CHARACTERIZATION AND SOURCE APPORTIONMENT OF PARTICULATE MATTER LESS THAN 10 MICRONS IN DIAMETER IN THE PRINCE GEORGE AIRSHED by Christine Breed BSc.(Agr)., The University of Guelph, 1994 THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in ENVTRONMENTALSCIENCE ©Christine Breed, 1998 THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA September 1998 All rights reserved. This work may not be reproduced in whole or in part, by photocopy or other means, without the permission of the author. ABSTRACT The susceptibility of the Prince George airshed to high concentrations of particulate matter less than 10 microns in diameter (PM10) have raised considerable concern because of the possible health impacts attributed to this air pollutant. This study examined the chemical and morphological characteristics of samples collected from two main PM10 sources and selected ambient samples from the archive of the Ministry of Environment to determine the contributions from these PM10 sources to the PM10 composition during Episodic and Non-Episodic events. The sources sampled included road dust taken from street sweepings, snow removed from city streets and unpaved roads and a beehive burner sample. PM10 samples from three Episodic events with 24 hour PM10 levels >50!-!g/m3 and three Non-Episodic events with 24 hour PM10 levels <50!-!g/m3 were examined in the bowl area ofPrince George (represented by three sampling sites: Plaza, Van Bien, and Lakewood) using a Scanning Electron microscope with Energy Dispersive system and Inductively Coupled Plasma Emission Spectroscopy. Episodes and Non-Episodes were also examined in the BCR industrial site. Results show that rounded, spherical and oval shaped particles were diagnostic of combustion sources, while amorphous shaped particles were dominant in all samples. The particle size distributions indicated that combustion sources contributed more to the fine fraction ofPM10 (<2.5!-!m) than road dust. The presence of a substantial amount ofPM10 with a diameter of3-4!lm is diagnostic for significant contributions of the road dust source to ambient PM10. The qualitative chemical analysis suggested that high concentrations of aluminum, silicon and magnesium were indicative of road dust while high concentrations of carbon, sodium and sulphur were indicative of combustion and industrial sources. Principal Component Analysis (PCA) was performed on the qualitative chemical data and four discernable sources were identified as contributing to the ambient PM10 in all locations: road dust, ii industrial, combustion, and salt. Most of the episodes examined were dominated by road dust while the non-episodes were influenced by industrial, combustion and road dust. The presence of sulphur in the ambient PMto sampled is a cause for concern due to the possible health implications. The methodology developed in this study can be applied to future source apportionment for the Prince George Airshed. TABLE OF CONTENTS Abstract ll - ill Table of Contents IV- V List ofTables VI List of Appendix Tables Vll List ofFigures Vl11 Acknowledgement IX Dedication X INTRODUCTION 1 Chapter One Chapter Two Chapter Three Literature Review Sources, Types and Composition ofPM10 Health Impacts ofPM10 PM10 Accumulation PMw in the Prince George Airshed Materials & Methods PM10 from Source Samples PM10 from Ambient Samples PMw Collection from Source Samples Morphology and in-situ Chemical Composition Total Chemical Composition Other Analyses Statistical Analysis Results and Discussion Morphological and Chemical Properties ofPM10 Sources Road Dust Beehive Burner Pulp Mill Comparison of Source Samples Morphological and Particle Size Characteristics Chemical Composition Characterization ofEpisodes and Non-Episodes in the Bowl Area Episode 1 Episode 2 Episode 3 4 8 11 12 14 15 18 19 22 24 27 30 36 37 38 39 46 52 55 IV Chapter 3 continued Chapter 4 Comparison ofEpisodes Non-Episode 1 Non-Episode 2 Non-Episode 3 Comparison ofNon-Episodes Comparison ofEpisodes and Non-Episodes Comparison ofEpisodes and Non-Episodes in the BCR site BCR Episodes BCR Non-Episodes BCR Episodes versus Non-Episodes Comparison ofBowl and BCR areas: Episodes and Non-Episodes Examination ofDifferences in Particle Size and Filter Location Comparison ofParticle Diameter and Mass 57 58 63 65 67 68 Conclusions and Recommendations 93 Literature Cited 76 82 84 89 90 91 100 Appendix A Meteorological Conditions on Study Dates 105 AppendixB Morphological Characterization 106 Appendix C Data from Carbon Coated Sample 111 AppendixD Blank Teflon Filter 113 AppendixE Standard Recoveries for Elemental Analysis (ICP) 115 AppendixF Teflon Blank for Quantitative Elemental Analysis (ICP) 116 Appendix G PCA Tables by Location 117 AppendixH ANOVA Results for Qualitative Chemical Analyses and Morphological! 131 Qualitative Chemical Analyses v LIST OF TABLES 5-6 Table 1: Composition ofNatural and Anthropogenic Sources of Particulates 16 Table 2: Location and Description of Ambient and Source Samples 25 Table 3: Comparison ofElemental Analysis (ICP) between Episodes and Non-Episodes 31 Table 4: Distribution ofMorphological Shapes in PM10 Sources 31 Table 5: Comparison of Average Particle Size for Sources 31-32 Table 6: Quantitative Chemical Composition ofPM10 Sources 35 Table 7: EDAX Qualitative Chemical Characterization ofPM10 Sources Table 8: Significant Correlation between Elemental Composition and Average Particle 40 Diameter in PM10 Sources 44 Table 9: PCA Eigenvalues and Primary Factors: Street Sweepings 44 Table 10: PCA Eigenvalues and Primary Factors: Snow Removal Table 11 : PCA Eigenvalues and Primary Factors: Unpaved Road Dust 45 45 Table 12: PCA Eigenvalues and Primary Factors: Beehive Burner 47 Table 13 : Distribution ofVarious Morphological types in selected Episodes 48 Table 14: Comparison ofMorphology between Episodes and Non-Episodes 49 Table 15 : Comparison ofParticle Size by location for Episodes and Non-Episodes 49-50 Table 16: Qualitative Chemical Characterization ofPMw Episodes Table 17: PCAEigenvalues and Primary Factors: Episode 1- 950121 52 54 Table 18: PCAEigenvalues and Primary Factors: Episode 2- 950328 Table 19: PCA Eigenvalues and Primary Factors: Episode 3 - 960227 57 59 Table 20: Distribution ofVarious Morphological types in selected Non-Episodes Table 21: Qualitative Chemical Characterization ofPM10 Non-Episodes 61-62 Table 22: PCAEigenvalues and Primary Factors: Non-Episode 1-960122 63 Table 23 : PCA Eigenvalues and Primary Factors: Non-Episode 2 - 960304 65 Table 24: PCA Eigenvalues and Primary Factors: Non-Episode 3 - 960509 67 Table 25 : Comparison of Qualitative Chemical Characterization in Episodes/Non-Episodes 71 Table 26: Comparison of Significant Correlation between Elemental Composition and Particulate Diameter 73 Table 27: Comparison of Qualitative Chemical Composition and Morphology in Episodes 74 Table 28: Comparison of Qualitative Chemical Composition and Morphology in Non-Episodes 75 Table 29: Comparison ofMorphology types: BCR site 77 Table 30: Comparison ofParticle Size for Episodes I Non-Episodes in the BCR site 77 Table 31 : Qualitative Chemical Characterization ofPM10 Episodes and Non-Episodes in the BCR site 79-80 Table 32: PCA Eigenvalues and Primary Factors: BCR Episodes 81 Table 33 : PCA Eigenvalues and Primary Factors: BCR Non-Episodes 84 Table 34: Comparison of Quantitative Elemental Analysis in the BCR site 85 Table 35 : Comparison ofMorphology between BCR Episodes and Non-Episodes 85 Table 36: Comparison of Significant Correlation between Elemental Composition and Particulate Diameter in the BCR site 88 Table 37: Comparison of Qualitative Chemical Composition and Morphology in BCR Episodes and on-Episodes 88 Table 38 : Comparison ofParticle Size Distribution on Different Filter Locations 90 vi LIST OF APPENDIX TABLES Table 1: Meteorological Conditions on Study Dates Table 2: Comparison of Average Sample Standards in Quantitative Analysis Table 3: Average Means I Standard Deviations for Blank Filter Table 4: PCAEigenvalues and Primary Factors: Episode 1-950121 Plaza Table 5: PCA Eigenvalues and Primary Factors: Episode 1- 950121 Van Bien Table 6: PCAEigenvalues and Primary Factors: Episode 1- 950121 Lakewood Table 7: PCA Eigenvalues and Primary Factors: Episode 2- 950328 Plaza Table 8: PCAEigenvalues and Primary Factors: Episode 2- 950328 Van Bien Table 9: PCA Eigenvalues and Primary Factors: Episode 2 - 950328 Lakewood Table 10: PCA Eigenvalues and Primary Factors: Episode 3 - 960227 Plaza Table 11 : PCA Eigenvalues and Primary Factors: Episode 3 - 960227 Van Bien Table 12: PCA Eigenvalues and Primary Factors: Episode 3 - 960227 Lakewood Table 13 : PCA Eigenvalues and Primary Factors: Non-Episode 1- 960122 Plaza Table 14: PCA Eigenvalues and Primary Factors: Non-Episode 1 - 960122 Van Bien Table 15 : PCA Eigenvalues and Primary Factors: Non-Episode 1 - 960122 Lakewood Table 16: PCA Eigenvalues and Primary Factors: Non-Episode 2- 960304 Plaza Table 17: PCA Eigenvalues and Primary Factors: Non-Episode 2- 960304 Van Bien Table 18: PCA Eigenvalues and Primary Factors: Non-Episode 2- 960304 Lakewood Table 19: PCAEigenvalues and Primary Factors: Non-Episode 3- 960509 Plaza Table 20: PCA Eigenvalues and Primary Factors: Non-Episode 3 - 960509 Van Bien Table 21 : PCA Eigenvalues and Primary Factors: Non-Episode 3 - 960509 Lakewood Table 22: PCA Eigenvalues and Primary Factors: BCR Episode 940408 Table 23 : PCA Eigenvalues and Primary Factors: BCR Episode 940923 Table 24: PCA Eigenvalues and Primary Factors: BCR Episode 950316 Table 25 : PCA Eigenvalues and Primary Factors: BCR Episode 950328 Table 26: PCA Eigenvalues and Primary Factors: BCR Episode 950831 Table 27: PCA Eigenvalues and Primary Factors: BCR Episode 960304 Table 28 : PCAEigenvalues and Primary Factors: BCR Episode 960813 Table 29: PCAEigenvalues and Primary Factors: BCR Non-Episode 960122 Table 30: PCA Eigenvalues and Primary Factors: BCR Non-Episode 960509 Table 31 : Krustal Wallis ANOVA results for Qualitative Chemical Analyses: Sources Table 32: Krustal Wallis ANOVA results for Qualitative Chemical Analyses: Bowl Episodes Table 33 : Krustal Wallis ANOVA results for Qualitative Chemical Analyses: Bowl Non-Episodes Table 34: Krustal Wallis ANOVA results for Qualitative Chemical Analyses: BCR Episodes I Non-Episodes Table 35 : Krustal Wallis ANOVA results for Morphological I Qualitative Chemical Analyses: Bowl Episodes I Non-Episodes Table 36: Krustal Wallis ANOVA results for Morphological I Qualitative Chemical Analyses: BCR Episodes I Non-Episodes 105 115 116 117 117 118 118 119 119 120 120 121 121 122 122 123 123 124 124 125 125 126 126 127 127 128 128 129 129 130 131 132 133 134 135 135 vii LIST OF FIGURES Figure 1: Sampling Locations for PM10 in the Prince George Airshed Figure 2: Filter Sampling Locations Figure 3: Particle Size Distribution: Street Sweepings Figure 4: Particle Size Distribution: Snow Removal Figure 5: Particle Size Distribution: Unpaved Road Dust Figure 6: Particle Size Distribution: Beehive Burner Figure 7: Particle Size Distribution: Episode 1 - 950121 Figure 8: Particle Size Distribution: Episode 2- 950328 Figure 9: Particle Size Distribution: Episode 3 - 960227 Figure 10: Particle Size Distribution: Non-Episode 1 - 960122 Figure 11 : Particle Size Distribution: Non-Episode 2 - 960304 Figure 12: Particle Size Distribution: Non-Episode 3 - 960509 Figure 13 : Particle Size Distribution: Episodes Figure 14: Particle Size Distribution: Non-Episodes Figure 15 : Particle Size Distribution: BCR Episodes Figure 16: Particle Size Distribution: BCR Non-Episodes Figure 17: Average Particle Size Distribution: Episodes and Non-Episodes Figure 18: Average Particle Mass Distribution: Episodes and Non-Episodes 17 20 33 33 34 37 51 53 56 60 64 66 70 70 78 83 92 92 viii ACKNOWLEDGEMENT Many thanks to the following people for their invaluable help during the preparation of this thesis. Jennifer Wilson David & Colette Purcell-Chung Ruman Carter David Sutherland Steve Lambie Dennis Fudge Guy Plourde Peter Jackson David Dick Peter McEwan Jill Craig Frank: Blues Mark Logan Richard Crombie Joselito Arocena BrunoZumbo Paul Broda Jane Hohenadel ix This thesis is dedicated to my family whose love and support mean everything Roger, Yvonna, & Allen Breed Charles & Marietta Gabriele Leslie & Gertrude Breed Wayne, D'arcy & Terry Gabriele X INTRODUCTION Particulate Matter is a collective term for the complex and varying mixture of air pollutants found in minute solid and liquid form. Particulate Matter contains both organic and inorganic compounds and varies in size, composition, origin and health hazards (Dockery & Pope, 1994). Examples of particulate matter include fine dusts which are formed from the mechanical breakdown of rocks (i.e. winter sanding materials) and smoke which is formed from combustion activities (i.e. fireplaces, vehicles, industry). Particulate Matter is considered to be a serious health concern for a considerable portion of the population. The World Health Organization concluded that there are over one billion people exposed to excessive levels of particulates and the advent and expansion of industrialization and urbanization continue to expose a greater portion of the population to these unacceptable conditions (French, 1990). Particulates, especially PM10 or the fraction less than 10 micrometres (J...lm) in diameter, pose the most significant health hazard because they can be inhaled into lung tissues and may interfere with lung functions. Prince George is highly susceptible to the accumulation ofPM10 due to local geography and meteorology, industrial activities in the city, and the severe winters which require significant application of sand to roads. To date, there is a lack of detailed information with respect to the characteristics and distribution ofPM10 in the Prince George airshed. Management ofPM10 has been identified by the Prince George Airshed Technical Management Committee as the first air quality management priority due to the high frequencies of unacceptable ambient air quality levels and current epidemiological studies indicating serious health impacts ofPM10 (PGATMC, 1996). The Northern Interior Health Unit (which includes Prince George) ranks nineteenth out of the twenty regions in B.C. for death rates, respiratory disease, and socioeconomic characteristics (PGATMC, 1996). The only air pollution health study to date in Prince George was a two part study completed in 1986 and 1991 and examined links 1 between total reduced sulphur (TRS), total suspended particulate (TSP) and respiratory disease. To date, no studies have focussed on the characterization and effects ofPMw on health in the Prince George region. The 1996 Draft Air Quality Management Plan for Prince George recommends studies to would identify the composition and sources ofPMw to aid in prioritizing reduction strategies. Comparisons of annual average ambient PMw levels between 1992 and 1996 show monitoring sites in Prince George 3 3 3 rank third (Plaza 261J.g/m ) , fourth (Van Bien 251J.g/m ) , and tenth (Gladstone 191J.g/m ) out of sixteen Canadian centers (Sutherland,1998). In 1995, at the British Columbia Railroad (BCR) site, the level B 3 3 -24 hour objective of 50 IJ.g/m was exceeded over 30% of the time (mean 411J.g/m ) ; at the Plaza it 3 was exceeded 10% of the time (mean 261J.g/m ) and in College Heights it was exceeded 3% of the time 3 (mean 171J.g/m ) (MELP,1997). The BCR site is an industrial park with extensive road system (paved and unpaved), beehive burners, sawmills, train tracks I traffic and various other industries. Between 1993-1995 the level A objective was exceeded an average of more than five weeks per year in Prince George (MELP,1997; MELP, 1995). The PM10 concentrations in the interior of the province corresponded to more than 5 weeks of poor to very poor air during 1993-1995 (MELP,1997). Knowledge of both the sources, and effects of meteorology are also crucial in characterizing the local air pollution problem. The health impacts and high concentrations ofPMw have been shown to be significant enough to warrant a study of this nature in the Prince George airshed. Source apportionment of the PMw in the ambient air will rectify the current lack of knowledge about the sources in the Prince George Airshed. This thesis is intended to provide knowledge of the morphology and composition ofPMw in the Prince George airshed. Objectives of this study are to a) determine the physical (e.g. , particle size distribution) and chemical (e.g. , elemental contents) composition of the major PMw sources in the 2 Prince George Airshed, and b) to determine the contribution from these major sources to PMw concentrations during episodic and non-episodic events in the bowl and the British Columbia Railroad (BCR) areas in Prince George. The BCR site was examined separately due to the high frequency of non-compliance of the Level B Objective at this location. 3 CHAPTER 1: LITERATURE REVIEW Sources, Types and Composition of PMto Natural sources ofPM10 include geological, oceanic, forest fire, volcanic, and biological emanations (See Table 1). Primary geological materials (soil) are largely contributed during summer and fall (Chow et al. , 1992). The composition of these crustal materials varies due to the distinctive elements found in different locations (Chow et a/. , 1992;Schroeder eta/. , 1987). Oceanic or marine sources can form aerosols with trace amounts of metals and sulphur (Bridgman, 1990;Schroeder et a/. ,1987). Forest fires can be large contributors during the summertime while volcanoes tend to be an irregular and unpredictable (although quite large) source (Chow et al. , 1992;Schroeder et al. , 1987). Biological emanations from leaves, peas, coniferous trees, soils, and pollen also contribute to PM10 in the environment (Schroeder et al. , 1987). Most natural sources produce PM10 in the coarse particle size fraction from 2.5!J.m to10!J.m diameter (Chow et a/.,1992). Coarse particulate often has basic pH, and is formed by the mechanical breakup of materials (Dockery & Pope, 1994). This is especially true of soil and crustal PM10 (Chow et a/. , 1992). It is believed that due to size and chemical composition, natural sources do not have the same adverse health effects as anthropogenic sources (Vedal, 1996). The chemical constituents found in the natural sources mentioned in the literature are summarized in Table 1. The elements found in natural sources vary not only between different sources, for example crustal sources contain aluminum and silicon while marine sources contain sodium, but also between similar sources, for example soil from two areas in Prince George may have quite different compositions (Table 1). 