Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An evaluation of metal transport from shoulder highway sections into roadside soils due to atmospheric… Preciado Cervantes, Humberto Feliciano 2005

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

Item Metadata

Download

Media
831-ubc_2005-105569.pdf [ 39.51MB ]
Metadata
JSON: 831-1.0063332.json
JSON-LD: 831-1.0063332-ld.json
RDF/XML (Pretty): 831-1.0063332-rdf.xml
RDF/JSON: 831-1.0063332-rdf.json
Turtle: 831-1.0063332-turtle.txt
N-Triples: 831-1.0063332-rdf-ntriples.txt
Original Record: 831-1.0063332-source.json
Full Text
831-1.0063332-fulltext.txt
Citation
831-1.0063332.ris

Full Text

AN EVALUATION OF METAL TRANSPORT FROM SHOULDER HIGHWAY SECTIONS INTO ROADSIDE SOILS DUE TO ATMOSPHERIC AND RUNOFF PROCESSES. by HUMBERTO FELICIANO PRECIADO CERVANTES B.A.Sc , Universidad Autonoma de Guadalajara, 1994 M.A.Sc., Universidad Autonoma de Queretaro, 1998 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (Civil Engineering) THE UNIVERSITY OF BRITISH C O L U M B I A August; 2005 © Humberto F. Preciado, 2005 ABSTRACT A comprehensive study of the migration pathways that contribute to the dispersal, accumulation and mobility of metals (Cu, Fe, Pb, Mn, and Zn) in roadside soils was performed at two highway sites with similar design, but different environmental, traffic and land use characteristics. Samples were collected from multiple media, which included: road sediment, atmospheric dustfall, atmospheric suspended particulates, stormwater runoff and roadside soils. Total metal concentrations, as well as the relative metal partitioning in different fractions, were evaluated to provide an estimate of their mobility and potential bioavailability across different environmental media. Metals showed an increasing degree of bioavailability with decreasing particle size in all sampled media at the two highway monitoring sites. Thus, metals showed low bioavailability in roaddust and roadside soils (except in highly contaminated spots in the case of roadside soil), intermediate metal bioavailability was found in dustfall, whereas metals in atmospheric suspended particulates and runoff were the most potentially bioavailable. These results stressed the importance of the contribution of atmospheric and runoff processes, particularly on surface water bodies, where a significant percentage of metal from deposited atmospheric particulates or incoming runoff may be readily available. Lead, was found to occur at the lowest metal concentration of the five metals measured in runoff and atmospheric samples. However, significant amounts of Pb remained in the roadside soils sampled. Most of the Pb contaminated soils exhibited greater amounts of labile metal and a distinct decrease in the proportion of the tightly bound "residual" extraction component. This pattern was also observed for metals Cu, Mn, and Zn at suspected anthropogenic metal input locations. A forensic investigation of the process of roadside soil contamination was achieved with the aid of Pb isotopic analyses and linked the accumulation of this metal with Cu and Zn. Additionally, a predictive methodology was proposed, which covered the coupled atmospheric and runoff metal loading processes and the main geochemical metal-roadside soil interactions. The methodology has applicability for identifying sensitive areas in highways systems and can be used as a predictive tool aiding in risk assessment or risk management activities when it is coupled with receptor toxicological data. ii / TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iii LIST OF TABLES vii LIST OF FIGURES viii LIST OF ACRONYMS xiii LIST OF SYMBOLS XV ACKNOWLEDGEMENTS xvii CHAPTER 1 INTRODUCTION 1 1.1 Statement of the Problem 1 1.2 Scope and Objectives 10 1.3 Research Plan 12 1.4 Research Contributions 17 1.5 Organization of the Thesis 18 CHAPTER 2 BACKGROUND AND LITERATURE REVIEW 19 2.1 Background 19 2.1.1 Inorganic Soil Components 19 2.1.2 Organic Matter 21 2.1.3 Sorption and Precipitation 22 2.1.3 Metal Surface Complexation Modeling 28 2.1.3.1 Hydrous Ferric Oxides 29 2.1.3.2 Humic Substances 32 2.1.3 Atmospheric Dispersion Modeling 35 2.1.4 Runoff Modeling 41 2.1.5 Infiltration Modeling 43 2.1.6 Multi-component Transport Modeling 45 2.2 Literature Review 2.2.1 Metals in the Highway Environment 46 2.2.2 Metals in Highway Runoff 49 2.2.3 Metals in Atmospheric Particulates 52 2.2.4 Metals in Roadside Soils 55 CHAPTER 3 METHODS AND MATERIALS 60 3.1 Site Description 61 3.1.1 Highway 17 Site 61 3.1.2 Trans-Canada Highway Site 63 3.2 Site Investigation 65 3.2.1 Meteorological Conditions 65 3.2.2 Mass Balance of Metals at Highway 17 65 3.2.3 Road Dust Collection 67 3.2.4 Atmospheric Dust Particulate Deposition 68 3.2.5 Atmospheric Suspended Particulates at Highway 1 68 3.2.6 Highway Runoff Measurement and Sampling 69 3.2.7 Soil Sampling 3.2.7.1 Detailed Surface Soil Sampling at Highway 17 70 3.2.7.2 Shallow Subsurface Soil Sampling at Highway 17 71 3.2.7.3 Subsurface Soil Drilling at Highway 1 71 3.2.8 Infiltration Studies 74 3.3 Laboratory Analyses and Investigation 74 3.3.1 Dust Deposition and Total Suspended Particulates 75 3.3.2 Highway Runoff Analyses 76 3.3.3 Physical and Chemical Characterization of Soils 77 .. 3.3.3.1 Soil Ph 77 3.3.3.2 Cation Exchange Capacity 77 3.3.3.3 Specific Surface Area 78 3.3.3.4 Mineralogy 78 3.3.3.5 Total Carbon Content 79 3.3.3.6 Humic/Fulvic Acid Ratios 79 3.3.3.7 Bulk Density and Specific Gravity 80 3.3.4 Batch Adsorption and Leaching Tests 80 3.3.5 Selective Sequential Extraction 82 3.3.6 Metal Analysis 84 3.3.7 Lead Isotopic Analyses 84 C H A P T E R 4 R E S U L T S A N D DISCUSSION 87 4.1 Highway 17 87 4.1.1 Meteorological Conditions 87 4.1.2 Road Dust 88 4.1.2.1 Total Metal 90 4.1.2.2 Metal Partitioning 92 4.1.3 Atmospheric Dust Particulates 94 4.1.3.1 Dust Particulate Loading 94 4.1.3.2 Metal Concentrations 96 4.1.3.3 Metal Partittioning 100 4.1.4 Highway Runoff 103 4.1.4.1 Rainfall-Runoff Relationships 103 4.1.4.2 Total Suspended Solids 105 4.1.4.3 Total Metal 107 4.1.4.4 Dissolved and Resin-Exchangeable Metal 110 4.1.5 Roadside Soil 112 i v 4.1.5.1 Surface and Subsurface Soil Samples 112 4.1.5.2 Metal partitioning in Soil Samples 121 4.1.6 Mass Balance of Solids and Associated Metals 125' 4.2 Trans-Canada Highway 130 4.2.1 Meteorological Conditions 130 4.2.2 Road Dust 130 4.2.2.1 Total Metal 132 4.2.2.2 Metal Partitioning 133 4.2.3 Atmospheric Dust Particulates 133 4.2.3.1 Dust Particulate Loading 133 4.2.3.2 Metal Concentrations 135 4.2.3.3 Metal Partittioning 138 4.2.4 Total Suspended Particulates 140 4.2.4.1 Suspended Particulate Loading 140 4.2.4.2 Metal Concentrations 142 4.2.4.3 Metal Partittioning 144 4.2.5 Exchangeable Metal in Road dust, Dustfall & TSP 146 4.2.6 Highway Runoff 147 4.2.6.1 Rainfall-Runoff Relationships 147 4.2.6.2 Total Suspended Solids 147 4.2.6.3 Total Metal 148 4.2.6.4 Dissolved and Resin-Exchangeable Metal 149 4.2.7 Roadside Soil 152 4.2.7.1 Subsurface Soil Samples 152 4.2.7.2 Metal partitioning in Soil Samples 158 4.3 Summary 160 CHAPTER 5 LEAD ISOTOPIC ANALYSES 164 5.1 Lead isotopes as tracers of pollution 164 5.2 Isotopic Signatures in Pb Ore Deposits 166 5.3 Lead Partitioning and Isotopic signatures , 1 6 8 5.4 Lead Isotopic Results 170 5.4.1 Highway 17 170 5.4.2 Trans-Canada Highway 172 5.4.3 Metal Enrichment Ratios 176 5.4.3.1 Metal Enrichment Ratios in Soils 176 5.4.3.2 Metal Enrichment Ratios in Atmospheric Samples 180 5.4.4 Identification of Past and Present Isotopic Trends 180 5.5 Conclusions and Recommendations 182 CHAPTER 6 INTEGRATED PREDICTIVE METHODOLOGY 185 6.1 Atmospheric Modeling 185 v 6.1.1 Emission Factor 187 6.1.2 Gaussian Atmospheric Dispersion 189 6.1.2.1 Highway 17 Modeling Estimates 190 6.1.2.2 Trans-Canada Highway Modeling Estimates 193 6.1.3 Limitations and Reliability 195 6.2 Runoff modeling 197 6.2.1 Rainfall characteristics 197 6.2.2 Runoff Quality 200 6.3 Infiltration modeling 204 6.3.1 Estimation of Infiltration Rates 204 6.4 Geochemical modeling 206 6.4.1 Modeling Approach 206 6.4.2 Experimental Adsorption Isotherm modeling 207 6.4.3 Surface complexation modeling on Hydrous Ferric Oxides 209 6.4.4 Surface complexation modeling on Humic & Fulvic 211 Substances 6.4.5 Current metal leaching scenario 214 C H A P T E R 7 C O N C L U S I O N S A N D R E C O M M E N D A T I O N S 221 7.1 Conclusions - 221 7.2 Recommendations & Future Work 224 R E F E R E N C E S 227 APPENDIX A. FLOW CHARTS OF SPECIFIC R E S E A R C H OBJECTIVES 239 APPENDIX B. ATMOSPHERIC D A T A 245 APPENDIX C. B A T C H TEST D A T A 256 APPENDIX D. MASS B A L A N C E OF SOLIDS A N D ASSOCIATED M E T A L S 277 APPENDIX E. SEQUENTIAL EXTRACTION C A L C U L A T I O N S A N D D A T A 287 APPENDIX F. SOIL MINERALOG1CAL CHARACTERISTICS 311 APPENDIX G. G E O C H E M I C A L D A T A 317 APPENDIX H . STATISTICAL A N A L Y S E S 323 APPENDIX I. ISOTOPIC A N A L Y S E S 333 v i 3 LIST OF T A B L E S Table 2.1 Key to atmospheric stability categories 39 Table 2.2 Common highway contaminants and their primary sources 47 Table 2.3 Summary of extraction procedures for atmospheric particulates 53 Table 2.4 GeoEnvironmental Site Assessments performed by the BC. MoTH 56 Table 3.1 Summary of Highway 17 characteristics in Delta, B.C. 62 Table 3.2 Summary of Trans-Canada highway characteristics in Surrey, B.C. 64 Table 3.3 Sequential chemical extractions for the partitioning of particulate metals . and their respective reagents 82 Table 4.1 Summary of meteorological conditions at Highway 17 during the monitoring period (March-November, 2002) 88 Table 4.2 Road dust physical characteristics at Hwy 17 89 Table 4.3 Total metal concentrations in road dust sweepings at Hwy 17 (Mar-Nov, 2002) 91 Table 4.4 Levels of association between Metals in dust deposition, .expressed by Pearson Correlation coefficients (upper box) and corresponding "P" values (lower box) 99 Table 4.5 Summary of Speciation of metals in dustfall samples at Hwy 17 102 Table 4.6 Rainfall-Runoff characteristics for rainfall events monitored at Highway 17 104 Table 4.7 Summary of Total Suspended Solids concentrations for the 8-month monitoring period (March-November, 2002) 106 Table 4.8 Summary of Event Mean Total, Dissolved and Chelex-Exchangeable metal concentrations in storm water runoff at Highway 17 111 Table 4.9 Summary of meteorological conditions at Highway 1 during the monitoring period (Apr-Sep, 2003) 130 Table 4.10 Summary of Measured Silt Loadings at Residential Roads, Arterial Roads, and Highways in the Canadian Lower Fraser Valley (g/m ) 131 Table 4.11 Total metal concentrations in road dust sweepings at T C H 132 Table 4.12 Summary of Speciation of metals in dustfall samples at T C H 140 Table 4.13 Summary of Speciation of metals in TSP samples at T C H 144 Table 4.14 Rainfall-Runoff characteristics for rainfall events monitored at T C H 147 Table 4.15 Summary of TSS values for runoff events at T C H 148 Table 4.16 Total and Dissolved Event Mean Concentrations at T C H 150 Table 4.17 Summary of Event Mean Total, Dissolved and Chelex-Exchangeable metal concentrations in storm water runoff at T C H 151 Table 5.1 i U O Pbr u ' Pb ratios in selected major lead bearing ores 167 Table 6.1 Values of k to predict emissions of different particles sizes 188 Table 6.2 Measured silt loadings (sL) and corresponding Emission Factors 189 Table 6.3 Source conditions used to develop the equation for the prediction of fugitive dust emissions 196 Table 6.4 Summary of precipitation amount statistics at the two Hwy study sites 199 Table 6.5 Summary of precipitation event duration statistics at the two Hwy study sites 199 Table 6.6 Summary of yearly precipitation statistics at the two Hwy study sites 200 Table 6.7 Summary of log-normal PDF statistics for metals Cu, Pb, Mn, and Zn 202 Table 6.8 Percentiles from PDF corresponding to total metal concentrations measured in runoff samples at Hwy 17 and T C H 203 vii LIST OF F I G U R E S Figure 1.1 Geo-environmental site assessments in the Vancouver, B.C. area 6 Figure 1.2 General research plan scheme 13 Figure 2.1 Schematic representation of ion and potential distribution in the double layer according to the theories of Helmholtz, Gouy, and Stern 25 Figure 2.2 Classification of adsorption isotherms 26 Figure 2.3 Schematic representation of ion binding on an oxide surface,. and Potential decay in the diffuse layer 30 Figure 2.4 Coordinate system showing Gaussian distributions in the horizontal and vertical directions 35 Figure 2.5 Small cube of space at the center of an atmospheric plume 36 Figure 2.6 Green-Ampt parameters and the conceptualized water content profile showing the sharp wetting front 44 Figure 3.1 Highway 17 study site location in Delta, B.C. 62 Figure 3.2 Trans-Canada Highway Site location in Surrey, B.C. 64 Figure 3.3 Track mounted Geoprobe sampling at the Trans-Canada Highway study site 73 Figure 3.4 Plastic inner liners retrieved from the split sampler and opened to expose soil samples 74 Figure 4.1 Particle size distribution of road dust samples collected at Highway 17 from the NorthBound and SouthBound corridors 90 Figure 4.2 Total metal trends in road dust throughout the Lower Mainland 91 Figure 4.3 Metal partitioning in road dust samples throughout the monitoring period for Northbound (NB) and Southbound (SB) locations 93 Figure 4.4 Background atmospheric deposition at Canadian Wildlife Services location 94 Figure 4.5 Dust deposition pattern on the right of way of Highway 17 for different sampling periods during the March 25 to July 11, 2002 monitoring period 95 Figure 4.6 Metal deposition loadings at different distances from the edge of the road for the sampling period between Jun 25th-Jul 11th, 2002 97 Figure 4.7 Particle size distribution curves of atmospheric particulates deposited on the right of way at different distances from the edge of the road 99 Figure 4.8 Summary statistics for metal concentration in dust deposition (mg/g) at different distances from the edge of the highway 100 Figure 4.9 Metal speciation in dust deposition rates: a) lead, b) copper, c) Manganese, d)Zinc 101 Figure 4.10 Evolution of TSS over a) First observation period: Mar-Jul, b) Second observation period: Jul-Nov, 2002 106 Figure 4.11 Association of TSS with different metals for the April 13th, 2002 discrete runoff sampling 107 Figure 4.12 Evolution of Event Mean Concentrations of metals in highway runoff over the 8-month monitoring period (Mar-Nov, 2002) 108 Figure 4.13 Total metal concentrations in highway runoff for the October 3 r d , 2002 event sampled during the July 1 lth-November 29th observation period 109 Figure 4.14 Total metal concentrations in highway runoff for the November 18 th, 2002 event during the July 1 lth-November 29th observation period 109 vm Figure 4.15 Comparison of dissolved metal concentrations for a discreet sampling event on November 18 th, 2002 at the Northbound location of Highway 17 110 Figure 4.16 Northbound metals concentrations on surface roadside soil at Hwy 17 112 Figure 4.17 Southbound metals concentrations on surface roadside soil at Hwy 17 113 Figure 4.18 Vertical profiles showing soil characteristics and metals concentrations with depth at 1NB (0 m) and 1SB (0 m) location 116 Figure 4.19 Vertical profiles showing soil characteristics and metals concentrations with depth at 2NB (5 m) and 2SB (6 m) location 117 Figure 4.20 Vertical profiles showing soil characteristics and metals concentrations with depth at 3NB (10 m) and 3SB (12 m) location 118 Figure 4.21 Vertical profiles showing soil characteristics and metals concentrations with depth at the Hwy 17 background location 119 Figure 4.22 Representative X-ray diffraction pattern from Hwy 17 roadside soil 120 Figure 4.23 Vertical distribution of Pb and its relative partitioning in the soil profile at the 1NB soil pit 121 Figure 4.24 Vertical distribution of Zn and its relative partitioning in the soil profile at the 1NB soil pit 122 Figure 4.25 Vertical distribution of Mn and its relative partitioning in the soil profile at the 1NB soil pit 123 Figure 4.26 Vertical distribution of Cu and its relative partitioning in the soil profile at the 1NB soil pit 124 Figure 4.27 Mass balance distribution of solids < 250 mm for the March-July, 2002 monitoring period at Hwy 17 126 Figure 4.28 Mass balance distribution of solids < 250 mm for the July-November, 2002 monitoring period at Hwy 17 127 Figure 4.29 Mass balance distribution of Zn for the March-July, 2002 monitoring period at Hwy 17 128 Figure 4.30 Mass balance distribution of Iron for the March-July, 2002 monitoring period at Hwy 17 129 Figure 4.31 Metal partitioning in road dust samples from the TCH,study site 133 Figure 4.32 Atmospheric deposition at the Trans-Canada Hwy background location 134 Figure 4.33 Dust deposition pattern on the right of way of T C H for different sampling periods 134 Figure 4.34 Metal deposition loadings at different distances from the edge of Trans-Canada Hwy for the sampling period from May 28th - Jul 2nd, 2003 135 Figure 4.35 Particle size distribution of atmospheric particulates deposited at different distances on the Trans-Canada Highway right of way 136 Figure 4.36 Summary statistics for metal concentrations in dust deposition at different distances from the edge of Trans-Canada Hwy 137 Figure 4.37 Metal speciation in terms of dust deposition rates for metals: Pb, Cu, Mn, and Zn 138 Figure 4.38 Comparison of Hwy 17 and T C H Particle Size Distributions from dust deposition samples collected at different distances from the road 139 Figure 4.39 Total Suspended Particulate measurements on T C H right of way 141 Figure 4.40 Comparison of particle sizes for TSP and dust deposition samples at T C H 142 Figure 4.41 Summary statistics for metal concentrations in Total Suspended Particulates at different distances from the edge of Trans-Canada Hwy .143 Figure 4.42 Metal speciation in Total Suspended Particulates for metals: Pb, Cu, Mn, and Zn 145 ix Figure 4.43 Figure 4.44 Figure 4.45 Figure 4.46 Figure 4.47 Figure 4.48 Figure 4.49 Figure 4.50 Figure 4.51 Figure 4.52 Figure 4.53 Figure 4.54 Figure 4.55 Figure 4.56 Figure 4.57 Figure 5.1 Figure 5.2 Figure 5.3 Figure 5.4 Figure 5.5 . Figure 5.6 Figure 5.7 Figure 5.8 Figure 5.9 Figure 5.10 Figure 5.11 Figure 5.12 Percentages of Exchangeable metal in Road dust (RD), Dustfall (DF), and Total Suspended Particulates (TSP) samples at the TCH site 146 TSS associations with different metals for the April 23rd, 2003 discrete runoff sampling event 148 Metals concentrations in highway runoff for the April 23rd, 2003 discrete sampling event 149 Dissolved metal concentrations for the July. 13th discrete sampling event 150 Vertical profiles showing soil characteristics and metals concentrations with depth at BH2 (0 m), and BH5 (0 m) location 154 Vertical profiles showing soil characteristics and metals concentrations with depth at BH3 (5 m), and BH6 (5 m) location 155 Vertical profiles showing soil characteristics and metals concentrations with depth at BH4 (10 m), and BH7 (12 m) location 156 Vertical profiles showing soil characteristics and metals concentrations with depth at the background location (Hwy No. 1 Study Site) 157 Vertical distribution of Pb soil concentrations and its relative partitioning in roadside soils at the BH2 location 158 Vertical distribution of Zn soil concentrations, and its relative partitioning in roadside soils at the BH5 location 159 Vertical distribution of Cu soil concentrations, and its relative partitioning in roadside soils at the BH6 location 159 Vertical distribution of Mn soil concentrations, and its relative partitioning in roadside soils at the BH6 location 160 Total metal concentrations in road dust at Hwy 17 and T C H 161 Mean total metal concentrations in dustfall at Hwy 17 and T C H 161 Summary of runoff metal concentrations recorded from discrete and composite sampling events at both study sites 162 Lead isotopic ratios in Road Dust and Soil Samples at Hwy 17 170 Metal concentrations across the soil profile and the corresponding 2 0 6 p b / 2 0 7 p b i s o t o p i c r a t i o s exhibited at the 1NB location 171 Metal concentrations across the soil profile and the corresponding 2 0 6 p b / 2 0 7 p b i s o t o p i c r a t j o s exhibited at the Ditch location 172 Lead isotopic ratios in Road Dust and atmospheric dust deposition samples at Hwy 17 172 Lead isotopic ratios in Road Dust and Soil Samples at Trans-Canada Hwy 173 'jot* Metal concentrations across the soil profile and the corresponding Pb/ Pb isotopic ratios exhibited at the BH5 location (at the edge of the hwy) 174 Metal concentrations across the soil profile and the corresponding Pb/ Pb isotopic ratios exhibited at the BH6 location (5 m away from the highway) 174 Lead isotopic ratios in Road Dust and atmospheric dust deposition samples at Trans-Canada Hwy 175 Lead isotopic ratios in Total Suspended Particulate samples at Trans-Canada Hwy 175 Summary of Pb/ Pb isotopic ratios for roadside soil samples at Hwy 17 and Trans-Canada Hwy study sites and their relationship with: a) Pb concentrations, and b) Pb Enrichment Ratios 177 Associations between Cu and Pb Enrichment Ratios at: a) Hwy 17, and b) Trans-Canada Hwy, study sites 178 Associations between Zn and Pb Enrichment Ratios at: a) Hwy 17, x and b) Trans-Canada Hwy, study sites 178 Figure 5.13 Linear correlations between: a) Mn and Pb soil concentrations at Hwy 17, and b) Mn and Pb Enrichment Ratios at Trans-Canada Hwy 179 Figure 5.14 a) Linear negative correlation between Pb concentration in atmospheric 206 207 , dustfall samples and Pb/ Pb isotopic signatures at both study sites, 906 907 and b) Pb/ Pb isotopic signatures and Pb Enrichment Ratios in atmospheric dustfall samples at both study sites 180 Figure 5.15 Summary of 2 0 6 Pb/ 2 0 7 Pb versus 2 0 8 Pb/ 2 0 6 Pb from different natural and anthropogenic sources 181 Figure 5.16 Summary of 2 0 6 Pb/ 2 0 7 Pb versus 2 0 8 Pb/ 2 0 6 Pb from different natural and anthropogenic sources. Triangular area that groups anthropogenic and natural sources in western Canada proposed by Simonetti et al. (2003) 182 Figure 6.1 F D M modeling results, using an emission factor of 3.68 g/VKT, and actual deposition data on roadside soils for the 1st monitoring period 191 Figure 6.2 F D M modeling results, using an average emission factor of 1.72 g/VKT, and actual deposition data on roadside soils for the 1 st monitoring period 191 Figure 6.3 F D M modeling results using an Emission Factor resulting from average collected silt loadings (2.07 g/VKT), plotted with actual deposition data on roadside soils for the 1 st monitoring period at Hwy 17 192 Figure 6.4 F D M model predictions of TSP based on an EF = 1.52 g/VKT, superimposed with actual TSP measurements on the T C H right of way study site at different distances from the road 193 Figure 6.5 F D M deposition predictions for an EF = 1.52 g/VKT, and actual deposition data on T C H right of way 194 Figure 6.6 F D M model predictions of TSP based on an EF = 0.81 g/VKT, superimposed with actual TSP measurements on the T C H right of way study site at different distances from the road 194 Figure 6.7 F D M deposition predictions for an EF = 0.81 g/VKT, and actual deposition data on T C H right of way 195 Figure 6.8 Scatter plots for predicted vs. measured: a) Total Suspended Particulates, and b) atmospheric deposition 196 Figure 6.9 a) Skewed Probability Distribution Function for daily precipitation data at Tsawwassen Station, and b) Log-transformed data with its corresponding normal distribution 198 Figure 6.10 a) Skewed Probability Distribution Function for daily precipitation data at Kwantlen Park Station, and b) Log-transformed data with its corresponding normal distribution 199 Figure 6.11 Probability plots for Cu and Zn sites E M C in North American: a) Rural sites, and b) Urban sites 201 Figure 6.12 Lead and Mn sites E M C in North American rural an urban sites: a) Probability plot, and b) Histograms for both metals 201 Figure 6.13 Measured infiltration rates in roadside soils and Green-Ampt model simulations for: a) Hwy 17, and b) Trans-Canada Hwy sites 205 Figure 6.14 Cumulative infiltration as a function of time and Green-Ampt model simulations for: a) Hwy 17, and b) Trans-Canada Hwy sites 206 Figure 6.15 a) Chromatographic breakthrough of metals in Hwy 17 roadside soils, b) relative concentration and breakthrough of flushing runoff metal solution 208 xi Figure 6.16 Figure 6.17 Figure 6.18 Figure 6.19 Figure 6.20 Figure 6.21 Figure 6.22 Figure 6.23 Figure 6.24 Figure 6.25 Chromatographic breakthrough of metals: a) Copper & Zinc, and b) Lead in T C H roadside soils 209 Metal breakthrough simulation in Hwy 17 roadside soils by surface complexation on HFO: a) Lead, and b) Copper and Zn 210 Scatter plots for experimental vs. predicted equilibrium concentrations 212 in batch adsorption tests for Hwy 17 soils: a) Cu, and b) Zn Scatter plots for experimental vs. predicted Pb equilibrium concentrations in batch adsorption tests 213 Scatter plots for experimental vs. predicted equilibrium concentrations in batch adsorption tests for T C H roadside soil samples: a) Cu, and b) Zn 214 Scatter plots for experimental vs. predicted Pb equilibrium concentrations in batch adsorption tests for TCH roadside soils samples 214 Actual and simulated Pb profiles at 1NB location corresponding to mobilization in highly permeable soil at the Hwy 17 study site 215 Evolution simulation of: a) Pb soil concentration profile over time and b) pore water at the bottom of the soil column 217 Actual and simulated Pb profiles at the BH2 location corresponding to mobilization in highly permeable soil at the T C H study site 218 Evolution simulation of: a) Pb soil concentration profile over time, and b) pore water, at the bottom of the soil column of the BH2 sampling location 218 i i Xll LIST OF A C R O N Y M S A C S V Anodic Cathodic Striping Voltametry ADT Average Daily Traffic A N O V A Analysis of Variance A S T M American Society for Testing and Materials A S V Anodic Striping Voltametry BET Brunauer Emmett & Teller method for Surface Area determination B H Borehole B M P Best Management Practice C A L T R A N S California Department of Transportation CEC Cation Exchange Capacity CSR Contaminated Sites Regulation EB Eastbound EF Emission Factor E G M E Ethylene Glycol Monoethyl Ether E M C Event Mean Concentration ER Enrichment Ratio E X A F S Extended X-ray Absorption Fine Structure F D M Fugitive Dust Model FHWA U.S. Federal Highway Administration G V R D Greater Vancouver Regional District HFO Hydrous Ferric Oxide H W Y Highway IAP Ion Activity Product 1CP Inductively Coupled Plasma ICP-MS Inductively Coupled Plasma - Mass Spectrometry ISE Ion Selective Electrode potentiometry L A Laser Ablation LA-ICP-MS Laser Ablation - Inductively Coupled Plasma - Mass Spectrometry M M T Methylcyclopentadienyl Manganese Tricarbonyl MoTH Ministry of Transportation and Highways MS Mass Spectrometry M T B E Methyl Tertiary Butyl Ether M W L A P Ministry of Water Land and Air Protection N B Northbound NURP National Urban Runoff Program P A H Poly Aromatic Hydrocarbons P D M Probabilistic Dilution Model PM-2.5 Particulate Matter with aerodynamic diameter < 2.5 microns PM-10 Particulate Matter with aerodynamic diameter < 10 microns PM-30 Particulate Matter with aerodynamic diameter < 30 microns PSD Particle Size Distribution PV Pore Volume PZC Point of Zero Charge ROW Right of Way RD Roaddust SB Southbound SC Surface complexation SOM Soil Organic Matter SPT Standard Penetration Test SSA Specific Surface Area SSE Selective Sequential Extraction STORM Storage Treatment Overflow Model S W M M Storm Water Management Model TCH Trans-Canada Highway TS Total Solids TSP Total Suspended Particulates TSS Total Suspended Solids U C L Upper Confidence Limit of the mean USEPA U.S. Environmental Protection Agency USGS U.S. Geological Survey U V Ultra Violet V K T Vehicle Kilometer Traveled WB Westbound LIST OF S Y M B O L S Symbol Brief Description c concentration CDO) concentration of species / in the diffuse layer CsO) concentration of species / in solution d effective atmospheric particle diameter F Faraday Constant = 96,458 C/mol Fd deposition rate g gravitational acceleration h head above the weir crest hf soil-water pressure head at the wetting front H mixing height Hs soil-water pressure at the surface l(t) cumulative infiltration at time t K turbulent dispersion coefficient K°pp apparent equilibrium constant K,nt intrinsic equilibrium constant Ks hydraulic conductivity K s o solubility product selectivity coefficient M humic molecular weight n distance in the direction of interest n v-notch weir constant N Avogadro's number = 6.02 x 10 2 3 formula units pK negative logarithm io of an equilibrium constant Q emission Q discharge or flow rate r radius of the humic molecule r a aerodynamic resistance r d deposition layer resistance R universal molar gas constant = 8.314 J/mol • K R ratio required for the sum of the counterion charges to balance the humic charge Z RA Bulk of a humic compound t time T length of exposure period T absolute temperature u wind velocity V volume of air XV Vd deposition velocity V D volume of the diffuse layer in a humic substance Vg terminal settling velocity W weight of filter x,y,z coordinate directions or lengths z vertical distance Z valence of a sorbing ion Z penetration of the infiltrating wetting front 5 Stern layer thickness e dielectric constant 8 0 permittivity of free space = 8.854 x 10"12 C/Vm r H sorption densities of protons TOH sorption densities of hydroxy] ions TM sorption densities of specifically sorbed cations r A sorption densities of specifically sorbed anions i"| atmospheric dynamic viscosity K Debye-Hiickel parameter 6S volumetric water content 60 initial water content p particle density a surface charge density a y horizontal dispersion coefficient o"z vertical dispersion coefficient *P surface potential T i total potential T z zeta potential xvi A C K N O W L E D G E M E N T S The author would like to express his gratitude to the following organizations: • The Mexican Council of Science and Technology (CONACYT), for providing the author with the sponsorship for the first four years of this work. • The B.C. Ministry of Transportation and Highways (MoTH) for covering the field investigation costs and numerous chemical analyses for this research. • Environment Canada, Aquatic & Atmospheric Sciences Division. • TELUS for providing access to their Surrey Tynehead facility to get electrical power for the monitoring equipment. • Main Road Contracting Ltd. for their support during traffic control activities The author would also like to express his gratitude to the following people: • My supervisor Dr. Loretta Y . L i , for her patience, guidance and encouragement throughout this odyssey, and for providing the author with a fellowship to complete this work. • My supervisory committee: Dr. Les Lavkulich, Dr. Ken Hall, and Dr. Roger Beckie for their willingness to educate and aid the author during our very illuminating meetings. • Rob G. Buchanan, Senior Geoscientist from B.C. MoTH, for his technical and practical suggestions, and for his enthusiastic and diligent work to allocate funds for this project. Scott Tomlinson for his logistical support in the early stages of the field work. • Susan Harper and Paula Parkinson from the U B C Environmental Lab for their instruction in laboratory procedures, analytical techniques and countless technical issues. • Dr. Dominique Weis for her valuable help during the isotopic investigation and analyses, much beyond her initial agreement with the author, and for numerous and very pleasant discussions. • Dr. Jane Barling for her help training the author to run the mass spectrometer. • Dr. Bruno Kieffer for his help supervising me and doing the chemistry for the isotopic analyses. • Wayne Belzer and Young Ryu from the Aquatic & Atmospheric Sciences Division of Environment Canada for their comments and recommendations regarding the atmospheric dust deposition monitoring phase of this study. Additionally for providing access to the Canadian Wildlife Services facility in Reifel, B.C., and for providing the author with meteorological and deposition data. • Ken Stubbs and Kelly Dehr from the Air Quality Monitoring and Assessment Division of the G V R D for providing the High-Volume air samplers and the road dust database. Additionally, Dave Ferguson, and the personnel from the Air Quality Instrument Shop of the G V R D for their assistance in road dust collection equipment and procedures. I would like to thank my classmates and friends from the Geo-Environmental group: Juan Garfias, Dylan MacGregor and Ranee Lai for sharing the learning journey and their knowledge with me. Finally I would like to thank my Family for their unconditional support, and Ellie for her love and companionship. xvn C H A P T E R 1 I N T R O D U C T I O N 1.1 Statement of the problem. The highway system is a potential source of a wide variety of contaminants to the surrounding environment through natural mechanisms such as atmospheric dust deposition or through the hydrologic cycle (Kobriger and Geinopolos, 1984). Contaminants then become pollutants when they interfere with the normal life cycle and function of organisms. The public and most of the engineering community are familiar with some of the environmental problems related to transportation like air pollution or accidental spills. However, there is little awareness of the pollution of soil, water and biota generated by significant amounts of contaminants deposited by vehicles in highways corridors and rights of way. Non-point pollutants in highways include heavy metals, suspended solids, micro-organics, oils and chlorides. These anthropogenic pollutants result from traffic activities, atmospheric deposition, engine exhaust, roadway degradation, and highway maintenance (Sansalone et al., 1995). From all contaminants present in the highway system, heavy metals are the most prevalent contaminant of concern due to their toxicity (Hathhora, 1996). Highway runoff may contain high concentrations of metals particularly: lead, zinc, iron, chromium, cadmium, nickel, and copper, that result from the ordinary wear of brakes, tires, and other vehicle parts (FHWA, 1998). Although the use of leaded gasoline has been suspended in North America and most of the industrialized world, lead continues to be used in approximately 74 countries around the globe, predominantly in sub-Saharan Africa, the Middle East and most of Asia (Hodes et al., 2003). Even with major future changes in automotive manufacturing, such as substitution of metal parts by plastic, graphite or other materials, reduction in cars dimensions, etc. heavy metals will persist as contaminants in the highway system since the last century trends show that car numbers are likely to keep increasing and the bioavailability of metals in contaminated media remain. 1 In 1976 the U.S. Environmental Protection Agency (EPA) promulgated regulations for 129 specific substances referred to as "priority pollutants". A relatively small number of revisions to this list have been made by the USEPA since 1977. A study developed at Milwaukee and Sacramento (Kobriger and Geinopolos, 1984) demonstrated that over 25% of those priority pollutants were present in the highway environment and that these migrated mainly from that environment via runoff from storm events and saltation processes. Surface waters such as lakes and streams are particularly vulnerable because they are directly exposed to contaminants released into the air and to direct discharges from point and nonpoint sources (Young et al., 1996). On the other hand, although contamination of groundwater tends to occur gradually because contaminants percolate downward through the soil slowly, contaminants can also reach groundwater rather quickly in fractured rock formations or sinkholes in Karst areas (Smoot et al., 1997). As part of the National Urban Runoff Program (NURP), the USEPA (1983) reported that a number of metals in storm water runoff exceeded water quality standards, but no information was provided on whether these concentrations were adverse to fish and aquatic life or on the speciation (bioavailability) of contaminants. Yousef et al. (1985) identified species of dissolved Pb, Zn, Cu, and Cd using an anodic stripping voltametry technique for rainfall, highway and bridge runoff in Florida. They found that possible metal bioavailability in bridge runoff followed the order Zn>Cd>Cu>Pb. They also found that Zn and Cd existed mainly as free metal ions in natural waters, Cu was present mainly as organic complexes, and Pb organic complexes varied between 15 and 44% of metal in solution. Morrison and Revitt (1987a) compared resin exchangeable metal concentrations in urban runoff to USEPA water quality standards finding significant acute threats for Cu and Cd, and chronic effects for Zn, Cd and Cu during snowmelt events, when the increase in ionic strength due to road salting encourages a continuous release of bioavailable metal species. The researchers consider chelex removable metal concentrations more acceptable than total metal concentrations, for a comparison with water quality standards, since the latter may not account for the high percentage of metals in non-bioavailable forms. More recently, Sansalone et al. (1996) instrumented lateral sheet runoff flow from a busy highway pavement section in Cincinnati, OH. 2 They reported the highest geometric means of dissolved/particulate-bound metal ratios for the metals Zn, Mn and Cu. Currently, there exists consensus among the scientific community that water quality standards have to be based more on the particular speciation of metals and their associated toxicological effect, rather than total concentrations. It appears that the regulatory framework on storm water runoff will eventually shift in that direction. Hence, in addition to concentration data about metals in highway runoff, information on metal speciation under different environmental conditions needs to be continuously updated, documented and be scientifically defensible for transportation agencies to use it to support planning and management decisions (Breault and Granato, 2000). Besides highway runoff, another metal migration process that requires further comprehensive work in highway environmental investigations on roadside soils and neighbouring areas is: atmospheric deposition and removal. Again, most of the literature has dealt with automobile exhaust emissions and long-range transport of breathable particulate matter (PM 2.5 and P M 10), due to their adverse health effects. A recent study on the significance of atmospheric deposition on highway runoff, stresses that much of the previously reported atmospheric deposition data may be unreliable, due to insufficient quality assurance and the use of inefficient deposition collection devices (Colman et al., 2001). Therefore, research that accounts for changes in deposition collection technology and updates on metal deposition rates is necessary to better understand the complex interaction among contaminant deposition rates, runoff loadings and roadside soil metal accumulation. Understanding these interactions should allow environmental and transportation agencies determine the best course of action to minimize the influence of highways on neighbouring sensitive ecosystems. Comprehensive data is necessary on the partitioning of atmospheric particulate matter and fugitive dust. Lum et al. (1982) performed sequential extractions on urban particulate matter standard reference material and found that high proportions of Cu, Pb, Zn and other metals were in soluble and/or exchangeable forms. They suggested that these metals might have immediate impact in receiving waters, because they are weakly bound and may equilibrate rapidly in water. •3 However just a few studies have addressed partitioning of particulate matter and fugitive dust on the immediate areas surrounding highways. Dreetz and Lum (1992) performed fractionations on air-intake filters from a building located in a street with busy traffic in Norway. The authors report the mobility of metals, based on the solubility in a sodium acetate solution, decreasing in the order: Zn>Pb>Mn>Cu>Fe. Hlavay et al. (1998) in another similar study of aerosol samples collected in the vicinity of a busy street, report mobile factors (exchangeable fractions) of metals following the order: Pb(40%) > Cu(13%) > Zn (12%)>Mn(5%). A regulatory approach to storm water runoff has been suggested for atmospheric particulates (i.e. consider the potentially bioavailable metal fraction). In addition, there are regulatory proposals to consider a so called "critical loads" approach for metals in different media, where environmental quality objectives are based on the same concept of potentially bioavailable metals rather than totals (Rieuwerts et al., 1998). However, the data on atmospheric partitioning for such a regulatory approach is scarce, particularly for atmospheric particulates genereated by highways. Thus, characterization of atmospheric particulates is necessary to better understand the evolution of metal transport in the highway environment, and to estimate the metal distribution between dissolved and particulate fractions. Little is known conclusively about the atmospheric source or how deposited material is transported from the highway to runoff (Colman et al., 2001). A mass balance calculation is an approach to understanding the processes that affect the distribution of metals along roadside soils resulting from atmospheric and hydrologic events. Mass balance calculations have been done in the past, under different contracts by the U.S. Federal Highway Administration (FHWA). Those studies (Kobriger and Geinopolos, 1984) identified highway runoff as the most important process for spread of contamination accounting for up to 80% of contaminant mass. However most of that information was gathered when leaded gasoline was still in use and the assessment of the atmospheric contribution was evaluated with what now would be considered inefficient dust collection technology. More recently, Turer et al. (2001) performed a mass balance calculation for the metals Pb, Cu, N i , and Cr comparing runoff and soil concentrations at Cincinnati, Ohio. According to their 4 calculations, 60% of Pb had been lost from roadside soils. Although the authors did not run an atmospheric monitoring program, they acknowledged that the most probable removal mechanisms for unaccounted metal levels in roadside soils were fugitive dust removal or surface runoff carrying metals into surface drainages bypassing roadside soils. Therefore, it is considered essential to update and expand the current inventory of roadside pollution due to highway sources that reflect changes in gasoline, automobile manufacturing materials, traffic counts and dust collection technology. In regards to roadside soils, Pb contamination in this media has been reported extensively due to the adverse health effects of this metal (Ward et al., 1975; Onyari et al., 1991; Sithole et al., 1993). Previous researchers have reported that lead content in roadside soils often exceeds local environmental regulations, that lead decreases exponentially with distance from the road (Motto et al., 1970; Musket and Jones, 1980; Kobriger and Geinopolos, 1984; Burguera and Burguera, 1988) and that the pattern of accumulation is correlated with traffic density (Rodriguez-Flores and Rodriguez-Castellon, 1982; Hafen and Brinkmann, 1996). Other researchers have reported high accumulation of metals Zn, Cu, Mn, and Pb in plants and soil located near highways (Albasel and Cottenie, 1984). Majdi and Persson (1989) reported higher concentrations of Pb and Cd on tree fine roots along a motor road in Sweden. Norrstrom and Jacks (1998) found that a large part of Pb, Cu and Zn in roadside soils is vulnerable to leaching when exposed to high NaCl concentrations, acidic or reducing conditions. Pagotto et al. (2000) analysed road dust and roadside soil near a major rural highway in France and found that Pb and Cu were not highly mobile but concluded that there was a significant risk of Zn mobilization under acidic conditions. On the other hand, in an assessment of soil and groundwater contamination from two old infiltration systems for road runoff in Switzerland, Mikkelsen et al. (1996) reported that the potential for groundwater contamination appeared to be limited, but soil and runoff sediment could be heavily contaminated. Contaminants that have accumulated in soil are often exposed by earth moving equipment as the right of way is cut to grade for subsequent construction (Russell, 1995). This causes transportation agencies to establish, prior to construction, an investigative stage to characterize the soil to be disturbed and compare analytical results with applicable regulations. In British Columbia, i f contaminants in the soil are above maximum allowable concentrations, the soil must then be treated as a waste and disposed of appropriately ( M o T H , 2000). These contaminants cause transportation agencies to incur a considerable expense in roadway improvement projects. In Geo-Environmental Assessments performed for roadway improvement projects in the Vancouver B . C . area, 18 out of 24 sites were found contaminated according to the Contaminated Sites Regulations and needed to be disposed off site and transported to landfills. In those assessments, lead contamination was the first cause for relocation followed by zinc and copper contamination ( M o T H , 2002) [Figure 1.1]. It is estimated that in the next 5 years, these actions wi l l add $25 million to the current costs of highway maintenance and expansion ( L i , 2002). Figure 1.1.. Geo-environmental site assessments in the Vancouver Area, B . C . (squares represent sites where metal contamination was found above industrial Contaminated Site Regulations standards). 6 An investigation by L i (2002) targeted to determine the environmental fate of lead in surface soils along highway corridors of B.C. reported significant.accumulation of this metal in roadside soils with concentrations decreasing to background level at a depth of 0.6 m. Lead partitioning through sequential extractions showed that a minor fraction of metal was exchangeable and that complexation with manganese and iron oxides, along with organic matter were the most important attenuation mechanisms in surface soils. Based on sequential extractions and batch desorption tests, L i concluded that Pb had a very limited mobility in highway soil and that leaving Pb contaminated soil in place might be acceptable. The author recommended organic matter addition as a remedial or preventive measure in dealing with Pb contamination. A l Chalabi and Hawker (2002) also suggested an important role of organic matter in controlling heavy metal contamination in Australian soils. However, questions remain as to: 1) how this metal attenuation might take place, 2) the strength of the organic matter metal associations, and 3) whether organic matter might enhance metal mobilization under certain conditions. The number of sorption sites in clay minerals decreases as sorption of metals increases. On the other hand, as organic matter is regenerated and decomposed, additional sites for metal sorption become available. The overall effect of organic matter on metal accumulation and mobility is not straightforward. For instance Igloria et al. (1996) report that metal mobility is greater in the presence of organo-metal complexes. In the studies on roadside soils of B.C, L i (2000; 2002) stressed that other metals present as non-point source pollutants exert a competition effect for adsorption sites in soil colloids and that these metals can substantially decrease attenuation for a particular metal. However, there is a lack of research that addresses the adsorption characteristics of roadside soils when subjected to multi-component metal solutions, which is the case when atmospheric and highway runoff loadings are continuously applied on these soils. Furthermore, the type of metal associations with different roadside soil components (i.e. retention/retardation mechanisms) also needs to be investigated in a more comprehensive manner. In the original FHWA reports, the authors found most metal accumulation within surface layers of soil and a decrease in water quality that percolated into the ground closer to the highway. 