UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An evaluation of highway stormwater runoff quality in the G.V.R.D. Onwumere, George Chukwudi 2000

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

Item Metadata

Download

Media
831-ubc_2000-486877.pdf [ 12.29MB ]
Metadata
JSON: 831-1.0063736.json
JSON-LD: 831-1.0063736-ld.json
RDF/XML (Pretty): 831-1.0063736-rdf.xml
RDF/JSON: 831-1.0063736-rdf.json
Turtle: 831-1.0063736-turtle.txt
N-Triples: 831-1.0063736-rdf-ntriples.txt
Original Record: 831-1.0063736-source.json
Full Text
831-1.0063736-fulltext.txt
Citation
831-1.0063736.ris

Full Text

A N E V A L U A T I O N OF H I G H W A Y STORMWATER RUNOFF QUALITY IN THE G.V.R.D. by GEORGE CHUKWUDI O N W U M E R E Diploma of Technology, B.C. Institute of Technology, Bumaby, 1982 B.Sc , University of Alberta, Edmonton, 1988 B.Sc , Special Certificate, University of Alberta, Edmonton, 1990 M . S c , The University of British Columbia, Vancouver, 1992 A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T 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 DEPARTMENT OF CIVIL ENGINEERING We accept this thesis as conforming to the required standard: THE UNIVERSITY OF BRITISH C O L U M B I A March 2000 © George Chukwudi Onwumere, 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The Department of Civi l Engineering The University of British Columbia 2324 Main Mall Vancouver, B.C. V6T 1Z4 Canada Date: Abstract In the Greater Vancouver Regional District (G.V.R.D.), highway stormwater runoff from bridge decks along the Trans-Canada Highway (#1) in .Burnaby and the New Westminster Highway (#91, east-west connector) in Richmond was assessed between 1995 and 1996. Discrete and composite samples of highway stormwater runoff, road dirt, surface soil sediment and grass clipping samples were collected manually from both sites. The stormwater runoff samples were analysed for total suspended solids (T.S.S.), chromium (Cr), nickel (Ni), cadmium (Cd), copper (Cu), iron (Fe), zinc (Zn), manganese (Mn), lead (Pb), calcium (Ca), oil/grease, pH and electrical conductivity (EC). The road dirt, soil sediment and grass clipping samples were analysed only for their metal content. A l l the parameters in highway stormwater runoff showed differences in seasonal concentration patterns except for Cu and Mn at both sites. However, these differences were not statistically significant at the 95% confidence level. Although concentrations of most pollutants were higher in the winter, LC50 daphnia bioassays were non-toxic. The non-winter Comp " A " runoff samples, on the other hand, had 70% and 57% survival rates after 24 and 48 hours respectively. Most contaminant concentrations exceeded the maximum allowable concentrations (MAC) set for drinking water or freshwater aquatic life protection. Between the two sites, the Burnaby site had higher rainfall amounts and runoff coefficients, thereby generating higher T.S.S., metal and oil/grease concentrations/loadings than the Richmond site. The Burnaby site grass drainage ditch was fairly efficient in its pollutant removal effectiveness which ranged from 48% for Cu to 77% for T.S.S. There were statistically significant differences in pollutant removal efficiencies for all the parameters except for Mn at i i the 95% confidence level. Pollutant concentrations forecasting, using single regression equations with individual environmental variable, yielded reasonably good predictions for T.S.S., Fe, and Mn at the Burnaby site; and T.S.S., and Ca at the Richmond site. Comparison between discrete sample and flow composite data indicated a significant difference only in the concentration of T.S.S. at the Burnaby site. i i i Table of Contents Abstract ii; Table of Contents l'v List of Tables vil'i List of Figures xii Acknowledgement xvi 1. Introduction 1 1.1 Population Growth and Vehicle Usage 2 1.2 Site Precipitation Patterns and Water Quality 9 1.3 Automobile-Related Water Pollutants and Their Impact 12 1.4 Purpose of This Study 18 2. Literature Review 20 2.1 Highway Surface Pollutants and Their Chemistry 22 2.1.1 Particulate 23 2.1.2 Heavy Metals 28 2.1.3 Organic Contaminants 35 2.1.4 Pathogenic Bacteria 38 2.1.5 De-icing Agents 42 2.1.6 Nutrients 46 2.1.7 Others (Asbestos, Rubber and Special Compounds) 49 2.2 Sources of Highway Contaminants 50 2.2.1 Rainfall 50 IV Table of Contents (cont'd) 2.2.2 Dustfall 54 2.2.3 Vehicles and Their Exhaust 56 2.2.4 Highway Maintenance 60 2.3 Highway Contaminants Migration Patterns 61 2.4 Causes of Variation in Contaminants Loading 66 2.5 Impacts of Pollutants on Water Quality 72 2.6 Stormwater Models 79 2.7 Stormwater Management Measures 82 2.7.1 Source Management Measures 83 2.7.2 Post-Deposition Measures Applied Prior to Runoff 86 2.7.3 Post-Runoff Measures 87 2.8 Summary 99 3. Materials and Methods 102 3.1 Experimental Overview 102 3.2 Site Selection Criteria 102 3.3 Site Description 110 3.4 Experimental Equipment and Procedures 114 3.5 Road Dirt, Soil Sediment and Grass Clipping Samples 122 3.6 Sample Custody and Laboratory Analysis 123 3.7 Quality Control/Quality Assurance 127 3.8 Statistical Analyses 129 4. Results and Discussion 132 v Table of Contents (cont'd) 4.1 Rainfall - Runoff Relationships 132 4.2 Composite Data Analyses 145 4.2.1 Total Suspended Solids Concentrations/Loadings 145 4.2.2 Heavy Metal Concentrations and Loadings 155 4.3 Discrete Data Analyses 169 4.3.1 Pollutographs 169 4.3.2 Oil and Grease 184 4.4 T.S.S. and Metal Removal Efficiency Study 191 4.4.1 T.S.S. and Metal Removal Efficiencies 191 4.5 Water Quality Assessment 196 4.5.1 Water Quality Data Comparison 196 4.5.2 Daphnia Magna Bioassay Data 200 4.6 Road Dirt, Soil Sediment and Grass Clipping Data 205 4.6.1 Highway Road Dirt Analyses 205 4.6.2 Soil Sediment Data Analyses 214 4.6.3 Grass Clipping Data Analysis 222 4.7 Regression and Correlation Analyses 231 4.7.1 Pearson Correlation Matrix Between T.S.S., Metals and Oil/grease 231 4.7.2 Regression Forecasting Model 235 V I Table of Contents (cont'd) 5. Conclusion and Recommendations 250 5.1 Objectives 250 5.2 Conclusions 251 5.3 Recommendations 258 6. Bibliography 260 7. Appendices 275 7.1 Appendix A-Summary of QA/QC Data 275 7.2 Appendix B-T.S.S. Concentrations for Discrete and Composite Data . . . . 278 7.3 Appendix C-Sample Calculations Using the Burnaby Site Data 279 vii List of Tables 2-1. Highway Runoff De-icing Agent Concentrations 45 2-2. Common Highway Runoff Constituents and Their Primary Sources 51 2- 3. Summary of Highway Stormwater Management Measures 101 3- 1. Burnaby and Richmond Sites Selection Criteria 112 4- 1. Burnaby Hydrological Data for Sampled Events 133 4-2. Richmond Hydrological Data for Sampled Events 134 4-3. Burnaby Runoff Coefficients for Sampled Events 136 4-4. Richmond Runoff Coefficients for Sampled Storm Events 137 4-5. Comparison of Runoff Coefficients 139 4-6. Measured and Estimated Total Runoff Data at the Burnaby Site 143 4-7. Measured and Estimated Total Runoff Data at the Richmond Site 144 4-8. Concentrations and Loadings of T.S.S. at Burnaby and Richmond Sites 147 4-9. T.S.S. Concentrations Comparison with Other Highway Studies 149 4-10. Comparison of T.S.S. Loadings in the G.V.R.D 152 4-11. Total Suspended Particulate for Burnaby and Richmond Sites From 1994- 1996 154 4-12. Seasonal Mean Metal Concentrations at Both Sites 156 4-13. Burnaby Site Stormwater Runoff Statistics 157 4-14. Richmond Site Stormwater Runoff Statistics 157 4-15. Comparison of Metal Concentrations in Highway Runoff 162 4-15. Comparison of Metal Concentrations in Highway Runoff (cont'd) 163 4-16. Seasonal Metal Loadings in Highway Runoff at Both Sites 165 vi 11 List of Tables (cont'd) 4-17. Highway Runoff Metal Concentrations and Loadings at Both Sites 165 4-18. Comparison of Metal Loadings in the G.V.R.D 167 4-19. Highway Runoff Oil and Grease Concentrations in the G.V.R.D., U.S.A. andU.K 185 4-20. Seasonal Oil and Grease Loadings and Statistics at Both Sites 188 4-21. Overall Loading of Oil and Grease in the G.V.R.D., U.S.A. andU.K 189 4-22. Seasonal Grass Drainage Ditch Pollutant Removal Efficiencies at the Burnaby Site 192 4-23. Summary of Mean Pollutant Removal Efficiencies at Other Sites 192 4-24. Average Water Quality Parameter Concentrations and Their Guidelines 197 4-25. Daphnia Bioassay Data at the Burnaby and Richmond Sites 202 4-26. Road Dirt Seasonal Metal Concentrations for Burnaby and Richmond Sites . . 207 4-27. Burnaby Site Road Dirt Statistics 207 4-28. Richmond Site Road Dirt Statistics 208 4-29. Seasonal Soil Sediment Metal Concentrations for Burnaby and Richmond Sites 215 4-30. Burnaby Site Sediment Statistics 218 4-31. Richmond Site Sediment Statistics 218 4-32. Seasonal Grass Clipping Metal Concentrations for Burnaby and Richmond Sites 224 4-33. Burnaby Site Grass Statistics 225 4-34. Richmond Site Grass Statistics 228 ix List of Tables (cont'd) 4-35. Bumaby Site Pearson Correlation Between Pollutants 232 4-36. Richmond Site Pearson Correlation Between Pollutants 232 4-37. Correlation Matrix of Environmental Variables at the Burnaby Site 236 4-38. Correlation Matrix of Environmental Variables at the Richmond Site 236 4-39. Results of Single Regression Model Analyses at the Burnaby Site 238 4-40. Results of Single Regression Model Analyses at the Richmond Site 238 4-41. Comparison of Observed Pollutant Concentrations Versus Those Predicted by the Single Regression Model at the Burnaby Site for Storm Event of 03/09/96 246 4- 42. Comparison of Observed Pollutant Concentrations Versus Those Predicted by the Single Regression Model at the Richmond Site for Storm Event of 03/09/96 246 5- 1. Summary of Highway Runoff Data for Burnaby and Richmond Sites 252 5-1. Summary of Highway Runoff Data for Burnaby and Richmond Sites (cont'd) 253 5-2. Summary of the Grass Drainage Ditch Pollutant Removal Efficiencies at the Burnaby Site 255 A - l . Summary of QA/QC Results for the Parameters Analysed 275 A-2. Metal Detection Limits for Sediment, Road Dirt and Grass Samples at Both Sites 276 A-3. Annual Stormwater Flow Estimates at Burnaby and Richmond Sites 277 B - l . Discrete and Flow Composite Concentrations of T.S.S. at Both Sites 278 C - l . Discrete Concentration and Flow Data for the Storm Event of 05-07-96 at the Burnaby Site 281 x List of Tables (cont'd) C-2. Discrete Concentration and Flow Data for the Storm Event of 06-10-96 at the Burnaby Site 282 C-3. Discrete Concentration and Flow Data for the Storm Event of 08-02-96 at the Burnaby Site 283 xi List of Figures 1-1. World Population Trend from 1950 - 2000 3 1-2. G.V.R.D. Population Trends/Forecasts for 1961-2011 6 1-3. Average Monthly Rainfall Data for Both Sites from 1977- 1997 13 1-4. Average Monthly Rainfall Data for Both Sites for 1995 and 1996 14 3-1. Burnaby and Richmond Site Locations 104 3-2. Summer Average Daily Traffic for Both Sites 106 3-3. Burnaby Site Monitoring Instrumentation with the Curbs 116 3- 4. Richmond Site Monitoring Instrumentation with the Curbs 117 4- 1. T.S.S. Concentration Pattern at the Trans Canada Highway, Burnaby, Event #1 (11-04-95) 172 4-2. Copper Concentration Pattern at the Trans Canada Highway, Burnaby, Event #1 (11-04-95) 172 4-3. Iron Concentration Pattern at the Trans Canada Highway, Burnaby, Event #1 (11-04-95) 173 4-4. Manganese Concentration Pattern at the Trans Canada Highway, Burnaby Event #1 (11-04-95) 173 4-5. Zinc Concentration Pattern at the Trans Canada Highway, Burnaby Event #1 (11-04-95) 174 4-6. Lead Concentration Pattern at the Trans Canada Highway, Burnaby, Event #1 (11-04-95) 174 4-7. Calcium Concentration Pattern at the Trans Canada Highway, Burnaby Event #1 (11-04-95) 175 xii List of Figures (cont'd) 4-8. Copper, Zinc and Manganese Concentration Patterns at the New Westminster Highway, Richmond, Event #2 (01 -06-96) 175 4-9. Iron and Calcium Concentration Patterns at the New Westminster Highway, Richmond, Event #2 (01-06-96) 176 4-10. T.S.S. Concentration Pattern at the New Westminster Highway, Richmond, Event #2 (01-06-96) 176 4-11. T.S.S. Concentration Pattern at the New Westminster Highway, Richmond, Event #9 (04-01-96) 177 4-12. Copper, Zinc and Manganese Concentration Patterns at the New Westminster Highway, Richmond, Event #9 (04-01-96) 177 4-13. Iron and Calcium Concentration Patterns at the New Westminster Highway, Richmond, Event #9 (04-01-96) 178 4-14. T.S.S. Concentration Patterns at the Trans Canada Highway, Burnaby Event #9 (04-01-96) 178 4-15. Chromium Concentration Patterns at the Trans Canada Highway, Burnaby, Event #9 (04-01-96) 179 4-16. Copper Concentration Patterns at the Trans Canada Highway, Burnaby, Event #9 (04-01-96) 179 4-17. Zinc Concentration Patterns at the Trans Canada Highway, Burnaby, Event #9 (04-01-96) 180 4-18. Iron Concentration Patterns at the Trans Canada Highway, Burnaby, Event #9 (04-01-96) 180 xi i' List of Figures (cont'd) 4-19. Manganese Concentration Patterns at the Trans Canada Highway, Burnaby, Event #9 (04-01-96) 181 4-20. Calcium Concentration Patterns at the Trans Canada Highway, Burnaby, Event #9 (04-01-96) 181 4-21. Road Dirt Metal Concentrations at Both Sites 209 4-22. Sediment Metal Concentrations at Both Sites 217 4-23. Grass Cilpping Metal Concentrations at Both Sites 227 4-24. Burnaby Regression Plot for T.S.S 239 4-25. Burnaby Regression Plot for Fe 239 4-26. Burnaby Regression Plot for Mn 240 4-27. Burnaby Regression Plot for Oil and Grease 240 4-28. Richmond Regression Plot for T.S.S 241 4-29. Richmond Regression Plot for Ca 241 4-30. Burnaby Residuals Plot for T.S.S 242 4-31. Burnaby Residuals Plot for Fe 242 4-32. Burnaby Residuals Plot for M n 243 4-33. Burnaby Residuals Plot for Oil and Grease 243 4-34. Richmond Residuals Plot for T.S.S 244 4-35. Richmond Residuals Plot for Ca 244 C - l . T.S.S. Concentration Patterns at the Trans Canada Highway, Burnaby Event #16 (05-07-96) 284 x'iV List of Figures (cont'd) C-2. T.S.S. Concentration Patterns at the Trans Canada Highway, Burnaby Event. #19 (06-10-96) 284 C-3. T.S.S. Concentration Patterns at the Trans Canada Highway, Burnaby Event #22 (08-02-96) 285 xv Acknowledgements This research project was funded by the Highway Environment Section of the B.C. Ministry of Transportation and Highways. I would like to thank Dr. S.O. (Dennis) Russell (my supervisor) for his assistance in the preparation of this thesis. His support, suggestions and experience in hydrology were very helpful in ensuring the success of this study. I am grateful to Drs. K . V . Lo, S.T. Chieng and K.J . Hall (Committee Members) for their support and initiative in reviewing the initial manuscript and providing the necessary advice. M y thanks to Dr. Rich Horner (University of Washington, Seattle) and Mr. Stan Woods (G.V.R.D.) for their advice in the planning phase of this research project. And my sincere • appreciation to Kurt Nielsen (Technician) for building the weirs and vandalism protection compartments for the equipment, Paula Parkinson for all the laboratory analyses, Patricia Keen of B.C. Research Corporation for daphnia bioassays and Zailani Danladi (field research assistant) for his effort in the sample collection at the Richmond site. Similarly, my gratitude to Dr. Steve Smith (Simon Fraser University) for reviewing the initial draft of this report and offering his valuable advice, and to Ibrahim Adamu for his help. Finally, I would like to thank Kathy Lasuik for her support and effort in typing this document, Ms. Debra Jackson for revising some of the chapters, and my family for always being there. xvi Chapter 1 INTRODUCTION In the past few decades, concern has been rising about water pollution from non-point sources (NPS) such as agricultural, urban stormwater and, more recently, highway runoff. Even though water pollution from non-point sources is hard to measure due to their dispersed nature and sporadic discharge, the potential contamination hazards to water quality and the aquatic environment have been identified, documented to some extent and found to be significant (B.C. Research Corporation, 1991). Corrective measures to ameliorate the impact of water pollution from non-point sources are more difficult due to the size and general configuration of most of these sources of pollution. Highway stormwater runoff contributes to a variety of problems-which range from direct pollution of receiving waters and overloading of treatment facilities to the incorporation of pollutants into sediments and their eventual travel up the food chain. To mitigate these problems, site-specific treatment facilities and programs will probably have to be designed and implemented to control water quality deterioration. Although much work has been done with urban stormwater runoff, little research has been done in highway runoff in Canada. But since maintaining high water quality standards is a priority for most of the provinces across Canada, a lot more work with highway runoff is likely to be done in the future. Water by itself has the natural ability to undergo self-purification. In this process, the microbes in the water body initiate biological waste degradation under suitable conditions. But over the course of human development, the capability of some water bodies to undergo this natural self-purification has been eroded greatly by higher "loads" of contaminants. These increased loads can be the result of high population density in the catchment areas, vehicular traffic and industrialization. The net result is a growing accumulation of waste substances in the 1 oceans, lakes and rivers. These substances from both point and, especially from non-point sources such as highway runoff, which are difficult to deal with because of their dispersed nature, are posing threats to water bodies. Of immediate concern is the threat that these pollutants pose to aquatic life. Reducing population and automobile growth would help in restoring water quality, but these are outside the control of water quality professionals. However, they can contribute to water quality improvement through first determining pollutant sources and then developing appropriate treatment methods and facilities. 1.1 Population Growth and Vehicle Usage Population growth creates colossal impacts on the welfare of a city, ranging from the rate of urbanization and transportation systems to the degree of environmental pollution. By 1980, the global annual population increment passed the 80 million mark. Rates of increase are greater in developing countries than in the developed countries of the world. The projection for the year 2000 is for the global population to be over six billion people (Figure 1-1). This is going to exert tremendous pressure for growth of urbanization and transportation systems and on the effort to keep environmental pollution associated with such growth to a minimum (Brown et al., 1988). In such cities as Rio de Janeiro, Brazil; Mexico City, Mexico; Bangkok, Thailand; New Delhi, India and Lagos, Nigeria, excessive population growth has not only affected the rate of urbanization, leading to the emergence of "squatter towns", but has also led to encroachment on potential agricultural and transportation "right of way" lands (Brown et al., 1988). Accompanying this population growth has been an enormous increase in the number of vehicles. While this has benefited the exchange of goods and services within the country, the increase in numbers of vehicles has created havoc with traffic congestion and has led to an increase in automobile-related environmental degradation. While the mitigative technologies associated with 2 o o o CM in 0) 0) o 0) 0) 10 00 ^ < o w 0) o 0) o (0 0) o in 0) C/5 § E—1 U § OH ON Ov oo ON Z Q m H o ON ^H s § O H < OH o OH Q o s PH < < Q o o o O H OH D (Nomia) NonvindOd automobile-related environmental degradation is still at a developmental stage in developed countries, this is the least concern of the developing countries. As a result, most developing countries are going to be facing severe environmental problems in the future, i f not already, unless proper precautionary measures are initiated and implemented as soon as possible. In developed countries such as Canada, although the natural population growth seems to be in a declining phase due to low birth rates, this decline has been offset by an influx of new immigrants into Canada over the last decade. With these new immigrants, the population has grown from 25 million in the late 1970s to the current estimate of almost 30 million people. The population distribution, however, is uneven among the provinces. As a result, while the populations of some of the provinces are growing, such as British Columbia (B.C.) and Ontario, other provinces, such as Alberta, Saskatchewan and Manitoba, see stable or decreasing populations. The provinces with noticeable increases in their population are experiencing the problems mentioned earlier, namely accelerated urbanization, traffic congestion and automobile-related environmental damage. However, there are some situations where automobile-related pollution is not a problem. For example, the U.S. Federal Highways Administration (1985/1985a) identified four conditions under which stormwater runoff from highways is going to have minimal negative impacts on a receiving water body. No public water supply involved in the receiving water; The highway has less than 30,000 vehicles as the average daily traffic (ADT); The highway runoff is conveyed by overland flow in an unlined or grassed channel over a distance of at least 60 m prior to discharging into the receiving stream; The dilution ratio between highway runoff and receiving streamflow is at least 100:1. 