4 (Kame et a/.,1992) Na, Fe, Mn, Pb, V, Zn, Cu, Ba, La, N03-, Organic C/Elemental C, Cl, Ti, Ni, Sr, Zr,Pd, Ag, Sn, Sb, Al, Si, K, Ca, soi- (Bridgman, l990;Kowalczyk et a/. ,1982;Schroeder eta/., 1987;Chow, l995) Al, Si, S, K, Ca, Fe, Ti, Cr, Mn, Ni, Zn, Li, Mg, P, Sc, Sn, Zr, Nb, Cs, As, Ba, Cl, Na, OC/EC (Kowalczyk et a/. ,1982;Xhoffer et a/. ,1991;Pierson & Brachaczek, 1983;Chow, 1995) Ocean/marine Soil dust fugitive dust, secondary ammonium nitrates, ammonia, limestone, N03-, NH/, Cr, Zn, Sr,soi-, Na, K, S, Cl, Mn, Ba, Ti, Al, Ca, Fe, Si, Organic C (Chow, et a/., 1992;Kowalczyk et a/. ,1982; Chow,1995) Cd, Cu, Mn, Ni, Pb, Zn (Kame et a/. ,1992) Ti (Xhoffer et a/., 1991) Ca, Mg (Kowalcyzk et a/. ,1982;Alpert & Hopke,1981) Ti, As, Mn, Fe, Zn, Pb, V, Cl, Ga, Se, Br, Rb, Cr, Zr, Cu, Ni, Co, P, K, Sr, Cd, Ba, Sb, Hg, OC/EC, Al, S, Ca, Si, NH/, N03-, S042(Kowalcyzk et a/., 1982;Xhoffer et a/., 1991 ; Schroeder et a/. ,1987;Chow,1995) fugitive soil, limestone, Cr, Mn, Zn, Sr, Ba, so/-, K, S, Ti, AI, Ca, Fe, Organic C, Si (Kowalcyzk et a/. ,1982; Chow,1995) S, V, Ni, Cr, K, Organic C/Elemental C, Cl, Ti, Co, Ga, Zn, Se, Na, Fe, Si, S02, NH/, N03-, S04' (Kowalcyzk et a/.,1982 ;Lowenthal & Rahn,1987; Kartal et a/. ,1993 ;Pierson & Brachaczek,l983 ; Chow et a/., 1992;Xhoffer et a/. ,1991;Chow,1995) Agriculture Anthropogenic Asphalt production Cement plants Coal fired boilers Construction projects Crude/residual oil combustion ANTHOPOGENIC SOURCE Cd Forest fires (Schroeder et a/. ,1987) Al, Ca, Fe, Si Zn, Hg, V, Ni, Cu, Cr, As, Pb, Mn, Fe, Co, Cd, Sb, volatile exudates, alkyl arsenic (Schroeder et a/. ,1987) Biological emanations Crustal ELEMENTS/COMPOUNDS NATURAL SOURCE TABLE 1 :Composition of Natural and Anthropogenic Sources ofParticulates 5 I ! Major: Al, Ca, Fe, K, S, Si Minor: As, Cr, Mn, Ni, Ti, Zn (Xhoffer et al., 1991 ;Alpert & Hopke, 1981) As, Cd, Cr, Pb, V, Zn (Schroeder eta/., 1987) Fly ash High temperature combustion Cu, V, Mn, Sb, Cr, Ti, Cd, Zn, Mg, Na, S02, Ca, K, Se, As, Pb, S (Kartal eta/., 1993 ;Harley et AI, Si, K, Ca, Ti, Mn, Fe (Chow et a/. ,1992;Chow,1995) Ti (Xhoffer et a/. ,1991;Aipert & Hopke, 1981) Ti, S, Ca, Fe, Zn, Pb, V, Mn, Cr, Cu, Ni, As, Co, Cd, Sb, Hg, Se, Br, Ba, AI, Si, P, K, NH/, OC/EC, Ag, sol- (Xhoffer et a/., 1991 ;Schroeder et a/.,1987;Kartal et a/., 1993; Chow,1995) alkyl selenides (Lowenthal & Rahn, 1987) P, Ca, Mn, Fe, Zn, Rb, Pb, NH/, Na, soluble potassium, Organic C/Elemental C, Sol-, Nol-, Br, Cl (Chow et a/. , 1992;Xhoffer et a/., 1991 ;Chow,1995) Pb, B, Mg, P, Br, Sr, Co, Ba, Ni, Zn, Fe, As, AI,Cr, Y, Si, Ca, S, Mn, NH/, N03-, sol -, methyl cyclopentadienyl manganese tricarbonyl, alkanes, unburned/oxygenated hydrocarbons, P AH, benzo(a)pyrene, Cl, 1,2dichloroethane, N-nitrosomorpholine (Chow et a/.,1992 ;OECD, 1995; Greenburg et a/. ,1993;Pierson & Brachaczek,1983 ;Harnilton et a/. , 1994; Williams et a/. ,1989b; Lowenthal & Rahn,1987;Chow,1995;Kowalczyk et a/. ,1982) Non-ferrous smelters Paved road dust Pigment spray Power plants Soil/sewage sludge Vegetative burning Vehicle emissions al., 1989;Chow, 1995) Fe, Zn, Cr, Cu, Mn, Ni, Pb (Schroeder et a/., 1987) Open hearth furnaces Zn, Cl, K, Ni, Ag, Sb, Fe, Hg, Pb, Ti, As, Cd, Co, Cu, Mn, V, Sn, AI, N03, Na, EC, Si, S, Ca, Br, La, S042-, NH/, Organic C (Schroeder et a/., 1987;Aipert & Hopke,1981 ;Kowalczyk et a/., 1982;Chow, 1995) K, organic carbon, retene, Zn (Chow et a/. ,1992;Lewis et a/.,1988) Fireplaces/wood smoke Incineration Fe (Xhoffer et a/. ,1991) ELEMENTS/COMPOUNDS Ferrous metallurgy ANTHROPOGENIC SOURCE TABLE 1: Composition of Natural and Anthropogenic Sources of Particulates continued 6 Anthropogenic or "man-made" sources can account for a significant portion of the PMw produced (See Table 1). Such sources include stationary fuel combustion (agriculture, oil & gas production, refining, manufacturing, industrial, electric utilities, residential); waste burning (agricultural debris, range/forest management, incineration); petroleum processing (storage/transfer, oil & gas extraction, petroleum refining); industrial processes (chemical, food, agricultural, mineral/metal processing, wood and paper industries, cement plants); miscellaneous processes (farming, construction, demolition, road dust, unplanned fires); mobile sources (passenger vehicles, heavy duty gas & diesel trucks, motorcycles, buses, trains, ships, aircraft) (Alpert & Hopke, 1981 ;Chow et a/., 1992). Anthropogenic sources tend to contribute finer PM10 (2.5J...Lm or less) than natural sources. These smaller particles tend to be acidic, for example soot particles or acid condensate aerosols (Dockery & Pope, 1994). Due to size and composition, this portion ofPM10 is the most hazardous to health (Vedal,1996). One example ofthis is vehicle exhaust. Eighty six percent ofthe particles emitted from diesel engines have an aerodynamic diameter ofless than 1J...Lm (Williams et a/., 1989a;1989b). There are two types of particles emitted from PM10 sources: primary and secondary particles. Primary particles undergo few changes in the atmosphere between sources and receptors (monitors) and the ambient concentration tends to be proportional to the quantities emitted (Chow eta/. , 1992). Secondary particles are formed through chemical conversions (gases to aerosols) in the environment and tend to produce fine aerosols (less than 0.1J...Lm- 2J...Lm) (Bridgman, 1990;Chow et a/.,1992). Aerosols are defined as small solid and liquid material that remains suspended for a period of time (Bridgman, 1990). Secondary aerosols can be transported over long distances affecting air quality and climate outside ofthe local (generating) area (Bridgman,1990). As the aerosols are transported, they often undergo interactions and coagulation to form particulates unique to the original source (Post & Bused<, 1984). It is believed that sulphates, nitrates, organic carbon compounds and acid aerosols mal 100J..lg/m3 (MELP,1997). The 12 annual mean concentrations ofPMw in the province ofBritish Columbia range from 15J...Lg/m3 to greater than 50J...Lg/m3 (MELP, 1997) The pollutants produced in the Prince George airshed are often concentrated and recirculated. The city contains numerous sources ofPMw which produce and emit particulates within the river valley. The particulates are often re-circulated in the "bowl area" contributing to the buildup of particulates. Often inversions occur covering the bowl area, generally caused by a warm air mass overlying cold or denser air. This decreases diffusion of the cold air containing PMw and forces it to remain stagnant. The longer the air is trapped, the higher the particulate levels become as the sources continue to produce and emit more PMw. When inversions occur for extended periods of time, the likelihood that pollution advisories will occur increases. 13 CHAPTER 2: MATERIALS & METHODS PMto Source Samples The three types of road dust sources included in the study were street sweepings, snow removal particulates, and unpaved road dust. Street sweepings were collected from a large pile next to the City works yard on 4th avenue, 2 hours after deposition in March 1997. Three 75-litre plastic pails full of materials were removed from five locations in the pile for chemical and physical analyses. All plastic pails used in this and subsequent procedures had been washed with distilled water and LiquinoxTM, and acid washed with 10% Hydrochloric acid previous to sampling. The street sweepings samples provided information on the contribution of the paved road dust to the composition ofPMto. Snow removal samples were collected at Carrie Jane Grey Park to provide information on the contribution of winter sands to the composition ofPMto. Materials were removed from several sections of one pile of melting snow containing winter sanding materials into three 75 litre plastic pails. Three 75 litre plastic pails, full ofunpaved road samples were collected from several locations on Northern Crescent and Willowcale Forest roads in the BCR site using a shovel to study the contribution from unpaved road dust to the composition ofPMto. Other sources ofPMto in the Prince George airshed are pulp mill emissions and beehive burners. A sample of Total Suspended Particulates (TSP) was provided by Canadian Forest Products Prince George Pulp mills. This sample was removed from the power boiler stack which produces the majority of the particulate matter emitted by the pulp mill. The beehive burner sample was obtained from an undisclosed site in the Central Interior ofBritish Columbia. 14 PMto from Ambient Samples The ambient PM10 samples from three locations in the bowl area and from the BCR site were provided by B.C. Ministry of the Environment Prince George Region (MELP) (Table 2). The three sampling locations in the bowl area were Plaza 400, Lakewood, and Van Bien (Figure 1). As discussed in the introduction, the BCR site was analyzed separately to determine the sources responsible for the frequent non-compliance of the 24 hour level A objective ofPM10 present at this location. The samples were collected on teflon coated borosilicate glass fiber filters which are used by the MELP for routine total particulate - PM10 Hi-volume monitoring (BC Environment, 1997). The ambient PM10 concentrations reported are based on the weight ofPM10 sampled in micrograms divided by the volume of air passed through the filter during the 24 hour sampling period in cubic meters. Three episodes (with average 24 hour ambient PM10 concentrations above 50Jlg/m3) and three non-episodes (with average 24 hour ambient PM10 concentrations below 50Jlg/m3) were chosen to represent unacceptable and acceptable PM10 levels, respectively (Table 2). The three episodes were the three highest PM10 episodes occurring between 1994 and 1997. The three non-episodes were chosen to represent good and fair air quality according to established criteria. Filter samples taken before 1994 were unavailable for analysis as they had been destroyed. The meteorological conditions on each date examined are summarized in Appendix A. Seven episodes and two non-episodes were chosen to represent the BCR sampling site. The episodes represented poor and very poor air quality and the non-episodes represented fair air quality. 15 TABLE 2: Location and Description of Ambient & Source Samples Van Bien Ambient Plaza 400 . #1 - January 21,1995 Bowl 54 f.Lg/m 3.. 60 f.Lg/m Lakewood 3 57 f.Lg/m 3 51 f.Lg/m 3 35 f.Lg/m 3 50 f.Lg/m 3 13 f.Lg/m 3 #2 - March 28,1995 Episodic 85 f.Lg/m 3 106 f.Lg/m 3 #3 - February 27,1996 61 f.Lg/m 3 63 f.Lg/m 3 #1 - January 22,1996 43 f.Lg/m 3 40 f.Lg/m Bowl Non - episodic 3 #2 - March 4,1996 32 f.Lg/m 3 44 f.Lg/m 3 #3- May 9,1996 BCR site #1 - April 8,1994 3 143 f.Lg/m #2- September 23,1994 3 110 f.Lg/m #3 - February 16,1995 3 139 f.Lg/m Episodic #4 - March 28,1995 3 181 f.Lg/m #5 -August 31,1995 3 104 f.Lg/m #6 - August 13,1996 3 101 f.Lg/m #7 - March 4,1996 3 85 f.Lg/m BCR site Non - episodic #1 - January 22,1996 3 47 f.LQ/m #2- May 9,1996 3 32 f.Lg/m Sources Road Dust (1) Street Sweepings (2) Snow Removed from City Streets (3) Unpaved Roads in BCR site Beehive Burner Undisclosed Site Pulp Mill Canadian Forest Products - Prince George Dates: * Date of collection by MELP; **Concentration ofPM10 collected over that 24 hour period. Advisories occurred March 29- April1 , 1995 & February 28- March 2, 1996. 16 Figure L Sampling Locations. fur PMto- in the Prince George Airsbed. • 1ur Monitoring sru: • 17 PMto Collection from Source samples PM10 from road dust samples were extracted using particle size analyses. The road dust samples were placed in a 2mm sieve and washed with de-ionized water to separate the materials into coarse fragments (>2mm) and the sand/silt/clay portion (<2mm). Materials smaller than 2mm were then passed through a 53~-tm sieve to separate the sand (>0.05mm) from the silt/clay (<0.05mm). The clay and silt portion was placed in 2L glass beakers and dried in an oven at 105°C overnight in order to concentrate the sample. The concentrated sample was gradually transferred to one 2L beaker which was topped up with de-ionized water. In order to separate the inhalable particulates ( Mill PMlO M ean % SD Mean % SD 1.12b 2.12b 0.020 H(5 ,n=18)=13 .35, p=0.0203 0.021 Aluminum 3 3 0.144 0.0 19 H(5 ,n=18)=14.99, p=0.0104 0.033 0.001 Barium 0.295 H(5 ,n=18)=9.95, p=0.0768 12.830 0.455 18.698 Calcium nd• 0.00025b H(5, n=18)=14.39, p=O.Ol33 nd 0.000 Cadmium 0.035b 0.246c H(5 ,n=18)=12.78, p=0.0255 0.005 0.003 Chromium 0.012b 0.026d H(5 ,n= 18)= 16.39, p=0.0058 0.000 0.000 Copper H(5 ,n= 18)= 10.23 , p=0.0690 1.300 0.056 3.170 0.030 Iron 0.00 16c 0.0016c H(5 ,n= 18)= 15.36, p=0.0089 0.000 0.000 Lithium 1.99ab 4.57c 0.033 0.056 H(5,n=18)=11.95, p=0.0355 M agnesium 0.99d 2.20° 0.020 0.020 H(5 ,n= 18)= 16.23 , p=0.0062 Manganese 0.016b 0.169c H(5 ,n= 18)= 15.13, p=0.0098 0.004 0.001 Nickel 1.46d 0.714c 0.009 0.015 H(5 ,n= 18)= 15.27, p=0.0093 Phosphorus 3.50b 0.230 0.027 H(5,n=18)=13.09, p=0.0225 1.06 3 Potassium 2.51d 0.3 5° 0.043 H(5,n= 18)= 16.25, p=0.0062 0.021 Sodium 0.044 0.001 H(5 ,n=18)=9.52, p=0.0902 Strontium 0.038 0.001 0.0169b 0.0076 3 0.002 0.001 H(5 ,n=18)=13 .63, p=0.0181 Tin 0.065c 0.089c H(5 ,n= 18)= 15.83, p=0.0074 Titanium 0.001 0.001 3 0.0034 0.003 0.0057 3 0.000 H(5 ,n=18)=13 .23 , p=0.0213 Vanadium 0.077d 0.222d Zinc 0.00 1 0.003 H(5 ,n=18)=16 .39, p=0.0058 nd nd 0.00 1 0.000 H(5 ,n=18)=9.75, p=0.0827 Zirconium Rows mean superscripted with different letters are significantly different (p<0.05); Each source (n=3) SD = Standard Deviation; nd = not detected 32 Figure 3: Particle Size Distribution: Street Sweepings 26 24.0% 24 t-··r--...,.-----------------··-··-··-··-··-··-·--·--------- 22 1--·· ------·-··-·-··-··-··-··-··-··-··-··-·-··-·-··- 20 1--·· 18 Ill Cll 0 :;::::;- 'f 6 --- --· 18.0% ·-·--·..,.,.,.vM"lcJl:O :--r---r·· ·-··-··-··-··-··-··-··-·-·-·-··- .... ctJ 14 -·· a.: ....·CII0 '. 12 --.... ..c 10 E z .. --- --6 --4 8 :::1 8.0% ---------------------·-1: t - -...... 6.0% f--··-··-,.m o-·-r----r·------- ------- 4.0% 2 : -·· 0 <= 2.5 (2:5,3) (3,4] (4,5] (5,6] (6,7] (7,8] (8,9) (9,10) > 10 AverageDiam.eter (micrometers) Figure 4: Particle Size Distribution: Snow Rem.oval 33 --------------------------------------------------31 .0% 30 - - 24 t--· ------------------------------------------------------------------------------------------------------------------------------------- Ill Cll 21 t--· ------~ ctJ 18 t--· ----- ~ .a. '. 0... 15 t--· il 12 1---· Cll :::1 z 9 1---· t-------------------------------------- ------------- 1-·------------------------------1- ----~-------------------------- 12.0% ~ 7.0% t----------------------- 6 1---- --------------------- 3 t--· - 1 t- 0 <=2.5 (2.5,3] (3,4) (4;5) (5,6] (6,7) (7,8] ~-- ~(8,9) (9,10) I 3.0% I > 10 Average Diameter (rriicrometers) 33 Figure 5: Particle Size Distribution: Unpaved Roads 33 -~ --------------------------------------------------- --27 --24 . --0 21 --1::. ra a. 18 --.... 15 --E 12 --:z 9 --6 . ---30 ------------------------------------------------ II) Cl) 0 ------ 18.4% --------------------------------------- ·-tSft-·-------------------------------- 14.6% Cl) ' ..Q ::J · 3 0 -~ ---------------------------- t---.-----,r----------------- . 5.8% --- 5.8% ~--------1 <='2.5 (2.5,3] (3,4] (4;5} (5;6] '· .. (6,7] (7,8] {8,9] (9,10} 2.9% I -- > 10 Average Diameter (micrometers) The quantitative values of barium, zinc, and sodium are higher and the iron is lower in the Prince George PM10 samples compared to those in the literature (Chow et a/., 1992; Chow, l995). This may be a result of variations in the amounts of these elements present in the crustal materials in Prince George. There is some uncertainty with these results due to problems with the filter blank (see Methods section). The qualitative chemical compositions (±10%) are consistent with the reported literature for carbon, chlorine, potassium, magnesium, and titanium (Chow eta/., 1992; Chow, 1995)(Table 7). There is however, more aluminum, sodium and silicon, and less calcium and iron than found in the reported literature (Chow eta/. , 1992; Chow, 1995). This could again be a result of natural variation or the large confidence intervals ofthe EDAX analysis. 34 Mean% 1.68. 1.41 ab 1.73. 0.95b SD 2.02 2.24 2.33 1.21 Mean% 4.01 3.45 4.46 2.96 SD 7.30 6.35 6.16 5.42 Mean% 16.42. 11.37b 12.28b 15.768 SD 11.38 11.77 11.45 11.91 SD 1.05 0.87 1.32 8.02 Mean% 48.28. 56.26b 54.10b 44.40. CI Mean% 0.02 nd nd nd Across rows, means superscripted with different letters are significantly different (p<0.05); Each source (n=100) Confidence Intervals for EDAX (±10%); ANOVA Results in Appendix I SD = Standard Deviation; nd = not detected Street Sweepings Snow Removal Unpaved Beehive Burner I TABLE 7: EDAX Qualitative Chemical Characterization of PMto Sources Ca Sources c AI Mean% SD Mean% SD Mean% 0.42. 18.72. 10.31 a 8.37 8.73 Street Sweepings 0.31. 21.16. 5.64b 4.72 10.20 Snow Removal 5.68b 0.61. 20.97. 9.36 6.08 Unpaved 1.92b 12.98b 19.93c 16.11 9.59 Beehive Burner SD 13 .98 16.77 13 .51 19.77 SD 0.18 nd nd nd Mean% 0.15 0.14 0.06 0.43 Fe Mean% nd 0.27 0.12 0.69 SD 1.02 1.10 0.42 2.71 SD nd 1.25 0.69 6.72 35 The quantity of road dust produced should be considered between the different types of road dusts. A distance of 1.6 kilometers (1mile) of travel on a paved road produced 0.0045kg (0.01 pounds) of dust while 1.6 kilometers (1 mile) of travel on an unpaved road generated 4.5kg (10 pounds) of dust (1000 times the dust) (Evans & Copper,1980). Beehive Burner The beehive burner sample analyzed contained three main morphological shapes: amorphous, oval, and flat (Table 4 & Appendix B). Amorphous or irregular shaped particles are often the result of combustion processes (ie fly ash) (Kautherr & Lichtman, 1984). The oval shape morphology identified has not been reported in the literature, however distinctive rounded or spherical shape particulates are produced by anthropogenic or combustion sources as a result of formation at high temperatures (Kautherr & Lichtman, 1984; Purghart et al., 1990; Xhoffer et a/., 1991). Spherical particles in combustion products such as fly ash indicate a complete melting of silicate materials (Fisher eta!. , 1978). The unique oval shaped particulates are most probably a result of the high temperature combustion which forms fine particulates (many are probably secondary particulates) through chemical conversions and condensation, however, this shape is not mentioned in the literature. The flat shaped morphology may either be a result of incomplete wood combustion or dust contamination, it is unclear which is responsible. The literature indicates that 50- 80% of total wood burning particulates are fine particulates which is consistent with the results (Figure 6) (Stevens, 1985;CHU, 1994). The larger particulates present in this sample may be the result of incomplete combustion or dust contamination (Dockery & Pope, 1994). The mean particle size varies significantly between burners depending on conditions and type of materials being burned (Boubel, 1968). One study 36 found a mean particle size of3 ~m which is consistent with the results in this study (Table 5) (Boubel, 1968). Figtlre 6: Particle Size Distributiort: Beehive Burlier 52 48 44 40 l/1 Q) 36 (j :e., 32 ....0 0.. 28 ... 24 . Q) .c 20 E ::::r z I ~ ··-··-··-··-··-··-··-··-·-···-··-··-·· 16 1~ 2" - ~--- - - - - - - - - - 8 4 0 <= 2.5 (2.5,31 (3,4] (4,5] .Average (5,6] (6,7] (7,8] m~t (micrometers) (8,9] (9;10] The quantitative and qualitative chemical compositions of the beehive burner sample was compared to vegetative burning as there was no published information on beehive burners in the literature. All the elements in the quantitative analysis except potassium were found in higher concentrations which could be a result of higher combustion temperatures or difference in material type (Table 6). The qualitative chemical compositions were also larger for most elements (Table 7) (Chow, 1995). The amount of carbon and iron were consistent with the published literature while chlorine and potassium were lower than expected (Table 7) (Chow, 1995). Pulp Mill The literature search conducted found no published literature on either chemical or physical characteristics of pulp mill particulates. Many ofthe elements were concentrated in the 37 PMw fraction of the sample including aluminum, barium, cadmium, chromium, copper, iron, magnesium, manganese, nickel, phosphorus, tin, vanadium, and zinc (Table 6). This concentration of elements seems to be consistent with the theory that trace metals tend to condense on the surfaces of fine particulates (Keyser et al. , 1978). Comparison of Source Samples Morphological and Particle Size Characteristics Examination of morphological characteristics confirm that the three types of road dust are different. The unpaved road dust sample contained more particulates with a flat morphology (47% compared to 21% in street sweepings and 3 5% in snow removal) which suggests a greater concentration of clay particulates in unpaved roads (Table 4). This is understandable considering the area sampled (BCR site) contains a high (60-70%) clay content (Pineview Clay) deposited while most of Prince George was under a glacial lake (Dawson, 1989). The snow removal materials also contained a larger proportion of particulates with flat morphology (3 5% compared to 21% in street sweepings) suggesting that materials removed from the roads during the winter may have contained more clays (Table 4). Amorphous morphology was more evident in the street sweeping sample (79% compared to 61% in snow removal and 54% in unpaved road dust) suggesting that mechanical breakdown of larger materials is more important on paved city streets in the spring (Table 4). The morphological characteristics indicate a significant difference between the beehive burner sample and the road dust samples. The appearance of the oval type accounts for 18% of the particulates from the beehive burner (Table 4). This sample also contains fewer amorphous type particulates (32%) compared to 79%, 61%, 54% in the street sweeping, snow removal, and 38 unpaved road dust samples respectively (Table 4). The larger percentage of flat particulates (48%) in the beehive burner compared to 21%, 3 5% in the road dust is possibly a result of incomplete combustion processes rather than clay particulates. However, the unpaved road dust contained a comparable amount of flat particulates suggesting that this shape would not necessarily be useful in distinguishing between road dust and beehive burner sources. The mean particle size and particle size distribution also illustrate that the road dust samples differ. The unpaved road dust is significantly smaller compared to the street sweepings (Table 5). There is no significant difference between the street sweepings and snow removal road dusts (Table 5). The unpaved road dust appears to contain more clay particles than the other road dusts which have smaller particle size and are flat shaped. Comparison of the particle size distributions in Figures 3-5 show a 6% increase in the fine particulate (<2.5!