7 They also reported some physico-chemical properties of soils but metal concentrations were reported as total concentrations in soil, grass and roots, without any particular reference to soil component and their ease of availability (Gupta et al., 1981; Kobriger and Geinopolos, 1984). As the toxicity of alkyl lead (used as an anti-knock additive in gasoline) and the halogenated scavengers became of concern, alternatives were considered such as methyl tertiary butyl ether (MTBE) and methylcyclopentadienyl manganese tricarbonyl (MMT). The most popular of these is M M T , which has been in use in Canada since 1976 and it is now approved for sale in 25 countries, including the U.S.A. (except California) and Australia (Kaiser, 2003). Studies in British Columbia developed as part of the Fraser River Action Plan (FRAP) [Hall et al., 1998] indicate that manganese has increased in stream and street sediments since 1973, and that this increase in stream sediments corresponds in time to the introduction of M M T as a replacement for tetra-ethyl lead, in 1974. In Canada where M M T has been in use for almost 30 years, a human risk assessment by Health Canada found: " A l l analyses indicate that the combustion products of M M T in gasoline do not represent an added health risk to the Canadian population" (Wood and Egyed, 1994). However, the environmental risks associated with M M T are not known. Additional studies, both by the M M T manufacturer and by independent researchers, continue and the U.S. EPA can change the approval of the additive in the U.S.A at any time when hazards are proven. Even the assumption that manganese is accumulating in soil and plants, as a result of M M T combustion, has not yet been proven. A study in Montreal, Canada shows that manganese concentrations in air samples were significantly correlated in time with traffic density, but the authors could not determine that those higher manganese concentrations were due to M M T , since manganese is practically ubiquitous in the environment (Loranger et al., 1993). In Utah, researchers found that roadside soil and plants were apparently contaminated by manganese oxides from Mn-containing motor vehicle exhaust. They also found that manganese concentrations in soil and in some plant species along impacted roadsides often exceeded levels known to cause toxicity (Lyttle et al., 1994). Brault et al. in their study about bioaccumulation of manganese by plants, suggest that the addition of M M T to gasoline may result in an increase in 8 exchangeable Mn in organic soils, but again they could not determine that the Mn accumulation was indeed due to M M T in gasoline (Brault et al., 1994). Hence, more information on manganese accumulation, its relationship with other metals, and the fate and bioavailability of newly deposited Mn is essential. This information can be linked with toxicological data as it becomes available and better estimates of environmental impact or a meaningful regulatory framework can be achieved. A l l roads have rights of way that are occupied by some fauna. Rights of way are long and relatively narrow and they can extend over very diverse areas with different land uses. An average right of way for inter-provincial roads can be up to 30 m wide and at least 33% of the land in the right of way is unpaved and can support plant and animal life. Thus, land associated with 12,000 km of roads in B.C. (and millions of kilometres around the world), potentially represent important wildlife habitats where metal contamination needs to be evaluated in a comprehensive manner. This thesis presents an approach to evaluating fate and transport of metals in roadside soils considering realistic or statistically sound metal loadings due to atmospheric and runoff processes. For this purpose, first atmospheric and hydrologic events were studied at two sites with similar highway design, but different environmental, traffic and land use characteristics. Meteorological conditions were recorded and samples were collected from: road sediment, atmospheric dustfall, atmospheric suspended particulates, stormwater runoff and roadside soils. The study of contaminant loadings was targeted to elevated flush shoulders type of highway section for three reasons: a) it is a simple design, b) it is the most prevalent section in highway systems, c) it allows a less heterogeneous spread of contaminant through atmospheric and runoff processes. The studies were focused on metals Cu, Pb, Zn, due to their prevalence in most highway environments, both locally and worldwide. Additionally Fe and Mn were studied for their ability to regulate the mobility of other metals and, in the particular case of Mn, for its possible association with the fuel additive M M T . Total metal concentrations, as well as the relative metal partitioning in different fractions, (solution + exchangeable, carbonates, oxides, organic, and 9 residual) were evaluated. The purpose of such analyses was to provide an estimate of metal mobility and potential bioavailability across different environmental systems (on-highway, atmospheric, hydrological, surface and subsurface media). Additionally, a forensic investigation of the process of roadside soil contamination was achieved with the aid of Pb isotopic analyses. These tests provided a comprehensive reconstruction of the contamination processes taking place in this environment and aided in the identification of some of the governing processes of contaminant migration from the highway pavement to surrounding areas and down the soil subsurface. Data from atmospheric and hydrologic monitoring programs were used to test the modeling capabilities of existing semi-empirical and statistical models. On one hand, the best suited atmospheric and runoff models were selected in order to evaluate the middle and long term metal input onto roadside soils. On the other hand, physico-chemical characteristics of roadside soils at the two study sites were determined. This information was used, along with sequential extractions and batch adsorption/desorption tests, to examine the dominant soil-metal interactions and thus to select the geochemical models that could mimic the observed experimental and field results. The end result of this comprehensive look at the process of roadside soil contamination was an integrated predictive methodology that addresses the overall problem of metal redistribution. It is believed that this methodology could be used by transportation agencies to assess long-term effects of metal accumulation in roadside soils (when coupled with toxicological data) at areas that have been identified as environmentally sensitive. 1.2 Scope and Objectives The specific objectives of this research are as follows: 1) Understand the processes that affect the distribution of metals along roadside soils resulting from atmospheric and hydrologic events. 2) Investigate metal-soil interactions 3) Develop a model to simulate these processes (mass balance model). 10 The importance of each objective is stressed herein. 1) Understand the processes that affect the distribution of metals along roadside soils resulting from atmospheric and hydrologic events. As stressed earlier, it is relevant to update and expand the current inventory of roadside soil pollution due to highway sources that reflect changes in gasoline, automobile manufacturing materials, traffic counts, and dust collection technology. This update is necessary for the second and third objectives of this study, so the accumulation and mobility of metals can be assessed and modeled based on realistic metal loadings from atmospheric and runoff processes. The isotopic variety of Pb provides for an excellent tracer element for tracking highway contamination in roadside soils during past and current times. The source of this element can be identified and its allocation or distribution through different environmental media can help understand the relevant metal migration processes. The present study, besides contributing to the database of emissions from paved roads, helps assess the validity of current fugitive dust and particulate models for the prediction of atmospheric particulates in roadside soils. Additionally, it provides information that could lead to a better understanding of atmospheric particulate removal processes. 2) Investigate contaminant-soil interactions. The importance of researching this matter in an organized and systematic form is highlighted by the fact that most Geo-Environmental studies performed by the B.C. MoTH in the Lower Mainland, found zinc contamination the second cause of soil relocation in highway improvement projects. A question that arises is whether zinc contamination will be the next headache for Canadian highway agencies. If metals input will likely continue in the future, the next question that arises is: what the influence on roadside soils retention/retardation processes will be from simultaneous input of metals. Metals show preferential sorption onto clay minerals, oxides, sesqui-oxides and organic matter surface sites in the presence of other metals (selectivity), and the amount of metal sorbed 11 depends on the concentration of other metals (competition), as well as complexing agents, pH, redox conditions, etc (Yong and Galvez-Cloutier, 1993). 3) Develop a model to simulate atmospheric, runoff and geochemical processes (mass balance model). A model that accounts for atmospheric, runoff and geochemical processes is useful to estimate metals loadings on neighbour ecosystems and roadside soils, and provides transportation and environmental agencies with a predictive tool that is useful for making sound and cost effective decisions about handling contaminated roadside soils. It also helps identify areas in the transportation system where management practices need to be implemented. This information can be used not only in formulating regulatory standards, but also is an aid to estimate the potential effectiveness of highway design changes and Best Management Practices for mitigating heavy metal contamination. 1.3 Research Plan. The investigation consists of five phases. During phase 1, a field investigation was conducted to assess the efficiency of the equipment to be used for atmospheric dust deposition in roadside soils. Phase 2 involved an environmental field monitoring investigation to study the processes of metal distribution along roadside soils at a simple site (i.e. in terms of meteorological conditions, geometrical configuration, other pollutant sources, traffic density, etc.). Phase 3 represents the laboratory investigation of contaminant-soil interactions, the determination of physico-chemical characteristics, the attenuation processes taking place in soils and also helps generate the database to calibrate the geochemical model. Phase 4 integrates the information gathered from the field and laboratory analysis to model results and predict long-term behaviour of metals in roadside soils under different environmental conditions. Phase 5 is a stage where the model is subjected to different environmental conditions present at another more complex site (i.e. greater traffic, metal loadings, more complex geometrical configuration). At this stage, model sensitivity is assessed and calibrated to simulate particular environmental conditions. The general research plan scheme is depicted in Figure 1.2. The specific objectives of each phase are shown in a flow chart format in Appendix A, and described thoroughly hereinafter. 12 OBJECTIVES FIELD MONITORING TESTS & ANALYSES DATA MODELING 1. Understand the processes that affect the distribution of metals along roadside soils resulting from atmospheric and hydrologic events. 2. Investigate metal-soil interactions 3. Develop a model to simulate these processes (mass balance model). Road dust Atmospheric Meteorological Runoff Soil Road dust characterization: Particle Size Distribution (PSD) Metal Analyses Sequential Extractions (SSE) Pb isotopic analyses Atmospheric loadings PSD Metal analyses SSE Pb isotopic analyses Discrete & Composite sampling Runoff volumes Metal analyses Physico-chemical characterization Metal analyses SSE Pb isotopic analyses Multicomponent batch adsorption/desorption tests Infiltration tests Metal concentrations Metal partitioning Silt loadings (<75um) Pb isotopic ratios Dustfall rate TSP concentration Metal cone. Metal partitioning Pb isotopic ratios Rainfall, wind speed, direction, temperature Runoff coefficients TSS Total/Dissolved & Chelex-Exchangeable metal cone. PSD, Surface area CEC, soil pH, Mineralogy Humic/Fulvic ratios Metal partitioning Pb isotopic ratios Metal adsorp/desdrp Infiltration rates Atmospheric rn Runoff H Geochemical M Figure 1.2. General research plan Phase 1. Background dust monitoring objectives. 1. To determine background dust deposition loadings. Dust loadings from non-vehicle related sources are necessary to assess the relative contribution of atmospheric particulates from roads to the right of way. One dry Frisbee dust gauge was installed at the Canadian Wildlife Services facility at Reifel, B.C. The facility is a bird sanctuary and dust loadings from anthropogenic sources at this location are only those from agricultural activities around the area. 2. To measure dust collection efficiency of Frisbee dust gauges vs. standard wet and dry deposition samplers. The dry Frisbee dust gauge was collocated with standard samplers to assess differences in collection under the same environmental and meteorological conditions. This validated the type of sampler to be used in the study of migration of atmospheric particulates from highway sources into roadside soils. Phase 2. Site investigation at highway 17. 1. To investigate road dust characteristics and accumulation rates. This helped identify the influence of road dust characteristics and management practices such as highway sweeping on runoff and atmospheric loadings on the right of way. 2. To investigate runoff characteristics and loadings. The quantity and quality of runoff needed to be determined to estimate the export rates of metals at this location via hydrologic events. Additionally, dissolved and chelex-exchangeable metal concentrations in highway runoff gave an estimate of metal bioavailability. 3. To investigate the amount and pattern of atmospheric dust loadings on roadside soils. Quantity and quality of atmospheric dust particulates loadings characterized export rates of dust and associated metals via atmospheric processes and helped delineate the area of influence. Selective extraction helped identify metal partitioning in atmospheric particulates and relative mobility and bioavailability from particulate matter, before it is deposited onto roadside soils. 14 4. To investigate shallow subsurface characteristics and collect soil samples for laboratory analysis. This objective achieved two purposes: a), helped delineate the area of atmospheric and runoff influence on neighbour soils and, b) gave a horizontal and vertical perspective of metal distribution in soils. Detailed surface soil samples were used for isotopic analysis of Pb and those analyses were compared with the isotopic ratio of the same elements in gasoline and lead based paint. Phase 3. Laboratory investigation. 1. To characterise roadside soils in terms of their physico-chemical properties. This characterisation allowed correlation of the attenuation processes of metal mobility with physico-chemical properties, plus it provided data for geochemical modeling based on surface complexation at colloids surfaces. 2. To determine the sorptive capacity of roadside soils for Zn, Cu, Pb, and Mn, through batch equilibrium tests. Multi-component batch adsorption tests provided information about the mass of metal sorbed per mass of soil under different pH and ionic strength conditions, in addition to the effects of other competing cations. This provided the experimental database to test and calibrate the geochemical model. 3. To determine the retention mechanism of metals in roadside soils through selective extraction. As mentioned for the partitioning of metal in atmospheric dust particulates, this procedure allows the geochemical speciation of metals in soils and provides an estimate of metal bioavailability and the feasibility of migration under different environmental conditions. Phase 4. Modeling. 1. To propose a predictive methodology for the assessment of multimedia redistribution of metals. The integrated predictive methodology covers the two metal migration pathways that affect roadside soils: 1) atmospheric deposition, and 2) runoff discharge. These pathways are modeled individually, but their coupled contribution to metals loading in roadside soils is considered as a whole. These coupled processes are evaluated over a certain time (e.g. a year) 15 and their metal predictions are used as an input to a mixing cell one-dimensional geochemical model. In this manner, the dominant soil-metal interactions can be evaluated and, to some extent, long-term behaviour of metals in roadside soils under different environmental conditions could be evaluated, particularly those observed in the laboratory such as different acidifying or ionic strength conditions, but also those less reproducible in the laboratory such as different redox environments. Phase 5: Site Investigation at Trans-Canada Hwy & 176th St. A different site was necessary i f the capabilities and consistency of the predictive methodology were to be assessed. Trans-Canada hwy represented a desirable testing site for the model since it was more complex in terms of meteorology, traffic, geometry and driving patterns. However, it was similar in the sense that both Hwy 17 and Trans-Canada are flush shoulder type of sections (i.e. surface drainage is as sheet flow over the pavement shoulders). This phase was similar to phase 2, except that: a) detailed surface soil sampling was not performed due to continuous disturbance of surface soil (top 5 cm) due to maintenance and landscaping activities, and b) atmospheric monitoring of particulates included not only atmospheric dust deposition loadings, but also the sampling of total suspended particulates (TSP). Based on results from phase 2 it was decided that TSP would provide information about the mass that can deposit not only on the right of way, but also in farther neighbour areas. Therefore extra mass not accounted for in the previous investigation would be considered. TSP collected in filters would provide extra mass of atmospheric particulate to further study metal-particulate associations, particle size and speciation. On the other hand, it provided more comprehensive information for the atmospheric migration of road pollutants and fugitive dust modeling. 16 1.4 Research Contributions This research provides data on metal speciation in multimedia (i.e. samples across different environmental media: air, water and soil), which help understand the processes of metal distribution in roadside soils due to hydrologic and atmospheric events. As part of the atmospheric monitoring studies, a new dust collection technology in North America has been validated and used to record more efficiently dust depositional processes. A series of recommendations has also been implemented, in particular, for the use of this technology to measure atmospheric dust-metal loadings. The forensic Pb isotopic investigation provided a comprehensive reconstruction of the contamination processes (historically and currently) taking place in roadside soil environments. This integrated Pb forensic investigation from "cradle to grave", to the knowledge of this researcher, has not been addressed before in the roadside soils of Canada. The investigation may also suggest some future research venues for the investigation of Mn accumulation in the Canadian environment, through the use of isotopic tracers. A multimedia predictive methodology for the accumulation and retention of metals in roadside soils had not been formulated or proposed before. It is believed this methodology should help in the assessment of the current consequences of years of lead deposition. The model also fits conceptually to predict the accumulation of other contaminants of concern such as Cu, Mn and Zn. Due to the buffering role of right of way soils in regards to atmospheric and runoff loadings, the Right of Way (ROW) must be regarded as a contaminant mitigating and storing unit. As such, the right maintenance must be provided at sensitive ecosystem areas to ensure the ROW adequate performance, but its contaminant buffer life expectancy has to be considered and re-evaluated over the course of time. The suggested predictive methodology, considers this evaluation process by taking into account the different pathways of metal migration and its characteristics in different environmental media. Practices oriented to minimize the effects of highway pollutants have focused on reducing the impact of runoff water in the environment through a combination of structural and non-structural 17 Best Management Practices (BMP). A predictive methodology at hand should help as a screening tool to identify sensitive areas, where highway design modifications can be implemented, before levels of heavy metals become unacceptable. On the other hand, i f those levels of metals are found to be above standards, modelling can complement risk based studies to determine the environmental conditions that would facilitate their mobility. 1.5 Organization of the Thesis This thesis consists of 7 chapters: Chapter 1 introduces the general problem, provides the rationale and objectives of this research, outlines previous related research and describes the plan for conducting this study. Chapter 2 provides background information about the problem in British Columbia and compiles information gathered by the Ministry of Transportation and Highways in multiple Geo-Environmental Assessments on roadside soils. This chapter also reviews the theory and concepts of soil-contaminant interactions that help interpret results and model future behaviour. Chapter 3 outlines the methods employed in different phases of this research and describes equipment and materials used. Chapter 4 presents results and a comprehensive discussion of those findings. Chapter 5 provides an extensive literature review on lead isotopic studies worldwide and locally, and presents the lead isotopic results for both monitoring and validation sites, linking the Pb isotopic chemistry with the anthropogenic accumulation of other metals. Chapter 6 presents the development of an integrated methodology that covers atmospheric, runoff and geochemical processes and the estimation of metal accumulation on roadside soils. Chapter 7 presents conclusions and recommendations for future research. The Disc compiles additional information to the appendices regarding metal concentrations (other than Cu, Fe, Pb, Mn, and Zn) that were measured during ICP scans. It also includes other useful and detailed information such as: isotopic mass fractionation data, input and output files from atmospheric and geochemical simulations, and infiltration simulations. 18 C H A P T E R 2 B A C K G R O U N D A N D L I T E R A T U R E R E V I E W 2.1 Background A general background review of established knowledge regarding inorganic and organic soil components is provided. The fundamental reactions that take place between metals and roadside soils under the conditions studied in this research have also been stressed, particularly the processes of sorption, precipitation and surface complexation. This is followed by the description of recent surface complexation modeling approaches on Hydrous Ferric Oxides and Humic Substances in soils. Subsequently, the conventional approach to atmospheric dispersion modeling and basic atmospheric concepts are described in greater extent, assuming that soil specialists may lack a formal education in the atmospheric sciences. The background review closes with runoff, infiltration and multi-component soil transport modeling approaches found in the literature, and the criteria for choosing the particular models used in this research. 2.1.1 Inorganic Soil Components Soils are complex assemblies of solids, liquids and gases (Sparks, 1995). The solid phase elemental composition is predominantly given by: O, Si, A l , Fe, C, Ca, K , Na and Mg (Schulze, 1989; Dixon and Weed, 1989). The inorganic components of soil represent about 90% of the total solid components and their properties like surface area, size and charge behavior play an important role in the physico-chemical reactions that take place in soils. Inorganic components of soils include primary and secondary minerals. A mineral can be defined as a natural inorganic compound with definite chemical, physical and crystalline properties (Sparks, 1995). Common primary minerals include quartz and feldspar, and other primary minerals found in smaller quantities include pyroxenes, micas, amphiboles and olivines. Primary minerals occur predominantly in sand and silt fractions of soil but may be found in slightly weathered clay-sized fractions (Sparks, 1995). Secondary minerals are derived from the weathering of primary minerals, some common secondary minerals in soils include alumino-silicates such as kaolinite or montmorillonite, oxides 19 such as gibbsite, goethite and birnessite, amorphous materials such as imogolite and allophane, and sulphur and carbonate materials (Sparks, 1995). Clay minerals are alumino-silicates (oxides of aluminium and silicon with smaller amounts of metal ions substituted within the crystal). The aluminium-oxygen combinations are the basic structural units; these units are bonded in such a way that sheets of each one result. The stacking of these sheets into layers, the bonding between layers and the substitution of aluminium or silicon for other ions give rise to the different minerals (Yong et al., 1992). Clay particles are plate shaped because the layer-lattice structure results in strong bondings in two directions and weak bonding between layers. The clay particle thickness depends upon those attraction forces between layers. Variation in specific surface area is primarily due to different thicknesses of the tabular particles. The specific surface area influences the activity of clay and its various colloidal properties. Theoretically speaking, this area could be estimated by measuring the size and shape distributions of the particles. Practically, the area is measured by estimating the amount of a liquid or gas required to cover the surface; for this purpose, the ethylene glycol method or the BET method are widely used (Yong et al., 1992). Substitution of one ion for another in the clay crystal lattice (isomorphous substitution) and imperfections at the surface, especially at the edges, lead to negative electric charges on clay particles (Yong et al., 1992). Soil minerals can exhibit two types of charge, permanent or constant charge, result of isomorphous substitution, which takes place when the mineral is formed, and variable or pH dependent charge. Examples of constant charge silicate minerals are smectites, vermiculite, mica and chlorite. Minerals such as kaolinite, and layer silicates coated with metal oxides and soil organic matter, are variable charge components of soil that change with protonation or deprotonation (pH changes) of-functional groups (Sparks, 1995). Cations from the pore water are attracted to the particle surface to compensate for the negative charge and reach equilibrium. These are the exchangeable cations and their number is the Cation Exchange Capacity (CEC) or the amount of negative charge per unit weight or per unit area of the clay particle (Yong et al., 1992). A major component of soil's CEC is due to inorganic secondary minerals and it affects the retention of inorganic and organic contaminants. The other major component is organic matter (Sparks, 1995). 20 Surface functional groups are molecular units that protrude from the solid surface into the soil solution (Sposito, 1989). The inorganic surface functional group of greatest abundance and reactivity in soil clays is the hydroxyl group exposed on the outer periphery of a mineral. This group is found in clay minerals and amorphous silicate minerals like allophane, but it is also important in other inorganic colloids such as oxides and oxyhydroxides. Clay minerals expose singly coordinated OH groups on the edge of surfaces created when crystallites are broken apart. In these broken crystallites, say kaolinite, at the edge of the octahedral sheet, OH groups are singly coordinated to A l 3 + cations and at the edge of tetrahedral units they are singly coordinated to S i 4 + cations. Because of the greater valence of Si, OH groups attached to them only tend to dissociate protons whereas OH's linked to A l tend to both bind and deprotonate. If a surface functional group reacts with an ion or a molecule dissolved in the soil solution to form a stable molecular unit, then a surface complex is formed and the formation reaction is termed surface complexation. If two or more functional groups of a single ligand are coordinated to a metal cation in a complex, the complex is termed a chelate. The complexes formed between surface functional groups and constituents of soil solution can be classified similarly to the complexes that form completely in an aqueous environment. That is, inner-sphere complexes will form when there is no water molecule bound between the surface functional group and the ion or molecule. Outer-sphere complexes result when there is at least one water molecule between the surface functional group and the ion or molecule. The bonding mechanisms for inner-sphere complexes involve ionic or covalent bonding or a combination of both, whereas outer-sphere complexes involve electrostatic bonding mechanisms (Sparks, 1995; Langmuir, 1999). 2.1.2 Organic Matter There are varying definitions for what constitutes Soil Organic Matter (SOM). Brady and Weil (1996) consider SOM encompasses all the organic components of soil: 1) living biomass (i.e. organic matter present as live microbial tissue), 2) dead roots and other recognizable tissues, and 3) largely amorphous and colloidal mixture of complex organic substances no longer identifiable as plant tissues called humus. For Stevenson (1982), SOM and humus can be considered 21 synonyms and account for the total organic compounds in soils excluding: undecayed plant and animal tissues, their partial decomposition products and the soil biomass. The decomposition of plant and animal residues is a biological process that is carried out by bacteria, actinomycetes and fungi. Part of the Carbon is used for the synthesis of body tissue (biomass) and part is incorporated into stable humus. The less resistant, identifiable biomolecules produced by microbial action are grouped as nonhumic substances, while the ill-defined, complex, resistant, polymeric compounds are called humic substances (Brady and Weil, 1996). In simple terms, humic substances are compounds in humus that are not synthesized directly to sustain the life cycles of the soil biomass (Sposito, 1989). . Humic substances have been classified into three chemical groupings based on solubility: 1) Fulvic acid, lowest in molecular weight and lightest in color, soluble in acid and alkali and most susceptible to microbial attack; 2) humic acid, medium in molecular weight and color, soluble in alkali but insoluble in acid and intermediate in resistance to degradation; and 3) humin, highest in molecular weight, darkest in color, insoluble in both alkali and acid and most resistant to microbial attack (Brady and Weil, 1996). Although humic substances are highly complex and much needs to be done to better define them, their functional groups and associated properties are better characterized. The most reactive functional groups with protons and metal cations include, in decreasing order of typical content: carboxyl, phenolic, alcoholic OH, quinone and ketonic carbonyl, amino and sulfhydryl groups. The total functional group acidity of humic substances is usually calculated as the sum of carboxyl and phenolic OH groups (Sposito, 1989). Fulvic and humic acids posses a much larger dissociable proton charge per unit mass than the typical cation exchange capacity of clay minerals, which translates into a high buffer capacity of both proton and metal cation concentrations in the soil solution. Additionally, since most of the total functional group acidity of humic substances dissociates between pH 5 and 7, humic and fulvic acid molecules generally bear a negative charge in soils (Sposito, 1989). Due to the high specific surface area and CEC, SOM is an important sorbent of plant macronutrients and micronutrients, heavy metal cations, and organic materials such as pesticides (Sparks, 1995). 22 Complexation of humic substances with metals can be beneficial or deleterious for the fate of metals in soils and waters, depending on whether the humic substance to which they strongly bind is mobile or not. Humic substances can decrease the toxicity of heavy metals to microbes (Sparks, 1995), but there has been evidence where these substances can enhance the mobility of Pt and other isotopes in soil systems (Cleveland and Rees, 1981). 2.1.3 Sorption & Precipitation Sorption can be defined as the accumulation of material at an interface between the solid surface and the liquid phase. Sorption is a general term that comprises chemical and physical adsorption, and precipitation (Sparks, 1995). Adsorption is a process whereby solutes in solution are attached to solids (organic or inorganic) surface to satisfy attraction forces. Adsorption differs from precipitation because it does not include the development of a three-dimensional molecular structure, even if such a structure grows on a surface (surface precipitate) (Sposito, 1989). These processes are governed by the solids surface properties and the physico-chemical properties of the aqueous solution (Yong et al., 1992). Adsorption Physical adsorption or retention via Van der Waals forces is weak and occurs when the ions or molecules in the soil solution are attracted to the soil components' surfaces (organic or inorganic) because of unsatisfied charges of the soil particles (Yong et al., 1992). These weak Van der Waals forces or London forces take place even in molecules with no permanent dipole because imbalances in electron distribution produce instantaneous dipoles i.e. i f two or more adjacent molecules synchronize their electric motion, electron-electron repulsion can be minimized while attraction is maximized. These forces decrease with distance according to 1/r 6 (Mc Bride, 1994). Physical adsorption occurs with polar or nonpolar molecules, particularly those of higher molecular weight (Sparks, 1995). Chemical adsorption refers to that attachment of solutes to solid surfaces by a covalent bonding i.e. an equal share of electrons. The interaction mechanism in chemical adsorption is difficult to distinguish from electrostatic adsorption except for higher energy levels required for chemical 23 adsorption. In general, the first adsorbed layer on a solid surface would be chemically bonded and subsequent layers would be held by Van der Waals forces (Yong et al., 1992). The matter that accumulates in two-dimensional molecular arrangements at an interface is the adsorbate. The solid surface on which it accumulates is the adsorbent. A molecule or an ion in the soil solution that has the potential to be adsorbed is called the adsorptive. Adsorption on soil particle surfaces can take place via three mechanisms: inner-sphere and outer-sphere complexes (explained earlier) and diffuse ion. The diffuse ion does not form a complex with the surface functional group but instead remains in solution and neutralizes charge in a delocalised sense. The diffuse ion swarm and the outer-sphere complexes involve almost exclusively electrostatic bonding or also called coulombic interactions (Sposito, 1989). Specific adsorption would be an equivalent of chemical adsorption and it is called specific because the covalent bonding that takes place in this mechanism depends on the particular or specific electron configurations of both the surface group and the complexed ion. On the other hand, physical adsorption can be related to the diffuse ion association and outer-sphere surface complexation, also called non-specific adsorption. Because of the presence of opposite charges on the colloid surface and in the liquid phase, an electric potential develops at the solid-liquid interphase, called the surface potential which is given by the Nerst equation and can be simplified to *F = 0.059 (PZC - pH) (2.1) where PZC is the point of zero charge or pH at which the resultant surface charge of the colloid becomes zero, and is expressed as potential in volts at 25° C. Theories that have been developed to study adsorption, ion exchange, and other surface reactions on clay silicates are the Helmholtz, Gouy-Chapman, and Stern double-layer theory (Sparks, 1995). In the Helmholtz double layer theory, the charge is considered to be uniformly distributed over the clay surface. The first layer of the double layer is formed by the charge of the clay and the second layer is in the liquid layer adjacent to the clay surface. The positive counterions in this zone are attracted to the clay surface, but at the same time they are free to distribute themselves 24 evenly throughout the solution phase. The potential decreases linearly with distance from the clay surface. The Gouy-Chapman double layer theory is similar to the double layer theory, but this one considers an exponential decrease of potential with distance away from the clay surface. In the Stern theory the first layer is similar to that of the previous theories. However, the second layer is divided into 1) a sublayer nearest the colloid surface, and 2) a diffuse layer. The decrease in potential is also divided into two parts. In the Stern layer (the first layer tightly packed with cations) the potential decreases with distance according to the Helmholtz theory. After that, in the diffuse layer, the decrease in potential follows the Gouy-Chapman theory. Figure 2.1 summarizes the layers and potential distributions considered in these theories. Helmhettt Qouy Stern Distance from particle surface! Figure 2.1. Schematic representation of ion and potential distribution in the double layer according to the theories of Helmholtz, Gouy, and Stern. denotes total potential; VF5, zeta potential; x, distance from particle surface; o, surface charge density; 8 thickness of Stern Layer (Krzic, 2000). Adsorption data are commonly expressed by an adsorption isotherm, which is a plot of the quantity of adsorbate retain by a solid vs. its equilibrium concentration in the solution phase. The shape of this isotherm suggests but does not confirm information about the adsorbate-adsorbent interaction. Isotherms have been classified in four major types. Other shapes have been observed but they are generally accepted as hybrids of the ones shown in Figure 2.2. 25 1- The L-type (Langmuir) isotherm reflects a relatively high affinity between the adsorbate and adsorbent and is usually indicative of chemisorption. 2- The S-type suggests adsorbate-adsorbate interactions that are greater than the adsorbate-adsorbent. This condition favours the clustering of adsorbate molecules at the surface because they bond more strongly with one another than with the adsorbent. 3- The C-type or constant partitioning isotherm suggests a constant relationship between adsorbate-adsorbent. This is usually observed only at low concentrations. 4- The H-type isotherm is indicative of very strong adsorbate-adsorbent interactions or chemisorption and it is often an extreme case of the L-type. Figure 2.2. Classification of adsorption isotherms (McBride, 1994). Thus far, Van der Waals or London forces, electrostatic or coulombic forces, and covalent bonds have been discussed in adsorption processes. Other types of bonds involved in adsorption include: Hydrogen bonds, ligand exchange, coordination bonding and hydrophobic bonding. Hydrogen bonds are a special kind of dipole-dipole bond in which H atom serves as a bridge between two electronegative atoms. These bonds are weak and additive and they can occur on 1) 26 O atoms of silicate surfaces, 2) edge OH groups, and 3) functional groups such as COOH, OH, and NH2 on organic colloids. The replacement of a ligand by an adsorbate due to stronger affinity of the adsorbate compared to the ligand is called ligand exchange. This exchange results in specific adsorption and leads to the formation of inner-sphere complexes. Similar to the coordinate covalent bonding, coordination bonding takes place when only one atom furnishes both electrons to the bond. This bond takes place when some organic ligands in nonhumic substances, fulvic and humic acids bind metal cations with more than 1 functional group, which leads to the formation of heterocyclic ring or chelate ring. Metals adsorbed by this mechanism are not easily exchanged because of the bonding strength. Hydrophobic bonding occurs when non-polar organic molecules prefer an environment less polar than that of the highly polar water. If some other less polar surface is present such as that of clay minerals or organic matter, then the uncharged organic molecule gets pushed away from water onto the soil. Precipitation Adsorption and precipitation are difficult to distinguish, but another difference between the two processes besides two and three-dimensional characteristics, is that precipitation is the formation of one compound that has a homogeneous chemical composition, whereas adsorption is usually considered as two chemical entities. As the concentration of an ion in solution increases, precipitation will not occur until the solubility product is surpassed. That is, some degree of supersaturation has to exist in order for new crystal nuclei to be formed, and as the crystal size increases its solubility decreases. This difference in solubility of the same solid phase arises from the fact that the small nuclei have higher surface energy than the larger crystallites. The status of the solution phase can be described by the ratio of the Ion Activity Product - IAP-(quotient of the ions in question) with the solubility product of the crystalline solid K s o . If 1AP/KS0<1 the solution is under saturated, i f IAP/K S 0=1 saturated and If IAP/K S 0>1 supersaturated. Because of the higher solubility of small crystal nuclei, precipitation can only 27 begin in a homogeneous solution if the solution is supersaturated by a large margin If IAP/KS O>100. In soil solutions heterogeneous nucleation (formation of crystal nuclei at surfaces of a different solid that is present before the initiation of precipitation) is more likely than homogeneous nucleation because other mineral and organic surfaces can catalyze the nucleation process of crystallization. The energy barrier to nucleation is decreased by these surfaces so this reduces the extent of supersaturation necessary for precipitation to be initiated. For many of the most abundant elements that are found in soils, such as A l , Fe, Si, Mn, Ca, and Mg, precipitation of mineral forms is common and can control the solubility of these elements. However, for most of the trace metals chemisorption is favored because of the natural low concentration of these elements in soils. Only when they are present in large quantities is precipitation favored. 2.1.3 Surface Complexation Modeling The adsorption isotherm and ion exchange models have limited applicability when working with complex and variable natural systems. More elaborate models based on double layer theory, which take a mechanistic and atomic scale approach to adsorption observations are called: electrostatic adsorption models or surface complexation models. These models can consider changes in pH, aqueous speciation, ionic strength, acid-base and complexing properties of different sites on the sorbing surface(s) (Langmuir, 1997). Common characteristics of surface complexation models are: consideration of surface charge balance, electrostatic potential terms, equilibrium constants, capacitances, and surface charge density. Differences among surface complexation models lie in the descriptions of the electrical double layer (Constant Capacitance, Double or Triple layer model), and how they describe changes in surface potential and surface charge from the surface of the sorbent phase to the bulk solution [Sparks, 1995; Langmuir, 1997]. The fundamental concepts common in all surface complexation models are (Dzombak and Morel, 1990; Langmuir, 1997): 28 • The sorbing surface is composed of specific functional groups (coordination sites) that react with sorbing solutes to form surface complexes in a manner analogous to the formation of aqueous complexes in the bulk solution. • Sorption reactions can be described with mass law equations, corrected for electrostatic effects using Electrical Double Layer theory. • Surface charge (a) and electrical potential ( ¥ ) are consequences of chemical reactions involving the surface functional groups. • The apparent binding constants (determined for the mass-law adsorption equations) are empirical parameters related to thermodynamic constants (also called intrinsic constants) via activity coefficients of the surface species. A brief explanation and discussion of the most common surface complexation approaches taken to predict metal adsorption behavior on hydrous ferric oxides and humic substances is presented in this section, along with a discussion of their fundamental assumptions and limitations. Hydrous Ferric Oxides (from Dzombak and Morel, 1990). Hydrous ferric oxides have been studied for their important sorption capabilities in deep soils, acid mine surface water and groundwaters. They are the dominant sorbents in nature because of their tendency to be finely dispersed and to coat other particles. For the sorbent in this situations, the application of the double layer model has been highly successful. A l l specifically sorbed ions are assigned to one surface layer and all non-specifically sorbed counter ions are assigned to the diffuse double layer, which is assumed to follow a Gouy-Chapman distribution (Figure 2.3). The net surface charge density is given by a = F [ rH - r 0 H + 2 (ZM rM) + I (ZA rA) ] (2.2) where F is the Faraday constant (96,485 C/mol); Z the valence of a sorbing ion; TH and TOH the sorption densities of protons and hydroxyl ions (mol/m2), respectively; and rM and rA the sorption densities of specifically sorbed cations and anions respectively. The surface charge density a (C/m2) is related to the potential at the surface ¥ (volts) by 29 a = (SRTee0c x 10 3) 1 / 2 • sinh (Z^F/IR!) (2.3) where R is the molar gas constant (8.314 J/mol • K), Tthe absolute temperature (K), e the dielectric constant of water (dimensionless), £o the permittivity of free space (8.854 x 10" C/Vm), and c the molar electrolyte concentration. Distdnce cr + Ci - o Figure 2.3. a) Schematic representation of ion binding on an oxide surface, and b) Potential decay in the diffuse layer. A = specific surface area (m2/g); S = Solid concentration (g/L) [Dzombak and Morel, 1990]. The surface protonation and deprotonation reactions at oxide surfaces can be expressed by = X O H 2 + = =XOH° + H + is app (2.4) =XOH° = =XO + H + (2.5) where =XOH 2 + , =XOH°, and XO" represent positively charged, neutral and negatively charged surface hydroxyl groups, and K ^ P and K ^ P are apparent acidity constants (defined earlier). The mass law equations corresponding to equations 2.4 and 2.5 are K: (= XOH°){H + } X O H 2 ) K a p p = ( S X O - ) { H + } (= XOH°) (2.6) (2.7) 30 where () represents concentrations and {} represent activities. K^p and K^p are apparent acidity constants because the include surface charge effects and hence are dependent on the extent of surface ionization. In order to obtain the intrinsic equilibrium constant, which is not conditioned to particular pH or ionic strength conditions, activity coefficients of the surface species need to be calculated. This is done by considering the coulombic term or the electrostatic work required in transporting ions through the interfacial potential gradient to the surface ^ i n t = A f l p pexp(AZPP//?7) (2.8) where AZ is the net change in the charge number of the surface species resulting from sorption, and the exponential term represents the Coulombic term or electrostatic work. This term allows the consideration of the effects of surface charge variations in surface complexation reactions. Adding this term, equations 2.6 and 2.7 change to = ( XOH°){H s } = e x p (_ j pv 1 / /^ 7 ;) (2.9) ( = X O H 2 ) o l K'n: = ( X ° ) { H n s } = exp(-.PF I RT) (2.10) a ] (=XOH°) where Hs + denotes a proton released at the surface but not yet transported to the bulk solution A similar approach is taken with adsorbed metal cations giving the following equations =XOH° + M 2 + = = X O M + + H + (2.11) =XOH° + M 2 + + H 2 0 = = X O M O H 2 + + H + (2.12) Solving surface complexation problems requires the solution of numerous simultaneous equations. These equations include: intrinsic constant adsorption expressions (mass action equations), material balance (mole balance equations), surface charge, and charge-potential equations. The problem may be formulated in a tableau format (Morel and Hering, 1993) and the resulting equations solved with a computer. 