4 The part of the Lower Fraser Valley region of the province of British Columbia (B.C.), more commonly known as the Greater Vancouver Regional District (G.V.R.D.), for example, has experienced a net increase in population of over 100% in the period from 1961 to present. Figure 1-2 shows the population trend/forecast for the period 1961-2011 (G.V.R.D., 1992). This increase in population is due primarily to interprovincial migration and an increase in foreign immigrants to the province. Accompanying this increase in population has been a corresponding increase in the number of motor vehicles—cars, buses and trucks. According to the statistical profile of Greater Vancouver, there has been a 29% increase in the total number of registered and insured vehicles in the G.V.R.D. for all rate classes (01 to 05) for the period of January 1986 to January 1993. Similarly, there have been 41% and 75% increases for vehicles insured for driving to/from work (rate classes 02, 03, and 04) and for business purposes (rate class 07) respectively for the same time period (G.V.R.D., 1993). This increase in the number of automobiles has been attributed to population growth in the suburbs and our increasing dependency on automobiles. The G.V.R.D. has experienced south-eastward sprawl in the last 20 years. The south and east sectors such as Surrey, Delta, White Rock, Langley and Fraser North experienced the heaviest growth, while the City of Vancouver, the North Shore, Burnaby and the City of New Westminster experienced modest growth. Figure 3-1 shows the map of the G.V.R.D. Based on this current population trend, the population of the G.V.R.D. is expected to reach 3 million people by the year 2021. Accompanying this population growth is a probable doubling of the number of automobiles to 2 million unless there are changes in current transportation habits. Most of this growth, in both population and the number of automobiles, is expected to be in suburbs (see Figure 3-1) such as Surrey, Langley, Mission, Abbotsford and Port Coquitlam (G.V.R.D., 1993). This pattern of growth is going to create an increasing number of problems for city planners, because it is very 5 CO o I N in csi CM in m (Noniii/M) NouvindOd o o CM CO 0) 0) d CO < CD LU 03 CO s SO as 2 o PH 00 H < a o on ON ON H o i—i H < CO CO OH o OH Q > 6 ( N Q oi > d § PL, < < Q CO 0) o M PH difficult to plan an effective, efficient and reliable transportation service for such a rapidly growing and diverse population sprawling away from the metropolitan core. The net effect is that more and more people will be automobile-dependent and the future outlook is for a growing dependency. However, this is not going to be without a price tag. According to travel surveys conducted by the G.V.R.D. in 1985 and 1992, there were noticeable patterns emerging from our travel habits. Some of the findings were: • Number of trips made during peak period grew by 37%, more than the population growth of 21 %; • Transit usage dropped by 1.3% on average; • Even though suburban trips to the City of Vancouver declined by 2.9%, trips from suburb to suburb gained 4%; and • The average trip distance to our workplace increased to about 14.0 km, with the average speed declining by 7% to about 34.7 km/h and an increase of about 20% to 24 minutes for the average trip time due to traffic congestion (G.V.R.D., 1993). Other impacts associated with current travel patterns are the effects the transportation demands are having on land use activities and the automobile-related environmental pollution associated with the transportation system. Transportation investments influence land use by providing better access to new growth areas. Where the markets for development are strong, like the Greater Vancouver Regional District in the last decade, this increases the population density and the land value of the new area. Such transportation investments tend to encourage urban sprawl by selectively providing better access routes. This tends to encourage a change in the land use activities of the new growth area. Due to population growth, most times the change is from agriculture-based land use to more residential, commercial-mall oriented and industrial usages. This usually increases the percent impervious surface area and vehicular traffic, thereby leading to an increase in automobile-related pollution. For example, it has been suggested that the opening of the Alex Fraser Bridge and the construction of the "new" Westminster Highway (the East-West connector) incited rapid developments south of the Fraser River. However, uncertainties exist as to the magnitude of the impact these two projects have with regard to development south of the Fraser River. The only indisputable fact is that more land is taken out of agricultural-based operations and put into urban usages. As a result, land prices are high, the potential for future land acquisition for transportation improvement does not look very encouraging (G.V.R.D., 1993) and the potential for automobile-related pollution is greater. Taking more agricultural land out of operation may be seen as an opportunity to reduce agriculture-associated water pollution. Chemical fertilizer application to cropland has been • attributed to a whole range of water pollution problems, from water eutrophication to nitrite-associated methaeoglobineamia disease in infants. However, there are also pollution problems associated with urbanization. As mentioned earlier, urban stormwater has been identified as a major non-point source of water pollution, second only to agricultural runoff (B.C. Research Corporation, 1991). Included in this category is runoff from the streets, residential, commercial and industrial areas. Recently, special attention has been focused on automobile-related pollution (Wang et al., 1982). There are two major types of automobile-related pollution of concern to engineers— air and water pollution. Currently, automobiles account for about 66% of total emission of five air pollutants, namely: oxides of nitrogen, oxides of sulphur, volatile organic compounds, carbon monoxide and particular matter (G.V.R.D., 1993). These pollutants can cause a whole range of problems for the region's air quality from smog formation to acid rain from oxides of sulphur in extreme conditions. However, despite the growth in the number of vehicles, the projection for 8 the future, at least for the G.V.R.D., is for cleaner air quality (G.V.R.D., 1993). The introduction of more fuel-efficient lines of automobiles, as well as the AirCare emission inspection, control and maintenance program, are expected to ensure reduction in these air pollutants. Offsetting the above trend, the production of "greenhouse" gases, such as carbon dioxide, are expected to rise due to demands on gasoline, diesel fuel, natural gas, propane and other carbon-based fuels. However, the overall prospects for a reduction in automobile-associated air pollution for the G.V.R.D. are quite good due to AirCare emission control, illustrating that a well-focused program of monitoring and control can be effective. Hopefully, a corresponding automobile-related water pollution program could also be effective. Urban stormwater runoff encompasses all the other types of surface runoff except agricultural runoff. However, with the increase in the number of automobiles and growing . environmental awareness, specific attention is now being paid to runoff from highways. Some research has been done in this field in the U.S.A. and Europe, but little or no research has yet been done with highway stormwater runoff in Canada. Even so, the research to date tends to show that runoff from highways contains a considerable amount of pollutants (Wang et al., 1982; Kobriger and Geinepolos, 1984a; Gupta et al., 1978/8la). 1.2 Site Precipitation Patterns and Water Quality With the exception of the prairie provinces, Canada is well-endowed with water resources. Canada has an average annual precipitation of approximately 600 mm. The average annual precipitation in the Arctic is about 100 mm, while along the Pacific region it is well over 3,600 mm. About 66.7% of this precipitation falls as rain, with the remaining 33.3% falling as snow, depending on the geographical location (Bertrand et al., 1985). British Columbia (B.C.) is probably one of the most water-rich provinces in Canada due to the amount of precipitation and 9 groundwater availability. In B.C., especially in the G.V.R.D., most of the precipitation falls as rain, with an average annual precipitation that ranges from 1,000 mm to well over 2,000 mm, depending on the location. These differences in the amount of precipitation influence site runoff generation. Rainfall runoff has been known to contain high, but variable, levels of contaminants, as will be explored in the next chapter. The amount of these contaminants depends on a range of variables from the average daily traffic and land use activities to climatic variations (Russell et al., 1979; Kobriger and Geinepolos, 1984a). Climatic variations affect both the quantity and quality of the runoff. The amount of rainfall varies with the season. The largest proportion of rainfall in the G.V.R.D. occurs from fall to early spring, with the least occurring during the summer months, as shown in Figures 1-3 and 1-4. Frequent, less intense rainstorms wash out considerable amounts of pollutants, whereas the more intense rainfall associated with fall and early spring tend to increase the solids concentrations, especially the suspended solids. These intense storms are responsible for the initial washout of contaminants, producing the "first flush" effect and the corresponding shock-loading to the receiving water (Tucker and Mortimer, 1978; Randall and Grizzard, 1983). According to past research, there is a relationship between the amount of contaminants in a first flush and the length of the dry period preceding the storm (Randall and Grizzard, 1983). Even though the relationship is variable, the trend points towards increasing contamination of the receiving water with an increase in the length of the dry period between storm events. In the G.V.R.D., the storms are frequent enough during the fall, winter and early spring that accumulation of contaminants may not be noticeable. This accumulation of contaminants, due to an increase in the length of dry period between rainstorm events, may be more observable during the summer months. However, there is a growing understanding that during dry periods, solids and other motor traffic pollutants are actively removed by wind action and vehicle-10 generated turbulence from highway surfaces. According to research reports from Norway, the United Kingdom and the U.S.A., active freeway lanes do not retain significant levels of pollutants due to saltation—an injection of sand-sized particles into the atmosphere by vehicle-generated turbulence (Lygren et al. 1984; Colwill et al. 1984; Kerri et al. 1985). Some question the correlation between the quantity of pollutants and the length of dry period between storm events. Asplund et al. (1982) reported solid pollutants to correlate with higher traffic, but only during storm events. There is less likelihood of solid pollutants being transported during rainstorm events by wind action or vehicle-generated turbulence. However, there will be some solids dispersal during rainstorm events from vehicle-generated splash action. The amount of water on the highway surface available for solids dispersal due to vehicle-generated splash action depends on the rainfall intensity and the drainage ability of the particular highway segment. Intense rainstorms have a tendency to provide more water on highway surfaces, leading to a lot of vehicular-generated splash action on high traffic highways. Solid pollutants dispersal by vehicular-generated splash action may help to reduce the amount of pollutants transported to a nearby water body. Any solid pollutant dispersed away from the drainage pathway through splash action may not be an immediate threat to a receiving water body. However, intense rainstorms are more likely to wash off more pollutants from highway surfaces. As a result, pollution of a nearby receiving water body is inevitable unless proper precautionary measures are in place. For example, Figures 1-3 and 1-4 show the precipitation pattern for both Burnaby Mountain and Richmond Nature Park areas, near the sites, for the last twenty years and two years during the study respectively. Based on data from Environment Canada, climatic data and these figures, the Burnaby Mountain area has approximately 33% more precipitation than the Richmond Nature Park area. As a result, the projection is for more pollutants to be washed off the highway surfaces in the Burnaby Mountain area than in 11 Richmond Nature Park. If the two locations are similar in terms of their proximity to a receiving water body, this should translate into more water pollution around Burnaby than Richmond. However, there are other factors such as average daily traffic, type of highway surface material, highway surface drainage ability and surrounding land use activities that influence the nature of water pollution from highway surfaces. Most of these influential variables will be explored in the next chapter, with the exception of highway surface drainage ability. In the past, highway drainage systems were designed for speedy transport of runoff water off the highway surfaces and their surrounding areas. In recent years, this traditional method of dealing with highway stormwater runoff has changed due to water quality concerns. According to Kobriger and Geinepolos (1984), stormwater runoff from highways is usually more polluting than general urban stormwater runoff. As a result, both speedy removal of stormwater runoff, as well as some kind of runoff treatment, may be required to mitigate water quality degradation. Designing a drainage system that will provide effective water removal, as well as a significant reduction in water pollution contamination, is going to be challenging. An effective solution starts with correct problem identification. Since highway stormwater runoff problems are both area and pollutant-specific, proper identification of automobile-related water pollutants will be required before an effective treatment/management program can be developed and implemented. 1.3 Automobile-Related Water Pollutants and Their Impact Until the last few years, there was a general lack of concern and data regarding urban stormwater quality, especially highway stormwater runoff quality. Recently, quite a number of research projects have been conducted in the U.S.A. and Europe with regard to the effects of automobile-related pollutants on receiving water bodies. There are many factors, such as vehicle wear and emissions, local land use conditions, average daily traffic and highway maintenance 12 400 350 Burnaby Mountain Site Richmond Nature Park Site 1 2 3 4 5 6 7 8 9 10 11 12 MONTH OF THE YEAR FIGURE 1-4. A V E R A G E M O N T H L Y R A I N F A L L D A T A FOR B O T H SITES FOR 1995 A N D 1996. (DATA F R O M ENVIRONMENT C A N A D A , 1995/97). 14 practices, to name just a few, that influence the amount of contaminants on highway surfaces. Similarly, the type and removal rates of these pollutants found on highway surfaces at any given time depend on other interrelated factors such as average daily traffic, local climate, wind action, vehicle speed, types of vehicles, land use activities and highway maintenance practices (Gupta et al. 1978; Hedley and Lockley, 1975; Horner et al., 1979). Based on research studies in the U.S.A. and Europe (Lorant, 1992), a brief summary of the highway-generated pollutants is presented below. A more detailed analysis will be presented in the next chapter. The contaminants of concern are biochemical oxygen demand (B.O.D.), nutrients (nitrogen, N and phosphorus, P), solids, de-icing agents, metals and organic contaminants. Some threshold levels have been set for most of these contaminants in the U.S.A. by the Environmental Protection Agency (EPA) (Lorant, 1992) and in the Canadian Water • Quality Guidelines (Environment Canada, 1987/95) by the Canadian Council of Resource and Environment Ministers. • Nutrients such as nitrogen and phosphorus have been known to cause accelerated eutrophication in water bodies. This can, in turn, lead to oxygen depletion, a change in the trophic status of the water body and potential fish kil l . With enough soil and drainage pathway retention time, the soil microorganisms and nutrient uptake by plants can help in reducing these nutrients before they enter a nearby water body (Onwumere, 1992). However, nutrient loadings are more of a problem with urban and agricultural runoffs than in highway runoff alone. • Solids, which include total solids and suspended solids, are mainly from pavement wear, vehicles, atmospheric fallout, highway maintenance and surrounding land use activities. Solids are the main carriers of pollutants in stormwater runoff, especially the fine particles. These solid sediments, especially 15 in the form of suspended load entering the surface waters, affect the quality of the receiving water by inducing turbidity. This increase in turbidity can, in turn, reduce prey capture for most predators that capture by sight, reduce photosynthesis in water column and clog gills of fish and invertebrates. Similarly, solids, being the main carriers of metal pollutants, can introduce a considerable amount of pollutants into the water column. This can be a major problem, since highway runoff tends to contain high amount of solids, sometimes comparable to urban stormwater runoff (Kobriger and Geinepolos, 1984a; B.C. Research, 1991; Wong, 1991). The only difference between the two is that urban stormwater runoff also includes the land constituents. Biochemical oxygen demand (B.O.D.), which is usually a measure of oxygen demand by bacterial activities, is usually low. B.O.D. data for highway runoff is limited, however, the general understanding is that highway B.O.D. data is lower than B.O.D. from urban stormwater runoff (Lorant, 1992). Sources of the oxygen- demanding organic matter contribution to highway runoff, which include plant litter, household waste, pet and bird droppings, are limited due to the general configuration of highway design. Where there is a considerable amount of oxygen-demanding substances in a body of water, however, the impact can be harmful to both aquatic life and water quality. De-icing agents, which are mainly chlorides, are only applied during the winter months in snowy regions. At elevated concentrations, the effects of de-icing agents on receiving waters and nearby soil/vegetation can be detrimental (Lorant, 1992a). However, with appropriate dilution during rain or snow melt conditions, the impact on both soil and water are minimal. 16 Metals attach themselves to solids. As a result, the observed high sediment metal contents can be attributed to rapid removal of metals through sedimentation. Similarly, there is a high metal content in sediment obtained from street sweeping. Most of these sediments loaded with metals may end up in receiving water where they affect both aquatic life and the water quality (Lorant, 1992). Insoluble forms of heavy metals in a stream do not pose an immediate problem, although the particulate metal forms may be transported and deposited in lakes, streams and rivers. Also, these contaminated sediments can be picked up by benthic organisms. Similarly, some of the insoluble metals may become soluble with increasing acidity, changes in the redox conditions and complexing agents. Soluble metals are associated with both chronic and acute toxicity to aquatic organisms at elevated concentrations. These metals come from several sources, as will be explored in the next chapter. Many of the organic contaminants found in urban/highway stormwater runoff are toxic to human and aquatic organisms. Some of these organic contaminants are carcinogenic and many accumulate in plant and animal tissues. Common organic contaminants associated with urban/highway stormwater runoff include oil and grease, aliphatic and aromatic hydrocarbons, pesticides, anti-sapstain chemicals for lumber, plasticizers and polychlorinated biphenyls (PCBs) (Cole et al., 1984; Kobriger and Geinepolos, 1984; Swain, 1983; Swain and Walton, 1991; Lawson et al, 1985). According to Cole et al. (1984), some anti-sapstain chemicals can persist by adsorption to sediments i f sheltered from photolysis, whereas plasticizers can accumulate in sediments and bioaccumulate in aquatic organisms. Similarly, Schueler (1987) found oil and grease to contain a variety of 17 hydrocarbons, with the aliphatic hydrocarbons being less toxic than the more accumulative polynuclear aromatic hydrocarbons (PAHs). Oil and grease is the only organic contaminant measured in this study. Overall, urban stormwater runoff, including highway discharge, can be detrimental to receiving waters. The extent of the runoff-associated hazard depends on such variables as average daily traffic, local climate, surrounding land use activities and highway maintenance practices, which will be explored in the next chapter. However, effective solutions can only start with correct contaminant identification and the appropriate treatment program implementation. 