UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A breath of fresh air? : cyclists' personal exposure to particulate matter along bicycle routes in Vancouver,… Thai, Amy 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-0611.pdf [ 14.56MB ]
Metadata
JSON: 831-1.0093063.json
JSON-LD: 831-1.0093063-ld.json
RDF/XML (Pretty): 831-1.0093063-rdf.xml
RDF/JSON: 831-1.0093063-rdf.json
Turtle: 831-1.0093063-turtle.txt
N-Triples: 831-1.0093063-rdf-ntriples.txt
Original Record: 831-1.0093063-source.json
Full Text
831-1.0093063-fulltext.txt
Citation
831-1.0093063.ris

Full Text

A B R E A T H O F F R E S H A IR? C Y C L I S T S ' P E R S O N A L E X P O S U R E T O P A R T I C U L A T E M A T T E R A L O N G B I C Y C L E R O U T E S IN V A N C O U V E R , BRIT ISH C O L U M B I A by A m y Nguyen Thai B.Sc. (Hons), University of British Columbia, 2005 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T S OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF M A S T E R OF S C I E N C E in The Faculty o f Graduate Studies (Geography) T H E U N I V E R S I T Y OF BRIT ISH C O L U M B I A August 2007 © A m y Nguyen Thai, 2007 T A B L E OF CONTENTS A B S T R A C T : ii L I S T O F T A B L E S . v L I S T O F F I G U R E S vi A C K N O W L E D G E M E N T S viii 1.0 I N T R O D U C T I O N 1 1.1 INTRODUCTION 1 1.2 P A R T I C U L A T E M A T T E R 1 1.2.1 Physical Characteristics 1 1.2.2 Spatial Patterns 3 1.3 A M B I E N T VS. PERSONAL EXPOSURE L E V E L S : 6 1.3.1 Cyclists' Personal Exposure : 7 1.4 OBJECTIVE .... ...9 2.0 M E T H O D O L O G Y 11 2.1 INTRODUCTION 11 2.2 ST U D Y L O C A T I O N 11 2.3 INSTRUMENTATION 13 2.3.1 G R I M M Portable Dust Monitor 13 2.3.2 P-Trak Ultrafine Particle Counter 15 2.3.3 Global Positioning System (GPS) 17 2.3.4 Tapered Element Oscil lating Microbalance ( T E O M ) 19 2.3.5 The Instrumented Bicycle '. 21 2.4 D A T A COLLECTION .- 22 2.4.1 Data Collection Route 22 2.4.2 Min imiz ing the Influence of Temporal Patterns 24 2.5 D A T A PROCESSING 25 2.5.1 G R I M M Data ..' 25 2.5.2 P-Trak Data 28 2.5.3 Land Use Analysis : : 28 2.5.4 Normalization 30 2.5.5 Model Comparison 31 3.0 P A R T I C L E C O N C E N T R A T I O N S A L O N G B I C Y C L E R O U T E S : D A Y - T O - D A Y V A R I A T I O N 33 3.1 INTRODUCTION 33 3.2 D A Y - T O - D A Y V A R I A T I O N 33 3.3 INSTRUMENT CA L I B R A T I O N 37 3.4 EMISSIONS 40 3.5 M E T E O R O L O G Y : 41 3.5.1 Precipitation 43 3.5.2 A i r Temperature 44 3.5.3 Wind Speed 45 3.6 S U M M A R Y 47 in 4.0 P A R T I C L E C O N C E N T R A T I O N S A L O N G B I C Y C L E R O U T E S : S P A T I A L V A R I A T I O N . . 48 4.1 INTRODUCTION 48 4.2 A V E R A G E CONCENTRATIONS A L O N G THE R O U T E .'. 4 8 4.2.1 P M 3 4 9 4.2.2 PM10-3 and P M 1 0 '. • 5 0 4.2.3 Ultrafine Particles 55 4.3 L A N D U S E A N A L Y S I S . . . . . 61 4.3.1 P M 3 62 4.3.2 PM10.3 and PM,o 64 4.3.3 Ultrafine Particles 67 4.4 SIZE DISTRIBUTION 7 0 4.5 M O D E L COMPARISON .' 71 4.6 S U M M A R Y 76 5.0 C O N C L U S I O N 78 5.1 INTRODUCTION 78 5.2 S U M M A R Y OF FINDINGS 78 5.3 F U T U R E R E S E A R C H - _.: 79 5.3.1 Duration of Particle Measurements 79 5.3.2 Extent and Location o f Sampled Bicycle Routes 79 5.3.3 Comparison to Motorists 80 5.3.4 Particle Composition 82 5.4 APPLICATIONS A N D R E C O M M E N D A T I O N S 82 R E F E R E N C E S • 84 A P P E N D I X : P - T R A K T I M E L A G C A L C U L A T I O N 88 iv LIST OF TABLES Table 1.1: Previous studies addressing cyclists' personal exposure to particulate matter 7 Table 2.1: Dates and times when data were collected 24 Table 3.1: Weather conditions and mean particle concentrations for each day particle measurements were taken. 41 Table 3.2: Pearson correlation coefficients for particle sizes and meteorological variables.. 42 Table 4.1: Land use codes and categories 62 LIST OF FIGURES Figure 1.1: A n idealized size distribution of ambient particulate matter 2 Figure 1.2: Particulate matter size distribution collected in a traffic-dominated area 3 Figure 2.1: Designated bicycle routes in Vancouver, British Columbia 12 Figure 2.2: G R I M M Portable Dust Monitor 13 Figure 2.3: Tri-modal size distribution o f particulate matter collected in a traffic-dominated setting showing G R I M M cutoff...... 14 Figure 2.4: P-Trak Ultrafine Particle Counter 16 Figure 2.5: Tal l buildings along Alberni Street 18 Figure 2.6: Example of multipath error 18 Figure 2.7: Kitsilano air monitoring station 19 Figure 2.8: T E O M s protruding above the station 19 Figure 2.9: The instrumented bicycle 21 Figure 2.10: The G R I M M mounted on bicycle handlebars 22 Figure 2.11: The P-Trak mounted on the rear rack of a bicycle 22 Figure 2.12: Map of data collection route and land use categories 23 Figure 2.13: Study route and annual mean PM2.5 as predicted by a model 32 Figure 3.1: P M 3 concentrations along data collection route 34 Figure 3.2: PM10-3 concentrations along data collection route 34 Figure 3.3: PM10 concentrations along data collection route 35 Figure 3.4: Ultrafine particles counts along data collection route 35 Figure 3.5: Background PM2.5 concentrations measured by the Kitsilano air monitoring station 36 Figure 3.6: Background PM10 concentrations measured by the Kitsilano air monitoring station 36 Figure 3.7: Regression of PM2 .5 concentrations measured at the Kitsilano site and PM3 concentrations measured by the G R I M M when passing the Kitsi lano site 38 Figure 3.8: Regression of P M 1 0 concentrations measured by the Kitsilano site and by the G R I M M when passing the Kitsilano site 38 Figure 4.1: Average P M 3 concentrations along the data collection route 49 Figure 4.2: Average PM10-3 concentrations along the data collection route 51 Figure 4.3: Average PM10 concentrations along the data collection route 51 Figure 4.4: Construction site at the northeast end of Cambie Bridge 53 Figure 4.5: Construction site along West 1 s t Avenue 53 Figure 4.6: Average ultrafine particle concentrations along the data collection route 55 Figure 4.7: Bicycle lane on Burrard Bridge 57 Figure 4.8: Seaside bicycle route 58 Figure 4.9: Chi lco bicycle route 58 Figure 4.10: Bicycle lane along West Georgia Street 59 Figure 4.11: Bicycle lane on Burrard Street 60 Figure 4.12: Bicycle lane on Cambie Bridge 61 Figure 4.13: Average PM3 concentrations for each land use category 63 Figure 4.14: Average PM10-3 concentrations for each land use category 64 Figure 4.15: Average PM10 concentrations for each land use category 65 Figure 4.16: Walkway leading from Pacific Street up to the Cambie Bridge 66 vi Figure 4.17: Average ultrafine particle concentrations for each land use category 68 Figure 4.18: Size distribution of particles in each land use category 70 Figure 4.19: Annual mean PM2.5 concentrations as predicted by a model along the study route. 72 Figure 4.20: Average P M 3 and modeled annual PM2.5 along the data collection route 73 Vll ACKNOWLEDGEMENTS I would first l ike to thank everyone who expressed a genuine interest or even disbelief in my research ("You're getting a master's degree for riding your bike!?"). Knowing that I'm not the only person interested in this topic was truly motivating. M y greatest thanks goes to my academic committee members (and fellow cyclists!) Ian McKendry and Michael Brauer, whose guidance, support, and even contrasting viewpoints made this study so much more than it would have been. I would especially like to thank my supervisor Ian McKendry, whose unwavering enthusiasm (even after I lost part of the G R I M M ) and occasional prodding, which we both know I needed once in awhile, got me through this endeavour. It was an honour and a pleasure to work with you. And thanks for letting me use your laptop and office during your sabbatical! Thank yous also go out to T im Jensen and A l Percival with the G V R D , who provided access to the Kitsi lano T E O M site and T E O M data, as well as shed light on the finicky nature of T E O M s . I would also like to thank Karen Bartlett at the U B C School of Occupational and Environmental Hygiene for lending me the P-Trak and trusting me to haul it around town on the back of my bike. There are also many individuals in the Geography Department for whose help I am grateful: Jason Su for providing the macro to match up G R I M M and G P S data, Jose Aparicio for always having a cheerful answer to my GIS questions and saving my thesis when my computer crashed, Brett Eaton for lending me his laptop with the MapSource software, and Brian Cheng for assisting with the grueling 'all-day ride'. Special thanks go to my family: thank you to my parents for supporting me through another couple years of school and letting me delay venturing into the 'real world ' . I'll v i i i eventually set foot outside the university bubble. A lso, thanks to my brother James, whose words of wisdom helped me get through grad school in two years flat. Special mention goes*to Buster, who has been with me since I was in elementary school, but passed away the weekend before I began grad school and is now in bunny heaven. Finally, thank you to Tom N g for being my roving photographer, but also for his patience, moral support, and sticking with me from my first month o f grad school until the bitter end and beyond, even though I wasn't the 3 r d year chemical engineer he thought I was. ix 1.0 INTRODUCTION 1.1 INTRODUCTION Particulate matter is an air pollutant defined as a mixture of minute solid or liquid particles that are small enough to remain airborne for an extended period of time. It originates from a variety o f sources that include both natural sources such as wind-blown dust or sea spray, and anthropogenic sources such as combustion. A s the behaviour o f particulate matter in an urban environment can be highly variable and complex, this chapter begins by providing a brief summary of the physical characteristics and spatial patterns of particulate matter in such an environment. This summary wi l l be followed by a discussion of ambient and personal exposure levels, particularly cyclists' exposure levels, and wi l l close with the identification o f the primary objective of this study and its implications. 1.2 PARTICULATE M A T T E R 1.2.1 Physical Characteristics For measurement, health, and exposure studies, particulate matter is often classified by size fraction or aerodynamic diameter, rather than actual size. Aerodynamic diameter is based upon the physical behaviour of particles, instead of their actual size. Particles are commonly grouped into three size fractions. First, inhalable particles are particles that are smaller than 1 0 um in aerodynamic diameter (referred to as PMio). Next, fine particles are those less than 2.5 um in aerodynamic diameter (PM2.5). Lastly, ultrafine particles are less than 0.1 urn in diameter (PM0.1). 1 Typically, the size distribution o f particulate matter in the atmosphere follows a bimodal distribution (Figure 1.1). Fine-mode particles are mainly produced by combustion or supersaturated conditions, while coarse-mode particles originate primarily from natural sources such as wind blown dust or dirt. 0.1 1 10 100 Particle Diameter (u.m) Figure 1.1: A n idealized size distribution o f ambient particulate matter. However, in a traffic-dominated area, a tri-modal size distribution occurs (Figure 1.2). First, the Aitken or nuclei mode consists of particles formed by rapid nucleation, commonly from combustion. This mode contains a very large number o f particles, but because of the minute size of each particle, this mode contributes only a small fraction to the total mass concentration (Harrison et al., 1996). Next, the accumulation mode is comprised of particles that are formed as a result of coagulation or condensation of the smaller particles from the Aitken mode. Particles in this mode are long-lived in the 2 atmosphere as removal mechanisms have a poor efficiency in this size region. Thus, these particles can play an important role in long range transport, as well as reducing visibil i ty as they are effective light-scatters (Harrision et al., 1996). Finally, coarse-mode particles are typically mechanically generated, such as road dust which is mobil ized by motor vehicles. These particles are short-lived and are greatly influenced by local conditions (Harrison et al., 1996). ^ 6 m 5 rr, >3. 4 <V « 3 OX) o 5 2 > * 1 Coarse M o d e Accumulation M o d e Nuclei M o d e 0.001 0.01 0.1 1 Particle Diameter (Lim) 10 100 Figure 1.2: Particulate matter size distribution collected in a traffic-dominated area. 1.2.2 Spatial Patterns It is often assumed that air pollutants, such as the smaller size fractions of particulate matter, are homogeneously distributed in large urban areas. In some cities, such as Philadelphia, this assumption is valid (Wilson et al., 2005). Furthermore, according to Harrison and Y i n (2000), in 1977, Friedlander introduced the concept of 'self-preserving aerosol size distribution'. Friedlander argued that over typical 3 atmospheric timescales, particulate matter wi l l tend towards a constant distribution o f size fractions: very small particles wi l l be lost by coagulation and very large particles wi l l be deposited, resulting in a more or less even distribution of intermediate particle sizes. However, this only applies to a time scale o f days. Processes in urban environments often occur over time scales of less than a day, thus one wi l l still expect to observe a variation of particle sizes in urban areas (Harrison & Y i n , 2000). In addition, more recent studies are revealing that there is a greater spatial variation of particulate matter than previously thought, and caution is now advised when using fixed site monitors (FSM) to represent personal exposure without first confirming the homogeneity o f the distribution (Wilson et al., 2005). In urban environments, the ambient concentration and thus the spatial distribution o f particulate matter varies depending on the size fraction, due to underlying mechanisms that operate upon these different size fractions. For example, atmospheric dilution is the main mechanism by which larger particles are lost from the atmosphere. Alternatively, coagulation reduces the number concentration of ultrafine particles. Other loss mechanisms that favour small particles include diffusion to surfaces and evaporation (Zhu et al., 2002). Diffusion rates can also differ, depending on the size fraction. For example, particles with a diameter of 10 nm diffuse approximately 80 times faster than particles with a diameter of 100 nm (Zhu et al., 2002). Coarse particles have an atmospheric lifetime of several minutes to hours, and tend to remain within tens of kilometres from emission sources (U.S. Environmental Protection Agency, 2007). Gravitational settling is the primary mechanism responsible for the spatial heterogeneity of larger particles. Larger particles are transported over 4 shorter distances as they settle out more quickly than smaller particles (Wilson et al., 2005). Therefore, this can result in greater spatial heterogeneity of these larger particles. In agreement with this concept, Monn (2001) found that in an urban area, there was a large spatial variation for the coarse mode of particulate matter ( P M 10-2.5), which he attributed to gravitational settling. The spatial distribution of fine particles (PM2.5) is relatively uniform compared to that of coarser particles (Wilson et al., 2005). This is because fine particles have a longer settling time and can remain in the atmosphere for days or even weeks. A s a result, they are able to travel hundreds to thousands of kilometres from their sources before settling out (U.S. Environmental Protection Agency, 2007). However, some cities, such as those with large emissions from diesel exhaust, can still show spatial variation in PM2.5, especially near emission sources (Monn, 2001). Ultrafine particles (PM0.1) can exhibit a large spatial variation (Monn, 2001). A reason for this could be because PM0.1 is very responsive to local sources. For example, measurements near combustion sources can show concentrations that are orders of magnitudes greater than ambient levels (Harrison & Y i n , 2000). Zhu et al. (2002) also found that the maximum number concentration of PM0.1 