31 Humic Substances Surface complexation modeling on humic substances follows the fundamental concepts applicable to SC on HFO, but with a greater degree of complexity due to the tri-dimensional nature of humic substances and the multiple type of binding sites. Such complexities have translated in different approaches to modeling; one of such attempts is the Humic-Ion Binding Model or Model V (Tipping and Hurley, 1992). Model V has been chosen to explain SC on Humic substances due to its relative simplicity, good fit to experimental data, and similarity with other more complex models. Addiditonally, Model V and its improved versions have been implemented in the computer codes WinHumicV (Gustaffson, 1999) and W H A M (Tipping, 1994), which facilitate cation speciation in water and soil environments where the dominant reactive components are humic substances. , In model V , humic compounds can bind metal ions either in monodentate or in bidentate coordination. The interactions are described in terms of intrinsic equlibrium constants and electrostatic terms. The dissociation of a proton on a humic compound can be expressed by the reaction: RAHZ <z> RAz~l + H + (2.13) where RA is the bulk of a humic compound with an acid carboxylate or phenolic group and Z is the humic charge. The dissociation reaction is described by a variable equilibrium constant, K(Z): K(Z) = [ R A Z ~ { ] ' { ^ + ) =KmX -e2wZ (2.14) RAHZ where ^ n t is the intrinsic constant and w is an ionic-strength-dependent electrostatic interaction factor. The proton-binding groups are heterogeneous, having a range of intrinsic pK values. Two types of acid group are distinguished, denoted by A and B. The four most strongly acid groups are referred to as type A groups, and consist mainly of carboxylic acid groups, while the remaining type B groups represent weaker acids such as phenolic acids. Within each A and B type there are four different groups, present in equal amounts, the pK values of which are described in terms of 32 a median value, pKA or pKB, and a factor, ApKA or ApKB, that defines the range of the values. For instance, the four type A groups have pK values given by: pKH(i) = pKA+^-9-ApKA (2.15) where i = 1, 2, 3 or 4. The humic content of type B sites («B) is fixed at one-half the content of the type A (« A ) sites. Metal binding can take place at single proton dissociating sites (monodentate complex) and when sites are sufficiently close at bidentate sites. The equilibrium reactions are expressed in terms of metal-proton exchanges: RAHZ + M z + <=> RAM^+z-'> + H (Z+Z-l) T T + (2.16) 1 + M <=> RA2W RA, RA, ' (Z+z-2) •>M + 2H H (2.17) where Z is the humic charge, and z the charge of the M metal ion. The extent to which bidentate binding can take place is constrained by the proximity factor (/pr), which defines the likelihood of pairs of proton-binding groups being close enough to form bidentate sites. The negative logarithms of the corresponding intrinsic constants are denoted by P K M H A and p K M H B for type A and type B sites respectively. In the model, the diffuse layer is a zone around the humic substances in which counter-ions reside. Co-ions are excluded. The volume of the diffuse layer V D is calculated from humic geometry (assuming spheres) and the Debye-Hiickel parameter. The volume is defined according to the following formula: 107V 4/r_ M ' 3 f 1 r + - r (2.18) where M i s humic molecular weight, TV is Avogadro's number, r is the radius of the humic molecule, and K is the Debye-Hiickel parameter 33 Since different cations may have different affinity for the humic charges a selectivity coefficient K s c i is introduced: C o ( ^ = K s e l ( z ) - i ? Z m o d < , ) (2.19) c,(0 where co(i) is the concentration of species i in the diffuse layer, cs(i) is its concentration in solution, R is the ratio required for the sum of the counterion charges to balance the humic charge Z and zmodO') is the modulus of the charge of species i. 34 2.1.3 Atmospheric Dispersion Modeling This section explains some of the fundamentals of Gaussian dispersion derivation and modeling. Initially a point source pollutant release is considered and the corresponding equations for the plume are derived, continuing with the definition of dispersion coefficients and atmospheric stability categories. Modifications to the Gaussian plume equation due to the effect of the ground and the introduction of depositional effects are also presented, closing with the particular case of a line source, which is the case of highway pollutant release. For the derivation of the Gaussian plume equation consider Figure 2.4 (De Nevers, 1995), where the origin is at the ground level, at or beneath the point of emission, with the x-axis extending horizontally in the direction of the mean wind. The y-axis is in the horizontal plane perpendicular to the x-axis, and the z-axis extends vertically. The plume travels along or parallel to the x-axis. Figure 2.4. Coordinate system showing Gaussian distributions in the horizontal and vertical directions (De Nevers, 1995). The mathematical model of the Gaussian plume, like many other processes, is based on the physical principles of mass conservation. If a small cube of space at the center of a plume (Figure 2.5) and a material that is neither created nor destroyed are considered, then the material balance for the cube is given by equation (2.20): z 35 Figure 2.5. Small cube of space at the center of an atmospheric plume Accumulation rate = £ (flow rates in) - H (flow rates out) (2.20) The accumulation rate is the time derivative of the amount of material contained, which is the product of the concentration and the volume. The volume of the cube is not changing in time, so Accumulation rate = —{cV) = V—- AxAvAz— dt dt dt (2.21) Assuming no convection into or out of the cube because the cube is moving with the local wind velocity, but considering there are flows through all six faces of the cube due to turbulent mixing, then a measure of turbulent mixing has to be defined. Defining turbulent mixing is quite a complex task, but this variable has been approximated by a theoretical model, which is analogous to the equations of heat and molecular diffusion. Using this approximation; the flux of material being mixed across any surface is given by: 8c Flux (time rate of mass flow per unit area) = -K dn (2.22) where c = concentration, n = distance in the direction considered (x,y,z), K = turbulent dispersion coefficient The net flow into the cube in the "x" direction is given by the difference of flux at location x and location x +Ax (Net flow into the cube in "x" direction) -Kdc dx / a t > •Kdc\ dx / at x + Ax' Aj;Az (2.23) 36 The net flow in the y and z direction can be estimated in a similar manner. Substituting equation (2.23) with the additional z and y direction and equation (2.21) into equation (2.20) and dividing both sides by AxAyAz gives dc 8t (Kdc^ v dx j f Kdc} A X v & ; Ax ( Kdc^ dy V s S&y+Ay (Kdc" v & j r Kdc^ a t , , V SZ J a t z + A z \ 3Z J Ay • + Az (2.22) knowing that limit Kdc dx Ax^tO alx+Ax V Kdc dx Ax K82c dx2 (2.23) Equation (2.22) becomes ^ = K ^ + K ^ + K.^ dt dx2 dy2 dz2 (2.24) which is the equation for heat conduction in a solid with variables renamed. The solution to this equation when the pollutant is released instantaneously at the origin (x = y = z = 0) at t = 0 is equivalent to the instantaneous source problem in the conduction of heat in solids, for which the mathematical solution has been determined (Carslaw and Jaeger, 1959). Using the analogous solution for a three-dimensional problem, the resulting concentration is X c - • %(ntfl2{KxKKz) 1/2 exp 2 2 2 A x y z — + —+ — yKx Ky KZJ (2.25) where: X= mass of pollutant (g), K = turbulent dispersion coefficient (x, y, and z directions), t time (s) Assuming the wind velocity u flowing in the x direction, the effects of Kx on the plume movement are negligible in this direction compared to those caused by the wind. This makes the plume a two dimensional spreading case, for which the equation is f j \( ..2 (_ TT\2 *\ c = Qlu exp / ,, ( ^ )2 (2.26) 37 where c = concentration (g/m3), Q = emission (g/s), w = wind velocity (m/s), z = elevation at which the concentration is being calculated, and H = source height. The turbulent dispersion coefficient has traditionally been expressed through the following equations: Ky=0.5cr2y- (2.27) x Kz=Q.5a2z- (2.28) x Where a y and o~z are called the horizontal and vertical dispersion coefficients or the standard deviations of plume concentration distribution in the horizontal and vertical planes. These dispersion values are based on experimental data and vary with the turbulent structure of the atmosphere, height above the surface, surface roughness, sampling time over which the concentration is to be estimated, wind speed and distance from the source (Turner, 1970). Substituting a for K in equation 2.26 gives c- — exp 2nuoyor which is the simplest expression to predict concentration in Gaussian plumes at considerable distances above ground. In the absence of direct measurements on the fluctuation of wind direction, Pasquill (1961) proposed a method to estimate atmospheric dispersion coefficients based on experimental data reported for different atmospheric stability conditions. Additionally, the author suggested that the atmospheric stability could be estimated from the wind speed at a height of about 10 m and the incoming solar radiation (during the day) or the cloud cover (during the night). This gave rise to the popular stability category classification, which is shown in table 2.1. 2a, • + 2cr2 2\ (2.29) 38 Table 2.1. Key to atmospheric stability categories (from Turner, 1970). Day Night Surface wind Incoming solar radiation speed (at 10 m) m/s Strong Moderate Slight Thinly overcast or > 4/8 Low cloud Clear or < 3/8 cloud 0-2 A A - B B — — 2-3 A-B B C E F 3-5 B B-C C D E 5-6 C C-D D D D >6 C D D D D The effect of the ground Equation 2.29 is a good predictor of plume concentrations at considerable distances above ground; however, concentrations at ground level are usually the most necessary to predict. For this purpose, the effect of the ground as a surface where pollutants can be reflected back to air has to be taken into account. This is achieved by considering the ground as a mirror image, where concentrations are the result of the sum of the plume above it plus whatever the mirror image of the plume would be if the ground were not there. Then the equation that considers the effect of the ground is c = Q 2n uoa^ y z exp •0.5 y 2 -0.5 (z-H\ exp \ ° y ) + exp -0.5 'z + H^ (2.30) The effect of Deposition In the mass balance or continuity principle expressed earlier, a more realistic view would have to consider sources and sinks of mass, which affect the net flow through the boundaries of the cube. The sources may include the transformation of reactive pollutants into other species, while the sinks may include the wash out of particles through rain or snow (deposition) and dry deposition. As it is the case for this research, the concern is mainly due to non-reactive particulate pollutants that are carriers of metals. Therefore, sources are not added in the plume mass balance equation, 39 while sinks such as dry deposition have to be considered, due to their pollutant contribution to roadside soils and other surrounding areas. Wet deposition effects on the other hand, are considered implicitly by adding background measurements of deposition in the area to predictions made by Gaussian dispersion modeling. Deposition flux or deposition rate Fd can be estimated as the product of the concentration X</ and a deposition velocity Vj, computed at a reference height. Fd = Xd-Vd (2.31) where Fd is expressed in units of mass/area-time (e.g. pg/m2-s), Xd (pg/m3), Vd (m/s) Deposition velocity is a variable that depends upon factors such as the type and size of the pollutant particles, the roughness of the terrain and the type of ground cover and meteorological conditions (Ermak, 1977). An important component of the Deposition velocity, especially for particle sizes greater than 10 pm (Harwell, 2001) is the gravitational settling velocity or terminal settling velocity Vg. Gravitational settling velocity is usually calculated by the following expression also referred as Stokes law: v £gd_ ( 2 3 2 ) 18// where Vg = terminal settling velocity, p = particle density, g = gravitational acceleration, d = effective particle diameter, r\ = atmospheric dynamic viscosity In order to include the effects of other factors in the definition of deposition velocity for finer particulates (particle size <10 pm), some researchers have proposed to include besides gravitational settling velocity, the effects of Brownian particle motion and inertial impaction in the definition of deposition velocity (Slinn and Slinn, 1980; Pleim et al., 1984). According to this criterion, the deposition velocity is estimated as the inverse of a sum of resistances to pollutant transfer through various layers, plus gravitational settling terms: vd=- + rd + rardVg (2.33) 40 where Vj = deposition velocity (cm/s), Vg = gravitational settling velocity (cm/s), ra= aerodynamic resistance (s/cm), ra= deposition layer resistance (s/cm) For large settling velocities, deposition velocity approaches settling velocity, whereas for small settling velocities, Vj tends to be dominated by the r a and rj terms. Gaussian dispersion and deposition model The Gaussian plume model to predict downwind pollutant concentration, considering the effects of deposition, for a point source located at (0,0,h) and emitting at a constant rate Q is (Ermak, 1977): c = Q -vx(z-h) Vga2 HK2 2K 2n<jy(Jzu -(z-h)2 -(z + h)2 + e 2a; 2a; v,(z+/)) v,2a; K +^K2 K erfc z + h 42K 42a, (2-Where erfc = complementary error function, and v/ is a factor given by v/ = Vj- (l/2)Vg In the trivial case when deposition velocity and gravitational deposition velocity are zero, equation 2.34 reduces to the Gaussian plume model presented in equation 2.30. 2.1.4 Runoff Modeling Quantitative and qualitative predictions of highway runoff can be organized in three categories: • Regression methods • Simulation models • Statistical techniques Regression methods are the simplest for runoff prediction, but their use is generally limited to the regions or site where they are developed. Driscoll et al. (1990) reviewed the database available from different sources such as: Federal Highway Administration (FHWA), University of Washington, California Department of Transportation (CALTRANS) and the United States Geological Survey (USGS). In that study, data from 42 storm events were available and each regression method was applied. The results were subsequently compared to evaluate the accuracy of each method. The authors could not get any definite conclusions, arguing that more case studies would need to be analyzed. Each of the methods evaluated was based on different independent variables. In the FHW A studies, predictive equations estimate the Total Solids (TS) load in runoff from individual storms, based on two major components: accumulation of pollutants (TS), followed by pollutant wash off. Accumulation (build-up) is modeled as a linear function of accumulation period and traffic volume, and wash off is expressed as an exponential function of the runoff rate. Any other pollutant than TS is predicted as a linear correlation from TS. In the Washington State Method, the major independent variable is the number of vehicles while the pavement is still wet, and the predicted variable is, again, TS. In this method it is assumed that vehicles kinetic energy rather than rainfall encourages transport of contaminants. The C A L T R A N S and USGS studies use multiple regression methods taking independent variables such as: ADT, dry days before storm, runoff volume, and storm duration. Simulation models can be regarded as an improvement over simple regression methods, they are the most demanding of site-specific data, including runoff quality data. Actually, most simulation models are no more than complex regression methods that also need to be calibrated, with the great difference that they can give spatial and temporal distributions of pollutants and are more flexible to incorporate changes in highway design or implementations that might affect resulting pollutant loadings. Driscoll et al., reviewed some of the widely used simulation models and concentrated on those that were operational, that had documentation, user support, and support by a government agency. Models that met those requirements were Storm Water Management Model (SWMM), Storage Treatment Overflow Runoff Model (STORM), Hydrological Simulation Program -FORTRAN (HSPF), and the FHWA Urban Highway Storm Drainage Model. A l l these models simulate build up of pollutants during dry periods, followed by wash off during storms. Statistical techniques have the advantage over regression methods, in that they can give a probabilistic distribution of quality parameters instead of just a mean value. This is particularly 42 useful when analysing impacts on receiving water bodies, because the occurrence probability or event frequency of a certain pollutant concentration in water might provide a more realistic evaluation than just considering the mean concentration value over a period of time.. The Probabilistic Dilution Model P D M (FHWA, 1990) belongs to this category of models and allows the user to compute the magnitude and frequency of occurrence of in-stream concentrations of a pollutant under the variable and intermittent discharges that are produced by stormwater runoff. The procedure compares the once in a 3-year concentration to an acutely toxic value that is specified by EPA criteria (Driscoll et al., 1990). 2.1.5 Infiltration Modeling The methods for the estimation of water infiltration rate through the vadose zone can be divided in three general categories: 1) empirical models, 2) Green-Ampt models, and 3) Richards equations models. In this section, the fundamental assumptions and considerations of the Green-Ampt model in particular are discussed, due to its simplicity and wide application in various hydrological problems. The use of more sophisticated models (such as those based on the nonlinear Richards equation) are considered too impractical and beyond the purpose of estimating runoff infiltration on roadside soils. Additionally, extensive work has been done to develop empirical equations to estimate the Green-Ampt model parameters, which facilitates its use and has encouraged its inclusion in multiple watershed models (USEPA, 1998). Green and Ampt assumed a piston-type water content profile (Figure 2.6) with a well-defined wetting front. The piston-type profile assumes the soil is saturated at a volumetric water content of 6S (except for entrapped air) down to the wetting front. At the wetting front, the water contents drops abruptly to an antecedent value of Oo, which is the initial water content. The soil-water pressure head at the wetting front is assumed to be ///(negative). Soil-water pressure at the surface, Hs, is assumed to be equal to the depth of the ponded water. At any time, t, the penetration of the infiltrating wetting front will be Z. Darcy's law can be stated as follows: dl v q = — = -K. dt 'hf-(H,+Z) I . Z (2.35) 43 where Ks is the hydraulic conductivity corresponding to the surface water content, and I(t) is the cumulative infiltration at time t, and is equal to Z(6S -Oo)- Using this relation for I(t) to eliminate Z from Eq. 2.35, and performing the integration yields, / = *,/-(A,-tf.xe.-ejiog, l -( ^ - ^ x o . - e o ) (2.36) \^ Wetting Front DepthZ -J e, 8< Actual r Green-Model a) Green-Amnt Parameters b) Water Content Profile Figure 2.6. Green-Ampt parameters and the conceptualized water content profile showing the sharp wetting front ( U S E P A , 1998a). Equation 2.36 is the statement of the Green-Ampt model. This equation has been adapted to represent different infiltration conditions such as: a) the G - A model for Layered Systems [ G A L A Y E R ] (Flerchinger et al., 1988); b) Explicit G - A model for cumulative infiltration and infiltration rates as implicit functions of time [ G A E X P ] (Salvucci and Entekhabi, 1994); and c) Constant Flux G - A model for simulating water infiltration for non-ponding conditions [ G A C O N S T ] . 2.1.6 Multi-component Transport Modeling 44 Multi-component transport models incorporate geochemical models into flow and transport processes in porous media. Many geochemical and transport models have been developed and upgraded versions are continually released. The description provided herein is not meant to be exhaustive; rather it shows the wide spectrum of programs that can be used to predict subsurface fate and transport of metals in roadside soils. Models that are multi-dimensional or targeted for wider scale applications such as diagenesis, convection of temperature dependent fluids, heat transport, etc. have been left out of this discussion because they have no use in the highway environment. Metal Speciation Equilibrium Model for Surface and Ground Water (MINTEQ) is a geochemical model capable of calculating equilibrium aqueous speciation, adsorption, gas phase partitioning, solid phase saturation states, and precipitation-dissolution of metals. The improved version, MINTEQA2 can solve a broad range of chemical equilibrium problems. The model contains an extensive thermodynamic database and seven different algorithms for calculating adsorption (USEPA, 1991). G E O C H E M (recently renamed SOILCHEM) is able to compute chemical equilibria in natural soil solutions, including precipitation/dissolution. Both redox reactions and solid phases can be included or excluded, and exchange coefficients can be estimated. The user can choose to have variable pH and ionic strength. The code and documentation are suited for users disinclined to modify the code. WATEQ includes an extensive thermodynamic database. Equilibrium distributions in natural waters are calculated with effects of temperature, pH, and redox potential. Data for reactions not contained in the database can be easily incorporated. Activity coefficients can be computed using the extended Debye-Huckel and Davies formula or the mean salt method of Pitzer. The temperature range where the model can be used is 0 to 100°C so the user is cautioned to be careful when using the model in temperatures that deviate significantly from 25°C (USGS, 2001). PHREEQE is an evolution of WATEQ where equilibrium calculations are based on aqueous phase concentrations and charge neutrality and where no mass constraints are imposed on hydrogen and oxygen. 45 Chemical Species Transport in Groundwater System C H E M T R N (Noorishad et al., 1987), model simulates one-dimensional solute transport in a saturated porous medium. The processes of advection, dispersion, sorption, aqueous phase complex formation, and dissociation of water are included in the model. No thermodynamic database is provided with the program. Thus the user must supply the equilibrium constants and stoichiometric coefficients for the reactions of interest. PHREEQC (Parkhurst and Appelo, 1999) is a revised version that combines the geochemical model PHREEQE with a one-dimensional mixing cell concept to model the transport, dispersion and diffusion of solutes. The mixing cell approach divides the domain into a group of cells. At each time step, the fluid contained within each cell is advected and mixed between neighbouring cells, with the mixing proportions between cells selected so solute dispersion is simulated (Parkhurst and Appelo, 1999). Additionally the model can make use of databases contained in WATEQ and MINTEQ, which expands the number of elements that can be modeled. PHREEQC contains a thermodynamic database for surface complexation reactions on hydrous ferric oxides derived from characterizations and modeling by Dzomback and Morel (1990). This makes it useful for subsurface transport of metals in roadside soils since these soils are generally comprised of sand and gravel with little or null amounts of clay minerals. Hence, most of the retention or retardation of metal movement in the inorganic component of soil is due to highly reactive iron and manganese oxide concretions and coatings on coarse particulates. The flexibility of the program to incorporate changes in the existing database, to include different databases, and a user friendly windows interface' make it quite suitable to model metal behavior in subsurface roadside soils. Additionally, an extensive user community provides constant upgrading and debugging of the program to make it work more efficiently. 2.2 Literature Review This section summarizes some of the most relevant research on highway environmental monitoring. It provides an overview of the most prevalent metals in highways, their sources, and attempts that have been undertaken to estimate their speciation or partitioning in multiple media. 46 2.2.1 Metals in the Highway Environment Sources of many contaminants that are generated or accumulated in the highway environment (highway surfaces, median areas and adjoining right of ways) have been documented in the literature (Young et al., 1996). A summary of common highway contaminants and their primary sources is provided in Table 2.2. More recently, Ozaki et al. (2004) have provided an update on metal sources in Japanese highways. They reported Pb coming from paint for road markings and from anticorrosive paint in guardrails. Diesel soot is an important source of N i , Cu, Zn and As, whereas tire rubber may have large quantities of Cd and Zn. The authors reported a sharp increase of these metals in alpine soils of some tourist areas, and attributed this to the significant increase of vehicles in recent years. Councell et al. (2004) estimated that approximately 10,000-11,000 tons of Zn are released from tire wear in the U.S. They also suggest that tire wear inputs to urban and suburban watersheds can be significantly greater than atmospheric inputs. The magnitude of these contaminants on roadways is a result of variables such as (Gupta et al., 1981): • Traffic characteristics (speed, volume, braking, etc.) • Climatic conditions (intensity and form of precipitation, wind, temperature) • Maintenance Policies (sweeping, mowing, repair, deicing, etc.) • Surrounding land use (residential, commercial, industrial, rural) • Percent pervious and impervious areas • Age of car and its maintenance • Littering laws and regulations covering car emissions and delivery trucks • Use of special fuel additives in vehicular operation • Vegetation type on the highway right of way • Accidental spills 47 Table 2.2. Common highway contaminants and their primary sources (Young et al., 1996) Constituent Primary Source Particulates Nitrogen Phosphorus Lead Zinc Iron Copper Cadmium Pavement wear, vehicles, atmosphere, highway maintenance Atmosphere, roadside fertilizer application Leaded gasoline (auto exhaust), tire wear (lead oxide filler material), lubricating oil and grease, bearing wear Tire wear (filler material), motor oil (stabilizing additive), grease Auto body rust, steel highway structures (guard rails, bridges, etc.), moving engine parts Metal plating, bearing and bushing wear, moving engine parts, brake lining wear, fungicides and insecticides Tire wear (filler material), insecticide application Chromium Nickel Metal plating, moving engine parts, break lining wear Diesel fuel and gasoline (exhaust), lubricating oil, metal plating, bushing wear, brake lining wear, asphalt paving Manganese Engine parts wear, combustion products of Methylcyclopentadienyl Manganese Tricarbonyl fuel additive (MMT) Cyanide Anticake compound (used to keep deicing salt granular) Sodium, Calcium, Chloride Deicing salts Sulphate Petroleum Roadway beds, fuel, deicing salts Spills, leaks or blow-by of motor lubricants, antifreeze and hydraulic fluids, asphalt surface leachate Previous studies have identified heavy metals and P A H (polyaromatic hydrocarbons) as the major contaminants of concern (USEPA, 1995; Maltby et al., 1995). Highways are sources of 48 considerable amounts of heavy metals, particularly lead (Pb), zinc (Zn), iron (Fe), cadmium (Cd), chromium (Cr), nickel (Ni) and copper (Cu). As these metals reach the ecosystem, they will undergo physical, chemical and biological transformations. They may be adsorbed on clay or oxide particles, taken up by plant and animal life or remain in solution. Toxicity depends largely on the physical and chemical form of the heavy metals, their availability to organisms and the existing conditions of the receiving environment. Water with high total metal concentrations may be less toxic than one having lower concentrations but with different species of the same metal. Ionic Cu, for instance, is more harmful to aquatic organisms than organically bound or elemental copper (Yousef et al., 1985). Therefore it is important to fully understand the different metal species present in the environmental media and to study the impact of those metal species on living organisms. 2.2.2 Metals in Highway Runoff In the past, numerous studies on highway runoff have concentrated on the determination of total metal concentrations and other constituents, mostly due to the sudden increase in the number of vehicles worldwide and the associated increase in contaminant release (Gupta et al., 1981; USEPA, 1983; Dupuis et al., 1985). Driscoll et al. (1990) examined the possible correlations between pollutants Event Mean Concentrations and TSS in the search for a method of estimating the concentrations of unmeasured constituents in highway runoff from suspended solids data. They found correlation coefficients of 0.67, 0.73 and 0.63 for Cu, Pb and Zn, respectively, between solids and metals concentrations. The order of association reported by these authors seems consistent with the ionic potential of these metals and their tendency to hydrate in solution and hence, their tendency to adsorb on to colloids surfaces (Langmuir, 1997). Driscoll et al. also suggested that a second benefit from the association of pollutants with suspended solids is to get a sense of the degree of removal that different control practices may exert. Those practices that target solids such as sedimentation or filtration would be anticipated to be effective in mitigating pollutants associated with them, while those pollutants that show no impact from these practices would be expected to be present in dissolved form. Local studies in the Brunette River watershed have shown that high concentrations of trace metals and hydrocarbons are associated with solids carried by storm water and that these pollutants 49 accumulate in the sediments when they are deposited throughout the watershed (McCallum, 1995; Larkin, 1995). In the final U.S. EPA Nationwide Urban Runoff Program (NURP) the Event Mean Concentrations E M C reported values were 34 ug/L for Cu, 144 pg/L for Pb and 160 pg/L for Zn (USEPA, 1983); however, most of that database was gathered when leaded gasoline was still in use. One particular pattern that researchers have found in highway runoff concentrations is that metal concentrations (either total or dissolved) are significantly greater at industrial sites than at rural, residential or commercial sites (Driscoll et al., 1990; Sanger et al., 1999). Metal Speciation in Highway Runoff Metals in polluted waters are distributed between the dissolved, colloidal and particulate phases, and living organisms. Metal speciation in highway runoff refers to the separation and determination of individual metal species and fractions (Morrison and Revitt, 1987). The speciation of heavy metals in highway runoff influences their transport, chemical reactions, bioavailability and toxicity. The toxicity of most metals to aquatic life is a function of the concentration of dissolved ionic forms of metals in water. Consequently, particulate metals are not directly toxic to most forms of aquatic life. Some factors that affect the relative concentration of dissolved and particulate bound metal include: type of metal, TSS, ionic strength, pH, alkalinity, hardness, bacterial activity, and amount of organic material (Morrison et al., 1990; Dempsey et al., 1993). A definite speciation of metals in natural waters is difficult to achieve. Some authors suggest the dissolved concentration of a trace metal (< 0.45 pm) is an approximate estimate of the fraction of total metal that is bioavailable (Charlesworth and Lees, 1999). However, the particulate bound metal has the potential too of becoming dissolved under the right conditions (e.g. pH, pE, ionic strength). Driscoll et al. (1990) summarized runoff events from 31 highway sites throughout the U.S. and suggested the following soluble (< 0.45 pm) fractions: Cu-40%, Pb-10%, and Zn-40%. Sansalone et al. (1996) instrumented lateral sheet runoff flow from a busy highway pavement section in Cincinnati, OH. They reported the highest geometric means of dissolved/particulate-bound metal ratios for the metals Zn, Mn and Cu. Lead showed intermediate values, while Fe 50 and A l were predominantly particulate-bound. They also found that dissolved/particulate-metal ratio tendencies are dependent on the runoff event duration and rainfall intensity. High rainfall intensities tend to carry greater amount of solids and associated metals, whereas milder and longer rainfall events produce an increase in dissolved / particulate-metal ratios. Prych and Ebbert (1986) reported similar results On metal variation throughout the storm hydrograph. Further fractionations of dissolved and particulate metal fractions have been suggested to provide a better estimate of metal speciation in waters. One successful scheme involves, on one hand, the fractionation of the dissolved phase into: 1) electro-chemically available [by Anodic Stripping Voltametry (ASV), Adsorptive Cathodic Stripping Voltametry (ACSV), Ion Selective Electrode (ISE) potentiometry (Weng et al., 2001)], 2) Chelex removable fraction, and 3) strongly bound fraction. On the other hand, the particulate metal fractionation involves a simplified scheme similar to the sequential extraction proposed by Tessier et al. (1979), where solids are separated into: 1) Exchangeable, 2) Carbonate and Hydrous Oxide, and 3) Organic fractions (Morrison and Revitt, 1987). Speciation or partitioning using chelating resins is based on the ability of the resin to remove metals by a process, which imitates metal uptake at the cell surface (Morrison, 1987). This technique has the advantage of being reliable, when the adsorption properties of the resin are well documented (Weng et al., 2001) and allows the use of standard laboratory procedures or equipment to analyze the trace metals attached to the resin. In a local study of runoff waters in the Brunette watershed, Yuan (2000) found that the low resin-exchangeable fraction of Cu indicated that it was largely in colloidal form. On the other hand, the author found that Zn showed high levels in ionic and weakly complexed forms. In another study, Morrison et al. (1984; 1988) separated toxic from non-toxic metal species in road runoff using anodic stripping voltametry, and ion exchange dialysis, reporting the percentages of dissolved ionic form with respect to total metal as: Zn (43%), Cd (74%), Pb (10%), and Cu (32%). The current U.S. EPA's policy approach is the use of dissolved metals to set and measure compliance with water quality standards (USEPA 1993). However, innovative regulatory approaches are emerging that combine factors such as: physical habitat, fish and macroinvertebrate assemblages, and chemical water quality. These approaches are labeled 51 "Biocriteria" (Davis, 1995; USEPA, 2004) and could become the dominant regulatory framework in years to come, particularly in sensitive ecosystems. 2.2.3 Metals in Atmospheric Particulates Metals are emitted to the atmosphere by natural and anthropogenic sources. Natural sources include: volcanoes, natural metal ore outcrops and weathering of rock material. Anthropogenic sources include: combustion of municipal solid waste, combustion of fuels in coal and oil fired power plants, releases from metal smelters, open cast mines, automobile emissions, and a large number of industrial stack emissions (Colman, 2001). Fuels and automobile exhaust emission contain several metals such as: A l , Ca, Co, Cr, Cu, Fe, K, L i , Mg, Mn, N i , Pb, Sr, and Zn (Hee, 1994). Ozaki et al (2004) reported that Zn is the metal exhibiting the highest concentration in exhaust soot from buses, regular and premium gasoline in Japan. Metals that followed in concentration included among others: Ni>Cu>Cd>Pb. In a comprehensive study on air quality in European cities, Kukkonen et al. (2003) found that Cu is mainly associated with gasoline exhaust. The authors also reported greater associations of metals Zn and Pb with fine fraction particulate matter (.< 1 pm), while Cu showed greater associations in the PM2.5 and PM10 range. Another air emissions study in France reported that the two main sources of Cu were road transport (48%) and other types of transport (31%). The main road transport contribution was from Diesel engine heavy-duty vehicles, which accounted for 14% of Cu air emissions (CITEPA, 2004). In a local study on the urban Brunette watershed precipitation quality, where the anthropogenic influence is predominantly due to vehicles, Belzer et al. (1997a) reported metal concentrations for precipitation samples, which showed the order: Fe>Zn>Mn>Cu>Pb. In another study, at the same location, Belzer et al. (1997b) reported that metals Zn and Pb in precipitation exceeded Health Canada water quality guidelines and that this could cause a short-term shock effect on aquatic systems. 52 Metal partitioning in atmospheric particulates Metals in suspended particulates are considered to be a health hazard when they are absorbed in the lung tissue. Wet and dry deposition act as an effective sink for metals in the atmosphere, but it results in metals accumulating in wide extensions of land and water. Although these measurements provide some indication of the general pollution level in an area, they do not provide information on the chemical speciation of metals (Fernandez et.al., 2002). Knowledge about the chemical speciation in airborne solids of metal and metalloids pollutants is urgently needed in order to develop appropriate environmental protection regulations and better air quality monitoring strategies for health protection (Lamoureux, 2004). Various partitioning schemes have been used to estimate the speciation of metals in atmospheric particulates (Table 2.3. Most of them have been modifications or exact reproductions of the method proposed by Tessier et al. (1979) for the partitioning of metals in river sediments. Some researchers have highlighted the limitations of such approach arguing that the extracting reagents may be too aggressive (Lum and Kokotich, 1987), or that the procedure may not mimic the conditions of deposition and solubilization into the lungs (Fernandez et al., 2002). Therefore, no established methodology exists for assessing the speciation of atmospheric particulates. Table 2.3. Summary of extraction procedures for atmospheric particulates Metal fraction released Authors Solution + Exchangeable Bound to carbonates Bound to Fe-Mn oxides Bound to organic matter Residual Lum et al., 1982 1 M MgCl 2 1 M NaOH/HOAc NH2OH«HCl in 25% HOAc H 20 2/HNO 3 + NH 4OAc/HN0 3 HN0 3/HC1/H 20 2 + HN0 3 /HF/H 2 0 2 Lum and Kokotich, 1987 Deionized distilled water H N O 3 + H 2 S0 4 Aqua Regia + HF Dreetz and Lund, 1992 1 M CH3COONa 1 M NaOH/HOAc NH2OH»HCl H 20 2 + NH 40Ac/HN0 3 HN0 3/HC10 4 Hlavay et al., 1996 1 M NH 4OAc NH2OH»HCl in 25% HOAc HN0 3 + HF Varga et al., 2000 Deionized distilled water 0.2 M NH 4OAc 2MHC1 6 M HN0 3 Aqua Regia + 22 M HF Poykio et al., 2002 Deionized distilled water + 1 M C H 3 C O O N H 4 NH2OH»HCl in 25% HOAc HN0 3 + HF + HC1 Fernandez et al., 2002 Deionized distilled water NH2OH«HCl in 25% HOAc H 20 2 + NH 4OAc HN0 3 /HC1/HC10 4 (6:2:5) 53 Dreetz and Lum (1992) performed fractionations on air-intake filters from a building located in a street with busy traffic in Norway. They reported the mobility of metals, based on the solubility in a sodium acetate solution, decreasing in the order: Zn>Pb>Mn>Cu>Fe. In a study about the fractionation of filter collected aerosol samples in a moderately polluted city in Hungary, Hlavay et al. (1996) reported the environmentally mobile character of elements based on solubility in a 1 mol/L NH40Ac solution. According to the authors, metal mobility decreases in the order Pb(72%) > Cu(18%) > Mn(10%) > Fe (2%). Hlavay et al. (1998) in another similar study of aerosol samples collected in the vicinity of a busy street, report mobile factors (exchangeable fractions) of metals following the order: Pb(40%) > Cu(13%) > Zn (12%) > Fe (18%) > Mn(5%). Varga et al. (2000) report that the exchangeable (water soluble) fraction of Zn can amount to 50-80%o of the total concentration, while Cu and Mn can reach 20-40% in certified reference material of urban particulate matter. On the influence of particle size on bioavailability, Hlavay et al. (1998) performed sequential chemical fractionations on different particle sizes of aerosol samples. The authors found maximum bioavailability of Pb in particles < 1 pm indicating traffic as the single anthropogenic source. They also reported that Mn would have differing bioavailable patterns among particles smaller or greater than 1 pm, reflecting anthropogenic as well as crustal sources of this metal in aerosols. In another study on the chemistry of atmospheric particulates, Querol et al. (1999) found that elements Cu, Mn and Zn exhibit increasing bioavailable metal concentration as particle size decreases, with 17-36% of their bulk levels in TSP, and 59-70% for the smaller PM10 and PM2.5 fractions. Lately, there are new promising techniques emerging for the chemical speciation of metals in solid airborne particulate matter, such as the Extended X-ray Absorption Fine Structure (EXAFS) spectrometry. E X A F S is a noninvasive, nondestructive, quantitative technique that allows analysis of samples in situ. Significant information such as the identity of the entities surrounding the excited atom, their distance from the excited atom as well as the number the backscatterers can be obtained. Lamoureux (2001) collected indoor air samples and surface dust. These samples were analyzed for lead, copper and zinc. Analysis of the dust sample from E X A F S for Cu near neighbor pairs indicated the presence of copper hydroxychloride 54 (Cu(OH)Cl) and for Zn near neighbor pairs indicated possible presence of a phosphate-containing Zn compound of the form Zn2Fe(P04)2-2.2.4 Metals in Roadside Soils Total Pb concentrations in roadside soils have been widely reported (Onyari et al., 1991; Sithole et al., 1993; Piron-Frenet et al., 1993; Hafen and Brinkmann, 1996; Fakayode and Olu-Owolabi, 2003), and to a lesser degree other metals as well, such as: Cd, Cu, Cr, N i , Mn, Pb, and Zn (Harrison et al., 1981; Ndiokwere, 1984; Albasel and Cottenie, 1985; Majdi and Persson, 1989; Mikkelsen et al., 1996; Turer et al., 2000). In roadside soils across British Columbia, the B.C. Ministry of Transportation and Highways has conducted GeoEnvironmental site investigations for highway improvement projects. Many of these sites have exhibited soil metal contamination above the Contaminated Sites Regulation (CSR) standards for industrial soils (B.C. M W L A P , 1997), which has triggered soil relocation or special handling procedures. A summary of metal contamination at these sites, along with traffic counts, soil pH and other data is provided in Table 2.4. As seen in this table, besides Pb contamination, a substantial presence of other metals has been found, most notably: Cr, Cu, Cd, and Zn 55 Table 2.4. GeoEnvironmental Site Assessments performed by the BC. MoTH Description of Highway project Soil CSR Location Metal Concentration pH Standard Traffic Vehicles/day (2002) TCH* Capilano w/b offramp North Vancouver Pb 454 5.7 . 250 21100 South Surrey Interchange Surrey Pb 268 5.9 250 26300 Lougheed Highway Cape Hornl/C Coquitlam Pb 922 5.7 250 50200 Lynn Valley Road on ramp widening North Vancouver Pb Cu Zn 1200 443 197 3.73 5.0 6.0 150 100 150 64260 TCH Sumas Vedder Canal Abbotsford Pb Cu 534 337 5.2 5.3 150 100 37715 TCH Cape Horn Coquitlam Cr 63 5.3 60 126748 Lougheed Ramps project (Coleman Ramp) Coquitlam Pb Zn Cu Cd 498 457 106 1.9 5.6 5.84 4.64 5.84 250 150 90 1.5 117753 264 St. Truck Lane construction (Hwy 13 widening) Langley Pb 413 5.3 100 6646 Intersection of Hwy 99 and Steveston Highway Richmond Pb Zn 549 • 230 5.7 6.2 250 . 