1.4 Purpose of this Study With the growth of environmental awareness, maintenance of basic water quality standards has become an important issue in water resource management. The population of the G.V.R.D. is growing very rapidly, with the number of automobiles growing even more rapidly. Research to date has linked certain contaminants with highway runoff. As a result, with the number of automobiles increasing with the population growth in the G.V.R.D., there is a growing interest in the pollutants which highways contribute to the surrounding receiving waters and how any adverse effects could be mitigated. According to Wong (1991), a drainage system designed for fast removal of highway runoff without any form of treatment is viewed as undesirable today. In the past, the B.C. Ministry of Transportation and Highway has used some criteria in the design of grassed drainage ditches, grassed channels and other forms of treatment to minimize highway pollution contributions. There is little available data on pollutants generated by B.C. highways or any other highways in Canada. Similarly, there are no Canadian data available on the effectiveness of currently used highway treatment facilities in terms of their pollution reduction potentials. The objectives of this study were first to conduct literature review on the whole 18 problem of contaminants contributed to the environment by runoff from highways, to collect and analyze data on selected pollutants such as T.S.S., Cr, N i , Cd, Fe, Cu, Zn, Pb, Ca, Mn and oil/grease at the Burnaby and Richmond sites, and to evaluate the extent of the problems and applicable solutions. Some detailed objectives were: 1. To evaluate pollutant concentration variation with seasons; 2. To investigate the effectiveness of grassed drainage ditches in pollution reduction; 3. To evaluate different trace metal removal mechanisms-soil/sediment adsorption versus biological plant uptake; 4. To contribute information required for developing best management practices by the B.C. Ministry of Transportation and Highways for the G.V.R.D.; and 5. To recommend planning approaches, design criteria, implementation and potential positive effects on highway runoff water quality, based on the data. In summary, Chapter 1 introduces the topic of highway stormwater runoff quality in the Greater Vancouver Regional District and states the objectives of the study. Chapter 2, Literature Review, is going to look at different types of contaminants, their sources, transportation or migration patterns, causes of changes in concentration/loading, impact on water quality, relevant models and stormwater management measures. Chapter 3 (Materials and Methods) will identify and describe the two sites, as well as the experimental equipment and procedures, while Chapter 4, Results and Discussions, will evaluate, analyze and compare the experimental data to other published stormwater studies. Chapter 5, Conclusion and Recommendations, will summarize the obtained research data and make recommendations on the use of the data for the future. The bibliography section lists all the references cited in this thesis and the appendix shows some of the equations used in the calculations and general data analysis. 19 Chapter 2 LITERATURE REVIEW Over the last decade, extensive research has been done on the impact of point and non-point pollution on receiving waters. Washington State Department of Ecology (WSDOE) (1992) defined non-point source pollution as pollutants that enter any waters from dispersed land-based or water-based activities including, but not limited to, atmospheric deposition, surface water runoff from agricultural lands, urban areas or forest lands, sub-surface or underground sources or discharges from boats or marine vessels. However, the U.S. EPA (1997) simply defined non-point source pollution as the pollution of waters caused by rainfall or snowmelt moving over and through the ground. Highway runoff is a typical non-point source. Point sources are easier to monitor and control and how effectively point source control programs translate into improvement in water quality can be measured. For example, in the U.S.A., about 758,000 stream miles and 16.3 million lake acres across the country were subjected to water quality investigations for a period of 10 years from 1972 to 1982. The study found that about 47,000 stream miles and 390,000 lake acres improved in quality due to sizeable investment in point source control programs (ASIPWCA, 1986). This returned water bodies once deemed unsuitable for people and animals into habitable environments. Non-point source control programs and how they translate into improvements in water quality are harder to assess. This is a result of the dispersed and variable nature of non-point source pollution. Even though considerable research has been done in this field, a consistent standard information base describing the extent of non-point source pollution, such as highway runoff, and the effects of existing control programs on water quality management has not been fully developed. In the case of highway runoff, the actual quantity and composition of the 20 pollutants deposited on a specific highway depend on such* variable factors as land use, geographical locale, weather, season, traffic volume, highway surface composition, and highway maintenance (McKenzie and Irwin, 1983; Kobriger and Geinepolos, 1984). Mar et al. (1981) found that highway runoff in Washington State was less contaminated than general urban runoff and that in most cases highway runoff did not create a significant impact on the quality of receiving water. However, published results from different studies indicated potentially short and long term effects of highway runoff on receiving water quality. Short term effects, such as acute toxicity to aquatic biota, could result from high pollutant concentrations during intense rainstorms. Long term effects, on the other hand, resulted from cumulative pollutant loadings over a period of time (Horner and Mar, 1985; McKenzie and Irwin, 1983). Although research to date has shown that both the short term (pollutant mass per unit water volume) and the long term (cumulative pollutant loadings) effects can significantly affect aquatic organisms under test conditions, there are still many uncertainties about the nature of the pollutants and the different conditions in which they exist in natural waters. In fact, there are certain pollutants, mostly metals, that at trace levels are essential to life, whereas higher levels can be detrimental. Iron, zinc and copper are among the most important essential elements required by the human body in the sense that these metals are required for the maintenance of life and cannot be wholly substituted with other elements. In fact, any serious deficiency will result in high cell malfunction or even death. This is also applicable to aquatic organisms, but the conditions under which various concentrations are helpful or harmful are complex and not completely understood. For example, these metal elements show different levels of toxicity to daphnia. Laboratory bioassays of daphnia showed a reduced Cu, Fe, Zn and Pb toxicity regulated by pH and the 21 presence of suspended solids. On the other hand, there was an increase in copper and zinc toxicity to daphnia when lead was also present in urban stormwater runoff (Kushner, 1993; Hall and Anderson, 1988). There are still many uncertainties about the way in which different pollutants affect different organisms and to what degree and under what conditions. An important key to reducing some of these uncertainties lies in looking at different highway runoff pollutants and understanding their chemistry. 2.1 Highway Surface Pollutants and Their Chemistry An overview of highway generated pollutants and their chemistry is given in this section. According to Gupta et al. (1981a/1981b), there are eight general categories of surface pollutants commonly associated with highway runoff. The categories with examples are: • Particulate, e.g. dust and dirt, stones, sand, gravel, grain, glasses, plastics, metals and fine residue. • Heavy metals, e.g. lead, zinc, iron, copper, nickel, chromium, cadmium and mercury. • PCB, pesticides, e.g. chlorinated hydrocarbons and organo-phosphorus compounds. • Pathogenic bacteria with indicator bacteria-like coliform. • De-icing agents or inorganic salts, e.g. CaCl 2 , NaCl, S0 4 , Br. • Organic matter, e.g. vegetation, dust and dirt, humus, roadway accumulations, fuels, oils and grease, and PAHs. • Nutrients, e.g. nitrogen and phosphorus. • Other, e.g. asbestos, rubber and special compounds. 22 2.1.1 Particulate Particulate in highway runoff, rivers and lakes are important materials as they are the main carriers for heavy metals. According to Kobriger and Geinepolos (1984), the solids are the main pollutant carriers. In their research studies in four U.S.A. cities, they found that the total solid content exhibits the highest degree of association with total pollutant quantity. Further analysis of other particle size categories revealed that the suspended solid content also showed close correlation with pollutant quantity. They also found that highway sites with higher average daily traffic (ADT) values gave higher concentrations and loadings of solids. Milwaukee, Wisconsin site with an ADT of 116,000 vehicles had mean concentrations of 381 mg/L and 157 mg/L for total and suspended solids respectively for non-winter conditions compared to the Efland, North Carolina site with an ADT of 25,000 vehicles which had mean solid concentrations of 125 mg/L and 23 mg/L for total and suspended solids respectively for the same time period (Kobriger and Geinepolos, 1984). However, the definitive impacts of A D T on particulate matter or solids are still questionable, since some research studies by Asplund et al. (1980), McKenzie and Irwin (1983), and Kerri et al. (1985) suggest that traffic during dry periods tends to remove pollutants from highway surfaces through vehicular generated turbulence. Also there was a distinct difference in both total and suspended solid concentrations for all the sites between winter and non-winter conditions. The generally higher winter concentrations of solids were attributed to the use of de-icing agents except for Sacramento, California where there were no noticeable winter conditions (Kobriger and Geinepolos, 1984). Similar observations were made by Hedley and Lockley (1975) and Colwill et al. (1984) for U.K. urban motorways; Lygren et al. (1984) from their work in Norway, and Stotz (1987) from his work on German highways. There are other variables such as rainfall intensity, the total volume of runoff discharged, highway maintenance practices, highway configuration and surrounding land use activities that 23 also influence the amounts of particulate material and associated contaminants removed from the road surface by each storm event. The impact of surrounding land use on particulate concentrations was reported to range from 60 mg/L for lightly used areas to greater than 5,000 mg/L for heavily used areas according to the Bedfordshire, U.K. study (Colwill et al. 1984). Surrounding land use activities were also found to exert a noticeable influence in studies conducted by Gupta et al. (1981a/1981b) and, later on, by Kobriger and Geinepolos (1984) in four U.S.A. cities. In Washington State, Asplund et al. (1982) found dust fall from urban areas and surrounding lands adjacent to the highway as major sources of pollution. They found deposition rates to range from 1.1 to 3.3 kg per ha per day in industrial areas and from 0.6 to 1.7 kg per ha per day in rural areas. Accordingly, surrounding land use activities can significantly influence the amounts of particulate material deposited on adjacent highway surfaces. The • impact of surrounding land use can be controlled to some extent by vegetating the lands next to highways to minimize wind blown transportation of particulate or solid materials. Other advantages of vegetating lands next to highways, such as erosion control and water quality mitigation, will be explored in subsequent sections. Highway maintenance practices affect the amounts of particulate material on road surfaces, but data on the effectiveness of highway maintenance practices in pollution reduction have been inconclusive. In 1974, Sartor et al. undertook a major study involving twelve cities across the eastern U.S.A. from Bucyrus, Illinois to Baltimore, Maryland. They found sweeping practices current at the time to be ineffective in the removal of dust and dirt fractions from streets. The removal efficiency of litter and debris ranged from 95 to 100 percent compared to 50 percent for the dirt and dust fraction. As a result, their general conclusion was that the current street sweeping practices were essentially for aesthetic purposes rather than for street/highway surface contaminants removal. The finding of Sartor et al. (1974) was validated by later studies 24 by the U.S. EPA in 1983 in which they found street cleaning to have virtually no effect on runoff pollutant loadings, since most of the pollutants were contained within the fine particles. Instead most of the pollution laden fine particles are removed by wind action or by runoff. Stotz (1987), in his experiments on German highways, concluded that only 5 to 20 percent of the total pollution from vehicular traffic was discharged with runoff with the remaining 80 percent of the total pollution removed by wind action and street cleaning. Asplund et al. (1982) estimated the pollutant removal range of street cleaning operations to be between 25 to 78 percent of the highway pollutant mass. But since the particle removal effectiveness by street cleaning is size dependent, the larger particles have higher removal probability (Shaheen, 1975; Howell, 1978; Gupta et al., 1978; Wilbur and Hunter, 1979). These pollutants in particulate are no threat to water quality and aquatic environment if they remain in the highway curb lanes and ditches. For example, a study conducted by Kerri et al. (1985) in Redondo Beach, Walnut Creek, and Sacramento, California, found that California highways did not produce large amounts of pollutants during storm runoff events. Consequently, the usually high cost treatment facilities required to meet water quality objectives were not needed in this case. The low pollutant load in the runoff was attributed to traffic generated turbulence which continuously sweeps the travelled lanes and highway shoulders. Also, this low pollutant loading in Californian highway runoff could be the result of insufficient rainfall to wash off the pollutant mass from the highway surfaces, a factor not considered by Kerri et al. (1985) in their study. The impact of vehicle-generated turbulence and rainfall intensity on pollutant removal from highway surfaces will be evaluated in a later section. Particulate Chemistry. Although the concentration of particulate in some highway runoff may be low, the chemical impact of particulate can be of major concern. Highway particulate or solid chemistry is very complicated since the material is an aggregation of pollutants and dirt/dust 25 particles. As discussed above, it has been found from past research that solids are the main carrier of pollutants in highway runoff. The total solid content exhibits the highest degree of association with pollutants, with suspended solid content also showing a close correlation. Particulate or solid materials contain contaminants ranging from chemical oxygen demand (COD), oil/grease and polynuclear aromatic hydrocarbons to metals. For example, lead is a metal that is predominantly associated with soils while the highest oil levels are found on sediments in the size range from 200 //m to 400 jum (Colwill et al., 1984). In order to understand particulate material chemistry better, individual pollutants such as metals or oil and their respective effects on water quality have to be studied separately. According to Colwill et al. (1984), the direct impact of sediments on water quality is probably not as significant as their capacity for adsorbing, transporting and releasing contaminants, particularly metals. Chemical Kpe.ciatinn in Sediments. It is more difficult to determine the speciation of the heavy elements in sediments than in solutions. Generally, the heavy metals can be associated with the sediments in a number of ways, the more important processes being chemisorption of heavy metals on to clays or Mn/Fe hydrous oxides, precipitation of discrete heavy metal compounds, and flocculation/complexation of heavy elements associated with reactive organic materials (Ferguson, 1990). Of the above mentioned processes, only chemisorption will be discussed briefly. Chemisorption involves the sorption of heavy metals onto clays. This process is controlled by the number of free-sorption sites on the clay surface which depends on the free or broken bonding positions, as well as on the proportion of atoms replaced with others of different valencies in the clays. Other factors that influence the sorption process are the pH, the nature of the heavy metal species such as their charge and hydration, and the clay type. The clays with expanding ability and large surface area, such as the montmorillonite, have greater capacity to 26 accumulate heavy metals than kaolinite (Laxen, 1983; Ferguson, 1990). Another major process for incorporating heavy metals into sediments is the chemisorption and eventual co-precipitation of heavy metals with the hydrous oxides of manganese and iron. Other significant processes involved in chemical speciation in sediments are: • Incorporation into detrital minerals which involves the embodiment of heavy metals into detrital minerals and other minerals such as clay. This may involve metal ion replacement such as lead replacing calcium or potassium. • Physical sorption, which may occur by electrostatic attraction, involves an electric double layer set up between a charged clay surface and the sorbed ionic or polar species. Consequently, there may be an alteration in the charged surface which subsequently influences the process. For example, the impact of zero point charge, which is the pH at which the surface has zero charge, is well noted by Ferguson (1990). The dispersion of particulate material from the charged surfaces on solids is the dominant process when the pH is away from the zero point charge (ZPC). As the ZPC is reached, however, there is a reduction in the repulsive forces between the like-charged surfaces, resulting in flocculation and coagulation. • Association with organic material is another prominent process in the speciation of the heavy elements in sediments. Organic materials can do this in two processes: - Solubilization of metal species by complexing the metal ions, and - taking metal ions out of solution and contributing them to the sediment. For example, the insoluble humic substances (complex polyphenols) contain organic groups such as polysaccharides and proteins which co-ordinate to heavy metal ions readily (Ferguson, 1990). These humic 27 substances quickly become incorporated into the sediment because of their size and insolubility. 