roof-top wind angle. For
rt
= 130° the wind angle at site 2 is greater than 180°, and as
rt
increases, the in-street angle at site 2 decreases, passing through 180° for a rooftop angle of just over 180° indicating channelled flow, and further decreasing until the flow starts to switch at roof-top winds from 240°. This apparent “reflection” of the flow may be attributed to a helical flow regime established within Marylebone Road up-wind of site 2, as has been reported in other studies (Dobre, et al.
2005, Johnson and Hunter, 1999).
The presence of the helical flow regime is also supported by data from sites 3 and 4 as shown in Figure 3 since the in-street wind angle has a negative dependence on the roof-top wind angle for almost all roof-top winds. Although there is some switching of the flow at sites 3 and 4 for
rt
around 90° and 270° it appears to be less extreme than within the intersection because of the recirculating component. Dobre et al. (2005) proposed that the in street wind angle θ can be related to the roof-top wind conditions using a simple model representing the superposition of the along canyon and across street components of the flow: tan
u u
2
1
u ref 1 uˆ u ref 2 uˆ
2
1
uˆ uˆ
2
1 tan
ref where uref1 is the component of the roof-top flow parallel to the street and uref2 is the perpendicular component. Assuming a uniform along street component, the hatted variables represent dimensionless functions of the canyon aspect ratio. Figure 3b demonstrates the agreement between the model (closed symbols) and observations (open symbols) for site 4. The good agreement supports the view that a helical flow regime is present which will clearly affect the dispersion of pollutants within the canyon. This complex dependence of in-street and intersection flows on roof-top wind direction is in agreement with previous wind tunnel studies (Robins et al. 2002) and implies that the dispersion of pollutants within the streets can be highly sensitive to small changes in above roof wind direction.
At the intersection, other processes are enhancing local turbulence intensity but not the magnitude of the mean winds. Janets point is that turbulence can be high under chanelled flows as a result of higher wind speeds. – so do we look at normalized turbulence with Uref and not Ti or as well as Ti?
Dispersion mechanisms depend not only on in street mean wind flows but also on the turbulence levels. The relative importance of turbulence compared to the mean wind is illustrated by the local turbulence intensity T i
= k
1/2
/U, where k is the turbulent kinetic energy and U is the magnitude of the local mean velocity. The current study shows that T i depends on site location as well as the roof-top wind direction. Figure 4 shows the dependence of T i
on roof-top wind direction within the intersection at site 2 compared to the in-canyon location, site 3. At site 2 a distinct minima in T i
occurs for
rt
≈180°. This is a result of flow being channelled along Marylebone Road leading to higher in-canyon wind speeds with reduced flow disturbance upwind of the intersection. For such channelled flows dispersion of pollutants by the mean wind would be dominant. The maximum T i
at site 2 appears to occur for
rt
just below 270° and just above 90° where the roof-top wind is perpendicular to Marylebone Road. Figure 2 showed that channelling along the direction of Gloucester place was minimal at site 2 and that flow switching occurred around these roof-top angles. It is not therefore surprising that the maximum local turbulence intensity is found under these perpendicular conditions. Turbulence is therefore likely to be relatively more important as a dispersion mechanism under perpendicular background flows than under channelled flows within the intersection (too naïve now rethink).
Constrast dispersion mechanisms at the street canyon and intersection sites.
In general, T i
is higher within the intersection than at the in canyon site and shows a greater degree of variability. Both sites show high T i
for
rt
between 200° and 270°, when the helical flow regime causes downdrafts to travel down the north side of Marylebone
Road. These downdrafts provide a mechanism whereby turbulent kinetic energy may be transported from aloft into the street canyon, thus contributing to the T i
. As
rt
increases above 270° the in street wind switches direction as shown for example in Figure 4b, causing site 3 to be on the leeward side of the canyon and T i
to drop to much lower levels.
The local turbulence intensity is therefore greater on the canyon windward side under flows with a recirculating component. The variations in T i
between sites and for different above roof wind directions provide additional influences on the dispersion of pollutants over and above the mean wind flows. Accurate representations of these variations in turbulence should be a requirement of any dispersion model for urban environments.
Vertical wind angle?