-!m) in the unpaved road dust compared to the street sweepings. The less positively skewed distribution in the road dust samples illustrates the presence of the larger "amorphous" particulates formed through mechanical breakdown oflarger particles (Chow, 1995). The mean particle size indicates a significant decrease between the beehive burner sample and any of the road dust samples (Table 5). This trend is illustrated in the particle size distribution (Figures 3-6) which indicate an increase of almost 20% in fine particulates in the beehive burner sample. This trend was expected due to the combustion nature of the beehive burner source. Chemical Composition There are significant correlation between elemental concentrations and average particle diameter (Table 8). However, it should be noted that the correlation coefficient values are indicating a very weak correlation (Mendenhall & Beaver, 1991). In the street sweepings, aluminum (r=0.22) is found to increase in concentration as particle size increases and sodium (r=- 39 0.21) is found to decrease in concentration as particle size increases (Table 8). In the snow removal sample, carbon (r=-0.25) seemed to be found in higher concentrations in the smaller particulates (Table 8). In the beehive burner sample, there are elements which have higher concentrations in smaller particulates ie carbon (r=-0.28) and sodium (r=-0.44) and elements that have higher concentrations in larger particulates ie magnesium (r=0.56), aluminum (r=0.20) and calcium (r=0.27) (Table 8). The carbon and sodium concentration in smaller particulates may represent small "secondary" carbon particulates (Chow, 1995). The magnesium, aluminum, and calcium concentration in larger particulates may be representative of dust particulates or perhaps incomplete combustion products (Chow, 1995). TABLE 8: Significant Correlation between Elemental Composition and Particle Diameter in PMto Sources AI(%)= 15.375 + 0.71473*Mean Diameter 20.864- 0.9510* Mean Diameter Calcium Carbon -0.28 0.56 -0.44 Ca (%)= -0.3471 + 0.72281* Mean Diameter C (%)= 23 .908- 1.451 *Mean Diameter Mg (%)= -0.2228 + 1.0175* Mean Diameter Na (%)= 21.348- 1.766*Mean Diameter There were significant differences between the contents of several elements in the three road dust samples (Table 6). There was significantly more manganese, phosphorus, and titanium, in the snow removal road dust than in the street sweepings (Table 6). These results suggest that these elements are most likely found in much greater concentrations in the winter sanding materials used by the city. The street sweepings did have significantly more zinc than the snow removal road dust, however, the concentrations of many other elements were consistent as was expected. The unpaved road dust had significantly more manganese and sodium than the other road dust samples suggesting that these elements are naturally present in higher concentrations in 40 the unpaved roads (Table 6). The literature suggests that the dominant elements composing road dust are aluminum, silicon, calcium, potassium, titanium, and iron which are also dominant in soil (See Table 1) (Chow, 1995). These elements were found in the road dust samples, however, iron and titanium were not found in the quantities expected. Many researchers use aluminum and silicon as tracer elements for dust sources, however the results for silicon could not be compared due to problems with the filter type, methodology and the percentage of aluminum was not significantly different between the road dust samples and the beehive burner samples (Table 6) (Chow, 1995). The beehive burner sample contained significantly more barium, lithium, manganese, potassium, sodium, tin, and zinc and significantly less copper than any of the road dust samples (Table 6). The higher levels of potassium are consistent with other studies which often use soluble potassium as a tracer element for wood combustion sources (Table 1) (Stevens, 1985; Chow,l995). The dominance of organic and elemental carbon in vegetative burning PMw can be used as an indication of combustion sources such as beehive burners. However, the ICP-AES is not able to analyze for this element (Chow,l995). Compared to the road dust and beehive burner samples, the pulp mill PM10 samples contained significantly more cadmium, chromium, copper, magnesium, manganese, and phosphorus suggesting that these elements are concentrated in the pulp mill processes. These elements may be useful in determining the pulp mill contribution to the PMw, however, the lack of information in the literature makes any comparisons impossible. The uncertainty caused by the problems with the blanks may be masking differences between the road dusts, beehive burner, and pulp mill samples and this uncertainty was considered when conclusions were drawn with this data. 41 The qualitative EDAX chemical composition means (±10%) showed some significant differences between the road dust samples (Table 7). The street sweepings contained significantly more carbon and sodium and significantly less silicon than the other two types of road dusts (Table 7). This could be due to addition of carbon by vehicle exhaust or deposits on the pavement and the addition of sodium by salt applied to paved roads. The beehive burner sample contained significantly more carbon and calcium and significantly less aluminum which would be expected from a combustion source (Tables 1&7) (Chow, 1995). Although fewer elements were analyzed and the method was qualitative with this technique, the results were considered more reliable than the quantitative analysis. The teflon blank contributions were included in Table 7 to indicate additional possible contributions to the recognized confidence intervals of ±1 0%. The results ofPCA for each of the three road dust samples indicated four factors or particle types present. The first factor in the street sweepings (accounting for 22.24% of the variance) and the second factor in the snow removal sample (17.62%) showed large loadings on silicon with corresponding negative loadings on aluminum and sodium (Table 9). The high loading on silicon suggests that this factor most likely represents "Quartz" or silicon dioxide. The second factor in the street sweeping (17.59%) and the first factor in the snow removal sample (22.62%) contained large loadings on aluminum and potassium with corresponding negative loadings on sodium which are elements present in minerals called "K-Feldspars"(Table 9&10) (Brady,1996). The third factor on the street sweeping (15 .35%) and the unpaved road dust (17.82%) also contained loadings on calcium and magnesium and corresponding negative loadings on sodium which are present in "Ca-Feldspars" (Table 9 &11) (Brady, 1996). The fourth factor in the street sweepings (11.31 %) contained loadings on calcium and chlorine which represents calcium chloride (Table 9). The third factor on the snow removal sample (15.49%) containing loadings on 42 calcium, carbon, and magnesium could represent either "Ca-Feldspar or Calcium Carbonate" it is unclear which, and the four factor on the snow removal (12.45%) containing high loadings on iron and titanium represents a "Clay mineral or Iron oxide"(Table 10) (Brady, 1996). The first factor on the unpaved road dust (26.35%) has high loadings on calcium, iron, potassium, and titanium representing "K-Feldspars or Iron oxide" and the second factor (18 .63%) with high loadings on silicon and corresponding negative loadings on carbon and sodium represents "Quartz" (Table 11) (Brady, 1996). The last factor (13 .22%) on the unpaved road dust contains high loadings on aluminum and potassium and corresponding negative loadings on silicon and probably represents "K-Feldspars"(Table 11) (Brady, 1996). The PCA analysis completed on the beehive burner determined three factors or three type of particulates present in the sample. The first factor (accounting for 25 .77% of the variance) contained high loadings on carbon and sodium and a corresponding negative loading on silicon and represents the expected organic carbon particulate (Table 12). The second factor (18 .69%) contains high loadings on calcium and magnesium and corresponding negative loadings on sodium as well as the third factor (14.57%) containing loadings on aluminum, magnesium, potassium and negative loadings on carbon could be dust contaminants and could not be interpreted (Table 12). The factors determined using the PCA were used as examples of possible relationships between the different variables/elements which may be characteristic of specific sources. There were several relationships seen in the road dust samples that were used in the determination of factors in the episodes and non-episodes. High loadings on silicon, aluminum, magnesium, potassium, iron, titanium in various combinations were considered to be representative of road dusts. High loadings on carbon and sodium were considered to be representative of combustion sources or a beehive burner. It was recognized that further resolution of the beehive burner 43 sample would require more organic carbon analysis as often carbon was found negatively related to the elements representing road dust. In these instances the highest loading present in the factor was considered to be most important and the relationships of the other elements in relation to that element determined the interpretation ofthe factor. Aluminum Calcium Carbon Chlorine Magnesium Potassium Silicon Sodium Titanium ramary Fac t ors: St ree t S m~ ues an dP' 2 3 1 Quartz K-Feldspar Ca-Feldspar 0.235853 -0.385746 -0.727238 0.02104 0.216242 0.434242 -0.254511 0.066238 0.013045 -0.0179 0.021926 0.020204 -0.169 0.045959 0.87634 -0.153556 0.178417 -0.780345 -0.155087 0.925532 0.095762 -0.58962 0.42697 -0.592113 0.075129 0.319792 0.261967 4 CaCh 0.273748 -0.402602 -0.172094 -0.794484 -0.000642 0.052515 0.294976 -0.000233 0.262254 Eigenvalue 2.001901 1.58271 22.24 22.24 17.59 1.381198 15.35 1.017577 % Total Variance 39.83 55 .18 TABLE 9: PCA E" ~ Factor Cumulative % 11.31 66.48 Numbers m bold mdtcate the amount and pattern of elemental loadmgs. Loadmgs of more than 0.71 (50% overlapping variance) are excellent, above 0.63 (40% overlapping variance) are very good, above 0.55 (30% overlapping variance) are good, above 0.45 (20% overlapping variance) are fair and above 0.32 (10% overlapping variance) are poor (Comrey & Lee,1992). TABLE 10 PCA Eigenvalues and Primary Factors: Snow Removal Factor Aluminum Calcium Carbon Iron Magnesium Potassium Silicon Sodium Titanium 1 K-Feldspar -0.560605 -0.041711 0.226843 0.00188 -0.034599 -0.879112 0.100726 0.434801 0.05285 2 Quartz 0.646837 -0.017216 0.192275 -0.095712 0.031329 0.025729 -0.937193 0.673709 0.146133 3 Ca-Feldspar 0.036581 0.539144 0.584015 0.250099 0.822765 -0.092366 -0.310894 -0.299675 -0.174331 4 Iron Oxide 0.284251 0.023874 -0.124624 0.645165 0.179537 -0.14056 -0.06923 -0.316012 0.839523 Eigenvalue 2.035569 1.585771 1.394461 1.120288 % Total Variance 22.62 17.62 15.49 12.45 Cumulative % 22.62 40.24 55 .73 68.18 For explanation of numbers in bold please see Table 9 44 TABLE 11: PCA Eigenvalues and Primary Factors: Unpaved Road Dust 3 4 2 Factor 1 Quartz Ca-Feldspar K-Feldspar K-Feldspar -0.221656 0.135617 0.000808 -0.898927 Aluminum 0.111 947 -0.452729 0.037285 -0.674218 Calcium -0.058023 0.048052 0.22966 0.711692 Carbon -0.154332 -0.134 186 0.0640 16 -0.811085 Iron -0.085845 -0.893256 -0.01944 1 0.11226 Magnesium 0.126022 -0.171125 -0.616231 -0.526038 Potassium 0.047764 0.098539 -0.855528 0.474723 Silicon 0.081441 0.604642 0.219517 Sodium 0.599026 0.2 12483 -0.088563 -0.836736 0.043886 Titanium 1.6037 1.189796 2.371355 1.676352 Eigenvalue 17.82 13 .22 26.35 18.63 % Total Va riance Cumulative % 26.35 44.97 62.79 76.01 For explanation of numbers in bold please see Table 9 TABLE 12: PCA Eigenvalues and Primary Factors: Beehive Burner Factor 1 2 3 Organic Other Other -0.222555 -0.11 3832 -0.822882 Aluminum 0.13134 Calcium -0.823181 0.264934 0.038805 Carbon 0.769954 0.35759 0.054956 0.033764 0.06981 Iron 0.014023 Magnesium -0.785683 -0.381262 0.008147 0.055472 -0.824935 Potassium 0.232724 0.076628 Silicon -0.94565 Sodium 0.580893 0.537579 0.08566 -0.00064 0.008674 0.02259 Titanium 2.31896 1.682383 1.311348 Eigenvalue 25 .77 18.69 14.57 % Total Variance 25 .77 44.46 Cumulative % 59.03 For explanation of numbers in bold please see Table 9 45 Characterization of Episodes and Non-Episodes in the Bowl Area Episode 1 Amorphous particulates were found to be the dominant shape in this episode, while oval, sphere, smooth-flat, flat, rod, rectangular, and cube shaped particulates were found in much smaller numbers (Tables 13 & 14). The presence of70% amorphous particulates suggests that road dust may be an important contributor, however many other sources can contribute to amorphous particulates population including uncontrolled combustion sources so this is not diagnostic (Dockery & Pope, 1994). Comparison of morphological data between monitoring sites indicate slight differences between them (Table 13). The number of"oval" shaped particulates was slightly smaller at the Lakewood site compared to either Plaza or Van Bien suggesting that combustion sources may have had less of an impact at that site (Table 13). The absence of the "rectangular" particulate at the Plaza also suggests another source contributes to ambient levels at the other sites, however, the identity of the source of the "rectangular" particulates is unknown (Table 13). The particle size data shows no significant difference between the different monitoring stations. The total mean particle size indicates that anthropogenic combustion sources forming fine particulates were the most important contributing sources to this episode (Table 5 & 15). This is further illustrated by the particle size distribution which indicates that over 73% of the total particulates were fine particulates (Figure 7). The particle size distribution is highly positively skewed which is consistent with other studies (Kim et al. , 1987). As indicated in the previous discussion combustion sources such as beehive burners tend to contribute to the fine particulate fraction while road dusts can be distinguished by significant contributions to coarse particulates. 46 Qualitative chemical composition averages show some significant differences between the different locations analyzed, however, this data has to be considered with some caution due to the large standard deviations and confidence intervals involved. The three most important elements present in the episode were carbon, sodium, and silicon (Table 16). This suggests that road dust (silicon) and combustion sources (carbon) may be the largest contributors to the PMw (Chow, 1995). The Plaza location has significantly more sulphur, potassium and calcium and significantly less carbon than the Lakewood site suggesting that the Plaza location was impacted by the proximity of industrial sulphur sources (Table 16). Table 13: Distribution of Various Amorphous Episode Round Sphere /o Flat 0 % % 6.00 5.67 3.00 2.33 1.67 4.00 5.00 6.33 4.33 16.33 16.33 12.67 79.33 82.00 78.00 0.00 0.00 0.00 0.00 0.33 0.33 0.33 0.00 1.00 20.33 17.67 20.67 960227 plaza 960227 vanbien 960227 lakewood 77.33 84.33 81.33 0.33 0.00 0.00 0.00 0.33 0.33 0.00 0.33 1.00 22.00 15.00 17.00 Episode Smooth Flat Cube Rectangle % % 0 % 950121 plaza 950121 vanbien 950121 lakewood 0.00 0.00 0.33 0.00 0.00 0.33 0.00 3.33 1.33 0.00 0.00 0.00 950328 plaza 950328 vanbien 950328 lakewood 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.22 0.33 0.00 0.33 0.00 0.00 0.00 % 0 /o 950121 plaza 950121 vanbien 950121 lakewood 70.00 65 .67 74.33 950328 plaza 950328 vanbien 950328 lakewood 0.00 960227 plaza 0.00 960227 vanbien 0.00 960227 lakewood Episodes (n=300); Totals (n=900) /o Rod 47 Non-Episode 2 Mean SD 262.67b 14.57 0.67b 0.58 1.67 1.53 11.00 8.00 24b 7.00 ob 0.00 0.00 0.00 Non-Episode 1 Mean SD 212.67" 3.22 11" 3.46 9.33 5.69 12.00 2.00 48.67" 4.73 3" 2.00 3.33 3.06 10.82 0.00 0.00 0.58 Non-Episode 3 SD Mean 262.33b 0.58 l.Ob 0.00 2.33 1.53 7.67 3.06 25 .67b 3.06 ob 0.00 0.00 1.00 54.00 0.00 0.00 0.33 Episode 3 Mean SD 243 b 10.54 0.33b 0.58 0.67b 0.58 1.33b 1.53 18.02 7.67 4.11 7.41 8.93 1.01 0.33 3.3 9 Total Non-Episode SD Mean 245.89 26.01 5.38 4.22 4.77 4.44 4.82 10.22* 32.78 12.76 1.80 1.00 1.44 2.19 230.78 5.00 3. 11 6.11 * 52.67 0.44 0.11 1.67 Total Episode SD Mean Across superscript (abc) indicates significant differences between means within Episodes and Non-Episodes Column Superscript(******) indicates significant differences between means of Episodes and Non-Episodes Total means for above Episodes and Non-episodes (n=300) Round Sphere Flat Smooth Flat Cube Rectangle Rod Amorphous Oval Round Sphere Flat Rectangle Rod Amorphous Oval Round Sphere Flat Smooth Flat Cube Episode 2 Mean SD 239.33b 6.11 ob 0.00 0.67b 0.58 1.33b 1.53 58.67 4.93 0.00 0.00 0.00 0.00 0.00 0.00 Episode 1 Mean SD 210" 13 .00 14.67" 4.93 8.0" 3.61 15 .67" 3.06 45 .33 6.35 1.33 1.53 0.58 0.33 4.67 5.03 I ~ t tt )~~~~1 t tt t t F(1 , 16)=2.05, p=0.171212 F(1 , 16)=0.062, p=0.806402 F(1,16)=0.40, p=0.534195 F(1 , 16)=1.95, p=0.181791 F(1,16)=14.68, p=0.00147 F(1, 16)=1.73, p=0.206983 F(1, 16)=1.0, p=0.332195 F(1, 16)=0.27, p=0.609669 F(1, 16)=3 .93 , p=0.064868 ~~~ ~~~t~ F(2,6)=33.41 , p=0.000559 F(2,6)=25 .16, p=0.001209 F(2,6)=4.39, p=0.066972 F(2,6)=0 .60, p=0.579120 F(2,6)=21.20, p=0.001905 F(2,6)=6.75, p=0.029131 F(2,6)=2.55, p=O.l58075 AN OVA Results F(2,6)=9.28, p=0.014586 F(2,6)=25 .58, p=0.001156 F(2,6)=11.81, p=0.00832 F(2,6)=44.02, p=0.00026 F(2,6)=2.27, p=0.184694 F(2,6)=2.29, p=0.182832 F(2,6)=1.0, p=0.421875 F(2,6)=2.38, p=0.173714 ANOVA Results 48 SD 18.76 17.24 19.38 18.15 Ca Mean% 1.27 1.64" 1.91" 0.25b SD 4.24 4.64 5.33 1.58 3.39 4.53 1.61 3.36 4.34 6.56 2.55 2.3 8 CI Mean% nd nd nd nd 0.03 0.08" 0.01b ndb ANOVA Results SD nd nd nd nd 0.29 0.50 0.07 nd 0.59 0.99 0.15 0.15 4.34" 0.74 27 .22" 4.08 0.06 950121 plaza 36.45b 3.22b nd nd 4.59 950121 vanbien 3.79"b 36 .58b 2.60 nd nd 950121 lakewood 13 .29 .. 16.06 .. 1.05 7.67 nd nd 12.78 2 1.22 14.37" 8.23 nd nd 15.72 12.61 950328 plaza 14.39" 6.55 nd nd 14.43 10.03 0.72 950328 vanbien 11.10b 7.69 nd nd 14.99 1.22 18.03 950328 lakewood 12.44 ... 14.88 .. 0.03 7.68 O.Dl 0.19 12.43 1.31 3 2.22" 0.06 7.24 nd nd 15.57 13.96 960227 plaza 12.53 0.98b 0.01 nd nd 7.08 15.05 11.19 12.68 960227 vanbien 0.74b 0.01 0.34 11.95 8.63 0.02 14.03 960227 lakewood 12.10 I Superscript down columns (abc)/(* ** ***)indicate significant differences between means (p<0.05); Confidence Intervals (±10%); nd =not detected TABLE 16: Qualitative Chemical Characterization ofPM10 Episodes AI Ba Episodes c Mean% Mean% SD SD Mean% 3.78· 33 .4( 3.87 0.43 0.02 1 Superscript across rows and down columns (123/* **)indicate significant differences between means (p<0.05) Plaza, Van Bien, Lakewood samples (n=299 - 300); Total Sample (n=900) TABLE 15:Comparison of Particle Size by Location for Plaza 49 Mean °/o nd nd nd nd 0.06 0.19 nd nd 0.04 0.07 0.03 nd 0.19 O.ll 0.17 0.29 0.43 0.22 0.09 0.16 0.19" nd" 0.29"b nd nd nd nd 1.40 2.43 nd nd 0.67 SD 2.76 2.14 0.80 1.24 0.98 nd 1.90 1.28 0.85 1.15 1.69 0.61 0.38 8 1.29b 0.88 .. 8 6.28" 5.44" 2.18b 0.76 •• Mean% 4.64. 1.42 1.37 1.45 1.48 1.35 1.44 1.41 •• 2.32" 2.65" 1.74b 1.42•• 10.02 10.12 12.22 6.41 3.75 4.02 1.73 4.76 2.82 3.16 2.93 2.19 SD 2.65 2.35 3.33 2.01 1.50 1.58 1.43 1.50 1.40 1.34 1.30 1.55 SD 42.09" 44.87b 39.75" 42.28 .. 26.88" 21.94b 25.10" 42.24 •• Mean% 24.64. 3.: 3.89 3.64 3.04 3.10" 3.56" 2.31b 0.18 0.06 0.10 2.99 •• o.n· Mg Mean% 14.41 14.25 16.90 14.96 13.52 13.42 17.15 12.98 13 .32 12.83 12.34 15.36 SD 6.52 5.84 5.71 1.23 0.59 1.03 4.95 5.22 5.16 4.38 0.99 SD I 0.03 0.02 0.02 0.06 0.06 0.04 0.08 0.02 Mean% 0.05 0.14" ndb ndb nd 0.01 0.02 nd nd nd nd nd nd' Mn Mean% nd· 0. 1.39 nd nd 0.2c SD nd 0.19 0.22 nd nd nd nd nd nd SD 0.30 0.24 0.25 0. 1.31 a 22.73 1.00 40.03" 0.87 23.59 1.00" 41.42" 0.34 0.57 0.33b 45.40b 22.50 nd 0.83 Superscript down columns (abc)/(* ** ***)indicate significant differences between means (p<0.05); Confidence Intervals (±10%); nd =not detected ANOVA results for Table 16 in Appendix I 10.64 9.32 ll.41 11.01 10.92 10.28 ll.80 20.92" 20.29" 24.55b 22.94··· 950328 plaza 950328 vanbien 950328 lakewood 3 960227 plaza 960227 vanbien 960227 lakewood z 30.53" 27.96b 30 .17" 21.92 •• 950121 plaza 950121 vanbien 950121lakewood . 8.77 8.91 8.95 8.23 10.65 960227 plaza 960227 vanbien 9602271akewood SD d nd nd nd nd nd nd nd nd nd nd nd 2.49 nd nd Mean% nd 0.14 nd nd nd nd nd 950121 plaza 950121 vanbien 950121 lakewood 2 950328 plaza 950328 vanbien 950328 lakewood TABLE 16: Qualitative Chemical Characterization ofPMto Episodes cont. Episodes Cu Fe K Mean% SD Mean% SD Mean% 2.24. ~ 1.44 0.25 2.07 50 Figure 7: Particle Size Distribution: Episode 1 • 950121 675 630 585 - 540 495 ~ 450 - ~.... 405 - ~ 360 • o. 315 • ..8""' 270 • ·E ;:, 225 z - 180 . 135 • • :90 . • ~----~~---- aCL..... <= 2.5 - - ....iQ'JL__ 45 . - 1~ (2.5,3] (3,4] (4,5] - - - ~- - - - ~~1 ~1~ (5,6] {6,7] - --- - - - --- - - - - - ---~ 3~ ~ {7,8] {B,9J - ~ ~~~~~~ (9,10] > 10 Average Diameter (micrometers) The PCA analysis determined four factors (or PM10 sources) which accounted for 61.19% of the total variance in the sample (Table 17). The first factor (accounting for 24.57% of the total variance) contained extremely high loadings of calcium, potassium, and sulphur and a corresponding negative correlation to carbon (Table 17). This is clearly an indicator for an industrial factor due to the presence of sulphur and the fact that it is the largest factor is consistent with the much smaller mean particle size results. A study by Chow et a/.