2.1.2 Heavy Metals Heavy metals in highway runoff are always associated with the fine particles. Research studies to date have shown that the fine silt and clay fractions carry more pollutants and contribute the biggest share to the total runoff pollutants load. These fine particles are easily mobilized by low intensity rainfall events, thereby contributing to the suspended solid concentrations in highway runoff. Concentrations of these suspended solids increase with high flow rates as road surface scouring takes place. As a result, the dominant factor in determining the concentration of suspended solids and the load discharged is the overall nature/type of the rainfall and runoff event. The impacts of rainfall and runoff events on pollutant transport will be further explored in a later section. The effects of traffic, land use activities, and highway maintenance practices will be evaluated in this section, as well as the chemistry of these heavy metals. Just like the effect of solids, the impacts of average daily traffic (ADT) on metals are still uncertain. The initial research report by Sartor et al. (1974) indicated that the amount of contaminant material at a given test site depended directly on the both antecedent dry days and traffic volume. They found the pollutant levels to increase with traffic volume. Their findings were supported later by the work of Kobriger and Geinepolos (1984 and 1984a). Again, just like solids, the highest concentrations of metals were found in the two sites with high traffic volumes. They found that, generally, the concentrations of lead and zinc were more heavily influenced by vehicular traffic than other metals such as iron, chromium, copper, cadmium, nickel and manganese. Milwaukee, Wisconsin with an ADT of 116,000 vehicles had lead and zinc 28 concentrations of 0.6 and 0.36 mg/L for non-winter conditions while Sacramento, California had lead and zinc concentrations of 0.4 and 0.26 mg/L respectively for the same period of time with an ADT of 85,000 vehicles. Efland, North Carolina with an ADT of only 25,000 vehicles, on the other hand, generated lead and zinc concentrations of 0.02 and 0.05 mg/L for the same time period (Kobriger and Geinepolos, 1984). One of the reasons for the generally higher lead concentrations could be attributed to the fact that lead is approximately eight times more abundant in materials deposited on highways than other metals. Another reason could be that lead from automotive exhaust occurs in both particulate fraction and an organic vapour phase, with the majority occurring in particulate inorganic fraction (Brief, 1962; Shaheen, 1975; Laxen and Harrison, 1977). In Canada, where Tetraethyl lead (TEL) has been banned in gasoline, total annual automotive lead emissions has gone from 14,360 tonnes at its peak in 1973 to about 6,500 tonnes in 1983 (Poon, 1988; McCallum, 1995). McCallum (1995) observed a considerable Pb decrease in stream, lake and street sediments in the Brunette River Watershed over the last 20 years, believed to be due to the banning of leaded gasoline. According to McCallum (1995), during the same time period, Zn, Cu, Mn and Hg increased in stream sediments by 45, 81, 130, and 290 percent respectively. Zinc, copper, manganese, mercury and iron concentrations are easily observable, while small or undetectable loadings of chromium, cadmium and nickel have been observed at most U.S.A. sites (Kobriger and Geinepolos, 1984). This lack of a conclusive pattern has led other researchers to question the effect of ADT on heavy metal pollution. Wong (1991) re-evaluated the data collected by Kobriger and Geinepolos (1984/1984a) and concluded that, with the exception of lead and zinc, there was little or no support for the initial generalization that high ADT highways generated more heavy metal pollution than the ones with a low ADT count. Prior to Wong's re-evaluation in 1991, Asplund et al. (1980), Gupta (1981), and McKenzie and Irwin (1983) have all questioned the impact of 29 traffic volume on heavy metal pollution with their results ranging from little or no significant impact to inconclusive effects. The current hypothesis tends to suggest that traffic during the dry periods appears to remove pollutants from highway surfaces rather than contribute to pollution. If this is the case, ADT may not have any significant impact on pollution generation or pollutant loading prediction during the dry weather period. This was found to be the case in a study conducted by Asplund et al. (1980) in Washington State. They found that traffic during dry periods is not significant in runoff concentration or loading generation or prediction. Instead, they found the amount of vehicles during storms (VDS) to be more important in estimating or predicting highway runoff concentrations or loadings. This issue will be evaluated in subsequent sections. While there may not be any definitive correlation between A D T and heavy metal loadings, there seems to be some noticeable difference between the non-winter and winter heavy metal loadings. Even though Colwill et al. (1984) did not find any statistically significant difference in lead, zinc and cadmium levels between winter and summer mean runoff events in their research at Bedfordshire, U.K. , many other researchers did. Kobriger and Geinepolos (1984) found that the median lead, zinc and iron concentrations were higher in the winter months, whereas chromium, cadmium, nickel and copper did not show any seasonal variations. Manganese was not measured in any of these studies and, as a result, no data was available on it. The findings of Hedley and Lockley (1975) on U.K. urban motorways and Lygren et al. (1984) on Norwegian highways supported the findings of Kobriger and Geinepolos (1984). The study by Hedley and Lockley (1975) on Ashton, Birmingham highways provided detailed data on monthly pollutant concentration and loading for metals such as zinc, lead, copper and iron. Just like the U.S.A. study of Kobriger and Geinepolos (1984), the concentrations of these metals were considerably higher in the winter months than in the rest of the year. There are no data available 30 on nickel, chromium, cadmium and manganese. Thus, seasonal pollutant concentration and loading seem to depend on the geographical location of the site. For example, there was no observable winter/non-winter variation with the data obtained in Sacramento, California by Kerri et al. (1985) or Orange County and Miami, Florida by Yousef et al. (1985), and McKenzie and Irwin (1983) respectively. The reasons given for higher winter pollutant concentrations and loadings where they occur are increased surface loading due to lack of sweeping, less atmospheric wind blow-off, more stop-and-go traffic due to dangerous winter conditions, reduced runoff caused by freezing winter roadway conditions, increased automobile body rusting due to winter de-icing agents, lack of regular highway maintenance to stop surface deterioration (Kobriger and Geinepolos, 1984; Lorant, 1992), and the use of studded tires which generated an estimated 20 to 50 g/km/vehicle of asphalt wear on Norwegian roadways (Lygren et al., 1984). There has not been any known seasonal variation in the impact of surrounding land use on heavy metal pollutant loadings/concentrations, even though surrounding land use practices have been known to influence the amount and type of other pollutants. In the study by Kobriger and Geinepolos (1984), the unusually high iron content in runoff from highways at the Harrisburg, Pennsylvania site was attributed to the local soil conditions. The iron concentration in the highway runoff at the Harrisburg site was found to be 16.1 mg/L compared to 4.92 mg/L for Milwaukee, 3.95 mg/L for Sacramento, and 2.13 mg/L for Efland for non-winter conditions. The slightly elevated concentrations of iron associated with these sites during the winter period could be due to the reasons given in the previous paragraph for higher winter pollutant concentrations and loadings rather than the impacts of the surrounding land use. The Harrisburg site showed the first clear indication of how local soil conditions can influence highway surface runoff. Other soil studies, especially with lead, by Motto et al. (1970), Zimdhal (1972), Getz et al. (1975), Scanlon (1977), and Gupta and Kobriger (1980), found the highest concentrations on 31 the soil immediately adjacent to the highway. They also found the concentrations of lead, as well as cadmium, nickel and zinc, to decrease rapidly with depth and distance from the highway. Scanlon (1977) found that metal concentrations decreased to background levels within 48 m from the highway on Virginia highways compared to 50 m on Illinois highways (Getz et al., 1975), 30 m on Denver highways (Zimdhal, 1972), and between 30 - 35 m on Milwaukee highways (Gupta and Kobriger, 1980). This is automobile-generated heavy metal pollution of local soil conditions as opposed to land use generated heavy metal pollution, as indicated at the Harrisburg site. These automobile-generated heavy metals are most likely to add to the metal background levels of the surrounding land, thereby increasing the heavy metal levels on land adjacent to the highways. Depending on the wind direction and the extent of vegetation on the land adjacent to the highway, some of the metals/particulate laden with metals can be blown back onto the highway surfaces. In this case, the cyclic effect will be characterized by constant removal of heavy metal from the road surfaces, deposition of the metals on land adjacent to the highway, uplifting of the particulate laden with metals by winds and re-deposition of the heavy metals onto the highway surfaces. However, there must be a cumulative effect, in which case metals released on the highway become trapped in soil sediment and vegetation adjacent to the highway. Laxen and Harrison (1977) reported that, generally, lead in the soil is effectively immobilized and confined to the top 15 cm depending on the distance from the roadway, the top 15.2 cm according to Motto et al. (1970) at 7.6 m away from New Jersey highways, and the top 10 cm according to Getz et al. (1975). The immobilization of lead by the soil, according to Zimdhal (1972) and Hassett (1974), depends directly on the soil cation exchange capacity (CEC) and it is inversely related to soil pH. As a result, there is a tendency for heavy metals, especially lead, to accumulate on lands adjacent to highways. Consequently, any form of transportation and re-deposition of these particulate materials laden with metals from land adjacent to highways 32 back onto highway surfaces will result in increased metal concentrations in highway runoff. However, with proper highway maintenance practices, some of these metals could be removed from highway surfaces. Since most of the heavy metals are associated with the dirt and dust particles found on the highway, highway maintenance practices have the same effects on both heavy metals and dirt and dust particles. As a result, the impact of highway maintenance practices on particulate fine matter evaluated in the previous section are also applicable to heavy metal pollution. The general consensus among researchers is that current highway maintenance operations such as street sweeping and flushing are inadequate in removing fine particles and associated heavy metals from highway surfaces. However, the designs of vehicles involved in the street cleaning operations have improved over the years. New studies are required to evaluate the effectiveness of these current vehicles in removing fine particles and their associated heavy metals. Meanwhile, special consideration should be given to heavy metal chemistry in order to better understand metal interaction with the environment. Heaw Metal Chemistry. There are two arbitrary categories of heavy elements in water: the filterable (<45//m), called the dissolved, and the suspended matter. The suspended matter, with time, will eventually undergo sedimentation to produce sediments. Highway runoff contains some metals such as lead (Pb), zinc (Zn), iron (Fe), cadmium (Cd), chromium (Cr), nickel (Ni) and copper (Cu) at much higher concentrations than background levels. Like other metals, these metals in highway runoff undergo physical, chemical and biological transformations on reaching the ecosystem. They may be adsorbed onto clay particles, picked up by plants and animals, or may remain in solution. In fact, the biological activity of both essential and toxic metals depends to a large extent on the ability of their ions to combine with other atoms and molecules as well as on their speciation in aqueous solution. As a 33 result, water with high total metal concentration may be less toxic than water with lower concentration of the ionic metal forms. For example, the biotoxicity of organically bound copper is low compared to that of ionic copper which is far more toxic to aquatic organisms (Yousef et al., 1985a; Ferguson, 1990; Kushner, 1993). From this example, it can be deduced that the toxicity of metal complexes tends to decrease with their increasing stability. Therefore, it is important to understand speciation of heavy metals (which is the transformation of a metal to different chemical and physical species) in aquatic systems as well as the bioavailability of the metal (Kushner, 1993). The relative importance of the factors affecting chemical speciation requires consideration of topics such as kinetics, thermodynamics, chemical equilibrium and stability constant data (Laxen, 1983; Morgan, 1987). These topics are beyond the scope of this research project. However, some of the factors such as pH, complexing agents and adsorption/desorption onto particulate matter may be used to explain trends in the observed data where applicable. For example, Yousef et al. (1985a), in their research at the Maitland Interchange and Interstate 4 in Florida, found that the presence of organic substances and sediments in natural waters played a key role in the detoxification of metals associated with highway runoff. However, the capacity of bottom sediments to retain most of these metals depends on the aerobic and redox conditions. Also, B.C. Environment (1992) stated that the presence of organic matter increases the cation exchange capacity of sediments, thereby increasing the amount of ion-exchangeable metals on the sediment particles. Depending on the metal and the species, metal toxicity often increases with pH. This was observed to be true in the research by Kushner (1993), in which he found the toxicity of copper (II) and cadmium (II) ions to bacterial and fungal species to increase with pH in the range from 5 to 9. On the other hand, nickel (II) ion was found to be more toxic to a number of micro-34 organisms with pH less than 5.5 than at higher pH. Babich and Stotzky (1985) explained that this low toxicity at high pH can be attributed to nickel (II) ions effectively binding to sites on nitrogenous organic compounds at high pH. Consequently, nickel (II) ions bound to nitrogenous compounds at higher pH would not be available to exert any toxic effects on micro-organisms. Zinc (II) ion toxicity may increase or decrease with increasing pH, depending on the aquatic micro-organism (Kushner, 1993). 2.1.3 Organic Contaminants Organic contaminants commonly found in urban and highway runoff include oil and grease, aliphatic and aromatic hydrocarbons, plasticizers (mainly phthalate esters), polychlorinated biphenyls (PCBs), pesticides, polynuclear aromatic hydrocarbon (PAHs) and anti-sapstain chemicals for lumber (i.e. pentachlorophenol, PCP). Many research studies have been conducted in the U.S.A. concerning the impacts of organic contaminants in general urban runoff, but few studies have been done on the impact of organic pollutants from highway runoff. However, based on the available research data, some of the organic compounds found in urban/highway runoff are known to be toxic to human and aquatic life (B.C. Environment, 1992). In U.S.A. studies, the most common organic contaminant found in urban runoff was the organic pesticide alpha-hexachlorocyclohexane (alpha-BHC). Due to limited highway research data, it is difficult to compare organic contamination in highway runoff to that of general urban runoff. Intuitively, lower levels of some organic contaminants such as pesticides and polychlorinated biphenyls (PCBs) would be expected in highway runoff than in general urban runoff. For example, Kobriger and Geinepolos (1984) found low levels of PCBs in highway runoff at all sites except for Milwaukee. The higher than normal concentration at the Milwaukee site was attributed to a power transformer station downwind of the site. In British Columbia, 35 Canada, low levels of PCBs were detected in highway associated samples of soil, vegetation, precipitation, surface dust and dirt and runoff, although higher levels were found in street surface sediment in the Brunette River basin (B.C. Environment, 1992) and in the residential area of Burnaby (Swain and Walton, 1991). Similarly, Larkin (1995) found hydrocarbon pollution to be prevalent in the Brunette River watershed. Larkin's analysis of total petroleum hydrocarbon (TPH) found elevated concentrations in lake cores, stream bed sediments, stormwater suspended solids and street surface sediments. Shopping mall parking lots and other parking lots have considerably higher TPH concentrations. Due to limited data on highway organic contaminants, as well as low levels of pesticide and PCBs concentrations, attention in this section is focused on oil and grease. Data on oil and grease is limited but some inferences can be made from the available data. According to Kobriger and Geinepolos (1984), there is no positive correlation between oil and grease concentrations and the average daily traffic (ADT). For example, Sacramento with only 85,000 vehicles per day, has mean oil and grease concentration of 11 mg/L, whereas Milwaukee, with 116,000 ADT, has 7 mg/L. Kobriger and Geinepolos (1984) did not give any explanation for the erratic relationship between ADT and oil and grease concentrations. Colwill et al. (1984) reported total oil levels ranging from 5 to 30 mg/L in their research study at Bedfordshire, U .K. and made an attempt to correlate the oil and grease concentrations with ADT. They found that at low initial flows the oil fraction was dependent on the antecedent dry period-traffic flow. Just like metals, the oil and grease content of the sediments varies with the particle size with the highest oil levels found in the sediment size range of 200 to 400 micrometers Cum). Colwill et al. (1984) estimated that, on average, as much as 70 percent of the oil is associated with sediments and that the percentage is higher after long dry periods. This observation suggests that antecedent dry periods have more impact on oil and grease concentrations than 36 ADT. Other factors that influence oil and grease concentrations are road surface temperature and the time of year. Since there is a positive correlation between oil/grease concentration and solid material, any seasonal variation that increases the suspended solids levels should also increase the level of oil and grease. Even though Colwill et al. (1984) reported total oil levels ranging between 5 and 30 mg/L, extremely high values such as 100 mg/L were recorded during high intensity rainfall events with their associated high suspended solids content. Kobriger and Geinepolos (1984) reported higher mean oil and grease concentrations for Milwaukee during the winter months (with their associated high rainfall events) than in non-winter months. They found the mean oil and grease concentration at the Milwaukee site to be 13 mg/L for the winter period compared to 5 mg/L for the non-winter. Both the works of Colwill et al. (1984) and Kobriger and Geinepolos (1984) are evidence that high intensity rainfall events with high suspended solids content, especially during the winter months, influence oil and grease concentrations. Swain (1983) reported a range of 3 to 27 mg/L in residential urban runoff in Vancouver and speculated that the range might be approximately the same with concentrations from the highways. This range, reported by Swain (1983), tends to agree with the levels recorded by Colwill et al. (1984) and Kobriger and Geinepolos (1984). The effect of surrounding land use activities on oil and grease levels has not been given much attention by other studies conducted to date, and it will not be part of the research objectives of this project. However, an attempt will be made to use the observations made about surrounding land use activities to explain the data obtained on oil and grease concentrations. Swain (1983a) stated that vehicles may not be parked along the highways for sufficient time for oil and grease leakage to occur as may occur with roads beside residential dwellings. However, 37 highways with high ADT are expected to have some oil and grease leaking from the passing vehicles. Highway maintenance, when effective, can be used to reduce oil and grease levels. Recall that, as described in the previous paragraphs, Colwill et al. (1984) estimated that a high percentage of the oil and grease fractions are associated with fine particles. Recall, also, from the particulate material section, that Howell (1978), Wilbur and Hunter (1979), and Asplund et al. (1982) all considered current highway maintenance practices, especially street cleaning, to be ineffective in removing fine particulate materials. As a result, the inefficiency involved in fine particulate material removal is also applicable to highway oil and grease pollution reduction. 