200 TCH No. 3 Road Abbotsford Pb Cu Zn 612 344 217 5.9 5.4 5.1 250 100 150 3240 Highway 10- King George Intersection Surrey Pb 2400 7.03 1000 25164 Highway 7 at Dewdney Trunk Rd Maple Meadows Way Maple Ridge Pb Zn Cu 1290 333 125 5.3 5.3 5.3 150 150 100 50200 Port Mann Bridge 5 Laning Environmental (Johnson cut) Coquitlam-Surrey Pb Zn Cr 4000 432 181 4.3 5.9 6.8 150 150 60 20368 Stanley Park causeway sampling Vancouver Pb Cu 266 97 3.7 3.5 150 • 90 61745 Lonetree Creek S Curves Lions Bay Pb 293 5.0 150 12947 Highway 99 widening, Beach Ave-Nicomekl Br Surrey Pb 1680 6.1 250 22516 Old Hope Princeton Way widening Hope District Pb 116 5.4 100 4331 Highway 101: Wharf Rd-Conveyor Sechelt District Pb 636 4.6 100 3455 Metal partitioning in roadside soils Despite, the abundant literature on metal contamination, the literature on various metal partitioning in roadside soils is scarce. Additionally, there are no studies linking the relative 56 partitioning in different migration processes (atmospheric and runoff) with roadside soils, or studies that use a consistent extraction methodology (following a particular set of reagent solutions). Harrison et al. (1981) reported partitioning for metals Cd, Cu, Pb, and Zn in road dust and roadside soils in England. They reported that the carbonate and Fe-Mn fraction dominated for Cd, Pb, and Zn with very little metal found in the organic and residual phases, whereas for Cu the organic and residual phases were the most important. They also reported metal mobility following the order: Cd > Pb « Zn > Cu. Norrstrom and Jacks (1998) reported similar results in roadside soils of Sweden, but their study showed that a large part of the Pb, Cu and Zn in these soils is vulnerable to leaching, when exposed to high NaCl concentration, reducing conditions, or a lowering in pH. Lee et al.,(1998) performed leaching tests and sequential extractions in roadside soils with significant differences in carbonate content. They reported a metal mobility following the order: Cd>Zn»Pb , suggesting that in carbonate-free systems moderately acid rainstorms may leach Cd and lower amounts of Zn from the border of the motorway: Pagotto et al. (2001) reported that Pb and Cu did not appear to be highly mobile in roadside soils due to the association of the former with carbonates and the latter with organic matter. However, they did not report the amount of carbonates in these soils to assess whether the metal association with carbonates was actually present or just an artifact from the sequential extraction method. They also reported; 1) a high risk of Cd leaching downwards in the soil profile due to its highly exchangeable characteristics, and 2) an intermediate risk of mobility for Zn due to its high sensitivity to acid pH. Sansalone et al. (2001) reported the exchangeable percentages of metals in roadside soils of an Ohio hwy as: Pb (4.5%), Zn (8.3%), Cu (6.9%), and Cr (3.7%). They considered that non-soluble organic matter was a more important holding agent than clays in these soils. Turer and Maynard (2003), reported for roadside soils in Ohio and Texas that most of the Pb and Zn, and much of Cu were tightly bound to an insoluble form of organic matter in the soils. They found the resuspension of fine particles of these soils during highway maintenance or construction activities to be a greater risk than the remobilization of these metals within the soil. 57 Leaching studies and modeling predictions in roadside soils A few studies have dealt with the actual leaching in roadside soils, whether through batch, column leaching or lysimeter tests, under different environmental conditions. Kobriger and Geinopolos (1984), as part of a series of F H W A studies, reported zero tension lysimiter data for four highway study sites throughout the U.S. The lysimeters were installed right below the root zone of grassy areas to collect water constituents that were beyond the recirculation influence of roots. They found that Na, CI, Ca and metals were percolating below the root layer in the locations closer to the road. Mean metal concentrations (pg/L) ranged from: Cu (40-1150), Pb (20-1500), and Zn (100-880) [where the low concentration corresponded to a low traffic site (25,500 vehicles/day) and the high concentration to a busy hwy site (116,000)]. Igloria et al.( 1996a) assessed the efficiency of infiltration systems in removing metals Cd, Cu, Pb, and Zn by subjecting large-scale soil columns (0.3 m in diameter and 1 m deep) to a synthetic stormwater solution. They reported that geochemical characteristics were more relevant than hydraulic properties of the soils under study. Although they measured low concentrations of metals at the bottom of the infiltrating columns, they considered that infiltration technology can: 1) work as long as the soil-water systems assimilative capacity is not taken for granted, and 2) the technology is not used as disposal means for any and all wastes. On the effects of Natural Organic Matter (NOM) on heavy metal transport during infiltration, Igloria et al. (1996b) reported that N O M was very efficient at retaining metals and that it should be included as a sitting condition for infiltration structures. Lee and Touray (1998) performed batch desorption tests and sequential extractions in roadside soils and sediments from a retention pond in a French highway. They reported greater desorption rates in sediments from the pond, where carbonates had been leached, than in roadside soils that were regularly subjected to a lower and shorter water exposure. They also found that concentrations of metals desorbed from pond sediments with a HNO3 solution (pH=2) were in good agreement with the sum of the first sequential extractions of duplicate samples (exchangeable + carbonates + Fe-Mn oxides). 58 Delmas et al. (2002) performed batch adsorption/desorption and column leaching tests, for metals Zn and Pb, on roadside soil and pond sediment samples, under different solution conditions (pH, ionic strength, EDTA flushing solution). They found that neither Zn nor Pb were mobile under exposure to solutions with < 1 mol NaCl. They also found Zn to be more mobile than Pb under slight acidification (pH<6), whereas Pb was only significantly mobile at pH 4. In contrast, E D T A better mobilized Pb than Zn. Interestingly; they reported significant differences in the mobilization of Zn depending on the source. Zinc in roadside soils, which originated mainly from tire wear, was less mobile than the Zn accumulated in a retention pond, which originated from the guardrails. Marcos et al. (2002) developed a column leaching experiment on pure sand subjected to a flushing solution of calcium bromide, zinc bromide and lead nitrate. They studied the competition effect between these metals under simplified conditions (no clay or organic material present in the soil) and found through modeling and experimentation that the main metal migration mechanism could be described by outer-sphere complexation reactions. 59 C H A P T E R 3 M E T H O D S A N D M A T E R I A L S This section consists of two phases: 1) Environmental site investigation and monitoring 2) Laboratory investigation. The environmental site investigation was carried out to: • Estimate the relative contribution of atmospheric and runoff processes to metal migration • Observe the impact of maintenance practices on metal migration • Provide updated local estimates of metal concentrations in highway runoff that can be incorporated into a provincial database that could be used for modeling and watershed management purposes • Characterize or speciate metal loadings from different migration pathways The environmental monitoring stage was implemented at two highway sites: Highway 17 in Delta, B.C., and the Trans-Canada Highway at the 176th Street intersection in Surrey, B.C. Both highway sites have similar design i.e. elevated highway sections with overflow flush shoulder type of drainage, but with different surrounding geology, land use, average daily traffic, and predominant meteorological conditions. Having two different monitoring sites served two purposes: 1) to observe the influence of different environmental conditions on metal loadings and characteristics, and 2) to assess the capability of a predictive methodology in estimating metal loadings on roadside soils. The laboratory investigation included the physico-chemical characterization of soils, analysis and speciation of metals in atmospheric particulates, highway runoff and soil samples. Soil-metal interactions were estimated through batch adsorption and leaching tests under varying conditions of acidity, which also provided the experimental information for calibration of the geochemical model. Anthropogenic metal contributions in soil samples and atmospheric particulates were evaluated with the aid of lead isotopic analyses. 60 In this chapter, a discussion of site selection criteria and site characteristics is followed by a detailed description of the field and experimental methods used throughout this research. 3.1 Site Description Site selection followed the recommendations by Gupta et al. (1981) and Kobriger (1984) when monitoring sites in the United States as part of an extensive program sponsored by the Federal Highway Administration FHWA to identify constituents of highway runoff. Factors to consider in site selection include: traffic characteristics, surrounding land use activities, meteorological conditions, pavement type and condition, drainage area, highway design characteristics, proximity to a receiving water body, highway maintenance practices, and logistical considerations (safety, accessibility, future development, etc.). For this study, both sites share most characteristics (except for traffic numbers, land use and meteorological conditions) and are thus similar enough to be suitable for comparison, but different enough to test the sensitivity of a predictive methodology. 3.1.1 Highway 17 Site The study area is the right of way on both sides of Highway 17 in Delta, B.C. at the intersection with 34B Ave. overpass (Figure 3.1). At the study site location, this highway is oriented in a north-south direction. On a bigger scale Hwy 17 connects Tsawwassen and the ferry terminal to highway 99, which in turn connects to the rest of the Lower Mainland. The area is located in the flat lowlands of the Fraser River delta. The underlying parent material is composed of soil deposits, primarily silts, clays and sands. These sediments were deposited over thousands of years by seasonal floodwaters that spreaded across these lowlands. They are important agricultural soils although in some.cases poor drainage can be a problem. The B.C. M O E L P has classified aquifers in the area as having moderate to high vulnerability to contamination, with low to moderate use (Geological Survey of Canada, 1998). Later stages of this study also investigated the use of realistic atmospheric and runoff loadings to assess the accumulation and mobility of metals in soils. Therefore, it was decided to study metal migration patterns from the road in a site with simple geometrical, meteorological, topographical 61 and traffic conditions so these results could be compared and calibrated with a more complex site with greater traffic counts. Additionally, the dominant wind directions (easterly and westerly) with respect to the highway orientation (N-S) provided ideal conditions for atmospheric dispersion of pollutants. Figure 3.1. Highway 17 study site location in Delta, B.C. Table 3.1. Summary of Highway 17 characteristics in Delta, B .C . Characteristic Highway 17 Type Rural Average Daily Traffic (ADT) 20,417 vehilces (Northbound-NB) 22,899 vehicles (Southbound-SB) Drainage area 1286m2 (NB) 1368m2 (SB) Surface pavement Asphalt Number of lanes/direction 2 Type of section Elevated flush shoulder Surrounding land use Agricultural The construction of Highway 17 was completed in the early seventies. Surrounding land use in the area is agricultural. According to MoTH records, traffic counts for Hwy 17 are 20,417 and 22,899 vehicles/day for Northbound (NB) and Southbound (SB), respectively. The surface comprises three lanes in each direction made of asphalt pavement with flush shoulder surface drainage and a middle concrete barrier. A farm is located approximately 500m away from the 62 highway which provided power for the meteorological station, necessary to monitor precipitation and wind conditions. Highway 17 characteristics are summarized in Table 3.1. Background monitoring for Highway 17 Site Environment Canada's Canadian Wildlife Services facility in Reifel, B.C., located 35 kilometres south of Vancouver, at the northern end of Westham Island, in the Fraser River Estuary was selected as a convenient background location for dust deposition monitoring due its proximity to the highway 17 study site (approximately 9 km northwest of the site), and logistics which include: 1) The facility has a meteorological station where precipitation, temperature and wind conditions are reported on a regular basis. 2) The facility is part of the National Atmospheric and Deposition Program that is run throughout North America. Therefore access to standard dust deposition samplers and measurement data was granted. 3) Anthropogenic dust loadings in the area are only those from agricultural activities rather than vehicle related. 4) Availability of Environment Canada personnel to collect dust deposition samples when the researcher could not be present at the site. 3.1.2 Trans-Canada Highway & 176th St. Site Description This site is located on the northern right of way of the Trans-Canada Highway TCH (or simply Highway 1) close to the intersection with the 176th Street overpass in Surrey, B.C. The highway corridor is oriented in a NW-SE direction and through the Port Mann bridge connects Surrey to Coquitlam and Port Coquitlam (Figure 3.2). The surface geology is classified as Capilano sediments, which are composed of marine to glaciomarine stony to stoneless silty loam to clayey loam with minor sand and silt. These sediments are normally less than 3 m thick, but in some areas may reach up to 30 m thick. Aquifers in the area have low vulnerability to contaminants and low use (Geological Survey of Canada, 1998). Construction of Highway 1 in this location was completed in the 1960's. Surrounding land use is a mixture of residential, agricultural, undeveloped and parkland. An industrial area known as 63 Port Kells lies 890 m north and 1.55 km east of the study site. MoTH records show traffic counts of 82,900 vehicles per day for the west-bound (WB) corridor and 73,100 for the east-bound (EB). The surface comprises three lanes of asphalt pavement for each direction with a middle grassy area, flush shoulder type of surface drainage and a concrete barrier that separates on and off-ramps from the main corridor in the proximity of the 176th street overpass. Figure 3.2. Trans-Canada Highway Site location in Surrey, B.C. A Telus facility for wireless communication networking located north of the Highway 1 and west of the 176th street overpass provided power for the meteorological station and the high-volume air samplers. Characteristics of the study site are summarized in Table 3.2. Table 3.2. Summary of Trans-Canada highway characteristics in Surrey, B.C. Characteristic Trans-Canada Hwy Type Mixed residential, industrial and parkland ADT 82,900 (Westbound-WB) 73,100 (Eastbound-EB) Drainage area 500 m 2 Surface pavement Asphalt Number of lanes/direction 3 Type of section Elevated flush shoulder Surrounding land use Agricultural, residential This site presents more complex conditions in terms of meteorology, geometry and traffic counts necessary to corroborate findings from Hwy 17 regarding the migration and accumulation of 64 metals from atmospheric and runoff processes into roadside soils. It also provided material for evaluation of geochemical model sensitivity and the prediction of atmospheric and runoff loadings. Background monitoring for Highway 1 Site The yard of the Surrey fire hall No. 5, located 1 km north of the Highway 1 study site was selected as background dust deposition monitoring station. The facility provided security for monitoring equipment; it had a space clear of buildings or other significant physical influences of atmospheric disturbance; and it only had what was considered minor anthropogenic influences from the occasional transit of fire vehicles. 3.2 Site Investigation 3.2.1 Meteorological Conditions Meteorological conditions at both highways were monitored with a Davis® Weather station equipped with an external humidity and temperature sensor, 3-cup anemometer and wind vane, tipping bucket rain gauge, and a weather monitor II® console with data logger. Wind direction vane was aligned to the magnetic north (approximately 20° east from the true north). At the Highway 17 location, the weather station was mounted on the top of a barn located 500 m east of the study site having a height of approximately 10 m. Data at this location were collected between September 2001 and February 2003. At the Hwy 1 site, the station was mounted on top of the Telus facility for wireless communication networking located 20 m north of the highway westbound. Data for this site were collected from April through September 2003. Parameters measured at both sites included wind speed (m/s), wind direction (16 compass directions), air temperature (0.1°C) and rainfall (0.2mm increments). Meteorological data were collected at 5 seconds intervals and averaged over ten-minute periods. 3.2.2 Mass Balance of Metals at Highway 17 Mass balance calculations of solids and associated metals were performed at Hwy 17 for two periods of four months to assess the distribution of metals on the right of way from elevated 65 flush shoulder highway sections under "ideal" conditions of atmospheric dispersion. The solids targeted for the mass balance were those with particle sizes smaller than 250 pm. The solids with particle sizes < 250 pm were considered only, and not the broad range of solids and debris that can be collected throughout the highway. This was based on the findings reported by Kobriger and Geinopolos (1984) in a comprehensive set of highway environmental studies developed for the FHWA. In those studies, the authors confirmed previous findings where most of the heavy metal load and other pollutants were linked to the < 250 pm particle size group. Since solids removed by runoff and atmospheric processes are mostly in this particle size range, the authors defined the term TS250 (Total Solids < 250 pm) for mass balance calculations. A 100 m length pavement surface of Highway 17 was swept at the start of the monitoring period (March 2002) with a Johnston® 610 series, road sweeper. Subsequently runoff and atmospheric samples were collected until July 2002. During the mass balance experiment, the maintenance contractor was asked to minimize disturbance to the area avoiding any maintenance practices on the pavement surface that could affect the build up of pollutant on the surface. Such activities included painting horizontal signs or highway infrastructure, paving, mowing the vegetation on the right of way, sod removal on the edge of the highway, etc. In July 2002, the highway was swept again and the mass of dust accumulated recorded. For another four months dust, solids and associated metals in atmospheric particulates and runoff were recorded until the end of the 8-month monitoring period in November 2002. The TS250 mass input processes to highway 17 included: 1) atmospheric deposition, which was estimated from the background atmospheric dustfall measurements; and 2) vehicle related deposition, which was calculated as the difference of the total TS250 accumulation on the pavement surface minus the atmospheric dustfall. The TS250 mass output processes included: 1) runoff, which was estimated from the average suspended solids measurements and the estimated amount of runoff occurring during the monitoring period; 2) Atmospheric removal, was estimated from the dust gauge samples installed on the right of way; and 3) Sweeping mass, which was estimated from the amount of TS250 collected in road dust sweeping operations. Mass balance calculations were intended to provide an estimate of the accumulation of solids and associated metals on the highway surface due to pavement wear and vehicle sources (such as: 66 brake wear, fuel exhaust, tire wear, brakes, body rust, etc.) and the relative contribution of atmospheric and runoff contaminant removal processes on the highway right of way. 3.2.3 Road Dust Collection at Hwy 17 & Trans-Canada Highway (TCH) The road dust collection program was developed to estimate the accumulation rate of particulate and associated metal over a defined period of time on paved surfaces and to inventory the mass of particulate emissions from the road. Therefore, road dust collection before and after each monitoring period was performed in two ways. The sample that would be analysed for metals and grain size was collected with a miele® S3121 industrial vacuum that had been used previously by the Greater Vancouver Regional District (GVRD) -Air Quality and Assessment Division- to remove available sediment from the road surface and estimate fugitive dust emissions from paved roads. The Air Quality and Assessment Division made some modifications to the vacuum to minimize the input from metal parts wearing to the road dust sample. Additionally, a preliminary experiment was run to investigate the possibility of metal contamination from the vacuum cleaner, which turn out to be not statistically significant (Appendix D). Road dust collection followed the Procedures for Sampling Surface/Bulk Dust Loading of the USEPA AP-42 document -Appendix C . l - (USEPA, 1993) and recommendations from G V R D personnel. Six road dust samples were collected from the northbound and southbound lanes of Hwy 17 in Delta and three from the westbound of Hwy 1 in Surrey, using the vacuum cleaner mentioned and subsequently taken to the soils laboratory. The samples were homogenized according to A S T M and grain size analyses were done with standard stainless steel sieves. Nine aliquots of the 2 mm fraction were separated for total metal analyses. Since the surface to be swept was extensive, the rest of road dust was collected with a Johnston® 610 series, road sweeper such as the type that is regularly used by the local maintenance contractor to sweep the highways of the Lower Mainland. Besides speeding the sweeping at the end of each monitoring period, the use of the sweeper would give clues about the efficiency of sweeping as metal pollution control practice. 67 3.2.4 Atmospheric Dust Particulate Deposition The atmospheric migration of dust and associated metals was monitored with six dry Frisbee dust deposit gauges on the right of way (ROW). The validation and rationale for using this sampling methodology as the preferred procedure for dust deposition collection are described in detail in Appendix B of this document. Three Frisbees were located in normal transects to the road at both sides of the highway, one at the edge of the shoulder (0 m) and at 5 m and 10 m for the northbound ROW and 0, 6, and 12 m for the southbound ROW. The spacing of dust gauges was intended to provide a spatial distribution of atmospheric loadings onto roadside soils. In this case, the collection period was dependent on the amount of particulate collected on the gauges (at least two weeks). An initial assessment of sampling variation among three collocated Frisbees on the edge of the highway resulted in a 10 to 15% sampling difference. No such assessment was done on Frisbees located farther away from the highway. However, it was assumed that sampling variability in these gauges would be lower due to a lower vehicle induced turbulence. Frisbees were sampled according to protocols recommended by the Stockholm Environment Institute at York (Vallack, 1995). At the end of each collection period (usually one month), wearing plastic gloves the collecting foam disc was removed, placed in a clean bag and labelled with the site location and date of retrieval and taken to the laboratory for rinsing. Distilled water was poured into the aluminum Frisbee making sure that all dust particulates were washed from the Frisbee's surface to a connecting pipe and subsequently to the collecting bottle. The collecting bottle was removed, labeled and replaced with a clean one containing a suitable biocide. 3.2.5 Atmospheric Suspended Particulates at Highway 1 Atmospheric particulate matter includes solids and liquids in the range of 0.005 to about 100 pm. Most of the fine particles 0.005-10 pm are produced by combustion, evaporation and condensation processes. Particles that have sizes from about 0.5 to 5 pm are considered lung damaging. Tests have shown that particles larger than about 10 pm are removed in the nose and throat and very few get into the trachea or bronchi (De Nevers, 1995). Total Suspended Particulates TSP provide a measure of the amount of suspended particulate of different sizes up to a diameter cutoff of approximately 30-50 pm. 68 In recent years the G V R D has concentrated monitoring efforts in the Lower Mainland on respirable particulates such as P M 10 and P M 2.5 (Particulate Matter with an aerodynamic diameter < 10 or 2.5 microns respectively) rather than TSP. However, since one of the focuses of this research is on the impacts of atmospheric particulates and associated metals in neighboring environments close to the highway source, it was decided that TSP would provide a better estimate, along with dust deposition measurements, of particulate matter that is removed from the air and transferred to surfaces on the right of way and neighboring areas. The high volume air sampler has been the recommended standard sampler for TSP monitoring in B.C., according to the B .C . Ministry of Water, Land and Air Protection (1996). TSP was sampled with General Metal Works High Volume Air Samplers calibrated, prior to use in the field, by the Air Quality Shop of the G V R D . The high-volume air sampler typically samples air through a 20.3 cm by 25.4 cm filter at a rate of 1,132 L/min. Each High-volume sampler was collocated with a Frisbee dust gauge in a transect normal to the highway. 3.2.6 Highway Runoff Measurement and Sampling Runoff volumes were measured with one right angle aluminum v-notch weir box at each side of the road. The v-notch weir boxes had been used for previous research in the Department of Civil Engineering at U B C (Onwumere, 2000), hydrotechnical program, and were repaired and recycled for use in this research. The dimensions of the weir box are: 122 cm long, 61 cm wide and 31 cm deep. The notch is at 7.5 cm from the bottom of the box. Water head represents the difference in elevation from the water level and the bottom of the notch elevation. The flow rate or discharge was determined using the following equation for a v-notch weir (Granet, 1989): Q=1.38h n (37) where Q = discharge in m /s h = head above the weir crest in m n = 2.5 for v-notch weirs water level was measured manually on-site. This procedure was considered more appropriate since it did not require special measures for safety of the measuring instrumentation equipment 69 and practical, since several water level measurements could be made between the ten-minute sampling intervals. Samples were collected manually with one Litre plastic bottles at the edge of the highway just before water fell into the weir box. Sample collection was performed for a minimum of 1.5 hr period, or for the duration of the rain event, when the latter was shorter than the minimum desirable time. Both discrete sampling to describe runoff characteristics temporally, and composite sampling to get a summary of runoff characteristics for the rain event were performed at the site. Composite samples were mixed based on a constant time/volume proportional to flow-rate method. The bottles were stored in a cooler at 4°C and taken to the laboratory for analyses. 3.2.7 Soil Sampling 3.2.7.1 Detailed Surface Soil Sampling at Highway 17 The rationale for the detailed soil sampling was a) to provide a better description of pollutant migration via atmospheric and runoff processes on the right of way and b) to collect surface soil samples for isotopic analysis of Pb that would be compared with the isotopic ratio of the same element in atmospheric particulates. It was assumed that these analyses would aide to assess the relative anthropogenic metal contribution from the highway onto roadside soils. Surface soil sampling was performed parallel to the dry Frisbee dust gauge transects at 1 m steps out from the highway. Five centimetres of surface soil from an area enough to collect approximately 1 kg of humid soil were dug with a stainless steel shovel and pick. After sample retrieval, the stainless steel shovel and pick were cleaned with a rion-phosphate detergent followed by a potable water rinse to prevent cross contamination. A new pair of disposable latex gloves was used for each soil sample collected. Each soil sample was placed in a sealed plastic bag and stored in an ice pack chilled cooler until they were taken to the laboratory for analysis. Samples were placed in plastic bags and stored at 4°C until further metal and isotopic analyses. 70 3.2.7.2 Shallow Subsurface Soil Sampling at Highway 17 Shallow soil samples were taken by hand digging the upper 0 to 30 cm layer and a representative sample was collected every 5 cm. Past experiences from MoTH and by L i (2002) in roadside soils of B.C. have shown that metal contamination from above ground sources is generally restricted to the surface layers. Therefore, this sampling technique was considered a convenient preliminary investigation. Further soil exploration and sampling would then be justified i f metal contamination was suspected beyond this surface layer. Three sampling locations parallel to the Frisbee dust collectors at each side of the road were chosen to describe the horizontal and shallow vertical distribution of metal in. roadside soil. Sampling locations were labelled 1NB, 2NB, and 3NB for the North-Bound right of way at 0, 5 and 10 m from the edge of the road respectively and 1SB, 2SB, and 3SB for South-Bound soils at 0, 6, and 12 m. Soils samples were taken every 5 cm. After sample retrieval, the stainless steel shovel and pick were cleaned and samples stored according to the procedure described in the previous section (3.2.7). 3.2.7.3 Subsurface Soil Investigation at Highway 1 Soil sampling was performed in transects normal to the road at two locations. Both transects were located on the northern right of way, east and west from the 176th Street overpasss. Three sets, (each set composed of four boreholes to provide enough sample) were drilled at 0, 5 and 10 m from the edge of the highway at both locations. The rationale for selecting two locations was to assess the influence of traffic behaviour on the subsurface characteristics of roadside soils. The location on the east side of 176th St. overpass is influenced by continuous driving behaviour on the two central lanes, but also by some breaking from vehicles exiting Hwy 1 through the third lane. The west side location is influenced by continuous traffic as well, but also by acceleration from vehicles incorporating from 176th St to Hwy 1. Sampling depth was limited to 1.2 m since a preliminary investigation performed in summer 2000 showed that metal contamination should be constrained to that depth. 71 Soil exploration and sampling was done'during summer 2 0 0 1 using a track mounted Geoprobe®. Several other Geotechnical exploration and sampling techniques were considered including backhoe digging, auger sampling, Sonic drill, and Standard Penetration Test with split spoon sampling. Backhoe digging was discarded since it could only be used for sampling distant areas of the right of way (due to the presence of utilities and underground infrastructure). In addition, it was considered that it could cause major disturbance of soil samples and cross contamination. Auger sampling could not provide a clean undisturbed sample for physico-chemical analyses. Sonic drill was a feasible technique that would have provided good amount and reliability of samples in a single operation. However logistical conditions in terms of availability and cost forced the project to adopt other options. SPT with split spoon sampling was convenient only for limited sampling. Detailed soil samples were to be taken every 5 cm. Therefore SPT sampling had to be done as soon as the split spoon was retrieved. Additionally, the amount of soil that could be sampled with the split spoon would have forced multiple penetrations in a small area in order to get the necessary amount of soil for analyses. The Geoprobe is a direct push technique that uses percussion hammers and static vehicle weight combined with hydraulic cylinders to advance tools to depth (Figure 3.3). It is an alternative method to conventional drilling for collection of soil and groundwater samples and the installation of monitoring wells in unconsolidated materials such as clay, silt, sand, and gravel (McCall, 2 0 0 2 ) . This technique was chosen because sampling was quick, reliable, and versatile. In general, sampling would take less than 15 minutes including track positioning, sampler push and retrieval. The split sampler could be lined inside with a plastic tube that would preserve sample integrity and would allow for transport of the entire soil sampled in case it needed to be stored for further analyses. The small track allowed for vehicle versatility in different topography and surface conditions without much disturbance of highway infrastructure. Limitations of this technique included: a) small amount of sample, which had to be compensated for by a larger number of boreholes, b) small diameter (1.5"«3.81 cm) caused interruption and sometimes relocation of boreholes whenever bigger pieces of gravel were found. 72 Figure 3.3. Track mounted Geoprobe sampling at the Trans-Canada Highway study site It was determined that in order to collect enough sample for laboratory analysis, four boreholes had to be pushed at each sampling location. Sampling configuration was a square shape of 50 cm per side, with the boreholes at each corner. Distance between boreholes was fixed at 50 cm to guarantee sample integrity and minimize disturbance, but at the same time close enough to represent the characteristics of the sampling location. Soil samples were cut every 5 cm in situ, retrieved from each of the probes, and poured into one plastic bag to form a composite soil sample for every depth (Figure 3.4). Samples were transported in an ice pack chilled cooler and stored at 4°C in the refrigerator of the Environmental Engineering Laboratory at UBC until further preparation and analysis was performed. 73 Figure 3.4. Plastic inner liners retrieved from the split sampler and opened to expose soil samples. 3.2.8 Infiltration Studies Roadside soils and ditches act as buffer zones for heavy metals and other pollutants that are transported via atmospheric and runoff processes. Precipitation and surface runoff may transport metals deposited on the pavement or grassy areas downwards into the soil. Therefore, double-ring infiltrometer studies were conducted at both sites to get an estimate of the actual infiltration rate, taking place in roadside soils and ditches. This method consists of driving two open cylinders, one inside the other, into the ground, partially filling them with water (or the particular liquid of concern) and maintaining the liquid at a constant level (ASTM, 1988). The volume of liquid added to the inner ring to maintain the liquid level constant is the measure of the volume of liquid that infiltrates the soil. The volume infiltrated during timed intervals was converted to an incremental infiltration velocity, expressed in cm/h and plotted versus elapsed time. 3.3 Laboratory Analyses and Investigation The laboratory investigation on atmospheric and runoff samples collected at both highway sites aided in estimating current metal inputs from highways onto neighbouring areas. Physical and chemical analysis of soil samples at different depths and distances from the highway provided insight into past accumulation and transport of metals through the roadside soil media. The speciation or partitioning of metals in atmospheric, runoff and soil samples provided with estimates of metal mobility or possible bioavailability with respect to the total amount of metal found in these samples. The soil-contaminant interactions investigated through batch adsorption 74 and leaching tests provided the experimental data to asses attenuation capacity of roadside soils sampled. 3.3.1 Dust Deposition and Total Suspended Particulates Dust Deposition After the collecting foam was retrieved from the dish dust gauge, the foam was thoroughly rinsed in the laboratory with 1 litre of distilled water in a clean, acid washed container. This water was later added to that collected in the field from wet deposition to form a bulk total dust deposition sample. A nine cm diameter quantitative filter was pre-weighted to the nearest 0.1 mg, after drying it on a watch glass in an oven for 1 hour at 80°C and equilibrating for 2 hours in a desiccator. The contents of the collecting bottle were filtered under suction using a three-piece funnel leading into a 1 litre Buchner flask. A wash bottle containing distilled water and a nylon rod with a rubber teat on the end, was used to loosen and rinse off any deposits inside the collecting bottle. Subsequent washings were passed through the filter. Once all the water with dust was passed through the quantitative filter, this was dried and equilibrated as before. The rate of dust deposition in mg/m2/day was calculated with the following expression: dust deposition filters were sectioned with a stainless steel cutter into eight triangle shaped pieces. Every other piece was selected. Thus, four pieces were composed into one sample and subjected to total metal digestion. The remaining four pieces of filter were subjected to a Sequential Chemical Extraction to estimate metal partitioning in dust deposition samples. In addition, some filters that were not used for metal analyses were subjected to particle size analysis on a Malvern Mastersizer 2000 laser diffractive instrument. ( W 2 - W i ) x 2 4 . 7 T (3.1) where W i = initial dry weight of filter (mg) W2= final dry weight of filter plus dust (mg) T = length of exposure period (days) 75 Total Suspended Particulates The procedure for the weighing of filters was based on USEPA 40 CFR 50 "Reference Method for the Determination of Suspended Matter in the Atmosphere (High-Volume Method)". After air samples had been collected, the 20.3 cm by 25.4 cm quartz fiber filters were returned to the laboratory and conditioned in a room of constant humidity and temperature. They were then gravimetrically tared. Once the post-field filter final weights had been obtained, the filter was subsampled by cutting a filter strip consisting of one-ninth of the overall filter and digested using a microwave technique according to USEPA "Compendium Method IO-3.1 for the Determination of Inorganic Compounds in Ambient Air". The extracts were analysed with a SpectrAA 220FS Atomic Adsorption Spectrometer (Varian Scientific Instruments). Additional filter strips were used for duplicate metal analyses and for Selective Extraction analyses to estimate metal partitioning in air suspended particulates. 3.3.2 Highway runoff analyses Total and dissolved (filterable) metal analyses were estimated on discrete and composite runoff samples along with measurements of pH, Suspended Solids and Specific Conductance according to the Standard Methods for Water and Wastewater (APHA, 1998). Metal speciation on the soluble water phase of composite samples was based on the scheme proposed by Morrison (1987). In this scheme, a runoff sample is put in contact with a Chelating Ion Exchange Resin (chelex) and hence, the removable metal fractions attached to the resin provide information on potentially bioavailable (free and weakly complexed) metal species in solution (Figura and McDuffle, 1980; Greenberg and Kingston, 1983). The resin manufacturer suggests either a column or batch method for extraction of free or weakly complexed trace metal and provides numerous references on the resin selectivity for trace metals over other cations. The column method was selected to achieve metal separation in order to have comparable results with previous work by Yuan (2000) on trace metal bioavailability in the Brunette River watershed in B.C. A composite runoff sample of 2 L was filtered through a 0.47 polymer filter, the remaining filtrate was exchanged with 6 g of Chelex-100 resin in a glass column of 25 cm length and 1.5 cm in diameter. After exchange, the resin was transferred to a 50 ml centrifuge tube and 20 mL of 2 M HNO3 acid were added. The resin and acid were shaken for 76 3 hr to elute trace metals and centrifuged afterwards for 20 min at 3500 rpm (2800 x g). The supernatant was poured through a Whatman # 4 filter into a 50 mL volumetric flask. Fifteen mL of de-ionized water were added to the resin remaining in the 50 mL centrifuge tube and shaken again for 0.5 hr and centrifuged for 20 min. The eluted metals in water were added to the 20 mL of HNO3 acid contained in the 50 mL volumetric flask and the sample made to volume. Resin extractable metals were analysed with an ICP-MS for (Cu, Fe, Pb, Mn and Zn) to give potentially bioavailable (weakly complexed) metal fractions. 3.3.3 Physical and Chemical Characterization of Soils This phase was undertaken to correlate physical and chemical characteristics of roadside soils with metal accumulation and attenuation processes, as well as to generate the data necessary for geochemical modeling. This characterization included the determination of: soil pH, cation exchange capacity, specific surface area, mineralogy, carbon content, humic/fulvic acid ratios, particle size analysis and specific gravity. 3.3.3.1 Soil pH Soil pH is one of the most informative measurements that can be made to determine soil characteristics (Thomas, 1996). The pH of soil samples were measured on air-dry samples sieved to < 2 mm following the distilled water method, where 10 g of soil and 20 mL of distilled water were placed in a container. The suspension was stirred for 30 minutes and let stand for about an hour to allow most of the suspended clay to settle before pH measurement. 3.3.3.2 Cation Exchange Capacity Cation Exchange Capacity CEC is the total amount of exchangeable cations that a soil can adsorb at a given pH (Sparks, 1995). The determination of CEC involves replacement of all exchangeable cations by another cation such as N r l / , removal of the excess NFL;+ cations with a dilute electrolyte solution, followed by replacement of all NH4+ ions with K + ion. Measurement of the number of displaced NFL;+ ions allows an estimate of the number of cation exchange sites per unit mass of soil material (Sumner and Miller, 1996). The concentrations of exchangeable Na, K, Ca and Mg cations were measured by atomic absorption spectrophotometer (Perkin Elmer 77 Model 306) and the N H 4 + concentration was measured by QuikChem A E Automated Ion Analyzer (Lachat Instruments). 3.3.3.3 Specific Surface Area SSA Specific surface area SSA is the total amount of area per mass of soil. SSA is a property dependent on particle size as the latter influences the area/volume ratio and it is composed of internal (within the lattices of secondary minerals) and external surface area. It is an important property for metal attenuation processes since it provides the space where contaminants can bind to reactive surfaces. SSA was measured by the Ethylene Glycol Monoethyl ether method (EGME) on the soil fraction that passed a No. 100 sieve. In this method, the weight of a monolayer of E G M E covering internal and external surfaces of soil is related to the weight of the same monolayer of E G M E covering 1 m 2 of surface to obtain the equivalent SSA of the soil (Sheldrick, 1984a). 3.3.3.4 Mineralogy It is essential to know the mineralogical characteristics of soils due to the influence of clay minerals on metal attenuation capabilities of soils. Since most soils in the right of way are composed of gravelly sandy fill material, it was suspected that the presence of clay minerals would not be substantial and that most of the geochemical reactions with metals would take place on oxides and sesquioxides surfaces, and organic matter. However, a minimum presence of clay minerals would still be important due to the high adsorption and buffer capacity of these colloids. Therefore, it was decided to run X-ray diffraction analyses on selected roadside soil samples at different depths and distances from the road that would aid to the mineralogical characterization of the media. A preliminary assessment was made on the < 2 mm fraction of roadside soils from the study sites by splitting out 1 to 2 g of soil sample, crushing it in a mortar and grinding it as a paste with propanol. The sample was loaded into a sample holder and left to dry for an hour. The X-ray diffraction pattern obtained gave an estimate of the proportion of clay and nonclay minerals. When the preliminary diffraction pattern was unclear, the nonclay minerals were separated from clay minerals by extracting a fine enough particle-size fraction through sedimentation methods. 78 Only when organic matter interference was suspected, this was removed using strong oxidizing agents such as sodium hypochlorite (NaOCl), but with the premise of doing as little as possible to the sample before exposing it to the x-ray beam. 3.3.3.5 Total Carbon Content In coarsely graded soils such as those in the study areas, most metal attenuation processes take place on organic matter, as well as in oxides and sesquioxides surfaces. Carbon content analyses were performed on samples taken at different depths and locations, to give a measure of organic carbon mass present in these coarse soils and hence, an estimate of the importance of metal-organic matter interactions. Total carbon measurements were done following the method of Sheldrick (1984b), using a L E C O induction furnace. This method is based on the measurement of CO2 as it gases off following the destruction of organic and inorganic carbon in samples under high temperatures. Although the method estimates total carbon (organic and inorganic), preliminary acid treatment of samples can be used to dissolve carbonates and thus remove their contribution to total carbon estimates. However, in this case, the soils studied tended to be acidic and a significant presence of carbonates was not observed, thus total carbon measurements were used as a direct indication of organic carbon. 3.3.3.6 Humic/Fulvic Acid Ratios Humic Substances are the major components of soil organic matter. They are amorphous, dark-coloured, hydrophilic, acidic, partly aromatic, chemically complex organic substances that range in molecular weight from a few hundred to several thousand (Shnitzer, 1982). It is widely accepted that organic functional groups present in humic substances are important complexing agents that affect metal mobility in natural soils (Sposito, 1989). Therefore, to further characterize organic carbon content and better understand the natural attenuation of metals in roadside soils, humic and fulvic acid contents were estimated on selected soil samples, following the method proposed by McKeaque (1968). Based on their solubility in alkali and acid, humic substances were partitioned into three main fractions: 1) Humic Acid HA, which is soluble in dilute alkali, but precipitated by acidification of the alkaline extract, 2) Fulvic Acid FA, which is 79 the humic fraction that remains in solution when the alkaline extract is acidified, i.e. is soluble in both dilute alkali and dilute acid, 3) humin, which is the humic fraction that cannot be extracted from the soil or sediment by dilute base and acid. Carbon determinations were made on H A and F A only, since these fractions are the most reactive and better suited to identify metal-humic substance interactions. 