next to a freeway was about 30 times greater than background levels. The spatial variation of very small particles could also be explained by coagulation (Harrison & Y i n , 2000; Monn, 2001). Ultrafine particles require time to coagulate. For example, high concentrations of ultrafine particles can be found near their sources because they have recently been emitted. However, as these particles spend more time in the atmosphere, they tend to coagulate into larger particles, resulting in spatial patterns based on how long these particles have 5 been i n the atmosphere. Therefore, rather than be ing lost f rom the atmosphere b y settl ing - out, they are lost because they coagulate into larger particles. D u e to rapid coagulat ion, particles less than 0.1 u m i n diameter have an atmospheric l i fe t ime o f less than 1 hour (Spurny, 2000). V e r y fine particles (less than 0.01 urn) can be lost b y coagulat ion at different rates i n different areas, resul t ing i n spatial var ia t ion. 1.3 AMBIENT vs. PERSONAL EXPOSURE LEVELS A s p rev ious ly discussed, particulate matter o f a l l s ize fractions m a y exhibi t a great spatial var ia t ion i n an urban environment. T h i s has serious impl ica t ions for the use o f F S M s , w h i c h measure ambient background concentrations, to represent personal exposure to particulate matter. F o r example, studies have found that F S M data do not correlate w e l l w i t h personal exposure, nor do they reflect peak exposure levels (Watson et a l . , 1997; A d a m s et a l . , 2001b). In particular, personal exposure levels i n urban areas can be m u c h higher than background concentrations. F o r instance, A d a m s et a l . (2001a) found that personal exposure levels to particulate matter o f ind iv idua l s t ravel ing b y b i cyc l e , bus or car i n L o n d o n , U K was about 100% higher than background concentrations. Furthermore, Gee and Raper (1999) demonstrated that personal exposure levels when t ravel ing b y b i c y c l e or bus were m u c h higher than ambient levels . G u l l i v e r and B r i g g s (2004) also showed that personal exposure to P M i o was higher w h i l e d r i v i n g a car or w a l k i n g , than levels recorded at a F S M . The same study also showed that the F S M under-predicted P M i 0 exposure levels even when it was located ve ry close to a sampl ing route. A reason w h y personal exposure to particulate matter is not consistent w i t h background levels cou ld be because ind iv idua ls m o v e through a variety o f 6 microenvironments everyday, each having different local influences on particle levels (Wilson et al., 2005). For example, mode of transportation, route traveled, and meteorological conditions can all influence personal exposure levels (Gee & Raper, 1999; Adams et al., 2001a; Gull iver & Briggs, 2004). 1.3.1 Cyclists' Personal Exposure Only a few studies have addressed the exposure of cyclists to particulate matter in urban environments. A list of some of these studies is provided in Table 1.1. To date, there have been no studies examining personal exposure of cyclists to particulate matter along bicycle routes. This can come as a surprise, as the amount of air pollution and traffic along a bicycle route is one of the main influences on the likelihood of an individual choosing to cycle (Teschke et al., 2007). Table 1.1: Previous studies addressing cyclists' personal exposure to particulate matter. '-' means that values were not available for these studies. Authors 1.ueal ion Particle Si/e Measured Mean Minimum Maximum Bevan et al., 1991 Southampton, U K Respirable suspended particulates (RSP) 1 3 0 p g m ' G e e & Manchester, P M 4 54 ug/m 3 16.8 ug/m 3 122 ug/m 3 Raper, 1999 U K Adams et London, U K P M 2 5 34.5 n g / n r 3 13.3 Ltg/m3 68.7 ug/m 3 al., 2001b (summer) 23.5 ug/m 3 (winter) (summer) 6.8 u:g/m3 (winter) (summer) 76.2 ug/m3 • (winter) Rank et al., Copenhagen, Total dust 44 ^g /m 3 21 ug/m 3 68 ug/m 3 2001 D K Vinzents et Copenhagen, Ultrafine 32 400 - -al., 2005 D K particles Pt/ml 7 The majority of studies dealing with cyclists' exposure were conducted in Europe and measured a range of particle sizes. However, the study conducted by Bevan et al. (1991) was the only one that focused exclusively on cyclists. Most studies compared cyclists' exposure to that of other modes of transportation such as cars (Adams et al., 2001b; Rank et al., 2001), buses (Gee & Raper, 1999; Adams et al., 2001b), and the subway (Adams et al., 2001b). In most cases, it has been demonstrated that levels of particulate matter in buses or cars are considerably higher than cyclists' exposure levels. For example, Gee and Raper (1999) found that bus riders' exposure to P M 4 was 4 to 6 times higher than cyclists' exposure. Furthermore, Rank et al. (2001) showed that total dust levels inside cars were 2 to 4 times as high as concentrations measured in cyclists' breathing zones. A possible reason for lower exposure is that cyclists often travel beside traffic rather than behind, reducing the amount of direct exposure to vehicle exhaust (Gee & Raper, 1999). Furthermore, in congested traffic conditions, cyclists are able to avoid becoming trapped behind slow-moving or idling vehicles, reducing their time in a congested and polluted environment, thus reducing their exposure to particulate matter (Adams et al., 2001b; Gee & Raper, 1999). Despite lower exposures, cyclists have a greater breathing rate.than drivers, hence more particulate matter is able to enter the respiratory system. A study by van Wijnen et al. (1995) determined that the breathing rate of cyclists-was about 2.3 times higher than that of drivers. When this higher ventilation rate of cyclists was taken into account, their delivered dose approached that of drivers. However, despite an increased breathing rate, Adams et al. (2001b) and Rank et al. (2001) were still able to show that cyclists' 8 delivered dose of pollutants was still lower than that of drivers. Again, this can be explained by the fact that cyclists are able to bypass congested traffic, spending less time being exposed to particulate matter in a congested environment. Some may argue that cars travel faster than cyclists, therefore drivers should receive a lower exposure to particulate matter over the same route. Yet, Rank et al. ( 2 0 0 1 ) showed that during rush hour in Copenhagen, drivers and cyclists travel at a similar speed. Given the advantage that cyclists have in being able to weave through congested traffic, the idea that cyclists spend less time in a congested environment than drivers is still supported. 1.4 OBJECTIVE Despite several studies comparing cyclists' exposure to other modes of transportation, there is still a lack of studies which examine cyclists' exposure more closely. For example, few studies identify spatial patterns in cyclists' exposure, or if cyclists are more likely to encounter higher levels of particulate air pollution in certain areas of an urban environment. As it has been demonstrated that particulate matter can exhibit a large spatial variation in urban environments, cyclists' exposure will also undoubtedly be varied over space. The primary objective of this study is to determine where cyclists are exposed to the highest levels of particulate matter along bicycle routes. Given to broad scope of this study, the outcome can have implications for several fields of research. First, this study applies to the field of geography and atmospheric science, as it can lead to a greater understanding of the behaviour of particulate matter in urban areas. Next, it is also 9 relevant to fields dealing with health, as it w i l l identify where cyclists may have to take extra precautions due to higher levels of particulate air pollution. Lastly, this study might be of use for urban planning, as it could assist in the planning of future bicycle routes to reduce exposures of cyclists to particulate air pollution. The following chapter w i l l outline the methodology o f this study, including a description of the study location, instruments used, and how data was collected and processed. Chapter 3 w i l l focus on between-day variation of particle concentrations and the influence o f instrument calibration, emissions, and meteorological conditions, while Chapter 4 w i l l discuss spatial variation of particle concentrations along the route as well as in different land use categories encountered along the route and particle size distribution in these land uses. Chapter 4 w i l l also include a qualitative comparison of measured fine particle concentrations to those o f a traffic-based model. Finally, Chapter 5 w i l l provide a summary of findings, discuss areas requiring further research, and make recommendations based on the results o f this study. 10 2.0 M E T H O D O L O G Y 2.1 INTRODUCTION This chapter outlines the methodology followed in this study. It w i l l begin with a description of the study location, instruments used to collect data, and data collection route. It wi l l then be followed by an explanation of how the data was processed. 2.2 STUDY LOCATION This study was carried out in the city of Vancouver, in the province of Brit ish Columbia, Canada. Vancouver is home to an avid cycling community, as seen by the numerous cycling events occurring throughout the year, such as B ike Month in June and the monthly Critical Mass rides. Vancouver also boasts an extensive network of bicycle routes, as seen in Figure 2.1. These routes are primarily residential side streets with modifications for cyclists, such as traffic signal push buttons closer to the road, • pavements markings, and directional signs. Other bicycle routes are in the form of designated cycling lanes on major roads, or off-road trails. 11 Designated Bicycle Routes Figure 2.1: Designated bicycle routes in Vancouver, Bri t ish Columbia. (Data source: Translink, D M T I Spatial, GVPvD.) Furthermore, these b i c y c l e routes pass through a variety o f urban environments. These environments range from quiet residential neighbourhoods and recreational trails, to the down town core, to t ruck-dominated industr ial areas. T h e variety o f settings through w h i c h V a n c o u v e r ' s b i c y c l e routes pass a l lows for particulate matter measurements to be taken along a range o f routes w i t h va ry ing condit ions, m a k i n g spatial patterns more apparent and a l l o w i n g for the compar i son o f concentrations measured at different locat ions. 12 2.3 INSTRUMENTATION 2.3.1 G R I M M Portable Dust Monitor T h e G R I M M dust moni tor (Series 1.108) (Figure 2.2) was used to measure particulate matter ranging from 0.3 to 20 u m i n diameter, at a f l ow rate o f 1.2 L / m i n . T h e basic p r inc ip le b y w h i c h the G R I M M instrument operates is that particles entering the device scatter l ight from a laser diode. T h e amount o f l ight scattered depends on the number and the size o f the particles. Therefore, this instrument is able to measure the number concentration as w e l l as the s ize dis tr ibut ion o f the particles, un l ike gravimetr ic devices w h i c h on ly measure mass. H o w e v e r , the G R I M M is also able to calculate mass b y assuming that particles are spherical and particle density is constant. Other advantages o f the G R I M M inc lude its portabi l i ty and its ab i l i ty to capture data at a h igh resolut ion (6 seconds). Figure 2.2: G R I M M Portable Dust Monitor. However , a disadvantage o f us ing the G R I M M is that the smallest particles it can detect are those 0.3 u m i n diameter. T h i s means that it misses ultrafine particles (those less than 0.1 u m i n diameter). Several studies have suggested that ultrafine particles m a y 13 pose a greater risk to human health than larger particles clue to these small particles' relatively large surface area (Harrison & Y i n , 2000; Zhu et al., 2002). Yet, the G R I M M still captures the right hand limb of the accumulation mode, which is where ultrafine particles could end up due to coagulation or condensation processes (Figure 2.3). £ a Q 0£ > 6 -5 -4 -3 -2 -1 -0 Coarse Mode Nuclei Mode 0.001 0.01 0.1 1 10 Particle Diameter (pm) 100 Figure 2.3: Tri-modal size distribution of particulate matter collected in a traffic-dominated setting showing GRIMM cutoff. Dotted line shows minimum particle size measured by the GRIMM. Another disadvantage o f the G R I M M is that in order to calculate mass, the instrument makes the assumption of spherical particles with a constant density. However, particles are often not spherical, nor do they always have a constant density. Therefore, one must be aware of these assumptions when analyzing the G R I M M ' s measurements. For this study, the G R I M M measured particles by mass concentration (^tg/m3). The G R I M M ' s output divided the measurements into 15 different size classes, from which metrics of PMio (particles less than 10 um in diameter), P M 3 (particles less than 3 14 um) and PM10-3 (particles between 10 and 3 urn) were calculated. P M 3 rather than the more widely used PM2.5 was used because the G R I M M does not have a 2.5 urn cut off. The G R I M M can be equipped with several different-sized nozzles to account for different wind speeds, ranging from 0.5 to 4 m/s. In this study, the wind is mainly a result of the turbulence caused by the G R I M M being stationed on a moving bicycle (section 2.3.5). However, wind tunnel tests conducted by Maletto et al. (2003) demonstrated that particle concentrations measured by the G R I M M was not dependent on nozzle size. Therefore, the 4 m/s nozzle was used for the majority of this study. 2.3.2 P-Trak Ultrafine Particle Counter The P-Trak particle counter (Model 8525) (Figure 2.4) measures particles between 0.02 and 1 um in diameter and records the number of particles per cubic centimeter. It has a sample flow rate o f 100 cm 3 /min and a total flow rate of 700 cm 3 /min. It operates in a similar manner as the G R I M M , using light scattering to count particles. However, with the P-Trak, the particles are mixed with an alcohol vapour prior to being measured. They then pass through a condenser tube where the alcohol . condenses on the particles, enlarging the particles. A s a result, when the particles reach the laser beam, they are large enough to scatter light and be detected. If the particles had not been enlarged by the addition of alcohol, it would not be possible for them to scatter enough light to be detected. 15 Figure 2.4: P-Trak Ultrafine Particle Counter One disadvantage o f the P -Trak is that it temporar i ly stops funct ioning i f it is not held i n a level pos i t ion . In other words , whenever the instrument is t i l ted, it w i l l stop recording data w h i c h w i l l result i n gaps i n the data. A s the route a long w h i c h particulate matter was measured inc luded several h i l l s that were steep enough to s ignif icant ly tilt the instrument, there are several gaps i n the data sets. Fortunately, these h i l l s were not ve ry long , so these gaps are re la t ive ly smal l compared to the overa l l data set. Ano the r potential p rob lem w i t h the P-Trak was that the intake nozz le was attached to a 127 c m tube leading into the instrument, w h i c h cou ld result i n a t ime lag between when a part icular sample o f air enters the nozz le and when it actually reaches the instrument 's measurement chamber. Howeve r , calculations (see A p p e n d i x ) showed that the t ime lag was negl ig ib le (less than 1 second). Thus , this study w i l l assume that the measurement o f a sample is instantaneous. There are fewer P -Trak data sets than G R I M M data sets because the P -Trak was not a lways avai lable on days when particle measurements were taken. 