2.1.4 Pathogenic Bacteria (Indicators) The Association of State and Interstate Water Pollution Control Administrators (ASIWPCA, 1986) defined pathogens as disease-causing organisms which include bacteria and viruses. Usually the presence of these organisms in water is detected by measuring their associated indicator pathogens. For example, fecal coliforms (FC), total coliforms (TC), and fecal streptococci (FS) are frequently used as indicator bacteria in testing for the presence or absence of pathogenic bacteria. The presence of fecal coliforms suggests contamination originating from human feces, whereas fecal streptococci appearance is an indication of animal feces contamination (BC Research Corporation, 1991). Research to date has indicated that there are high levels of these indicator micro-organisms in general urban runoff. Athayde et al. (1983) found total coliforms in undiluted samples at every site exceeded U.S.A. EPA criteria when it rained. Dutka and Rybakowski (1978) concluded that a significant portion of intermittent microbial contamination to receiving waters was from urban runoff, a conclusion that was supported by high levels of pathogenic 38 bacteria found in urban runoff by Field and Pitt (1990). Research in Ontario, Canada found microbial concentrations in urban runoff to be similar to those found in dilute sewage (OMOE, 1980). Dutka and Rybakowski (1978) found higher densities of indicator bacteria such as 190,000 fecal coliforms per 100 ml and 1,400,000 fecal streptococci per 100 ml in urban runoff in Ontario; Waller (1972) found fecal streptococci concentration of 1,400,000 per 100 ml in urban roof runoff samples in Halifax, Nova Scotia compared to the most probable number (MPN) fecal coliforms range counts from less than 20 to greater than 24,000 per 100 ml in runoff from a residential catchment (Swain, 1983), and from less than 200 to 54,000 per 100 ml in runoff from an industrial catchment in British Columbia (Lawson et al, 1985). These bacterial concentration numbers from general urban runoff are considerably higher than the indicator bacterial levels found in the limited data available on highway stormwater runoff. Kobriger and Geinepolos (1984) reported ranges of fecal coliform and streptococci counts from 250 to 660 and 5,300 to 127,000 per 100 ml respectively for Harrisburg, 350 to 4,600 and 2,400 to 11,000 per 100 ml respectively for Sacramento, 570 to 730 and 11,000 to 16,000 per 100 ml respectively for Efland, and 1 to 100,000 and 40 to 100,000 per 100 ml respectively for Milwaukee. Based on this data, it appears that the bacterial counts for fecal coliforms are generally lower than the counts for fecal streptococci. As a result, apart from using total and fecal coliforms and fecal streptococci to indicate bacterial contamination, fecal coliforms and fecal streptococci (FC/FS) ratios can also be used. Low FC/FS ratio (with numbers less than 0.7) is usually an indication of animal rather than human sources as the origin of bacterial contamination (Gannon and Busse, 1989). However, due to different die-off rates for fecal coliform and fecal streptococci, the FC/FS ratios may be deceiving (Dutka and Rybakowski, 1978). Therefore, the use of fecal coliforms as an indication of health hazards may not be as accurate as initially believed (Athayde et al., 1983; Field and Pitt, 1990); hence U.S. 39 EPA has suggested the use of E. Coli or enterococci in evaluating freshwater quality (Gannon and Busse, 1989). These indicator micro-organisms can remain either in the water column or can be adsorbed to the sediments. Under favourable conditions, such as abundance of nutrients and shelter from the sunlight (Marsalek, 1986), these organisms tend to concentrate at the sediment-water interface. Re-suspension of these organisms is possible with a certain degree of turbulence (Schillinger and Gannon, 1985). In a non-water environment like dirt and dust, fecal coliforms, according to Kobriger and Geinepolos (1984), can remain viable for at least seven weeks, although the conditions that facilitated their survival for such an extended period of time were not given. With highway runoff there is no clear correlation between the ADT and the indicator bacterial counts, especially with regard to fecal streptococci. However, there was a weak correlation between ADT and fecal coliform counts based on the data of Kobriger and Geinepolos (1984). Milwaukee, with an ADT of 116,000 vehicles, had a fecal coliform range count of 1 to 100,000 per 100 ml, followed by Sacramento with an A D T of 85,000 vehicles and fecal coliform count of 350 to 4,600 per 100 ml, Harrisburg with 27,000 vehicles and 250 to 660 per 100 ml and Efland with 25,000 vehicles and fecal coliform range count of 570 to 730 per 100 ml. Based on this weak correlation and limited data availability concerning highway runoff indicator bacterial contamination, it is safe to say that the effect of ADT on indicator micro-organisms highway pollution contribution is inconclusive. Kobriger and Geinepolos (1984) found a more conclusive result between bacterial counts and the time of year. Highway runoff bacterial counts were found to be generally higher in the fall than in the spring or at any other time of year. These higher bacterial counts in the fall 40 samples could be a result of lower ambient temperature, better moisture conditions (to avoid desiccation), and lack of regular maintenance during the fall and winter months. Even though the data on the impacts of surrounding land use activities may also be limited, surrounding land use may have more of an impact on the level of indicator bacteria than ADT. Kobriger and Geinepolos (1984) listed trucks carrying livestock and stockyard waste as one possible source of indicator micro-organisms on highway surfaces. If this assumption is right, highways located in a predominantly farming area should have higher levels of bacterial contamination. Although the impact of surrounding land use activities on indicator bacteria is not part of the objectives of this research, future evaluation on this topic, as well as the impact of maintenance, is highly recommended. The extent to which regular maintenance, such as sweeping, affects bacterial counts in highway runoff is unknown. However, since there is some possible adsorption to suspended solids and sediments by these micro-organisms, the same inconclusiveness noticed earlier in the effectiveness of highway maintenance on particulate pollution reduction may also be applicable in this case. For example, where there is an adsorption of micro-organisms to sediments/dirt and dust fraction, the effectiveness of street sweeping to remove this bacterial-adsorbed sediment may depend on the size of the particle. Research to date has shown that the larger the particle size, the greater the probability of being removed through regular maintenance (Shaheen, 1975; Howell, 1978; Gupta et al., 1978; Wilbur and Hunter, 1979). But like the metals, indicator bacteria are associated with the fine particles which have lower removal efficiencies through street sweeping operations. Consequently, more research is needed to evaluate the effectiveness of current highway maintenance practices, including street sweeping, in removing both fine particulate and indicator bacteria from highways. 41 2.1.5 De-Icing Agents De-icing agents are used extensively during the winter months, either alone or with abrasives, to promote safe, driveable highway conditions. The de-icing agents commonly used are sodium chloride (NaCl) and, to a lesser extent, calcium chloride (CaCl 2) because they are effective and relatively cheap (Murray and Ernst, 1976). Even though it is not economically feasible to melt all the snow and ice through salt application, the primary purpose of salt application is to break the ice-pavement bond to enhance plowing. Apart from the monetary costs, the use of highway de-icing agents can cause corrosion of highway infrastructures and vehicles, contamination of surface and subsurface waters, contamination of soils, and damage to roadside vegetation (Hanes et al., 1970; Jones et al., 1986; Lord, 1989) depending on the site-specific conditions. Some of the factors that influence the impacts of de-icing agents on the surrounding ecosystem are (Davis et al., 1991): • site, soil and vegetation characteristics (slope aspects, soil permeability, texture, structure, vegetation tolerance and age); climate (amounts and patterns of rainfall, temperature, wind and snow cover); highway characteristics (slope length, average daily traffic (ADT), speed and curve radius); • highway de-icing practices (kinds, amounts, rates, timing of salt applications and snow removal procedures); and • time. Both current and past research studies suggest reduction in the amounts of de-icers used and the use of alternate chemical de-icers as possible mitigative measures. Lord (1989) listed a whole range of possible alternatives which included: pavement heating; 42 incorporation of de-icing agents into the highway surface fabric; alternative (less harmful) de-icing chemicals; reduced chemical use; improved operating maintenance and mechanical approaches (modification of plow equipment). The use of alternate chemical de-icers may create new problems or may be more expensive to implement. Implementation of the other possible alternatives suggested by Lord (1989) may not be economically feasible or may lead to more environmental degradation. However, OMOE (1980) and B.C. Research Corporation (1991) have both endorsed reduced chemical use as one of the most successful approaches so far. Even though de-icing agents are not measured directly in this research project, electrical conductivity measurements are used to make inferences about potential hazards from de-icers. Electrical conductivity (EC) is related to the concentration of dissolved mineral salts, and with regard to soil is used as an index of soil salinity. Colwill et al. (1984) used electrical conductivity to evaluate de-icing salt applications to U.K. highways while Kobriger and Geinepolos (1984) used the concentrations of sodium, calcium and chloride (mg/L) to do a similar evaluation on U.S.A. highways. Although the methods of analysis were different, the general conclusions were similar. The research studies of Asplund et al. (1982), Colwill et al. (1984), and Davis et al. (1991) did not use average daily traffic (ADT) to predict de-icing agents pollution contribution. Colwill et al. (1984) suggested that the effects of traffic flow variation and ADT were overshadowed by other more dominant factors which included storm characteristics, antecedent dry period and time of year. Kobriger and Geinepolos (1984) monitored traffic volumes and de-icing agent concentrations such as sodium, calcium and chloride on four U.S.A. highways, 43 however, their attempts to correlate ADT and the effects of de-icing agents were unproductive. Milwaukee, with the highest ADT, had the highest sodium, calcium and chloride concentrations for both winter and non-winter. Efland, on the other hand, with the lowest ADT,-had more sodium and chloride concentrations than Harrisburg with the third highest A D T for the winter months, whereas Harrisburg had more calcium concentration than Efland for the same time period, as can be seen in Table 2-1. Consequently, it is safe to assume that the impact of ADT on de-icing agents is inconclusive. Unlike ADT, the time of year has a direct effect on the de-icing agent concentrations. As can be seen in Table 2-1, the sodium, calcium and chloride concentrations were higher in the winter months than the non-winter period for all the sites with the exception of Sacramento (no winter). Kobriger and Geinepolos (1984) attributed these winter high values to the addition of sodium and calcium chlorides to highway surfaces to improve road safety. The application rate of these de-icing agents varies from place to place. Also the rate of application depends on the road ice conditions as well as on the rate of snow fall/accumulation. Colwill et al. (1984) estimated application rates to vary from 20 to 40 g/m2 in U.K. highways. Asplund et al. (1982) estimated application rates in Washington State to be approximately 170 kg per lane-km (600 lbs per lane-mile) per storm, of which 20 percent is salt and 80 percent is sand. In B.C., sodium chloride is the predominant de-icing agent used and the salt is applied at the rate of 60 to 130 kg salt per lane-km depending on the local site conditions (Davis et al., 1991). Accordingly, these sanding and de-icing operations are the largest single source of both solids and de-icing agents during the winter months at many of the sites. During the summer, routine use of calcium and magnesium chlorides for dust control is the major source of de-icing agents. This was observed to be the case by Davis et al. (1991) in their research studies in Lillooet and Goldbridge in B.C. 44 Table 2-1. Highway Runoff De-icing Agent Concentrations (mg/L) ADT Sod ium Calcium Chloride Non-Winter Winter Non-Winter Winter Non-Winter Winter Milwaukee 116,000 159 1,459 14 59.3 82.9 2,360 Sacramento 85,000 18.6 19 _ 12.6 Harrisburg 27,000 16 43.9 15 20 22.1 61.1 Efland 25,000 3.1 90.4 13.8 19.2 202 Source: Kobriger and Geneipolos (1984) Note: 1) — = Data not available 45 The surrounding land use activities have little or no direct impact on the highway runoff de-icing agents contribution. Indirectly though, surrounding land with predominantly trucking activity use may necessitate regular de-icing and sanding operations to improve road safety during the winter months. During the non-winter months, calcium and magnesium chlorides may be required for dust control purposes (Davis et al., 1991) as part of the highway maintenance. Highway maintenance practices can greatly influence the concentrations of de-icing agents on highway runoff. Highway maintenance practices such as street sweeping can be used to remove solid materials during the non-winter months. However, in some areas the routine use of calcium and magnesium chloride may be required to control dust particles during the non-winter period. This is a practice that was observed in Lillooet and Goldbridge, B.C. by Davis et al. (1991). During the winter months, most highway maintenance practices are centered around snow plowing and sanding/de-icing salt application. As a result, large quantities of de-icing agents are added to the highway environment as part of regular highway maintenance (Asplund et al., 1982; Kobriger and Geinepolos, 1984; Davis et al., 1991), as can be seen in Table 2-1. Consequently, this practice adds highway de-icing agents as sources of environmental pollution. 2.1.6 Nutrients In contrast to metals, pathogenic bacteria and organic contaminants, nutrients such as nitrogen and phosphorus are generally present in urban and highway runoffs in dissolved form. As a result, settling out the solids is ineffective in nutrient removal from urban and highway runoff (Stahre and Urbonas, 1990). Previous research studies, such as Kobriger and Geinepolos (1984), have suggested that nutrient levels in highway runoff are similar to levels found in general urban runoff. Athayde et al. (1983) supported that claim. OMOE (1980) found the levels of nitrogen and phosphorus in urban runoff to be less than the levels found in municipal sewage, 46 but still significant. Similarly, Swain (1983) found total nitrogen (N) and total phosphorus (P) concentrations of 1.61 mg/L and 0.42 mg/L respectively in residential urban runoff and Lawson et al. (1985) found total N and P concentrations of 1.31 and 0.09 mg/L respectively in industrial runoff in B.C. to be similar to the total N and P concentrations of 1.5 and 2.0 mg/L N and 0.33 and 0.36 mg/L P respectively found in the general urban runoff studies of Stahre and Urbonas (1990) and Bastian (1986) in the U.S.A. In Seattle, Washington, Chui et al. (1982) reported the total N and P concentrations of 1.2 to 1.5 and 0.2 mg/L respectively found in highway runoff to be similar to values found in general urban runoff. The values of 2.72 mg/L N and 0.59 mg/L P found by Shelley and Gaboury (1986) in urban highways, and the findings of Chui et al. (1982), tend to support the conclusion that nutrient levels are comparable for urban and highway runoff. However, in contrast to this . generalization, Swain (1983a) found the total N and P concentrations from highway runoff in Kelowna, B.C. to be considerably higher than runoff from the residential area. Swain's (1983a) finding was supported by the median nitrogen and phosphorus concentrations of 2.72 and 0.59 mg/L respectively in highway runoff according to the U.S. F H W A data base, as opposed to median nitrogen and phosphorus concentrations of 1.5 to 2.0 and 0.33 to 0.36 mg/L respectively in general urban runoff (B.C. Research Corporation, 1991). Kobriger and Geinepolos (1984) looked at the impacts of other variables such as ADT on nutrient levels, but the correlation between ADT and nutrient concentrations was not clear. Milwaukee, with the highest ADT value of 116,000 vehicles, had mean phosphate, nitrate and nitrite and total kjeldahl nitrogen concentrations of 0.33, 0.66 and 2.88 mg/L respectively, compared to Harrisburg with an ADT of 27,000 with mean phosphate, nitrate and nitrite, and total kjeldahl nitrogen concentrations of 1.88, 4.37 and 2.19 mg/L respectively. The unusually high nutrient levels in the Harrisburg site have less to do with the A D T and more with local site 47 conditions. Harrisburg has a median precipitation value of 14 mm compared to Milwaukee with 6.4 mm. The probability of contaminants being washed off completely from highway surfaces is thus higher at the Harrisburg site than the Milwaukee site (Kobriger and Geinepolos, 1984). The time of year had more of an impact on the Milwaukee site than at other sites. While the other sites did not show any significant nutrient concentration differences, Milwaukee showed a noticeable difference in the concentrations of nutrients between winter and non-winter periods. The winter mean phosphate, nitrate and nitrite, and total kjeldahl concentrations for Milwaukee were 0.54, 0.83 and 3.41 mg/L compared to the non-winter values of 0.33, 0.66, and 2.88 mg/L respectively. With no winter conditions in Sacramento, the only other site that showed noticeable variation was Harrisburg with a non-winter nitrate and nitrite mean concentration of 4.37 mg/L compared to a winter value of 6.20 mg/L (Kobriger and Geinepolos, 1984). Clearly, more research is needed to study the impact of time of year on the nutrient pollution contribution from highway runoff. Very little data exists on the effect of surrounding land use activities on nutrient pollution generation from highways. Past research suggests that runoff from predominantly agriculture-based surrounding land use tends to have higher nutrient concentrations. This is due to fertilizers/animal manure being washed off the fields into general urban runoff. However, where local conditions encourage denitrification, which is the biochemical reduction of nitrate and nitrite to nitrous oxide and molecular nitrogen through microbial activities such as heterotrophic bacteria (Loehr et al., 1979), the nitrate and nitrite contribution to urban runoff or through leaching is minimal (Onwumere, 1992). Whether nutrients in runoff from highways in a predominantly agricultural area will be higher or lower is thus an open question and a recommended subject of future research. 48 Current highway maintenance practices such as street sweeping do not affect highway runoff nutrient levels. The only highway maintenance practice that could affect the nutrient levels in runoff is summer ditch or median fertilizer addition to enhance grass growth. 2.1.7 Others (Asbestos, Rubber and Special Compounds) The definition of the term "special compounds" depends on the researcher. Not much published data exists on special compounds and asbestos concentrations in urban and highway runoff. As a result, little can be said about the impacts of ADT, time of year, surrounding land use activities and highway maintenance on highway asbestos and special compound concentrations. However, Kobriger and Geinepolos (1984) reported average asbestos loadings on highways from brake and clutch wear for passenger cars, light trucks, medium trucks and heavy trucks to be 45.9, 141, 468 and 1,530 micrograms per km respectively. Rubbers levels, on the other hand, are affected by ADT. Kobriger and Geinepolos (1984) reported surface rubber loadings of 7.25 kg per km per lane in Milwaukee with an ADT of 116,000; 1.51 kg per km per lane in Sacramento with an ADT of 85,000; 0.096 kg per km per lane in Harrisburg with an ADT 27,000; and 0.088 kg per km per lane in Efland with an ADT of 25,000 vehicles. Like ADT, the time of year with the most traffic is more likely to generate the highest surface rubber loading. In the G.V.R.D., there tends to be more traffic during the summer months. Consequently, more surface rubber loading would be expected during this time of the year than the winter months with lower ADT. Surrounding land use activities probably have little impact on highway surface rubber generation directly. Indirectly, it is expected that watersheds with predominantly trucking activities/businesses are more likely to have higher surface rubber loadings from their highways 49 than watersheds with no trucking activities. There has not been any known definitive research done to evaluate the impact of surrounding land use activities on highway surface rubber loadings. As a result, whether the surrounding land use activities impact on highway rubber generation directly or indirectly is a subject for future research. Highway maintenance operations such as sweeping and flushing are likely to have a direct impact on the level of rubber on highway surfaces. Depending on the rubber particle size, street sweeping and flushing may reduce the amount of rubber on highway surfaces, thereby reducing the level of rubber in highway runoff waters. 2.2 Sources of Highway Contaminants The major sources of contaminants to highway runoff are: • Rainfall Dustfall • Vehicles (i.e. metals, oil and grease) and their exhaust • Highway maintenance - sanding/de-icing agents (Asplund et al., 1982; Kobriger and Geinepolos, 1984). Although Table 2-2 shows a comprehensive list of highway runoff constituents and their primary sources, the literature review in this section is based on the above-mentioned four sources of highway pollutants. 