3.3.3.7 Bulk Density and Specific Gravity Bulk density and specific gravity determinations were useful for volumetric/gravimetric relationships to estimate porosity and pore volumes for runoff infiltration simulations in roadside soils. Bulk density was measured directly in the field by determining the weight/volume ratios of soil samples retrieved by the Geoprobe. Although a certain amount of compaction was expected from the blows of the sampler, it was considered a fair approximation to original field conditions. Specific gravity was determined on soil samples according to U.S. Standard A S T M 854-98. 3.3.4 Batch Adsorption and Leaching Test The adsorption capacity of roadside soils was evaluated through batch adsorption tests under different pH conditions and provided the experimental information needed for an experimental approach to geochemical modeling and for comparison with a surface-complexation modeling approach. Leaching tests were also carried out to evaluate the strength of soil-metal interactions and the kind of attenuation process taking place under different conditions of acidity. Batch adsorption tests Batch adsorption tests procedure followed USEPA guidelines (Roy et al., 1992) and was conducted at pH 2, 4 and 6 accounting for natural acidic soil conditions at the sites and the possibility of further acidification of roadside soils, through accidental spills of acid on the road or acid rain. Multi-element solutions (Cu, Mn, Pb, Zn,) were prepared with concentrations reflecting the relative proportion of these metals in highway runoff measured at each study site. A stock solution of 1000 ppm in 0.5% nitric acid was prepared for each metal from reagent grade chemicals in their nitrate form, and diluted to the desired working concentration. Adjustments of 80 pH were made with reagent grade HNO3 acid or NaOH. Ionic strengths of 0.001-0.01 M were achieved with a weak electrolyte solution of NaCl. Metal concentrations in solution for adsorption tests on samples from both sites followed the order Zn>Mn>Cu>Pb, with a range of concentratidns that attempted to reflect the possible metal contribution from highways in the span of 50 years. Factors considered in the design of metal solutions for the batch tests included: average precipitation at each study site, Site Median Concentration, Percent of dissolved (non-filterable) metal and runoff coefficient. A detailed account of the factors considered and calculations used for designing the concentration ranges used in batch tests for this research is provided in Appendix C. Batch samples were prepared in a 1:10 soil/solution ratio (2 g of soil:20 ml of metal solution). Samples were shaken for 24 hours using an end-over-end rotator at 29 rpm to reach equilibrium conditions at ambient laboratory temperature (approx. 25°C). Soil-solution separation was accomplished by centrifuging for 15 minutes at 3500 rpm (2800 x g). The supernatant was filtered through a Whatman No.4 filter paper to discard organic matter that remained floating after centrifugation, and analyzed for copper, lead, manganese and zinc with a SpectrAA 220FS Atomic Absorption Spectrometer (Varian Scientific Instruments). The amount of metals adsorbed from solution on roadside soils was calculated based on the differences between concentrations of metals in the initial solution and in the batch supernatant. Leaching Tests Leaching tests were conducted following the same soil/solution ratio as batch tests (1:10), resembling the German method listed in the Compilation and Evaluation of Leaching Test Methods document (USEPA, 1978). Kinetic studies were performed to assess the minimum equilibrium period, which resulted in approximately 16 h but which for convenience was set at 24 h. Although this method was not as realistic as performing a column leaching test, it was relatively simple and provided a measure of the effects of pH on metal release. It also proved useful as an initial estimate of metal mobility and as a calibration database for geochemical modeling in roadside soils. 81 3.3.5 Selective Extraction Sequential Extraction in Soils Selective Sequential Extraction SSE is a procedure developed for the partitioning of particulate trace metals into five fractions: 1) exchangeable, 2) bound to carbonates, 3) bound to Fe-Mn oxides, 4) bound to organic matter and 5) residual (Tessier et al., 1979). As the extraction proceeds from exchangeable to residual, so does the strength of the extracting chemical solution, which provides an operationally defined speciation of particulate trace metals. The sequential extraction procedure used in this research was that proposed by Yong et al. (1993), which is a slightly modified approach to that proposed by Tessier et al. (1979) [Table 3.3]. Use of such a procedure in previous Pb speciation studies of roadside soils (Li, 2002), justified its application here as it was considered to render comparable results. One of the modifications was the first step of extraction, where a K N O 3 solution is adjusted to the same pH value as in the original soil solution pH, instead of the direct extraction of exchangeable metals with a solution of 1 M M g C l 2 at pH 7 or a solution of 1 M NaOAc at pH 8.2. Table 3.3. Sequential chemical extractions for the partitioning of particulate metals and their respective reagents Fraction Metal Species Extracting Reagent Treatment Exchangeable 1 M K N O 3 adjusted to natural soil pH 25°C 1 hour agitation Carbonate 1 M NaOAc, adjusted to pH 5 with HOAc 25°C 1 hour agitation Oxide Organic Residual 0.04 M NH 2 OHHCl in 25% (v/v) HOAc 30%H,O 2(pH2 w/ HNO3) + 0.02 M HNO3 HNO3/HCI 96°C, 6 hours with intermittent agitation 1) 85°C, 2 h intermittent agitation; 2) addition of (H 20 2) 3 h intermittent agitation; 3) 3.2 M (NH4OAc) in 20% (v/v) HNO3 continuous agitation for 30 min Complete digestion 82 After the first treatment with K N 0 3 , the soil-extractant sample was centrifuged at 3500 rpm (2800 x g) for 15 min and the clear supernatant was analysed for Cu, Fe, Pb, Mn and Zn by Atomic Absorption Spectrometry. The residue was washed with distilled water, the pH of which was adjusted to that of the previous supernatant. Then it was centrifuged and the supernatant from the washing discarded. The soil residue was exposed to subsequent extractions as outlined in Table 3.3. Another change in the method by Tessier et al. was the use of HNO3 /HCI instead of HF-HCIO4. This change was mainly operational, since the use of HF-HCIO4 acid required dedicated facilities that were not available for this procedure. Additionally, HNO3/HCI acid digestion was considered suitable for residual extraction because total metal concentrations obtained using this acid compared well to the sum of the five sequential extractions. The nitric/hydrochloric acid digestion is considered a total "recoverable" metals digestion, which does not dissolve silicate minerals (USEPA, 1986). Hence, the procedure could not extract all metals bound to the crystal structure of some primary and secondary minerals. However, these metals are released very slowly and are not of environmental concern. Sequential Extractions in Atmospheric Particulate Matter The sequential extraction procedure for atmospheric dust particulates and Total Suspended Particulate matter was carried out as outlined above, except that the filter/solution ratio was much lower than the soil/solution ratio for which the sequential extraction procedure was originally designed. This modification was due to the need for a minimum amount of solution to carry out metal analyses. It is acknowledged that the lower solid/solution ratio could result in greater fractions of metals attributed to the first sequences of extraction i.e. the most labile or bioavailable fractions. Therefore, results by this approach represent only qualitative conservative estimates of metal partitioning in atmospheric particulate matter. After total metal digestion of four of the eight pieces of each deposition filter (as outlined previously in section 3.3.1), the remaining four pieces of each filter were folded and placed in centrifuge tubes for sequential extractions to be carried out. An analogous procedure was carried out for Total Suspended Particulate filter strips. After the extraction procedure was completed 83 and the solution-filter samples were centrifuged, the supernatant was carefully removed with a pipette, filtered and stored for metal analyses. 3.3.6 Metal Analysis Metal analyses on batch adsorption tests, solutions and sequential extractions were carried out using a Varian SpectrAA 220 Multi-element Fast Sequential Atomic Absorption Spectrometer. Lead concentrations in runoff samples were determined using Inductively Coupled Plasma-Mass Spectrometry on a Thermo Jarrel instrument. Glass and plastic ware used for preparation, experimental or storage purposes were washed, then acid washed for 24 hr, rinsed four times with tap water, rinsed three times with distilled water and let to air dry. QA/QC procedures and protocols consisted of 1 replicate analysis in 20 samples, method blanks, blank spikes and matrix spikes, 1 per batch. Multipoint calibrations were performed daily and verified every 20 samples, with a tolerance of ± 10% of initial calibration. The accuracy of the methodology was evaluated by analyzing a certified reference soil (BCSS-1). 3.3.7 Lead Isotopic Analysis Lead isotopic ratios were measured in atmospheric and soil samples to aid in the identification of anthropogenic and natural Pb sources. It was hypothesized that this identification would also aid in the interpretation of the anthropogenic contribution of other metals such as Cu, Mn and Zn. Iron and manganese oxides have been identified in some studies on highway pollution as the sediment fraction where lead accumulates most predominantly (Backstrom et al., 2004). Other authors arrived at a similar conclusion with a different approach to metal speciation on road sediments using a laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) analysis technique (Rauch et al., 2000). Therefore, to make the identification of anthropogenic Pb more viable, it was decided to conduct isotopic analyses on Pb samples obtained from the Fe and Mn oxides fraction of the sequential extraction procedure (fraction #3). Lead isotope method The identification of Pb isotopes from different sources is possible due to the variations in the isotopic compositions of Pb in different rocks and minerals arising from the radioactive decay of 84 2 j S U , and 2iZTh atoms to 7 U b Pb, z u / P b and ' °Tb . The abundance of Pb isotopes is often expressed in the environmental literature as the ratio of the^radiogenic isotopes ( 2 0 6 Pb/ 2 0 7 Pb, 2 0 8 Pb/ 2 0 6 Pb) or among geologists, as the ratio of the radiogenic Pb with the non-radiogenic 2 0 4 Pb isotope [ 2 0 6Pb/ 2 0 4Pb, 2 0 7 Pb/ 2 0 4 Pb, 2 0 8 Pb/ 2 0 4 Pb] (Monna et al., 1997). In this report, the environmental convention has been followed since it was sufficient to aid in the identification of the different sources of Pb. Sample preparation A 10 ml sample of hydroxylamine hydrochloride solution and the associated metals from fraction # 3 of the SSE was placed on a teflon® beaker and let dry on a hot plate. Seven ml of 24 M HF plus 1 ml of 14 M HNO3 acid were added to the solid residue and let stand for 24 h on a hot plate at 120°C to dissolve any silicates present. The sample was dried down and 8 ml of 6 M HC1 acid added. The acid solution was again let stand with a lid on for 24 h on the hot plate at 120°C. The sample was then dried down and 1 ml of 0.5 M HBr acid were added. The sample was then ready for Pb chemical purification on mini columns filled with exchange resin (200-400 mesh). The purpose of the chemical purification was to eliminate other elements from the sample solution and their possible interference during isotopic analyses. Before use, the resin was pre-treated with two washings of water, HBr and HC1 acid, followed by an extra 1 mL of 0.5 M HBr washing to eliminate any possible contamination. The Pb sample, in HBr solution, was passed through the resin column and the eluant discarded. The resin with complexed Pb was rinsed with 1 ml of 6 M HC1 and the eluant collected. The HCl-Pb eluant was dried down on the hot plate and the residue re-dissolved in 0.05 M HNO3 acid and stored for isotopic analysis. Analytical Method An initial Pb analysis was conducted by ICP-MS to determine the amount of Pb in the HNO3 acid matrix after the chemical purification was completed. Samples were then diluted to normalize all concentrations to around 200 ppb. Samples and standards were then doped with 50 ppb of Thallium so that the TI/ TI ratio could be determined simultaneously with the Pb isotopes of interest. The TI/ TI ratio was used to determine the instrumental mass bias, which 85 was then used to correct Pb isotopic ratios for instrumental mass bias. The validity of this correction was determined by running several Pb standards of known isotopic composition at the beginning of each day and after every second sample. The data were then corrected off-line for any drift in the standard value during the run (Appendix I). Isotopic ratios measurements were conducted on a Nu plasma Multi-collector ICP-MS in the Department of Earth and Ocean Sciences at UBC. Isotopic ratios were measured as 2 0 8 Pb/ 2 0 4 Pb, 2 0 7 Pb/ 2 0 4 Pb and 2 0 6 Pb/ 2 0 4 Pb. Sample preparation and analyses were performed under "clean room" conditions. A clean room is defined in ISO 14644-1 as "a room in which the concentration of airborne particles is controlled, and which is constructed and used in a manner to minimise the introduction, generation, and retention of particles inside the room and in which other relevant parameters, e.g. temperature, humidity, and pressure, are controlled as necessary" (Whyte, 1999). 3.4 Statistical Analyses Statistical techniques were used to aid in the sampling design, experimental design and interpretation of results. The computer software MINITAB® along with EXCEL® were used for all statistical computations. The statistical techniques used consisted mainly of elementary tools such as: a) simple descriptive statistics; b) probability and probability distributions; c) confidence intervals and sample size; d) hypotheses testing; e) correlation and regression; and f) analyses of variance (ANOVA) for testing hypotheses concerning the means of three or more populations. These tests give statistical significance to the results. They are mentioned whenever they have been used throughout this thesis, and they are described in further detail in Appendix H. 86 CHAPTER 4 RESULTS AND DISCUSSION Results from the field and laboratory investigations from both study sites (Hwy 17 and T C H in sections 4.1 and 4.2, respectively) are presented in this chapter along with discussions, which provide the basis for assessing the: 1) contribution of atmospheric and runoff processes on metal migration to neighbour roadside soils, 2) metal accumulation on the soil and distribution throughout the soil profile, 3) soil-metal interactions and their strength (mobility), and 4) the impact that some maintenance practices have on metal loadings. Results generally follow the sequence of methods described in Chapter 3. Some of these results provided justification for subsequent isotopic analyses with the aid of mass spectrometry to get further insight into the origin and pattern of accumulation of metal in the highway system (Chapter 5). 4.1 Highway 17 4.1.1 Meteorological Conditions Meteorological conditions at Highway 17 during the period of March-November of 2002 are summarized in Table 4.1. Temperatures range from a minimum of-5.2°C in March to a maximum of 28.3°C in July. The driest conditions occurred during August and the wettest in November, typical of this geographical region where summer accounts for the driest season and the winter months for the most precipitation (snow/rainfall). Wind directions were predominantly westerly (blowing from the west), consistent with the typical wind pattern in the region which tends to be easterly or westerly depending in general on the season of the year and more particularly on the time of the day (Environment Canada, 2002). Weather conditions in Delta were optimum for observing the impact of management practices such as sweeping on atmospheric and runoff export of metals, since dryer conditions at this site, resulted in greater pollutant build up on the pavement surface, while dominant wind direction and highway orientation facilitated greater dispersion of atmospheric particulates. 87 Table 4.1. Summary of meteorological conditions at Highway 17 during the monitoring period (March-November, 2002) Month Temperature Total Average Maximum Dominant Range Precipitation Wind Wind Speed Wind CO (mm) Speed (m/s) direction Max-Min (m/s) March 13.3,-5.2 22.0 3.3 19.7 ENE April 19.2,1.4 38.8 3.1 18.3 W May 21.9, 1.6 28.4 2.7 17.9 S June 27.2, 7.3 23.8 3.1 14.8 ESE July 28.3, 7.4 12.2 2.7 13.4 ESE August 27.3,9.1 8.0 2.3 11.2 W September 23.2, 5.0 26.4 2.3 15.2 W October 17.6, -2.9 18.0 1.5 12.1 W November 15.5,-4.6 73.6 2.6 18.8 W The parameters summarized on Table 4.1 such as precipitation and wind directions were important to understand runoff and atmospheric removal patterns, which are discussed in the following sections, while temperature and wind speed were important parameters to estimate atmospheric stability class and removal energy for subsequent atmospheric predictions. 4.1.2 Road Dust Physical characteristics from composite road dust samples taken during the 8-month monitoring period from the Highway 17 NB and SB corridors are summarized in Table 4.2. The total mass of road dust collected during the March-July period was the highest, suggesting that a greater amount of dust is removed by runoff during the wetter July-November period. Higher amounts of dust generation in the area and hence, dry deposition is expected during the drier spring and summer season. Additionally, since no salt spread over the pavement was recorded or other dust generating maintenance activity, the only sources of extra dust were those due to agricultural activities in the area, which tend to intensify in spring and summer months. Road dust accumulation rates for Highway 17 ranged from 52 g/m2 for the spring and summer season to 41 g/m for fall and winter. "Silt loadings" or the fraction of road dust < 75 mm vacuumed from the transited paved lanes at Hwy 17 ranged from 0.17 g/m during spring and 2 summer to 0.053 g/m in the fall and winter monitoring periods. 88 Table 4.2. Road dust physical characteristics at Hwy 17 Date of Collection Location Mass of road dust (Kg) % Finer 2 mm 1 mm 425 pm 250 pm 125 pm 63 pm 25/03/2002 NB N . A . 91 77 42 18 5 2 SB 96 88 55 23 4 1 7/11/02 N B 137 90 82 63 37 11 3 SB 95 87 68 41 14 6 10/29/02 NB 109 75 60 34 14 4 1 SB 81 70 43 20 5 2 Ball et al. (1996) suggested that the deposition and removal of road dust is a continuous process and that sediment accumulation on the road surface attains a maximum when equilibrium is reached. The authors added that equilibrium is achieved when the rate of deposition on the road surface is equal to the rate of removal and the sediment load is at maximum. In another study on some Australian roads, Birch and Scollen (2003) found that dust accumulation rates were irregular and non-linear, but increasing with time, further supporting Ball et al. findings. However, the authors could not report an equilibrium dust load due to an unusual wet season, which did not allow for further dust accumulation Although Hwy 17 had comparable ADT to the medium traffic road reported recently by Birch and Scollen, road dust accumulation rates were much greater. This difference was thought to be mainly due to agricultural activities in the area, which generated greater amounts of fugitive, dust that could deposit on the pavement surface. The attainment of equilibrium dust load conditions was hinted by the maintenance contractor sweeping records. This was later supported by a comprehensive road dust study report by the G V R D (section 4.2.2 of this document). Results from sieve analyses presented in Figure 4.1, show that road dust from each season had a distinct size distribution depending on the date of collection, with summer samples (July 11 t h, . 2002) exhibiting the finest distribution, the spring sample (March 25 t h, 2002) showing intermediate particle sizes and the winter sample (November 29 t h, 2002) showing the coarsest distribution. These results may be attributed to greater number of dry days during summer that favor deposition and transport of finer fugitive dust emissions whereas more humid conditions favor the aggregation of fine particulates for which greater kinetic energy is required to remove them from the pavement surface by moving traffic. Additionally, the curves not only showed the 89 differences in particle size but also the similarities in distribution shape, which was an indication of a similar material source or dust generation procedure. Part ic le S ize D is t r i bu t ion 0.01 Particle Size (mm) - # — NB (25/03/02) SB (25/03/02) - * - NB (11/07/2) - ^ S B (11/07/02) - * - N B (29/10/02) SB (29/10/02) Figure 4 .1. Particle size distribution of road dust samples collected at Highway 17 from the NorthBound (NB) and SouthBound (SB) corridors. Besides estimating road dust accumulation rates, results from this section were aimed to characterize what is called in atmospheric sciences as "silt loadings" over the pavement surface to get Particulate Matter emission factors for this particular site, and to aid in the allocation of metal mass for the mass balance calculation. 4.1.2.1 Metal Concentrations in Road Dust Total metal concentrations in road dust sampled by highway vacuuming during the highway sweeping operations were quite consistent throughout the 8-month monitoring period following the general order Fe>Zn>Mn>Cu > Pb (Table 4.3), although copper and lead concentrations did not show a statistical difference (t = 0.57, P = 0.584). A s mentioned in the literature review, these metals are linked with different sources: lead and nickel with gasoline, zinc with tires, copper with brake linings, manganese with moving metal parts and gasoline additives, etc. (Kobriger and Geinopolos, 1984). Additionally, asphalt wear is identified as a source of trace metals and other pollutants that originate from bitumen and rock aggregates (Lindgren, 1996). Table 4.3. Total metal concentrations in road dust sweepings at Hwy 17 (Mar-Nov, 2002) Metal (mg/Kg) Mar-25 Jul-11 Nov-29 NB SB NB SB NB SB Copper 101 70 87 126 104 127 Iron 30,150 29,567 22,800 30,200 31,000 30,200 Lead 98 72 84 51 181 55 Manganese 277 237 293 322 257 266 Zinc 394 270 336 435 346 411 Local studies investigating metal concentrations in streams and highway runoff in British Columbia reported total metal concentrations on road dust from streets and highways with traffic counts ranging from 19,000 to 80,000 vehicles per day (McCal lum, 1995; Onwumere, 2000). Interestingly, road dust total metal concentrations from those highways showed a similar trend as that reported for total metal concentrations at Highway 17 (Figure 4.2). A n earlier study by Wilber and Hunter (1979) in Lodi , New Jersey U . S . A found concentrations in road dust following the order Fe>Pb>Zn>Mn>Cu, which except for lead (now banned from gasoline) is the same as that reported for Highway 17. 1200 C o p p e r L e a d M a n g a n e s e Z i n c • Burnaby- (A) • R i c h m o n d - ( B ) O H w y 1 7 - ( C ) • R u p e r t S t - (D) H W i l l i n g d o n - ( E ) • C a n - w a y @ B r e n t w o o d - ( F ) I I B r e n t w o o d Mal l - (G) • L o u g h e e d Mal l - (H) m Nor th Rd- ( l ) H C a n - w a y @ s p e r l i n g - ( J ) • B ra id St- (K) 0 C a n - w a y @ M a y f i e l d - ( L ) Figure 4.2. Total metal trends in road dust collected in 1995 and 2000 throughout the Lower Mainland of Vancouver, B . C . Canada (McCal lum, 1995; Onwumere, 2000). 91 This similarity may indicate that the rate of metal accumulation follows a pattern consistent with metal generation from a particular vehicle source, except o f course in those areas where industrial or natural sources of metals have been identified. For example, considering that tires are the main source of zinc in the road environment and copper is mainly associated with brake lining wear (Gustafsson, 2002), the accumulation of zinc should be greater in highways since tires are wearing permanently, whereas brakes are mainly used in curves and intersections. The origin of manganese and iron in road dust is more obscure since these metals are present in high background concentrations in most environments. Further discussion on manganese origin and attempts to distinguish background from anthropogenic sources is presented in subsequent sections. The influence of particle size on road dust collected from Highway 17 was not directly investigated, although it was indirectly taken in to account by analyzing metal concentrations on deposited or suspended atmospheric particulates on the right of way (see sections 4.1.3, and 4.2.5). 4.1.2.2 Metal Partitioning Metal partitioning in road dust samples estimated through sequential extractions are presented on a percent basis in Figure 4.3. The carbonate fraction is not reported in sequential extraction analyses for Highway 17, since carbonates presence in road dust and sediments in the area are negligible and, as a result, sediments were mostly of an acidic nature. Road dust partitioning showed a consistent distribution of metals with M n and Fe present dominantly in the oxide and residual fractions, Cu predominantly attached to organics, Pb showed almost equal partitioning between residual and the other more labile fractions, and Z n showed a high degree of association with oxide and organic fractions. The highest exchangeable fraction in road dust corresponded to lead, followed by copper and zinc. This partitioning was also consistent with results reported by Wilber and Hunter (1979) where Pb exhibited the highest exchangeable percentages in road dust followed by Z n and Cu. However, Pb was still in use as alkyl lead in gasoline at the time, impacting road dust total Pb concentrations which reached up to 8,300 mg/Kg. Birch et al. (2003) reported similar findings when assessing bioavailability o f metals through E D T A and dilute HC1 extractions. The authors 92 reported that metals more readily released by this method were Pb (76-87%), Z n (50-64%) and C u (41-66%). According to Y i n g et al. (1992) dilute HC1 and E D T A extractions predict the uptake of Zn , Pb and C d by deposit feeders and a species of burrowing polychaete worm. Lead Copper o Q. E ra in — tn 3 73 T3 ra s. Jul 11 (NB) a Jul 11 (SB) a. E Mar 25 (NB) re tn *-» Mar 25 (SB) tn 3 T3 T3 Nov 29 (NB) ra i Nov 29 (SB) 0 20 40 60 80 100 • Exchangeable O Oxides • Organic • Residual Zinc (°/Q a EL E re tn TS •a ra £ Z I 20 40 Jul 11 (NB) Jul 11 (SB) Mar 25 (NB) Mar 25 (SB) Nov 29 (NB) Nov 29 (SB) 100 20 40 Jul 11 (NB) Jul 11 (SB) Mar 25 (NB) Mar 25 (SB) Nov 29 (NB) Nov 29 (SB) 100 O Exchangeable El Oxides • Organic • Residual Manganese (%) a a. E re in in 3 •D T3 ra £ Jul 11 (NB) Jul 11 (SB) Mar 25 (NB) Mar 25 (SB) Nov 29 (NB) Nov 29 (SB) 20 40 60 80 100 I Exchangeable • Oxides • Organic • Residual Exchangeable • Oxides • Organic • Residual Figure 4.3. Metal partitioning in road dust samples throughout the monitoring period for Northbound (NB) and Southbound (SB) locations Extraction analyses indicated that Pb and C u were the most labile metals from road dust generated on the pavement surface; however, the percentage of exchangeable metal was in general less than 3% (Figure 4.3). These results were subsequently compared to what was observed in samples from atmospheric and runoff mobilization mechanisms, and with the metal distribution in soil profiles (sections 4.1.3, 4.1.4, and 4.1.5). 93 4.1.3 Atmospheric Dustfall Particulates Monitoring of atmospheric dust loadings from highway sources was considered necessary to assess the importance of this removal process onto roadside soils by using more efficient dry Frisbee gauges than those apparatuses used in earlier studies. If atmospheric removal resulted in greater loadings than reported in the past, this would give further evidence to implement Best Management Practices ( B M P ' s ) as means to control this migration pathway of metals, especially in highways close to residential areas. Thus, the atmospheric migration pathway was an essential component in the estimation and characterization of metal input onto roadside soils. Background A tmospheric Dustfall Background estimates of pollutant loadings from atmospheric processes (rainfall and dustfall) were Collected at the Canadian Wildlife Services (CWS) facility at Westham Island. Atmospheric particulate collection by the dry Frisbee dust deposit gauge at the C W S background location were at least 40% greater than collection by the standard dry and wet deposition samplers altogether (Appendix B) . 50 -, ~ 45 •S 40 ^ 35 o) 30 £ 2 5 .2 2 0 •I 15 o 2- 10 Q 5 0 •Pi Mar Apr May Jun Jul Aug Monitoring Period Sep Oct Nov Figure 4.4. Background atmospheric deposition at the Canadian Wildlife Services facility in Reifel, B . C . during 2002. Hal l et al. (1993) demonstrated through wind tunnel experiments that standard dust samplers create significant turbulence around them to hinder dust deposition and hence can underestimate 94 deposition loadings. Vallack (1995) conducted a field evaluation for Frisbee-type of gauges in rural areas of Yorkshire, U . K . against standard dust samplers and concluded that the dry Frisbee with foam insert was superior in sampling efficiency to Frisbees with sticky coatings, dry Frisbee with no insert and standard dust samplers as well . Figure 4.4 shows background dust deposition measurements during the 8-month monitoring period (Mar-Nov, 2002). Roadside Atmospheric Dustfall Atmospheric loadings next to Highway 17 showed in most cases an exponential distribution, consistent with an atmospheric dispersion process. Extensive literature on lead contamination near highways has reported similar deposition patterns (Cholak et al., 1968; Atkins, 1969; Little and Wiffen, 1978), which has reflected on the same type of lead accumulation pattern in soils (Motto et al.1970; Musket and Jones 1980; Kobriger and Geinopolos 1984; Burguera and Burguera, 1988; Rodriguez-Flores and Rodriguez-Castellon 1982; Hafen and Brinkmann 1996). Figure 4.5 shows dustfall loadings at the beginning of the first four-month period (Mar-Jul, 2002), starting with a highway surface just recently swept and loadings close to the end of the first half of the period when road dust had accumulated on the pavement surface. Although the dust accumulated on the pavement surface plays a role in the amount of dustfall exported to roadside soils, the combination of vehicle wake and wind direction are also determinant factors (Moosmiiller et al., 1998). 500 n o 100 4 D . 0 i '• 1 1 1 1 1 1 1 0 2 4 6 8 10 12 14 Distance (m) Figure 4.5. Dust deposition pattern on the right of way of Highway 17 at the Northbound (NB) and Southbound (SB) locations, for different sampling periods during the March 25 to July 11, 2002 monitoring period. • 95 In the sampling periods depicted in Figure 4.5, the resultant average wind speed normal to the highway was 2 K m / h for the March 25-Apri l 8 period, and 6.8 K m / h for the June 3-June 25 sampling period. Therefore, the increase in dust deposition on the right o f way might be caused by a combination of the effect of wind blown-off and the increasing amount of mass accumulation of road dust on the pavement surface. This increase in deposition was significant even after considering the 10-15% sampling variability found on Frisbees located at the edge of the highway. Additionally, it can be observed that dust deposition at the edge of the highway was still about three times greater than those reported for the C W S background location. The deposition pattern throughout the 8-month monitoring period (Mar-Nov, 2002) confirmed that deposition was consistent overall with wind directions, i.e. when the dominant wind direction was westerly (coming from the west) dust deposition loadings tended to be higher on the northbound (NB) corridor, than on the southbound (SB) and vice versa, when winds showed easterly wind directions (Appendix B) . Exponential equations described deposition patterns with R 2 values that were generally greater than 0.9. Previous research has shown that Total Particulate Matter and associated metals decrease rapidly with distance from the road edge (Kobriger and Geinopolos, 1984). The authors found that particulate matter deposition followed a logarithmic distribution from the edge of the pavement at four study sites in the U . S . A . and that deposition would reach background levels around 35 to 50 m away from the highway. The authors also reported that atmospheric removal of contaminants onto neighbor areas accounted for up to 30% of the total removal processes, with highway runoff accounting for the rest. More recently, researchers studying the attenuation of particulate matter < 10pm in size (PMio) in unpaved roads report that 90% of the mass is lost as deposition within 50 m from the source (Watson et al., 1996). Results from this section stressed the hwy right of way as the main buffer area for the mitigation of road atmospheric particles and its associated pollutants. 4.1.3.2 Metals in Atmospheric Deposition Metal concentrations in fugitive dust particulates showed the order: Fe>Zn>Mn>Cu>Pb (Figure 4.6 & 4.8). Belzer et al. (1997a) reported a similar order of metal concentrations for precipitation 96 samples collected in the urban Brunette watershed, where the anthropogenic influence is predominantly due to vehicles. These similarities may indicate a characteristic trend of sites influenced by vehicular activity in the Lower Mainland. In another study, Belzer et al. (1997b) reported that metals Z n and Pb in precipitation exceeded Health Canada water quality guidelines and that this could cause a short-term shock effect on aquatic systems. The authors acknowledge there are no currently recognized standards for deposition rates, however Hoff (1996) has suggested an interim program to identify the regions in Canada where deposition fluxes for metals (except for Hg) exceed 100 pg/m /yr. Metal deposition fluxes at the H w y 17 study site exceeded the suggested 100 pg/m 2-yr limit during all the 8-month monitoring period, even for sampling locations at the limit Of the right of way (12 m). These results highlight the importance of restricting human-hwy interaction for prolonged periods of time in the proximity of the right of way. On the formulation of dustfall guidelines, Vallack and Shillito (1998) have proposed a method for monthly dustfall guidelines, based on the background levels normally expected. The authors propose adopting the concept of "likelihood of complaint", already in use for rating the impact of noise, avoiding the various difficulties in defining dust nuisance standards. However, this proposed guideline addresses exclusively dust loadings and not the particulate associated pollutants. £ 25 - -k- - Zn ---x--- Mn —e—Cu —e—Pb Figure 4.6. Metal deposition loadings at different distances from the edge of the road for the sampling period between Jun 25 t h -Jul 1 l m , 2002 i th 97 The order of metals found in atmospheric particulates (Fe>Zn>Mn>Cu>Pb) was consistent with concentrations found in road dust, except that for road dust there was a clear statistical difference between M n and C u concentrations (Fe>Zn>Mn>Cu>Pb). This may be due to the greater association of Cu with finer particulates, which were exported as fugitive dust and later collected on the Frisbee gauges. On this influence of particle size on metal content, El l is and Revitt (1981) did not find significant differences in M n and Fe concentrations for sediments with size fractions smaller or greater than 250 pm, which may explain the consistent rank of M n in road dust and dustfall samples. On the other hand, McCa l lum (1995) reported for Brunette watershed sediments that the increase in metal concentration, due to the increase in silt and clay content, was greater for C u and Zn than for M n and Fe, which may explain the observed increase in C u with respect to M n concentrations in atmospheric dustfall samples of Hwy 17. The observed relationship, between road dust and dustfall metal concentrations in the right of way of Highway 17, is partly a result of saltation as one of the dominant processes of pollutant export within the right o f way. Saltation occurs when sand-sized particles are lifted by vehicle wake and later re-deposited, disrupting aggregate structure and ejecting smaller particles (<10pm) that are not usually suspended by direct wind action, but that can travel long distances once suspended (Young et al., 2002). Bul l in et al. (1979) found similar Fe and T i ratios between suspended particulates and road dust from roadways in Texas, suggesting that their origin was the resuspension of roadway silt. More recently, Young et al. (2002) found that resuspension of contaminated soils near highways and industrial facilities can be an important source of airborne Pb. Figure 4.7 shows particle size distribution curves of particulates deposited on the right of way at different distances from the edge of the road. The curves were similar in shape, but the curves also showed that a greater amount of fine particulates were deposited as the distance from the road increased. This is consitent with the saltation and resuspension process mentioned earlier, where at a distance of 10 m the amount of deposited particles smaller than 10 pm increased up to 10%. 98 Part ic le S ize (pm) Figure 4.7. Particle size distribution curves of atmospheric particulates deposited on the right of way at different distances from the edge of the road. Pearson correlation analyses were performed to assess the level o f association between metals collected at different distances from the road. A l l metals showed high degrees of correlation with Fe and M n , which highlights their natural association with iron and manganese oxides in local sediments and the importance of Fe and M n coatings in regulating the mobility of these metals. Copper and Z n showed high degrees of association (0.68-0.90) in the coarser particulates that deposited at the edge of the highway and at the middle of the right of way. Lead and Zn, showed the greatest association at the edge of the road (0.76). Lead and M n presented a similar pattern showing a Pearson correlation coefficient of (0.70). These degrees of association are presented in Table 4.4 for the edge of the road sampling location. Table 4.4. Levels of association between Metals in dust deposition, expressed by Pearson Correlation coefficients (upper box) and corresponding "P" values ( ower box) Cu Fe Pb Mn Fe 0.91 >0.01 Pb 0.72 0.65 0.02 0.04 M n 0.94 0.98 0.70 >0.01 >0.01 0.02 Zn 0.97 0.86 0.76 0.90 >0.01 >0.01 0.01 >0.01 99 a. c « -4—» c O c o O O m 5 m 1 0 m O m 5 m 1 0 m Copper Lead O m 5 m 10m O m 5 m 10m Manganese Zinc Figure 4.8. Summary statistics for metal concentration in dust deposition (pg/g) at different distances from the edge of the highway. (Note: Circles represent mean concentrations, boxes are 90% confidence interval of the mean, bars across the boxes are medians, and whiskers are minimum and maximum values). Figure 4.8 shows summary statistics of metal concentrations from 10 dustfall samples, retrieved throughout the 8-month observation period, at different distances from the edge of the road. The figure shows the greater scatter of values closer to the road, but also the general order of concentrations mentioned earlier, which resembled the order of road dust concentrations. 4.1.3.3 Metal Partitioning in Atmospheric Deposition Metal speciation in atmospheric dustfall particulates was carried out following a slightly modified (Yong et al., 1993) sequential extraction procedure proposed by Tessier et al. (1979). Speciations in atmospheric deposition samples are summarized in Figure 4.9 according to their relative partitioning in different fractions. Although a five-step extraction procedure was performed (exchangeable, carbonate, oxides, organic and residual), the less mobile carbonate-oxide, as well as the organic-residual results have been combined to aid on the graphic display of metal speciation deposition rates. 100 1SB 2 S B ^ T ~ S - « T - ^ l o c a t i o n Ex • Car & Oxi • Org & Res • Total 3SB l o c a t i o n 1NB 2NB 3NB Ex • Car & Oxi • Org & Res • Total Note: Ex = Exchangeable, Car&Oxi = Carbonate and Oxides, Org&Res = Organic and Residual Figure 4.9. Metal speciation in dust deposition rates from Hwy 17 samples during the Jul-Nov, 2002 monitoring period: a) lead, b) copper, c) Manganese, d) Zinc Figure 4.9 shows metal speciation on deposition samples gathered at different distances from the road at Highway 17. This metal speciation in dustfall particulates resembled the speciation shown previously for road dust. Metals partitioned predominantly among oxide, organic and silicate phases, with the exception of Zn , which seemed to be more exchangeable in dustfall than 101 in road sediment samples. A s particle size decreases, more metal is expected to leach, due to the anthropogenic nature o f finer atmospheric particulates (Paciga and Jervis, 1976). If the percentage of exchangeable metal is used as a mobility indicator, this metal leachability effect in dustfall samples from H w y 17 was not as evident for metals other than Zn, perhaps due to a masking effect of the mass of road resuspended sand sized particles that were also deposited on dust gauges. In the past, when Pb was used as an antiknock additive, Biggins and Harrison (1979) reported that lead sulfate compounds were dominant. They also reported that these compounds had a significant solubility under M g C k extraction conditions. L u m et al. (1982) reported that although great amounts of Pb are leached from urban particulate matter, little Pb may leach from road dust or roadside soils due to significant compositional differences between these materials. These differences result in a greater impact from metal deposition at interfaces such as forest canopies and at the air-water interface than in soils (Lum et al., 1987). Table 4.5. Summary of Speciation of metals in dustfall samples at H w y 17 for the Jul-Nov, 2002 monitoring period Mean Mean Carbonate Organic Total Total & & Cone. Loading Exchangeable Oxides Residual Element (M-g/g) (pg/m2/day) % % % C u 57+ 13 1 0 ± 4 3 50 47 Fe 5 6 3 6 ± 9 5 9 1031± 319) 47 53 Pb 2 8 ± 6 5+1 4 61 35 M n 6 2 ± 1 1 11 + 4 2 54 44 Z n 125 ± 2 7 23 ± 8 9 55 36 Note: Mean total metal concentrations are based on 30 dust deposition samples. Speciation percentages are based on the mean of six dust deposition samples. Table 4.5 summarizes average total metal concentrations on a per gram basis or as the traditional depositional flux units. Confidence intervals (95%) of the mean total concentrations and the average percentages that were obtained for each chemical extraction are reported as well . Separate HNO3 /HCI acid metal digestions were performed to compare with total metal 102 concentrations obtained by addition of the sequential extractions. Differences between separate sequential extractions were below 20%, which for the natural variability of the deposited material was considered satisfactory (Appendix E). The effects of particle size on metal speciation were further investigated in this research, by performing sequential extractions on road dust, roadside soil, dust deposition and Total Suspended Particulates collected at the busier traffic conditions o f T C H (section 4.2). 4.1.4 Highway Runoff 4.1.4.1 Rainfall-Runoff Relationships Flush shoulder pavement sections are designed with surface transverse slopes that allow for runoff to flow to the shoulders and drain through the sides along the highway. Therefore, all runoff is expected to drain equally along the highway on horizontal terrain. In the case of Highway 17, runoff was collected at two lower points on each side of the highway where a slight longitudinal water flow was also induced in order to have enough water sample and a continuous runoff measurement during a rain event. Therefore, in this case, runoff measurements rather than providing the rate of pollutant export per meter length of highway, represent runoff at that particular point of measurement for a greater contributing area of pavement. This allows for comparison between expected and the actual runoff caused by the particular rainfall event. The expected runoff is the product of the precipitation (m) during the rain event and the contributing area, while the actual runoff also depends on the losses from permeable surfaces, splash off from driving vehicles and errors in the determination o f the catchment area. The term runoff coefficient is a measure of the ratio between the actual and expected runoff volumes, and it is a useful and simple parameter for estimating pollutant exports from catchment areas where more complex variables such as evaporation, retention, infiltration, etc. have not been accounted for (Chow et al., 1988). Runoff coefficients for monitored rainfall events at highway 17 are presented in Table 4.6. 103 Table 4.6 shows that runoff coefficients from the southbound location are twice or more times greater than the northbound location, which may be a result of greater losses through the permeable shoulders in the northbound versus southbound catchment area. Gupta et al. (1981) reported runoff coefficients in the range of 0.30 to 0.43 for highways with percentages of surface paved varying from 27 to 37 %, while for local highways with a 100% of paved surface in Richmond and Burnaby, B . C . , Onmwumere (2000) reports runoff coefficients ranging from 0.55 to 0.70 respectively. Therefore, runoff coefficients measured at Hwy 17 along with those reported by other researchers (Scharff et al., 2003), would result in a runoff coefficient <50 % for hydrologic modeling or pollutant export purposes in flush shoulder sections. Table 4.6. Rainfall-Runoff characteristics for rainfall events monitored at Hig^ hway 17 Date& Location Northbound (NB) Southbound (SB) Time sampling started Daily Rainfall (mm) Site Rainfall (mm) Duration (min) Dry days Before Rain Event Dry days Between Rain Events Estimated Runoff Coefficient 13/04/02 (NB) 23:00 6.2 1.00 72 none 8 0.34 (SB) 0.64 16/04/02 (NB) 12:04 2.4 0.40 60 none none 0.14 (SB) 0.39 26/04/02 (NB) 12:30 5.6 1.50 75 9 9 0.15 (SB) 0.41 28/06/02(NB) 14:00 10.8 1.72 90 none 42 0.10 (SB) 0.27 03/10/02 (NB) 00:10 6.4 2.88 120 none 74 0.20 (SB) 0.38 06/11/02 (NB) 14:30 5.4 1.40 110 none 13 0.17 (SB) 0.38 29/01/03 (NB) 12:10 7.6 2.7 105 2 ' 41 0.37 Another explanation for the difference in runoff coefficients on both sides of the highway are sources of error which may include: catchment area estimation and runoff/rainfall measuring devices. Both northbound and southbound sides were carefully surveyed and the catchment area determined instrumentally with a transit and visually by recording the contributing area during rain events. On the other hand, runoff discharge measurements were made manually several times and averaged to estimate the right height of water over the V-notch weirs. However, as the head over the notch decreases, the uncertainty in the discharge coefficient of the weir increases. 