16 2.3.3 Global Positioning System (GPS) A GPS device (Garmin GPS 76) was used during data collection in order to assign a location to each particle measurement. Coordinates were recorded at a 10 second resolution. The G P S clock was synchronized with the P-Trak clock, but as the seconds on the G R I M M clock could not be changed, the time difference between the G P S and G R I M M clocks were noted and times were adjusted afterwards. With any GPS device, the clock wi l l constantly reset to ensure it is synchronized with the satellites. A s it was not possible to monitor the GPS during data collection to note when the clock reset, it wi l l be assumed that the G P S clock and clocks on the instruments are relatively in sync. This is a valid assumption because in this study, the measurements wi l l be averaged over each land use zone, which is discussed in detail in section 2.5. Thus, the exact location o f each specific measurement is not essential. Another problem encountered with the GPS device is due to multipath error. In urban areas, satellite signals can bounce off structures before it reaches the receiver, resulting in erroneous measurements. During this study, multipath error was apparent in the downtown core, mainly along Alberni Street and Burrard Street, where many tall and reflective structures, such as off ice buildings, are located. Figure 2.5 shows the route passing between tall buildings along Alberni Street. As a result, G P S coordinates were scattered when traveling through this area, as seen in Figure 2.6. Therefore, when graphing measurements by distance, any coordinates that deviated more than 30 metres from the route were removed. (A detailed description of this process can be found in section 2.5.1.) This reduced the influence of these stray points on the total distance of the route. 17 Figure 2.5: T a l l buildings along Alberni Street. a.) b) « * « » « » - fill ll III II ii . «d \ \ / Figure 2.6: Example o f multipath error. Different coloured points represent different data sets, a) Precise G P S readings in a recreational/residential area, b) Imprecise G P S readings, due to multipath error, in the downtown core. (Data source for road network: D M T I Spatial.) 18 2.3.4 Tapered Element Oscillating Microbalance (TEOM) Tapered element osc i l l a t ing microbalances ( T E O M s ) are c o m m o n l y used to measure ambient particulate matter i n B r i t i s h C o l u m b i a . The air mon i to r ing station w h i c h measures both P M ] 0 and P M 2 5 and is closest to the study route is the K i t s i l a n o station (T002) (Figure 2.7 and 2.8). It is located i n a residential neighbourhood, on the grounds o f K i t s i l a n o Secondary Schoo l . Backg round particulate matter measurements from the K i t s i l a n o station were used as a baseline to w h i c h experimental data were compared and normal ized (section 2.5.4). Figure 2.7: Kitsi lano air monitoring station. Figure 2.8: T E O M s protruding above the station. W i t h T E O M s , air is first drawn through an inlet, and then heated to remove water from the particles. T h e particles are then deposited onto a filter w h i c h rests on top o f a tapered glass element. T h i s glass element is forced to oscil late, l i ke a tuning fork, and the frequency o f osc i l la t ion depends on the mass o f the particles on the filter: as the mass 19 increases, the frequency of oscillation wi l l decrease. Therefore, particle mass can be calculated from the oscillation frequency. One weakness of the T E O M is its sensitivity to noise. Vibrations, such as those caused by walking into the building which houses the T E O M s , can affect the oscillation frequency (A l Percival, personal communication, 2007). Furthermore, temperature' changes inside the building, such as whether or not the air conditioning is turned on, can also affect the oscillation. Another potential source of error in T E O M measurements is caused by the heating stage: semi-volatile material can be lost when a sample is heated to remove water from the particles. The T E O M which measures P M 2 . 5 at the Kitsi lano site is equipped with a dryer to eliminate moisture. However, this dryer heats samples to only 30°C, instead of the more common 40°C, to reduce the loss of volatile material (A l Percival, personal communication, 2007). The T E O M s at the Kitsilano station collected data at 1 minute resolution. However, due to the instruments' sensitivity, this data is not stable and is only used for quality assurance and quality control, then discarded. Thus, for this study, hourly T E O M data were used, which has an accuracy of 2 to 2.5 ug (A l Percival, personal communication, 2007). Lastly, the T E O M data were recorded in Pacific Standard Time (PST), and thus had to be converted to Daylight Savings Time (DST) for the majority of the days when particle measurements were taken. 20 2.3.5 The Instrumented Bicycle Part ic le measurements were taken by r i d ing an ' instrumented b i c y c l e ' (F igure 2.9) a long a f ixed route. T h i s b i c y c l e was equipped w i t h the G R I M M , P-Trak , G P S , as w e l l as a digi ta l vo ice recorder to record quali tat ive observations, such as i f the cycl is t was wa i t ing beh ind a motor veh ic le at an intersection, passing a l e a f b lower , or was i nvo lved i n any other situation that cou ld cause unusual ly h igh levels o f particulate matter. Figure 2.9: The author with the instrumented bicycle used for data collection. The G R I M M was mounted on the b i cyc l e ' s handlebars, w h i l e the P-Trak was placed on the rear rack (Figures 2.10 and 2.11). The intakes o f both instruments were placed wi th in the cyc l i s t ' s breathing zone (wi th in 30 c m o f the cyc l i s t ' s mouth) . E v e n though the P-Trak had a l ong tube leading from the intake to the instrument, the t ime lag between when a sample enters the intake and when it reaches the instrument is negl ig ib le (as discussed i n section 2.3.2). 21 Figure 2.10: The G R I M M mounted on bicycle Figure 2.11: The P-Trak mounted on the rear rack handlebars. of a bicycle. 2.4 DATA COLLECTION 2.4.1 Data Collection Route The route a long w h i c h particulate measurements were taken (Figure 2.12) is p r imar i l y on designated b i c y c l e routes, except for areas where there is no connect ion between designated routes. T h e route was chosen to pass through as many different urban environments as possible , such as those discussed i n section 2.2. 22 Study Route Major Roads Road Residential- Single Family and Duplexes Residential-Towihouse and low-riseApartments Residential- High-rise Apartments Commercial - Residential/Mbted | Commercial j Institutional j Industrial | Transportation, Communication and Utilities Recreation and Protected NaturalAreas Open and Undeveloped | Lakes and Water Bodies Figure 2.12: M a p o f data collection route and land use categories, Vancouver, Brit ish Columbia. (Data source: G V R D , D M T I Spatial.) T h e route was r idden dur ing m o r n i n g rush hour on random weekdays between 7:30 and 9:00 am and covered a distance o f approximate ly 20 k m . It was r idden at about 15 k m / h and at a relat ively constant speed. T h i s resulted i n approximately 4 G R I M M measurements and 3 P-Trak or G P S measurements per 100 m . Data were col lected i n Augus t through the end o f October 2006 to capture a variety o f meteorological condi t ions , such as wa rm and dry h igh pressure condit ions w h i c h are typ ica l ly seen i n the summer, and cool and wet l o w pressure condit ions c o m m o n i n the fall and winter. A s a result, mean concentrations w o u l d be representative o f overa l l weather patterns i n Vancouve r , and not just one part icular type. 23 Over this 3-month span the route was ridden 16 times during morning rush hour. However, due to instrument malfunctioning, data from 2 rides were discarded, resulting in a total of 14 data sets. Table 2.1 lists the dates and times when data were collected, as well as the total distance traveled on each day. The total distance was not available for October 12 due to a calculation error. Table 2.1: Dates and times when data were collected, and total distances traveled. All data were collected in 2006. The total distance was not available for October 12 due to a calculation error. Date Start Time (a.m.) Knd Time (a.m.) Total Distance (km) August 11 • 7:39 . 9:06 20.2 August 17 7:35 9:07 20.8 August 18 7:40 9:07 20.6 September 7 7:34 9:01 20.3 September 8 7:33 9:08 20.7 September 11 7:39 9:07 20.6 September 21 7:29 8:54 20.3 September 29 7:29 8:54 20.4 October 2 7:35 9:00 20.7 October 12 7:31 9:01 -October 13 7:47 9:10 19.1 October 20 7:51 9:27 21.1 October 23 7:44 9:38 21.2 October 30 7:32 8:58 20.5 2.4.2 Minimizing the Influence of Temporal Patterns As a major objective of this study was focused on spatial patterns o f particulate matter, steps were taken to ensure any patterns observed were due to spatial factors, not temporal factors. First, the study route was ridden in both directions to ensure patterns were visible regardless of at which end of the route data collection began. This would rule out the question of i f patterns were present because of time elapsing throughout data collection. Next, on one day, the route was ridden multiple times to determine i f the spatial patterns existed throughout the day, and were not simply an artifact o f the morning 24 hours. In both cases, spatial patterns were comparable, regardless of direction or time of day. Therefore, it can be assumed that patterns are primarily spatial in nature and not a consequence of the changing concentrations over the duration of the ride. To minimize the day-to-day variation, data were also normalized based on the T E O M measurements, as discussed further in section 2.5.4. 2.5 DATA PROCESSING 2.5.1 G R I M M Data The G R I M M output was downloaded from the instrument using G R I M M Software 1.177, version 3.20. It was then exported to Microsoft Office Excel 2003 for data processing. As mentioned in section 2.3.3, the G R I M M clock could not be synchronized with the GPS clock prior to data collection. Thus, synchronization of the clocks had to be carried out afterwards when the data were exported into Excel . Next, each G R I M M measurement was assigned a G P S coordinate. First, the G P S coordinates were converted from decimal minutes to decimal degrees to facilitate future mapping. A s the G R I M M data were collected at a 6 second time resolution while the G P S data were at a 10 second resolution, it was not possible to simply combine the data sets. Using a code, each G R I M M measurement was assigned the GPS coordinate that was closest to it in time. This resulted in some G R I M M measurements having identical G P S coordinates, but as mentioned in section 2.3.3, the exact location of each measurement is not essential, and a general location is adequate for the purpose of this study. 25 Once the G R I M M measurements were matched with a G P S coordinate, data were plotted in a geographic information system (GIS). For this study, ESRI ' s ArcGIS (version 9.0) was used. The G P S had recorded data using W G S 84 (World Geodetic System) as the coordinate system, while existing GIS data used U T M (Universal Transverse Mercator). Therefore, in order to make the data sets compatible, ArcCatalog was used to specify that the G P S data originally used W G S 84, but wi l l use U T M when being inputted into ArcMap. Maps of G R I M M measurements were then created. The next step was to graph particle concentrations along the length of the route for each day, to visualize any peaks or troughs in particle concentrations. However, some data points strayed drastically from the data collection route, due to multipath error (section 2.3.3 and Figure 2.6). This could affect the total distance of the route and shift other points away from their actual location. To reduce the effect of this problem, any points more than 30 m away from the study route were removed. A distance of 30 m was chosen because it was large enough to still include small deviations from the collection route due to day-to-day variation, but small enough to exclude points that were extremely scattered. To eliminate the stray points, the route along which particles were measured was first digitized in ArcMap. Then, using the Select by Location function, points within 30 m of the route were selected. The selection was then switched, so that points further than 30 m away from the route were selected. These points were then noted down, and deleted from the data sets in Excel . However, these points were only removed for graphing purposes. A l l points were used for statistical calculations. 26 Using these modified data sets, the distance between neighbouring data points was calculated using the spherical law of cosines (Moveable Type, online): d = acos [sin(lati).sin(lat2).+ cos(lati).cos(lat 2).cos(long 2-longi)].R Where: d - distance between two points lati = latitude of start point, in decimal degrees lat 2 = latitude of end point, in decimal degrees long] = longitude of start point, in decimal degrees long 2 = longitude of end point, in decimal degrees R = radius of the Earth, in metres (6 371 000 m) The distances between each point in a data set were then summed to calculate a cumulative distance for each point. In other words, it was possible to identify where a particular data point lay on the route, with respect to the start point. This made it possible to plot average concentrations of P M | 0 , P M ] 0 - 3 , and PM3 along the length o f the route to show i f any peaks or valleys in particulate concentration occurred at the same location on different days. Despite a common start and end point for the study route, there were some discrepancies between data sets with respect to the total distance traveled, as seen in Table 2.1. This could be due to minor deviations from the route traveled during data collection, such as when the cyclist had to use the sidewalk due to obstructions on the road, or using a crosswalk to make a left turn at a busy intersection instead of turning left with the traffic. Differences in total distance could also be attributed to multipath error, as discussed in section 2.3.3, because points that were less than 30 m away from the data collection route were not discarded. In addition to graphing particle concentrations along the route for each day, the average concentration for all days was calculated. Measurements had not been taken at 27 the exact same locations for each ride, therefore it was not possible to simply calculate an average for specific locations along the study route. Instead, average concentrations were calculated for each 250 m leg of the route. This yielded over 80 averages, which was enough for plotting average particulate concentrations along the length o f the route for PMio, PM10-3, and P M 3 . 2.5.2 P-Trak Data P-Trak data were downloaded from the instrument using TSI TrakPro, version 3.41, then exported into Excel . As the P-Trak often temporarily stopped functioning throughout data collection when it was tilted (section 2.3.2), there were gaps in the data sets. Therefore, the data first had to be cleaned up by manually deleting these gaps. Next, similar to the handling of the G R I M M data, G P S coordinates were assigned to each P-Trak measurement. Both of the GPS and the P-Trak data had a time resolution of 10 seconds, therefore the times for both data sets were simply lined up when the data sets were combined. If the GPS clock had reset itself and the time was no longer identical to that o f the P-Trak clock, the P-Trak data were assigned the GPS coordinate that was closest to it in time. P-Trak data were entered into a GIS then mapped and graphed using the same method as outlined in section 2.5.1. 