2.2.1 Rainfall The studies of Pham et al. (1978) in New Jersey, Owe et al. (1982) in Syracuse, and Kobriger and Geinepolos (1984) in Milwaukee, Sacramento, Harrisburg, and Efland provide the best data so far on pollutant concentration and loading from precipitation. The six sites studied 50 Table 2-2. Common Highway Runoff Constituents and their Primary Sources Constituent Primary Source Particulate Pavement wear, vehicles, atmosphere, highway maintenance. Nitrogen, phosphorous Atmosphere, roadside fertilizer application. Lead Leaded gasoline (auto exhaust, tire wear, lead oxide filler material), lubricating oil and grease, bearing wear. Zinc Tire wear (filler material), motor oil (stabilizing additive, grease). Iron Autobody rust, steel hwy. structures (guardrails, etc.) moving engine parts. Copper Metal plating, bearing and bushing wear, moving engine parts, brake lining wear, fungicides and insecticides applied by maintenance operations. Cadmium Tire wear (filler material), insecticide application. Chromium Metal plating, moving engine parts, brake lining wear. Nickel Diesel fuel gasoline (exhaust) and lubricating oil, metal plating, bushing wear, brake lining wear, asphalt paving. Manganese Moving engine parts and gasoline Bromide Auto exhaust. Cyanide Anti-cake compound (ferric ferrocyanide), prussian blue or sodium ferrocyanide, yellow prussiate of soda used to keep de-icing salt granular. Sodium,calcium De-icing salts, grease. Chloride De-icing salts. Sulphate Roadway beds, fuel, de-icing salts. Petroluem products Spills, leaks or blow-by of motor lubricants, antifreeze and hydraulic fluids, asphalt surface leachate. Polychlorinated bipnenyl, pesticides Spraying of highway right-of-ways, background atmospheric depositions, PCB catalyst in synthetic tire. Pathogenic bacteria (indie.) Soil, litter, bird droppings, truck hauling livestock and stockyard waste. Rubber Tire wear. Asbestos Clutch and brake lining wear. Note: The above table was compiled from U.S.A. and European literature. 51 have different climatic, surrounding land use, highway design and traffic characteristics. However, the precipitation results indicated that all the sites had considerable levels of particulate, metals, oil and grease as well as de-icing agents in their precipitation. Kobriger and Geinepolos (1984) reported total solids concentrations of 44, 64, 30 and 12 mg/L for Milwaukee, Sacramento, Harrisburg and Efland respectively in their precipitation analysis. According to B.C. Research Corporation (1991), the major source of solids in precipitation is from atmospheric dustfall which is entirely site-specific and highly influenced by surrounding land use and ADT. This explains why the total solids concentrations in the two urban sites (Milwaukee and Sacramento) were higher than the concentrations found in the two rural sites (Harrisburg and Efland), even though there are appreciable variations in the mean precipitation values (Milwaukee - 6.44mm, Sacramento - 29 mm, Harrisburg - 13.7 mm and Efland - 29.2 mm). Colwill et al. (1984) also reported high total suspended solids concentrations in their rainwater quality analysis in Bedfordshire, U.K. They concluded that the concentration of suspended solids in an individual event was highly influenced by rainfall intensity, with the highest concentration usually coinciding with the peak rainfall intensity. Also, there may be total solids contribution from vehicle splash-off depending on the height and distance of the rain gauge from the highway. This may be the reason for the considerably high total suspended solids values of 557 and 156 mg/L taken from precipitation stations 1 and 2 metres away from the highway respectively. Similarly, there may have been some contribution of solids, especially to the precipitation station that is 1 metre away from the road and, hence, from traffic generated spraying. Apart from solids, high levels of metals were also found in precipitation. Kobriger and Geinepolos (1984) reported zinc (Zn), iron (Fe), and chromium (Cr) levels in precipitation in Milwaukee to be 0.17, 0.20, and 0.04 mg/L respectively; Sacramento to be 0.11, 0.18, 0.035 52 mg/L respectively; Harrisburg to be 0.08, 0.30 and 0.04 mg/L respectively and Efland to be 0.03, 0.08, and 0.03 mg/L respectively. The results of Kobriger and Geinepolos (1984) supported the findings of Pham et al. (1978) who reported zinc, cadmium, lead, and copper concentrations in precipitation to vary from 0.05 to greater than 0.25 mg/L in storms in their study in New Jersey. Black (1980) also found metal levels to vary during a single storm event, but reported the highest metal concentration just after the storm began. With Owe et al. (1982) reporting metal levels in the range of 0.02 to 0.186 mg/L for lead, zinc, copper, and cadmium, and Colwill et al. (1984) reporting metal levels in the range of 0.013 to 10.4 mg/L for cadmium, chromium, copper, iron, lead and zinc in rainfall samples, it seems that precipitation can contribute a significant amount of metals to highway runoff. No mercury (Hg) was reported in precipitation samples in the U.S.A. by Kobriger and Geinepolos (1984) nor in the U.K. by Colwill et al. (1984). This is probably because Hg is extremely volatile in the environment. Unlike the solids and metal concentrations, the limited data available showed levels of oil and grease in precipitation to be undetectable. Both Kobriger and Geinepolos (1984) and Colwill et al. (1984) reported undetectable levels of oil and grease in their precipitation analysis in the U.S.A. and U.K. This tends to suggest that all of the oil and grease found in highway runoff are either vehicular generated or from oil spills. De-icing agents, on the other hand, have been detected in precipitation samples, especially chlorides (CI"), at all the U.S.A. and U.K. sites. Kobriger and Geinepolos (1984) observed the following chloride concentrations: 9 mg/L for Milwaukee, 3 mg/L for Sacramento, 1 mg/L for Harrisburg and 3 mg/L for Efland, and also noticed a seasonal variation. The highest chloride concentrations were found during the winter months (November and January) at the Milwaukee site. Colwill et al. (1984) also noticed a seasonal variation in the chloride concentrations in their precipitation samples analysis in the U.K. with the highest concentrations 53 observed during the winter months. Swain (1983) observed the same trend in residential precipitation data in Vancouver, B.C. The sodium and chloride results for Vancouver were higher than at other British Columbia sites. Swain (1983) and Lawson et al. (1985) attributed the high sodium and chloride concentrations in Vancouver precipitation to the proximity of the site to the Pacific Ocean (a probable source of salt). Swain (1983) also reported finding about 0.00002 tonnes/kmVmonth of Hg in residential precipitation samples in Vancouver, B.C. This finding could not be substantiated by Lawson et al. (1985) due to lack of Hg data in their precipitation samples in Burnaby, B.C. Also present in rainfall are polynuclear aromatic hydrocarbons (PAHs). However, neither Kobriger and Geinepolos (1984) in U.S.A. nor Colwill et al. (1984) in U.K. reported finding PAHs in precipitation samples. According to Smith (1994), surface runoff and fallout from-the air are the main sources of high molecular weight P A H , while petroleum spillage is the main source of total P A H . 2.2.2 Dustfall Atmospheric dustfall pollution contributions to highway runoff are site-specific. The main sources of dustfall are from urban areas, the land adjacent to the highway, and ADT (Asplund et al., 1982; B.C. Research Corporation, 1991). As a result, atmospheric deposition of airborne pollutants could be a significant source of highway runoff contamination (Bellinger et al., 1982). According to Kobriger and Geinepolos (1984), and Colwill et al. (1984), both wet and dry deposition rates are higher in urban sites than in rural sites. For example, Kobriger and Geinepolos (1984) reported total particulate matter loadings in wet and dry precipitation for the two urban sites to be 130 and 160 mg/m2 for Milwaukee and Sacramento respectively, whereas the rural sites had total particulate matter loadings of 30 and 40 mg/m2 for Harrisburg and Efland 54 respectively. According to Lord (1987), however, the dry deposition of contaminants on highways has more of an influence on runoff contamination and thus is more important than wet deposition. In western Washington State, Asplund et al. (1982) estimated total suspended solids (T.S.S.) deposition rates from dustfall to be 0.90 kg per ha per day for rural areas and 2.24 kg per ha per day for urban areas. Similarly, they found deposition rates ranging from 1.1 to 3.3 kg per ha per day in urban industrial sites compared to values of 0.6 to 1.1 kg per ha per day for rural sites. These values are higher than the dustfall measurement range of 0.6 to 2.5 kg per ha per day reported by Kobriger and Geinepolos (1984) in their studies across the U.S.A. In Vancouver, B.C., Lawson et al. (1985) reported particulate deposition rates to range from 0.33 to 8.3 mg/dm2 per day. In England, Harrison and Wilson (1985) reported atmospheric deposition contributed up to 48 percent of suspended solids in highways runoff. Kobriger and Geinepolos (1984) observed higher particulate levels at the Milwaukee site in the fall season than at any other time of year. Consequently, atmospheric dustfall can contribute significant amounts of pollutants to highway runoff, especially when the surrounding land is highly urbanized. Like the particulate matter loadings, metals were found in higher concentrations in dustfall from urban areas. For example, the two urban sites - Milwaukee and Sacramento - had higher concentrations of lead, zinc, iron, chromium, cadmium and nickel in the wet and dry precipitation data analysis than the two rural sites of Harrisburg and Efland (Kobriger and Geinepolos, 1984). The higher loadings in the fall season at the Milwaukee site were attributed to higher winds, thus suggesting a seasonal influence. However, Lawson et al. (1985) in their studies in Vancouver, B.C. found no evidence of seasonal influence on metals, but they found significant amounts of insoluble lead, copper and zinc in their dustfall analysis. In contrast to metals, they found a seasonal influence on particulate, which is consistent with the work of Kobriger and Geinepolos (1984). Both Kobriger and Geinepolos (1984) in the U.S.A. and 55 Colwill et al. (1984) in U.K. did not report finding any Hg levels in their samples. However, Swain (1983) in Vancouver, B.C. reported finding about 0.00002 tonnes/km2/month of Hg in his residential dustfall samples. Swain's finding was supported by Lawson et al. (1985), even though analytical problems hindered obtaining an adequate number of Hg results in their research. 2.2.3 Vehicles and their Exhaust Vehicles contribute significant amounts of pollutants to highway runoff. There are direct and indirect (or acquired) sources of pollutants deposited by vehicular traffic. According to Asplund et al. (1980), the direct sources originate from normal vehicular operations and factional wear of parts, whereas the indirect sources result from pollutants that are carried and later deposited by vehicles, especially during storm events. This is probably why the total • suspended solids and metal loading models from the Washington State studies suggest vehicle during storm (VDS) as a more important factor than vehicle counts during the dry period (Asplund etal., 1980). The direct pollutant sources are major contributors of pollutants ranging from chemical oxygen-demanding (COD) compounds to metals. Oil leakage and gasoline/gasoline associated products contribute contaminants such as chemical oxygen demand, oil and grease, nitrates, sulphate, phosphorus and organic contaminants to highway runoff (Asplund et al., 1980). The levels of oil and grease and nutrient contribution from vehicular traffic have been reported in the previous sections. Kobriger and Geinepolos (1984a) divided particulate of vehicular origin as 37 percent from tire wear, 37 percent from pavement wear, 18.5 percent from engine and brake wear and 7.5 percent from settleable exhaust. This supported earlier work by Christensen and Guinn (1979) who reported vehicular tire wear contributions of 0.0030 g zinc per vehicle-km and 0.0049 g lead per vehicle-km. Other research studies by Hedley and Lockley (1975), Ellis and 56 Revitt (1982) and Colwill et al. (1984) in England and Helsel et al. (1979), Hoffman et al. (1984a) and Cole et al. (1984) in the U.S.A. reported wear particles from tires, brake linings and corrosion of metal parts as significant sources of zinc, copper, chromium, cadmium, nickel and lead. Pigment from road marking paints was found to contain chromium, lead and zinc, whereas road salts (de-icing agents) were found to contribute such contaminants to highway runoff as zinc, nickel, chromium, lead, cyanides and a considerable quantity of suspended solids from road salts/sand applications (B.C. Research Corporation, 1991). In Canada, the phase-out between 1973 and 1990 of Tetraethyl lead (TEL), a gasoline additive to enhance octane ratings, led to a decline in automobile- associated Pb pollution. At the peak of TEL usage, Poon (1988) estimated the total Pb emissions from Canadian automotives to be about 14,360 tonnes annually. Lead has been completely removed as a gasoline additive since 1990, however, the natural levels of Pb in gasoline at approximately 10 mg/L may still contribute trace amounts of Pb to urban environments (Lee and Jones-Lee, 1993). The banning of leaded gasoline meant something else had to be used. In 1974, Canadian refineries started using methyl cyclopentadienyl manganese tricarbonyl (MMT) as a substitute for Pb in gasoline (Forget et al., 1994). Since then, several studies have found significant correlations between increased environmental M n concentrations and traffic density in Canada (Joselow et al., 1978; Forget et al., 1994; Loranger et al., 1994), unlike in the U.S.A. where leaded gasoline is still used. In fact, McCallum (1995) reported an increase in extractable M n between 1973 and 1993. He attributed this M n increase in the environment to the obvious use of M M T as a gasoline additive in Canada. Huntzicker et al. (1975) reported leaded gasoline average exhaust emissions of 0.107 kg per 1000 vehicle-km in a Los Angeles basin of which 45 percent is particulate lead and the rest is organic hydrocarbon and metallic compounds. Kobriger and Geinepolos (1984) reported leaded 57 gasoline emissions of 0.094 kg per 1000 vehicle-km of which lead was less than 33 percent of total particulate, carbon compound was about 35 percent and the rest was organic hydrocarbons and metallic compounds. They reported unleaded gasoline exhaust to emit about 0.103 kg per 1000 vehicle-km of suspended particulate. G.V.R.D. (1993a) reported that motor vehicle emissions of particulate matter and carbon compounds should be of concern to the region's air and water quality. This was based on the findings of Kobriger and Geinepolos (1984) in which they reported that carbon emissions from unleaded gasoline cars represent 20 percent of the total particulate. Lower particulate emission rates have been observed with alcohol-operated motor vehicles, according to Asplund et al. (1980), while higher particulate emissions were observed with diesel-engine motor vehicles. The exhaust emission rates depend on vehicle size, age, fuel type, condition of the vehicle, highway driving speeds, vehicle acceleration rates and highway slopes (Asplund et al., 1980). In summary, direct pollutants from vehicles are a major source of pollution to highway runoff. Other, though lesser, sources of vehicular generated pollution are the indirect or acquired sources. The pollutants from this source are from adjacent land use, atmospheric/wind blown solids, litter and debris, pavement wear, and vehicular picked up and later deposited contaminants. Kobriger and Geinepolos (1984) attributed the high iron concentration at the Harrisburg site to the high iron content in the surrounding local soil and the higher winter loadings of contaminants at all the sites to less atmospheric blow-off/reduced winter wind actions. These findings were supported by the studies of Hedley and Lockley (1975) and Colwill et al. (1984) in U.K. motorways, and Asplund et al. (1982), Horner et al. (1979) and Kerri et al. (1985) in the study of U.S.A. highways. For example, Asplund et al. (1982) reported atmospheric deposition rates from surrounding land use activities to vary from 0.6 kg per ha in the rural areas to 3.3 kg per ha in urban sites such as Seattle, Washington. Similarly, Kobriger 58 and Geinepolos (1984) reported that 37 percent of the particulate from vehicular origin came from pavement wear. Although Asplund et al. (1982) noticed an association between particulate availability and studded tire impacts, an attempt to quantify the relationship was unsuccessful. However, Lygren et al. (1984) estimated pavement dust from the use of studded tires in Norway during the winter months to range from 20 to 50 g per km per vehicle. Pavement wear could also be a source of chemical oxygen demanding substances and oil and grease depending on the composition of the asphalt/concrete mix. According to Asplund et al. (1982), asphalt based highway surfaces tend to exert greater influence on chemical oxygen demand and oil and grease than concrete highways. Other acquired sources of pollution such as vehicular pick-up and later deposited pollutants depend on where the vehicle has been, what it picked up and when it is washed off the vehicle. Potential sources of contaminants for this type of acquired or indirect pollution are unpaved parking lots, urban and industrial sites, dirt roads, construction sites and agricultural sites (Asplund et al., 1980). Usually these pollutants are washed off from the underside of the vehicle during rainfall events. This is why Washington State models suggest using vehicle during storm (VDS) in estimating total suspended solids loading in highway stormwater runoff. Both direct and indirect sources of major pollutants are of concern for this research project. In summary, metals and their sources from vehicles are: lead and nickel from gasoline and exhaust emissions; lead, nickel and zinc from lubricating oils; zinc and lead from grease; zinc and cadmium from tire wear; various metals and solids from concrete wear; nickel and vanadium from asphalt wear; copper, chromium and nickel from brake lining wear; copper and lead from bearing wear and iron, manganese, chromium and cobalt from wear of vehicle engine parts (Kerri et al., 1976; Hopke et al., 1980; Novotny and Chesters, 1981; Wang et al., 1982; 59 Colwill et al., 1984; Kobriger and Geinepolos, 1984). Other contaminants and their main sources are listed in Table 2-2. 2.2.4 Highway Maintenance In the previous sections, some of the highway maintenance practices such as street sweeping and flushing have been described as contributing to pollutant removal. According to Asplund et al. (1982), 25 to 78 percent of the highway pollutant mass can be removed by street sweeping operations. However, these street sweeping operation are more effective in removing larger particles than the pollutant laden fine particles. Sartor et al. (1974) found street cleaning operations at the time to be ineffective in removing the fine particles. Also, their research indicated that there may be a crushing of larger particles into fine particles by the mechanical actions of the street cleaning equipment. If this is the case, street sweeping could actually add to pollution. According to Hedley and Lockley (1975), Gupta et al. (1978) and Horner et al. (1979), most of the pollutants, especially metals, found on the highway are associated with the dirt and dust fractions. Therefore, i f the current street sweeping operations increase the amount of fine particles instead of reducing them, they could inadvertently add to the highway pollutant loadings. Apart from street sweeping operations, other highway maintenance practices such as sanding and de-icing operations have been major and significant sources of total solids and salt pollution during the winter months in temperate regions. Studies by Hedley and Lockley (1975) and Colwill et al. (1984) in U.K. ; Asplund et al. (1982) and Kobriger and Geinepolos (1984) in the U.S.A. and Davis et al. (1991) in B.C., Canada, all concluded that salts used for de-icing the highway surface are the most concentrated contaminants in highway runoff in the temperate regions during the winter months. Sanding deposition rates and frequencies vary from site to site 60 as well as with the weather conditions, road conditions, traffic volume and the local transportation department's operation policies. For example, Colwill et al. (1984) estimated application rates on U.K. highways to vary from 20 to 40 g per m 2; Asplund et al. (1982) estimated that sanding and de-icing operations add about 825 kg per lane-km (600 lbs. per lane-mile) on Washington State highways whereas Davis et al. (1991) gave a range of 60 to 130 kg salt per lane-km for British Columbia, Canada highways. Consequently, sanding and de-icing operations in temperate regions do contribute significantly to total solids and salts loading in highway runoff during the winter months. Similarly, during the non-winter months, the use of calcium and magnesium chloride for dust control purposes can add to highway runoff salt pollution. Davis et al. (1991) observed that both calcium and magnesium chloride were routinely used for dust control on B.C. highways, however, they did not indicate the extent of their usage. But Kobriger and Geinepolos (1984) reported calcium chloride concentrations ranging from 2.1 to 82.9 mg/L in four of the U.S.A. highway sites during the non-winter months. Based on this observation, calcium and magnesium chloride use during the non-winter period for dust control can be a source of salt pollution. However, the extent of this pollution needs further research. 