104 Therefore limitations in runoff coefficients determinations should be considered, particularly for the events that exhibited less amount of rain. The possibility of splash-off influence generated by vehicles can be considered by comparing the difference between runoff coefficients measured during day traffic hours and those measured in late hours. Runoff coefficients for the northbound side are greater during late hours sampling than most other day hour sampling. Similarly, the highest runoff coefficient for both the northbound and southbound sides was measured during late hours, when the influence of traffic volume and speed is not generally significant. The number of dry days before the rain event or pre-storm history influences the losses due to surface depression storage, the time to wet the pavement surface and the permeability of unpaved surrounding areas. Gupta et al. (1981) considered dry days from the last storm event, and site characteristics to be determinant variables for a series of equations to predict runoff volumes based on rainfall data. This influence was not significant for the monitored events at Highway 17, except for the two late hours sampling events where the event of A p r i l 13 t h , 2002 had only a few dry days between rain events and had the greatest runoff coefficients, versus the October 3 r d , 2002 event that had 74 dry days between sampling events and where paved and unpaved surfaces were prone to retain more water. 4.1.4.2 Total Suspended Solids (TSS) Summary values of TSS concentrations for the 8-month monitoring period are presented in Table 4.7. Average TSS concentrations ranged from 37 to 53 mg/L with the greatest TSS variation during the spring and summer monitoring periods. Gupta et al. (1981) report similar TSS concentrations for an urban highway in Harrisburg, U . S . A with paved areas and runoff coefficients also in the same order as those reported for Highway 17. Onwumere (2000) reports greater TSS average concentrations ranging from 98 to 259 mg/L on similar B . C . highway sections, but with average daily traffic greater than 42,000 vehicles/day and areas fully paved. Additionally, the author reports coefficients of variation ranging from 0.33 to 0.50 for Burnaby and Richmond highways respectively. In summary, it could be said that TSS concentrations from Hwy 17 were consistent with other studies in that percent pervious area and daily traffic are major factors in the amount of sediment washed off the pavement. 105 Figure 4.10 shows how TSS concentrations tend to increase on both sides of the highway, as the paved area of study evolves, from a recently swept surface at the beginning of the monitoring period, to a surface where road dust and associated pollutants have accumulated throughout each 4-month observation period. This phenomenon was consistent for all locations for both observation periods. Significant differences at 95% confidence level were encountered between discreet event mean concentrations at the beginning and at the end of each monitoring period. Table 4.7. Summary of Total Suspended Solids concentrations for the 8-month monitoring period (March-N ovember, 2002) Monitoring period Tota Suspended Solids (mg/L) Location Weighted Average Maximum Minimum Coefficient of Variation Spring and Summer N B 37.4 57.0 19.0 0.6 SB 53.3 105.0 23.0 0.7 Fal l and Winter N B 46.5 115.5 27.5 0.4 SB 43.6 110.0 25.5 0.4 Sample after sweeping \ 0 4 , , 1 , , , , Apr13 Apr16 Apr26 Jun28 Oct3 Nov6 Nov18 Jan29 S a m p l i n g date Figure 4.10. Evolution o f TSS over a) First observation period: Mar-Jul, b) Second observation period: Jul-Nov, 2002. Highway sweeping was an effective maintenance practice to reduce TSS loadings in runoff processes at the H w y 17 study site. Significant differences were found between the suspended solids mean concentrations sampled on the last monitored event (Nov 18 t h) before sweeping took 106 place on Nov 29* , and the runoff sampled several days after sweeping (Jan 29 ). This difference was significant despite road dust accumulation for over two months, which showed that the suspended solids concentrations in highway runoff were highly associated with the amount of road dust accumulated on the pavement surface. 4.1.4.3 Tota l Me ta l Total metal concentrations in runoff at Highway 17 were highly correlated to TSS trends during the observation period, exhibiting coefficients of determination R 2 above 90% as shown, in this case, for the Apr i l 13 t h rain event discrete sampling (Figure 4.11). The results were consistent with findings reported by McCa l lum (1995) in a study of the impacts of soil use change in the Brunette River watershed. E c o "•4-* ro C Ci o c o o ra <-> o 0.2 -I 0.18 - Zn 0.16 - • Mn 0.14 - C u 0.12 - A P D 0.1 -0.08 -0.06 -0.04 -0.02 -0 -0 25 50 T S S (mg/L) 75 100 Figure 4.11. Association of TSS with different metals for the A pr i l 13 , 2002 discrete runoff sampling. Metal concentrations also showed a similar trend to Suspended Solids, i.e. runoff quality (as described by total metal concentrations) decreased as the accumulation of road dust on the pavement continued (Figure 4.12). Log-transformed data for all metal concentrations showed significant differences at 95% confidence level at the beginning and at the end of each stage of the monitoring period. Additionally, number of dry days between sampling events seem to have 107 an influence on runoff quality, since the events of November 6 and November 18 did not have dry days between sampling events and did not show statistical difference in metal concentrations at 95% confidence level. Apr13 Apr16 Apr26 Jun28 Oct3 Nov6 Nov18 Jan29 S a m p l i n g date Figure 4.12. Evolution of Event Mean Concentrations of metals in highway runoff over the 8-month monitoring period (Mar-Nov, 2002) Total metal concentrations followed the order Fe>Zn>Mn>Cu>Pb as illustrated in discrete sampling events shown for the metals Zn , M n , C u , and Pb on Figures 4.13 and 4.14. This order of metal concentrations were very similar to those found in road dust and dust deposition loadings on the right of way of Highway 17. In this case, however Pb concentrations were clearly lower than Cu , which reflects the strong attachment of lead to colloids, its precipitation under close to neutral pH environments and its low leachability from road dust particulates. A l l concentrations exceeded the Canadian Council of Ministers of the Environment ( C C M E , 2002) water quality parameters for the protection of freshwater aquatic life. In another relatively recent study on highway storm water runoff in Cincinnati, O H . Sansalone et al. (1996) found for a heavily transited H w y (150,000 vehicles) instrumented over a month, a similar order of Event Mean Concentrations of metals in five rainfall events, with Z n and C u exceeding surface quality discharge standards for all rainfall events and Pb having two exceedehces. 108 •A— Zinc Samp l ing t ime Figure 4.13. Total metal concentrations in highway runoff for the October 3 , 2002 event sampled during the July 1 l t h -November 29 t h observation period. •A— Zinc 5 ro -sr uo O r - CN ro LT> S a m p l i n g t ime Figure 4.14. Total metal concentrations in highway runoff for the November 18 t h , 2002 event during the July 1 l t h -November 29 t h observation period. 109 4.1.4.4 Dissolved and Resin-Exchangeable Metal Dissolved (filterable) Metal Dissolved concentrations in this section refer to metal concentrations on the 0.45 mm filtrate of runoff samples, which rather than being a true dissolved concentration, is an operationally defined speciation of metals in water ( A P H A , 1998); hence, in the remaining of this document "dissolved metal" is used as a synonym for the 45 pm "filterable" metal fraction in solution. Dissolved metal concentrations followed a similar order to that found in road dust, atmospheric dust deposition, and total metal in storm water runoff, i.e. Zn>Mn>Cu>Pb, as depicted in Figure 4.15, for the November 18 t h, 2002 discrete sampling event. However, looking at the relative percentages dissolved with respect to total metal concentrations, results show that Z n dissolved/total metal ratio is just slightly higher than Cu , followed by M n , Pb and Fe. The rank in dissolved metal concentrations found at this study site was also consistent with dissolved metal values reported by Sansalone et al. (1996) for a busy highway pavement section in Cincinnati, O H , and those summarized by Driscoll et al. (1990) for several U .S . hwys. Table 4.8 summarizes Event Mean dissolved and resin-exchangeable metal concentrations, and their ratios with other metal fractions. -A—Zinc Manganese - • — Copper - © — Lead -Q- -©-oo LO o c\i CM h— CO CN -0-c\i LO CN S a m p l i n g t ime Figure 4.15. Dissolved metal concentrations for a discreet sampling event on November 18 , 2002 at the Northbound location of Highway 17 110 Table 4.8. Summary of Event Mean Total, Dissolved and Chelex-Exchangeable metal concentrations in storm water runoff at Highway 17 Metal Total (ug/L) Dissolved (ug/L) Dissolved / Total (%) Chelex-Exchangeable (ug/L) Chelex / Dissolved (%) Chelex / Total (%) Copper 45 15 42 6 43 18 Iron 2679 52 2 4 8 0 Lead 18 1 4 0.3 41 2 Manganese 59 20 28 15 80 22 Zinc 148 80 45 53 63 29 Resin-Exchangeable Metal In this section, resin-exchangeable metals refer to the fraction of metals in the 0.45mm filtrate that exchange preferentially with a chelex-100 resin, providing information on the potentially free and weakly complexed metal species in solution (Figura and McDuffle, 1980; Greenberg and Kingston, 1983). Chelex-100 resin possesses carboxylic acid and amine functional groups, as well as a porous structure. Speciation by means of this chelating resin is based on the ability of the resin to remove trace metals by a process that imitates metal uptake at the cell surface (Florence and Batley, 1980). Bioavailability, expressed by resin-exchangeable fractions of metals in Highway 17 runoff, gave the order Zn>Mn>Cu>Fe>Pb (Table 4.8). Zinc had high levels in ionic and weakly complexed forms, while Cu showed intermediate and the low resin-exchangeable fraction of Pb and Fe indicated that they were largely in colloidal form. In a study of the bioavailability of trace metal contaminants in the Brunette River watershed in B . C . , Yuan (2000) found similar results, with resin-exchangeable/total metal ratios which followed the sequence: Zn(45%) > Mn(21%) > C u ( l l % ) > F e ( 3 % ) . Resin-exchangeable metal fractions were useful in this study to estimate the relative percentage of ionic or weakly complexed metal, which may be of greater environmental concern because of the toxicity and bioavailability of these metal species. These concentrations also depend on 111 rainfall specific factors such as: rainfall intensity and duration, ionic strength and water p H (Sansalone, 1996). However, another important factor is the background geology and the resulting lack or abundance of clay minerals in sediments, which when transported or deposited as fugitive dust may regulate the partitioning of metals in stormwater during runoff events. 4.1.5 Roadside Soil 4.1.5.1 Surface and Subsurface Soil Samples Detailed Surface Soil Samples Surface soil samples provided a detailed description of recent deposition events, while subsurface metal concentrations provided the long-term effects of metal accumulation and evidence of potential mobility. Surface soil results also helped identify the extent and amount of metal removal from the highway onto the surface of roadside soils under optimum dispersion conditions, due to the type of highway section without obstructions and the dominant wind directions that were generally across the road. Metals concentrations on surface soil samples of the northbound and southbound right-of-way are presented in Figures 4.16 and 4.17, respectively. 300 -, * - M n * - Z n ^ - P b • • - C u 250^ 0 0 2 4 .6 Distance f rom Highway (m) 8 10 Figure 4.16. Northbound metals concentrations on surface roadside soil at Hwy 17 Surface roadside soils also showed, although more scattered, exponential decrease of metal concentrations with distance from the edge of the highway. Previous research has shown that 112 metal levels decrease rapidly with distance from the road (Motto et al., 1970; Kobriger and Geinopolos, 1984; Little et al., 1994; Hafen and Brinkmann, 1996). Most of the literature has dealt with Pb pollution in roadside soils, but in another study Brault el al. (1994) found that exchangeable M n was significantly higher in an organic soil located near a busy road than at other sites. Correlations among all metals were significant with Cu and Z n showing the highest level of association (R 2 = 0.84). These metal associations agreed with Geo-Environmental site assessments performed by B . C . M o T H personnel indicating similar relationships among different metals, particularly among Pb, C u and Z n in contaminated soils throughout the B . C . Lower Mainland ( M o T H 2002). Fagotto et al. (2000) also found high associations among those metals and estimated high risk of Z n mobility, and moderate risk for C u and Pb under acidifying conditions in roadside soils from a rural area in France. 350 n 0 4 , , , , , _, : , 0 2 4 6 8 10 12 Distance from Highway (m) Figure 4.17. Southbound metals concentrations on surface roadside soil at Hwy 17 Metal concentrations on southbound (SB) right of way were higher. These results reflected the influence of dominant wind direction, which for this site is to the west and possibly a minor influence of traffic volume, which is also slightly higher for the SB direction. Although no major modifications have been made to the site, the scatter of metal concentrations close to the road for the SB area might be due to road maintenance activities, as sediment and vegetation is routinely 113 scraped from the shoulder to facilitate surface drainage. N o such activity was observed on the northbound N B right of way since little vegetation grew on the coarse soil next to the shoulder. Subsurface Soil Samples Profiles of total metal concentrations, pH's and some physico-chemical properties up to 30 cm deep are presented in Figures 4.18 through 4.21 for the S B and N B location. Soils tended to be more acidic as distance increased from the edge of the road. Soil pits 1NB and 1SB showed pH's neutral or basic while locations 2 and 3 for SB showed pH's below 7 and as low as 5. This may be expected as the layer of soil fill material decreases with distance from the edge of the road and natural delta deposits appear. Most likely these deposits have leached base forming cations leaving the colloidal complex dominated by H + and A l 3 + cations (Sparks, 1995). Soils were predominantly gravelly sands to silty sands, as expected in roadside environments where inert non-plastic soils are used for road construction. Organic carbon profiles showed a smooth decrease with depth, consistent with the presence of fibrous grass roots throughout the soil profile. The surface area vertical distribution showed a similar trend to either organic carbon or total iron profiles, highlighting its relationship with the presence of organic matter, clay minerals, and iron and manganese oxide coatings. Since roadside soils were mainly composed of sand, gravel and some silt, no significant presence of clay minerals was anticipated. X-ray diffraction analyses seemed to confirm this assumption (Appendix F). Additionally these analyses revealed the presence of silicate minerals in the feldspars group such as albite and quartz, and some phyllosilicates such as muscovite and chlorite. Therefore, it was expected that most metal-soil interactions would take place among organic matter reactive functional groups and amorphous iron and manganese oxides. A representative X-ray diffraction pattern from Hwy 17 roadside soil is shown in Figure 4.22. Total metal profiles were slightly higher for the SB location, which is in agreement with greater runoff and atmospheric loadings observed during the monitoring period on the SB right of way. Highest metals concentrations were located at the edge of the highway where coarser soils are present. Lead was the anthropogenic metal that showed the highest concentration in roadside soils, which reflects past accumulations of this metal in soils from past automobile exhaust 114 emissions. The peak Pb concentration was found at 25 cm for 1NB and beyond 30 cm for the 1 SB location (both at the edge of the Highway), which indicates some degree of Pb mobility in coarse permeable soils. This mobility may be in the form of leached metal from the ground surface or sediment bound metal that percolates through the porous gravelly soil matrix. The total Lead profiles at the 1NB and 1SB locations resembled a pulse release of Pb that was subsequently diffused downwards by rainwater and runoff infiltration. This Pb pulse was consistent with a phenomenon of constant Pb input, (which occurred for many years as alkyl lead fuel additive in the past) and a subsequent rinse of soil with water that had no Pb present (corresponding to the phase out of Pb in gasoline). Maintenance activities such as sod removal from the edge of the highway may have contributed to the absence of Pb on the roadside surface soil, however this activity is generally constraint to the top 5 cm of soil or grass. Therefore, sodding alone did not explain the low surface Pb concentrations up to 20 cm deep and the sudden peak in lead from 20 to 30 cm, but rather a piston flow displacement of particulate-bound Pb. Other persistent metals such as Cu and Z n showed multiple peaks, which might be a result of a continuous input from the highway through runoff and atmospheric processes and some degree of mobility or relocation by soil invertebrates. In most locations Cu and Z n showed similar shape of vertical concentration profiles, which may be related to their similar ionic potentials and tendency to form aqueous complexes, as well as their affinity to organic matter (Langmuir, 1997). Significant statistical differences were observed between metal concentration profiles at the highway shoulder and at the background location. Metal concentrations in vertical profiles showed greater degree of variance in the right of way than in the background location. Even at the right of way boundary (12 m) vertical profiles of metals C u and Pb still showed significant differences at 90% level of confidence with respect to the background concentration. The background soil was composed of the same gravelly-sand material used for highway construction, but it was located 150 m away from Hwy 17. Therefore, metal differences should have originated from traffic activities rather than differences in soil composition. 115 %soil < 4.75 mm pH Organic C Surface Area (m2/g) Fe (mg/kg) Pb (mg/kg) Total Metal (mg/kg) Figure 4.18. Vertical profiles showing soil characteristics and metals concentrations with depth at a) 1NB (0 m) and b) 1 SB (0 m) location. Figure 4.19. Vertical profiles showing soil characteristics and metals concentrations with depth at a) 2 N B (5 m) and b) 2 SB (6 m) location. %soil<4.75mm pH Organic C{°/o) Surface Area (m2/g) Fe (mg/kg) Pb&Cu (mg/kg) Mn & Zn (mg/kg) o Pb 40 60 80 5 5.5 6 1.5 2.5 3.5 35 45 55 65 16000 24000 0 50 0 100 200 300 %soil<4.75mm pH Fe (mg/kg) Pb&Cu (mg/kg) Mn&Zn (mg/kg) 40 60 80 5.5 6 12000 22000 0 50 100 0 200 400 Figure 4.20. Vertical profiles showing soil characteristics and metals concentrations with depth at a) 3 N B (10 m) and b) 3 SB (12 m) location. Figure 4.21. Vertical profiles showing soil characteristics and metals concentrations with depth at the Hwy 17 background location. 5000 -} 4000 1 3000 A 2-Thela - Scale f\t|C:\Diffoat1\>f!D User\Hwy17-1NB-0-5.F!AW -File: Hwy17-1 NB-0-5.RAW - Type: 2Th/Th locked - Start 3.000 ° - End: 70.000 ° - Step: 0.040° - Steptime: 2.s - Temp.: 25°C( I Operations: Import • 78-1252 (C) - Quartz apha.syn - Si02 -Y : 93.75% - d xby: 1. -WL: 1.54056-Hexagonal-l/ lcPDF 3.2-• 09-0466 ( * ) - A I M e, ordered-NaAIS308-Y: 37.50 % - d x b y : 1.-WL: 1.54056-Trie inic- l / lc PDF2.1 -J. 75-0943 (C) - Muscovite - KAI3Si3O10(OH)2 - Y: 17.97% - d xby: 1. - W L 1.54056 - Monoclinic-l/lc PDF 2.9-72-1234 (C) - Chlorite - Mg2.5Fe1.65AI1.5S22AI1.8O10(OH)8 - Y: 4 .17% - d xby: 1. - W L : 1.54056 - Moncc Inic- l / lc PDF 1.2 -x 42-1369 (l)-Magnesicartvedsorite- Na3(Mg,Fe)5S8022(OH)2 - Y : 6 . 2 5 % - d x b y : 1.-WL: 1.54056 - Monoclinic-Figure 4.22. X-ray Diffractogram for soil sample (depth 0-5 cm) collected at the 1NB location (0 m from the edge of the Hwy). to o 4.1.5.2 M e t a l part i t ioning in Soil Samples The same metal partitioning procedure applied to roaddust and atmospheric dust particulates, was applied to roadside soils with a few modifications. The carbonate extraction was combined with the oxide extraction and reported as such. This change was based on the fact that the procedure was originally designed for more arid or carbonaceous soils were the carbonate metal phase has always been present. In the granitic and acidic soils sampled at Hwy 17, however, the carbonates were very low or non-existent. Therefore, most metals extracted with 1 M N a O A C and presumed to be in the carbonate phase were actually most likely associated with the oxide phase. Lead total concentrations and sequential extraction results for the northbound side of Hwy 17 are presented in Figure 4.23. Partitioning results expressed on a percentage basis show the effectiveness of the sequential extraction procedure in identifying possible anthropogenic inputs of metal. The procedure of sequential extraction of metals assumes that the residual fraction corresponds to metal that is tightly bound to the crystal lattice of primary or secondary minerals, and that it is environmentally unavailable (Tessier, 1979). Pb (mg/kg) Pb (°Q 0 20 40 60 80 100 • Exchangeable • Oxides • Organic • Residual Figure 4.23. Vertical distribution of Pb and its relative partitioning in the soil profile at the 1NB soil pit. In the 1NB soil location, where the accumulation of metal is more pronounced, Figure 4.23 shows that the relative percentage of residual lead decreases where the accumulation of 121 suspected anthropogenic lead increases. Additionally, the amount of exchangeable Pb increased significantly for the 25-30 cm depth, where the greatest Pb accumulation was found. On the assumption that residual lead or tightly bound lead corresponds to lead that has been deposited through geological rather than anthropogenic processes, changes in residual lead should not be as pronounced, except for a change in the geological features across the vertical soil strata. Similar abrupt changes in the residual fraction were observed for metals Zn and M n , where an anthropogenic input of these metals was suspected, as displayed in Figures 4.24 and 4.25. Zn (mg/kg) Zn • Exchangeable • Oxides • Organic • Residual Figure 4.24. Vertical distribution of Z n and its relative partitioning in the soil profile at the 1NB soil pit. As explained earlier, Z n and M n show multiple peaks, which may be a result of their mobility and relocation by soil invertebrates, but again the percentage of the more extractable metal fractions increases where anthropogenic input of metals is suspected. In the case of Pb, this increased extraction of the exchangeable, oxide and organic fractions at depths of 20-30 cm, may have been due to past accumulations of the sulfate form of Pb compounds. These Pb compounds were present in past automobile emissions and tended to have greater solubility in chloride solutions (Lum et al., 1982). Another factor that could have led to an increase in Pb extraction at these depths is the finer nature of anthropogenic atmospheric particulates, which are generated by high temperature combustion processes (Paciga and Jervis, 1976), and are later deposited on roadside soil. 122 A similar retention mechanism given for Pb may apply to M n accumulation, since M n is now used as an antiknock substitute for A l k y l lead additives. A report by Health Canada estimated that the main form of M n in automobile exhaust was inorganic M n in the form of airborne particles of Mn304, with traces of manganese sesquioxide Mn203 (Health Canada, 1978). However, more recently in a car exhaust study Zayed et al. (1999) found that the frequency of M n oxide was only 2% as opposed to 8% for M n phosphate, 16% for M n sufate and 54% for a mixture of M n phosphate and sulfate. In another study by the manufacturer of M M T , Lynam et al. (1999) reported similar results with respect to the M n species present in automobile exhausts. Manganese oxides, and M n phosphates have very low or null water solubilities, whereas M n oxysulfates and M n sulfates have medium to high solubilities. Therefore, it is plausible that some M M T exhaust products are deposited and then leached through the soil profile, and that this process results in a greater amount of extractable M n from roadside soil at particular depths. Figure 4.25 depicts total M n concentrations and its respective partitioning through sequential extractions. A significant increase of M n in the more labile metal phases is apparent between 15 and 25 cm, which seemed to match Pb maximum accumulation in the same soil profile. Mn (mg/kg) Q. 8 M n ("/<) • Exchangeable • Oxides • Organic • Residual Figure 4.25. Vertical distribution of M n and its relative partitioning in the soil profile at the 1NB soil pit. However, M n is one of the more abundant elements in the lithosphere. M n is likely to occur in soils as oxides and hydroxides forming coatings in soil particles or nodules of varying diameter. 123 These M n concretions are reported to accumulate Fe and several trace elements (Kabata-Pendias and Pendias, 1991). Manganese is also used to regulate metal ductility and hence is ubiquitous in almost every metal part of automobiles ( F H W A , 1998). Therefore, anthropogenic accumulation of M n in roadside soils may not be exclusively related to M M T , but also due to traffic activity and the wearing of automobile metal parts in general. In the case of Copper (Figure 4.26), sequential extractions in soils showed a significant affinity from this metal to organic matter, consistent with Cu partitioning in road dust and dustfall samples. Harrison et al. (1981) have reported similar associations, which result from the formation of stable organic complexes predominantly with humic substances (Phillips and Chappie, 1995). However, the presence of anthropogenic C u based on a pronounced increase in leachable versus residual Cu , was not as clear as the extractions of metals Pb, M n and Z n (Figures 4.23-4.25). Cu (mg/kg) 0 20 40 60 80 100 • Exchangeable • Oxides • Organic • Residual Figure 4.26. Vertical distribution of C u and its relative partitioning in the soil profile at the 1NB soil pit. This lack of a clear anthropogenic Cu identification may be related to time effects on metal redistribution among different soil phases. McLaren and Ritchie (1993) reported on the long-term fate of Cu in a lateritic sandy soil through sequential extractions. The authors found that much of the C u was initially associated with the organic matter fraction. However, after approximately 20 years, a significant amount of this organically bound Cu redistributed to the 124 residual soil fraction. According to the authors, this redistribution of metal resulted in a decrease in available C u for plants. For the soils profiles thus far presented, it appears that the deeper metal peaks refer to older metal deposition, as illustrated by Pb accumulation between 15 and 25 cm. Therefore, it is possible that for Cu , some of this older metal has already redistributed to the residual soil fraction and that it results in a decrease in organically bound Cu. This effect would mask the identification of anthropogenic C u in the deeper depths of the soil profile. The anthropogenic C u accumulation would be clearer for the more recent Cu deposition, which is illustrated by the greater organically bound Cu fraction in the upper soil horizons. Additional metal partitioning profiles for the rest of the sampled locations may be found in Appendix E . A one-way analysis of variance ( A N O V A ) was done on the percentages of metal in the solution + exchangeable fractions, from the sequential extraction procedure (P<0.0001). Assuming that the percentages of metal in the solution and exchangeable fractions are an indicator of metal mobility, then metal mobility in H w y 17 soils followed the order: Zn>Cu>Mn >Pb. The only exception to this order was present at peak Pb concentrations in the soil profile, where a significant anthropogenic Pb accumulation was suspected and where exchangeable Pb reached up to 7% of the total Pb concentration (600 mg/Kg). These results are consistent with previous results reported by other researches in soils from other geographical locations (Phillips and Chappie, 1995; Pagoto et al., 2000), which may indicate a more general trend in metal accumulation and mobility in roadside environments. Results from soil sequential extraction analyses were compared with those obtained at the validation site. Additionally, hypotheses regarding the possible anthropogenic source of metals in the soil profile were tested and verified with Pb isotopic analyses. 4.1.6 Mass Balance of Solids and Associated Metals The information in this section was aimed at establishing the accumulation of solids and associated metals on the highway surface due to pavement wear and vehicle sources (such as: brake wear, fuel exhaust, tire wear, brakes, body rust, etc.), evaluate the relative contribution of 125 atmospheric and runoff contaminant removal processes on the highway right of way and finally, evaluate the effectiveness of highway sweeping as a best management practice. Mass balance of Solids The mass calculations involved two main mechanisms: 1) atmospheric and vehicle related deposition; and 2) removal processes, which included atmospheric and runoff removal as well as highway sweeping. Total Solids of particle size < 250 pm (TS250) were recorded for the March-July and the July-November (2002) monitoring periods. Results from mass balance calculations for both monitoring periods are presented in Figures 4.27 and 4.28. Further mass balance calculations for all metals in both monitoring periods and graphs are presented in Appendix D of this report. E CO 1600 j 1200 - -800 400 0 -400 -800 -1200 - L Atmospheric input Runoff Vehicle & Pavement input + Atmospheric 1 removal Sweeping Figure 4.27. Mass balance distribution of solids < 250 pm for the March-July, 2002 monitoring period at Hwy 17 (columns represent mean quantities and whiskers represent 95% upper and lower confidence limits of the mean) Both monitoring periods showed the small relative contribution of atmospheric input in the area, mostly caused by regional agricultural activities, when compared to the amount of solids generated by vehicles activity and pavement wear. Kobriger and Geinopolos (1984) also reported vehicle generated deposition as the main source of solids and associated pollutant on U.S . highways. The U S E P A (1977) in a previous study reports that approximately 37% of vehicle related deposition was attributed to tire wear, 37 to pavement wear, 18.5% to brake and engine component wear and 7.5% to deposited exhaust. 126 Figures 4.27 and 4.28 highlight the importance of atmospheric export processes in removing solids from the pavement and further depositing them in neighboring areas. These results contrast with results reported earlier by Kobriger and Geinopolos (1984), where runoff had been identified as the main removal mechanism of TS250- This contrast may be due to different sampling efficiency in atmospheric deposition devices used in the previous studies and that used for this research. The use of the more aerodynamic Frisbee deposition gauge made a difference in the contribution allocated to different solids removal processes. Mass balance of solids showed that highway sweeping accounted for up to 50% of the total removal mass for the March-July, 2002 monitoring period. Hence, H w y sweeping could be considered an effective maintenance practice in reducing the amount of solids available for runoff and atmospheric removal processes. However, the effectiveness of this maintenance practice was greater for the drier March-July period than for the wetter July-November monitoring period, as displayed in Figure 4.28. E o IT) CM 1200 T 800 + 400 + 0 -400 + -800 ->-Atmospheric input + Runoff Vehicle & Pavement input Sweeping Atmospheric removal Figure 4.28. Mass balance distribution of solids < 250 pm for the July-November, 2002 monitoring period at Hwy 17 (columns represent mean quantities and whiskers represent 95% upper and lower confidence limits of the mean) A contrasting result to the first monitoring period was the increase in the relative amount of atmospheric removal for the second "wetter" monitoring period. Greater precipitation would 127 mean that the dominant removal of solids should be due to runoff rather than atmospheric processes. This apparent contradiction was due to a greater amount of precipitation during the second period, but at the same time a greater number of dry days. This resulted in solids being removed constantly by saltation and blown-off processes. It also resulted in a great amount of solids being removed by the isolated stronger rain events, and a smaller amount of solids left on the pavement surface for highway sweeping maintenance activities (Figure 4.28). Mass balance of Solids Associated Metals In contrast to TS250 mass distribution, metals showed a very different mass allocation. For all metals monitored in this study, except for Fe, runoff was the main removal mechanism, reflecting the strong association of metals with the finer runoff suspended solids. Figure 4.29 shows the mass allocation for Zn , which was similar to other metals. Other metal mass balance calculations and graphs for the rest of the metals recorded in this study are presented in Appendix D . 1200 800 ? 400 -400 -800 Atmospheric input Runoff Vehicle & Pavement input | H W | M 1 Atmospheric [ removal Sweeping Figure 4.29. Mass balance distribution of Zinc for the March-July, 2002 monitoring period at Hwy 17 (columns represent mean quantities and whiskers represent 95% upper and lower confidence limits of the mean) This contrast from TS250 observations may also have been the result of atmospheric deposition gauges not only collecting vehicle and pavement generated particulates, but also collecting particulate matter that did not contain a substantial amount of metal (such as background dust, fine organic debris, etc.). It would seem that these particulates did not influence metal mass 128 removal, but they did influence solids mass measurements next to the highway. Metal mass balance calculations reflected a similar metal order observed for road dust, atmospheric and runoff samples (i.e. Fe>Zn>Mn>Cu>Pb). A higher efficiency in iron removal by sweeping was observed for the first monitoring period (Figure 4.30). This was probably due to the background, crustal nature of this metal at the monitored site, where sweeping was capable of removing the coarser Fe-rich sediments. On the other hand, sweeping removal efficiency diminished for those metals that tend to exhibit a more soluble nature such as Zn. Sweeping metal removal efficiency followed the order: Fe>Mn>Pb>Cu>Zn. 50000 j 40000 - -30000 ---g- 20000 I) 10000 0 -10000 -20000 -30000 J -Atmospheric input Runoff Vehicle & Pavement input Atmospheric removal Sweeping Figure 4.30. Mass balance distribution of Iron for the March-July, 2002 monitoring period at H w y 17 (columns represent mean quantities and whiskers represent 95% upper and lower confidence limits of the mean) The metal mass balance calculations showed runoff as the main pollutant removal mechanism, and this highlights the importance of mitigation practices that target runoff export processes from highways to urban watersheds such as: infiltration trenches, porous pavements, detention and/or retention basins, vegetated swales, wetlands, etc. 129 4.2 Trans-Canada Highway 4.2.1 Meteorological Conditions Meteorological conditions at Trans-Canada H w y during the period of April-September of 2003 are summarized in Table 4.9. Temperature ranged from a minimum of 1.9°C in A p r i l to a maximum of 33.5°C in June. The driest monitored month was August and the wettest A p r i l , which was more than double the maximum amount of rain registered at H w y 17. The differences in meteorological characteristics with the H w y 17 site can be summarized as: 1) warmer temperatures, 2) greater total precipitation, 3) lower average wind speeds, and 4) easterly wind directions. Greater precipitation could be attributed to the geographical proximity of the study site to the Coast Mountains, specifically Burke and Coquitlam Mountains. The proximity of these mountains produces an increase in relative humidity as air is lifted, causing moisture condensation and in some instances precipitation. Lower average wind speeds were also the result of topography, the influence of surrounding forested areas and to some extend of atmospheric stability during the monitored period. Table 4.9. Summary of meteorological conditions at Highway 1 during the monitoring period (April-September of 2003) Month Temperature Total Average Maximum Dominant Range Precipitation Wind Speed Wind Speed Wind direction (°Q (mm) (m/s) (m/s) A p r i l 20.7, 1.9 120.8 0.8 11.2 N E M a y 23.4, 3.9 62.2 0.7 9.8 E June 33.5,8.7 27.4 0.9 7.6 SE July 33.2, 9.7 10.6 0.9 10.3 SE August 30.4, 9.4 5.8 0.9 8.0 SE September 31.5,9.1 48.2 0.8 7.6 E 4.2.2 Road Dust Road dust collection was accomplished using the industrial vacuum cleaner. A total length of 450 m over two lanes 3.5 m wide were necessary to collect the minimum 200 g of sample specified in the "Procedures For Sampling Surface/Bulk Dust Loading" ( U S E P A , 1997). Three road dust samples were collected, one before the start of the monitoring period in February, one in A p r i l and a last one in mid June of 2003. The samples were sieved and the fraction smaller 130 than 75 pm recorded. The fraction smaller than 75 pm is designated in the U S E P A AP-42 document as "silt loading" and it is routinely used for fugitive dust emission estimations from paved roads. Silt loadings for the A p r i l and June sampling events were 0.02 and 0.008 g/m 2 respectively. The A i r Quality and Assessment Division of the Greater Vancouver Regional District ( G V R D ) has measured silt loadings throughout the Lower Fraser Valley, as part of a regional effort to update their particulate emissions inventory ( G V R D , 2000). This Division provided the researcher with a summary of silt loadings for arterial, residential roads and highways for all seasons (Table 4.10) [Kelly Der, personal communication]. The silt loadings trends can be summarized as: residential > arterial > highway. It is also important to note that silt loadings collected for this research In Hwy 17 (0.17 g/m 2 and 0.053 g/m 2) and in Trans-Canada H w y & 176 t h St. (0.02 and 0.008 g/m 2) were within the range of those collected by the G V R D for similar highways in the Lower Mainland (Table 4.10). Table 4.10. Summary of Measured Silt Loadings at Residential Roads, Arterial Roads, and Highways in the Canadian Lower Fraser Valley (g/m2) City/Municipality Street Location Street Type Spr ing 2000 S u m m e r 2000 A u t u m n 2000 Win te r 2000 Spr ing 2001 S u m m e r 2001 Vancouver 4641 Knight St Cambie Street near 55th Ave 1800 West Georgia Street 1100 West King Edward 46th Ave near Victoria Arterial Arterial Arterial Arterial Residential NA 0.050 NA 0.034 NA 0.093 0.143 0.078 0.028 1.940 0.197 0.095 0.082 0.087 1.192 0.099 0.122 0.067 0.041 1.631 0.073 0.077 0.068 0.018 0.687 0.117 0.192 0.024 0.026 . 1.657 Richmond 5660 Francis Road 10251 No.1 Road Arterial Arterial 0.042 0.059 0.098 0.056 0.066 0.039 0.087 0.084 0.064 0.048 0.016 0.042 West Vancouver 2100 West Marine Drive Arterial 0.047 0.054 0.043 0.052 0.063 0.042 Burnaby 5700 Royal Oak 1050 BoundryRoad Arterial Arterial 0.134 0.051 0.131 0.131 0.079 0.042 0.077 0.061 0.093 0.054 0.294 0.073 Delta On Highway 99 south of Highway 17 Highway 0.103 0.072 0.067 0.088 0.044 0.023 Surrey On TransCanada Hwy near 190 St 176 Street south of 24th Avenue On 88th Ave near 127th St Highway Arterial Arterial 0.021 0.049 0.063 0.011 0.066 0.055 0.034 0.110 0.059 0.053 0.180 0.102 0.007 0.061 0.007 0.065 0.065 Port Coquitlam 1128 Terra Ct Residential 0.707 0.785 0.232 0.695 0.915 0.936 Coquitlam 500 Lougheed Hwy Arterial 0.093 0.112 0.143 NA 0.290 0.038 Maple Ridge On Highway 7 near 105th Ave Highway 0.116 0.118 0.070 0.171 0.073 0.270 Langley Township On 93a Ave near 210 St Residential 0.218 0.130 0.489 0.065 0.183 0.191 Abbotsford 33330 Marshall Road Arterial 0.084 0.028 0.047 0.067 0.055 0.030 Mission 32752 7th Ave Arterial 0.074 0.066 0.065 0.096 0.088 0.070 Chilliwack Evans Road Arterial 0.104 0.068 0.070 0.112 0.142 0.082 131 These results were consistent with those by B a l l et al. (1996), Birch and Scollen (2003), where deposition and removal of road dust can be considered as a continuous process and where sediment accumulation on the road surface attains a maximum when equilibrium is reached. Kantamaneni et al. (1996) also report consistent silt loadings with overall standard deviation of less than 10% for some of the neighbor Washington state highways. 4.2.2.1 Metal Concentrations in Road Dust Average total metal concentrations in road dust samples at T C H followed the order: Fe>Zn > Mn>Cu > Pb (Table 4.11). This concentration order, overall, was consistent with those reported earlier for H w y 17 and with the general trend of road dust concentrations in the Lower Mainland (section 4.1.2.1, Figure 4.2). N o statistical difference at 95% confidence level was found for road dust metal concentrations between the two study sites, except for M n (PO.01) . Manganese showed significantly higher concentrations in road dust at the busier T C H site than those recorded at Hwy 17. Furthermore, M n concentrations were significantly higher in dustfall and runoff samples at the T C H study site than at the Hwy 17 site. These differences in M n concentrations between the two study sites were consistent with the strong association reported between M n and vehicle activity for the Brunette river watershed in Burnaby, B C . (Hall et al. 1998). Additionally, they reported that the increase in M n concentrations in the watershed sediments correspond in time, to the introduction of M M T as a replacement for tetra-ethyl lead. Table 4.11. Total metal concentrations in road dust sweepings at T C H (2003) Metal (mg/Kg) February April June Copper 77 80 108 Iron 33,300 31,700 31,900 Lead 64 55 58 Manganese 415 402 427 Zinc 493 318 491 132 4.2.2.2 Me ta l Part i t ioning Road dust samples from the T C H study site exhibited very similar metal partitioning pattern to those collected at Hwy 17 (Figure 4.31). A l l samples exhibited low percentages of exchangeable metal, which may be due to the constant removal of finer sediment particulates by atmospheric and runoff processes from the pavement surface. Figure 4.31. Metal partitioning in road dust samples from the T C H study site [Samples # correspond to: 1) Feb/03, 2) Apr/03, 3) Jun/03]. 4.2.3 Atmospheric Dust Particulates Background Dustfall Dustfall was measured at a background site located approximately 500 m northeast of the study site. The background sampling location was relatively flat, with no major obstacles or buildings surrounding the dust gauge, which could create significant turbulence to influence deposition measurements. Background deposition measurements at T C H were greater than those recorded for the background site at Hwy 17 reflecting a greater influence of anthropogenic dust generating activities in the area (Figure 4.31). Additionally, deposition increased during the months of July through September, reflecting the drier conditions of those months, and perhaps some influence from the construction of a commercial development in the neighborhood. 133 120 -, Apr May Jun Jul Aug Sep Monitoring Period Figure 4.32. Atmospheric deposition at the Trans-Canada Hwy background location in Surrey, B . C . , during the monitoring period (Apr-Sep, 2003). Roadside Dustfall Dust particulate loadings on the right of way of T C H followed an exponential pattern consistent with those reported for Hwy 17 (Figure 4.32). However deposition curves for the monitored months at T C H were not as scattered as those of Hwy 17, perhaps due to greater atmospheric stability reported for this site and lower road dust loadings on the pavement surface. According to the U S E P A , greater amounts of silt loadings and deposition are expected from roads with lower Average Daily Traffic ( A D T ) versus high A D T roads ( U S E P A , 1997). Dust in high A D T roads is continuously ground and re-suspended, hence finer dust particulates are generated that can be transported to distances beyond the immediate vicinity of the right of way. 0 5 10 15 20 Distance (m) - « — A p r - 2 3 — • — M a y - 2 8 Jul-02 Aug-07 - * - 1 7 - S e p Figure 4.33. Dust deposition pattern on the right of way of T C H throughout the monitoring period (Apr-Sep, 2003) 134 Dustfall levels at the closest sampling location from the highway (3.5 m) came close (172 2 2 mg/m -day), but did not exceed the Desirable Dustfall B . C . A i r Quality Objectives (175 mg/m -day). Therefore, considering the trend of the deposition curves, these air quality objectives were most likely exceeded for a strip area < 3.5 m wide along the T C H sampling location. 4.2.3.2 Metals in Atmospheric Deposition Dust deposition metal concentrations on the T C H right of way showed the order: Fe>Zn>Cu>Mn>Pb (Figure 4.34). This metal order was similar to that registered at H w y 17, except for a clear significant difference observed between C u and M n concentrations. This contrasted with the order of metals found in road dust were concentrations clearly followed the order: Fe>Zn>Mn>Cu>Pb. The difference in order highlighted the greater association of C u to finer sediments, stressed earlier in the discussion for metals in atmospheric deposition at H w y 17. £ 6 0 . 0 -, N g 5 0 . 0 - J 3; 4 0 . 0 -| c .2 3 0 . 0 in § . 2 0 . 0 CO - 10 .0 ra .*-» 0.0 10 Dis tance (m) — i — 15 - £ - - Z n -OK-- - Mn - - e — 2 0 Figure 4.34. Metal deposition loadings at different distances from the edge of Trans-Canada H w y for the sampling period from M a y 28 t h - Jul 2 n d , 2003. Despite lower dustfall measurements at the T C H right of way, when compared to the H w y 17 site, metal deposition loadings were still much higher at the T C H site at equal distances from the edge of the Hwy, reflecting the influence of a greater traffic activity at this site. Mean metal 135 concentrations from dustfall measurements at T C H were approximately 4 times higher for Cu , 2.5 for Pb, 1.5 for M n , and 2 times higher for Zn , than at the less busy Hwy 17 site. Similar to results from H w y 17 study site, particle size distribution curves of deposited material on the right of way showed distinct differences with finer material dominating as the distance from the highway increased (Figure 4.35). 