2.5.3 Land Use Analysis To investigate whether or not cyclists are exposed to higher concentrations of particulate matter in certain land uses, particle measurements were overlaid on land use 28 data in ArcMap. The land use data were provided by the Greater Vancouver Regional District ( G V R D ) and is current as of 2001. Whi le collecting data, it was observed that some areas which had been classified as 'open and undeveloped' in the land use data set were now developed into residential areas. Therefore, the land use data set was updated accordingly. Some of the data points were shifted slightly away from their correct land use, possibly due to day-to-day variation in atmospheric conditions or multipath error, both of which can affect the functioning of a G P S . Therefore, these shifted data points were manually placed in their correct land use by using the Edit function in ArcMap. After ensuring that all data points were in their correct land use category, A rcMap was used to spatially join the points to the land use data. In other words, the particle measurements were assigned a land use code, based on the land use category in which they were located. Bridges were not classified in the land use data set. Therefore, measurements taken along Burrard Bridge and Cambie Bridge were manually selected and assigned the code ' B ' to symbolize that they occurred on a bridge. Next, all data sets were exported to Excel , compiled into one file, and sorted by land use code. Average concentrations of particulate matter were then calculated for each land use code. The size distribution of particles in each land use was also examined. As mentioned in section 2.3.1, the G R I M M ' s output was divided into 15 different size classes, ranging from 0.3 urn to greater than 20 um. From these 15 size classes, it was possible to plot the size distribution of the particles measured in each land use to show which size class, i f any, had the most variable concentration when compared by land use. 29 2.5.4 Normalization Data were collected on days with varying temperatures, wind speeds, and other meteorological conditions that could affect particle concentrations. Therefore, in addition to examining the raw data, the data were also normalized to remove day-to-day variation, which is a common procedure in this field of research (i.e. Larson et al., 2007). Normalization was based on background levels recorded by the Kitsilano air monitoring station. As the T E O M s at the Kitsi lano site did not measure ultrafine particles, there was no reference instrument for this size fraction. Therefore, normalization was only carried out for the G R I M M data and not the P-Trak data. It is also important to note that the G R I M M measured P M 3 , while the T E O M measured PM2.5, which could explain some of the differences between measurements taken by the two instruments. However, for the purpose of normalization, P M 3 and PM2.5 wi l l be treated as identical metrics. Using the T E O M data from the Kitsilano site, a correction factor was calculated for PM2.5, PM10-2.5, and PM10 for each day, and can be defined as: PM , Xu = ^7 Where: X,j = correction factor for day i and particulate matter size j PM j - mean T E O M concentration for all days for particulate matter s i z e / PMy = mean T E O M concentration for day i and particulate matter size j This correction factor was then applied to the corresponding G R I M M data sets. As a result, this process removes day-to-day variation caused by different meteorological conditions. Any variation in particulate concentration that remains can be attributed to local sources such as emissions. 30 2.5.5 Model Comparison P M 3 data were compared to a traffic-based land use regression model o f annual mean PM2.5 (Henderson et al., 2007). The model was developed by collecting PM2.5 measurements at 25 locations over one week periods. At each location land use characteristics were also measured for circular buffers of different radii in a GIS. These land use characteristics were regressed on the measured concentrations. Other variables involved in the creation of this model included road length and vehicle density. Though the model estimated annual P M 2 5 and the measurements in this study were for P M 3 and only had a 3-month span, this model was the most compatible and readily available. Due to the differences in metrics, a direct comparison between modeled and measured particle concentrations was not possible. Therefore a more qualitative approach was used. Model comparison was conducted in a GIS by first overlaying the study route on the model (Figure 2.13). Next, the Convert Paths to Points feature in the Hawth's Tools extension was used to convert the route to a series of points at 50 m intervals, which is similar to the resolution of the G R I M M data. However, the model was provided in a raster format, while the particle measurements in this study were in vector format. Therefore, ArcToolbox's Spatial Analyst extension was used to analyze these data sets. The Extract Values to Points function joined the value of a raster cell to the corresponding point along the data collection route (which had been converted to points). A s a result, the particle concentrations of the model appeared in the attribute table of the study route. These values could then be exported to Excel and plotted alongside measured particle concentrations for visual comparison. 31 Figure 2.13: Study route and annual mean P M 2 5 as predicted by a model, Vancouver, Bri t ish Columbia. (Data source: Michae l Brauer, Sarah Henderson, D M T I Spatial.) 32 3.0 P A R T I C L E C O N C E N T R A T I O N S A L O N G B I C Y C L E R O U T E S : D A Y - T O - D A Y V A R I A T I O N 3.1 INTRODUCTION This chapter describes the analysis o f the variation of particle concentrations between the 14 days of measurements. Day-to-day variation can be attributed to several factors, such as instrument calibration, emissions, and meteorological conditions. Specifically, precipitation, air temperature and wind speed wi l l be considered. Discussion of spatial patterns and individual peaks and troughs in particulate concentrations measured along the route wi l l be discussed in Chapter 4. 3.2 DAY-TO-DAY VARIATION Figures 3.1 to 3.4 show particle concentrations measured over the length of the route during the 14 days of observations. The left hand side of each graph corresponds to the western end of the route shown in Figure 2.12. Peaks at specific locations along the route wi l l be discussed in section 4.2. October 12 was omitted from all graphs due to faulty distance calculations. Yet, from these figures, one can still see that all particle sizes exhibit some degree of variation between days. 33 s 2 0 0 . 180. 160. 140. 120. 100. 80. 60. 40 . 20 . 0. 11-Aug 17-Aug 18-Aug 7-Sep — 8-Sep 11-Sep 21-Sep 29-Sep 2-Oct 13-Oct 20-Oct 23-Oct 30-Oct 5 0 0 0 1 0 0 0 0 1 5 0 0 0 Distance (m) 2 0 0 0 0 Figure 3.1: PM3 concentrations along data collection route. Left hand side corresponds to the western end. 9 0 0 . 0 I B i 1 o 0.0 -(\ fl u 1 1 1 v m \m I E «J 1 T 5 0 0 0 1 0 0 0 0 1 5 0 0 0 Distance (m) 20000 — 11-Aug 17- Aug 18- Aug 7-Sep —8-Sep 11-Sep 21-Sep — 29-Sep - 2-Oct 13-Oct - 20-Oct — 23-Oct 30-Oct Figure 3.2: PM10-3 concentrations along data collection route. Left hand side corresponds to the western end. 34 1000.0 900.0 500.0 100.0 0.0 -1* 1 I 1 1 il \ n U i i i 11-Aug — 17-Aug 18-Aug 7-Sep — 8-Sep 11-Sep — 21-Sep — 29-Sep — 2-Oct 13-Ocl 20-Oct — 23-Oct 30-Oct 5000 10000 15000 Distance (m) 20000 Figure 3 .3 : P M | 0 concentrations along data collection route. Left hand side corresponds to the western end. -CU a> a c BS 350000 300000 250000 200000 150000 100000 50000 (» - l l - A u g 17- Aug! 18- Aug 8-Sep II-Sep • 2 -Oc t - 2 0 - O c l whim . y 5000 10000 Distance (in) 15000 20000 Figure 3 .4 : Ultrafine particles counts along data collection route. Left hand side corresponds to the western end. Figures 3.5 and 3.6 show P M 2 . 5 and P M 1 0 , respectively, as recorded by the T E O M s at the K i t s i l a n o air mon i to r ing site dur ing the days and times when data were col lected a long the route. T h e P M J O data from Augus t 18 was omit ted because it was 35 incomplete . T h e T E O M s also show that day-to-day var ia t ion exists w i t h background levels , con f i rming that this var ia t ion is not mere ly an artifact o f the G R I M M or P-Trak instruments. Figure 3.5: Background PM 2 5 concent ra t ions measured by the Kits i lano air monitoring station. 7:05:00 7:55:00 Time 8:45:00 11 -Aug 17-Aug 07- Sep 08- Sep 11- Sep 21-Sep 29- Sep 02-Oct 12- Oct 13- Oct 20-Oct 23-Oct 30- Oct Figure 3.6: Background P M , o concentrations measured by the Kits i lano air monitoring station. 36 In this study, variation between days can be explained by three main factors: instrument calibration, emissions, and meteorological conditions. These wi l l be discussed in the following sections. 3.3 INSTRUMENT CALIBRATION One source of between-day variation is instrument calibration. Variation could be simply a result of instrument functioning, rather than actual variations in particle concentration. Here, measurements recorded by the G R I M M are compared to those by the T E O M s , which are standard reference instruments. This was not a direct comparison, therefore one would not expect these instruments to record identical measurements. The main goal of this comparison was to show the variation in concentrations measured by the two instruments. Figures 3.7 and 3.8 show regression plots between mean particle concentrations measured by the G R I M M when cycling along West 10 t h Avenue between Trafalgar and Larch Streets, near the Kitsi lano air monitoring station, and the measurements by the T E O M s at the Kitsi lano site corresponding to the same period of time. Data from October 23 were excluded because the cyclist did not pass the Kitsi lano site that day. It was not possible.to compare ultrafine particles because there is no reference instrument for this size fraction at the Kitsi lano site. 37 Figure 3.7: Regression o f P M 2 . 5 concentrations measured at the Kits i lano site and P M 3 concentrations measured by the G R I M M when passing the Kits i lano site, r = 0.44. • 0 5 1 0 15 2 0 2 5 3 0 3 5 PM10 (TEOM) (Hg/m3) Figure 3.8: Regression o f P M 1 0 concentrations measured by the Kits i lano site and by the G R I M M when passing the Kitsi lano site, r = 0.20. 3 8 The correlation coefficient between PM2.5 and PM3 measurements is 0.44, while the correlation coefficient between PM10 measured by the two instruments is 0.20. These low correlation coefficients suggest that the G R I M M and T E O M instruments do not record identical measurements, nor are their differences in concentration consistent, making it impossible to apply a single correction factor to the data. A comparison of these two instruments was also carried out by Maletto et al. (2003), showing that the G R I M M and T E O M do not differ systematically. PM2.5 and PM3 measurements from the two instruments may be more strongly correlated than PM10 because PM2.5 is typically more uniformly distributed over space than PM10, which wi l l be discussed in detail in sections 4.2.1 and 4.2.2. As a result, the location of a sampling instrument might not influence PM2.5 concentrations as much as PM10 concentrations. The G R I M M instrument typically measured concentrations greater than those measured by the T E O M . This is most l ikely because G R I M M measurements were taken on a road while the T E O M recorded roadside measurements. In addition, these instruments function on very different principles and make different assumptions, as described in sections 2.3.1 and 2.3.4. Yet, these instruments are still positively correlated. In other words, when one records a high concentration, the other does as well . However, as instrument variation cannot be controlled and the G R I M M had been recently calibrated, it wi l l be assumed that instrument variation is negligible in this study. 39 3.4 EMISSIONS Emissions can also contribute to day-to-day variation of particle concentrations. Emissions encountered in this study are primarily from traffic, as measurements were taken along roads used by motor vehicles. Traffic emissions can vary depending on a variety of factors, such as time of day or day of the week. For instance, Woo et al. (2001) found that ultrafine particle concentrations were elevated during rush hour, and concentrations were also higher on weekdays than on weekends. A study by Levy et al. (2001) also showed that PM2.5 concentrations are higher during morning rush hour and on weekdays. Land use can also affect emissions, as different land uses are associated with different sources of emissions. For example, Sun et al. (2004) measured P M | 0 at an industrial site, traffic site, and residential site. They found that there was a significant difference in PM10 levels between these sites, with concentrations being highest at the residential site during the winter, and the industrial site during the summer. Furthermore, Chan et al. (2001) measured PM10 and PM2.5 at sites classified as urban-residential, urban-commercial, urban-industrial, and new town and found that concentrations between sites were highly variable. However, during this study data were only collected on weekdays and at approximately the same time each day, as described in section 2.4.1 and Table 2.1. Data were also collected on random weekdays to reduce day of the week effects. A lso , data were collected along a fixed route, passing the same sequence of land uses each time. Therefore, it wi l l be assumed that emissions are relatively constant throughout the study and do not contribute significantly to between-day variation of particulate concentrations. 40 3.5 METEOROLOGY Table 3.1 lists the average precipitation, temperature, and wind speeds for each day when particulate measurements were taken, as provided by Environment Canada (Environment Canada, 2007), and the corresponding mean particle concentrations measured on each day. As data collection did not occur when it was raining (due to instruments being unable to function in wet conditions), precipitation was defined as the total amount of precipitation which fell over the previous two days. Average temperatures and wind speeds were calculated from hourly data between 7:00 and 9:00 am D S T , except for October 30 where data were between 7:00 and 9:00 am PST. Table 3.1: Weather conditions and mean particle concentrations for each day particle measurements were taken. Precipitation refers to total precipitation over the previous two days, while average temperature and wind speed were calculated between 7:00 and 9:00 am D S T unless otherwise stated. ' - ' means data were not available for that day. Date Precipitation Average Average Mean Mean Mean Mean (mm) IVmperaluir Wind PM, I'M m-l I'M in ritrafiiie ("O Speed (km.h) (fig/in1) (fig/in') (fig/m') Particle Count (Pl/cc) August 11 2.8 11 » 17.7 32.1 27.4 59.5 21547 August 17 0 12 16.3 31.0 30.0 61.0 18830 August 18 0 15.7 5 23.2 19.5 42.7 22379 September 7 0 13.5 10.3 33.6 41.3 74.8 -September 8 0 12.2 16.3 19.0 32.1 51.1 25535 September 11 4.6 11 4.7 18.1 39.0 57.1 45293 September 21 18.6 12.1 27.7 10.6 11.0 21.6 September 29 0 12.9 6.3 25.2 45.3 70.5 -October 2 0.2 8.8 1.3 18.4 39.4 57.8 57692 October 12 0 8.8 2.7 20.4 31.3 51.7 -October 13 0 9.1 7.7 19.9 27.2 47.1 -October 20 8 10.7 2 11.9 23.2 • 35.1 46168 October 23 0 9.2 5.7 23.9 29.7 53.6 -October 30 1.6 -0.1 9 7.3 18.4 25.7 -(7:00-9:00 am, PST) A s most of the measurements were taken during the dry summer months, 8 days recorded no precipitation over the previous two days. The maximum amount of 41 precipitation over the two days prior to a day when particle measurements were taken was 18.6 mm. Particle measurements were taken during the morning, when temperatures did not yet reach their daily highs. A s a result, average temperatures ranged from -0.1 to 15.7°C. Average wind speeds were more variable, ranging from 1.3 to 27.7 km/h. These three meteorological variables were chosen because they have been shown to influence particulate matter concentrations. For example, precipitation can remove particulate matter from the air by rainout or washout. It can also reduce the amount of road dust that becomes airborne. Next, Adams et al. (2001a) found that air temperature had an effect on PM2.5 concentrations when the data collection route was not fixed, and Kittelson et al. (2002) found that temperature and ultrafine particle concentrations were negatively correlated. Lastly, Rank et al. (2001), Adams et al. (2001a), and Zhu et al. . (2002) demonstrated that wind speed influenced concentrations of PM10, PM2.5, and PM0.1, respectively, due to the ability of the wind to dilute and transport particles. Table 3.2 shows the Pearson correlation coefficients and significance for each particle size and meteorological variable. Values in bold indicate significance atp = 0.05 level. Table 3.2: Pearson correlation coefficients for particle sizes and meteorological variables. Values in bold indicate significance atp = 0.05 level. n Precipitation Air Temperature Wind Speed P M 3 14 -0.51 0.57 0.06 P M 1 0 - 3 14 -0.58 0.22 -0.44 P M 1 0 14 -0.63 0.43 -0.24 Ultrafines 7 0.42 -0.76 -0.81 In the following sections, the influence o f precipitation, temperature, and wind speed wi l l be discussed in detail with respect to each particle size. 42 3.5.1 Precipitation Correlations between precipitation and particle concentrations were statistically significant for PM10-3 and P M i 0 . These particles were negatively correlated with precipitation, suggesting that greater amounts of precipitation are associated with lower particle concentrations and vice versa. These results are comparable to those of van Wijnen et al. (1995), who found that for cyclists, the duration of rainfall influenced benzene and toluene concentrations, which were components o f the PM10 fraction measured in their study. The PM10-3 and PM10 size fractions refer to coarse particles, which include road dust, particles from tire wear, and brake linings. For instance, Harrison et al. (1999) report that according to the United States Environmental Protection Agency, 98% of particles from brake linings are within the PM10 range. A s coarse particles are mainly mechanically generated, precipitation can affect their concentration by inhibiting their ability to be mobilized from the ground. In other words, extended wet conditions would lower their concentration by preventing particles from becoming airborne, while dry conditions would increase concentrations by encouraging the formation and accumulation of these particles. P M 3 and ultrafine particle concentrations may not have been significantly correlated with precipitation because these particles are mainly formed by coagulation or combustion. Therefore, precipitation may not have as large an effect on these processes as it has on processes that form coarse particles. Though not statistically significant, ultrafine particles were positively correlated with precipitation, unlike other particle sizes which were all negatively correlated. This could be because precipitation could lower 43 temperatures, which in turn would increase ultrafine particle formation by favouring nucleation. Detailed effects of air temperature on particle concentration wi l l be discussed in section 3.5.2. 