2.3 Highway Contaminant Migration Patterns According to Gupta et al. (1981b), a large percentage of potential contaminants to highway/street runoff was found in fine, silt-like particles. Even though these fine particles by weight accounted for only about 6 percent of the total street/highway solids, they contained approximately 25 percent of the total oxygen demand, 33 to 50 percent of the algal nutrients, more than 50 percent of the heavy metals, and nearly 75 percent of the total pesticides/hydrocarbons (Gupta et al., 1981b). The level of contaminants in these fine particles 61 is important since Sartor et al. (1974), Gupta et al. (1981), and Asplund et al. (1982) found the conventional street sweeping operations at the time to be ineffective in removing these pollutant-laden fine particles. Depending on whether the highway surface is wet or dry, the physical migration of these pollutants from highway surfaces occurs through natural wind action/vehicular generated turbulence and washoff by rainfall/snow melt. The removal of pollutants through natural wind action/vehicular generated turbulence and the mechanical scrubbing action of the tires are most likely the major pollutant removal mechanisms in low precipitation regions (Asplund et al., 1982). This finding by Asplund et al. (1982) was supported by later research on completely paved highway sites at Redondo Beach, Walnut Creek and Sacramento, California by Kerri et al. (1985). Based on the data from seven rainfall events, they concluded that most urban highways in California do not produce large amounts of contaminants during storm events. Therefore, costly treatment facilities are not required in order to meet the current water quality criteria. They attributed this low pollutant loading to traffic-generated turbulence which constantly scours and transports contaminants off the travelled and shoulder lanes. The injection of fine particles into the atmosphere by vehicular generated turbulence is called saltation and it is affected by both wind speed and direction. Kobriger and Geinepolos (1984) suggested that saltation played a major part in the removal of particulate off the four highway sites they studied in the U.S.A. during dry periods. Colwill et al. (1984)made similar observations in their study in the U.K. , but further stated that the rate of deposition of these airborne pollutants on adjacent lands rapidly falls off with distance. They estimated that the saltation process disperses 630 kg per km per year of insoluble solids, 57 kg per km per year of oil and 2 g per km per year of polynuclear aromatic hydrocarbons within a distance of approximately 5 m from the hard shoulder of the highway in their study site in the U.K. Laxen and Harrison (1977) and Little and Wiffen (1978) 62 found that most of the pollutants deposited on the travelled or driving lanes are easily blown off either onto the distress strip or completely off the highway. Where there is a curb in place, however, they found the majority of the pollutants within 0.9 m (3 ft) of the curb. For example, Zimdhal (1972) and Getz et al. (1975) reported high levels of lead concentrations on road soil adjacent to the highway near Denver, Colorado and Illinois, but found the concentrations to rapidly diminish to background levels with increasing distance from the highway. Both Scanlon (1977) and Gupta and Kobriger (1980) found similar patterns in saltation pollutant dispersal. Kobriger and Geinepolos (1984) found the quantity of saltating particles to be influenced by average daily traffic, wind speed and direction, available surface pollutant load, highway drainage design, time of year, proximity of travel lanes to right-of-way and landscape and its effects on wind. They found that ADT has the greatest influence on the amount of solids removed from highway surfaces through saltation. They reported that 9.3, 11.0, 1.2 and 0.7 kg per km per day of total particulate were moved by an ADT of 115,000, 90,000, 32,000 and 27,000 respectively on four U.S.A. sites. As stated earlier, they also found wind speed and direction to significantly affect the saltation process. In another example, Getz et al. (1975) found lead concentrations along Illinois highways to decrease to background levels within 50 m on the downwind side and within 20 m on the upwind side. The available surface load and the type of highway drainage design affect the amount of pollutants dispersed through saltation. The more fine particles available on highway surfaces, the greater the amount of particles dispersed through vehicular generated turbulence. The studies of Laxen and Harrison (1977), Little and Wiffen (1978), Asplund et al. (1982) and Kobriger and Geinepolos (1984) all agreed that traffic during dry periods tends to remove pollutants from highways; however, curbs can trap pollutants. Therefore, highways with curbs tend to generate 63 higher pollutant loadings in runoff because the curbs can retain more air-blown material than non-curbed sections of the highway. Kobriger and Geineplos (1984) found that large amounts of dust and dirt build-up during the winter months usually resulted in the highest total highway saltation rates in the spring, while Asplund et al. (1982) report on Envirex studies on 1-50 at Sacramento found 87 percent of the total solids mass on the median and distress lanes. This provides evidence that the proximity of the travelled lanes to right-of-way, as well as time of year, affects the quantity of saltating particles. Finally, where the surrounding landscape is highly vegetated, re-suspension of saltating particles by wind may be minimal. Vegetation can act as sinks for dust and dirt particles, thereby lessening their probability of re-suspension. As a result, once the particles are blown completely off the highway surfaces, they probably remain within the vegetation. Solid removal during wet periods from highway surfaces results from scrubbing actions on the pavement by rainfall, mechanical scrubbing activities of the vehicle tires or both. Gupta et al. (1981) reported the rate of particulate matter removal by rainfall from street/highway surface to depend on rainfall intensity, street/highway surface characteristics and particle size, as well as the amount of material available for washoff (McKenzie and Irwin, 1983). Asplund et al. (1982) reported Envirex studies in which a rainfall event of 25.4 mm (1.0 inch) or greater with a peak intensity of 13 mm per hour (0.5 inch per hour) for at least one hour was required to remove approximately 90 percent of the solids. McKenzie and Irwin (1983), in their research at 1-95 in Miami, Florida, found two different storms with the same total runoff, peak discharge and similar runoff volumes to transport two significantly different total solids loads. The storm of November 3, 1979 had a total storm runoff volume of 48,700 L with total solids, dissolved solids, chemical oxygen demand and total organic carbon masses of 35.5, 33.6, 14.1 and 4.80 kg respectively, whereas the storm of May 1, 1981 had a total storm runoff volume of 48,100 L with total solids, 64 dissolved solids, chemical oxygen demand and total organic carbon masses of 8.0, 4.5, 7.7 and 1.2 kg respectively. McKenzie and Irwin (1983) attributed the higher total solids, dissolved solids, chemical oxygen demand and total organic carbon loads of November 3, 1979 to higher parameter concentrations (the amount of material available for washoff). Consequently, rainfall intensity, as well as the amount of material available for washoff, influences the rate at which rainfall washes off loose particulate matter and pollutants from highway/street surfaces. However, parameter concentration (the amount of material available for washoff) may be a more influential factor than rainfall intensity and runoff volume. For example, the research studies of Kerri et al. (1985) in California found that because traffic generated turbulence continuously sweeps the travel lanes (low parameter concentration), very low amounts of pollutants were generated during storm runoff events. Similarly, the scrubbing actions of the automobile tires against the highway surfaces help in grinding some of the larger particulate matter into finer particles. Thus, it is easier to transport fine particulate matter by both saltation and rainfall washoff. Asplund et al. (1982) estimated the amount of energy transferred by 1,300 vehicles per hour per lane to equal the energy delivered by a 13 mm per hour (0.5 inch per hour) rainstorm. As a result, when rainfall intensities are low, the scrubbing actions of the tires and the grinding of large particulate matter into finer particles with their eventual washoff may be the primary mechanism of pollutant removal from highway surfaces. Finally, the highway/street surface characteristics may ease or hinder the ability of rainfall events to wash off pollutants from highways. For example, Asplund et al. (1980) suggested that the most influential single factor in evaluating the amount of contaminants removed through storm runoff is probably the curbing or non-curbing of highways. Although in the present research project the two sites under investigation are both curbed, visual observations 65 by the researcher support the findings of Asplund et al. (1980) about the ability of curbs to retain air blown material. 2.4 Causes of Variation in Contaminant Loading One of the main objectives of this research project is to evaluate the quality of highway runoff water as it leaves the highway. Therefore, it is very important to look as some of the factors that cause variation in highway runoff contaminants loadings. They are (Gupta et al., 1981a and 1981b; McKenzie and Irwin, 1983; Kobriger and Geinepolos, 1984; Hall et al., 1976): Effect of surrounding land use; Impact of ADT and number of lanes; Climatic effects, i.e. rainfall intensity/snowmelt/discharge; Parameter concentration; Time of year; Drainage type; Pavement type and design characteristics; Antecedent factors, i.e. number of dry days and atmospheric dryfall. Effect of Surrounding Land Use. There has been some research done on the use of trees and shrubs in controlling the effect of wind and snow drifting during the winter months. The extent to which shrub cover protects the land soil adjacent to the highway from wind erosion is still questionable. However, the research studies of Asplund et al. (1982) and Kobriger and Geinepolos (1984) found total particulate matter in urban precipitation samples to be considerably higher than in those from rural sites. Asplund et al. (1982) estimated dustfall deposition rates from urban industrial areas to the highway to range from 1.1 to 3.3 kg per ha per day (1.0 to 3.0 lb per acre per day) as opposed to 0.6 to 1.1 kg per ha per day (0.5 to 1.0 lb per 66 acre per day) from rural sites. Kobriger and Geinepolos (1984) found urban precipitation samples to contain, on average, four times the total particulate matter of the rural precipitation samples. Even within the same urban environment in Vancouver, B.C. , Swain (1983) and Lawson et al. (1985) found total particulate matter to exceed the British Columbia Air Quality Objectives and Guidelines for desirable particulate levels of 1.75 and 2.90 mg per dm2 per day respectively. They attributed these high particulate matter levels to their introduction from unpaved parking areas/lands and subsequent suspension in air by heavy vehicular traffic. Similarly, analysis of urban stormwater runoff trace metal toxicity to aquatic invertebrates by Hall and Anderson (1988) showed commercial runoff to be the most toxic, followed by industrial, residential and open space. Therefore, the surrounding land use activities can significantly cause variation in dustfall deposition rates as well as other pollutant loadings. . Impact of ADT and Number of Lanes. Even though Colwill et al. (1984) in the U . K and Kerri et al. (1985) could not find any correlation between average daily traffic (ADT) and contaminant loading, urban areas generate more traffic than rural areas. The higher the number of vehicles, the more likely that more traffic lanes will be required to ease congestion. For example, the two urban sites - Milwaukee and Sacramento - in the Kobriger and Geinepolos (1984) study, with an ADT of 116,000 and 85,000 respectively, had eight lanes each. On the other hand, the two rural sites, Harrisburg and Efland with an ADT of 27,000 and 25,000 respectively, had four lanes each. Doubling the number of lanes means requiring twice the amount of sanding and de-icing agents needed during the winter months to keep these lanes safe (Kobriger and Geinepolos, 1984). Consequently, the number of lanes may be a significant factor in causing variation in contaminant loading, since during the winter months more particulate, de-icing agents and metals are introduced into highway runoff due to the extra lanes. 67 Climatic Effects. The amount of rainfall, intensity, storm discharge and quality of rainfall, as well as snowmelt, are important factors that affect the quantities and concentration of contaminants in highway/street runoff. Under normal circumstances, the higher the rainfall, runoff volume and/or the snowmelt, the greater the level of contaminant loading (Kobriger and Geinepolos, 1984) although, according to McKenzie and Irwin (1983), the influence of those factors on contaminant loading depends to a large extent on parameter concentration (the amount of contaminants available for washoff). Kerri et al. (1985) found generally low contaminant loading in urban California highways due, in part, to saltation which lowers the parameter concentration and also due to low precipitation. McKenzie and Irwin (1983) found that high intensity storm and runoff events on November 20, 1979 and May 20, 1981 transported nearly 70 to 80 percent suspended solid loads within a short period of time. This indicates that climate can be an important factor in causing variability in contaminants loading. Parameter Concentration. Parameter concentration (amount of material available for washoff) has the dominant influence on contaminant loading. As indicated in an earlier section, McKenzie and Irwin (1983) found two storm events with almost the same total runoff and peak discharge to transport significantly different concentrations of total and dissolved solids, chemical oxygen demand and total organic carbon and they attributed the difference to higher parameter concentrations. Other research studies by Hedley and Lockley (1975), Asplund et al. (1982), Kobriger and Geinepolos (1984a) and Kerri et al. (1985) have all indicated the importance of parameter concentration in highway runoff pollution generation. Time of Year. The time of year influences the concentration of contaminants in highway runoff. Characteristically, the concentrations of suspended and total solids, heavy metals and salts are higher during the winter months. Kobriger and Geinepolos (1984) found the concentrations of the above-mentioned parameters to be considerably higher during the winter months. They 68 attributed these higher winter concentration loadings to the use of de-icing agents and the effect of frost which reduces the movement of solids by wind and vehicle generated turbulence. As noted earlier, saltation and wind actions help in reducing the amount of pollutants in travel lanes by blowing pollutants off the road. Similarly, an earlier research study by Hedley and Lockley (1975) indicated increases in suspended solid and heavy metal concentrations during the winter months on U.K. highways. In fact, their analysis of the monthly distribution of the total annual loading showed 10, 42 and 17 percent of the total annual loading to occur in December, January and February respectively. Other studies by Asplund et al. (1982) in Washington State, Colwill et al. (1984) in U.K. and Lygren et al. (1984) in Norway found considerable amounts of pollutants, especially sanding/de-icing agents, metals and particulates from the use of studded tires and de-icing operations, in highway runoff during the winter months. Time of year is, • therefore, an important factor in causing variability in highway pollutant loadings. Another influential factor is the time taken for the suspended solids to be transported. Hoffman et al. (1984a) found the highest concentration of suspended solid transport to occur with the first flush which is the first major peak in flow rate. They further estimated the loading relationship to be linear with the total rainfall. McKenzie and Irwin (1983) found high-intensity rainfall events with sharp-peak flows to transport available material much more rapidly than low-intensity storm events. They concluded that approximately 90 to 100 percent of the cumulative suspended solids loads were transported in about 10 to 18 minutes after the initial runoff in their analysis of 5 storm events in Miami, Florida. They found other contaminants to follow a similar transportation pattern. Consequently, the time taken to wash off contaminants, as well as the time of year, can significantly influence highway pollutant loading. Drainage Type. The type of drainage channel can influence the type and amount of pollutants reaching a nearby water body. Past research studies have indicated lower concentrations of 69 ) contaminant in highway runoff channelled through grass and bare soil ditches than those measured in paved channels or in storm sewers. In fact, bare soil ditches were found to be the most efficient with regard to pollution concentration reduction with increased retention time. Although bare soil ditches are not as effective as grass ditches in flow attenuation, which is needed to enhance infiltration rates, grass drainage ditches work well in reducing the amounts of pollutants. For example, Wang et al. (1982) found a grass drainage ditch-length of 60 m to be effective in removing 60 to 80 percent of lead, zinc and copper from highway runoff in Washington State. Yousef et al. (1985) found grass drainage ditches built on high ground with good drainage and high infiltration rates to be very effective in highway contaminant removal. Pavement Type and Design Characteristics. According to Kobriger et al. (1981), the U.S. Federal Highway Authority uses factors such as rainfall-runoff, runoff duration, pollutant accumulation, pollutant wash-off and constituent loading characteristics to classify sites into three types: Type I Sites: Urban, elevated bridge deck, 100 percent paved, characteristic of urban highway systems. Type II Sites: Mountable curb, paved and non-paved areas, inlets spaced along the highway surface. Type III Sites: Rural, flush shoulders, paved (about 25 percent) and non-paved, runoff through grassy ditches with inlets spaced along the ditch. Theoretically, type I sites are associated with high runoff coefficients, increased peak flows in streams, erosion and sediment transport into receiving water bodies. Types II and III sites were less polluting due largely to unpaved areas which allowed infiltration and thus reduced flows. As a result, unpaved areas, usually vegetated, help in pollutant loading reduction. This supports the 70 findings by Wang et al. (1982), Kobriger and Geinepolos (1984) and Yousef et al. (1985) described in the previous section. Data analysis by Kobriger and Geinepolos (1984) on the influence of pavement type on runoff quality was inconclusive. The Sacramento and Harrisburg sites were concrete highways, whereas the Milwaukee and Efland were asphalt pavements. They found pavement type to have no significant impact on the quality of the runoff waters with their limited data set. Antecedent Factors. Number of dry days, traffic volume and atmospheric dryfall are some of the antecedent environmental factors that influence parameter concentration (the amount of material available for washoff) according to Asplund et al. (1982), McKenzie and Irwin (1983), Colwill et al. (1984), Kobriger and Geinepolos (1984) and B.C. Research Corporation (1991). Kobriger and Geinepolos (1984) found A D T counts to influence the quality of highway stormwater runoff among the four sites; McKenzie and Irwin (1983) suggested that antecedent dry days and traffic volume do not relate directly to runoff concentrations or loads; Colwill et al. (1984) found that during the dry periods, traffic generated turbulence aids in the dispersal of pollutants from highway surfaces in the U.K. , and Kerri et al. (1985) found traffic volume and speed to increase the rate of saltation in California highways, which reduces parameter concentration and, hence, the potential contaminants in highway stormwater runoff. Even though the influence of antecedent dry days and ADT on highway pollutant generation is inconclusive, most of the studies agree on the impact of atmospheric dry fall on highway stormwater runoff quality. As seen in the earlier section on sources of highway contaminants, atmospheric dryfall, especially at urban industrial sites, can be a significant source of highway contaminant loading. 