1 10 100 1000 Part ic le S ize (pm) Figure 4.35. Particle size distribution of atmospheric particulates deposited at different distances on the Trans-Canada Highway right of way. However, a significant association between particle size and metal concentration could only be observed for Pb. There was a significant difference at 95% level (P = 0.04) between Pb concentrations in dust particulates collected at 3.5 m and 15 m, highlighting its association with fine, combustion generated atmospheric particulates. Summary statistics for metal concentrations in dust deposition are shown in Figure 4.36. 136 3.5 7.0 15.0 Copper i 1 r 3.5 7.0 15.0 Lead T r 3.5 7.0 15.0 Manganese 3.5 7.0 15.0 Zinc Figure 4.36. Summary statistics for metal concentrations in dust deposition at different distances from the edge of Trans-Canada Hwy. (Note: Circles represent mean concentrations, boxes are 90% confidence intervals boxes are 90% confidence intervals mean, bars across the boxes are medians, and whiskers are minimum and maximum values). 137 4.2.3.3 Metals Part i t ioning in Atmospheric Deposition Metal partitioning of dust deposition samples at T C H was similar to Hwy 17 samples, except for a significant increase in the fraction of exchangeable metal (Figure 4.37). This result is believed to be related to: 1) the high traffic characteristic of T C H , and hence a significant increase in anthropogenic metal loadings that can have greater leachability, and 2) the difference in particle size of dustfall collected at both sites (Figure 4.38). 3.5 m 7.0 m 15 m 3.5 m 7.0 m 15 m l o c a t i o n l o c a t i o n • Ex • Car & Oxi • Org & Res E3 Total Ex • Car & Oxi • Org & Res • Total 7.0 m 15 m 3.5 m 7.0 m 15m l o c a t i o n l o c a t i o n Ex • Car & Oxi • Org & Res • Total Ex • Car & Oxi • Org & Res • Total Figure 4.37. Metal speciation in terms of dust deposition rates for metals: a) lead, b) copper, c) Manganese, d) Zinc 138 Figure 4.38 shows that practically all samples collected at Hwy 17 were coarser than those from T C H , which to some extend stressed the finer anthropogenic nature of vehicle related particulates collected at the T C H site. This speciation differences from both sites also confirmed the possibility of speciation masking effects from sand sized saltating soil particles collected at Hwy 17, and mentioned earlier (section 4.1.3.2). This particle size effect was further investigated in the Total Suspended Particulate section (section 4.2.4). 1 10 100 1000 Particle Size (pm) Figure 4.38. Comparison of H w y 17 and T C H Particle Size Distributions from dust deposition samples collected at different distances from the road. Table 4.12 summarizes total metal concentrations in dust deposition samples and the relative partitioning of these loadings in: exchangeable, carbonates and oxides, and organic and residual phases. A s mentioned earlier, total metal concentrations in dustfall were significantly higher for the busier T C H site, as well as the exchangeable metal fraction for metals: Cu , Pb and Zn. 139 Table 4.12. Summary of Speciation of metals in dustfall samples at T C H during the moniroting period (Apr-Sep, 2003) Mean Mean Carbonate Organic Total Total & & Cone. Loading Exchangeable Oxides Residual Element (ug/g) (pg/m2-day) % % % C u 223 ± 5 1 2 6 ± 8 13 59 28 Fe 8 7 5 4 ± 1016 993 ± 2 6 2 0.2 49.1 50.7 Pb 88+ 12 1 0 ± 2 15 69 16 M n 110+ 16 1 3 ± 4 2 54 43 Z n 346 ± 7 2 3 8 ± 10 16 52 32 • Note: Mean total metal concentrations are based on 15 dust deposition samples. Speciation percentages are based on the mean of three dust deposition samples. 4.2.4 Total Suspended Particulates 4.2.4.1 Suspended Particulate Loading Suspended particulates were collected to provide insight into the relationship between particle size and possible bioavailability, as well as provide information for a methodology to estimate metal loadings in roadside soils and neighboring areas due to atmospheric and runoff processes. Results from the Total Suspended Particulate sampling program at T C H are summarized in Figure 4.39. TSP measurements were overall consistent with an atmospheric dispersion and deposition process showing exponential trend lines with R 2 > 0.90. Two groups of curves can be identified in Figure 4.24: a) an upper group of curves corresponding to the sampling events of A p r i l 29 and June 26/2003, where winds where predominantly blowing in the direction of the samplers (NW); b) a lower group, in which variable wind conditions were present during the 24 hour TSP sampling events. 140 E o> c o ra CD O c o O 100 90 80 70 H 60 50 -40 -30 -20 -10 0 •-Apr-15 •— Apr-29 ± - M a y - 1 8 x— May-25 Jun-26 • Jul-17 H—Aug-11 — Sep-11 0 - r — 2 — i — 4 - r — 8 — i — 10 Distance (m) 12 — i — 14 16 Figure 4.39. Total Suspended Particulate measurements on T C H right of way during the monitoring period (Apr-Sep, 2003) None of the TSP measurements on the right of way reached the maximum acceptable air quality objective of 120 pg/m 3 ( C C M E , 1995) for a 24 hr monitoring period. These results agreed with a preliminary local study by Lynch and Wyse (2001), where TSP concentrations did not exceed air quality objectives, registering a maximum of 113 pg/m 3 at 5 m from the road edge. This showed the importance of deposition in areas closer to the road, where up to 90% of the suspended particulate mass can be lost (Watson et al., 1996). Total Suspended Particulates samples exhibited maximum particle sizes, which ranged from approximately 30 pm closer to the road to 16 pm at the 15 m sampling location, reflecting depositional effects and the effectiveness of this process in removing larger particles from the air. Figure 4.40 shows the contrast in particle size distributions for TSP (upper curves) and dust deposition particulates (lower curves) from collocated samplers. Suspended Particulates at the 7 and 15 m location exhibited similar size distributions indicating a lower rate of deposition for the finer breathable particulates such as those in the P M 10 and P M 2.5 group. 141 1 10 100 Part ic le Size (pm) Figure 4.40. Comparison of particle sizes for TSP and dust deposition samples at T C H 4.2.4.2 M e t a l Concentrations in T S P Total metal concentrations in TSP samples followed the order: Fe>Cu>Zn>Mn>Pb (Figure 4.41), which, except for the metal Cu , resembled similar trends observed in road dust (Fe>Zn>Mn>Cu>Pb) and dust deposition samples (Fe>Zn>Cu>Mn>Pb) collected at T C H . Results highlighted Cu as the second most important metal contribution from the road in TSP. This change in the rank of C u was also consistent with findings for road dust (section 4.1.2.1) and dustfall metal concentrations (4.1.3.2) collected at Hwy 17, where C u concentrations appeared to increase in dustfall samples versus road dust samples i.e. as particle size decreased. Previously, Lynch and Wyse (2001) reported for suspended particulates at T C H an inconsistent pattern of Cu concentrations in spatial terms, with higher C u concentrations in locations farther away from the highway. These findings provided more evidence to suspect an important association of C u with finer (< 30 pm) atmospheric particulates in areas close to roads. On the other hand, a documented C u contamination problem exists arising from the wear of H i -V o l motor armatures, as some cross contamination may exist between sampled and exhausted air ( U S E P A , 1997b). This problem was addressed in this research by the application of exhaust hoses on H i - V o l motors as recommended by the referred U S E P A document (1997b). However, no statistical difference was found between copper concentrations measured in this study and 142 those reported earlier for the same site by Lynch and Wyse (2001). In that study, the authors could not discard C u cross contamination as an explanation for erratic C u concentrations in suspended particulates. c CD O C o U 25CH £ 200\ ~Sb lOfl 3.5 7.0 15.0 Copper 3.5 7.0 15.0 Lead 3.5 7.0 15.0 Manganese 3.5 7.0 15.0 Zinc Figure 4.41. Summary statistics for metal concentrations in Total Suspended Particulates at different distances from the edge of Trans-Canada Hwy. (Note: Circles represent mean concentrations, boxes are 90% confidence intervals of the mean, bars across the boxes are medians, and whiskers are minimum and maximum values). Therefore, assuming cross contamination in this research was avoided or at least diminished by the use of exhaust hoses, the increase of C u in suspended particulates could be due to traffic activity sources. It has been documented that the main atmospheric C u contribution in roads is from Diesel engine heavy-duty vehicles, which account for 14% of Cu air emissions ( C I T E P A , 2004). This Cu contribution from roads is further studied by comparing Cu concentrations in roadside and background soil. The heavy-duty traffic influence and resuspension of fine C u contaminated roadside soil could be proposed as possible explanations for the increased C u concentrations in TSP. Despite the proximity of H i - V o l (High volume) air samplers to the road, none of the metal concentrations in TSP filters exceeded the 24 hr B . C . A i r Quality Objectives of 4000 ng/m 3 for Pb and 5000 ng/m 3 for Zn. The G V R D reports maximum 24 hr concentrations in TSP of 100 143 ng/m for Pb and 190 ng/m3 for Zn , with comparable concentrations for C u and M n in the Lower Fraser Valley ( G V R D , 2002). None of the TSP metal concentrations measured on the right of way exceeded these maximum concentrations. 4.2.4.3 Metal Partitioning in TSP Metal partitioning of TSP samples collected at T C H , displayed an increasing amount of metal in the exchangeable fraction (Figure 4.42), when compared to dustfall samples (Figure 4.37). If the percentage of exchangeable metal is used as an indicator of potential bioavailability (Hlavay et al., 1996 and 1998; Querol et al., 1999; Varga et al., 2000), then potentially bioavailable metal increased as particle size decreased (Figure 4.28, Table 4.12). A summary of metals concentrations and their relative partitioning in Total Suspended Particulate samples at T C H is provided in Table 4.13. The greatest increase in metal bioavailability in TSP samples takes place among metals Cu , M n and Zn, with increases in the exchangeable fraction of 29%, 41% and 40%, respectively. The rank of bioavailability, as expressed by the solution + exchangeable metal fraction, followed the order: Zn>Mn>Cu>Pb>Fe. Table 4.13. Summary of Speciation of metals in TSP samples at T C H for the moniroting period (Apr-Sep, 2003) Mean Mean Carbonate Organic Total Total Solution + & & Cone. Loading Exchangeable Oxides Residual Element (ng/m3) (pg/g of filter) % % % C u 106126 85 + 23 42 40 19 Fe 6 0 2 ± 1 3 8 454 ± 1 0 5 0.9 70 29 Pb 4 ± 1 3 ± 1 13 74 12 M n 1 8 ± 4 1 4 ± 3 43 45 12 Z n 3 2 ± 7 2 4 ± 5 56 34 10 . Note: Mean total metal concentrations are based on 24 Total Suspended Particulates samples. Speciation percentages are based on the mean of 9 TSP filters. 144 15 location (m) 15 location (m) Figure 4.42. Metal speciation in Total Suspended Particulates for metals: a) lead, b) copper, c) Manganese, d) Zinc The increasing bioavailability with decreasing atmospheric particle size is a cause of concern particularly i f metal release can be expected after deposition on water bodies or during scavenging of aerosols by wet deposition. However, the limitations of the sequential extraction procedure itself must be taken into consideration. The method was originally developed for river sediments and specific soil/solution ratios and extracting solutions have been suggested (Tessier et al., 1976; Yong et al., 1993). 145 4.2.5 Exchangeable Me ta l in Road dust, Dustfall & T S P The exchangeable metal fractions from sequential extraction tests in road dust, dustfall and TSP samples from T C H are shown in Figure 4.43. The association of particle size and potential bioavailability is highlighted, as the road dust samples with a particle size < 2 mm (shown on the far left of Figure 4.43), exhibited the lowest exchangeable fraction, dustfall samples with an approx. particle size < 0.5 mm exhibited intermediate exchangeable metal fraction, whereas TSP with an approximate particle size < 30 pm exhibited the highest % of exchangeable metal with up to 60% in the case of Z n (far right side). 70 | 60 | 50 S a> 40 f 30 a | 20 " 10 0 Q cr I I T T U 3 • DC Q I Q or D_ i LL Q O I LL Q r5 I LL Q -O CL I 3 o DL p DL P I p Figure 4.43. Percentages of Exchangeable metal in Road dust (RD), Dustfall (DF), and Total Suspended Particulates (TSP) samples from the T C H site. A n important limitation of the percentages of exchangeable or potentially bioavailable metal for atmospheric particulate matter reported in the literature, as well as in this research, is the difficulty or impossibility to follow the suggested soil/solution ratios, in the SSE method. The amount of particulate collected on filters hardly exceeds decimal fractions of a gram, which results in a substantial decrease of particulate/extracting solution ratio. From this standpoint, the method may overestimate the actual bioavailability of metals collected in atmospheric particulates. On the other hand, the actual environmental deposition of atmospheric particulates on surface water bodies, for instance takes place under even lower particulate/solution ratios. 146 Therefore, the results from atmospheric particulate partitioning are mostly applicable in environmental processes such as rain wash-out of atmospheric particulates, the deposition of these particulates on forest canopies and atmospheric deposition on surface water bodies. 4.2.6 Highway Runoff 4.2.6.1 Rainfall-Runoff Relationships Highway runoff sampling at T C H reflected the busier traffic conditions of the site. Table 4.14 shows rainfall/runoff relationships recorded at this site. Runoff coefficients were greater than those recorded at H w y 17 and were comparable to those measured by Onwumere in other busy highways of B . C . (Onwumere, 2000), where the catchment area was for the most part impermeable. Additionally, recorded rainfall events tended to be longer and there were fewer dry days for the monitored period at T C H . Table 4.14. Rainfall-Runoff characteristics for rainfall events monitored at T C H Sampling date Time sampling started Daily Rainfall (mm) Site Rainfall (mm) Duration (min) Dry days Before Rain Event Dry days Between Rain Events Estimated Runoff Coefficient 20/02/03 15:10 160 13/04/03 13:30 11.2 1.9 170 none .5 . • 23/0.4/03 15:30 13.8 2.7 140 2 3 .8 22/05/03 21:50 12.4 0.6 100 none 17 .7 13/07/03 18:00 6.0 0.8 140 none 34 .4 4.2.6.2 Total Suspended Solids The amount of TSS removed by runoff was much greater than any runoff event sampled at Hwy 17, which may be a combined result of longer rainfall events, greater amounts of precipitation per event and greater amount of solids generation by traffic activities. Summary values of TSS are provided in Table 4.15. 147 Table 4.15. Summary of TSS values for runoff events at TCH Total Suspended Solids (mg/L) Sampling date Weighted Average Maximum Minimum Coefficient of Variation 20/02/03 132 200 83 0.3 13/04/03 110 553 40 1.1 23/04/03 447 1080 168 0.5 22/05/03 n.a n.a n.a n.a 13/07/03 98 151 48 0.4 4.2.6.3 Total Metal Similar to H w y 17 results, TSS showed strong associations with all the metals studied in this research for all sampling events (Figure 4.44). Metal concentrations followed the order: Fe>Zn>Mn>Cu>Pb (Figure 4.45), also consistent with road dust concentrations at T C H and runoff concentrations recorded previously at the less busy Hwy 17. These results highlighted the importance of road dust characterization throughout watersheds to infer contaminants of concern and metal inputs to surface water bodies due to highway runoff. 0 500 1000 1500 TSS (mg/L) Figure 4.44. TSS associations with different metals for the Apr i l 23 r d , 2003 discrete runoff sampling event. 148 2 o o o o o o o o o o o o o o Sampl ing t ime Figure 4.45. Metals concentrations in highway runoff for the A pr i l 23 r d , 2003 discrete sampling event The results also contrast with the order of metal concentrations in dustfall (Fe>Zn>Cu>Mn>Pb) and suspended particulates (Fe>Cu>Zn>Mn>Pb). This contrast in metal order supports previous observations at H w y 17, where the continuous process of atmospheric removal of finer particulates leaves the coarser particulates (road dust) for runoff removal, reflecting in runoff metal concentrations. Similar to Hwy 17, all runoff concentrations exceeded the C C M E water quality parameters for the protection of aquatic life ( C C M E , 2002). 4.2.6.4 Dissolved and Resin-Exchangeable Metal Dissolved metal Dissolved metal concentrations followed the same order observed for total metal concentrations (Zn>Mn>Cu>Pb>Fe) at H w y 17 and T C H , except for Fe, which decreased substantially in the dissolved/filterable form since this metal tends to be particulate bound or forms precipitates under oxic environments (Figure 4.46). Figure 4.46. Dissolved metal concentrations for the July 13th discrete sampling event The effects of rainfall intensity and duration on dissolved concentrations can be assessed when comparing metal concentrations for the Apr i l 23 r d and July 13 t h event (Table 4.16). For these sampling dates, despite total E M C being greater for the more intense and longer A pr i l 23 r d rain event, dissolved E M C still were greater for the July 13 t h event, which had less precipitation, less intensity and lower runoff coefficient. This effect was consistent with results reported by Sansalone et al. (1996), and it is the result of lower sediment transport and lower dilution taking place during milder and longer runoff events. Table 4.16. Total and Dissolved Event Mean Concentrations at TCH Total EMC Dissolved EMC Sampling Cu Fe Pb Mn Zn Cu Fe Pb Mn Zn Date pg/L pg/L pg/L pg/L pg/L pg/L pg/L pg/L pg/L pg/L 20/02/03 47 5400 35 120 278 9 50 1 36 47 13/04/03 33 3440 32 84 223 12 32 1 4 98 23/04/03 108 16100 91 351 663 42 47 7 18 24 22/05/03 48 4230 40 91 256 29 35 2 9 44 13/07/03 74 3340 26 89 399 62 60 4 9 155 150 Resin-Exchangeable Metal The order of metal bioavailability, when the absolute concentration of Chelex-Exchangeable metal is considered, follows the order: Zn>Cu>Mn>Fe>Pb. On the other hand, i f the Chelex-Exchangeable /Dissolved ratios are considered, the order is: Z n > Mn>Cu>Pb>Fe (Table 4.17). Both ranking criteria were consistent with runoff observations at H w y 17, providing a framework for estimating the relative fraction of bioavailable metal concentrations that can be discharged to streams, or as is the case in this research, flushed over roadside soils. Table 4.17. Summary of Event Mean Total, Dissolved and Chelex-Exchangeable metal concentrations in storm water runoff at T C H Metal Total Dissolved (ug/L) Dissolved / Total (%) Chelex-Exchangeable (Pg/L) Chelex / Dissolved (%) Chelx / Total (%) Copper 62 31 48 16 61 26 Iron 6502 45 1 6 14 0 Lead 45 3 7 1 52 4 Manganese 147 15 12 12 78 10 Zinc 364 74 24 52 79 17-Assuming that Chelex-Exchangeable metals exclude nonlabile metal complexes and metals strongly associated with colloidal matter (Liu and Ingle, 1989); then the Chelex/Dissolved (%) indicated the fraction o f ionic and weakly complexed metal in the 45pm fraction o f stormwater runoff. This ratio can be useful to evaluate the effectiveness of H w y Best Management Practices such as sedimentation techniques, which target the suspended solids as a measure to decrease metal loadings from hwy runoff. The Chelex/Total (%), on the other hand, indicated how much metal could be potentially bioavailble from the direct hwy runoff discharge into surface water bodies such as streams, rivers or lakes. The Chelex-Exchangeable metal or weakly complexed metal order (Zn>Mn>Cu>Pb) found in hwy runoff was consistent with the tendency of these metals to attach to soil colloids. For instance, the metal binding over a wide p H range and ionic strength conditions, in hydrous ferric oxide expressed by the cation surface complexation constants follows the order Pb>Cu>Zn>Mn (Dzombak and Morel , 1990), while these metals binding on humic substances follows the order Cu>Pb>Zn>Mn (Tipping and Hurley, 1992). 151 4.2.7 Roadside Soil 4.2.7.1 Subsurface Soil Samples Profiles of metals concentrations along the soil profile as well as physico-chemical characteristics are presented in Figures 4.47 through 4.50. Similar to soil samples collected at H w y 17, soils were predominantly acidic, particularly as the distance from the edge of the road increases, showing compositional differences between the coarse sandy fi l l that conforms the pavement structure, and the silty roadside soil on the right of way. Iron content and surface organic carbon increased with distance from the highway, as well as the amount of sand and silt. Mineralogy of roadside soils interpreted through X-ray diffractograms also showed the predominant presence of silicate minerals such as quartz, feldspars such as albite, and phyllosilicates in the mica group such as chlorite and muscovite (Appendix F). Maximum Pb concentrations at T C H were comparable to those found previously at H w y 17 (1000 mg/kg, p H = 6), but in the case of T C H , higher metal concentrations persisted at greater distances from the highway (Figure 4.49). Lead concentrations on surface soil at H w y 17 decreased down to 20 mg/Kg at 12 m, while for an equivalent distance, Pb concentrations were 10 times greater at the T C H site. Lead still showed the characteristic pulse release, or piston flow displacement recorded earlier at H w y 17, particularly at the coarser soil locations next to the edge of the road (BH2, BH5) . A t these locations, maximum Pb concentrations were located at 20 cm depth. The total extent of this metal migration down in to the soil profiles was approximately 60 cm for. the coarser gravely-sand soils at the edge of the H w y and 35 cm for the finer silty glacial tills in roadside soils. In a preliminary study at another location with similar design characteristics and geology on the T C H , L i (2000) found that Pb concentrations in soil profiles decreased to background levels at about 50 to 85 cm depth. The author also reported peak Pb concentrations (1000 mg/Kg, p H = 4) at 25 cm depth on a topographic depression of the highway median grassy area. Surface Zinc concentrations reached a maximum of 200 mg/Kg and likewise Pb, concentrations decreased with distance from the highway, in some cases diffusing through the soil column down to 40 cm (BH2). Zinc and C u exhibited remarkably similar accumulation in soil profiles, with the • ' •-* 152 former showing greater concentrations. The similar accumulation patterns of these two metals along soil profiles stresseed their similar ionic potential and their constant anthropogenic input from atmospheric and runoff sources, as evidenced by the strong association (0.97) between Z n and C u found in dust deposition samples. Both metals diffused back to background levels at about 1 m depth, consistent with results reported by L i , for a similar site (2000). Manganese and Fe showed similar shape in their concentration profiles, reflecting their association with oxides or hydroxides species. According to Kabata and Pendias (1991), M n can concentrate in different soil horizons, particularly in those rich in Fe oxides and hydroxides, but it is also accumulated in top soils as a result of its fixation by organic matter. Manganese also showed some uncharacteristic concentration spikes in surface soils located at the bottom of the Hwy ditches (BH3, BH6) , which may be from an anthropogenic origin. This hypothesis is further elaborated in the isotopic analyses section. 153 %soil<4.75mm P H Organic C (°/o) Surface Area (m2/g) Fe (mg/kg) Pb (mg/kg) Total Metal (mg/kg) 3 70 90 5.9 6.9 0 2 4 6 0 50 14000 24000 n 400 800 0 200 400 600 Figure 4.47. Vertical profiles showing soil characteristics and metals concentrations with depth at a) B H 2 (0 m) and b) BH5 (0 m) location. 4^ %soil<4.75 mm pH 0 - I • i i . 1 • • 0.2 - • • • 0.4 • ? pth 0.6 • • 0.8 -1 -• 1.2 • 9 0 —i Organic C (°/Q Surface Area (m2/g) 2 4 o 20 40 60 20000 35000 Fe (mg/kg) Pb (mg/kg) Total Metal (mg/kg) 0 500 1000 0 a) Organic C (°/<$ Surface Area (m2/g) Fe (mg/kg) Pb (mg/kg) Total Metal (mg/kg) 6 0 20 40 60 80 20000 35000 I « . . . I Figure 4.48. Vertical profiles showing soil characteristics and metals concentrations with depth at a) B H 3 (5 m) and b) B H 6 (5 m) location. %soil<4.75mm PH Organic C (° /o) Surface Area (m2/g) Fe (mg/kg) Mn (mg/kg) Total Metal (mg/kg) 75 100 5 7 9 0 2 4 0 20 40 60 20000 35000 o 500 1000 0 50 100 i i a ) %soil<4.75mm PH Organic C (°/o) Surface Area (m2/g) Fe (mg/kg) Mn (mg/kg) Total Metal (mg/kg) 50 80 5 6 0 20 o 100 14000 29000 o 1000 0 100 200 Figure 4.49. Vertical profiles showing soil characteristics and metals concentrations with depth at a) B H 4 (10 m) and b) B H 7 (12 m) location. %soil<4.75 mm 50 0 0.2 -f 0.4 0.6 0.8 -f 1 1.2 80 i — i — • • PH Organic C (%) Surface Area (m2/g) 10 o r H • • 20 i — i — i • • Fe (mg/kg) 40 14000 19000 i • • I i Mn (mg/kg) Total Metal (mg/kg) 1000 0 f - 1 50 X X X X •-ev ' ' • A • O A • • A }*<\ < 0 A f O A O A • O A 100 1—I • Pb o Cu A Z n Figure 4.50. Vertical profiles showing soil characteristics and metals concentrations with depth at the background location (Hwy No. 1 Study Site). 4.2.7.2 M e t a l Part i t ioning in Soi l Samples Metal partitioning in T C H roadside soils confirmed previous observations from Hwy 17, where an anthropogenic metal input was characterized by a decrease in the residual fraction, or an increase in the sum of exchangeable, oxide and organic fraction. Lead was mainly associated with oxide phases and greater total concentrations of this metal were linked to greater exchangeable amounts of Pb, reaching up to 12% at B H 2 , located at the edge of the highway (Figure 4.50). Pb (mg/kg) 0 100 200 300 400 500 600 700 800 Pb o 0.05 ? 0.1 0.15 0.2 0.25 Q. a 0.04 0.11 0.18 0.23 0% 20% 40% 60% 80% 100% • Exchangeable D Oxides • Organic • Residual Figure 4.51. Vertical distribution of Pb soil concentrations, and its relative partitioning in roadside soils at the B H 2 location. Zinc exhibited a slightly greater degree of mobility than Pb and Cu , expressed by the exchangeable fraction in surface soils. Zinc, like Pb also showed predominant associations for the oxides fraction, but Z n showed greater partitioning to organics (Figure 4.52. A s expected, Cu was predominantly partitioned in the organic and residual fractions. In general, partitioning results at T C H soils seem to be qualitatively consistent with observations made for Hwy 17, however the percentage of exchangeable metal seemed to be greater at T C H since metal input from the road and accumulation onto T C H soils is also greater (Figure 4.53. Pagotto et al. (2001), reported a significant risk of Z n mobilization under acidifying conditions and minimal risks of mobilization for Pb and C u in roadside soils in France. 158 Zn (mg/kg) 50 100 150 Zn (°/o) 0.05 •= 0.1 CL 2 0.15 0.2 0.03 0.08 0.13 0.18 0% 20% 40% 60% 80% 100% • Exchangeable o Oxides n Organic • Residual Figure 4.52. Vertical distribution of Z n soil concentrations, and its relative partitioning in roadside soils at the B H 5 location. Cu (mg/kg) 0 20 40 60 80 100 0.05 £ 0.1 a & 0.15 H 0.2 -I Cu (°/o) & 0.03 0.08 0.13 0.18 0% 20% 40% 60% 80% 100% sidual Figure 4.53. Vertical distribution of C u soil concentrations, and their relative partitioning in roadside soils at the B H 6 location. Manganese exhibited multiple peaks, which obscured identification of possible anthropogenic accumulation. However, in two boreholes located on ditch locations, surface M n concentrations were significantly greater than the rest of M n concentrations along the soil profiles (Figure 4.54). Precisely at these locations the amount of exchangeable M n increased substantially, which may be an indication of significant input from atmospheric and highway runoff depositional processes. 159 Mn (mg/kg) M n (°/o) 200 300 400 500 600 700 800 Figure 4.54. Vertical distribution of M n soil concentrations, and its relative partitioning in roadside soils at the B H 6 location 4.3 Summary Despite different meteorological conditions at the two study sites, common patterns were identified regarding the distribution of metals along roadside soils that result from atmospheric and hydrologic events. Detailed statistical analyses of road dust, dustfall and runoff for both sites is provided in Appendix H . However, summary charts that display the general trends observed at both sites are presented in this section. Road dust accumulation and characteristics such as moisture, metal content and leachability provided important clues to identify the impact from atmospheric and runoff migration processes on neighbor environments. Road dust may be a good indicator of potential areas that require attention in terms of implementing or assessing B M P and environmental impacts since its metal print showed in most cases a close resemblance with runoff and atmospheric quality. A summary of total metal concentrations observed at both sites is provided in Figure 4.55. Manganese concentrations were significantly different at both sites at the 95% confidence level reflecting a possible significant influence from traffic activity. Dustfall exceeded the desirable B . C . A i r Quality Objectives (175 mg/m 2-day) for most of the sampling events in a 6 m and a 3.5 m strip along Hwy 17 and T C H , respectively. On the other hand, none of the TSP sampling events at T C H reached the Maximum Acceptable A i r Quality 160 Objective for a 24 hr monitoring period ( C C M E , 1995). Therefore, the right of way was sufficient to buffer dust deposition exposure at both sites. Additionally atmospheric deposition was an important removal mechanism that lessened the highway's contribution of suspended particulate loadings to the airshed. Figure 4.56 shows the estimated metal concentrations in dustfall collected on the right of way at both sites. A l l of the metals studied showed significant differences from a light to a heavy traffic site. c o CO c 0) u c o O oi CO O) o 3 CO D Q TJ ra o 600.0 j 500.0 - -400.0 --300.0 200.0 --100.0 --0.0 + + Hwy 17 TCH Hwy 17 TCH Hwy 17 TCH Hwy 17 TCH C o p p e r Lead M a n g a n e s e Z inc Figure 4.55. Total metal concentrations in road dust at H w y 17 and T C H (bars are mean concentrations and whiskers are 95% confidence intervals of the mean). </) Q c o ra 4-1 C 0) o c o o ra 0) 400.0 j 350.0 300.0 - -_ 250.0 --I! 200 .0 . - -^ 150.0 --100.0 - -50.0 -0.0 111 Hwy 17 TCH C o p p e r Hwy 17 TCH Lead Hwy 17 TCH M a n g a n e s e Hwy 17 TCH Zinc Figure 4.56. Mean total metal concentrations in dustfall at H w y 17 and T C H (bars are mean concentrations and whiskers are 95% confidence intervals of the mean). 161 Copper and Z n exhibited a greater degree of association with decreasing particle size, especially at T C H . Copper showed the most distinct concentration increase, from coarser road dust and dustfall samples to the finer suspended particulate matter collected at the busy traffic site. This sharp increase in C u concentrations might have been related to the significant increase in diesel engine heavy-duty vehicles at this site. Runoff characteristics may depend on several variables such as: rain amount, duration, dry days before the event, and traffic numbers at the time of sampling among others. However, a significant positive impact on runoff quality was observed from sweeping operations. Constant accumulation o f road dust on the pavement resulted in a similar increase in solids and associated metals. After wet sweeping operations, metal concentrations even several days after sweeping decreased down to at least half the pre-sweeping concentrations. Figure 4.57 shows a summary of total metal concentrations in highway runoff recorded at both study sites. A l l metals showed a clear statistical difference, except for Pb, which indirectly related traffic characteristics with runoff metal quality. C o E c CD O c o o re o I-£ o c 500.0 j 400.0 -300.0 --200.0 100.0 0.0 Hwy 17 TCH Hwy 17 TCH Hwy 17 TCH Hwy 17 TCH C o p p e r L e a d M a n g a n e s e Z i n c Figure 4.57. Summary of runoff metal concentrations recorded from discrete and composite sampling events at both study sites (bars are mean concentrations and whiskers are 95% confidence intervals of the mean). In roadside soils, Pb exhibited a pulse release accumulation in coarse and permeable soils from both study sites, resulting from past exhaust vehicle emissions and subsequent deposition when alkyl Pb was widely used in gasoline. This anthropogenic metal input was not only hinted by 162 total metal soil concentrations, but it was also observed in Sequential Extraction tests. Most of the Pb contaminated soils exhibited greater amounts of labile metal and a distinct decrease in the proportion o f tightly bound "residual" extraction component. This pattern was also observed for metals Cu , M n , and Z n at suspected anthropogenic metal input locations. Most soil metal profiles diffused down to background concentrations at about 60 cm on coarse locations and 30 cm on the silty and less permeable soils. Analyses in different environmental media along highways helped identify Z n as the metal of future potential concern. Zinc was only second in concentration to Fe in most dust, air and water samples. Zinc was the metal that exhibited the highest mobility or potential bioavailability (as expressed by the percentage of exchangeable metal) in sequential extractions and the highest resin exchangeable proportion in runoff samples. In addition the role o f Fe and M n oxides, as well as organic matter, were highlighted as key elements in the natural attenuation of metals in roadside soils by the application of sequential extraction tests. 163 CHAPTER 5 LEAD ISOTOPIC ANALYSES Lead isotopic analyses were performed for three main purposes: Differentiate anthropogenic from lithogenic Pb in roadside soils. Identify possible past and present sources of Pb deposition. Relate Pb anthropogenic deposition and its accumulation with other metals that do not exhibit sufficiently variable isotopic composition such as Zn , Cu , and M n to infer their anthropogenic origin. This chapter presents relevant literature review specific to isotopic studies as tracers for anthropogenic input of metals in the environment, with particular emphasis on the highway environment. This is followed by results and discussion of isotopic analyses performed on roadside soils and atmospheric samples from both highway study sites, and finally closes with some conclusions and recommendations specific to this chapter findings. 5.1 Lead Isotopes as tracers of pollution The key to differentiating anthropogenic from lithogenic sources of Pb is the distinct isotopic ratios that Pb from mined ores or Pb from industrial sources exhibit compared to background Pb present in rocks. The radiogenic isotopes of Pb (206, 207 & 208) result from the radioactive decay of U and Th atoms. The Pb isotopic composition of a mineral therefore depends on its Pb /U and Pb/Th ratios and its age. Lead tends to exhibit highly radiogenic isotopic ratios in very old U and Th rich minerals. On the other hand, in ore deposits where Pb has been sequestered and concentrated by geologic processes and where Pb /U and Pb/Th ratios are high (e.g. galena), Pb tends to be less radiogenic (Faure, 1986). This radiogenic characteristic w i l l depend on the age of the ore deposit formation, with older ore deposits having lower isotopic ratios than more recently formed Pb ores. 164 Lead isotopic analyses have been successfully applied to identify Pb sources in different materials such as: soils (Chow 1970; Erel et al., 1997; Sutherland et al., 2003), water (Gelinas and Schmit, 1997; V a n de Flierdt et al., 2003), lake sediments (Graney et al., 1995; Moor et al., 1996), lichens (Carignan et al., 2002; Simonetti et al., 2003), ice cores (Vallelonga et al., 2002), and even biological materials such as blood (Gulson et al., 1995). Lead contamination in roadside environments has been extensively documented (Motto et al., 1970; Musket and Jones, 1980; Rodriguez-Flores and Rodriguez-Castellon, 1982; Kobriger and Geinopolos, 1984; Burguera and Burguera, 1988; Ffafen and Brinkmann, 1996). However, concentrations alone do not prove anthropogenic input from traffic sources, since other anthropogenic activities such as smelting, coal burning, mining, etc. can spread significant amounts of Pb into the environment through atmospheric deposition (Aberg, 2001). Patterson (1965) was one of the first researchers to point out that the automobile exhaust was the main source of environmental lead pollution and raised concern about the risks of long-term exposure to large quantities of Pb. Later, Chow (1970) observed that the isotopic composition of lead aerosols was similar to that of the tetra-ethyl lead additive in gasoline. This author also reported similar results for the Pb isotopic composition from the top 10 to 15 cm of roadside soil. In a study at a grassland site in the U . K . , Bacon et al. (1996) found that Pb isotopic signatures tended to increase with depth reflecting a more radiogenic signature in uncontaminated soil. Hansman and Koppel (2000) reported a sharp Pb isotopic contrast between unpolluted soil and anthropogenic contaminants in soil profiles in Switzerland and estimated that Pb pollution reached to depths between 20 and 30 cm. Erel et al. (1997) estimated through Pb isotopic tests in roadside soil samples in Israel that anthropogenic Pb migrated into the soil profile at a rate of 0.5 cm/y. Sturges and Barry (1987) identified the isotopic signatures of alkyl Pb in atmospheric particulates over eastern U . S . A and Canada. The authors reported distinct Pb isotopic compositions between atmospheric particulates from both countries, mainly due to the origin of Pb ores used for the production of alkyl lead additives in each country. In general, the authors stated that the U.S . ore sources reflected more radiogenic Pb isotopic ratios than Canadian Pb Sources. 165 In subsequent years, Reid et al. (1993) reported that Pb deposition loadings in Ontario declined substantially after the introduction of Pb-free gasoline in 1974 and continued a steady decline until Pb was completely phased out in 1990. Blais (1996) reported that Pb concentrations in lake waters in southern Ontario matched this Pb decline in atmospheric particulates. The author considered deep lake sediments the main sink of previous alkyl lead anthropogenic inputs since these sediments displayed a Pb isotopic signature consistent with this source. 5.2 Isotopic Signatures in Pb Ore Deposits Since the introduction of tetraethyl lead (TEL) in 1923, consumption of Pb in gasoline grew steadily until the seventies when studies by Patterson (1965), Chow (1970), the U .S . National Academy of Sciences (1972) and later on by an overwhelming number of researchers in the scientific community, raised enough concern among the general public to put pressure on authorities to decrease and finally eliminate the use of leaded gasoline in North America and most of the industrialized world by 1990. During this time, the alkyl Pb manufacturer in North America, used Pb from several ore deposits, which had characteristic Pb isotopic ratios that have been reflected on the different media where leaded gasoline consumption has had an influence. Most of these Pb ore sources were located in the U . S . A , Canada, Mexico, Peru and Australia (Sturges and Barrie, 1987). The 2 0 6 P b / 2 0 7 P b ratios in selected major Pb bearing ores are presented in Table 5.1 (Sturges and Barrie, 1987; Chow & Johnstone, 1965). A s highlighted by the studies of Sturges and Barrie (1987, 1989), Pb pollution due to leaded gasoline in Eastern Canada has been consistent with isotopic signatures from Canadian Pb bearing ores used in the production of alkyl lead, with some degree of mixture from U . S . Pb ore sources in areas where regional atmospheric patterns have spread contamination across the border. According to data shown in Table 5.1 and excluding the low radiogenic Pb ores from Ontario, the average Pb/ Pb isotopic ratio for Canadian ores would be 1.14 ± 0.02. Outridge (2000) reported an annual average airborne Pb/ Pb ratio of 1.163 in Ontario, Canada, after the elimination of gasoline lead additives, which indicates an increase in the 166 isotopic ratio when Pb from natural or crustal sources dominates. Gelinas and Schmit (1997) reported 2 0 6 P b / 2 0 7 P b isotopic signatures ranging from 1.184 ± 0.022 to 1.274 ± 0 . 0 1 2 for soils, streams and suspended solids in agricultural areas, while this ratio ranged from 1.160 ± 0.016 to 1.167 ± 0.015 for urban streams in southern Quebec. Erel et al. (1997) reported similar results for 2 0 6 P b / 2 0 7 P b ratios in Israel, with lead aerosols from the combustion of European gasoline exhibiting ratios lower than 1.15 and those of natural sediments and carbonate bedrock exhibiting ratios greater than 1.20 Table 5.1. Pb/ Pb ratios in selected major lead bearing ores (from Sturges and Barrie, 1987) Canadian Automobile Pb Sources Z U 0Pbru /Pb North Star, B . C . „ 1.06 Sullivan, B . C . 1.07 Bluebell, B . C . 1.13 Monarch, B . C . 1.19 Brunswick no. 6, N . B . 1.16 Reserve, N . B . 1.16 Manitouwadge, Ontario 0.92 Buchans, Newfoundland 1.15 F l in Flon, Manitoba 1.16 US Automobile Sources Durango, Mexico 1.20 Taxco, Mexico 1.19 Cerro de Pasco, Peru 1.20 Mt . Isa, Australia 1.04 Mississippi Valley, Missouri 1.39 Another important source of anthropogenic Pb in the environment, including highways, has been lead based paint (Yafee et al., 1983). The color stability properties o f Pb compounds made them the desired pigment to put in paint. The two most common forms of Pb used in paint pigments were: white lead (Pb carbonate) and red lead (Pb oxide). The amount o f Pb in paint pigment was very high often reaching 38% of the dry weight of paint (Pesce, 1995). 167 Lead based paint was widely used in houses around the world, particularly before the 1950's, with lower lead levels used until 1978, when the U.S . Consumer Product Safety Commission completely banned its use in households. In Canada, the paint industry voluntarily stopped using lead in interior and exterior household paints in 1991 ( I O M C , 1998). In the highways of British Columbia, lead based paint was used due to its durable properties in signs, barriers and pavements markings until its phase out in recent years (Buchanan personal communication). Therefore it was expected that some of the lead based paint isotopic composition was left on the roadside soils of the highway sites investigated in this research. A s was the case with leaded gasoline, Pb used in paint came from a wide variety of Pb bearing ore sources. According to Rabinowitz (1995) paints made at different times and places have shown different isotopic ratios. In a study to apportion sources of lead in house dust wipe samples Adgate et al. (1998) found that paint isotopic ratios varied more between homes than interior wipe sample, soil or street dust isotopic ratios. The authors reported an isotopic mean Pb/ Pb = 1.17 in interior house paint, with a range of 1.092 to 1.267 and a coefficient of variation of 3.5%. Sutherland et al. (2003) summarized the 95% isotopic confidence interval of documented Pb based paint sources as 1.16 < 2 0 6 P b / 2 0 7 P b < 1.18, consistent with the mean ratio reported earlier by Adgate et al. Throughout the course of this chapter, the Pb/ Pb ratio < 1.16 is considered the benchmark for Canadian Pb bearing ores, which were predominantly used for the manufacture of alkyl lead in Canada (Sturges and Barrie, 1987). The literature on lead based paint isotopic ratios seems to report slightly greater ratios with a mean Pb/ Pb ratio of 1.17. Hence, Pb isotopic signatures within that range in soils and atmospheric particulates sampled at Hwy 17 and T C H are estimated to have a predominant anthropogenic influence caused by these two main lead sources. 5.3 Lead Partitioning and Isotopic signatures Metal partitioning through selective sequential extraction has been highlighted as a procedure that can provide useful insight into the mobility and bioavailability of metals according to their possible association as exchangeable cations, precipitated or coprecipitated as carbonates, adsorbed to the surface of Fe and M n oxides or organic matter, and integrated into the crystal lattice of soil or rock minerals. This sequential procedure also sheds light onto the possible 168 anthropogenic source of metals by correlating the more removable metal extractions with the marked increase of metal input from human activities (Howard and Sova, 1993; Garcia et al., 1996; Phillips and Chappie, 1995; Pagotto et al., 2001). Isotopic analyses can enhance the results provided by sequential extractions, by apportioning more effectively the different natural and anthropogenic sources of Pb through the identification of the distinct isotopic signatures that each source exhibits. Using Pb isotope ratios as tracer of 206 207 bioavailability in Southern Quebec, Gelinas and Schmit (1997) reported lower Pb/ Pb isotopic ratios in the exchangeable Pb fraction of soils leached with 1 M sodium acetate than in the total Pb leached with an aqua regia solution. The lower isotopic ratio in exchangeable and more bioavailable fraction was linked with that from vehicle atmospheric emissions, while the higher isotopic ratios found in the aqua regia leaching were considered to have a mix of natural and anthropogenic sources. Similarly, Erel et al. (1997) reported lower Pb isotopic ratios for an initial 0.5 M HNO3 leaching from roadside soils than the ratios in subsequent soil acid leachings. In analyses of concentrations and isotopic compositions associated with different soil fractions in Czech soils, Emmanuel and Erel (2002) observed that anthropogenic Pb was primarily associated with surface bound and organic matter fractions. Backstrom et al. (2004) reported for roadside soils in Sweden that the labile fractions (exchangeable, reducible and oxidizable) had very similar isotopic ratios and that these contrasted sharply when compared to the isotopic signature of the residual fraction. The authors also observed that the isotopic ratio increased with both depth and increasing extraction strength. A s mentioned earlier in the Methods and Materials section, isotopic analyses were performed only in the extraction associated with Fe -Mn oxides, since a significant amount of Pb and other metals were consistently linked to this fraction in soil and atmospheric samples analyses. It has been suggested that the incorporation of anthropogenic Pb into soils, as characterized by Pb isotopic analyses, follows the order exchangeable/carbonates > (hydr)oxides > organic matter > residual (Backstrom et al., 2004). However, as it is discussed herein the isotopic tests performed in this research, solely for the Fe -Mn oxide extraction, still preserve the degree of anthropogenic influence with depth and distance from the highways. 169 5.4 Lead Isotopic Results 5.4.1 Highway 17 Lead isotopic ratios for road dust and soil samples at different distances from the highway and at different depths, are presented in Figure 5.1. In this figure it can be observed that the least radiogenic Pb/ Pb ratios corresponded to depths of 20-30 cm at the 1NB location, right at the edge of the highway. These isotopic ratios correspond to those of Canadian Pb-bearing ores ^ A y ; 0/1*7 ( Pb/ Pb < 1.16) used in the manufacture of alkyl Pb additives mentioned earlier. For this location, the top 20 cm seem to have lost its less radiogenic (anthropogenic) Pb (whether through dust resuspension, leaching or particulate bound Pb infiltrated through macro pores and/or preferential pathways) and isotopic ratios were more radiogenic and closer to natural background Pb ratios. 