3.5.2 Air Temperature P M 3 and ultrafine particles were significantly correlated with air temperature. However, P M 3 was positively correlated, while ultrafine particles were negatively correlated. In other words, an increase in temperature would be associated with an increase in P M 3 , or a decrease in ultrafine particle concentration. A positive correlation between fine particles (PM2.5) and temperature was also found by Adams et al. (2001a). They showed that commuter cyclists were exposed to increased levels of PM2.5 with an increase in temperature, with r = 0.23. Though their correlation coefficient is much lower than the one calculated in this study, it is still significant due to their larger sample size. A s P M 3 is formed by photochemistry, coagulation, combustion, and other chemical reactions in the atmosphere, an elevated temperature may increase the rate of these formation processes, resulting in higher P M 3 • concentrations. The negative correlation between ultrafine particles and temperature can be explained in a similar method. Whereas increased temperatures favour the formation of P M 3 , it favours the loss of ultrafine particles. This is because ultrafine particles are primarily formed by nucleation, but are lost by coagulation. In other words, when ultrafine particles grow into larger particles, they end up in the accumulation mode, where fine particles reside, instead of remaining in the nucleation mode. This results in a 44 loss of ultrafine particles, even though they are not being removed from the atmosphere. Higher temperatures may decrease nucleation or increase coagulation rates, resulting in lower concentrations of ultrafine particles. Vinzents et al. (2005) also found that personal exposure to ultrafine particles when cycling in traffic was negatively related to temperature, stating that this is because of increased condensation of gases at lower temperatures. Likewise, Kittelson et al. (2004) also showed a negative correlation between temperature and ultrafine particle concentrations, finding that colder temperatures increased ultrafine particle formation by favouring nucleation. Correlations between temperature and coarse particles (PM10-3 and PM10) may not have been significant because coarse particles are mainly mechanically generated and are lost by gravitational settling or dilution. These processes are more physical than chemical, therefore would not be influenced by temperature as much as the chemically based formation and loss processes of fine and ultrafine particles. Harrison et al. (1997) also calculated a low correlation coefficient between PM10 and temperature, with r = 0.23. 3.5.3 Wind Speed Wind speed was only significantly correlated with ultrafine particles. There was a negative correlation, indicating that an increase in wind speed is associated with a decrease in ultrafine particle concentration or vice versa. These findings are similar to those of Vinzents et al. (2005), who found that exposure to ultrafine particles when cycling in traffic was inversely related to wind speed due to dispersion by high wind speeds. Zhu et al. (2002) also reported that wind speed 45 played an important role in determining levels of ultrafine particles near a freeway. They suggested that higher wind speed can lead to greater dilution rates, and in turn lower concentrations of ultrafine particles. Reasons provided by these studies could also explain why a negative correlation was found in this study. A s ultrafine particles were shown to have a heterogeneous spatial distribution in this study (Figures 4.6 and 4.17), this suggests that there could be pockets of 'c leaner ' air scattered between the areas of high ultrafine particle concentration, allowing for the dilution of ultrafine particles at high wind speeds. Though not statistically significant, correlations between wind speed and coarse particles (PM10-3 and PM10) were also negative, most l ikely due to atmospheric dilution at higher wind speeds. Harrison et al. (1997) also found a negative correlation between PM10 and wind speed, with r = -0.32. However, the correlations in this study may not be significant because high wind speeds can also resuspend coarse particles, such as road dust (Harrison et al., 1997), which could indicate a positive correlation. Consequently, the effects of atmospheric dilution and resuspension may cancel each other out, resulting in neither a significant positive nor negative correlation. The very low and not significant correlation between P M 3 and wind speed could be explained by the relatively homogeneous spatial distribution of P M 3 in this study, which is further discussed in section 4.2.1. Adams et al. (2001a) showed that wind speed lowered the concentration of fine particles by dilution and calculated a statistically significant correlation coefficient of -0.34. However, i f the particles exhibit a homogeneous spatial distribution, high wind speeds may only be transporting particles 46 with a similar concentration from one location to another, instead of diluting them, as with the case of ultrafine particles. 3 . 6 SUMMARY Overall, there are three main factors contributing to day-to-day variation of particle concentrations in this study: instrument calibration, emissions, and meteorological conditions. A s instrument calibration could not be controlled for, and emissions were assumed to be constant because of the design of this study, this suggests that meteorology is the main factor responsible for day-to-day variation. Specifically, precipitation was negatively correlated with PM10-3 and PM10, air temperature was positively correlated with P M 3 and negatively correlated with ultrafine particles, and wind speed was negatively correlated with ultrafine particles. In the following chapter, spatial patterns of particle concentrations measured along the route wi l l be investigated. Chapter 4 wi l l also include a discussion of concentrations in specific land uses and size distributions in these land uses, as well as a qualitative comparison of measured P M 3 levels to concentrations of annual mean PM2.5 as predicted by a traffic-based land use regression model. 47 4.0 P A R T I C L E C O N C E N T R A T I O N S A L O N G B I C Y C L E R O U T E S : SPATIAL V A R I A T I O N 4.1 INTRODUCTION This chapter wi l l focus on spatial patterns o f particle concentrations along the study route. The particle measurements to which are referred in this chapter have been normalized according to background averages as described in section 2.5.4 to remove day-to-day variation caused by meteorology, unless otherwise stated. Therefore, any patterns wi l l be discussed with respect to emissions. First, average concentrations along the route wi l l be analyzed, including an explanation of peaks and troughs in particle concentrations. Next, average concentrations in different land use categories wi l l be discussed, followed by a description of the particle size distribution in different land use categories. The discussion of spatial patterns wi l l close with a qualitative comparison of measured PM3 to those predicted by a traffic-based land use regression model of annual mean PM2.5. 4.2 AVERAGE CONCENTRATIONS ALONG THE ROUTE The following sections focus on average concentrations of each particle size measured along the length of the route during the 14 days of particle measurements. Averages were calculated as described in section 2.5.1. 48 4.2.1 P M 3 Figure 4.1 shows the average concentrations of PM3 over the length o f the route for all 14 days. The left hand side corresponds to the western end of the route shown in Figure 2.12. Error bars represent one standard deviation. 70.0 60.0 50.0 10.0 0.0 -10.0 Q1Q_ 5000.0 10000.0 15000.0 20000.0 . Distance (m) Figure 4.1: Average P M 3 concentrations along the data collection route. Left hand side corresponds to the western end. Error bars represent one standard deviation. PM3 shows a relatively homogeneous spatial distribution, compared to other particle sizes measured in this study (Figures 4.2, 4.3, and 4.6). Numerous other studies, such as Bari et al. (2003), DeGaetano & Doherty (2004), and Martuzevicius et al. (2004), have also found that PM2.5 exhibits little spatial variation. Furthermore, a study conducted in Vancouver by Ebelt et al. (2000) also showed that PM2.5 exhibited low spatial variability. The relatively homogeneous spatial distribution of P M 3 can be -attributed to the fact that fine particles have a relatively long atmospheric residence time, as removal mechanisms have a poor efficiency in this size region. Consequently, these particles can remain airborne for extended periods of time and disperse l ike a gas, resulting in similar concentrations over the study area. j 49 The overall average P M 3 concentration measured throughout this study was 22.6 ug/m 3 , with a 95 t h percentile of 29.8 p.g/m3. The proposed Canada-wide standard for PM2.5 is 30 pg/m (Canadian Counci l of Ministers o f the Environment, 2000), to be achieved by 2010. However, this standard is for a 24-hour averaging time, while the averaging time for this study was approximately 1.5 hours. Yet, it is important to note that the P M 3 concentrations measured in this study are still below the proposed standard for the majority of the route, as demonstrated by the 95 t h percentile. The P M 3 average measured over the route is also lower than concentrations measured in similar studies. Gee & Raper (1999) recorded an average of 54 pg/m for PM4 in Manchester, U K , and Adams et al. (2001b) measured a summer average of 34.5 ug/m 3 for PM2.5 in London, U K . A s these studies were conducted in different cities, differences in concentration can be due to a variety of factors such as different climates, emission sources, and land uses. Yet, P M 3 concentrations measured along the majority of the route in this study are still below these averages. 4.2.2 P M 1 0 3 and P M 1 0 PMio- 3 and PM10 wi l l be discussed together because the mass of PM10 is dominated by coarse particles (i.e. PM10-3). Consequently, PM10-3 and PM10 show similar spatial patterns. Figures 4.2 and 4.3 show the average concentrations of PM10-3 and PM10, respectively, over the length of the route for all 14 days. The left hand side corresponds to the western end of the route shown in Figure 2.12. Error bars represent one standard deviation. Labels correspond to specific locations along the route, which wi l l be later discussed in detail. 50 s 100.0 o -100:o -200.0 5000.0 Dis tance (m) Figure 4.2: Average PM10.3 concentrations along the data collection route. Left hand side corresponds to the western end. Error bars represent one standard deviation. 1: Bus along Universi ty Boulevard on Sept. 29. 2: Construction site at the northeast end o f Cambie Bridge. 3: Construction site along West l s l Avenue. 400.0 300.0 -Sc 200.0 o 100.0 0.0 ( -100.0 MB 5000.0 10000.0 15000.0 20000.0 Dis tance (m) Figure 4.3: Average P M ) 0 concentrations along the data collection route. Left hand side corresponds to the western end. Error bars represent one standard deviation. 1: Bus along University Boulevard on Sept. 29. 2: Construction site at the northeast end of Cambie Bridge. 3: Construction site along West 1st Avenue. Both PM10-3 and PM10 show a more heterogeneous spatial distribution than PM3. This is consistent with several other studies, such as Grivas et al. (2004) and Sun et al. (2004), which found that PM10 did not have a uniform spatial distribution and 51 concentrations differed significantly between sites. Coarse particles may exhibit a non-uniform spatial distribution because of their short gravitational settling time. Because they settle out more quickly than smaller particles, they are transported over shorter distances and thus tend to remain near their sources. As a result, higher concentrations of coarse particles are found near their sources, and lower concentrations are found further away, creating a heterogeneous spatial distribution. The peaks measured along the route are from road dust and construction sites, and are labeled in Figures 4.2 and 4.3. Peak 1 corresponds to when a bus passed very close to the bicycle lane along University Boulevard on September 29, l ikely resuspending road dust. Though this was the only occasion when a passing bus had such a large effect on particle concentrations, it still managed to significantly affect the average particle concentrations for all days, resulting in a noticeable peak. The large error bars suggest a large variation in concentration at that location. That variation is due to the extremely high concentrations measured on that one day, which contrasted with the lower concentrations on all other days. Peak 2 is a result of coarse particles originating from a construction site at the northeast end of the Cambie Bridge, as seen in Figure 4.4. Likewise, Peak 3 occurred next to the construction site at the southeast end of False Creek along West 1 s t Avenue, shown in Figure 4.5. The combination of exposed soil and industrial machinery at both of these sites can act as a prime source of coarse particles, which can be mobil ized by wind or industrial activity. This is in agreement with a study conducted by Baek et al. (1997), who found that industrial areas had higher particle concentrations than urban residential and suburban areas. 52 Figure 4.4: Construction site at the northeast end o f Cambie Bridge. Figure 4.5: Construction site along West 1 s t Avenue. E v e n though these peaks also have large error bars w h i c h suggest h i g h l y variable concentrations at these locations, Figures 3.2 and 3.3 show that the large var ia t ion is not due to outliers as i n the case o f Peak 1. The variat ion is due to different m a x i m u m concentrations recorded on different days. Y e t , it is important to note that regardless o f the value o f the m a x i m u m concentration measured on a part icular day, the m a x i m u m concentration o f most o f the rides s t i l l occurred at the locat ion o f Peak 2 or Peak 3. 