(1992) found that sources of sulphur dioxide were just as important to ambient PM10 as sources of primary materials such as dusts. The three Pulp mills and Husky oil refinery produce 94% of the sulphur dioxide emissions in the Prince George Airshed (PGATMC, 1996). These sources most probably account for this factor. The second factor (17.47%) contains high loadings of silicon, aluminum, and sodium and a corresponding negative correlation to carbon (Table 17). This is a road dust factor probably indicating "Na-Feldspars". The third (10.01 %) and fourth (9.14%) factors also represented "Iron 51 and Magnesium oxides" in road dust (Table 17). A PCA was performed on each location and the results were similar to those found above (Appendix G) . The only difference of interest was that the Lakewood site was impacted more by road dust than by the industrial source. TABLE 17· PCA Eigenvalues and Primary Factors: Episode 1-950121 2 3 4 Factor 1 Road Dust Road Dust Road Dust Industrial Mg Oxide Sulphur Source Na-Feldspars Iron Oxide 0.067686 0.670223 0.332950 0.195519 Aluminum 0.114253 -0.052970 -0.036382 0.009692 Barium -0.020303 -0.192732 0.133028 -0.870979 Calcium -0.086391 -0.071784 -0.835988 0.511455 Carbon -0.137072 0.170480 -0.008943 0.717078 Copper -0.011859 0.065998 0.026635 0.612403 Iron 0.074969 -0.036698 0.022679 0.753564 Magnesium -0.028499 0.025150 -0.112908 -0.845694 Potassium -0.006699 -0.097429 0.226530 0.823898 Silicon 0.210878 -0.275128 0.014745 0.604545 Sodium -0.219404 0.076706 0.057721 -0.928687 Sulphur -0.020109 -0.033728 0.096757 0.749117 Titanium 1.096906 2.948397 2.096229 1.201292 Eigenvalue 24.57 17.47 10.01 9.14 % Total Variance 24.57 42.04 52.05 61.19 Cumulative % For explanation of numbers m bold please see Table 9 Episode 2 Amorphous particulates were found to be the dominant shape in this episode, while sphere, flat, and round shaped particulates were found in much smaller numbers (Table 13 & 14). The presence of amorphous and flat particulates suggest that road dust may be an important contributor to the episode. The morphological data showed little difference between the monitoring sites suggesting that each site was equally affected by the main sources (Table 13). The particle size data shows a significant difference between the locations. The Lakewood location had a significantly smaller mean particle size compared to the Plaza and Van Bien sites (Table 15). This suggests that sources contributing larger particle sizes (likely road dust) are more important at the Plaza and Van Bien locations. This also corresponds to the amount ofPM10 52 being sampled as the Van Bien location had twice the amount ofPM10 of the Lakewood location. The particle size distribution illustrates a noticeable peak at the 3-4 J.lm range which suggests that the source contributing quite substantially to this episode is road dust (Figure 8). This positively skewed I bimodal distribution is consistent with other studies and is seen in most of episodes examined in this study (Kim et a/.,1987). The qualitative chemical composition averages indicate some significant differences between the locations. The four most significant elements present in this episode were aluminum, carbon, sodium, and silicon (Table 16). This suggests that road dust (silicon, aluminum) and combustion sources (carbon) may be the largest contributors to the PM10 (Chow,1995). The Lakewood location had significantly less silicon, aluminum, and magnesium and significantly more sulphur and sodium than either Plaza or Van Bien locations (Table 16). This suggests that Figure 8: Particle SizeDistribution: Episode 2 • 950328 360 " 288 264 240 ] 216 n:l 192 a. ...0 .168 .c 144 E 120 2 96 72 48 336 312 1/1• Q) ~ Q) :J 0 <;; 2.5 (2.5,3] (3,4] (<(5] (5,6] (6,7] (7,8] (8;9] (9)0] •• >10 Average Diameter (micrometers) 53 industrial sources were impacting this location and the road dust source (characterized often by silicon, aluminum, and magnesium) was not as important. The PCA determined five factors which accounted for 62.87% of the total variance (Table 18). The first factor ( 17.91%) contained high loadings on aluminum and potassium and corresponding negative loadings on carbon and represents the road dust source "K-Feldspar" (Table 18). The second factor (14.85%) which has high loadings on calcium and sulphur and a low loading on phosphorus represents an industrial source which has been discussed previously (Table 18). Factor three (11.95%) had extremely high loadings on silicon and smaller negative loadings on carbon, chlorine, and sodium represents road dust "Quartz" (Table 18). The fourth factor (9.58%) with loadings on magnesium and sodium also represents road dust "Magnesium Oxide" (Table 18). It is unclear what the fifth factor represents in this case. A PCA was performed on each location and the results were similar to those found above (Appendix G). The only difference of interest was that the first factor at Lakewood site was industrial opposed to road dust at the other sites TABLE 18 : PCA E. ~ . de 2- 950328 nmary Fac t ors: E;paso ues an dP. Factor 1 2 3 4 Road Dust Industrial Road Dust Road Dust K-Feldspar Sulphur Source Quartz Mg Oxide -0.031509 Aluminum -0.823241 0.080392 0.3 13102 0.041241 0.007017 Calcium -0.914956 0.042907 0.112625 Carbon 0.546088 0.644555 0.152618 0.022124 0.018748 -0.060019 Chlorine 0.3398 0.040154 0.148446 0.074004 Iron 0.291549 -0.219916 0.008475 0.103446 Magnesium 0.762741 -0.063028 0.080184 0.048348 Phosphorus -0.3407 -0.736094 0.028645 -0.015791 -0.005744 Potassium 0.072865 0.23911 -0.087712 Silicon -0.940982 -0.003123 0.075792 0.521777 -0.686618 Sodium 0.028815 0.076107 -0.028283 -0.851322 Sulphur 0.113245 -0.137628 0.024038 Titanium 0.326628 2.149674 1.781623 1.43425 1.149502 ~ 17.91 14.85 11.95 9.58 % Total Variance 17.91 32.76 44.71 Cumulative % 54.29 For explanation of numbers in bold please see Table 9 5 Other 0.007178 -0.177648 -0.033988 0.031853 -0.444053 0.020051 -0.669094 -0.00213 0.072411 0.042447 0.173114 0.576403 1.029822 8.58 62.87 54 suggesting that this location was affected differently during this episodes, which is consistent with the lower levels ofPM10 present at this site compared to the other sites. Episode 3 Amorphous particulates were found to be the dominant shape in this episode, while oval, sphere, flat, round, rectangular shaped particulates were found in much smaller numbers (Tables 13 & 14). The presence of amorphous and flat particulates suggest that road dust may be an important contributor to the episode. The morphological data between monitoring sites showed little difference (Table 14). The particle size data shows no significant difference between the locations (Table 15). The Lakewood location in this case is not exceeding the ambient objective A of 50!J.g/m3 in this episode suggesting that sources affecting the other two locations enough to cause an episode do not influence this site as severely. The particle size distributions show a distinct peak at the 3- ~ m range which is indicative of road dust being an important source for this episode (Figure 9). The qualitative chemical composition averages indicate two significant differences between the locations. The four most significant elements present in this episode were aluminum, carbon, sodium, and silicon (Table 16). This suggests that road dust (silicon, aluminum) and combustion sources (carbon) may be the largest contributors to the PM10 (Chow,l995).The EDAX chemical composition averages indicate that the Lakewood location contained significantly more silicon and significantly less sulphur than the other sites suggesting that this site was influenced more by road dust than industrial sources (Table 16). 55 Figiu:e 9: Particle Siz.e Distribution: Episode 3 - 960227 420 390 360 330 - :!l 300 - &. 240 - ... ~ 270 - .....0 --- - t t ~---------------------------------------------- <= 2.5 (2.5,3] (3,4] . ~ --------------------------------------- (4,5] (5,6] (6,71 (7,8] (B,9J (9o10J > 10 Average Diameter (micrometers) The PCA determined six important factors which accounted for 58 .24% of the total variance (Table 19). The first factor (15 .09%) has high loadings on sodium and chlorine and a negative loading on magnesium and represents Salt (Table 19). The source of salt could be either industrial or from winter salting activities. The second (10.98%) and third (8 .66%) factors are the road dust factors representing "Quartz and K-Feldspar" seen before (Table 19). The fourth factor (8.43%) has high loadings on sulphur and calcium, which as discussed previously probably represents an industrial source (Table 19). It is unclear what the last two factors represent. A PCA was performed on each location and some differences were found at the Plaza location (Appendix G) . The Plaza location had a significant combustion factor which wasn't found at either of the other sites. 56 - nmary Fac t ors: E~ . d e 3 960227 TABLE 19 : PCA E"1genval ues an dP. 4 3 2 1 Factor Industrial Road Dust Road Dust Salt K-Feldspar Sulphur Source NaCl Quartz -0.139364 0.217995 0.129919 -0.785749 Aluminum 0.146662 0.084694 0.63205 -0.044995 Barium -0.120629 0.068794 0.30532 0.511571 Calcium -0.105339 0.032191 0.373464 -0.794867 Carbon -0.065534 0.043865 0.108511 -0.585407 Chlorine 0.020914 0.04094 0.013198 -0.058613 Iron -0.190018 0.093802 -0.088305 0.522589 Magnesium 0.044893 -0.05836 0.05065 -0.085596 Manganese -0.16102 0.001169 -0.088146 0.04075 3 Phosphorus 0.016404 0.033652 -0.047558 -0.781684 Potassium -0.196475 0.126647 0.902381 0.185196 Silicon -0.285655 0.023005 0.135523 -0.790422 Sodium -0.048946 -0.178194 -0.047045 0.762513 Sulphur -0.01692 -0.029005 0.093974 0.056353 Titanium 2.112819 1.537874 1.212316 1.180689 Eigenvalue 15.09 10.98 8.66 8.43 % Total Variance 15.09 26.08 34.74 43 .17 Cumulative % For explanation of numbers in bold please see Table 9 5 Other 6 Other -0.047751 0.118454 -0.364646 0.077232 -0.299382 -0.003864 -0.182595 0.184535 -0.597112 0.139413 0.118116 0.088916 0.049253 -0.643683 1.058253 7.56 50.73 -0.134528 0.052751 0.005616 0.096476 -0.362146 -0.454904 -0.485628 -0.601021 0.249276 0.226054 0.185649 0.050268 -0.052999 -0.052381 1.052121 7.52 58.24 Comparison ofEpisodes The three episodes show considerable differences in chemical composition, morphology, and particle size. Episode 1 contained significantly less amorphous particulates and significantly more oval, round, and spherical particulates (Table 14). Episode 1 had a much larger industrial/combustion component as represented by the more "rounded" -featured morphologies. Mean particle size and particle size distributions illustrated that road dust strongly influenced Episode 2 and to a lesser extent Episode 3 (Table 15;Figures 7-9) .There is a recognizable peak at the 3- ~ m range indicating the influence of road dust on the ambient air (Figures 7&8). The significantly smaller mean particle size and large percentage of fine particulates in Episode 1 are consistent with the influence of the industrial I combustion component (Table 15). The difference between the episodes is well illustrated by comparing the fine particulate fraction in Episode 1 (74%) compared to Episode 2 (39%) and Episode 3 (49%). The presence of a large proportion of 57 fine particles has important health considerations because they are more likely to be deposited deeply in the lungs and are believed to remain in the lungs for long periods of time (Dockery & Pope,l994; Vedal, 1996). Episodes 2 and 3 contained significantly larger mean particle sizes which is consistent with the impact of road dust (confirmed by the particle size distributions and PCA) (Table 15). The qualitative chemical composition also indicates significant differences between the episodes which are consistent with the observation that Episode 1 was impacted by industrial I combustion sources while episodes 2 and 3 were impacted more by road dust. Episode 1 had significantly less aluminum, magnesium, silicon (large components in road dusts) and significantly more carbon, potassium, sodium, and sulphur (large components in combustion/industrial sources) (Table 16). The PCA performed on the three episodes also confirm the differences in source contribution to the three episodes. Non-Episode 1 Amorphous particulates were found to be the dominant shape in this Non-Episode, while oval, round, sphere, flat, rod, and rectangular shaped particulates were found in much smaller numbers (Table 14 & 20). The presence of70% amorphous particulates suggests that road dust may be an important contributor, however many other sources can contribute amorphous particulates including uncontrolled combustion sources so this is not diagnostic (Dockery & Pope,1994). The morphological data between locations show a few differences. The Plaza station had less "round" , more "flat", and no "rod" shaped particulates (Table 14). It could be an indication of more clay (road dust) particulates at the Plaza location. The Lakewood station had 58 fewer "oval" particulates suggesting that combustion sources may not be as important at this site (Table 14). TABLE 20: Distribution of Various Amorphous Non-Episode % Oval % in Selected N Sphere Round % % 3.11 Flat /o 0 960122 plaza 960122 vanbien 9601221akewood 92 71.67 71.33 69.77 3.66 4.33 4.33 2.33 1.00 3.67 4.65 3.99 4.00 3.33 4.65 16.21 18.00 15.00 15.62 960304 plaza 960304 vanbien 960304 lakewood 82.00 91.00 89.67 0.00 0.33 0.33 1.00 0.67 0.00 6.33 1.00 3.67 10.67 7.00 6.33 960509 plaza 960509 vanbien 960509 lakewood 87.67 87.33 87.33 0.33 0.33 0.33 1.33 0.33 0.67 2.56 1.67 2.33 3.67 8.33 9.67 7.67 Non-Episode Smooth Flat % Cube % Rectangle % Rod % 960122 plaza 960122 vanbien 960122 lakewood 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.33 1.66 0.00 2.00 1.33 960304 plaza 960304 vanbien 960304 lakewood 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.67 0.00 0.00 0.00 0.00 960509 plaza 0.00 960509 vanbien 0.00 960509 lakewood Non-Episode (n=300); Totals (n=900) 0.00 0.00 0.33 The particle size data shows no significant differences between the locations (Table 15). The particle size distribution shows a substantial amount (70%) of the total particulate in the fine fraction suggesting that anthropogenic sources such as combustion are important contributors (Figure 10). 59 The qualitative chemical composition averages indicate some significant differences between the locations. The four most significant elements present in this Non-Episode were carbon, sodium, sulphur and silicon (Table 21). This suggests that both road dust (silicon) and combustion/industrial sources (carbon, sulphur) were significant contributors to the PM10 (Chow, 1995). The Van Bien location had significantly less calcium and sulphur than the other sites suggesting that perhaps the industrial source is not as important at this site (Table 21). The PCA determined five important factors which accounted for 71 .48% ofthe total variance (Table 22). The first factor (21.72%) represents an industrial source (seen previously) (Table 22). The second factor (16.82%) is considered a road dust "Mica or Feldspar" source despite the very low loading on sulphur which was not considered important (Table 22). The third factor (11.81%) has high loadings on carbon and negative loadings on sodium and represents an Figure 10: Particle Size Distribution: .Non-Episode 1 -960122 645 . 602 - 559 - "516 - 473. . 430 - ~ 344 - ~ 0 387 0... 301 . ..8· 258 § •215 2 · 172 - <= 2.5 (2.5,3) (3,4) (4,5] (5,6] (6,7) {7,8) {8,9) (9,10) > 10 Average Diameter (micrometers) 60 SD Mean% nd I I 16.28. 14.35ab 13.35. 14.02 12.38 12.92 14.66 •• 27.55 26.92 26.11 13 .11 ·· Mean% 26.86* c SD 15.50 11.78 12.67 17.15 17.02 18.09 16.32 12. 13.49 11.39 11.46 I 6.27 3.20 5.51 4.40 3.01 0.88 1.30. 1.02. 0.25b 1.56 0.76 1.02 4.01 2.67 3.55 SD Ca Mean% 1.49* 1.80. 0.92b 1.74. I Mean% nd· I I SD Mean% SD SD 1 nd n nd nd 960122 plaza 0.02. nd nd nd nd nd 0.28 nd 960122 vanbien 0.24b 960122 lakewood nd nd nd nd nd 2.05 nd 2 960304 plaza o.o2• 960304 vanbien nd nd 0.25 2.26 0.27 nd nd 0.17b 960304 lakewood 0.58 0.15 0.84 0.03 0.12 2.01 1.03 nd· 3 nd nd nd nd r 960509 plaza nd nd nd nd nd nd 0.27 3.13 960509 vanbien nd nd nd nd nd nd 0.15 1.30 960509 lakewood nd nd nd nd nd 0.11 0.93 nd Superscript down columns (abc)/(* ** ***)indicate significant differences between means (p<0.05); Confidence Intervals (±10%); nd = not detected; ANOVA results for Table 21 in Appendix I I ~ TABLE 21: Qualitative Chemical Characterization ofPMto N Non-Episodes SD Me!:% SD nd nd 5.08 8.12 nd nd 960122 plaza 960122 vanbien 4.11 4.06 nd nd 960122lakewood 3.92 3.61 nd nd 10.37 •• 7.92 0.01 0.26 8.o8• 960304 plaza 5.33 nd nd 12.39b 960304 vanbien 0.45 8.86 0.03 10.64° 8.49 nd nd 960304 lakewood 8.94··· 7.08 nd nd 960509 plaza 8.91 7.29 nd nd 8.74 960509 vanbien 7.03 nd nd 6.92 9.17 960509 lakewood nd nd 61 nd nd nd nd nd nd nd 0.02 nd nd 0.04 nd nd nd nd 0.44 nd nd 0.76 nd nd nd SD 1.75 3 1.52 3 0.54b 1.82 3 0.43b 0.25b 1.27 •• 6.69 3 4.25b 8.183 0.83 •• 11.09 11.56 9.64 11.63 2.77 3.84 2.12 1.53 4.34 4.46 5.35 2.68 SD 38.343 41.30b 4l.89b 36.863 42.54b 43 .61 b 40 .51 •• 26.94 28.35 25 .71 41.00 .. 21.oo· Mean% 12.88 17.11 16.94 12.53 13 .80 11.97 11.44 13 .19 12.82 13 .53 13 .13 16.02 SD 0.03 nd 0.05 nd nd nd nd nd nd nd nd nd SD 0.03 0.02 0.02 0.04 0.05 0.05 0.02 Mean% 0.003 nd nd 0.01 0.02 29.85 28.96 30.18 30.41 Na Mean% 31. 29.77 32.66 31.16 29.01 33 .91 3 25.64b 27.48b nd nd 0.14 0. 0.40 0.25 0.21 0. 0.87 0.48 0.24 o.m SD 10.73 12.56 11.00 14.03 11.47 14.41 14.66 11. 12.62 11.66 10.81 SD Superscript down columns (abc)/(* ** ***)indicate significant differences between means (p<0.05); Confidence Intervals (±10%); nd = not detected; ANOVA Results for Table 21 in Appendix I 960304 plaza 960304 vanbien 960304 lakewood 3 960509 plaza 960509 vanbien 960509 lakewood z 960122 plaza 960122 vanbien 960122lakewood . TABLE 21: Qualitative Chemical Characterization of PMto Non-E isodes cont. Non-Episodes K Mg Mn Mean% SD Mean% SD Mean% 0.65. 1.66 3.22 0.001 3 1.54 1.52 0.60 2.21 nd 960122 plaza l.77ab 960122 vanbien 1.73 0.67 3.82 0.003 1.90b 1.72 0.70 3.42 960122lakewood nd 1.55·· 2.9(. z 1.73 6.48 nd 3 3 1.76 1.85 1.99 4.36 nd 960304 plaza l.57ab 3.71 b 1.72 nd 7.75 960304 vanbien 1.32b 3.04b 1.59 960304 lakewood 6.76 nd 1.76. 1.53 ... 1.46 3 4.55 nd 1.46 960509 plaza 1.81 1.79 4.84 nd 1.70 1.48 960509 vanbien 1.26 4.16 nd 1.76 1.45 1.56 4.61 960509 lakewood nd 62 organic/combustion source (Table 22). It is unclear what the fourth factor represents; however, the fifth factor (9.74%) represents road dust "Iron oxide" (Table 22). A PCA was performed on each location and these results were similar to those above (Appendix G) . TABLE 22· PCA Eigenvalues and Primary Factors: Non-Episode 1 - 960122 3 4 2 1 Factor Other Combustion Industrial Road Dust Sulphur Source Mica or Feldspar 0.185347 -0.112057 -0.082603 -0.731939 Aluminum 0.068284 -0.00024 0.239079 0.823817 Calcium 0.182375 0.535083 0.613017 -0.511879 Carbon -0.04035 -0.01939 -0.030758 0.029482 Iron 0.105447 0.047085 0.07135 -0.824367 Magnesium -0.03736 1 -0.02355 -0.071 328 -0.698072 Manganese -0.279021 0.081481 0.165646 0.732949 Potassium -0.122703 0.132209 -0.128613 -0.866575 Silicon 0.093395 -0.201671 0.11 3541 -0.945421 Sodium 0.006405 -0.256343 0.399058 Sulphur 0.806972 -0.042583 0.052121 0.033356 0.067261 Titanium 2.3 89593 1.850272 1.298811 1.251996 Eigenvalue 21.72 16. 82 11.81 11.38 % Total Variance 21.72 38.54 50.35 61.73 Cumulative % For explanation of numbers in bold please see Table 9 5 Road Dust Iron oxide -0.192337 -0.061116 0.074749 -0.831406 -0.003701 0. 071341 0.09405 0.148229 0.025557 -0.114334 -0.5478 1.071776 9.74 71.48 Non-Episode 2 Amorphous particulates were found to be the dominant shape in this episode, while oval, round, sphere, and flat shaped particulates were found in much smaller numbers (Table 14 & 20). The presence of 87% amorphous particulates suggests that road dust may be an important contributor, however many other sources can contribute amorphous particulates including uncontrolled combustion sources so this is not diagnostic of a particular source (Dockery & Pope, 1994). The morphological data between the locations indicated some differences between location. The Van Bien location had fewer "sphere" shaped particulates suggesting combustion may have been less important at this site while the Plaza location contained more flat particulates suggesting that road dust may have had a greater influence on this site (Table 14). 63 The mean particle size shows no significant difference between the locations (Table 15). The particle size distribution illustrates a small peak at the 3-4J..lm range which is indicative of road dust, however, dominance of fine particulates (<2.5 J..lm) accounting for 61% of the total particulates indicates that other anthropogenic sources are more important (Figure 11). The qualitative chemical composition showed that the four most abundant elements were aluminum, carbon, sodium, and silicon (Table 21). This suggests that road dust (silicon, aluminum) and combustion sources (carbon) may be the largest contributors to the PM10 (Chow, 1995). The Plaza site had significantly more sodium and sulphur and significantly less aluminum, magnesium, and silicon suggesting it was more highly influenced by industrial sources rather than by road dust (Table 21 ). Figure 11: Particle Size Distribution: Nc;n-Episode 2 -960304 .555 518 481 444 407 Ill 370 Cl) 0 :w 333 a. 296 0 259 Cl) .c 222 E ::;, 185 .,.. .... .. 2 61.3% ..,..;;,;,;,o,;,o,...----------------------------------------------------------. . . . . . . 148.. . 