2.11 -, 2.10 n g°" 2.09 ° t -£L 2.08 -f CO o CM 2.07 2.06 • 25-30 • 20-25 X RD m 0-5 X RD • 1NB @ 0m H3NB @ 10m A Ditch @ 12m] XRoad Dust A 0-5 - I I I 1 I I— 0-5 • 15-fo _ l I I I 1_ k 25-30 1.14 1.15 1.16 1.17 1.18 2 0 6 p b / 2 0 7 p b 1.19 1.2 Figure 5.1. Lead isotopic ratios in Road Dust and Soil Samples at H w y 17. (Numbers next to the data points correspond to depths in the soil profiles at the specified location, except for road dust samples labeled R D ) . Road dust samples showed intermediate ratios consistent with the isotopic ratios that surface soils exhibited in the 1NB, 3 N B and Ditch locations. This isotopic similarity highlights road dust as the likely source of metal deposition in neighbour areas, in contrast to the past Pb accumulation due to leaded gasoline emissions still present in the deeper permeable soils at the edge of the road. The farther 3 N B and Ditch soil samples show a mixed influence of past Pb accumulations particularly at the surface. This is evidenced by the less radiogenic isotopic ratios present at the surface, compared to the 25-30 cm depth soils. 170 The different accumulation of the anthropogenic Pb along the soil profiles in 1NB and 3 N B soil sampling locations was mainly due to the different soil gradation at these sampling points. While the 1NB location consisted of a porous coarse soil, the 3 N B and Ditch locations at 10 and 12 m respectively, consisted of sandy silt with significant organic matter content. These results are further emphasized, with the display of soil metal concentrations along with Pb isotopic signatures across the 1NB soil profile at the edge of the road (Figure 5.2). Besides the sharp contrast in isotopic ratios, it is worth noting the association of Pb accumulation with that of M n . Total Metal (mg/kg) Pb (mg/kg) n t-nn 0 200 400 600 ' * 206 207 Figure 5.2. Metal concentrations across the soil profile and the corresponding Pb/ Pb isotopic ratios exhibited at the 1NB location. Farther away from the highway at the ditch location (12 m), the anthropogenic influence is not as definitive, but there exists a slight difference in Pb isotopic signatures from the top and bottom. The lowest Pb ratio was present in the topsoil consistent with a slight gradient in other metal concentrations, included with Pb, across the soil profile (Figure 5.3). The low Pb ratio in the topsoil suggested a mixed anthropogenic influence from previous leaded gasoline emissions and road dust deposition sources. The co-located atmospheric dust deposition samples, on the other hand exhibited the clear influence of road dust in current soil Pb accumulation (Figure 5.4). In this figure, road dust samples exhibit the least radiogenic 2 0 6 P b / 2 0 7 P b ratios and the dust deposition samples closer to the road are marked with this signature. However as the distance from the road increased, the isotopic signature became more mixed and radiogenic, reflecting the resuspension of more radiogenic background or crustal derived material. The background dustfall sample had the most 171 radiogenic isotopic signature, a result of the high U/Pb and Th/Pb ratios of soils with lower Pb concentration levels. Mn & Zn (mg/kg) Pb & Cu (mg/kg) 0 0.05 0.1 •'• 0.15 - : 0.2 0.25 0.3 -t 0 200 400 i i • i i i i i i 0 20 40 • i f -*— Mn -A— Zn i i • • i i ^ 1 7 8 5 4 5 l E ^ ) 206 207 Figure 5.3. Metal concentrations across the soil profile and the corresponding Pb/ Pb isotopic ratios exhibited at the Ditch location. -Q Q. <o o _Q Q . 0 0 o CM 2.10 -r 2.09 - j 2.09 - j 2.08 - j 2.08 -; 2.07 2.07 X RD • 0 m • NB B SB xRoad Dust ABkgrd 6 m • 0 m 12 m RD • 5 m . • 10 m • Bkgrd _i i • • 1 ' \—> • • • i '—•—'<-1.16 1.165 1.17 1.175 2 0 6 p b / 2 0 7 p b 1.18 1.185 Figure 5.4. Lead isotopic ratios in Road Dust and atmospheric dust deposition samples at H w y 17. (Numbers next to the icons correspond to distances from the edge of the Highway at the specified location, except for road dust samples labeled R D ) . 5.4.2 Trans-Canada Highway The validation site was expected to exhibit even lower radiogenic isotopic signatures due to the significantly greater traffic activity present at the site (80,000 vehicles/day). Figure 5.5 shows 206 207 Pb/ Pb ratios for road dust and soil samples collected at T C H . Most of the isotopic signatures recorded for road dust and soil were indeed lower than those recorded at H w y 17, reflecting 172 overall a stronger anthropogenic influence. Road dust presented an intermediate radiogenic signature between Pb accumulation, attributed to alkyl Pb emissions, and lithogenic Pb isotopic signatures (in this case the lithogenic signature could be considered the 50-55 depth soil, which did not have signs of Pb contamination). However, road dust gave a distinct signature that did not fit with the overall slope of isotopic ratios. This contrasting result is further developed below in the interpretation of Total Suspended Particulates and atmospheric dustfall isotopic analyses. 2.13-r 2.12 -j Q_ 2.11 -• 2.10-'• of- 2.09 O CM 2.08 -; 2.07 1.12 25-30 A R D 0 10-15 5 ^ 0 • 25-30 • 0-5 1.13 0-5 • BH6 @ 0m • BH5 @5m A Road dust ' ' ' I ' -f- 1-1.14 1.15 1.16 2 0 W 0 7 P b 50-55 1.17 1.18 Figure 5.5. Lead isotopic ratios in Road Dust and Soi l Samples at Trans-Canada Hwy. (Numbers next to the icons correspond to depths in the soil profiles at the specified location, except for road dust samples labeled R D ) . A s recorded for H w y 17, those soil depths exhibiting the oldest Pb accumulation at each borehole location also exhibited the lowest Pb/ Pb ratios. Figure 5.6 displays the soil profiles for the B H 5 location right at the edge of the highway. The topsoil exhibits higher ratios reflecting the migration downwards of anthropogenic or low radiogenic Pb previously deposited and the substitution for more radiogenic road dust and crustal derived Pb at the surface. Lead isotopic signatures become again more radiogenic as the influence from anthropogenic Pb diminishes at 50-55 cm depth. Other metals concentrations appear to be diffusing down across the soil profile, included M n , which shows a concentration peak closely resembling that of Pb soil concentrations. Borehole B H 6 at a farther location (5 m), had a gradual and slower diffusion of all metals, mainly due to the finer nature of the silty soil and its associated lower permeability (Figure 5.7). However the soil Pb sampled at the leading wedge of the diffused metal concentration profile 173 still give a distinctively low radiogenic isotopic signature, from the time when this anthropogenic input was constant in the highway environment. Noticeable too is the association of anthropogenic Pb with the other metals studied in this research. Total Metal (mg/kg) Pb (mg/kg) 0 200 400 600 200 400 1.17444±2E-5 Anthropogenic isotopic signature Lithogenic signature 7(\f\ 7fi7 Figure 5.6. Metal concentrations across the soil profile and the corresponding Pb/ Pb isotopic ratios exhibited at the B H 5 location (at the edge of the highway). Total Metal (mg/kg) Pb (mg/kg) 0 500 1000 0 200 400 2E-5 Figure 5.7. Metal concentrations across the soil profile and the corresponding Pb/ Pb isotopic ratios exhibited at the B H 6 location (5 m away from the highway). Lead isotopic signatures on dustfall samples showed road dust as an important source of current Pb and associated metals loading. As shown previously for atmospheric dust deposition samples at Hwy 17, isotopic signatures showed increasingly radiogenic ratios as the distance from the road increased, with the road dust signature showing the least radiogenic ratio (Figure 5.8). 174 Q. <o o "X. _Q Q. 0 0 o CM 2.13000 -r 2.12000 -i 2.11000 -; 2.10000 -; 2.09000 - ; 2.08000 - j 2.07000 - : 2.06000 -; 2.05000 + 1. B RD «> Deposition • Road Dust + 3.5 m • 7.5 m • 15 m Background 1 1 1 1 I—11 1 1 I 1 1 — 1 ' I 1 14 1.15 1.16 1.17 2 0 6 P B / 2 0 7 P B • • ' • 1.18 1.19 Figure 5.8. Lead isotopic ratios in Road Dust and atmospheric dust deposition samples at Trans-Canada Hwy. (Numbers next to the data points correspond to distances from the edge of the Highway at the specified location, except for road dust samples labeled RD). Total Suspended Particulates collected at different distances from the highway (Figure 5.9), on the other hand, showed mixed Pb isotopic ratios that could not be as clearly associated to the highway source as the atmospheric dustfall results presented above. This isotopic mixture resulted from the finer nature of Suspended Particulates (approximately < 30 pm), some of which can be transported several kilometers from the source where they are emitted, but also from the influence of local eddies and turbulence that enhanced the mixture of particulates from different origin. 2.13000 2.12000 £ 2.11000 ^ - 2.10000 » ~ 2.09000 CM 2.08000 2.07000 A RD 15 m • • 3.5 m • TSP1 • TSP2 A Road Dust 15 m B 3.5 m 7.5 m 1.14 1.15 1.16 1.17 2 0 6 P B / 2 0 7 P B • 7.5 m 1.18 1.19 Figure 5.9. Lead isotopic ratios in Total Suspended Particulate samples at Trans-Canada Hwy (Note: TSP1 = sampling event 1 [May 25,h & 26th, 2003]; TSP2 = sampling event 2 [Jun 26th & 27th, 2003]). 175 5.4.3 Metal Enrichment Ratios A method of enrichment ratios was used to better identify the anthropogenic accumulation of other metals (Cu, M n , Zn) coupled with Pb isotopic analyses. The enrichment ratio (ER) is defined as: [Metal/Al] s am Pie / [metal/Al] c r ust, and it is a method for identifying and quantifying the influence of anthropogenic activity with respect to natural element cycles (Zoller et al., 1974). This method was originally used by the authors to compare the chemical composition of atmospheric particulates, collected at the South Pole, to the composition o f the earth's crust or the ocean used as a baseline for natural sources. However, some authors (Taylor and McLennan, 1995; Reimann and De Caritat, 2000) have discouraged the indiscriminate use of E R in atmospheric particulates as a measure of anthropogenic activity, due to the inability of the method to account for natural geochemical differences such as large extensions of land with different underlying lithologies, and biogenic processes where plants can actively change their chemical environment and selectively accumulate some elements in organic soil horizons. In this study, limitations of the E R method have been considered; therefore an approach suggested by Simonetti et al. (2003) has been adopted, where A l values or crustal background reference values are used from locally gathered data rather than globally reported values. Additionally, E R data are compared with Pb isotopic signatures, and raw data (concentrations) as suggested by Reimann and De Caritat (2000). The authors discourage the indiscriminate use of E R and suggest using extensive raw data over a large area that are not affected by an interfering second element, which may show locally high values for a variety of reasons. 5.4.3.1 Metal Enrichment Ratios in Soils Enrichment ratios by the criteria defined above were calculated for soils of both observation (Hwy 17) and validation (TCH) study sites. Results showing the relationships between Pb concentrations in roadside soils and Pb isotopic signatures for both study sites are shown in Figure 5.10a. A s presented for some soil profiles previously, an increment in Pb concentration is well correlated with a decrease in the Pb/ Pb isotopic ratios, which in turn is also consistent with low radiogenic Pb input from anthropogenic origin. A similar plot is shown, but using Lead 176 E R as a defining characteristic, in lieu of concentration (Figure 5.10b). The figure shows a similar trend, but the correlation with Pb isotopic signatures is slightly improved. Figure 5.10. Summary of Pb/ Pb isotopic ratios for roadside soil samples at H w y 17 and Trans-Canada H w y study sites and their relationship with: a) Pb concentrations, and b) Pb Enrichment Ratios The strong relationship in the Canadian environment as well as in other regions of the world between low radiogenic Pb and anthropogenic Pb has been established (Sturges and Barrie, 1987; Monna et al., 1995; Monna et al., 1997; Jaeger et al., 1998). This relationship has also been confirmed earlier in soil and atmospheric dustfall samples at both study sites. Hence, relationships between Pb accumulation and other metals are explored herein, as an aid for the possible anthropogenic identification of metals: Cu , M n , and Zn . In the case of soil samples, the relationship between anthropogenic Pb and other metals is limited by the chromatographic effect that metals exhibit as they diffuse through the soil profile (i.e. accumulation of anthropogenic Pb at a particular location may not necessarily imply accumulation of other metals due to the different mobility that each metal exhibits across the porous media). Nevertheless, it is believed that these relationships are useful tools to provide supporting evidence of other less clear anthropogenic accumulation of ubiquitous metals such as M n . Lead and Copper Enrichment Ratios Lead and C u Enrichment Ratios showed significant correlations, particularly for T C H soil samples (Figure 5.11). Also noticeable are the higher E R ' s for C u and Pb at the T C H compared 177 to those calculated at H w y 17, reflecting the greater influence of traffic activity on the neighboring roadside soils at the T C H study site. Enrichment Ratios for C u at the T C H site reached up to 12 times the local crustal value compared to around 3 times for the less busy H w y 17. Pb Enrichment Ratio Pb Enrichment Ratio Figure 5.11. Associations between Cu and Pb Enrichment Ratios at: a) H w y 17, and b) Trans-Canada Hwy, study sites Known primary highway sources of C u to the environment include: metal plating, bearing and brushing wear, moving engine parts, brake lining wear, and fungicides and insecticides applied by maintenance operations ( F H W A , 1984). Lead and Zinc Enrichment Ratios These metals E R ' s also showed some degree of positive linear or exponential association among each other, and an increase in calculated E R ' s corresponding to busier traffic conditions as shown by the higher Z n E R ' s for T C H compared to those of H w y 17 (Figure 5.12). J — 1 — I 600 Pb Enrichment Ratio Pb Enrichment Ratio Figure 5.12. Associations between Z n and Pb Enrichment Ratios at: a) H w y 17, and b) Trans-Canada Hwy, study sites 178 Zinc highway sources include: tire wear (filler material), motor o i l (stabilizing additive) and grease, galvanized metal in guardrails, roofing materials and other automobile components ( F H W A , 1984). Councell et al. (2004) estimate that about 285,000 tons o f Z n were released from tire wear in the U.S . between 1936 and 1999 and that 10,000 tons of Z n were released in 1999 alone. Lead and Manganese Enrichment Ratios Enrichment Ratios associations for these metals were weaker than those exhibited for C u or Zn . They are however, significant at the 90% confidence level in the case of samples collected at the busier T C H site. Samples collected at Hwy 17 did not show a statistically significant linear correlation when using E R ' s , however a positive linear correlation was found when using soil concentrations rather than E R ' s [as suggested by Reimann and De Caritat, 2000] (Figure 5.13). The inadequacy o f M n E R to exhibit any correlation with Pb E R at the H w y 17 study site may have been due to the difficulty in finding an adequate local crustal value for this metal. 600 -r 500 4j c o I -3 400 8 I) 3 0 0 o 200 o c 5 100 0 a) -+-0 400 800 1200 Pb Concentration (mg/kg) 200 400 Pb Enrichment Ratio 600 Figure 5.13. Linear correlations between: a) M n and Pb soil concentrations at H w y 17, and b) M n and Pb Enrichment Ratios at Trans-Canada H w y Known sources of M n in the Canadian highway environment include: engine moving parts and the antiknock octane enhancer fuel additive Methylcyclopentadienyl Manganese Tricarbonyl ( M M T ) . M M T was formulated in the research laboratories of Ethyl Corporation in the 1950's and it has been widely used across Canada from the mid 1970s (Kaiser, 2003). However its input to the environment may date back to the days of extensive leaded gasoline use, when the "Tetra Ethyl Lead Motor 33 M i x " additive contained about 57.5% tetraethyllead, 17.6% ethylene dichloride, 16.7% ethylene dibromide, 7.0% M M T and 1.2% dye (Harwood, 1963). 179 5.4.3.2 Metal Enrichment Ratios in Atmospheric Samples Enrichment Ratio associations were not as useful as expected in identifying the possible anthropogenic source of other metals based on their association with anthropogenic Pb due to various reasons: 1) the removal of alkyl lead in current atmospheric emissions, 2) the current absence of lead based paints in horizontal and vertical signs on B . C . highways (Buchanan personal communication), and 3) the finer nature of atmospheric particulates, which can be transported great distances from the source where they are emitted and mix with locally sourced material altering the isotopic signal. It is estimated that a more comprehensive atmospheric study would be necessary to calculate reliable Enrichment Ratios for other possible anthropogenic metals. Figure 5.14a shows the association of atmospheric dustfall Pb concentrations collected at both study sites and their association with Pb/ Pb isotopic signatures. Higher Pb atmospheric dustfall concentrations are significantly correlated with lower Pb radiogenic signatures, which imply some degree of anthropogenic origin. Enrichment Ratios, however are too low and the association with anthropogenic signatures not significant (Figure 5.14b). n o. n a. 5 10 ' Pb Concentration (mg/Kg) 15 b) • Hwy 17 m TCH 1 1 1 2 Pb Enrichment Ratio Figure 5.14. a) Linear negative correlation between Pb concentration in atmospheric dustfall samples and 2 0 6 P b / 2 0 7 P b isotopic signatures at both study sites, and b) 2 0 6 P b / 2 0 7 P b isotopic signatures and Pb Enrichment Ratios in atmospheric dustfall samples at both study sites 5.4.4 Identification of Past and Present Isotopic Trends A summary of all Pb isotopic ratios obtained for road dust, soil, TSP, and dustfall samples for both study sites along with some previously reported isotopic data from different anthropogenic sources are shown in Figures 5.15 and 5.16. The purpose of organizing all data in one plot was to 180 identify past and present trends in isotopic signatures by comparing it with historical and current Pb isotopic data. It may be observed from Figure 5.15 that a significant number of samples show isotopic ratios *)(\f\ 907 below the designated benchmark for Canadian anthropogenic sources C U D P b r ' P b < 1.16) and even a greater number are below the estimated Pb based paint isotopic composition ( 2 0 6 P b / 2 0 7 Pb< 1.17). These results would put the majority of atmospheric and soil samples collected from the highway study sites in the anthropogenic activity influenced range. It can also be observed in this figure that most samples align on a Pb-Pb isotope ratio plot, which can signify mixed origin between two end-member Pb sources. The end members are based on results reported by Simonetti et al. (2003) in a study on epiphytic lichens to trace atmospheric sources in western Canada. These end-members would correspond to highly radiogenic American ore lead sources and low radiogenic Pb bearing ores in British Columbia. However, a few samples tend to divert from this main isotopic line, which hints at an additional Pb source (Figure 5.15). The samples that tend to divert most strongly from this line are road dust and uncontaminated soil samples, which reflect a different Pb origin in road dust resulting from highway pavement and vehicle wear, and the different Pb source present in uncontaminated soil. 0. to o CM a. .Sullivan BC 1.06 • BH6 ( • BH5( ARDTi OTSP xDust • RD 17 + 1NB A3NB • Gasoline - Natural Om 5m 1.11 1.16 2 0 6 p b / 2 0 7 p b - Natural 1.21 1.26 Figure 5.15 Summary of Pb/ 7 Pb versus Pb/ Pb from different natural and anthropogenic sources. Unleaded gasoline Pb isotopic value is the average of values reported by Erel et al. (1997). French gas Pb ratio as reported in Monna et al. (1995). American gas values from Sturges and Barrie (1987). Natural Pb isotopic signatures according to values reported by Godwin and Sinclair (1982) and later used by Simonetti et al. (2003). 181 Simonetti et al. (2003) reported that isotopic ratios for lichens in western Canada (which ranged in location from the artic circle, all the way down to the Canada-USA border) could be grouped in a triangular area marked by the influence of American Pb ores, B . C ores and natural isotopic signatures from the Canadian Cordillera (Godwin and Sinclair, 1982). It can be seen that the authors proposed triangular area, reproduced in Figure 5.16, accurately covers isotopic signatures of atmospheric and soil samples collected at both highway study sites. 1.06 1.11 1.16 1.21 1.26 206p b / 207p b Figure 5.16. Summary of Pb/ Pb versus Pb/ Pb from different natural and anthropogenic sources. Triangular area that groups anthropogenic and natural sources in western Canada proposed by Simonetti et al. (2003). 5.5 Conclusions and Recommendations Past Pb anthropogenic sources, such as paint and leaded gasoline, were successfully linked to significant Pb accumulation in roadside soils of Hwy 17 and Trans-Canada Highways study sites, through Pb isotopic tests. These tests were useful in discarding the possibility of previous contamination of imported soil used for the construction of the highway, by giving a clear record of the past use and subsequent phase out of the effects of these Pb sources across soil profiles. This was achieved by identifying the distinct Pb isotopic composition in older versus newer Pb accumulation at different depths. Older Pb accumulations exhibited lower Pb/ Pb isotopic ratios consistent with Canadian Pb bearing ores, while newer Pb accumulations reflected a 182 mixture of the more radiogenic Pb/ Pb ratios of road dust with the less radiogenic Pb of uncontaminated soil. Isotopic signatures in soils and atmospheric particulate samples exhibited mixed compositions from different sources, that could be grouped in a triangular area marked by: the influence of highly radiogenic American Pb ores, low radiogenic B . C ores and Natural isotopic signatures from the Canadian Cordillera, as suggested earlier by Simonetti et al. (2003). Isotopic analyses were helpful in identifying road dust as the dominant current source of Pb (and other road dust associated metals) in roadside soils, through the comparison of the isotopic signatures derived from road dust and atmospheric dust particulates. The known association of Pb with anthropogenic sources was used to indirectly relate other metals (Cu, M n , Zn) to the same source by means of the Enrichment Ratio method. Significant positive correlations at the 90-95% confidence level were found between C u , Z n and Pb Enrichment Ratios in roadside and dust deposition samples. Weaker correlations were found between M n and Pb due to the difficulty in defining a background crustal value for M n and the ubiquitous nature of this metal. The direct identification of M M T derived manganese poses a challenging task since numerous studies have indirectly related its accumulation to traffic activities (Loranger et al., 1994; Lytle et al., 1995; Loranger and Zayed, 1997; Hal l et al., 1998). However, to this researcher's knowledge no method has successfully been used to directly link M n accumulation in the environment to the combustion products of M M T in gasoline. It has been suggested that a technique similar to that used to directly determine organo-Pb compounds (Lobinski et al., 1993) could be used for the direct determination of organo-Mn compounds (Veysseyre et al., 1998). However, this technique has only been used in ice-core samples, where according to Lobinski et al. (1993) organo-Pb compounds remain stable due to very low temperatures and the virtual absence of biological life and U V radiation. In order to better identify the possible anthropogenic source of manganese from M M T and evaluate its distribution and accumulation in different environmental media, it is suggested that approaches similar to those proposed by Suarez et al. (1998) or W u et al. (1998) be used. The former group of researchers added an Ir tracer in fuel to determine soot exposure of commuters 183 in Baltimore, while the latter group has used the same tracer technique to determine size distribution of aerosols emitted by motor vehicles. A full scale environmental study in a small Canadian community could be carried out, where an Ir tracer could be added to fuel containing M M T and the deposition of emitted M n with Ir-containing particles be tracked to assess the environmental accumulation and mobilization of M n in air, water, soils and plants (Barling personal communication). 184 C H A P T E R 6 I N T E G R A T E D P R E D I C T I V E M E T H O D O L O G Y This chapter outlines a proposed integrated methodology to forecast metal accumulation in roadside soils, based on results provided in chapter 4 and 5. Additionally, modeling simulations are used to provide further insight on the mobilization mechanisms of metals through different environmental media. The integrated predictive methodology covers the two metal migration pathways that affect roadside soils: 1) atmospheric deposition, and 2) runoff discharge. These pathways are modeled individually, but their coupled contribution to metals loading in roadside soils is considered as a whole. These coupled processes are evaluated over a certain time (e.g. a year) and their metal predictions are used as an input to a mixing cell one-dimensional geochemical model that simulates the dominant soil-metal interactions taking place in roadside soils. 6.1 Atmospheric Modeling Traffic atmospheric pollutants are solid, l iquid and gaseous matter that when emitted in the air change its natural composition. These pollutants may be classified as those that behave like reactive gases, those that are non-reactive and those that have properties of small particles. A non-reactive primary pollutant does not undergo significant chemical reactions in the span of one day during diffusion, dispersion and transport processes. Non-reactive gases would include emissions of N O , C O and hydrocarbon compounds (De Nevers, 1995). Reactive gases or secondary pollutants are sulphur dioxide SO2 (associated with diesel exhaust), nitrogen dioxide N O 2 and O3 formed in the atmosphere by a complex set of reactions (Hamilton and Harrison, 1991). Very fine-grained solid airborne material that does not react with the environment, such as ash and dust, fall in the category of inert particulates or non-reactive. A i r quality modeling procedures, on the other hand, can be classified in four generic classes: Gaussian, numerical, statistical or empirical and physical. Gaussian models are the most used technique for predicting the impact of non-reactive pollutants. Numerical models may be more 185 convenient than Gaussian models for reactive pollutants, but they require much more extensive input databases and resources. Statistical or empirical models are used mostly when a lack of understanding of the physical and chemical processes exist or when the required database is not present to use a Gaussian or numerical model. Physical modeling involves the use of scaled models and representation of the possible environmental conditions ( U S E P A , 1997). Regulatory models approved by the U.S . E P A to assess air quality impacts near transportation facilities are based on the Gaussian approximation. The Gaussian models most widely used are semi-empirical. These models are derived from scientific principles (e.g. conservation of mass) but rely on empirically defined parameters (e.g. dispersion rates). Models in this category include the California Line Source Dispersion Model or C A L I N E 3 and its improved versions C A L I N E 4 and C A L 3 Q H C . These models can be used to predict carbon monoxide and particulates concentrations near highways and arterial streets given traffic emissions, site geometry and meteorology. B y default these models ignore deposition calculations since the main purpose is to get conservative estimations of pollutants near highways. However, data about deposition and settling velocity can be incorporated in the model to account for this process. In order to predict metal loadings in roadside soils, deposition processes need to be accounted for in model calculations. Therefore, modifications to these models make them useful for the purpose of this research. Dispersion models, besides meteorological conditions and other empirical variables (dispersion rates, deposition velocity, etc.), require an emission factor, which is expressed usually in units of mass per length (g/mile, g/Km). The emission factor is another input variable that needs to be measured or estimated from other so-called emissions models. Similar to air quality modeling, emissions models can vary in the level of complexity and the amount of input data required for calculations. Models such as M O B I L E 6 (Mobile Source Emission Factor Model) and P A R T 5 are . comprehensive models that require detailed characteristics of vehicular fleet to estimate emission factors for several pollutants such as hydrocarbon (HC), carbon monoxide (CO), oxides of nitrogen (NOx), exhaust particulate matter (which consists o f several components), tire wear particulate matter, brake wear particulate matter, sulfur dioxide (SO2), ammonia (NH 3 ) , etc. ( U S E P A , 2002). 186 On the other hand, statistical algorithms such as those in the AP-42 document for emission factor documentation ( U S E P A , 1997) are based on regression analysis of numerous emission tests. Sources tested include paved roads, as well as controlled and uncontrolled industrial paved roads. In this procedure, prediction of particulate matter (PM-2.5, PM-10 , PM-15) and Total Suspended Particulate TSP, is based on the fine dust loading on paved surfaces and other input variables such as the vehicle type distribution and average weight of vehicles traveling the road. In previous chapters it has been shown that metals loading by atmospheric processes on roadside soils are primarily associated with dust and particulate matter, models that account for other primary pollutants are considered too elaborate and beyond the scope of this research. Additionally, detailed hourly estimates are useful for receptor modeling scenarios, (where exceedences of air quality parameters need to be predicted with higher accuracy) while in terms of particulate loadings next to highways, average air quality conditions throughout the year provide a better description. Therefore, a procedure that comprises a statistical model such as the AP-42 algorithm to estimate emission factors, in conjunction with a Gaussian dispersion model to describe particulate transport, was selected to simulate atmospheric deposition on roadside soils. 6.1.1 Emission Factor Fugitive emissions are defined as those air pollutants that enter the atmosphere without first passing through a stack or duct designed to direct or control their flow (Kinsey and Cowherd, 1992). Thus, particulates are injected into the atmosphere by natural wind blowing erosion of land surfaces and by vehicles activities over paved, unpaved roads, agricultural and industrial sites (Gillies etal., 1999). Particulate emissions from paved roads originate from loose material present on the surface. A s this particulate source is removed it is also continuously replenished by other sources such as: deicing maintenance activities, wind erosion and deposition from surrounding areas, vehicle and pavement wear, etc. ( U S E P A , 1997a). 187 According to the AP-42 document ( U S E P A , 1997a) on the compilation of air pollution emission factors, the quantity of dust emissions from vehicle traffic on a paved road may be estimated using the following empirical expression: E F = k ( sL/2) U M (W/3) 1.5 (6.1) Where: E F emission factor for particle size category i ( g /VKT) V K T : Vehicle Kilometer Traveled W = k = sL = empirical factor for particle size category i ( g /VKT) road surface silt loading* (g/m 2) * silt loading is the fraction of road dust with a particle size < 75 pm average weight of the vehicles traveling the road (tons) The particle size categories and their corresponding empirical factor for the paved road equation are shown in Table 6.1. Using the silt loadings recorded at both the Hwy 17 and T C H study sites reported in chapter 4 (sections 4.1.2 and 4.2.2, respectively) in the referred equation, with a k multiplier for TSP, renders the predicted emission factors shown in Table 6.2 (For detailed E F calculations refer to Appendix B) . The direct measurement of E F ' s involves the installation of Particulate Matter samplers, all around the road section that is to be monitored. Gillies et al. (1999) used Particulate Matter samplers installed on an "inverted u " supporting structure to record emissions produced by the road in all directions. Kantameni et al. (1996) reported the use of tracer ratio techniques along with upwind and downwind samplers to measure roadway P M 1 0 emission rates. Table 6.1. Values of k to predict emissions of different particles sizes PM-2.5 PM-10 PM-15 PM-30 or TSP Size range M u l t i p l i e r k ( g / V K T ) 1.1 4.6 5.5 24 188 Table 6.2. Measured silt loadings (sL) and corresponding Emission Factors Predicted EF (g/VKT) 3.68 1.72 1.52 0.84 Actual Emission Factors were not directly measured in this research, due to the extensive effort that the estimation of this sole parameter would require, compared to the potential contribution o f this migration pathway to roadside soils. However, E F ' s can be indirectly estimated by the field measurements of deposition on roadside soils, in both Hwy 17 and T C H study sites, and by the additional collection of Total Suspended Particulates at the T C H validation site. This approach is further explained in section 6.1.2 of this chapter. 6.1.2 Gaussian Atmospheric Dispersion In terms of atmospheric dispersion modeling, highways can be conceptualized as a line source of pollutants that are constantly emitted to the atmosphere. A line source can be considered as the sum of an infinite series of point source releases over a particular length. Models such as the California Line Source Dispersion Model C A L I N E (and all its different versions) divide the highway in a series of elements for which, concentrations are computed at each one element according to the well-known Gaussian dispersion equation for a point source. Later, each element concentrations are summed to form a total concentration estimate from the line source for a particular receptor location ( F H W A , 1989). The Gaussian dispersion model selected to estimate particulate concentrations and deposition on roadside soils was the Fugitive Dust Model F D M ( U S E P A , 1993). This model was used due to its capabilities to compute concentration and deposition impacts from diverse fugitive dust Study Silt loading (sL) Month of collection Site (g/m2) July 0.17 Hwy 17 November 0.053 Apr i l 0.02 T C H June 0.008 189 sources such as: point, line and area sources. The model also incorporates an improved gradient-transfer deposition algorithm (Ermak, 1977), which is a valuable feature due to the greater importance of deposition processes when dealing with fugitive dust emissions. Additionally, F D M like many U S E P A technical resources is free software, and its public release encourages the development of a user network that can keep and upgrade the program. Emissions for each line source (highway) are provided by the user. The modeler also provides other relevant data such as: number of sources and receptors, coordinates of the sources and receptors, meteorological conditions, particle size classes ( if known), roughness length, particle density, etc. A detailed description of the F D M model is available in the user's guide manual ( U S E P A , 1993). In the modeling runs for this research, the emission values used were the ones estimated in the previous section from the AP-42 algorithm. The source of fugitive dust was the highway stretch under shady; the receptor locations were those corresponding to the dust deposition gauges and TSP samplers. Meteorological conditions were those specific to each study location, during each observation period. Particle size classes were based on the average size distribution of suspended particulates reported earlier for the atmospheric sampling programs. The roughness length considered was 25 cm, which corresponded to uncut grass conditions in roadside soils (Stull, 1988). Some input and output runs of F D M for both study sites are provided in Appendix B of this thesis. 6.1.2.1 Highway 17 Mode l ing Estimates Modeling estimates of deposition for the two sets of emission factors obtained at H w y 17 and for different sets of atmospheric conditions are presented in Figures 6.1 and 6.2. Additionally, real deposition data collected throughout the monitoring periods were superimposed on those figures. When the Emission Factor of 3.68 g / V K T for the end of the first monitoring period at H w y 17 was used, the deposition modeling results were approximately double the actual deposition data measured during the same period (Figure 6.1). On the other hand, the E F = 1.72 g / V K T obtained for the end of the second period, gave lower predictions when compared to actual deposition data collected (Figure 6.2). 190 Figure 6.1. F D M modeling results, using an emission factor of 3.68 g / V K T , and actual deposition data on roadside soils for the 1 s t monitoring period. 6 0 0 -i Distance (m) Figure 6.2. F D M modeling results, using an average emission factor of 1.72 g / V K T , and actual deposition data on roadside soils for the 1 s t monitoring period. In order to get reliable modeling results, several Emission Factors must be calculated based on numerous road silt loadings collection. A s mentioned earlier in Chapter 4, the A i r Quality and Assessment Division of the Greater Vancouver Regional District ( G V R D ) has measured silt loadings through out the Lower Fraser Val ley, as part of a regional effort to update their particulate emissions inventory ( G V R D , 2000). If information on silt loadings, gathered by the G V R D (2000) for local highways and arterial roads is used, an estimate of average dust loadings in H w y 17 can be obtained based on loadings in a similar nearby road (Table 16, Chapter 4). 191 Silt loadings reported for H w y 99 (south of Hwy 17) give an average of 0.07 g/m 2 , which is more consistent with the silt loading measurement for the 2 n d monitoring period (Jul-Nov., 2002). This average silt loading would give an E F = 2.07 g / V K T , which for the 1 s t monitoring period would result in the deposition predictions shown in Figure 6.3. It can also be observed from this figure that deposition predictions are significantly improved by using a lower emission factor. 600 o Model X Data - - -Expon. (Model) Expon. (Data) Distance (m) Figure 6.3. F D M modeling results using an Emission Factor resulting from average collected silt loadings (2.07 g / V K T ) , plotted with actual deposition data on roadside soils for the 1 s t monitoring period at H w y 17. In all these figures, exponential trend lines of the measured and modeled data points have been plotted with the sole purpose of illustrating the differences between mean model predictions and mean measured data. Correlations of these summary trend lines for the whole set of data points are low, and would seriously limit the assessment of air quality in roadside soils based on a summary of air concentrations derived from these curves; specially since air quality impacts are assessed based on maximum event concentrations rather than average quality values. However, since predictions of atmospheric inputs on roadside soils need to be based on average trends (deposition over an extended period of time), the trend lines provide insight into the influence of model input parameters such as the Emission Factor. These findings along with model limitations are further discussed in the section 6.1.3, where F D M is evaluated after being applied at two different sites to predict the concentration of suspended particulates along with atmospheric deposition. 192 6.1.2.2 Trans-Canada Highway Mode l ing Estimates Prediction estimates of Total Suspended Particulates and deposition on the right of way of T C H , based on the Emission Factors calculated earlier (Table 6.2), are shown in Figures 6.4 through 6.6. Figure 6.4 shows that for silt loadings of 0.02 g/m 2 , and a resulting E F = 1.52 g / V K T , model predictions were about double the actual TSP measurements. Model predictions had an outlier for the atmospheric conditions present on the July 17 t h sampling event, which had higher southerly winds that could disperse fugitive dust more efficiently (upper data points in Figure 6.4). However, these atmospheric conditions did not impact the actual TSP measurements. The sharp contrast in modeling versus real TSP data for this day was most probably due to previous rain events (July 13 t h and 14 t h), which washed off the pavement surface and caused a decrease in the actual fugitive dust Emission Factor generated by the road. This change in E F is hard to quantify, particularly i f it is considered that only a few measurements of silt loadings can be typically made to characterize emissions from roads. However, as mentioned earlier for the purposes of this research, the average atmospheric contribution would be more representative than sporadic spike atmospheric pollution events. 200 i E 180 -160 -3-c 140 -o 120 -re i _ +-» 100 -c CD 80 -O c o 60 -o 40 -Q. CO 20 -J— 0 -o g 6' o Model X Data - - -Expon. (Model) Expon. (Data) 1^ 10 Distance (m) — i — 15 —i 20 Figure 6.4. F D M model predictions of TSP based on an E F = 1.52 g / V K T , superimposed with actual TSP measurements on the T C H right of way study site at different distances from the road. 193 The resulting F D M deposition predictions are also greater than the actual deposition measurements recorded in dust gauges during the study period (Figure 6.5). This is expected since higher predictions of TSP emissions than those registered should result in greater deposition rates, under similar deposition velocity conditions. 450 "re 4 0 0 350 H 300 O) J. 2 5 0 c 200 o 5 150 (A o 100 6 50 H 0 0 o o Model X Data - - - Expon. (Model) Expon. (Data) — i — 10 — i — 15 Dis tance (m) 20 Figure 6.5. F D M deposition predictions for an E F = 1.52 g / V K T , and actual deposition data on T C H right of way. 0 -I , , • 1 0 . 5 10 15 20 Distance (m) Figure 6.6. F D M model predictions of TSP based on an E F = 0.81 g / V K T , superimposed with actual TSP measurements on the T C H right of way study site at different distances from the road. 194 If on the other hand, the lower E F = 0.81 g / V K T previously calculated is used (table 6.2), the resulting TSP predictions are slightly improved (Fig. 6.6). Additionally, deposition predictions using this lower E F are greatly improved as shown in Figure 6.7. Considering six other silt loadings colleted by the G V R D -Table 4.10- ( G V R D , 2000) a silt loading average, sL = 0.02 g/m 2 (and a resulting E F = 1.5 g / V K T ) for T C H near 190 t h St., would place most predictions of emissions and deposition generated by this road, slightly above those recorded for the monitoring period (Apr-Sep, 2003). 300 •g 250 -1 i CM -§ 200 -| — 150 c o £ 100 (A O S" 50 Q 0 o O Model X Data - - »Expon. (Model) Expon. (Data) 5 10 Dis tance (m) — i — 15 20 Figure 6.7. F D M deposition predictions for an E F = 0.81 g / V K T , and actual deposition data on T C H right of way. 6.1.3 Limitat ions and Rel iabi l i ty A general assessment of TSP and deposition predictions can be observed in the scatter plots shown in Figure 6.8, where data aligned with the 45° slope would indicate a perfect match between model predictions and actual data. It can be observed that predictions for atmospheric deposition on roadside soils were significantly better than TSP prediction. This is deemed acceptable for the purpose of predicting atmospheric pollutant inputs on highway neighbor soils, but not acceptable for air quality evaluation purposes. It is believed that the over prediction of TSP by the model was mostly attributable to limitations of the E F predicting equation in the AP-42 document, rather than limitations in the Gaussian dispersion and deposition model. For example, It has been reported that equation 6.1 has a high reliability rating i f it is applied within the range of conditions specified in Table 6.3 ( U S E P A , 195 1997a). As seen from that table, all conditions for using the equation are met, except for the June sampling at T C H , where silt loadings were below the specified range. Figure 6.8. Scatter plots for predicted vs. measured: a) Total Suspended Particulates, and b) atmospheric deposition Table 6.3. Source conditions used to develop Equation 39 ( U S E P A , 1997a) Silt loading (sL) (g/m 2) Average vehicle weight [W] (Tonnes) Average vehicle speed (km/hr) 0.02-400 2.0-4.2 16-88 However, Venkatram (2000) provided a critical examination of empirical emission factor models such as the AP-42 , arguing that the model: 1) lacks a mechanistic basis, 2) its formulation is highly dependent on the dataset used to derive it, and 3) the accuracy of the model is determined by the methods used to measure emissions. Venkatram et al. (1999) estimated that PM10 predictions of emissions using the AP-42 model can deviate by as much as an order of magnitude from actual measurements. On the other hand, dispersion modeling limitations may include among other factors: 1) the use of empirical dispersion values derived from the prairie grass study conducted in 1956 (Pasquill, 1961; Hilsmeier and Gifford, 1962), 2) the concept of artificial uniform mixing of particles throughout the lowest atmospheric layers, which causes the models to underestimate deposition and overestimate road dust concentrations (Dong et al., 2003), and 3) lack of understanding and characterization of dispersion induced by vehicular motion. 196 For the Fugitive Dust dispersion Model used here, Winges and Gombar ( U S E P A , 1990) report in validation studies of the model that TSP predicted results are within a factor of two for measured results for 95% of the values, provided that an accurate and detailed emission inventory exists. Hence, this approach to atmospheric predictive modeling "as is" is considered appropriate only for evaluation of average impacts on roadside soils and neighboring areas. The better characterized the Emission Factors and the simpler the terrain features, the more reliable predictions wi l l be. Although the AP-42 model for E F calculations and Gaussian dispersion both have several limitations, these models have been shown to give reasonable approximations of actual atmospheric pollutant inputs along roadside soils of two study sites, which varied significantly in dust accumulation, traffic and meteorological conditions. 6.2 Runoff modeling In this research, a st