53 Therefore, in terms of spatial patterns, one can conclude that these areas have elevated levels o f coarse particles, compared to other locations along the route. The overall average PM10-3 and PM10 concentrations measured throughout this study were 32.0 ug/m 3 and 53.9 pg/m 3 , respectively. PM10-3 had a 95 t h percentile of 71.1 |j,g/m3, while PM10 had a 95 l h percentile of 100.2 ug/m 3 . Canada does not have a standard for PM10-2.5, nor does it have a Canada-wide standard for PM10. However, the province of Brit ish Columbia has a PM10 standard of 50 ug/m 3 , for an averaging time of 24 hours (Environment Canada, 2002). Though the 95 t h percentile o f PM10 exceeds this standard, high concentrations were mainly encountered near construction zones and in the Transportation, Communication and Utilities and Industrial land use categories, which wi l l be further discussed in section 4.3.2. P M i 0 concentrations measured along more than half of the route (60 t h percentile = 50 ug/m 3) were equal to or below the provincial standard. There are no studies on cyclists' exposure to the coarse fraction of particles ( P M 10-2.5), and very few measuring PM10. Therefore, PM10 wi l l be discussed here as it includes the coarse fraction. One study by Bernmark et al. (2006) of bicycle messengers found that cyclists were exposed to a mean concentration of 55 ug/m 3 for PM10 in Stockholm. A lso, as previously mentioned, Gee & Raper (1999) measured an average of 54 jrg/m 3 for P M 4 in Manchester, U K . Though not directly measuring P M ] 0 , by definition PM10 concentrations would exceed those of PM4, providing an indication of PM10 levels in that city. Similar to P M 3 , average P M ) 0 concentrations measured in this study are similar to or lower than those in other studies. 54 4.2.3 Ultrafine Particles Figure 4.6 shows the average concentrations of ultrafine particles over the length of the route for all 14 days. The data used to calculate these averages were not normalized because there was no reference instrument for ultrafine particles. The left hand side corresponds to the western end of the route shown in Figure 2.12. Error bars represent one standard deviation. Labels correspond to specific locations along the route, which wi l l be later discussed in detail. 140000 120000 3 100000 I 5 20000 0 -20000 J 5000.0 10000.0 15000.0 20000.0 Distance (m) Figure 4.6: Average ultrafine particle concentrations along the data collection route. Left hand side corresponds to the western end. Error bars represent one standard deviation. 1: University Boulevard. 2: Burrard Bridge. 3: West Georgia Street. 4: Burrard Street. As with PM10-3 and P M | 0 , ultrafine particles exhibit a heterogeneous spatial distribution. This is in agreement with several studies, such as Zhu et al. (2002) and Kittelson et al. (2004), which also found that ultrafine particles did not have a uniform spatial distribution. In particular, these studies showed that the concentration of ultrafine particles is higher near roads with heavy traffic. While the heterogeneous spatial distribution of coarse particles can be attributed to their short gravitational settling time, 55 the distribution of ultrafine particles is mainly due to nucleation and coagulation. Ultrafine particles are commonly formed by nucleation, mainly from combustion by motor vehicles. As a result, higher concentrations are often found near areas with heavy traffic (Zhu et al., 2002; Kittelson et al., 2004). Furthermore, Zhu et al. (2002) argue that motor vehicles are the primary direct emission sources of ultrafine particles. However, as ultrafine particles are transported further away from their sources, they are lost by coagulation and grow into larger particles that end up occupying the fine particle size fraction. Consequently, higher concentrations of ultrafine particles are found near their sources, before the particles have the chance to coagulate into larger ones. The concept that motor vehicles are major sources of ultrafine particles is supported by this study. The main peaks in ultrafine particle concentration measured along the route are labeled in Figure 4.6. Peak 1 occurred along University Boulevard, which is frequented by cars and diesel buses. Diesel vehicle emissions contain a significant amount of ultrafine particles, compared to other particle sizes (Harrison et al., 1999). Therefore, it is not surprising to find higher levels of ultrafine particles along this section of the route. Peak 2 corresponds to the Burrard Bridge. As seen in Figure 4.7, the bicycle lane on this bridge is located directly beside the road, where the cyclist is often less than a metre away from fast and heavy traffic, especially during the morning rush hour when particle measurements were taken. A s previously mentioned, combustion from motor vehicles is a major source of ultrafine particles. Therefore, being near busy traffic could account for the elevated levels of ultrafine particles recorded while crossing this bridge. 56 Figure 4.7: Bicycle lane on Burrard Bridge. The large drop i n ultrafine particle concentration di rect ly after this peak is due to the route changing from Burra rd B r i d g e to the Seaside and C h i l c o b i cyc l e routes. F igure 4.8 shows the Seaside b i c y c l e route, w h i c h runs a long E n g l i s h B a y and is for the most part w e l l separated from traffic. F igure 4.9 shows the C h i l c o b i cyc l e route, w h i c h is located on a smal l , residential street, border ing Stanley Park and on the edge o f the down town core. A s a result, very li t t le traffic was encountered on this section o f the route. Therefore, the drop in ultrafine particles immedia te ly after crossing the Burra rd Br idge can be explained by the Seaside and C h i l c o b i cyc le routes be ing separated from or not frequented by heavy traffic. 57 Figure 4.8: Seaside bicycle route. Figure 4.9: Chilco bicycle route. H i g h levels o f ultrafine particles were encountered again when the route traveled onto Wes t Geo rg i a Street (Peak 3). Wes t Georg ia Street, seen i n F igure 4.10, has a designated b i c y c l e lane o n each side o f the street. It is classif ied as a h ighway and is the largest road sampled dur ing this study. Furthermore, it leads to the L i o n s Gate B r i d g e , w h i c h is the p r imary route to and from V a n c o u v e r ' s d o w n t o w n core for residents o f the N o r t h Shore. A s a result, this street often had a steady stream o f traffic dur ing the 58 m o r n i n g rush hour per iod when particle measurements were taken, account ing for the sharp increase i n ultrafine particle concentrations. Figure 4.10: Bicyc le lane along West Georgia Street. T h e sudden drop i n ultrafine particle concentration immedia te ly after Peak 3 occurred when the route traveled onto Cardero Street, a m u c h smaller road than Wes t Geo rg i a Street w i t h less motor vehic le traffic. However , ultrafine part icle concentrations increased again when the route returned to Burra rd Street (Peak 4). T h e b i cyc l e lane on Burra rd Street is located i n the m i d d l e o f the road, as seen i n F igure 4 .11. W i t h two lanes o f traffic to the left and one lane to the right, the cycl is t is often surrounded by motor vehicles , result ing i n h igh concentrations o f ultrafine particles. 59 Figure 4.11: Bicycle lane on Burrard Street. T h o u g h h igh concentrations o f ultrafine particles were measured on Burra rd Br idge , there was no peak i n ultrafine particle concentration for the C a m b i e Br idge , located approximate ly between 17 700 and 18 400 m a long the route. Th i s cou ld be expla ined b y the fact that the C a m b i e B r i d g e has a w ide r b i cyc l e lane further away from traffic, as seen i n F igure 4.12. Consequent ly , cycl is ts are not as close to motor vehicles as they are when cross ing the Bur ra rd Br idge , reducing their exposure to ultrafine particles. In addit ion, the Bur ra rd B r i d g e has two b i c y c l e lanes, one on each side o f the bridge, meaning that cycl is ts w i l l a lways be t ravel ing i n the same direct ion as motor vehic le traffic. H o w e v e r , the C a m b i e B r i d g e on ly has a b i c y c l e path on one side, meaning that sometimes cycl is ts are t ravel ing i n the opposite direct ion o f traffic, as demonstrated i n F igure 4 .12. T h i s cou ld reduce the amount o f direct exposure to ultrafine particles, leading to lower concentrations on C a m b i e Br idge than on Burrard Br idge . 60 Figure 4.12: Bicycle lane on Cambie Bridge. The overa l l average concentration o f ultrafine particles measured a long the study route was 33 899 Pt /cc, w i t h a 9 5 t h percentile o f 56 417 Pt /cc . A s the study o f ultrafine particles i n terms o f air po l lu t ion is s t i l l re lat ively new, Canada does not have a standard for this part icle size. O n l y one other study deal ing w i t h cyc l i s t s ' exposure to ultrafine particles was found. V i n z e n t s et a l . (2005) showed that cycl is ts were exposed to an average o f 32 400 Pt /cc, when c y c l i n g a long a f ixed 2 0 - k m route i n central Copenhagen, D K dur ing m o r n i n g and/or afternoon rush hours. The length o f Vinzen t s et al . 's (2005) route and t ime when part icle measurements were taken are very s imi la r to those i n this study, therefore it is not surprising that ultrafine particle concentrations are s imi la r as w e l l . T h i s suggests that the Copenhagen and V a n c o u v e r routes have s imi lar traffic condi t ions and other sources o f ultrafine particles. 4.3 LAND USE ANALYSIS In addi t ion to average concentrations measured a long the route, average part icle concentrations were also calculated for each land use category as defined by the G V R D . 61 Table 4.1 provides a list o f land use codes and their corresponding categories. The distribution of these land uses along the route where particulate measurements were taken can be seen in Figure 2.12. Table 4.1: Land use codes and categories, as defined by the GVRD. Land I se Code L a n d Use Category B Bridge R100 Recreation and Protected Natural Areas S110 Residential - Single Family and Duplexes S130 Residential - Townhouse and Low-rise Apartments S135 Residential - High-rise Apartments S200 Commercial S210 Commercial — Mixed Use S300 Industrial S400 Institutional S500 Transportation, Communication and Util it ies U100 Open and Undeveloped The following sections wi l l discuss in which land use category the highest average particle concentrations were measured for each size fraction. The size distribution of particles in each land use category wi l l also be addressed. 4.3.1 P M 3 Figure 4.13 shows average concentrations of P M 3 measured in each land use category. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. 62 50.0 40.0 10.0 0.0 T B R100 SI 10 S130 S135 S200 S210 S300 S400 S500 U100 L a n d Use Code Figure 4.13: Average PM 3 concentrations for each land use category. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. the Institutional land use category (S400) to 30.2 pg/m in the Transportation, Communication and Utilities areas (S500). There is no land use category that has a distinctly higher PM3 concentration than others. Compared to PM10-3, PM10, and ultrafine particles (Figures 4.14, 4.15 and 4.17), PM3 concentrations are similar over all land uses. A s discussed in section 4.2.1, PM3 exhibited a relatively homogeneous spatial distribution over the length of the route due to the long atmospheric residence time of these particles, allowing them to disperse like a gas. This could explain why there is not a large difference in concentration between land uses. These results are in agreement with several other studies. For example, DeGaetano & Doherty (2004) sampled PM2.5 in a variety of land uses, including residential, commercial, industrial, and forest areas. They found that there was very little spatial variation of PM2.5, and there were significant Average PM3 concentrations were relatively similar, ranging from 18.7 pg/m in 63 correlations between sites. Bari et al. (2003) and Martuzevicius et al. (2004) also found little spatial variation of PM2.5. 4.3.2 PMio-3 and P M 1 0 PM10-3 and PM10 wi l l be discussed together because the mass of P M | 0 is dominated by coarse particles (i.e. PM10-3). A s a result, these two size fractions show similar patterns with respect to average concentrations in particular land uses. Figures 4.14 and 4.15 show average concentrations of PM10-3 and PM10, respectively, measured in each land use category. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. 200.0 150.0 cm 3. f i 100.0 4 S 50.0 0.0 -50.0 • • • • • • • • • < • B Rl 1 1 1 1 1 1 00.S110 S130 S135 S200 S210 S3 1 . 1 1 tX) S400 S500 U100 L a n d Use C o d e Figure 4.14: Average PM10.3 concentrations for each land use category. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. 64 250.0 200.0 s 100.0 PH 50.0 4 0.0 -50.0 • > < • -• > < • • • > B R l 00 SI 10 S130 S135 S200 S210 S300 S400 S500 UlOO Land Use Code Figure 4.15: Average P M i o concentrations for each land use category. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. Compared to P M 3 , PM10-3 and PM10 concentrations are more variable. Average PM10-3 concentrations range from 18.8 pg/m 3 in the Residential - Single Fami ly and Duplexes category (SI 10) to 98.3 pg/m 3 in the Transportation, Communication and Utilities category (S500). Average PM10 concentrations range from 39.0 pg/m 3 in the | 3 i Residential - Single Family and Duplexes category to 128.0 ug/m in the Transportation, Communication and Utilities category. The land use showing the highest average particle concentration for both PM10-3 and PM10 is the Transportation, Communication and Utilities land use category. The average PM10-3 concentration was 98.3 pg/m while the average PM10 concentration was 128.0 ug/m . This land use corresponds to the area at the northeast end of the Cambie Bridge, as seen in Figure 2.12. One reason that this land use has the highest concentration of coarse particles could be because of the large construction site located in 65 this area, shown i n F igure 4.4. A s discussed i n section 4.2.2, the exposed so i l and heavy machinery at construct ion sites can act as a source o f coarse particles. Furthermore, the b i c y c l e route traveled a long an elevated w a l k w a y (Figure 4.16) leading from Pac i f i c B o u l e v a r d up to the C a m b i e Br idge . T h i s w a l k w a y was par t ia l ly covered, w h i c h cou ld hinder the dispersion o f coarse particles, increasing their concentration i n this area. Therefore, the combina t ion o f the construction site and the possible tendency o f the covered w a l k w a y to trap particles resulted i n h igh particle concentrations i n the Transportation, C o m m u n i c a t i o n and Ut i l i t i e s land use. Figure 4.16: Walkway leading from Pacific Boulevard up to the Cambie Bridge. T h e land use wi th the second highest average coarse particle concentration was the Industrial land use (S300). T h e average PM10-3 concentration was 59.3 u g / m 3 and the average PM10 concentration was 85.0 u g / m 3 . A l l industrial areas a long the route are shown i n F igure 2.12. Spec i f ica l ly , the m a i n area classified as an industrial zone is located at the southeast end o f C a m b i e Br idge , and contains the sections o f the route w h i c h run a long West 5 t h Avenue , C o l u m b i a Street, West 1 s t A v e n u e , and Ontar io Street. There are also smaller areas at the intersection o f West 10 t h A v e n u e and Arbutus Street, 66 as well as at the southeast end of Burrard Bridge. Similar to the Transportation, Communication and Utilities land use, there is a major construction site in the Industrial land use category. It borders West 1 s t Avenue and is shown in Figure 4.5. A s a result, coarse particles from this construction site could increase the overall concentration of coarse particles measured in the industrial areas along the route. However, as there are many sections of the route which pass through industrial areas but are not near this construction site, the average concentration for the Industrial land use is not as high as that of the Transportation, Communication and Util it ies land use. 4.3.3 Ultrafine Particles Figure 4.17 shows average ultrafine particle concentrations measured in each land use category. The data used to calculate these averages were not normalized because there was no reference instrument for ultrafine particles. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. 67 90000 80000 j - . -£T 70000 T : u T is 5j. 60000 j : w « 50000 T — : : : u ,, ,, (2 40000 T T : a* «a 30000 „ = : — , . U " . 5 20000 — - ' • : : 10000 — J - 1 1 1 0 T 1 1 1 1 1- 1 1 1 1 1 B R100 SI 10 S130 S135 S200 S210 S300 S400 S500 U100 Land Use Code Figure 4.17: Average ultrafine particle concentrations for each land use category. Land use codes are defined in Table 4.1. Error bars represent one standard deviation. A v e r a g e ultrafine part icle concentrations ranged from 22 296 Pt /cc i n the Resident ia l - S ing le F a m i l y and Dup lexes category (SI 10) to 46 240 Pt /cc i n the O p e n and Undeve loped category (U100) . The land use category that had the highest average ultrafine particle concentration is the Open and Undeve loped land use, w i t h an average o f 46 420 Pt /cc. H o w e v e r , the C o m m e r c i a l - M i x e d U s e (S210) and B r i d g e (B) categories showed s imi la r concentrations, w i t h 44 211 Pt /cc and 43 975 Pt /cc , respectively. T h e Open and Undeve loped land use areas are shown i n F igure 2.12. Spec i f ica l ly , they are located at the southeast end o f Bur ra rd B r i d g e , a long Pac i f i c Bou leva rd between G r a n v i l l e Street and H o m e r Street, and a long Wes t 1 s t A v e n u e between C o l u m b i a Street and Ontar io Street. M o s t o f these areas were i n the process o f be ing developed dur ing the t ime when 68 particulate measurements were taken, and diesel-powered industrial equipment such as cherry pickers and backhoes were being operated at these sites. As previously mentioned, combustion is a major source of ultrafine particles. Similar to motor vehicles, industrial equipment also have emissions due to combustion processes. A s a result, these emissions can contribute to high ultrafine particle levels in the Open and Undeveloped land use areas. High ultrafine particle concentrations measured in the Commercial - M ixed Use and Bridge land use categories can be explained by high volumes of traffic in these areas, as motor vehicles are a major source of ultrafine particles (Zhu et al., 2002). Several studies such as those conducted by Zhu et al. (2002) and Kittelson et al. (2004) have also shown that ultrafine particle concentrations are higher in areas with heavy traffic. The Commercial - M ixed Use areas are primarily located in the downtown core, including the sections of the study route which run along West Georgia Street, Alberni Street, Burrard Street and part of Pacif ic Boulevard. As measurements were taken during the morning rush hour, there was a large volume of motor vehicle traffic in the downtown core, which could increase ultrafine particle concentrations. Similarly, the Burrard Bridge, which leads into the downtown core, also had a high volume of traffic during the morning rush hour. However, the Cambie Bridge had lower ultrafine concentrations, as discussed in section 4.2.3, which could explain why the average ultrafine particle concentration for the Bridge land use is not as high as that of other land uses. 69 4.4 SIZE DISTRIBUTION The size distribution of particles in each land use category was also analyzed. The data for this analysis were not normalized because there was no reference instrument for each size category. Only measurements taken by the G R I M M were used because the P-Trak did not record size distributions. Therefore, only particles between 0.3 and 20 um in diameter were included in this analysis. Figure 4.18 shows the size distribution of particles in each land use category, and is consistent with the idea that fine particles are more homogeneously distributed in space than coarser particles (sections 4.2.1 and 4.3.1). This is mainly due to the ability of fine particles to remain airborne longer than coarse particles and thus disperse like a gas, while coarser particles tend to settle out near their sources. Particle concentrations below 1 um in diameter are relatively similar in all land use categories. However, as particles increase in size, their concentrations are more varied over different land use categories. B - B r i d g e R100-Parks SI 10- Residential, single family S130- Residential, townhouse and lowrise apart ments S I35 - Residential, highrise apartments S200 - Commercial S2I0 - Commercial. rrn>ed use S300-Industrial S400-Institutional S500 - Transportation, comm, utilities UlOO-Open and undeveloped 0.1 1 10 100 Size (pm) Figure 4.18: Size distribution o f particles in each land use category. 70 When compared to a theoretical size distribution for particles in a traffic-dominated area, such as one shown in Figure 1.2, the size distribution measured in this study shows similar peaks. A detailed discussion about the origin of these peaks can be found in section 1.2.1 Both theoretical and measured size distributions show a mechanically generated peak at approximately 7 pm, and the measured size distribution shows the right-hand tail of an accumulation peak for particles less than 1 pm. It is not surprising that these peaks appear for particles measured in this study because the study was conducted in an urban area, and majority of the route along which particles were measured were in areas frequented by motor vehicle traffic. Therefore, it was expected that a size distribution similar to that of a traffic-dominated setting would occur. 4.5 M O D E L COMPARISON As described in section 2.5.5, PM3 data were also compared to a traffic-based model of annual mean PM2.5 (Henderson et al., 2007). Figure 4.19 shows the study route superimposed over the model, and the model's specific annual mean PM2.5 concentrations along the route. Red points indicate areas with high particle concentrations (greater than 8 ug/m3) while green points show areas with low particle concentrations (less than 2 pg/m3). Sections along the study route that were predicted to have the highest annual mean PM2.5 concentrations are at the north end of Denman Street along West Georgia Street and the northeast end of Cambie Bridge, which showed annual mean PM2.5 concentrations of 9.6 to 10 pg/m3. The Burrard Bridge also showed relatively high concentrations, with values ranging from 9 to 9.5 pg/m . 71 J Annual Mean PM2 5 Annual Mean PM2 5 N uglm*3 (along study route) ug/mA3 A • 0 - 1.99 • 2 00 - 3.99 M • 4 00 - 5 99 HI Low 0 • 6 00- 7.99 Major Roads and Highways • 8 00- 10 00 0 0 5 1 2 Kilometers 1 i i i I i i i I Figure 4.19: Annual mean P M 2 5 concentrations as predicted by a model along the study route. (Data source: Michael Brauer, Sarah Henderson, DMTI Spatial.) A s the metrics i n this study and the mode l were not ident ical , a qualitative approach was used to compare concentrations. F igure 4.20 shows the average PM3 concentration measured a long the length o f the study route and the annual P M 2 . 5 concentrations as predicted b y a traffic-based mode l a long the same route. 72 80.0 70.0 ^ 60.0 S 50.0 ~5JD 3 40.0 S 30.0 C M 20.0 10.0 0.0 - P M 3 • Annual PM2.5 (Model) 0.0 5000.0 10000.0 15000.0 Distance (m) 20000.0 Figure 4.20: Average PM 3 and modeled annual PM2.s along the data collection route. Left hand side corresponds to the western end. Error bars represent one standard deviation. While the average P M 3 measured along the route was 22.6 pg/m 3 , the average annual PM2.5 predicted by the model was only 4.64 pg/m . The annual mean PM2.5 as predicted by the model is consistently lower than PM3 measured in this study for several reasons. The main reason might be because particle measurements for this study were taken on the road, while the model is based on roadside measurements. Therefore, as measurement devices used for the development of the model were further away from exhaust sources, particle concentrations would be lower. Next, the model is predicting PM2.5, not P M 3 . Therefore, omitting particles between 2.5 and 3 pm in diameter would reduce modeled concentrations compared to those measured in this study. Furthermore, the model is for particulate matter concentrations averaged over a year, while measurements for this study only spanned the length of three months. A longer averaging time can lower mean concentrations by reducing the influence of high outlying values. Lastly, even though the three months during when particle concentrations were measured included several warm dry days and cool wet days, meteorological conditions still may 73 not be representative of the entire year. Much lower particle concentrations could be encountered during the winter, when temperatures are cooler. (The effect of air temperature on P M 3 was discussed in section 3.5.2). Particle concentrations may have been higher during the study period compared to other times during the year, which could also explain why measured values were higher than modeled values. Likewise, particle measurements on which the model was based were only taken between March 5 and May 8. These months can still be relatively cool and wet in Vancouver, resulting in low particle concentrations which are not representative of the entire year, which in turn produced low modeled PM2.5 values. The model shows three distinct peaks along the route which correspond to the areas of high PM2.5 concentrations mentioned above. The first peak at approximately 9500 m corresponds to south end df Burrard Bridge, the second peak between 13 000 m and 14 000 m was measured along West Georgia Street, and the third peak at about 17 500 m corresponds to the north end of Cambie Bridge. Both measured and modeled particle concentrations show a peak along West Georgia Street. However, the measured particle concentrations gradually increase along Cambie Bridge, rather than showing a strong peak at the north end as seen in the model. Similarly, measured particle concentrations show a broad increase over Burrard Bridge, rather than a sharp peak at the south end as seen in the model. The peak in particle concentrations along West Georgia Street which was predicted by the model could correspond to measured concentrations because West Georgia Street is the largest road sampled during this study and often has heavy traffic, especially during morning rush hour when particle measurements were taken. Therefore, 74 concentrations predicted by the model would correctly reflect actual concentrations because both the model and the actual measurements are traffic-dominated. The model captured an increase in particulate matter concentrations at the north end of Cambie Bridge and the south end of Burrard Bridge, but not along their lengths as measured during this study. This could be explained by the fact that measurements for this study were more precise than those used to estimate particle concentrations for the model. A s a result, this study was able to capture changes in particle concentrations at smaller scales than the model. A s previously discussed, the model is based on different land use characteristics in circular buffers of different sizes. The smallest buffer had a 100 m radius. Therefore, i f the land use or traffic characteristics do not change over the small scales where changes in particle concentrations were measured (i.e. increases in the middle of bridges), the model would not be able to capture these changes. Peaks in particulate matter that were measured but not seen in the model were primarily due to irregular or non-traffic-based sources. For example, the peak in PM3 seen at approximately 5500 m was due to a bus passing very close to the cyclist on September 29. Clearly, the model would not have predicted this as it was an isolated occurrence. A lso , the peak in PM3 near the end of the route at slightly before 20 000 m could have been due to emissions from industrial machinery at a construction site along West 1 s t Avenue, seen in Figure 4.5. Though the model shows a slight increase in the area corresponding to the construction site, this may reflect the fact that the construction site was in an industrial area. However, as construction sites can be relatively short-l ived, they were not accounted for in the model. Thus, this plateau did not reflect the high particle concentrations which were attributed to the construction site. 75 Overall, as the model was primarily traffic-based, it corresponded with actual peaks in particle concentrations in traffic-dominated areas. However, at other locations where particle concentrations could be subjected to non-traffic influences, the model is not as representative of the actual particle concentrations because these areas have particulate matter sources which the model did not account for. 4.6 SUMMARY In general, the spatial variation o f particles varied depending on their size. PM3 was found to have a relatively homogeneous spatial variation, which can be attributed to the ability of fine particles to remain airborne for extended periods of time. A s a result, they can disperse further away from their sources. On the other hand, PM10-3 and PM10 exhibited a more heterogeneous spatial distribution. This is because these coarse particles have a shorter settling time and are not transported as far away from their sources as fine particles. Higher concentrations were measured in areas with construction sites and locations where road dust was resuspended. Ultrafine particles also showed a heterogeneous spatial distribution. This is due to the formation o f these particles by combustion from motor vehicles, then their loss by coagulation as they travel away from their sources. The average concentrations o f PM3, PM10-3, and PM10 measured in this study are for the most part comparable to those measured in similar cycling exposure studies conducted in different cities. Though federal or provincial particulate matter standards do not exist for all particle sizes measured in this study, when they do exist, particle 76 concentrations measured during this study were lower than or similar to these standards for the majority of the study route. When compared to a traffic-based model of average annual PM2.5, actual particle concentrations measured in this study corresponded to the model in traffic-dominated areas. However, along other sections of the route which may have had sources of particulate matter other than traffic, the model was not as representative o f actual concentrations. 77 5.0 CONCLUSION 5.1 INTRODUCTION This chapter sums up the findings of this study with respect to the objective laid out at the beginning of the study, followed by a discussion of future research which could be conducted to provide a more in-depth picture o f cyclists' exposure to particulate matter along bicycle routes. This chapter wi l l close with recommendations and potential applications of these results. ^ 5.2 SUMMARY OF FINDINGS The primary objective of this study was to determine where cyclists are exposed to the highest levels of particulate matter along bicycle routes. Results showed that different areas can exhibit different levels of particulate matter, depending on the particles' size. Whi le concentrations of P M 3 were found to be relatively spatially r uniform over the length of the study route, P M 10-3 and P M ip showed a more heterogeneous spatial distribution. Specifically, construction sites and areas susceptible to the suspension of road dust have higher concentrations of these coarse particles. Ultrafine particles were also heterogeneously distributed in space, with areas with heavy traffic volumes having the highest concentrations. Yet, overall particle concentrations measured in this study were comparable to those measured in similar cycling exposure studies in other cities. This study also showed that meteorology may have an impact on cyclists' exposure to particulate matter. Results indicated that P M 3 was positively correlated with air temperature, PM10-3 and PM10 were negatively correlated with precipitation, and 78 ultrafine particles were negatively correlated with both air temperature and wind speed. These results suggest that not only would the route influence cyclists' exposure to particulate air pollution, but weather conditions in which they cycle could also play a role. 5.3 FUTURE RESEARCH Though this study sheds some light onto the exposure of cyclists to particulate matter along bicycle routes, there are several aspects of this study that could be improved upon to provide a more complete understanding of this topic. 