Concentrations at intersection are less dependant on background w direction but there is some dependency which could be related to the strength of updrafts within the intersection.
Complex flow pattern at intersection – likely to lead to more sheer () and therefore mixing.
Apparent drop in concentrations for background wind directions where a strong up draft is present?
For NE direction, as you crank up wind speed the streamlines shift the vertical updraft increases in angle at the intersection and concetrations drop.
Calculate vertical wind speeds from data (sqrt (sqr(total)-sqr(horiz)).
Pollutant Concentrations
Concentrations of the traffic related pollutant CO are shown to be heavily influenced by the flow structures established within the street – combination of this and local source distribution – needs a qualitative description of source distribution. At site 3, located within the Marylebone Road canyon, the sector averaged [CO] is strongly dependant on the roof-top wind angle as shown by the concentration rose in figure 5a. The highest [CO] occurs on the leeward side of the canyon i.e. 0° ≤
rt
< 180°, and for in-street winds from
180° to 360°, indicating the presence of a recirculating flow regime leading to flow
reversal at the canyon floor. Interestingly, the highest concentrations are found when the roof-top wind angle is oblique, rather than perpendicular to the street. For these conditions
(120° ≤
rt
< 150°), where a helical flow regime is suggested, the concentrations are about
3 times those for the opposing roof-top wind direction where the site is in a windward position. The presence of the helical flow structure therefore has a strong influence on local concentrations within the canyon and consequently exposure. Figure 1(b) demonstrated that conditions where 120° ≤
rt
< 130° were the most frequent during the measurement period and therefore site 3 might be expected to experience relatively high concentrations of CO for much of the campaign.
In contrast, figure 5b shows that within the intersection the sector averaged [CO] is less dependent on the roof-top direction. In addition, the dependence of [CO] is similar for instreet and roof-top wind directions, although this does not necessarily indicate that flow reversal is not a feature at this site – see below from Jethro’s thesis. Is this because the sampler is close to an area source – or at least one which extends in several directions. This needs to be re-written to take account of some wind directional influences with wind speed shown in the bivariate plots.
Asymmetry due to Gloucester place being one way.
How do we explain this? The lack of dependence on wind direction means that different dispersion features must come into play for different flow scenarios. Figure 2 shows that for many background wind directions there is no preferred flow direction at site 1 indicating that the flows here are highly dependant on local effects and may be quite intermittent.
The highest [CO] found at the intersection is similar to that found at the leeward location within the canyon. The lowest sector averaged concentrations found within the intersection are however, higher than those found when the monitor at site 3 is in a windward position.
This indicates that in general, average concentrations found within the intersection are
higher than those found within the adjacent canyon. Need to investigate why – check out local wind speeds.
Where channelled flows exist along Marylebone Road i.e. in street winds close to 0° and
180°, there are different effects at site 3 depending on the direction of channelled flow. For winds from ≈ 0/360° the two curves in figure 5a effectively cross on the axis confirming the result shown in figure 2a that there is a good correspondence between background and in street angles for
rt
≈ 0°. For
rt
≈ 180° a wide spread of in street wind angles was experienced at site 3 leading to differences in [CO] for this sector in figure 5a. Why? Local building structures? Refer to reliability of reference measurement – met paper.
Figure 6 shows the mean diurnal [CO] for each site used in the present study, where 15 minute concentrations have been averaged across the same time period during weekdays only. The figure shows that the highest concentrations occur at the morning and afternoon rush hours with a minor late-evening peak. There are significant differences between the mean levels of [CO] for each of the sites, with [CO] at site 1 at least twice that for site 13, and 3 times higher than at site 16 which is situated in a low trafficked side street adjacent to Marylebone Road. High concentrations at the intersection are probably due to the
proximity to queuing traffic from 3 directions. Within Marylebone Road, Site 1 shows the highest mean concentrations throughout the diurnal cycle followed by the in canyon site 3 which, due to the roof-top winds experienced during the study, was most predominantly the leeward location when compared to site 4. This result shows that background meteorology can have a strong influence on spatial variability in concentrations within the canyon over very small distances. The lowest in canyon [CO] occurred at a mid canyon location furthest away from the intersection i.e. site 13. Site 9 is close to the location of the
Supersite within Marylebone Road and is therefore representative of concentrations compared with air quality standards on regular basis. The site is on the Southern side of
Marylebone Road and is therefore influenced by the winds being predominant from the
North Easterly direction. In fact the concentrations at site 9 are similar to those measured at site 4 but are clearly much lower than those measured within or close to the intersection.