111 . ------~~--------------------------------------------- ..Ll'L-••-~11 37 • ~ ...... <=2.5 (2:5,3] ~ ......J..... (3,4] (4,5] - t (5,6] ---------------------------------- t (6,7] ~ --- - (7,8] (8,9] -----------~ ~ ~ 111 ~ ~ t -- (9,10] . > 10 AvenigeDiameter (micrometers) The PCA determined six important factors which accounted for 60.53% of the total variance (Table 23). The first factor (15 .75%) was a combustion source indicative ofthe large 64 carbon loading and negative silicon and aluminum loadings (Table 23). Factors 2 "Feldspar"(11.16% ), 4 "Iron oxide"(8 .23% ), and 6 "Ca-Feldspar"(7.54%) represented road dust (Table 23). Factor 3 (10.17%) represents an industrial source and factor 5 (7.67%) was salt (Table 23). A PCA was performed at each location which indicated some differences in the importance of sources (Appendix G) . As expected the Van Bien location was influenced greater by road dust source (which was consistent with the mean particulate size) and contained no combustion factor (Appendix G) . The Lakewood location was influenced by a salt factor and road dust source far more than either combustion and industrial sources (Appendix G) . . d e 2 - 960304 ramary F actors: Non-E;paso TABLE 23 : PCA E" ~ ues an dP" 3 Factor 1 2 4 Combustion Road Dust Industrial Road Dust Feldspar Sulphur Source Iron oxide 0.183092 0.016479 -0.339675 -0.655003 Aluminum 0.017783 0.114206 0.067545 Barium -0.769963 0.179929 -0.045304 0.100452 0.102127 Calcium -0.3 13336 0.151004 0.057033 0.836986 Carbon -0.063849 0.156996 -0.645835 0.019088 Chromium -0.059182 -0.066785 -0.078649 0.082704 Chlorine 0.038461 0.02561 -0.088047 0.029826 Copper -0.181695 0.007863 -0.682213 0.06226 Iron 0.006194 -0.107274 -0.047001 -0.627457 Magnesium -0.048594 0.103153 -0.070466 0.811913 Potassium 0.153844 -0.206241 -0.816336 0.10411 Silicon 0.3 18645 0.209965 -0.117216 Sodium 0.37007 0.217014 0.152032 0.044586 Sulphur 0.697756 -0.098619 -0.00871 -0.141978 0.138095 Titanium 2.205583 1.562109 1.424358 1.152662 ~ 15.75 11.16 10.17 8.23 % Total Variance 15.75 26.91 37.09 45 .32 Cumulative % 5 Salt NaCl -0.251503 0.019492 0.046704 -0.223798 0.089968 0.81367 -0.014465 -0.050109 0.028114 -0.234671 -0.34481 0.655536 0.18395 -0.037479 1.073641 7.67 52.99 6 Road Dust Ca-Feldspar 0.036213 0.048118 -0.725378 0.036552 0.25451 -0.087542 -0.009782 -0.112962 -0.503023 0.047254 0.109922 0.276704 -0.220325 -0.5049 1.055448 7.54 60.53 For exrplanation of numbers in bold please see Table 9 Non-Episode 3 Amorphous particulates were found to be the dominant shape in this Episode, while oval, round, sphere, flat shaped particulates were found in much smaller quantities (Table 14 & 20). The presence of 87% amorphous particulates suggests that road dust may be an important 65 contributor, however many other sources can contribute amorphous particulates including uncontrolled combustion sources so this is not diagnostic of a particular source (Dockery & Pope, 1994). The morphological data indicates little difference between the three locations analyzed. The mean particle size indicates no significant difference between locations (Table 15). The particle size distribution illustrates that the fine particulate is dominant (76%) even in the cleanest of air, which is represented by this Non-Episode (Figure 12). The qualitative chemical composition indicates that the four most abundant elements were aluminum, carbon, sodium, and silicon (Table 21). This suggests that road dust (silicon, aluminum) and combustion sources (carbon) may be the largest contributors to the PM10 (Chow, 1995). There is significantly less silicon at the Plaza location and significantly less sulphur at the Lakewood location (Table 21 ). This is consistent with the vicinity of industrial sources to these locations. Figure 12: Particle Size Distribution: Non•Episode 3 - 960509 690 ___.12.2.2!! _________________________________________________ 644 598 552 506 Ill Cll 460 u 414 ~ n:l a.: 368 ...0Cll 322 ;p E :::1 z 276 230 184 138 92 46 0 -·-··-•j'ft··-·-----··-··-··--·----------------·-·· _jj_'.l!. <= 2.5 (1.5,3] (3;4] (4,5] {5,6] (6,7] (7,8] (8,9] (9;10] > 10 Average Diameter 66 The PCA determined four important factors which accounted for 57.28% of the total variance (Table 24). The first factor (18.28%) was a combustion source, while the second factor "Mica or Feldspar" (16.67%) and the fourth factor "Iron oxide" (9.89%) were road dust sources (Table 24). The third factor (12.44%) was an industrial source (Table 24). A PCA was performed at each location which indicated that the combustion source was more influential at the Plaza location than the other sites (Appendix G). TABLE 24: PCA Eigenvalues and Primary Factors: Non-Episode 3 - 960509 3 4 2 Factor 1 Industrial Road Dust Road Dust Combustion Iron oxide Mica or Feldspar Sulphur Source -0.215107 -0.035597 -0.312653 0.710797 Aluminum 0.049704 0.112692 0.25147 0.695974 Calcium -0.030629 0.112018 0.079386 Carbon 0.865114 0.00625 -0.018065 0.067092 -0.756789 Iron 0.162375 -0.039793 0.08778 0.722679 Magnesium 0.073158 0.107738 0.089235 0.033281 Phosphorus 0.193647 0.116731 -0.047459 Potassium -0.65575 0.055646 0.093326 -0.579232 Silicon -0.629076 -0.079645 -0.063879 0.035406 Sodium -0.818941 -0.257774 -0.19155 -0.053435 0.784045 Sulphur 0.008844 0.102006 0.002656 Titanium -0.719186 2.011002 1.833428 1.368153 1.087702 Eigenvalue 18.28 16.67 12.44 9.89 %Total Variance 18.28 34.95 47.39 57.28 Cumulative % For explanation of numbers in bold please see Table 9 Comparison ofNon-Episodes The three Non-Episodes show some differences in morphology, particle size, and chemical composition. The morphological information suggests that combustion sources (such as beehive burners) may have been more influential in Non-Episode 1 compared to Non-Episodes 2 and 3 because of the larger numbers of oval and flat shaped particulates (Tables 4 & 14). The mean particle size data indicates statistically significant differences between the three non-episodes. The third Non-Episode had a significantly smaller mean particle size than the other 67 two Non-Episodes (Table 15). Non-Episode 2 had a large peak at the 3- ~ m range indicating that road dust was dominant on this date (Figures 10-12). As the ambient PM10 decreases, the proportion affine particulates (<2.51J.m) increases suggesting that the ambient air normally contains a much larger proportion affine particulates (Figures 7-12). This has implications for health effects because even at low ambient PM10 levels, there are potentially detrimental effects on health, perhaps due to the large number of fine particulates present (Kao & Friedlander, 1995). The qualitative chemical composition averages did differentiate between the three NonEpisodes. Non-Episode 1 had significantly more carbon, calcium, and sulphur and significantly less aluminum, magnesium, and silicon compared to Non-Episodes 2 and 3 which was consistent with the large industrial factor present (Table 21 ). Examination of the PCA also indicates the importance of the Industrial source in the first Non-Episode (Tables 22- 24). The three Non-Episodes were highly influenced by the same three sources (combustion I industrial I road dust) which appear to have the most influence on the ambient PM10 in Prince George. This finding is consistent with the MELP estimates that road dust, beehive burners, and pulp mills are the three largest sources ofPM10 in Prince George (MELP, 1996). Comparison ofEpisodes and Non-Episodes There are significant differences between the Episodes and Non-Episodes with respect to morphology, particle size, and chemical composition. The morphological data indicates the only significant difference between episodes and non-episodes is in the amount of spherical shaped particulates which are indicative ofbeehive burner/combustion sources (Table 4 & 14). There are significantly more spherical shaped particulates in the Non-Episodes which suggests that beehive burners I combustion sources are more influential in Non-Episode conditions (Table 14). The 68 particle size data indicates that Episodes have a significantly larger mean particle size than NonEpisodes (Table 15). The influence of road dust to the PM10 is responsible for this increase in the mean particle size. Comparison of the particle size distributions illustrates this road dust influence as a decrease in fine particulates and an increase in the peak found between 3-4j...lm (Figures 13 & 14). The mean particle size is important due to the belief that fine particulates have a larger impact on health because they are able to penetrate deep into the lungs and remain there for long periods of time. Comparison of the qualitative chemical composition averages indicate some significant differences between Episodes and Non-Episodes (Table 25). The Episodes have significantly more aluminum, carbon, and magnesium. The aluminum and magnesium are indicators of road dust while the carbon is an indicator of combustion sources (Chow, 1995). The Non-Episodes have significantly more sulphur and sodium suggesting that normally the industrial particulates are a more important contributor to PMw (Table 25). All the significant correlation between elements and particle diameters are summarized in Table 26. These values are extremely small and only indicate very weak correlation. The PM10 sampled in this study is reasonably uniform in elemental composition across the particle sizes which is unexpected. Other studies have found crustal related elements (aluminum, silicon) and metallic elements (cadmium, copper, lead, manganese, and iron) have bimodal distribution patterns (Kao & Friedlander,1995; Infante & Acosta,1991). Some studies have found substantial co-variation between PM2.s and sulphate, which was also not seen in this data (Ostro et al. , 1991). The assumption that elemental composition is dependent on particle size was not illustrated in this data perhaps due to the large degree of uncertainty inherent in qualitative analysis. 69 Figure 13: Particle Size Distribution: Episodes 53~~ 1470 ~ 1372 ~- ------------------------------------------------- 1274 r-1176 r-- 1076 ~Ill ~ ::";(J ...ltl Q.. ..... 0 ... Q) J:l E :::1 z 980 r-882 r-784 ~686 r-- 588 r-490 ~392 t-294 196 98 0 - ---~ --- -- 10.i.% --- - -- - - ~ -- - - -- - - - - - -- - - ------------------------------ ~----------------------------- •. I <=2:5 (2.5,3) (3,4] (4,5) (5,6) I - ~< ---- ~ . {6,7] {7,8) {8,9) -----.....~--0·6% I ' {9,10J I > 10 Average Diameter (microllleters) Figure 14: Particlo $ize Distribution: 1875 ~ .--.....- ..-----------------------------------------------. . 69.2% 1750 ~- 1' ... 1625 r-1500 r-- 1375r-- :!! 1250 ~ u 1125 r-. :e ltl a. 1000. ~- 0... 875 r-- ..8 750 r-- :::1 625 E· :z 500 375 ·-··-·--·-··-··-·--·-··-··-·--·-··-··-··-··-·--·I' ..· · - ..,_ ·- -- 9.0% :--r---.·-··-·-·-··-··-----------------------5.4% 5.3% <= 2.5 (2.5,3] (3,4] (4,5] (5,6) (6,7] (7,8] (8,9] (9,10] > 10 Average Diameter (micrometers) 70 Other studies used bulk analysis of different portions of the PMw to distinguish these patterns (Kao & Friedlander, l995 ; Infante & Acosta, l991). Another possibility is that the road dust (dominant source) in the Prince George area may contain a uniform chemical composition. Aluminum Barium Calcium Carbon Chlorine Chromium Copper Iron Magnesium Potassium Silicon Sodium Sulphur son of Qualitative Chemical Characterization in Episodes I Non-Episodes Episode Non-Episode SD Mean% SD ANOVA Results 7. 7.4 H(l,n=5397)=103 .71 , p=O.OOOO 9.84" 7.92 0.0009 0.27 0.003 0.15 H(l,n=5397)=1.047611, p=0.3061 4.04 H(1 ,n=5 397)=2.80, p=0.0941 4.01 1.15 1.21 18.2lb 21.45" 17.17 15.66 H(l ,n=5397)=59.76, p=O.OOOO H(1 ,n=5397)=3.02, p=0.0821 0.019 0.38 0.037 0.18 H(1 ,n=5397)=1.15, p=0.2830 nd nd 0.004 0.19 0.016 0.83 0.097 2.42 H(1 ,n=5397)=2.72, p=0.0988 0.198 1.58 0.145 1.62 H(l ,n=5397)=4.14, p=0.0419 1.7b 2.21" 4.78 5.02 H(l ,n=5397)=94.44, p=O.OOOO 1.69 1.97 1.68 1.62 H(1 ,n=5397)= 12.59, p=0.0004 16.68 36.18 15.41 H(1 ,n=5397)=0.43 , p=0.5132 36.39 30.03b 24.8" 12.49 10.73 H(1 ,n=5397)=271.17, p=O.OOOO 2.83b 2.09" 6.63 7.08 H(1 ,n=5397)= 12.11, p=0.0005 Superscript across rows indicate significant differences between means (p<0.05); Confidence Intervals (±10%); nd =not detected The qualitative composition of different morphological shapes was compared to further define the sources of ambient PMw. Only those elements showing significant differences between morphological shapes were reported (Tables 27 & 28). Amorphous particulates dominated the ambient samples and were contributed by many sources including road dust (accounting for the aluminum, calcium, magnesium, and silicon) and combustion (carbon) (Chow, 1995) (Tables 27 & 28). The oval and spherical shaped particulates, which are diagnostic of combustion sources, contained significantly more carbon and significantly less aluminum, magnesium, and silicon compared to the amorphous particulates (Table 27). The flat particulates in the Episodes contained significantly more aluminum, calcium, magnesium, and silicon and significantly less sodium and carbon than the combustion morphological shapes and indicates that these are clay 71 particles (Chow, 1995). The rectangular shapes contained very high levels of sulphur and calcium indicative of an industrial source. All the morphological shapes identified except (smooth-flat) contained some level of sulphur suggesting that there is an interaction occurring between sulphur (S0 2) and the fine particulates in the ambient air (Table 27 & 28). The sulphur may be coating the surface of the particulates (Keyser eta/. , 1978). The distinctions between morphological shapes in the Non- Episodes are not as evident most probably due to the contributions of many different sources instead of just a few sources seen in the Episodes. The PCA performed on the Episodes and Non-Episodes illustrate that importance of source differs between locations and dates (Tables 17-19;22-24). In Episode 1, there was an industrial source providing the most significant PMw contribution while Episodes 2 and 3 were influenced more by road dust. The Non-Episodes were all influenced by combustion, industrial, and road dust sources. Overall, the main sources seem to remain quite consistent between all dates sampled except the combustion factor was more evident in the Non-Episodes. 72 0.09 S = 1.4447 + 0.21142*Diameter 0.13 0.22 0.1 -0.07 Calcium Magnesium Manganese Potassium -0.33 0.14 -0.17 0.17 Carbon I Na = 33.835 - 1.552*Diameter S = 1.8990 + 0.37813*Diameter Si = 38. 581- 0.9780*Diameter Ca = 0.67335 + 0.19566*Diameter Mg = 0.70873 + 0.40482*Diameter Mn = -0.0012 + 0. 00064 *Diameter K = 1.7886- 0.0431 *Diameter C = 15.746 + 1.0067*Diameter and Particulate Diameter 73 Silicon Mean(%) SD 37.37" 16.40 26.92b 10.68 29.58bc 12.0 1 27.55b 11.18 35 .05c 18.04 33. 89abc 3.96 o.oob 0.00 20.98b 18.99 na na Magnesium Mean (%) SD Particulates 2075 2.21" 4.55 0.07b 45 0.44 o.oob 28 0.00 0.46b 55 1.99 2.75c 474 5.98 o.oo•bc 4 0.00 14.39d 1 0.00 1.89abc 16 7.56 0 na na Amorphous Oval Round Sphere Flat Smooth Flat Cube Rectangle Rod ANOV A Results summarized in Appendix H Amorphous Oval Round Sphere Flat Smooth Flat Cube Calcium Mean (%) SD 1.11" 3.65 o.oo• 0.00 1.30ab 6.15 0.30" 1.83 l.57b 4.98 1.89ab 3.78 o.oo•b 0.00 10.16c 9.08 Particulates 2075 45 28 55 474 4 1 16 Aluminum SD Mean (%) 9.99" 7.44 3. 56b 2.09 5.46b 5.30 5.18b 5.71 10.83c 9.88 5.56abc 1.19 o.oo•bc 0.00 3. 62b 6. 79 TABLE 27: Comparison of Qualitative Chemical Composition and Sodium Mean (%) SD 24.99" 10.61 29.97b 6.97 31.28b 11.36 28.86b 8.40 22.68c 10.92 38.3 1bd 4.30 58.53d 0.00 18.61 c 15.45 na na Carbon Mean (%) SD 20.68" 16.34 35.70b 17.42 26.09ac 15.11 33 .64bc 19.14 22 .07ad 19.36 18.oo•cd 8.76 2 .45abc 0.00 15.44ad 16.54 Sulphur Mean (%) SD 1.75" 5.60 1.33"b 3.30 4.25b 10.93 1.66ab 5.40 2.86b 8.42 o.oo•b 0 .00 23 .28c 0.00 22.52c 21.61 na na Copper Mean (%) SD o.oo• 0.00 0.96b 6.42 o.oo• 0.00 o.oo· 0.00 o.oo• 0.00 o.oo• 0.00 o.oo• 0.00 o.oo· 0.00 74 Magnesium Particulates Mean (%) SD 1.8 1 a 2212 5.10 0.06b 0.36 38 0.28ab 40 1.36 1.13ab 3.54 92 1.6l ab 295 5. 50 o.oo·b 9 0 0.56ab 13 1.41 9 13 38 40 92 295 2212 ANOV A results summarized in Appendix H Amorphous Oval Round Sphere Flat Rectangle Rod Amorphous Oval Round Sphere Flat Rectangle Rod alitative Chemical Aluminum Mean (%) SD 8.23. 7.46 5.01 b 1.63 5.16b 2.51 6.82ab 4.88 6.75b 8.32 4.38ab 2.45 4.66ab 2.97 39.08. 32 .76ab 30.14ab 29.76cb 16.89 15.63 14.93 15.22 Silicon Mean (%) SD 36.62ac 15.34 35.39ab 10.48 34.72ab 14.61 Calcium Mean (%) SD 1.12. 4.02 o.s5• 1.29 0.97. 2.62 1.60. 6.71 1.24. 3.31 5.40b 5.67 0.66. 1.76 19.23 27.48b Sulphur Mean (%) SD 2.54. 7.10 2.26ab 5.16 4.04ab 8.72 2.56. 7.11 4.59b 9.43 12.23 c 14.99 5.55•b 10.71 15.02 19.71 7.23 16.93. 21.68b 16.2l ab Carbon Mean (%) SD 17.66. 15.08 22.08ab 13.00 20.08ab 13 .60 75 Comparison of Episodes and Non-Episodes in the BCR site BCR Episodes Amorphous particulates were found to be the dominant shape in these Episodes, while round, sphere, smooth-flat, flat, and rectangular shaped particulates were found in much smaller numbers (Table 29). The presence of 86% amorphous particulates suggests that road dust may be an important contributor, however many other sources can contribute amorphous particulates including uncontrolled combustion sources so this is not diagnostic of any particular source (Dockery & Pope, 1994). Comparison of the morphology data between episodes indicates few differences (Table 29). In two episodes 950831 and 960813 there seems to be a larger proportion of "flat" particulates which may be a result of increased unpaved road dust levels (Table 4 & 29). Analysis of the mean particle size data indicates no significant differences between the episodes (Table 30). The average particle size is quite large ( 1- ~-tm) which is illustrated in the particle size distributions which show a very large peak between the 3- ~-tm range (Table 30; Figure 15). This suggests that road dust was an important source. The qualitative chemical composition indicated that the most abundant elements were aluminum, carbon, magnesium, sodium, and silicon (Table 31). This suggests that road dust (silicon, aluminum, magnesium) and combustion sources (carbon) were likely the largest contributors to the PM10 (Chow, 1995). The qualitative chemical composition averages indicate some significant differences between the various episodes. In most cases this difference should be considered cautiously due to the uncertainty involved in the qualitative analysis. The episodes occurring on 940923 and 950831 had significantly more carbon suggesting a combustion/industrial source is a larger contributor to these episodes (Table 31 ). 76 Sphere Flat Smooth Rectangle 0.00 0.33 0.33 0.00 0.33 0.33 0.00 10.3 3 10.33 12.33 12.00 14.67 11.00 17.67 0.00 0.33 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.67 7.67 0.00 0.00 72.33 3.00 960122 10.71 0.00 0.00 1.00 0.67 87.63 930509 Episode I Non-Episode (n=300); Total Episode (n=2100); Total Non-Episode (n=599) 0.33 0.00 940408 940923 950316 950328 950831 960304 960813 89.67 88.67 87.33 88.00 83 .67 88.67 82.33 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.33 0.00 0.00 1.33 0.00 0.00 Columns with different superscripts (* ** ***) indicate significant differences between means (p<0.05) BCR samples (n=300); BCR Episode Total (N=2100); BCR Non-Episode Total (N=599) 77 Figure 15: Particle Size Distribution: BCR Episodes -· 29.2% 574 -· 533 -· 492 -· 615 451 U) . Q) 410 u 369 ... ;::; . Ill ......oiiiii.Ooo...· - · - · - · - · - · - · · - · - · · - · - · - · - · - · - · - · - · - · - · · - · - · - · - · - · - · -· -· ~- a:.. 328 ~- ...0 287 Q) ..Q E ·-·-·· J.ll.J.L... ··-·-··-·-·-·-·-·-·-·-·-·-··-·-·-·-·-·-· ·-·-·· ..13.a.-··-·-·-·-·-·-·-·-·-·-··-·-·-·-·-·-· ~- 246 1-· .....9.. .6% -....-·-·-·-·-·-·-·-··-·-·-·-·-·-· :z:· 205 . ~::l --e:'T'J!j·-·-·-·-·-·-··-·-·-·-·-·-· · 164 ~- ... ·-·-·-·-·-··-·-·-·"!fft"""·-· 3.7% . T::. 123 1-· 82·' 1-· • • 41 1-· 0 <=2;5 (2.5,3) (3,4) (4;5] (5,6J (6,7) (7;8) • I ?.!'IIi (8,9) •• • I 1.8% -· / • (9;10) -· / > 10 Average Diameter (micrometers) The PCA indicated seven important factors (sources) accounting for 63.68% ofthe total variance. Factors 1 "K-Feldspar" (12.96%), 3 "Iron oxide" (9.18%), 4 "Quartz" (8 .7%), and 5 "Sodium" (7.82%) all represent types of road dusts (Table 32). Factor 2 (11.21 %) was an industrial source. The last two factors were not interpreted because those combinations of elements were not seen previously (Table 32). A PCA was performed on each episode and the results were in most cases consistent with those above (Appendix G). In most cases road dust was the most important contributor to the ambient PM10 while an industrial factor was also evident (Appendix G). Contrary to the qualitative chemical averages, the PCA performed on Episode 950831 contained four factors all of which represented road dust and no factors representing combustion, however, it is unclear what the source of carbon is. The PCA performed on Episode 960304 indicated that industrial and combustion sources were more important contributors to this episode (Appendix G). The above episode did not exceed the 50j...Lg/m3 objective to the extent of 78 the other BCR episodes analyzed suggesting that the very high levels of ambient PMto in the BCR site have a high road dust component while lower levels ofPMto are more influenced by the industrial and combustion sources in the area to a greater extent. TABLE 31: Qualitative Chemical Characterization of PMto Episodes and Non-Episodes in the BCR site AI Ba c Ca Episodes {%) (%) (%) (%) Mean SD Mean SD Mean SD Mean SD 3.68 0.01 13 7.91 0.50 1.74a 12.18a 15.93 a 4.35 1.43 0.08 8.58 8.60 940408 14.85abc 7.52 1.87a 17.57b 14.25 nd 5.39 nd 940923 J5 .46ab 7.55 13.35a 9.06 0.86b nd 2.36 nd 950316 14.33 bc 7.20 0.99b 12.69a 7.67 nd nd 1.97 950328 14.70abc 7.57 18.25b 17.52 0.99b nd nd 3.04 950831 15.56ab 8.76 1.65a 9.85a nd nd 11.94 4.07 960304 1.73a 14.02c 8.29 13 .82c 12.47 nd nd 4.32 960813 940923 950316 950328 950831 960304 960813 Non-Episodes nda 0.01ab nda nct• 0.03 b nda nd 0.20 nd nd 0.25 nd nd nd nd nd nd 0.04 nd nd nd nd nd 0.66 nd nd nd nd nd nd nd nd nd nd nd nd 0.16 0.02 0 .19 0 .01 0 .07 0 .08 1.21 0.25 0.32 0.18 0.58 0.71 0.02 0.15 nd 2.53 nd 0.29 0.19 960122 2.09 nd nd nd 960509 nd nd nd 0.06 0.80 Superscript down columns (abc I * ** ***) indicate significant differences between means (p<0.05) Confidence Intervals: (±10%); (nd=not detected) ANOVA results summarized in Appendix H 79 TABLE 31: Qualitative Chemical Characterization of PMto Episodes and Non-Episodes in the BCR site K Mg Mn Na (%) (%) (%) (%) Episodes SD Mean SD Mean SD Mean SD Mean 4.67 3 4.19 3 4.oo•b 3.13b 4.o5•b 3.95ab 4.21 3 6.79 6.67 5.88 4.70 7.3 8 5.70 8.04 nd nd nd 0.13 nd nd nd nd nd 0.23 nd nd nd 18.o5· 14.45b 18.01 3 21.83 c 15.23b 19.46 3 18.123 10.99 10.19 11.59 9.98 10.74 12.16 11.07 nd 1.40 1.53 nd nd 2.27 0.39abc ndb 0.49cd 0.79d 1.86 0.84 1.92 nd 3.36 4.29 44.54 46.49 45.33 45 .13 47.44 45 .36 16.28 14.24 13.28 14.99 17.01 17.08 0.28 0.04 0.18 0.19 0.05 0.03 2.71 0.45 0.22 2.48 0.58 0.58 nd 2.85 6.39 3 0.86b 11.84 4.62 24.04 3 45.15b 12.85 14.52 nd 0.01 nd 0.13 1.46 3 1.90b 1.55• 1.17c 1.46 3 1.46 3 1.62 3 1.59 2.00 1.82 1.25 1.62 1.77 2.30 940923 950316 950328 950831 960304 960813 Non-Episodes nd 0.13 0.09 nd nd 0.18 960122 960509 nd 0.16 940408 940923 950316 950328 950831 960304 960813 Non-Episodes Superscript down columns (abc/* ** ***)indicate significant differences between means (p<0.05) Confidence Intervals (±10%); (nd=not detected) ANOVA results Summarized in Appendix H 80 1 dP" 2 F Road Dust Industrial K-Feldspar Sulphur Source -0.819599 0.059259 Aluminum 0.035679 Barium -0.728365 0.008074 -0.285618 Calcium 0.065034 Carbon 0.377601 0.075098 0.11 3333 Chlorine -0.027001 0.02935 Chromium -0.05534 0.00186 Iron -0.086237 0.001739 Magnesium -0.201013 0.033335 Manganese -0.002234 0.151284 Phosphorus -0.013333 -0.759766 Potassium 0.115002 0.189 123 Silicon 0.078292 0.002973 Sodium 0.025094 -0.768555 Sulphur 0.015304 0.010624 Titanium 1.681125 Eigenvalue 1.9444 12 12.96 11 .21 % Total Variance 12.96 24.17 Cumulative % For explanation of numbers in bold please see Table 9 PCA E" TABLE Factor BCREoisod -3 4 Road Dust Road Dust Quartz Iron oxides -0.042 135 0.043077 0.008507 -0.004364 -0 .016052 0.06019 -0.008856 0.76413 0.12202 0.026219 -0.018856 -0.158985 0.019424 -0.802375 0.020287 0.239025 -0.032592 0.054746 -0.043304 0.002 182 -0.05863 -0.045034 0.016434 -0.935349 0.103828 0.323518 0.004513 0.0563 14 -0.004257 -0.817711 1.377258 1.30473 1 9. 