5.3.1 Duration of Particle Measurements One aspect of this study that could be expanded upon is the duration o f particulate matter measurement. Due to the time constraint of this study, measurements were only taken between August and October. Therefore, particle concentrations may not have been representative of the entire year because of different meteorological conditions occurring in other months. If possible, future studies should measure particulate matter over the span of an entire year to capture more representative particle concentrations along bicycle routes in Vancouver. 5.3.2 Extent and Location of Sampled Bicycle Routes Next, a greater variety of bicycle routes should be sampled. Practical reasons caused the route along which particles were measured in this study to be restricted to the northwest corner of Vancouver. The route also only passed through one major Industrial 79 zone and one major Transportation, Communication and Util it ies zone. Furthermore, there were several construction sites along the route, which could have acted as a confounding variable, skewing particulate concentrations in some land uses. However, the bicycle route network covers the entire city, as seen in Figure 2.1. Therefore, to provide a more accurate idea of particulate pollution along Vancouver's bicycle routes, a more extensive network of bicycle routes should be sampled. In addition, the route should include several sections of each land use category to ensure that any concentrations measured in a particular land use is typical of that land use and not due to characteristics of a single section. A lso , the route should avoid construction sites to reduce the tendency of these sites to elevate particle concentrations, producing concentrations that may not be typical of the land uses where these sites are located. 5 . 3 . 3 Comparison to Motorists The comparison of motorists' exposure to cyclists' exposure would be of great interest, as it is a popular misconception that cyclists are exposed to higher levels of particulate matter than drivers of motor vehicles. However, several studies have demonstrated that cyclists' personal exposure levels to particulate matter are significantly lower than drivers' exposure levels (Gee & Raper, 1999; Adams et al., 2001b; Rank et al., 2001). Reasons for this include the fact that cyclists often travel beside traffic instead of directly behind motor vehicles, which reduces their direct exposure to vehicle exhaust (Gee & Raper, 1999). A lso , cyclists are able to bypass slow-moving or idling vehicles, reducing their time in a congested and polluted environment (Adams et al., 2001b). 80 It would not be difficult to compare cyclists' exposure to particulate matter measured in this study to motorists' exposure, as the comparison would simply involve placing the same instruments in a motor vehicle and driving along the same route traveled by the cyclist. A comparison of cyclists' and motorists' exposure to particulate matter in Vancouver could also provide a basis of comparison to the literature, and would be of great interest to the cycling community as well as individuals who spend much of their time in traffic and are concerned with their health. Cyclist ventilation rate should also be taken into account, van Wijnen et al. (1995) determined that the breathing rate of cyclists was about 2.3 times higher than that of drivers. Vedal (1995) stated that patterns of particulate deposition in the respiratory tract can depend on an individual's behaviour. For example, heavy breathing, which can occur during exercise, can increase airflow in the larger airways. Consequently, this behaviour wi l l concentrate a larger amount of particles in these larger airways rather than deeper in the lungs. A lso, breathing through one's mouth, as one often does during exercise, w i l l increase the amount of particles that reach the lung, as when one breathes through the nose, particles are caught in the nasal passage. Therefore, in order to evaluate the actual amount of particles that cyclists ingest, breathing rate must be taken into account. Travel time should also be considered. Despite Rank et al. (2001) showing that drivers and cyclists travel at similar speeds during rush hour, this might not hold true over longer distances or non-rush hour times. Cyclists wi l l most l ikely have a longer travel time than drivers, which could lead to longer exposures and greater doses of particulate matter air pollution. 81 5.3.4 Particle Composition Finally, the composition of particles to which cyclists are exposed should be analyzed. It is possible to determine the source as well as the toxicity o f particulate matter from the particles' composition. Though this study investigated the size o f particles as a means of determining their source, particle composition may be able to pinpoint more specific sources. For example, in an urban environment, carbon is mainly found in particulate matter from diesel engines. These particles, which are also known as soot, are composed of a core of elemental carbon and a surface coating o f semi-volatile organic carbon which condenses from exhaust gases (Harrison & Y i n , 2000). Some argue that adverse health effects can be caused by organic compounds that are adsorbed onto the carbonaceous core of these sooty particles (Obbt et al., 2002). Therefore, revealing the composition and thus the specific source from which particles originate could lead to a greater understanding of where cyclists are at greatest risk due to high levels of particulate air pollution. 5.4 APPLICATIONS AND RECOMMENDATIONS Though there are several aspects of this study which can be expanded upon, a few recommendations can still be made based on the study's preliminary results. A s previously mentioned, the amount of air pollution and traffic along a bicycle route is one of the main factors influencing the likelihood o f an individual choosing to cycle (Teschke et al., 2007). Therefore, i f Vancouver wishes to encourage more individuals to cycle, the city should consider planning future bicycle routes away from areas with high levels o f particulate pollution, such as streets with heavy traffic. In 82 addition, bicycle routes along bridges should be located further away from traffic i f possible, as seen on Cambie Bridge, to reduce cyclists' exposure to particulate pollution when traveling alongside the fast and heavy traffic which is commonly seen on bridges in Vancouver. A s it is not possible to anticipate where construction sites wi l l crop up throughout the city, it would not be feasible to plan future bicycle routes away from these sites. However, i f a major construction site does occur along a bicycle route, a safe and accessible detour route should be available for cyclists to reduce their exposure to particulate pollution from these sites. Lastly, though there is still much to be learned about where cyclists are exposed to the highest levels of particulate pollution, this is one of very few studies focusing on cyclists' exposure to air pollution along bicycle routes. It is hoped that this study wi l l draw attention to the need for more research to be carried out in this field. It is only with a greater amount of research and understanding that cyclists wi l l know where they can truly take a breath of fresh air. 83 R E F E R E N C E S Adams, H.S. et al. 2001a. Determinants of fine particle (PM2.5) personal exposure levels in transport microenvironments, London, U K . Atmospheric Environment. 35: 4557-4566. Adams, H.S. et al. 2001b. Fine particle (PM2.5) personal exposure levels in transport microenvironments, London, U K . The Science of the Total Environment. 279: 29-44. Baek, S. et al. 1997. A quantitative estimation of source contributions to the concentrations of atmospheric suspended particulate matter in urban, suburban, and industrial areas of Korea. Environment International. 23: 205-213. Bar i , A . et al. 2003. Measurements of gaseous H O N O , HNO3, S 0 2 , HCI , N H 3 , particulate sulfate and PM2.5 in New York, N Y . Atmospheric Environment. 37: 2825-2835. V Bernmark, E. et al. 2006. Bicycle messengers: energy expenditure and exposure to air pollution. Ergonomics. 49: 1486-1495. Bevan, M.A . J , et al. 1991. Exposure to carbon monoxide, respirable suspended particulates, and volatile organic compounds while commuting by bicycle. Environmental Science & Technology. 25: 788-'/'91. Canadian Council o f Ministers of the Environment. 2000. Canada-wide Standards for Particulate Matter (PM) and Ozone. C C M E Council of Ministers, Quebec City. Chan, L .Y . et al. 2001. Spatial variation of mass concentration of roadside suspended particulate matter in metropolitan Hong Kong. Atmospheric Environment. 35: 3167-3176. DeGaetano, A .T . and O . M . Doherty. 2004. Temporal, spatial and meteorological variations in hourly PM2.5 concentration extremes in New York City. Atmospheric Environment. 38: 1547-1558. Ebelt, S.T. et al. 2000. Exposure of chronic obstructive pulmonary disease patients to particulate matter: relationships between personal and ambient air concentrations. Journal of the Air & Waste Management Association. 50: 1081 -1094. Environment Canada. Canadian Climate Data. [Online.] www.climate.weatheroffice.ec.gc.ca/climateData/canada_e.html. Accessed November 2007. Environment Canada. The Green Lane: National Ambient A i r Quality Standards: proposed decisions on particulate matter and ozone. [Online.] www.ec.gc.ca/pdb/can_us/qual/matter.html. Accessed 16 July 2007. 84 Gee, I.L. and D.W. Raper. 1999. Commuter exposure to respirable particles inside buses and by bicycle. The Science of the Total Environment. 235: 403-405. Grivas, G . et al. 2004. Spatial and temporal variation of PMio mass concentrations within the greater area o f Athens, Greece. Water, Air, and Soil Pollution. 158: 357-371. Gull iver, J . and D.J. Briggs. 2004. Personal exposure to particulate air pollution in transport microenvironments. Atmospheric Environment. 38: 1-8. Harrison, R . M . and J . Y i n . 2000. Particulate matter in the atmosphere: which particle properties are important for its effects on health? The Science of the Total Environment. 249: 85-101. Harrison, R . M . et al. 1999. Source Apportionment of Airborne Particulate Matter in the United Kingdom. Airborne Particle Expert Group, London. Harrison, R . M . et al. 1997. Sources and processes affecting concentrations of PMio and PM2.5 particulate matter in Birmingham (U.K.). Atmospheric Environment. 31: 4103-4117. Harrison, R . M . et al., 1996. Airborne Particulate Matter in the United Kingdom. The Quality of Urban A i r Review Group, London. Henderson, S.B. et al. 2007. Application of land use regression to estimate long-term concentrations o f traffic-related nitrogen oxides and fine particulate matter. Environmental Science & Technology. 41: 2422-2428. Kittelson, D.B. et al. 2004. Nanoparticle emissions on Minnesota highways. Atmospheric Environment. 38: 9'-19. Larson, T. et al. 2007. A spatial model of urban winter woodsmoke concentrations. Environmental Science & Technology. 41: 2429-2436. Levy, J.I. et al. 2001. Fine particulate matter and polycyclic aromatic hydrocarbon concentration patterns in Roxbury, Massachusetts: a community-based GIS analysis. Environmental Health Perspectives. 109: 341-347. Maletto, A . et al. 2003. Profiles of particulate matter size distributions using a balloon-borne lightweight aerosol spectrometer in the planetary boundary layer. Atmospheric Environment. 37: 661-670. Martuzevicius, D. et al. 2004. Spatial and temporal variations of PM2.5 concentration and composition throughout and urban area with high freeway density - the Greater Cincinnati study. Atmospheric Environment. 38: 1091-1105. 85 McKendry, L G . Case Study SizeLSegregated Vertical Profiling of Particulate Matter in the PBL. Unpublished lecture, G E O G 402: A i r Pollution Meteorology, U B C , March 2005. Monn, C . 2001. Exposure assessment of air pollutants: a review on spatial heterogeneity and indoor/outdoor/personal exposure to suspended particulate matter, nitrogen dioxide and ozone. Atmospheric Environment. 35: 1-32. Moveable Type Ltd. Calculate distance, bearing and more between two latitude/longitude points. [Online.] www.movable-type.co.uk/scripts/latlong.html. Accessed 7 Apr i l 2007. ' Obot, C .J . et al. 2002. Surface components of airborne particulate matter induce macrophage apoptosis through scavenger receptors. Toxicology and Applied Pharmacology. 184:98-106. Oke, T.R., 1987. Boundary Layer Climates. Routledge, London. Percival, A . G V R D A i r Monitoring and Computer Services Supervisor. Personal communication. 19 March 2007. Rank, J . et al. 2001. Differences in cyclists and car drivers exposure to air pollution from traffic in the city o f Copenhagen. The Science of the Total Environment. 279: 131-136. Spurny, K.R. , 2000. Aerosol Chemical Processes in the Environment. Lewis Publishers, New York. Sun, Y . et al. 2004. The air-borne particulate pollution in Bei j ing - concentration, composition, distribution and sources. Atmospheric Environment. 38: 5991-6004. Teschke, K., et al. Cycling in cities: Understanding people, neighbourhoods and infrastructure to promote an epidemic of change. Report to Transport Canada, Mov ing on Sustainable Transportation Program. Apr i l 2007, 60 pp. U.S. Environmental Protection Agency. Science Topics: Particulate Matter: Background. [Online.] http://es.epa.gov/ncer/science/pm. Accessed 22 August 2007. van Wijnen, J .H. et al. 1995. The exposure of cyclists, car drivers and pedestrians to traffic-related air pollutants. International Archives of Occupational and Environmental Health. 67: 187-193. Vedal , S. 1995. Health Effects of Inhalable Particles: Implications for British Columbia. Ministry of Environment, Lands and Parks, Vancouver. 86 Vinzents, P.S., et al. 2005. Personal exposure to ultrafine particles and oxidative D N A damage. Environmental Health Perspectives. 113: 1485-1490. Watson, J . G . et al. 1997. Ambient particles and health: Lines that divide. Journal of the Air & Waste Management Association. 47: 995-1008. Wi lson, J . G . et al. 2005. A review o f intraurban variations in particulate air pollution: Implications for epidemiological research. Atmospheric Environment. 39: 6444-6462. Woo, K .S . 2001. Measurement of Atlanta aerosol size distributions: observations o f ultrafine particle events. Aerosol Science and Technology. 34: 75-87. Zhu, Y . et al. 2002. Study of ultrafine particles near a major highway with heavy-duty diesel traffic. Atmospheric Environment. 36: 4323-4335. 87 APPENDIX: P - T R A K T I M E L A G C A L C U L A T I O N The fol lowing calculation determines the time lag (if any) between the initial uptake of a sample and when it reaches the measurement chamber to be recorded. -1 I'-li.ik Figure A. 1: Schematic of P-Trak and measurements. Figure not drawn to scale. Total sample f low rate = 700 cm 3 /min Tube length =127 cm Tube diameter = 0.3 cm Handle length = 15.2 cm Nozzle length (not extended) = 15.6 cm Volume = 7rr h .. . = 7i(0.15cm)2(127+15.2+15.6cm) =11.15 cm 3 11.15 cm 3 =0.0159 min = 0.95 sec 700 cm 3 /min The time it takes for a sample to travel from the intake to the measurement chamber is less than 1 second, therefore the time lag between sample uptake and sample measurement is negligible. The cyclist traveled at approximately 15 km/h while making particle measurements, therefore measurement error is approximately 4 metres. 15.2cm 127cm I 1 15.6cm 1. I 88 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0093063/manifest

Comment

Related Items