This raises some concern that the Supersite measurements are perhaps an underestimation of the highest concentrations found within Marylebone Road even when taking into account longer term trends in meteorology – in abstract?.
Reference surjit Kauer, Colville.
The marked differences between the locations suggests that any attempt to model personal exposure within the area would require accurate representations of traffic related emissions under varying traffic regimes as well as complex turbulent dispersion mechanisms and location within the streets. The concentrations within the side streets parallel to the main canyon can be lower than those within the intersection or on the canyon leeward side by almost a factor of 4 at peak times. This has implications for the control of personal exposure within the streets and suggests that the use of side streets rather than main roads by pedestrians and cyclists traveling through Central London could significantly lower overall exposure.
4. CONCLUSIONS
Depending on the roof-top wind direction, both flow channelling and recirculation regimes were identified within the canyon and intersection. At the intersection, merging of the channelled flow from the two canyons increased the flow complexity and turbulence intensity, compared with the street canyons. Oblique flows across the intersection caused both flow channelling from the adjoining streets as well as evidence of the influence of helical flows from the adjoining canyons. Strong evidence of a helical flow regime was observed in the street canyon locations. For all sites, it was possible for small changes in the roof-top wind direction to cause large changes in the in-street mean wind angle and turbulence intensity, and consequently, significant variation in [CO]. Along Marylebone
Road, the highest [CO] were recorded within the intersection, followed by the most predominantly leeward site. Make the point that this is a diurnal mean – across all background wind directions and that the bivariate plots show higher max [CO] concentrations at the canyon site leeward samplers compared to the intersection site.
Queuing traffic contributed to higher intersection concentrations, relatively independent of roof-top wind direction not proven! JFB. Concentrations within the parallel side-streets were, on average, a factor of three lower than the highest found within Marylebone Road.
The experimental study and analysis shows that modeling exposure within urban areas requires accurate representations of both the temporal and spatial variations in vehicular emissions and the dispersion of pollutants due to the complex air flow regimes that are established within such complex built environments.
5. ACKNOWLEDGEMENTS
We acknowledge EPSRC for DAPPLE funding and JIF support to the LANTERN
consortium, and financial support from NERC. We also thank DAPPLE colleagues, staff at
WCH, Transport for London and the School of Earth and Environment at the University of
Leeds.
6. REFERENCES
Arnold, S., et al. 2004 Dispersion of air pollution and penetration into the local environment, DAPPLE, Sci. Tot. Environ., 332, 139-153.
Boddy, J.W.D., Smalley, R.J., Dixon N.S., Tate, J.E., Tomlin, A.S., 2005 The spatial variability in concentrations of a traffic-related pollutant in two street canyons in York.–
Part I: influence of background winds, Atmos. Environ., in press.
Dobre, A., Arnold, S.J., Smalley, R.J., Boddy, J.W.D., Barlow, J.F., Tomlin, A.S. and
Belcher, S.E. 2005
Flow field measurements in the proximity of an urban intersection in London, UK, submitted to Atmos. Environ.
Johnson, G.T., Hunter, L.J., 1999. Some insights into typical urban canyons airflows,
Atmos. Environ. 33, 3391-3999.
Louka, P., Belcher, S.E., Harrison, R.G., 2000. The coupling between air flow in-streets and the well- developed boundary layer aloft, Atmos. Environ. 34, 2613-79.
Pavageau, M., Schatzmann, M., 1999. Wind tunnel measurements of concentration fluctuations in an urban street canyon. Atmos. Environ. 33, 3961-3971.
Robins, A., Savory, E., Scaperdas, A., Grigoriadis, D., 2002. Spatial variability and source- receptor relations at a street intersection. Water, Air, and Soil Pollution, Focus 2, 381-393.
Sini, J.-F., Anquetin, S., Mestayer, P.G., 1996. Pollutant dispersion and thermal effects in urban street canyons. Atmos. Environ.30, 2659-2677.