18 8.7 33 .35 42.05 5 Road Dust Sodium 0.232219 0.064975 0.077653 0.059728 0.063272 0.004977 0.063433 0.760782 -0.139784 0.0 16561 0.0 17741 0.097581 -0.816587 -0.06749 1 -0.014316 1.173345 7.82 49.87 7 Other 0.152748 -0.040719 0.099806 -0.230579 0.471438 0. 730371 0.195862 0.224641 -0.137743 -0.169032 -0.137262 -0.144 189 0.192849 0.0 16225 -0.100612 1.010437 6.74 63 .68 6 Other 0.0755 1 0.215 118 -0.782876 0.102 159 -0.377037 0.165122 0.0428 19 -0 .088843 -0.047825 -0.622552 0.01222 1 0.161228 0.053263 -0.296266 -0.068193 1.060679 7.07 56.94 81 BCR Non-Episodes The two Non-Episodes were extremely different from each other with regards to morphology, particle size and chemical composition. Amorphous particulates were found to be the dominant shape, while oval, sphere, flat, and rectangular shaped particulates were found in much smaller quantities (Table 29). The presence of 80% amorphous particulates suggests that road dust may be an important contributor, however many other sources can contribute amorphous particulates including uncontrolled combustion sources so again this is not diagnostic of any source (Dockery & Pope, 1994). Comparison of morphological data indicates that the NonEpisode 960122 was influenced more by combustion sources due to the larger percentages of "oval" and "spherical" shaped particulates (Table 29). Episode 960122 also contained more "flat" particulates which with the "round" shaped particulates can be indicative of combustion sources such as beehive burners (Table 4). The mean particle size shows no significant differences between the two Non-Episodes (Table 30). The particle size distribution indicates a large proportion of fine particulates (59%) in the Non-Episodes (Figure 16). The qualitative chemical composition indicates that the most abundant elements were aluminum, carbon, sodium, sulphur and silicon (Table 31). This suggests that road dust (silicon, aluminum) and combustion I industrial sources (carbon, sulphur) may be the largest contributors to the PMw (Chow, 1995). The qualitative chemical composition averages show significant differences between the two Non-Episodes analyzed (Table 31). Episode 960122 contained significantly more carbon, sodium and sulphur and significantly less aluminum, magnesium, and silicon which is also consistent with the morphological and particulate size results (Table 31 ). Episode 960122 appears to have been highly influenced by a combustion I industrial source. 82 Figure 16: Particle Size Distribution: BCR Non-Episodes 360 336 ---~m...----·--·-··-··-···-··-··-··-··-··-··-··-··-··• ----·-··-··-···-··-··--------·-··-··-·-··- 312 • 288 . ·-··-··-··-··-···-··-··-··-··-··-··-··-··-··- . . Cll 240 0 216 . ... a. 192 . 0 168 . Qj 144 . E :::r 120 . . 264 Ill C'CI ..Q z ··-··-···-··-··-··-··--·-··-··-··-··. <= 2.5 (2.5,3] {3,4] (4;5] {5,6] {6,7) (7,8] . {8,9] {9,10] > 10 Average Diameter (micrometers) The PCA determined five important factors accounting for 59.07% of the total variance (Table 33). Factors 1 "Mica" (19.71%) and 5 "Iron oxides"(7.81%) represent road dust while factors 3 (9.57%) and 4 (8 .63%) represented industrial and combustion sources (Table 33). It is unclear what source factor 2 represented as the combination of calcium and phosphorus was not diagnostic of a particular source. A PCA was performed on each of the Non-Episodes which confirmed that they were different. In Non-Episode 960122 the first factor was a industrial source (21.12%) and the third factor was combustion source (12.63%) which is consistent with the other analysis completed (Appendix G). The results ofNon-Episode 960509 was similar to the PCA completed above (Table 33). 83 . d es TABLE 33: PCA Eigenvalues and Pramary F actors: BCRNon- E )ISO 4 2 3 1 Factor Combustion Industrial Other Road Dust Sulphur Source Mica 0.103842 -0.233608 0.338654 Aluminum 0.721978 0.015412 0.078747 0.422669 -0.731369 Calcium -0.09765 -0.791347 0.029186 -0.295134 Carbon -0.002937 0.174885 0.057501 -0.567659 Chlorine -0.016663 -0.004108 -0.010038 -0.014302 Copper 0.041863 0.131461 0.035955 0.080938 Iron 0.156017 -0.08279 0.06799 0.75958 Magnesium -0.102539 0.018474 -0.006269 -0.804326 Phosphorus 0.079389 0.218498 0.548521 0.326517 Potassium 0.090119 0.358154 -0.587593 0.54718 Silicon 0.186859 0.110425 -0.038407 -0.815283 Sodium -0.05094 -0.133232 -0.002018 0.885211 Sulphur -0.098389 0.102079 -0.132927 -0.135055 Titanium 2.562497 1.735712 1.244378 1.122053 Eigenvalue 19.71 9.57 8.63 13 .35 % Total Variance 19.71 33 .06 42.64 51.27 Cumulative % For explanation of numbers in bold please see Table 9 5 Road Dust Iron oxide -0.113542 0.018609 0.127121 -0.202524 0.084723 0.891598 0.101109 0.019515 -0.342648 -0.175183 0.052866 -0.01941 -0.173419 1.014908 7.81 59.07 BCR Episodes versus Non-Episodes There are significant differences between the BCR Episodes and Non-Episodes which are a result of the influence of the main PM10 sources present at the BCR location. Comparison of elemental analysis (ICP) indicated few significant differences between the concentrations ofthe elements tested (Table 34). The average concentrations of most elements are smaller in the non-episodes however the differences were not statistically significant (Table 34). This may be a function ofthe variation seen in the filter blank (Appendix F). Comparison of morphology between Episodes and Non-Episodes indicates that in NonEpisodes there are significantly more oval and spherical shaped particulates (Table 35). The influence of combustion sources is greater in Non-Episodes than Episodes which seems to be overwhelmed by road dust. The mean particle size data indicates that Episodes have a significantly larger particle size than Non-Episodes supporting the conclusion that road dust plays an important role in Episodes of the BCR site (Table 30). The particle size distributions illustrate this point 84 TABLE 34: Comparison of Quantitative Elemental Analysis in the BCR site Non-Episode ANOVA Results Element Episode Mean% Mean% SD SD H(l,n=9)=3 .09, p=0.0790 3.957 5.595 13 .214 8.279 Aluminum H(l,n=9)=2.34, p=O.l263 nd 3.373 3.887 nd Barium 9.635 13.625 H(1 ,n=9)=0.34, p=0.5582 15.324 8.551 Calcium H(1 ,n=9)=3 .19, p=0.0740 nd nd 0.004 0.006 Chromium H(1 ,n=9)=0.10, p=0.7484 0.007 0.009 0.012 0.003 Copper 0.924 0.344 H(1 ,n=9)=2.14, p=O.l432 2.458 0.961 Iron 0.004 0.007 H(1,n=9)=0.09, p=0.7697 0.006 0.010 Lithium 3.385 1.637 2.374 H(1,n=9)=0.77, p=0.3798 1.813 Magnesium 0.030 0.007 0.009 H(1 ,n=9)=2.62, p=0.1059 0.063 Manganese 0.0022b 0.0083" H(1 ,n=9)=4.2, p=0.0404 0.003 0.003 Nickel 0.044 H(1 ,n=9)=0.34, p=0.5582 0.107 0.075 0.051 Phosphorus H(1 ,n=9)=3 .19, p=0.0740 3.176 2.193 Potassium nd nd 5.230 11 .580 14.850 4.320 H(1 ,n=9)=1.37, p=0.2416 Sodium 0.169 H(1,n=9)=2.16, p=0.1416 0.091 nd nd Strontium 0.005 0.009 H(1 ,n=9)=1.09, p=0.2967 nd nd Tin H(l,n=9)=0.34, p=0.5582 0.468 0.149 0.320 0.3 50 Titanium ndb 0.0075" 0.003 Vanadium nd H(l,n=9)=4.24, p=0.0396 2.180 1.755 1.413 1.998 H(1 ,n=9)=0.54, p=0.4623 Zinc Superscnpts across rows indicate significant differences in means (p<0.05) BCR Episodes (n=7); BCR Non-Episodes (n=2) TABLE 35: Comparison of Morphology between BCR Episode Non-Episode Mean SD Mean SD 260.71 Amorphous 8.36 239.50 31.82 H(1 ,n=9)=1.77, p=O.l840 5.50b o• Oval 0.00 4.95 H(1 ,n=9)=7.88, p=0.005 0.71 1.50 0.00 0.00 Round H(1 ,n=9)=0.64, p=0.4227 13.00b 0.57" Sphere 0.54 H(l ,n=9)=4.75 , p=0.0292 14.14 37.86 8.05 Flat 41.00 12.73 H(1 ,n=9)=0.086, p=0.7688 0.14 0.38 Smooth 0.00 0.00 H(1,n=9)=0.29, p=0.593 Rectan 0.00 0.00 0.50 .5, p=0.0614 0.71 Superscript across rows indicates significant differences between means (p<0.05) Episode (n=2100); Non-Episode (n=599) Means are based on Total Particulate number above through the difference in the composition affine particulates, 29% versus 59% (Figures 15 & 16). The qualitative chemical composition averages show significant differences between Episodes and Non-Episodes (Table 31). Episodes contain significantly more aluminum, magnesium, and silicon (road dust indicators), while the Non-Episodes contain significantly more 85 carbon, sodium, and sulphur (industrial/combustion indicators) especially the Non-Episode 960122 (Table 31). The correlation between elemental composition and particulate diameter were analyzed to determine the significant correlation found in Table 36. The weak correlation identified between elemental composition and diameter indicated that in Episodes and Non-Episodes aluminum and magnesium were found in larger concentrations in larger particulates and sodium is found in larger concentrations in smaller particulates (Table 36). In the Episodes phosphorus was found in larger concentrations in larger particulates while in Non-Episodes carbon and chlorine were found in larger concentrations in larger particulates (Table 36). There were expectations oflarger correlation which would indicate that elements are concentrated on certain size fractions however, this was not the case in this data set. The qualitative composition of different morphological shapes was compared to further define the sources of ambient PM10. Only those elements showing significant differences between morphological shapes were reported (Table 37). Amorphous particulates dominated the ambient samples and were contributed by many sources (Table 37). It is interesting that there is less carbon and sodium in the amorphous particulates in the Episodes compared to the Non-Episodes which suggests that the source of amorphous particulates in the Episodes is road dust while in the Non-Episodes it is road dust and combustion. The oval and spherical shaped particulates which represent combustion sources contained significantly more carbon compared to the amorphous particulates (Table 37). The flat particulates in the Episodes contained less carbon which suggests that they may be clay particles (Chow, 1995). 86 The PCA performed illustrate the dominance of road dust in Episodes in the BCR site (Table 32 & 33). The industrial/combustion source still influences the ambient PM10 in the BCR site, but not to the extent seen in the Non-Episodes. 87 -0.24 Sodium Na = 21.321 - 0.7930*Diameter P = -0.0667 + 0.03027*Diameter - SD 11.08 10.61 19.88 11.27 0.00 479 nd nd 19.09 15.49 27.84 14.65 Amorphous nd nd 24.01 32 .85 5.90 11 12.29 Oval 7.34 26 nd nd 24.43 11.55 33.03 Sphere 27.79 17.42 nd nd 20.45 17.65 82 Flat 48.82 nd 7.59 0.00 1 nd 0.00 Rectangle ANOVA results summarized in Appendix I Calcium results in Non-Episodes showed no significant differences and were not included, nd = no difference ~ TABLE 37: Comparison of Qualitative Chemical Composition and Morphology in BCR Episodes & N . Calcium Sodium Carbon Mean(%) Mean(%) Particulates Mean(%) SD SD 18.39"b 1.43" 3.82 13 .55" 10.71 1824 Amorphous 10.63bc 52.53b o.oo• 36.65 0.00 5 Round 24.61 ac o.oo• 28.00" 0.00 19.42 4 Sphere 15.94c 14.37c 1.21" 3.91 19.14 265 Flat 6.73 abc 9.53b 6.37"c 1 0.00 0.00 Smooth Flat 0.09 Phosphorus Magnesium Carbon Chlorine TABLE 36: Comparison of Significant Correlation between Elemental Composition and Particulate Diameter in the BCR site ~---~---~ ~ ~ - ~------------------------------- --------------------------------- 88 Comparison of Bowl and BCR areas: Episodes and Non-Episodes Comparison of morphology between the bowl and BCR locations indicates that during Episodes there are more amorphous and less oval, round, sphere, flat, smooth-flat, and rectangular shaped particulates at the BCR site compared to the bowl area (Tables 14 & 29). This is consistent with the conclusion that road dust is the main source contributing to the BCR site. The morphological composition ofNon-Episodes is consistent between the two areas suggesting that in normal ambient air, similar sources influence each location equally (Tables 14 & 29). The mean particle size measurements show a similar trend between Episodes and NonEpisodes in both the bowl and BCR locations. The episodes in both locals have significantly larger particle sizes than the Non-Episodes (Tables 15 & 30). The BCR location had larger particle sizes for both Episodes and Non-Episodes than the Bowl area which is consistent with the conclusion that road dust (which contributes to coarse particulates) is a more important contributor at the BCR site than at the Bowl Location (Table 15 & 30). This trend is illustrated in the particle size distributions (Figures 11-14). Comparison ofthe Episodes indicates that there is a much larger proportion of coarse particulates at the BCR site (Figures 11 & 13). The Non-Episodes show a similar trend except the Bowl Location had 10% more fine particulates than the BCR location (Figures 12 & 14). The qualitative chemical analyses indicate that during episodes, the BCR location contained more aluminum, magnesium, and silicon and less carbon, sodium, and sulphur than the bowl area which suggests that road dust has a greater influence in the BCR site (Tables 16 & 31). During Non-Episodes there were few differences between the two locations which is consistent with the morphological and particle size information. The PCA performed on the Episodes and Non-Episodes indicate the same general trends at both locations. During Episodes road dust and 89 industrial factors are dominant while during Non-Episodes road dust, industrial, and combustion factors are all significant (Tables 17-19,22-24,32-33). Examination of differences in Particle Size and Filter Location To determine the importance of filter location on randomization of results, the mean particle size was analyzed across locations on the filter of the Bowl area results. The filter was sampled in three locations (Figure 2). In the Non-Episode filters, there was a significant difference between the outside edge location (A) and the inner locations (B & C) (Table 38). The outside edge of the filter was receiving smaller particle sizes which may have been either a function of the small amounts of particulates being sampled. Overall results from the bowl area again indicate there is a significant difference between the different locations on the filter (Table 38). The difference is quite small (0.24- 3 ~m) and should not have too much impact on the overall results. Therefore, in future studies, location of sample for SEM EDAX analysis can be taken at any location on the filter. TABLE 38: Comparison of Particle Size Distribution on Different Filter Locations Episode ANOVA Results 2. 2.78 2. 2.70 Episodes I Non-Episodes: A, B, C (n=900); Total (n= 1800) Superscript indicates significant differences between means (p<0.05) 90 Comparison of Particle Diameter and Mass As illustrated in Figures 17 & 18 the average particle size distribution is not similar to the average particle mass distribution. The mass of each particle was determined by calculating the volume of the particle (4/3nr3) and multiplying by the average particle density found in soils (2.65g/m3) . These figures indicate that particle mass has a similar distribution as size except for a small portion of larger particles which contribute significantly to the total mass. This suggests that contrary to the particle size where fine particulates dominate the distribution, they do not dominate the amount of mass present in the ambient air. These results should however be considered cautiously due to the assumptions required to determine the mass. As illustrated in this study, most of the particulates are not spherical in shape and mass is a function of elemental composition which varies significantly between particles (Linton et al. , 1980). 91 Figure 17: ~ Particle Size Distribution: Episodes and Non-Episodes 61.6% 3330 t - - , - - - - . - - - - - - - - - - - - - - - - - - - - - - 3108 2886 2664 Ill Q) u :c 2442 .2220 ...cu 1998 ... 1776 1554 ~ ... ·0 Q) ,. c· 1332 E ;:, z 1110 888 666 -- 11 7.8% 444<. ~------------------- 222 0 -- <=' 2.5 {2.5,3] (3,4] (4;5] (5;6) (6,7) -11 (7,8) (8,9) - -~- (9,10) > 10 Average Particle Diameter (micrometers) Figure 18: A'le.rage Particle Mass Distribution: Episodes and Non-Epi.sodes 39.3% 2130 1--r---r·-··-·-·-··-··-·-·-··-··-·-·-··-··-··-··-·-·-··1988 1846 1704 1562 - - - - - - - - - - - - - - - --~ --- 1420 ,!!" . 0 :;::: 1278 tel 1136 Q; ....0 994 852 Q) Ill ... ... J:l E ;:, z 710 568 426 10.6% '6:3'l'O·-··----·-··-·-·-·-··-··-··-··-· 284 ~ -3 11 · 142 0 <=5 (10,15) (5,10) - . 