Wind fields and turbulence statistics in an urban street canyon
Atmospheric Environment, Volume 40, Issue 1, January 2006, Pages 1-16
I. Eliasson, B. Offerle, C.S.B. Grimmond and S. Lindqvist
60 o 15
90 o
120 o
30 o
10
150 o
5
0 o
15 10 5 5 10 15
180 o
5
330 o
10
210 o
300 o 15
270 o
240 o
(a) (b)
Figure 1 (a) Site schematic, indicating positions of fixed monitoring equipment (not to scale) (b) percentage frequency contribution of wind directions, in 10° sectors, measured at the WCH library roof.
360
270
180
90
0
0 90 180
Wind direction (WCH roof)
270 360
(a)
180
90
360
270
0
0 90 180
Wind direction (WCH roof)
270 360
(b)
360
270
180
90
0
0 90 180
Wind direction (WCH roof)
270 360
(c)
Figure 2 Dependence of in-street wind direction on roof-top wind direction for two intersection sites (a) Site 1 lower (b) Site 1 upper (c) Site 2. All data are 15 minute averages.
360
270
180
90
0
0 90 180
Wind direction (WCH roof)
270 360
(a)
360
270
180
90
0
0
-90
90 180 270 360
Wind direction (WCH roof)
(b)
Figure 3 Dependence of in-street wind direction on roof-top wind direction for (a) Site 3
(b) Site 4, open symbols data, closed symbols model prediction.
100
10
1
0.1
0 90 180
Wind direction (WCH roof)
270 360
(a)
100
10
1
0.1
0 90 180
Wind direction (WCH roof)
270 360
(b)
Figure 4 Dependence of T i
on roof-top wind direction for (a) site 2 lower and (b) site 3.
(a)
30 o
60 o 2.0
90 o
1.5
1.0
120 o
150 o
0.5
0 o
2.0
1.5
1.0
0.5
0.5
0 0.5
1.0
1.5
2.0
180 o
330 o
300 o
1.0
1.5
2.0
270 o
240 o
210 o
30 o
60 o 2.0
90 o
1.5
1.0
120 o
150 o
0.5
0 o
2.0
1.5
1.0
0.5
0.5
0 0.5
1.0
1.5
2.0
180 o
330 o
1.0
1.5
210 o
300 o 2.0
270 o
240 o
(b)
60 o 2.0
90 o
1.5
120 o
30 o
150 o
1.0
0.5
0 o
2.0
1.5
1.0
0.5
0.5
0 0.5
1.0
1.5
2.0
180 o
330 o
300 o
1.0
1.5
2.0
270 o
240 o
210 o
60 o 2.0
90 o
1.5
120 o
30 o
150 o
1.0
0.5
0 o
2.0
1.5
1.0
0.5
0.5
0 0.5
1.0
1.5
2.0
180 o
330 o
300 o
1.0
1.5
2.0
270 o
240 o
210 o c) (d)
Figure 5 – Dependence of sector averaged [CO] on ▲ roof-top
○ in street wind direction for (a) Site 1 lower, (b) Site 2 (c) Site 3, (d) Site 4.
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
00
:0
0
01
:0
0
02
:0
0
03
:0
0
04
:0
0
05
:0
0
06
:0
0
07
:0
0
08
:0
0
09
:0
0
10
:0
0
11
:0
0
12
:0
0
13
:0
0
14
:0
0
15
:0
0
16
:0
0
17
:0
0
18
:0
0
19
:0
0
20
:0
0
21
:0
0
22
:0
0
23
:0
0
Time of day
4000
3500
3000
2500
2000
1500
1000
500
0
Site 1
Site 2
Site 3
Site 4
Site 13
Site 16
Site 9
Traffic flow
Figure 6 - Comparison of vehicles per hour (Marylebone Rd.), diurnally averaged [CO] for several sites.
60 o 0.6
90 o
120 o
30 o
0.4
150 o
0.2
0 o
0.6
0.4
0.2
0.2
330 o
0.4
0.2
0.4
0.6
180 o
210 o
300 o 0.6
270 o
240 o
Figure 7 – Dependence of sector averaged [CO] on ▲ roof-top for Site 16