2 3% - - - (25;30) - - - ~ n OL. -~-- ~ - 1- (30,35) (20,251 (15,20] - 2.6% - (40,45] (35,40] >50 (45,50] Mass (micrograms) 92 CHAPTER 4: CONCLUSIONS AND RECOMMENDATIONS Source Characterization Morphological and chemical examinations of the major PM10 sources in the Prince George Airshed indicated the presence of some distinguishing features between the various sources present. Anthropogenic combustion sources such as beehive burners form more spherical and oval shaped particulates which is related to the high temperatures involved in their formation . In general, the majority of particulates examined had an amorphous shape which is not diagnostic for any individual source. Flat morphology was also detected in all sources and suggesting road dust or perhaps anthropogenic (incomplete combustion) origins. The particle size distribution was the most informative and reliable data acquired in this study. The four sources ofPM10 examined indicated different particle size distribution patterns. The beehive burner sample was dominated by fine particulates (<2.5J...Lm) which was consistent with data published for combustion sources. The road dust samples contained significantly more particulates in the coarse fraction (>2.5J...Lm), and is consistent with the behavior of the mechanical breakup of soil particulates. The presence of clay particulates account for the smaller size fraction found in the road dust samples (especially in the unpaved road dust) . The average road dust and beehive burner qualitative chemical composition from SEMEDAX analysis were useful in recognizing differences between sources. These measurements were qualitative in nature with high standard deviations due to the methodology, and the large variation in chemical compositions within the particle samples. Despite the qualitative nature of the data, there were recognizable differences between the mean concentrations of many elements. In general, the beehive burner sample had more carbon while the road dust samples had more 93 aluminum, magnesium, and silicon which is consistent with the literature. These differences were used to identify the relative contribution of sources in the ambient samples. The ICP bulk quantitative analysis was not considered informative due to the problems encountered with extraction. The teflon coated glass fiber filters contributed extensive contaminants during the extraction procedure which masked much of the information for the PMw . Filters with significant PMw samples produced more interpretable results because the blank did not significantly mask the sample. The ICP results indicated some differences between the sources, especially the pulp mill PMw suggesting different elemental composition with respect to chromium, magnesium, nickel, and phosphorus. The results from the BCR site showed few significant differences between elemental composition which also may have been attributable to interference from the filter. The quantitative analysis of sources and ambient PMw is important for discerning differences and possible tracer elements, however this analysis must be replicated using a different filter media for satisfactory results. Episodic and Non-Episodic events Morphological and chemical examination of the ambient PMw in the Prince George Airshed illustrated the contribution from major PMw sources. The Episodes tend to be dominated by amorphous shaped particulates, while Non-Episodes show a large variety of particulate shapes such as spherical and oval. The other particulate types (rectangular, round, rod, and cube) were rarely seen and it was unclear as to their origins. Overall, due to the predominance of amorphous particulates, the use of morphological features to characterize the ambient PMw in Prince George was not as useful as other techniques. The mean particle size and particle size distributions illustrated a definite trend between most Episodes and Non-Episodes. Most of the Episodes examined contained a bimodal 94 distribution with a large concentration of particulates in the fine fraction ( <2.51-!m) and a second smaller peak at the 3- ~- m range. The fine particulates generally represent anthropogenic sources such as combustion while the coarse size fractions represent crustal materials such as road dust. Although, road dust source contributes some fine fraction ofPMto to the ambient air, its major contribution to the coarse size fractions is diagnostic for its presence in ambient PMto. All but one of the Episodes examined contained this second peak indicating that road dust was an important factor in Episodes. The first Episode (950121) for the bowl area was dominated by anthropogenic sources as indicated by the distinctive small mean particulate size. The Non-Episodes examined were highly positively skewed and contained a large peak in the fine fraction ofPMto and a much smaller generally indiscernible peak at the 3-41-!m diameter range. In Non-Episodes, anthropogenic sources influenced the ambient PMto as indicated by the mean particle size and particle size distribution. The fine fraction which is believed to cause considerably more health problems, dominates most of the Episodes/Non-Episodes examined. There is evidence that PMw ambient levels less than 2011g/m3 may have health impacts and the dominance ofPM2.s in instances of lower ambient PMto levels may be one explanation for this. The Episodes also illustrated that road dust and industrial sources influence the PMw levels differently at various locations and during Episodic/Non-Episodic events. The mean qualitative chemical composition was useful in recognizing the importance of different sources in Episodes and Non-Episodes. The influence of the road dust source was associated with a dominance of silicon, aluminum and magnesium while predominance of carbon indicated the contribution from combustion sources. The presence of sulphur in the particulates was expected considering the industrial sources present in Prince George, however, the amount of sulphur in the Non-Episodes was slightly higher than in the Episodes suggesting that sulphur 95 particulates are constantly present in the ambient air. The presence of sulphur in the fine fraction (which dominate non-episodes) may have health implications. It is unclear whether the particulates themselves originate from a specific source or the PMto is interacting with sulphur aerosols to form sulphur coated PMto. The correlation of mean particle diameter and chemical composition revealed very weak relationship suggesting that the Prince George PMw is reasonable uniform chemically in all size ranges. The qualitative nature of the chemical composition may have affected the relationships. The comparison of morphology and chemical composition revealed some relationships between morphological shapes seen in the ambient PMto and chemical composition. The episodes examined indicated that percentages of silicon, aluminum, and magnesium in amorphous particles were larger in those Episodes dominated by road dust. The rectangular shapes contained very high levels of sulphur and calcium indicative of an industrial source. All the morphological shapes identified except (smooth-flat) contained sulphur suggesting that there is an interaction occurring between sulphur dioxide (S0 2) which is coating the fine particulates in the ambient air. If sulphur is being transported with the fine particulates it may be causing health impacts additional to those caused by PMto. The above trends with respect to morphology, particle size, particle size distribution, and chemical composition were also present in the BCR site. The dominance of the road dust source was especially evident in the BCR episodes. Contribution from Various Sources to Ambient PMto Composition The final objective of this study was to determine the contribution of various sources during Episodic/Non-Episodic events. Principal Component Analyses (PCA) show four discernable sources contributing to the ambient PMw : Road Dust, Industrial, Combustion, and 96 Salt. These sources were not identical in elemental loadings throughout the various PCA due to variability in source composition and meteorological conditions. The particulate emitted from a source often undergoes changes due to temperature, relative humidity, and the presence of aerosols which may react with it. The four main sources (factors) were identified by interpreting the pattern and extent of loadings of particular elements and the correlation between loadings (positive/negative). Most ofthe Episodes analyzed were dominated by various types of road dusts. The BCR site Episodes were characterized by high levels of road dust. Episode 1 (950122) for the bowl area and the Non-Episodes, contained more particles of anthropogenic origin (industriaVcombustion). Generally, Non-Episodes have more distinct sources ofPM10 compared to the Episodes because road dust is less dominant. The salt factor could be a result of several different sources. The salt could be a result of either industrial sources or winter salting applications. The combustion source has to be considered a combination of all possible combustion sources (beehive burner, vehicles, fireplace burning, etc ... ). Study of organic particulates would be required to distinguish between these sources. The combined results of the various analyses indicate it is possible to determine source apportionment using the microscopic techniques described in this study. The combined use of morphological, particulate diameter, and particulate elemental composition can be used to distinguish between road dust and industriaVcombustion sources present in the PM10 in the Prince George Airshed. 97 RECOMMENDATIONS FOR FUTURE STUDY 1. In order to expand the knowledge about the sources and the ambient PM10 further studies are required. Any analysis using ICP would be much more successful if a different filter type was used during the collection. The glass fiber filter normally used by the Ministry of the Environment contributes too much contamination for quantitative analysis. A cellulose or pure teflon filter should be used for future analysis (Chow, 1995). In order to examine the different size fractions quantitatively, a cascading or dichotomous collector could be incorporated into sampling procedure. 2. Future definition ofthe organic portion (examination for tracer compounds unique to specific sources) of ambient PM10 would help to characterize combustion sources and their contributions to total PM10. This analysis would be most successful if glass fiber filters and foam (PUP) were used to trap the volatile and solid organic PM10. 3. For a complete study ofPM10 in the Prince George airshed, concurrent sampling using Teflon filters (Microscopic), Glass fiber filters (Organic), and Cellulose filters (Elemental- ICP) would produce a complete characterization of the ambient PM10 for specific periods of time. 4. Further analysis of the PM10 incorporating organic composition in the BCR site should be considered due to the high levels ofPM10 in the area. Further definition of source apportionment in this area would provide useful information that could be applied to reduction strategies. There are a considerable number of people working in that area being exposed to these PM10 levels that are considered detrimental to health. Serious consideration should be given to decreasing the PM10 levels by paving roads. 5. Improved source profiles of the major PMlO contributors using organic and elemental analyses would be useful in future source apportionment. 98 6. A health study examining the effects ofPMw on health in the Prince George area would be useful. This study could be incorporated into the complete study ofPMto (Recommendation #3) which would allow researchers to compare levels ofPMw over a long period oftime with health indicators. 99 LITERATURE CITED Alpert,D.J. & Hopke P .K.1981 . A determination of the sources of airborne particles collected during the regional air pollution study. Atmospheric Environment .V15:No5 .pp675-687. B.C. Environment. Methodology Analysis: Total Particulate- PM10- HiVol:5305 . B.C. Environment. 1998. Meteorological Data from Provincial Database: Victoria, B.C. Boubel,R.W. 1968. Particulate Emissions from Sawmill Waste Burners. Bulletin #42. Engineering Experiment Station: Oregon State University: Corvallis, Oregon. Brady,N .C.1996. The nature and properties of soils: 11th Edition. Prentice Hall: Upper Saddle River, N .J. 740pp. Bridgman,H.1990. Global Air Pollution problems for the 1990' s.Belhaven Press:Pinter Publishers.London (TD883 .b74 1990) Cariboo Health Unit.1994.Aerosol Characterization.CHU#15 .Williams Lake. Chow,J.C., Liu,C., Cassmassi,J., Watson,J., Lu,Z ., & Pritchett,L.1992. A Neighbourhood-Scale Study ofPM1 0 Source Contributions in Rubidoux, California. Atmospheric Environment. V26A:No4.pp693-706. Chow,J.C.1995.Measurement Methods to Determine Compliance with Ambient Air Quality Standards for Suspended Particles. Journal of Air & Waste Management Association. V45.pp320-382. Comrey,A.L. & H.B.Lee.1992. A first course in factor analysis :Second Edition. Hillsdale,NJ: Erlbaum. Dawson,A.B. 1989. Soils of the Prince George- McLeod Lake Area. British Columbia Soil Survey 0840-9730; Report 23 . Ministry ofEnvironment and Parks: Victoria. Dockery,D .W. & C.A.Pope.1994. Acute Respiratory Effects ofParticulate Air Pollution. Annual Review Public Health.V15 .pp107-132. ~ 1 5 DX-4 Users Manual: Revision 5. EDAX International, Mahwah, N.J. EPA.1984. The Research behind a clean air proposalVlO (May).pp29-31. Econornist.1995. The way to Dusty Death.Feb.18.pp82-83 . Environment Canada. 1994-1996. Monthly Meteorological Summaries- Prince George Airport. Atmospheric Environment Service: Ottawa. 100 Evans,J.S . & Cooper,D .W.1980. An inventory of particulate emissions from open sources. Journal of the Air Pollution Control Association.V3 0 :No 12.pp 1298-13 03. Fisher,G.L. , Prentice,B.A, Silberman,D. , Ondov,JM., Bierman,AH., Ragaini,RC. , &ARMcFarland. 1978. Physical and Morphological studies of size-classified Coal Fly Ash. Environmental Science & Technology. V12 :No4:pp447-451. French.H.F.1990. Worldwatch Paper 94: Clearing the Air: A Global Agenda. Worldwatch Institute. 1776 Massachusetts Avenue, N .W. Washington D .C. 20036 USA (TD.883 .f69.1990). Harnilton,RS., Kershaw,P ., Segarra,F., Spears,C. , & Watt,J1994. Detection of airborne carbonaceous particulate matter by scanning electron microscopy. The Science of the Total Environment V146/147.pp303-308. Harley,RA, Hunts,S., & Cass,G.1989. Strategies for the Control of particulate air quality: LeastCost Solutions based on receptor-oriented models. Environmental Science & Technology. V23 :No8 .pp 1007-1014. Harman,JN.1989. ICP Emission Spectroscopy.p89-92. in JP.Lodge (ed.) Methods of Air Sampling and Analysis: Third Edition: Lewis Publishers, Inc: Chelsea, Mi. pp763 . Hileman,B.1981. EST Outlook Particulate Matter: The inhalable variety. Environmental Science & Technology. V15 :No9.pp983-986. Hopke,P .K , Lamb,RE. , & D.F.S.Natusch. 1980. Multielemental Characterization ofUrban Roadway Dust Environmental Science & Technology. V14 :No2:ppl64-172. Infante,R& Acosta,I.1991 . Size distribution of trace metals in Ponce, Puerto Rico air particulate matter. Atmospheric Environment V25B:No 1. pp 121-131. Kao,AS . & Friedlander,S.K1995. Frequency Distribution ofPM10 Chemical Components and their sources. Environmental Science & Technology. V29 :Nol.ppl9-28 . Kartal,S., Dogan,M., Rojas,C. , & Grieken,R1993. Composition and sources of atmospheric particulate matter at Kayseri, central Turkey. The Science of the Total Environment V133 .pp83-97. Karue,J., Kinyua,A, & El-Busaidy,A1992. Measured Components in Total Suspended Particulate Matter in a Kenyan Urban area. Atmospheric Environment V26B :No4.pp505-511. Kaufherr,N ., & D.Lichtman. 1984. Comparison ofMicron and Submicron Fly Ash Particles using Scanning Electron Microscopy and X-Ray elemental Analysis. Environmental Science & Technology. V18 :No7 :pp544-547. 101 Keyser,T.R. , Natusch,D.F.S., Evans,Jr,C.A., & R.W.Linton. 1978. Characterizing the surfaces of environmental particles. Environmental Science & Technology. V12 :No7 :pp768-773 . Kim,D .S., Hopke,P .K. , Massart,D .L., Kaufinan,L. , & G.S. Casuccio. 1987. Multivariate Analysis of CCSEM Auto Emission Data. The Science of the Total Environment. V 59.pp 141-15 5. Kowalczyk, G., Gordon,G.E ., & Rheingrover,S.1982. Identification of Atmospheric particulate sources in Washington D .C. using Chemical Element Balances. Environmental Science & Technology. V16 :No2.pp79-90. Lewis,C.W., Baumgardner,R. , & Stevens,R.1988. Contribution ofWoodsmoke and Motor Vehicle emissions to ambient aerosol mutagenicity. Environmental Science & Technology. V22 :N o8 .pp968-971 . Li,C-S, Hsu,L-Y. , Chuang,Y-Y.1993. Elemental profiles ofindoor and outdoor particulate matter less than 10um (PM10) and 2.5um (PM2.5) in Taipei. Chemosphere.V27 :No1l.pp2143-2154. Lichtman,D. & Mroczkowski,S.1985. Scanning electron microscopy and energy- dispersive X-Ray spectroscopy analysis of submicrometer coal fly ash particles. Environmental Science & Technology. V19:No3 .pp274-277. Linton,R.W., Farmer,M .E ., Hopke,P .K., & Natusch,D.F.S.1980. Determination ofthe sources oftoxic elements in environmental particles using microscopic and statistical analysis techniques._ Environment International. V4.pp453-461. Lowenthal,D.H., & Rahn,K.1987. A Quantitative Assessment of Source Contributions to Inhalable particulate matter in Metropolitan Boston. Atmospheric Environment. V21:Nol.pp257-265 . Mage,D .T.1985. Concepts ofHuman exposure assessment for airborne particulate matter. Environmental International. V11.pp407-412. Mendenhall,W., & R.J.Beaver.1991. Introduction to Probability and Statistics: Eighth Edition. PWSKent Publishing Company: Boston.pp716. Ministry of the Environment, Lands, & Parks (MELP).1995. Prince George Air Quality Management Background Report. British Columbia. Ministry ofEnvironment Lands & Parks (MELP)- Air Resources Branch. 1997. Air Quality Report for British Columbia: Fine Particulate (PM10) levels (1990-1995); Victoria, B.C. OECD - Organization for Economic Co-operation and Development.1995. Motor Vehicle Pollution Reduction strategies beyond 2010.0ECD. 102 Oke,T.R. 1987. Boundary Layer Climates. 2nd edition. Methuen :London. 435pp. Ostro,B.D., Lipsett,M ., Wiener,M, & Selner,J.1991. Asthmatics responses to Airborne Acid Aerosols. American Journal of Public Health. V81 :no6. pp694-702. Pierson, W .R. & Brachaczek, W .W.1983. Particulate Matter Associated with Vehicles on the road.II. Aerosol Science and Technology. V2.ppl-40. Post,J.E., & P .R. Buseck. 1984. Characterization of Individual Particles in the Phoenix Urban Aerosol using Electron- Beam Instruments. Environmental Science & Technology. Vl8 :Nol.pp35-42. Prince George Airshed Technical Management Committee (PGATMC). 1996. Prince George Air Quality Management Background Report. Prince George, British Columbia:66pp. Purghart,B.C., Nyffeler,U. , Schindler,P ., Van Borrn, W., & Adams,F.1990. Metals in Airborne Particulate Matter in Rural Switzerland. Atmospheric Environment. V24A:No8 .pp2191-2206. Schlesinger,R.B. 1990. The Interaction oflnhaled Toxicants with Respiratory Clearance Mechanisms. Critical Reviews in Toxicology. V20.pp257-286. Schroeder,W .H., Dobson,M ., Kane,D ., & Johnson,N .1987. Toxic Trace Elements associated with airborne particulate matter: A review. Journal of the Air Pollution Control Association.V37 :Noll.ppl267-1285 . Spengler,J.D., Treltman,R. , Tosteson,T. , Mage,D., & Soczek,M.1985. Personal exposures to respirable particulates and Implications for air pollution epidemiology. Environmental Science & Technology. Vl9 :No8 .pp700-707. Stevens,R.K.1985. Sampling and analysis methods for use in source apportionment studies to determine impact of wood burning on fine particulate mass. Environment International.Vll.pp271-283 . Sutherland,D .1998. Personal Communication. Swift,D .L. & D .F.Proctor. 1982. Human Respiratory Deposition ofParticles During Oronasal Breathing. Atmospheric Environment. Vl6 :No9.pp2279-2282. Tabachnick,B.G., & Fidell,L.S .1996. Using Multivariate Statistics: Third Edition. HarperCollins College Publishers: New York: New York. pp880. Valtink,P . & Liegmahl,H.1989. Analysis of traffic-induced airborne particulate matter with Energy Dispersive X-Ray Fluorescence Spectrometry EDXRF. Journal ofEnvironmental Science and Health. V A24 :No7.pp679-693 . 103 VanBorm,W.A. & Adams,F .C.1988. Cluster Analysis ofEiectron Microprobe Analysis data of individual particles for Source Apportionment of Air Particulate Matter. Atmospheric Environment .V22 :No 10. pp2297-2307. Vedai,S. 1995. Health Effects oflnhalable Particles: Implications for British Columbia.Prepared for Air Resources Branch, BCMELP. Ministry ofEnvironment, Lands, and Parks. UBC. Vedal,S. 1996. Evaluation ofHealth Impacts Due to Fine lnhalable Particles (PM2.5). Prepared for Health Canada. UBC. Vancouver Hospital and Health Sciences Center. Warren,C.J., Xing,X., & Dudas,M.J. 1990. Simple Microwave Digestion Technique for Elemental Analysis ofMineral Soil Samples. Canadian Journal of Soil Science. V70 :pp617-620. Williams,D.J., Milne,J., Roberts,D ., & Kimberlee,M.1989. Particulate Emissions from 'In-Use' motor vehicles- I. Spark Ignition Vehicles. Atmospheric Environment.V23 :Nol2. pp2639-2645. Williams,D .J., Milne,J., Quigley,S., Roberts,D ., Kimberlee,M.1989 . Particulate Emissions from 'In-Use' motor vehicles- II. Diesel Vehicles. Atmospheric Environment.V23 :Nol2.pp2647-2661. Xhoffer,C., Bernard,P ., Grieken,R., & Auwera,L.1991. Chemical Characterization and Source Apportionment of Individual Aerosol Particles over the North Sea and the English Channel using Multivariate Techniques. Environmental Science & Technology. V25 :No8 .ppl470-1478. Zumbo,B.D., & D .Coulombe.1997. Investigation of the Robust Rank-Order Test for Non-Normal Populations with Unequal Variances: The Case ofReaction Time. Canadian Journal of Experimental Psychology. V5l:No2.ppl39-149. 104 -34.5 to -19 -18.1 to -0 .5 - 14 to -4 -1.7 to 6.9 BCR I Bowl Non-Episode 1 Bowl Episode 3 Bowl Non-Episode 3 I BCR Episode BCR I Bowl Non-Episode 3 BCREpisode January 22,1996 February 27, 1996 March 4, 1996 May 9, 1996 August 13, 1996 (B. C. Environment, 1998; Environment Canada 1994-1996) - - - - 0-5 .7 5 to 21 BCREpisode August 31 , 1995 8 to 22 0.1-1.7 -1 to 14 Bowl Episode 2 I BCR Episode March 28, 1995 0.3-3 .7 0.3-5.7 1-6.2 1.0-4 0-3 0-3 March 16, 1995 0-3 -2 to 12 Bowl Episode I January 21 ,1995 12 to 20 0-5 .5 BCREpisode BCREpisode September 23 ,1994 -2 to 13 Wind Speed (m/s) 0-1 BCREpisode AprilS, 1994 oc Temperature -10 to -7 Episode I Non-Episode BCR I Bowl Dates TABLE 1: Meteorological Conditions on Study Dates Appendix A South North North North North South North South-West NIA South North Wind Direction No indication of inversion 105 Excellent air circulation, precipitation Very good air circulation Night-time cooling may have caused an inversion, dissipating by lOAM No indication of inversion No indication of inversion Night-time cooling may have caused inversion, dissipating by 9AM Night-time cooling and calm winds may have caused inversion, dissipating by 7 AM Conditions probably promoted stable boundary layer - causing inversion Night-time cooling may have caused inversion, strong wind speeds during day Wind speeds high, no evidence of inversion Interpretation ! Appendix B: Morpbological Characterization AMORPHOUS OVAL 106 ROUND SPHERE 107 FLAT 108 CUBE RECTANGLE 109 ROD 110 APPENDIX C: Data from Carbon Coated Sample (950121 Van Bien) Particulate Sodium Magnesium Aluminum Silicon Sulphur 2d1 63.14 0.00 0.00 15.51 12.84 8.75 38.95 0.00 0.00 52.30 2d2 27.49 5.68 5.50 0.00 38.54 2d3 13 .76 13 .98 0.00 0.00 66.99 2d4 46.04 0.00 9.64 0.00 44.33 2d5 21.16 0.00 0.00 70.26 0.00 2d6 45.41 0.00 54.59 0.00 0.00 2d7 41.84 1.89 26 .44 3.90 25 .94 2d8 10.21 23.93 0.00 0.00 16.19 2d9 50.96 2.13 7.58 0.00 37.23 2d10 37.59 0.00 0.00 0.00 62.41 2dll 14.52 14.32 0.00 0.00 63 .3 1 2d12 41.60 5.97 11.66 0.00 40.76 2d13 33.81 12.14 5.20 0.00 43 .31 2d14 51.58 0.00 0.00 8.46 38.17 2d15 2.64 79.18 5.28 0.00 12.89 2d16 28.98 6.14 5.05 0.00 55 .05 2d17 21.35 13.77 0.00 0.00 57.30 2d18 9.65 12.18 0.00 0.00 78.17 2d19 8.25 46 .71 3.17 0.00 39.71 2d20 17.09 12.29 0.00 0.00 64.94 2d21 45.03 2.43 7.15 0.00 43 .83 2d22 11.43 29.91 0.00 0.00 36.89 2d23 34.31 8.84 5.21 0.00 47.58 2d24 52.67 3.72 8.69 0.00 32.20 2d25 31.36 8.18 0.00 5.87 49.50 2d26 32.22 13.66 0.00 0.00 54.12 2d27 26.35 18.45 4.78 0.00 40.38 2d28 35 .89 10.73 0.00 0.00 53 .38 2d29 3.80 44.32 0.00 0.00 15 .36 2d30 44.29 4.39 7.31 0.00 40.79 2d31 5.95 32.81 9.57 0.00 47.93 2d32 8.95 44.70 0.00 0.00 45 .02 2d33 45.14 0.00 8.37 0.00 46.49 2d34 42.21 2.55 7.68 0.00 45.43 2d35 35.49 0.00 0.00 8.23 56 .28 2d36 20.34 27.04 3.05 0.00 27.60 2d37 1.34 45 .17 0.00 0.00 53.48 2d38 21.45 16.06 0.00 0.00 56.29 2d39 7.93 54.50 4.51 0.00 29.34 2d40 20.04 9.17 0.00 3.58 61.90 2d41 34.13 5.99 0.00 51.32 6.09 2d42 33 .23 5.50 5.67 0.00 52.68 2d43 7.22 49.37 1.79 0.00 38.59 2d44 7.46 0.00 46.90 0.00 44.07 2d45 26.71 10.94 0.00 0.00 58.86 2d46 30.56 13 .83 6.18 0.00 43 .58 2d47 44.22 8.36 0.00 6.96 37.13 2d48 31.23 4.99 4.92 5.71 51.69 2d49 Potassium 8.51 0.00 2.44 5.28 0.00 8.58 0.00 0.00 0.00 2.10 0.00 7.85 0.00 5.55 1.80 0.00 4.78 7.58 0.00 2.15 5.68 1.57 14.53 4.06 2.72 5.10 0.00 1.98 0.00 13 .61 3.21 3.74 1.33 0.00 2.13 0.00 11.79 0.00 6.20 3.72 5.32 2.46 2.92 3.02 1.57 3.50 5.85 3.34 1.46 Calcium 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 49.67 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 7.25 0.00 0.00 0.00 0.00 8.06 0.00 22 .91 0.00 0.00 0.00 0.00 0.00 0.00 10.19 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Titanium 0.00 0.00 20.35 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Iron 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 111 2d50 2d51 2d52 2d53 2d54 2d55 2d56 2d57 2d58 2d59 2d60 2d61 2d62 2d63 2d64 2d65 2d66 2d67 2d68 2d69 2d70 2d71 2d72 2d73 2d74 2d75 2d76 2d77 2d78 2d79 2d80 2d81 2d82 2d83 2d84 2d85 2d86 2d87 2d88 2d89 2d90 2d91 2d92 2d93 2d94 2d95 2d96 2d97 2d98 2d99 2dl00 49.03 25.25 67.84 62.23 35.43 43.38 46.64 45.94 44.04 31.91 32.83 29.14 71.65 35.36 48.44 29.96 37.88 50.24 15.70 49.52 28.69 26.70 22.60 19.82 8.87 36.24 18.33 13 .26 16.78 20.59 16.53 29.91 51.25 65 .58 57.55 61.84 46.05 45 .78 57.08 32.39 47.93 58.69 17.25 54.33 26.04 63 .75 33.30 28.29 61.72 67.30 66.65 0.00 33 .73 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 35.61 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 6.41 16.79 0.00 0.00 5.12 7.76 6.97 7.29 7.67 7.25 0.00 0.00 0.00 7.64 6.73 2.14 6.92 8.42 0.00 6.11 0.00 2.80 0.00 0.00 25 .87 7.90 1.95 0.00 7.59 0.00 8.28 8.25 0.00 0.00 0.00 0.00 7.47 0.00 0.00 4.46 0.00 0.00 0.00 0.00 3.68 0.00 0.00 2.94 0.00 0.00 0.00 36.63 15.21 13 .68 0.00 48.11 48.87 41.11 43.42 44.16 33.00 2.98 7.24 6.54 51.50 37.29 15.07 41.34 40.06 6.15 38.33 6.08 16.37 4.72 0.00 25.49 47.31 11.21 9.71 62.84 7.91 65 .87 50.08 48.75 21.54 29.45 26.50 42.55 0.00 37.80 25.46 35.47 11.48 8.33 15.34 66.52 27.45 8.26 20.78 38.28 11.51 18.53 4.28 4.46 12.23 37.77 8.34 0.00 3.19 1.77 2.66 20.98 35 .09 41.85 13.84 3.17 5.38 32.04 10.04 0.00 42.40 6.04 35.72 32.14 39.73 46.19 2.05 5.74 36.66 42.57 4.21 37.49 3.40 7.92 0.00 7.64 9.56 11.67 2.29 33.67 5.11 18.00 16.60 21.51 40.33 30.32 3.76 8.80 34.47 23 .56 0.00 15.87 9.92 3.65 1.29 6.24 0.00 3.00 0.00 2.09 1.58 1.47 6.85 1.76 0.00 7.96 2.33 2.16 12.43 3.82 1.28 14.33 0.00 18.47 12.24 21.50 16.81 0.52 2.80 9.87 16.00 5.92 17.36 4.18 3.84 0.00 5.23 3.44 0.00 1.65 20.55 0.00 9.43 0.00 8.32 12.58 0.00 0.00 0.00 12.24 14.02 0.00 5.32 4.90 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 27.34 21.77 0.00 0.00 0.00 8.36 0.00 0.00 21.43 0.00 11.04 9.76 11.46 17.18 0.00 0.00 21.97 18.46 2.66 16.65 1.75 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 10.26 0.00 0.00 21.51 0.00 0.00 0.00 11.73 10.41 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.26 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 112 APPENDIX D: Blank Teflon Filter Particulate# Carbon Oxygen Fluorine Sodium Aluminum Silicon Potassium 0.00 8.53 1.20 50.57 8.65 27.45 3.60 TBl 0.24 8.59 9.04 1.31 27.48 50.06 3.28 TB2 0.25 7.18 9.84 1.66 33 .32 45.00 2.75 TB3 8.02 0.26 1.22 8.56 52.10 26.37 3.46 TB4 10.66 0.42 1.48 46.33 8.80 29.36 2.95 TBS 7.10 0.26 0.93 60.89 5.99 19.83 5.00 TB6 9.96 0.41 1.61 27.28 49.20 8.60 2.94 TB7 6.14 0.00 6.44 0.92 60.73 4.44 21.33 TB8 0.43 4.10 0.00 3.74 16.72 69.64 5.37 TB9 0.25 7.82 1.23 8.75 30.08 49.24 2.63 TBlO 5.77 0.22 8.32 1.26 54.17 3.35 26.92 TBll 0.27 1.42 8.61 9.69 35 .14 42.38 2.49 TB12 5.67 0.15 1.06 29.78 53.18 8.26 1.92 TB13 5.07 0.14 4.27 0.64 19.32 66.18 4.40 TB14 14.38 0.17 7.23 2.19 10.15 65.67 0.20 TB15 10.56 0.36 11.34 1.82 34.71 40.18 1.03 TB16 8.59 0.3 1 1.29 7.96 28.46 50.65 2.74 TB17 9.32 0.36 48.08 8.30 1.40 2.70 29.85 TB18 1.74 11.16 0.43 42.41 9.97 32.39 1.90 TB19 8.58 0.31 7.90 1.26 24.89 53 .72 3.34 TB20 13 .91 0.70 2.01 38.97 32.88 10.16 1.38 TB21 1.54 10.61 0.25 30.64 9.3 1 1.04 46.61 TB22 20.15 0.71 16.70 10.72 3.16 0.51 48.04 TB23 0.28 10.74 1.62 9.11 32.45 43 .33 2.47 TB24 12.13 0.49 48.99 23 .04 11.85 1.86 1.64 TB25 0.35 1.74 9.90 10.09 1.85 31.63 44.45 TB26 9.96 0.3 5 35.05 11.72 1.63 1.78 39.50 TB27 9.79 0.28 27.66 12.55 1.87 1.21 46.64 TB28 1.85 11.66 0.40 37.04 36.87 10.73 1.45 TB29 11.05 0.45 32.34 42.60 9.76 1.81 2.00 TB30 7.30 1.69 12.15 0.72 27.74 47.21 3.19 TB31 7.21 0.23 48.29 9.26 1.11 2.64 31.26 TB32 9.48 0.30 43 .51 33 .26 10.41 1.59 1.46 TB33 2.11 12.46 0.50 10.89 37.90 34.58 1.57 TB34 48.78 24.82 12.25 1.74 10.99 0.40 1.01 TB35 1.75 9.79 0.37 2.33 33.45 41.54 10.78 TB36 7.22 1.29 9.56 0.40 3.19 25 .26 53.09 TB37 1.73 0.47 36.14 38.65 10.02 11.71 1.28 TB38 7.33 1.43 10.71 0.48 25.41 51.43 3.19 TB39 13.95 10.82 2.12 0.50 0.64 53.05 18.92 TB40 2.01 12.84 0.54 27.01 46.51 8.64 2.46 TB41 0.72 5.95 0.15 16.14 66.82 4.03 6.19 TB42 2.56 15.46 0.29 0.50 52.85 19.19 9.16 TB43 1.04 8.06 0.23 36.38 42.86 9.15 2.29 TB44 4.89 18.78 64.24 6.43 0.80 0.16 4.70 TB45 8.39 1.13 9.85 0.30 2.58 32.37 45 .3 9 TB46 7.24 2.27 14.96 0.28 40.03 1.62 33 .61 TB47 2.15 14.26 0.33 1.04 38.56 34.99 8.68 TB48 0.75 6.03 18.08 64.61 5.36 0.19 4.99 TB49 1.07 8.25 0.28 23 .53 55 .57 7.3 1 3.98 TBSO 6.96 0.18 23.96 56.14 8.02 1.01 3.73 TB51 2.20 15.07 0.34 60.17 13.39 8.05 0.77 TB52 113 TB53 TB54 TBSS TB56 TB57 TB58 TB59 TB60 TB61 TB62 TB63 TB64 TB65 TB66 TB67 TB68 TB69 TB70 TB71 TB72 TB73 TB74 TB75 TB76 TB77 TB78 TB79 TB80 TB81 TB82 TB83 TB84 TB85 TB86 TB87 TB88 TB89 TB90 TB91 TB92 TB93 TB94 TB95 TB96 TB97 TB98 TB99 TB100 Mean SD 3.85 4.45 1.06 1.88 1.76 2.08 0.89 3.02 4.01 3.51 2.24 5.14 1.59 3.05 3.00 1.65 3.56 0.92 0.97 3.83 2.03 1.48 2.00 6.83 1.40 1.33 4.60 0.78 4.06 1.14 1.00 1.87 1.06 1.93 1.27 1.77 0.81 2.11 3.97 1.81 1.70 3.95 1.32 1.65 0.78 0.73 1.04 2.62 2.41 1.37 24.83 22.83 26.83 28.00 39.66 31.34 32.34 29.18 23 .92 24.24 35 .84 16.95 42.86 25.83 26.54 38.83 25 .23 29.39 35 .57 19.58 30.68 41.17 30.40 9.15 25 .68 47.21 23 .31 45.82 24.39 34.11 39.41 37.08 43 .56 33.60 48.05 30.33 52.33 28.14 20.46 34.95 34.07 20.43 45 .02 32.51 38.01 54.66 45.42 28.49 33.03 10.35 52.87 58.20 44.42 44.51 34.03 41.16 31.10 46.72 56.24 57.29 41.14 64.98 30.78 50.11 51.64 34.89 54.19 39.07 34.67 63 .26 46.3 1 31.93 36.14 78.19 49.34 26.04 56.12 22.76 55 .98 40.14 32.17 37.93 27.01 43 .99 26.14 42.50 18.54 45 .96 61.99 39.69 35 .79 61.03 26.09 39.46 30.48 13 .69 24.58 50.39 42.89 13.79 7.34 6.82 7.36 8.82 10.99 9.63 8.17 9.92 8.20 8.08 9.91 5.95 11.19 8.12 8.83 10.46 8.27 10.79 11.61 7.04 10.07 11.92 10.09 2.67 8.45 11.92 8.01 11 .54 8.26 11.48 12.28 10.09 10.69 10.20 13 .09 9.43 14.26 10.14 6.99 11.40 9.72 7.84 12.65 8.96 9.46 13.46 12.45 8.34 9.19 2.12 1.54 0.84 2.26 1.78 1.62 2.06 3.05 1.45 1.05 0.93 1.66 0.89 1.63 1.47 1.44 1.79 1.23 2.73 2.32 0.97 1.61 1.86 2.69 0.38 1.95 1.70 1.15 2.49 1.14 1.95 1.94 1.67 2.17 1.34 1.52 1.71 2.10 1.81 0.92 1.67 2.40 1.17 2.05 2.07 2.74 2.30 2.19 1.36 1.62 0.55 9.30 6.64 17.14 14.31 11.52 13.21 23.11 9.39 6.36 5.77 8.94 5.92 11.55 10.86 8.30 11.94 7.28 16.34 14.27 5.13 8.99 11.26 17.94 2.77 12.58 11.37 6.56 15.94 5.98 10.94 12.66 10.91 14.73 8.66 9.78 13.56 11.57 ll.41 5.43 10.13 15.77 5.47 12.41 14.56 17.74 14.66 13.86 8.51 10.48 3.69 0.28 0.23 0.94 0.70 0.42 0.53 1.35 0.32 0.23 0.18 0.26 0.17 0.40 0.57 0.25 0.45 0.24 0.77 0.59 0.20 0.30 0.38 0.74 0.00 0.60 0.44 0.25 0.67 0.19 0.22 0.53 0.45 0.79 0.28 0.15 0.70 0.39 0.43 0.24 0.35 0.56 0.13 0.47 0.78 0.79 0.50 0.46 0.28 0.38 0.22 114 WQB-11 Recovery (%) I 0.004 0.266 0.00 1 I N/A N/A N/A 0.004 138.000 I 0.3 51 0.003 178.000 0.002 30.900 0.263 I 0.252 0 .750 7.581 I 0.252 87.000 0.780 70.000 I 4.745 0.266 N/A 13.817 2.694 88.000 I 0.263 79.000 N/A 0.256 0.001 102.000 N/A N/A 0.089 N/A 0.140 0.076 47.900 51.300 N/A 0.00 1 0.009 0.003 153.000 0.003 200.000 450.000 0.010 0.006 101.900 0.003 116.000 76.200 Appendix E: Standard Recoveries for Elemental Analysis (ICP) N/A 0.051 0.003 20.500 0.005 24.800 39 .000 4.780 2.191 92 .000 1.421 94.000 96.000 N/A 0.228 0.003 90.000 N/A 0.002 89.000 0.23 6 0.015 87.000 O.Oll 95.000 90.500 0.013 105.000 0.009 96.000 N/A 0.013 42.900 N/A 0.000 N/A 9.900 0.000 0.005 0.385 0.052 87.000 0.046 88.000 89.400 99.800 0.232 0.005 89.000 N/A 0.048 88.300 l15 Appendix F: Teflon Blank for Quantitative Elemental Analysis (ICP) TABLE3 : A AI ppm 161.132 Cr ppm 0.152 Li ppm 0.174 Ni ppm 0.064 Sr ppm 3.542 Zr ppm 0.725 ~ SD 74.005 SD 0.053 SD 0.066 SD 0.026 SD 3.542 M eans/St an d ar d D ev1a . f 100 ~or Bl an k F·tt 1 er Ba Ca Cd ppm ppm ppm SD SD SD 87.464 37.915 202.595 89.721 -0.003 0.001 K Cu Fe ppm ppm ppm SD SD SD 0.195 0.038 4.486 2.230 52.948 20.983 Mg Mn Na ppm ppm ppm SD SD SD 39.720 18.913 -0.115 0.048 142.830 60.983 p Si Sn ppm SD ppm ppm SD SD 1.657 0.529 383 .066 159.768 0.389 0.389 Ti Zn v ppm ppm ppm SD SD SD 5.411 5.411 0.192 0.192 78.325 78.325 SD 0.725 116 Appendix G: PCA TABLES by Location TABLE 4: PCA Eigenvalues and Primary Factors: Episode 1-950121 Plaza 3 4 2 Factor 1 Barium Road Dust Industrial Road Dust Iron oxides Sulphur Source Mica -0.29 1415 0.126817 0.704569 -0.341755 Aluminum 0.072874 0.033 444 0.066444 0.916084 Barium -0.1571 26 -0. 174712 0.045186 Calcium 0.816396 0.065066 0.114193 -0.818742 -0.499798 Carbon -0.024507 -0.060664 0.033827 -0.751883 Iron 0.053575 -0. 110194 0.023942 0.05 1904 Magnesium 0.289539 0.020078 0.137959 Potassium 0.675509 0.132592 0.16291 -0.271 96 0.826889 Silicon 0.286462 -0.137363 0.0304 195 -0.427855 Sodium -0.231559 -0.102658 0.03847 0.915337 Sulphur 0.09 1795 0.037967 -0.007045 Titanium -0.661648 2.610731 1.892325 1.46728 1.050734 Eigenvalue 23 .73 17.2 13 .34 9.55 % Total Variance 23 .73 40.94 54.58 63 .83 Cumulative % For explanation see Table 9 5 Road Dust Magnesium oxides 0.179826 -0.086137 0.104452 0.150504 -0.189504 0.869051 -0.029671 -0.041536 -0.493042 0.045182 0. 189835 1.004492 8.13 72.96 TABLE 5: PCA Eigenvalues and Primary Factors: Etlisode 1-950121 Van Bien Factor 1 3 2 4 Industrial Road Dust Road Dust Iron Sulphur Source Na-Feldspar Magnesium oxide 0.136162 -0.675716 -0.127691 0.265586 Aluminum Calcium -0.927375 0.197159 0.01043 1 0.060004 0.070113 Carbon 0.602207 0.759376 0.192738 0.005733 0.042788 -0.117756 -0.795669 Copper 0.02 1705 0.055355 -0.029029 Iron -0.901726 0.015225 -0.106091 Magnesium -0.797655 0.112806 Potassium -0.899888 0.018277 0.020786 -0.001237 0.269251 Silicon -0.825731 0.022596 -0.00513 Sodium 0.422204 -0.556101 0.111025 -0.377627 Sulphur -0.95416 0.219 181 0.026998 0.027521 3.148544 Eigenvalue 1.957879 1.296073 1.068036 34.19 19.58 % Total Variance 12.96 10.68 34.19 Cumulative % 53.76 66.73 77.41 For explanation see Table 9 117 TABLE 6:PCA Eigenvalues and Primary Factors: Episode 1-950121 Lakewood 3 2 Factor 1 Industrial Road Dust Road Dust Sulphur Source Na-Feldspar Na-Feldspa r -0.093375 0.169039 0.849428 Aluminum -0.165572 -0.090432 -0.898782 Calcium -0.189169 0.32444 -0.9151 Carbon 0.093848 0.046767 0.051733 Iron -0.007997 -0.05007 0.891158 M agnesium 0.051225 0.013595 -0.921371 Potassium 0.200556 -0.241376 Silicon 0.856076 -0.001532 0.632275 0.480198 Sodium -0.223275 0.307036 -0.857012 Sulphur 1.167142 2.933013 2.482965 Eigenvalue 32.59 27.59 12.97 % Total Variance 73 .15 32.59 60.18 Cumulative % For explanation see Table 9 . . d e 2- 950328 PI aza TABLE 7: PCA E'agenva ues an dP nmary Fac t ors: E;JllSO Factor 1 2 3 4 Road Dust Industrial Road Dust Road Dust Quartz Sulphur Source Iron oxide K-Feldspar -0.055483 0.050021 0.298079 -0.802033 Aluminum 0.005 18 0.038137 0.07648 Calcium -0.781587 0.243803 0.112771 0.624336 Carbon 0.483786 -0.066409 -0.063677 0.036857 Chlorine 0.573602 0.057672 0.23109 -0.15419 0.40554 Iron 0.090861 -0.020866 -0.190722 Magnesium 0.760068 -0.074779 -0.138629 0.029465 0.130923 Phosphorus -0.021863 0.109966 0.044362 Potassium -0.724651 -0.87879 0.142878 -0.205632 -0.075846 Silicon 0.157939 0.562334 -0.537618 0.122333 Sodium 0. 110745 -0.888604 0.005445 0.014824 Sulphur -0. 110529 -0.11622 0.225067 Titanium 0.463147 2.304538 1.848237 1.433655 Eigenvalue 1.066018 19.2 15.4 11.95 8.88 % Total Variance 19.2 34.61 Cumulative % 46.55 55.44 For explanatiOn see Table 9 5 Other 0.025486 -0.492373 -0.034904 0.135412 -0.258451 0.015474 -0.850776 0.035086 0.222394 0.089424 0.077103 0.369104 1.021766 8.51 63 .95 118 . d e 2- 950328 V an B.lCD TABLE 8 : PCA E.u!:env al ues an dP. nmary Fac t ors: E;JllSO 3 4 2 Factor 1 Road Dust Industrial Road Dust Road Dust Quartz Sulphur Source K-Feldspar Magnesium oxide 0.153863 0.036381 0.803332 0.372885 Aluminum 0.010075 0.120626 0.045965 -0.869732 Calcium 0.052518 -0.005825 -0.513817 Carbon 0.574705 0.194407 0.033108 -0.08 1463 -0.036582 Chlorine 0.054522 -0.055603 0.059933 0.922348 Magnesium -0.075571 -0.095033 -0.125613 0.790296 Potassium -0.192867 -0.101285 0.153075 Silicon -0.951602 0.106182 -0.025201 0.699934 -0.52856 Sodium 0.049579 0.023844 -0.88714 -0.036429 Sulphur -0.133577 0.021733 0.144873 0.012077 Titanium 1.995771 1.842872 1.383276 1.153286 Eigenvalue % Total Variance 19.96 18.43 13.83 11.53 19.96 38.39 52.22 63.75 Cumulative % For explanation see Table 9 . ;plSOde 2- 950328L akewood TABLE 9 : PCA E"1genva ues an dP nmary F actors: E. 1 2 3 4 Factor Industrial Road Dust Road Dust Road Dust Sulphur Source K-Feldspar Quartz Magnesium oxide 0.136887 -0.059002 Aluminum -0.826459 0.316788 0.021429 Calcium -0.94134 0.052582 0.056728 0.100148 Carbon 0.447099 0.744414 0.116865 0.021788 -0.066254 Iron 0.098305 -0.100455 0.110187 -0.305522 0.137335 Magnesium 0.762944 -0.037221 Potassium -0.020129 -0.123181 -0.757115 0.23072 -0.016155 0.123921 Silicon -0.958216 0.069311 -0.014701 Sodium 0.363964 -0.631931 0.041793 Sulphur -0.944168 0.081524 0.004299 -0.194997 0.207133 Titanium 0.06828 0.488286 2.194023 Eigenvalue 1.731547 1.452894 1.236898 21.94 17.32 % Total Variance 12.37 14.53 21.94 39.26 Cumulative % 53.78 66.15 For explanation see Table 9 5 Other 0.107824 0.060865 -0.371143 -0.543152 0.055081 -0.184386 0.103602 0.175348 0.002101 -0.736164 1.023791 10.24 73.99 5 Other 0.090247 0.017515 -0.259745 -0.812693 0.021247 -0.169614 -0.079616 0.536441 0.003244 0.217018 1.036339 10.36 76.52 119 . d e 3 - 960227 PI aza nmary Fact ors: E,plSO TABLE 10 : PCA E"1genval ues an dP. 2 3 4 Factor 1 Industrial Combustion Road Dust Salt NaCI Ca-Feldspar Sulphur Source 0.138659 -0.129359 -0.669079 0.353685 Aluminum 0.25692 0.06577 0.283588 0.507162 Calcium 0.14652 -0.059346 0.06 11 93 0.845678 Carbon -0.023708 -0.796057 -0.051145 0.064577 Chlorine 0.117714 -0.07 124 -0.038879 0.6786 Iron -0.078309 -0.023309 -0.1289 12 0.831944 Magnesium -0.137055 0.104206 0.079385 0.327685 Phosphorus -0.142129 0.205348 0.237889 Potassium -0.493391 -0.596592 -0.347463 -0.328618 0.344754 Silicon 0.013722 0.0400 19 -0.546338 -0.682405 Sodium 0.026716 0.074087 -0.082874 0.782872 Sulphur 0.02983 0.040393 -0.011922 0.005732 Titanium 2.136813 1.693778 1.240 174 1.14 111 9 Eigenvalue 10.34 17.81 14.42 9.5 1 % Total Variance 42.26 17.81 31.92 51.77 Cumulative % For explanation see Table 9 5 Road Dust K-Feldspar 0. 104945 -0.138604 0.201046 -0.055152 -0.298203 0.027674 -0.068458 0.322505 -0.263678 0.06786 0.199069 -0.798411 1.013245 8.44 60.21 " . d e 3 - 960227 V an B"lCD TABLE 11 : PCA E"1genval ues an dP nmary F actors: E;piSO 3 Factor 1 2 4 5 Road Dust Industrial Other Road Dust Other Magnesium oxides Sulphur Source Quartz 0.138536 0.188444 -0.14 148 0.051002 -0.339831 Aluminum -0.135407 -0.019536 -0.125388 Calcium -0.804698 0.102481 0.06461 0.1457 16 -0.086199 Carbon 0.7461 -0.089412 Chlorine 0.10 1691 -0.16533 1 0.007295 0.027763 -0.777398 -0.117258 -0.009628 0.11895 0.000783 Iron -0.747684 -0.063 192 -0.04019 1 0.133443 Magnesium -0.840207 0.010648 -0.240 116 -0.092131 -0.000464 0.249142 Manganese 0.097556 -0.042369 0.060429 0.024275 -0.00401 Phosphorus 0.856545 0.044524 -0.051639 -0.11519 Potassium 0.01442 -0.076041 -0.039336 0.14106 -0.080828 -0.939632 Silicon 0.038916 0.02733 5 0.028834 Sodium 0.73196 0.394435 0.128427 0.01424 Sulphur -0.821612 0.03103 0.227488 -0. 1579 19 0.09 1606 -0.070027 -0.009387 Titanium 0.836744 0.006 118 2.186647 1.628453 1.516118 Eigenvalue 1. 262686 1.183 442 16.82 12.53 % Total Variance 11.66 9.71 9.1 16.82 29.35 41.01 50.72 Cumulative % 59.83 For explanation see Table 9 6 Road Dust K-Feldspar -0.70965 0.032867 0.362661 0.019735 -0.052241 -0.038745 -0.067649 0.027602 -0.846489 0.08305 0.114092 -0.0045 -0.038391 1.038543 7.99 67.81 120 TABLE12 : PCAE"1genval ues an dP" nmary Fact ors: E~ t. d e 3 - 960227 Lakewoo d 2 3 4 5 Factor 1 Road Dust Road Dust Industrial Road Dust Road Dust Quartz Sulphur Source K-Feldspar Magnesium oxide Titanium -0.091393 0.202965 -0.021682 0.178383 -0.823052 Aluminum 0.012271 0.044333 0.059966 0.045319 0.772248 Barium -0.034068 -0.04404 -0.022595 0.444555 -0.652171 Calcium -0.180917 0.275301 -0.117905 -0.057385 -0.71264 Carbon -0.059109 -0.07084 -0.02842 0.036934 0.009187 Iron 0.129378 -0.004039 -0.108534 0.67829 -0.296082 Magnesium -0.022263 0.046236 0.008659 0.070578 0.072879 Manganese -0.010868 -0.18343 0.006702 0.081808 Potassium -0.812528 -0.096956 0.214041 -0.193323 0.149632 Silicon 0.909085 0.138795 0.16132 -0.049473 Sodium -0.760947 0.188303 -0.09419 0.02438 -0.006517 -0.088558 Sulphur 0.839281 0.015421 -0.120323 0.089486 0.027703 Titanium -0.820643 2.189174 1.666525 1.456901 1.277486 1.095427 Eigenvalue 16.84 12.82 % Total Variance 8.43 11.21 9.83 16.84 29.66 40.87 50.69 Cumulative % 59.12 For explanation see Table 9 6 Other 0.120635 0.057641 -0.078406 0.132692 0.538107 0.287447 0.743997 -0.180977 -0.183625 -0.134115 -0.039263 -0.047731 1.047173 8.06 67.17 TABLE 13: PCA Eigenvalues and Primary Factors: Non-Episode 1 - 960122 Plaza Factor 1 2 3 4 Industrial Road Dust Road Dust Road Dust Sulphur Source K-Feldspar Mica Iron oxide -0.3 11196 Aluminum -0.156129 0.302924 0.706255 0.830579 0.184324 0.056948 -0.029799 Calcium -0.1733 Carbon -0.920888 0.013616 0.243563 0.014024 0 .028795 -0 .004335 Iron -0.695994 0.302634 -0.027655 -0.346553 Magnesium 0.487465 0.258067 0.008447 Potassium 0.63058 0.437476 -0.650368 0.082888 0.090711 Silicon 0.610525 -0.139656 Sodium 0.212556 -0.79621 -0.387918 0.125787 Sulphur 0.944649 0.016788 0.06418 1.814044 1.370271 Eigenvalue 2.3 22892 1.033842 25 .81 % Total Variance 20.16 15.23 11.49 25.81 45 .97 61.19 72.68 Cumulative % For explanatiOn see Table 9 121 . de 1 - 960122 V an B a' en ' ry F actors: N on-EGpaso T ABLE 14 : PCA E"agenva ues an dP nma 2 3 4 Factor 1 Other Industrial Road Dust Road Dust Sulphur Source Mica or Feldspar Ma211esium oxide 0.062672 -0.039147 0.059089 0.804204 Aluminum -0.126819 0.069688 0.140658 -0.884279 Calcium 0.231346 0.647537 -0.544702 Carbon 0.44871 2 0.007562 0.021432 -0.158511 0.023517 I ron -0.08967 0.029624 -0.057236 -0.867188 M agnesium 0.08096 0.05934 0.055306 -0.73178 Manganese 0.121123 0.127232 0.192 184 -0.838432 Potassium 0.103985 -0.051668 0.064381 Silicon 0.906122 0.1639 -0.121312 0.07698 -0.945512 Sodium -0.057813 -0.775425 -0.371681 -0.354779 Sulphur 2.0146 1.461276 1.296782 2.435205 Eigenvalue 14.61 12.97 24.35 20.15 % Total Variance 24.35 44.5 59.11 72.08 Cumulative % For explanatiOn see Table 9 TABLE 15: PCA E igenvalues and Primary Factors: Non-Episode 1 - 960122 Lakewood 3 Factor 1 2 4 5 Industrial Road Dust Combustion Road Dust Road Dust Sulphur Source K-Feld spa r Iron oxide Ma211esium oxide 0.305935 0.530886 0.269809 0.03219 -0.569951 Aluminum -0.082392 -0.148085 0.791484 0.287477 0.124054 Calcium -0.034505 -0.859681 -0.180087 0.121311 0.419698 Ca rbon 0.069839 0.058952 0.085881 0.718402 Iron 0.548899 0.285821 0.0825 0.19387 M agnesium 0.513046 -0.699081 0.178045 P otassium 0.363559 0.627014 -0.42214 0.124823 0.145678 -0.188943 0.576407 -0.098925 -0.697441 Silicon -0.003647 -0.325738 -0.193049 0.083806 0.044175 Sodium 0.122348 0.197008 -0.086193 Sulphur 0.91173 -0.056725 0.0783 15 0.139598 0.13655 0.069029 Titanium 0.479547 2.601143 1.89752 1 1.234 147 1.147135 Eigenvalue 1.081648 26.0 1 18.98 12.34 11.47 % Total Variance 10.82 26.01 44.99 57.33 68.8 79.62 Cumulative % For explanation see Table 9 122 TABLE 16 : PCA E"1genva ues an dP" nmary Fact ors: N on- E~ . d e 2 - 960304 PI aza Factor 1 2 3 4 5 Combustion Road Dust Industrial Salt Road Dust Iron oxide Sulphur Source NaCI Titanium -0.140493 0.310137 -0.113652 -0.615036 0.335953 Aluminum 0.20028 0.199963 0.144292 -0.07228 Calcium -0.64276 -0.025208 -0.309298 0.85714 0.32853 0.003094 Carbon 0.07389 0.157915 -0.112843 -0.179117 -0.744845 Chlorine -0.012003 -0.029695 -0.013878 0.742947 0.331177 Iron -0.295482 0.069134 0.003435 -0.037428 Magnesium 0.78357 -0.159089 -0.010132 0.215552 0.033345 Potassium 0.817777 -0.127956 -0.187696 0.293149 0.133698 -0.821743 Silicon 0.088749 -0.721417 0.316972 -0.415465 0.165109 Sodium 0.216661 0.025984 -0.222393 0.822631 -0.124832 Sulphur 0.025733 Titani1 -0.655572 1.134402 1.031171 Eigenv "U :r iil z 0 10.31 %Total V 9.37 Cll oo 0 :Jcr IUU iii :;,CliO -., Cumulat 60.88 70.25 10 ~~ ~~~- '