71 2.5 Impact of Pollutants on Water Quality Documented information on the impact of highway runoff on receiving waters is limited. But much research has been done on the impact of urban stormwater runoff on receiving waters and most of that research may be applicable to the potential impact from highway runoff. Research to date has found undiluted urban/highway stormwater runoff to often exceed both the acute or shock (short term) and chronic (long-term) toxicity criteria for most water bodies (Gupta et al., 1981; B.C. Research Corporation, 1991). Acute (short term) toxicity is reduced by dilution of the runoff before it reaches the receiving water. Many of the pollutants in urban/highway runoff such as metals, organic contaminants and pathogens have been found to persist in the sediments. Pollutants adsorbed or incorporated with the sediments have been found to have chronic (long term) effects through either their re-release to the water column or incorporation into the food chain through consumption by benthic organisms (Helsel et al., 1979). Terstriep et al. (1986) concluded that urban runoff affects the receiving water bodies by raising metals, nutrients and suspended solid concentrations well above their background levels. In the Puget Sound region of Washington State, stormwater analysis found that more than 90 percent of the annual runoff pollutant load to the rivers was generated by storms of six month frequency or less (WSDOE, 1990). Since urban/highway runoff, i f undiluted, is a significant source of both harmful toxic elements and nutrients, it is a potential threat to the environment (Simpson and Stone, 1988). According to Gupta et al. (1981) and Livingston (1989), urban/highway runoff is intermittent, relatively short in duration, site-specific and storm-specific and its impact depends on rainfall quantities/qualities and the sensitivity of the receiving waters. Due to the similarities between urban and highway stormwater runoff quality, the impact of contaminants such as suspended solids, oxygen demanding substances, metals, organic 72 contaminants, nutrients and pH/electrical conductivity (EC) on receiving waters are summarized in this section. Suspended Solids. Particulate matter, especially in the form of suspended solids, can contribute to a variety of problems such as objectionable appearances and increased turbidity which can, in turn, lower dissolved oxygen concentration and reduce prey capture for predatory sight feeders. Other effects can include the clogging of fish gills and invertebrate filters, reducing spawning and juvenile fish survival as well as angling success, smothering of bottom dwelling aquatic organisms such as benthal organisms and, above all, introduction of solid-laden pollutants into the water column (Asplund et al., 1982; B.C. Research Corporation, 1991). The effects of suspended solids and their associated contaminants vary from site to site as well as with the seasons. Suspended solid generation rates variation with sites were reported in the works discussed earlier by Homer et al. (1979), Asplund et al. (1982) and Kobriger and Geinepolos (1984) in the U.S.A., Colwill et al. (1984) in U.K. , and Lygren et al. (1984) in Norway. This site variability in suspended solid generation and the impacts on receiving waters make it necessary to generate local data for effective stormwater management. Oxygen-Demanding Substance. Estimates of the amount of oxygen consumed by oxidation of the organic matter in solution by chemical and biological means are usually done by chemical oxygen demand (COD) and biochemical oxygen demand (BOD) respectively. But both measurement techniques have major drawbacks. According to Marsalek (1986), oxygen demands on stormwater runoff cannot be accurately measured using BOD test because of possible inhibition of biological activity by contaminants contained in stormwater runoff, as well as by the inability to detect delays in oxygen demands resulting from benthal activities. Similarly, the COD test has its own drawbacks. According to Schueler (1987), the COD test may reduce inorganic substances as well as include non-biodegradable organic matter in its 73 analysis. The COD test is, however, a more appropriate indicator of long-term oxygen demand in stormwater runoff than BOD. Oxygen-demanding substance analysis in highway runoff by Chui et al. (1982) and Shelley and Gaboury (1986) reported no data for BOD but reported a range of 0.23 to 1291 mg/L for a variety of sites ranging from rural to urban highways from their COD analyses. In the G.V.R.D., Swain (1983) and Lawson et al. (1985) found mean BOD and COD values of <10 and 14, and 33 and 78 mg/L respectively in their urban stormwater runoff samples. Hall and Anderson (1988) reported a COD range of 46 to 1,031 mg/L at their sites. The lowest COD level of 46 mg/L was reported for one of the open space sites, whereas the highest COD level of 1,031 mg/L was reported for one of the commercial sites. If these concentrations are comparable to highway stormwater BOD and COD concentrations, then oxygen-demanding substances in highway runoff must be taken seriously. Neither BOD nor COD were measured in this research project due to limited budget. Organic materials from highway runoff in a receiving water body can stimulate bacterial growth, which may consume most of the available oxygen, resulting in depressed oxygen levels in water which can adversely affect aquatic life forms that require oxygen (B.C. Research Corporation, 1991). Under severely depressed oxygen conditions, gas formation, discoloration and odours may be apparent. If this happens in a lake, it can change the trophic status of the lake from eutrophic to oligotrophic. Again, site-specific COD data are required for realistic estimates of the loading to a receiving water body. Metals. Many metals that are required as micronutrients by organisms at reasonable concentrations can be toxic to aquatic organisms at elevated concentrations. The toxicity levels depend on the metal and metal species, the organism exposed, the age of the organism and the chemical environment. Metals have been known to cause lethal and non-lethal stress to fish and other aquatic organisms in water column and bottom sediments, cancer and immunological 74 changes, and to induce morphological transformations in organisms including chromosome breakage (Helsel et al., 1979; B.C. Environment, 1992; Kushner, 1993). Gi l l damage to fish is usually a result of acute metal toxicity. According to Davies (1986), a breakdown of physiological metabolism and other biochemical functions due to metal toxicity can lead to chronic effects on aquatic life such as long-term lethality, effects on reproduction, growth and physical/behavioral abnormalities. At subtoxic levels, cadmium has been known to render commercial fish and shellfish highly unmarketable (Wilber and Hunter, 1977). Owe et al. (1982) found lead, zinc, copper and cadmium concentrations in parking lot runoff waters of New York State to be higher than the U.S.EPA water quality standards for aquatic life protection. Other effects of metals have to do with bioaccumulation and related food chain effects, and their persistence in the sediments of receiving water, as well as possible entry into the food chain. Metals in sediments are easily consumed by benthic biota. These metals, in turn, can subsequently travel up the food chain where they can impact other organisms. Iron and manganese act as important scavengers as they are able to incorporate other metals into sediment oxide coatings (Helsel et al., 1979). Other sediment metals of interest are zinc, lead, copper, cadmium, chromium and nickel since they are more toxic. Again, local metal data is required to evaluate their impacts on receiving waters. Organic Contaminants. As indicated in an earlier section, some researchers have found organic contaminants in runoff from heavily used highways and parking lots to be more polluted than the general urban runoff. Hoffman et al. (1984) found the annual load of polynuclear aromatic hydrocarbons (PAHs) to account for more than 50 percent of pollutants to Pawtuxet River in a study on 1-95 in Rhode Island. Owe et al. (1982) found hydrocarbon concentrations in runoff from a mall parking lot in New York State to exceed the recommended U.S. EPA water quality 75 standards for aquatic life protection, whereas Stotz (1987) in Germany found PAHs concentrations in highway runoff to be 50 to 60 times that of the receiving waters. Even though the data for organic contaminants in highway runoff is limited and the few existing data sets pertain to oil/grease and hydrocarbons, the impacts of organic contaminants on receiving water and organisms have been documented. According to Schueler (1987), oil and grease contain a wide variety of hydrocarbons. Of all the hydrocarbons, aliphatic hydrocarbons are generally less toxic to aquatic life and are more biodegradable. Polynuclear aromatic hydrocarbons (PAHs) can bioaccumulate in the food chains and tend to be more toxic to marine life (Fam et al., 1987). In a study in Jamaica Bay, New York in 1977, Tanacredi found aromatic hydrocarbons from waste petroleum in tissue extracts of marine benthic organisms. Some of the organic pollutants found in urban/highway runoff are toxic to human and aquatic life. In B.C., Canada, hydrocarbons found in stormwater, according to Larkin (1995), were predominantly aliphatic and particulate associated. Some organic contaminants are carcinogenic and many can bioaccumulate in the tissues of plants and animals over an extended period of time (B.C. Research Corporation, 1991). The bioaccumulation of these organic contaminants in plant and animal tissues may be harmful to the organisms directly or indirectly and may subsequently travel up the food chain. Other organic contaminants such as anti-sapstain chemicals and plasticizers can adsorb to sediments, where anti-saptain chemicals can persist i f protected from photolysis, and plasticizers can accumulate in the sediments and bioaccumulate in aquatic life (Cole et al., 1984). Nutrieiits. The direct impact of nutrients (nitrogen and phosphorus) on water quality is minimal compared to their indirect effects. Some of the direct impacts of excessive nutrients on water quality are increased chlorine demand due to presence of ammonia, possible increase of nitrate content of the water with 76 potential methemoglobinemia problems, and possible coagulation of particulate in the water with iron and aluminum salts due to the presence of phosphate species, especially condensed phosphates (Lee, 1970). The indirect impact from nutrient inputs can result in accelerated eutrophication which can lead to water quality deterioration through excessive growth of nuisance aquatic plants (Lee, 1970; Schueler, 1987). For example, the presence of algae can cause an increase in turbidity and reduce light penetration in water. In water treatment plants, their removal usually involves addition of coagulants like iron and aluminum salts and filtration through sand or other media filters, all of which can be costly. Lee (1970) also found algae and other plant remains present in water to react with chlorine species, leading to formation of organochlorine compounds and oxidation products. This increases the water chlorine demand. Similarly, there is a potential for the presence of algae and other organisms in a water supply to protect pathogenic organisms from chlorine destruction in drinking water. Other water quality problems, such as depressed dissolved oxygen concentrations, release of toxins, fish kills, water discoloration and odour generation, are also apparent (Lee, 1970; Schueler, 1987). pH/Electrical Conductivity (EC). The pH of a solution is a measure of the hydrogen ions (FT) present, whereas electrical conductivity (EC) is related to the dissolved mineral salts concentration. Changes in both the pH and EC of the water can affect aquatic biota. According to Kushner (1993), the outer layer of micro-organisms, as well as all cells, have a net negative charge due to carboxylic acids and phosphate groups. These outer cells usually attract metal cations. With the lowering of pH, however, the available hydrogen ions (H+) compete with the metals for sites. The microbial organisms, in an effort to maintain a near neutral internal pH when surrounded by extremely low pH values, are exposed to stressful conditions which might affect their reaction to toxins. For example, Hall and Anderson (1988) reported an increased 77 metal toxicity to daphnia with a decrease in pH from 8 to 5. Similarly, a decrease in pH tends to increase the solubility (or bioavailability) of toxic metals. Other effects of pH are (Kushner, 1993): • pH changes can affect metal ion hydroxylation, i.e. metal cation of form M 2 + at lower pH will add OH groups as pH increases resulting in M(OH) + , M(OH) 2 , M(OH) 3- and M(OH) 4 2 \ • pH changes can affect metals speciation which will subsequently affect the attraction of these metals to bacterial surfaces. The overall impact of increasing or decreasing pH depends not only on the type and species of metal, but also on the micro-organism (Collins and Stotzky, 1989). For example, as pH rises, the toxicity of dissolved zinc to fish increases. But at a pH of 8.5 or greater, the dissolved zinc is replaced by zinc precipitate which has low toxicity to fish (Bradley and Sprague, 1985). Unlike pH, EC, which measures salinity of the solution, has a more direct impact on both the receiving water and aquatic organisms. Chloride and salts of calcium, magnesium, sodium and potassium are responsible for increase in the salinity of a solution. According to Schueler (1987), chlorides at high concentrations are toxic to many aquatic micro-organisms. Manahan (1984) attributed the detrimental effects of salinity on aquatic biota to osmotic impacts. Other possible effects of increased salinity on receiving waters are (Marsalek, 1986): • prevention of mixing due to establishment of density gradients; • bottom layer stagnation from lack of mixing; • pH increases; and • shifts in aquatic communities within the ecosystem. These effects depend on a range of other factors such as alkalinity (acid-neutralization capacity) and water hardness. Natural waters in the Greater Vancouver Regional District (G.V.R.D.) area 78 have both low alkalinity and hardness values (B.C. Research Corporation, 1991) and may therefore be more susceptible to pH changes. 2.6 Stormwater Models For the general public, highway stormwater runoff polluting a nearby surface water body spells an environmental hazard. Research to date has established the influence of the following factors on water quality: differences in precipitation among sites, precipitation patterns, volume and intensity, ADT, local surrounding land use, geological characteristics, highway maintenance practices and highway drainage design (Aye, 1979; Horner et al., 1977; Kobriger et al., 1981a; Asplund et al., 1980; Meinholz et al., 1978). The extent of water quality degradation resulting from these factors varies from site to site. For the engineers, hydrologists and planners involved in pollution management, this creates a major problem in trying to design treatment facilities for the different sites with limited budget/data. But, with the availability of computer models, both time and costs can be reduced considerably. Models, when properly used, can be a useful tool in predicting environmental problems and in developing solutions to deal with the problems. Computer models provide the ability to describe the performance of an entire stormwater system in much greater detail than would have been possible using hand calculations (Urbonas and Stahre, 1990). Where practical solutions are not available, models can suggest alternative remedial actions. However, the precision and accuracy of the results from a model depends on the goodness of the input data, how representative the model and the experience of the user. Due to the great variety of sources of pollution, including urban and highway stormwater runoff, there have been a number of models developed for both pollution and control planning purposes and system design. 79 For highway stormwater runoff, three predictive models, namely a deposition-accumulation/washoff model (Sartor and Boyd, 1973; Sylvester and De Walle, 1972; Meinholz et al., 1978; Kobriger et al., 1981a), a pollutant loading over time model (Asplund et al., 1980; Chui et al., 1982), and a linear regression pollutant forecasting equation (Kerri et al., 1985), have been proposed for highway stormwater runoff quality estimation. The earlier deposition-accumulation/washoff model developed by Sartor and Boyd (1973), and Sylvester and De Walle (1972) was later refined by Meinholz et al. (1978) and Kobriger et al. (1981a) into what is now known as the Envirex Model. Predictive models for the quantity and quality of stormwater from highways have limitations which must be recognized. For example, some of the obvious limitations of the Envirex Model are (Kobriger et al., 1981a; Kerri et al., 1985): • The assumption that highway area is uniform; • Long antecedent dry periods and overlapping storms make it difficult to estimate prestorm surface load; • Model estimates of the initial surface load at the start of the storm have to be fairly accurate; • Construction activities/effects are hard to account for; • The exponential washoff equation is insufficient for modelling pollutants that are continuously added to the runoff from the undercarriage of VDS, atmospheric fallout and surrounding land use activities. Like the Envirex Model, the pollutant loading over time model, or the Washington State model, has its own limitations. Even though the model works best over long-term application (monthly or annually), some of the obvious limitations are (Chui et al., 1982): • Inability to accurately model individual storms; 80 • Inadequacy of the model in accommodating sanding operations and atmospheric fallout; • Model does not account for road surface pollutants prior to the storm events. Consequently, attention will be focused on linear regression modelling. The linear regression pollutant load forecasting equations developed by Kerri et al. (1985) use pollutant constituents such as lead, zinc, cadmium, oil and grease as dependent variables and the number of vehicles both before and during the storm as independent variables. They found that before the storm, when road surfaces were dry, there was an increased adherence of materials to the engine, undercarriage and wheel walls of the vehicles. Similarly, the road surfaces are constantly being swept by traffic-generated turbulence. The result was poor correlation between pollutants and ADT during dry periods. However, an analysis of lead vs. number of vehicles during storm events showed better correlation. Studies by Kerri et al. (1985) with other constituents as dependent variables and vehicles during storm as independent variables resulted in these two equations: C L = a + b (VDS) C L - a + b (TR) where: C L = Cumulative constituent or pollutant load a & b = Regression coefficients where "a" represents an initial load (g), and "b" represents the constituent washed off during the storm event VDS = # of vehicle during the storm from start to the end of runoff TR= Total residue The use of this pollutant forecasting equation methodology may be time consuming as well as costly. The determination of the pollutant loads requires knowing both the storm hydrograph and 81 the pollutograph (plot of pollutant concentrations against time) during the runoff period (Kerri et al., 1985). The accuracy of the models varies widely and the models are, at best, only as good as the data used in their validation. At this stage, most of the models are used in evaluating the impacts of increased traffic, highway configuration changes or other design changes, as well as surrounding land use activities and highway maintenance practices on the loading of solids and other pollutants to the receiving water. Where there is potential for problems, stormwater management and implementation programs can be initiated. 2.7 Stormwater Management Measures According to Burch et al. (1985) and B.C. Research Corporation (1991), there are three types of management practices commonly used to reduce highway runoff pollution: 1. Source Management Measures Planning measures Design and operations Regulations 2. Post-Deposition Measures Applied Prior to Runoff Street cleaning Debris removal and spill clean up Ditch maintenance 3. Post-Runoff Measures Infiltration system Vegetative controls a. Grassed channel 82 b. Filter strips c. Overland flow Wetland Detention ponds Filtration systems Flow attenuation/alteration Catch basins More attention will be focused on grassed channels (post-runoff measures) in this investigation. 2.7.1 Source Management Measures Data to quantify the effectiveness of source management measures for the mitigation of highway stormwater runoff pollution are difficult to obtain and, hence, are scarce. However, most of the source management control measures have been rated moderate to probably high in their ability to remove pollutants from highway runoff. Transportation planning measures such as mass transit, car pools, bicycles, walking and toll installation are all source management measures that have potential benefits in reducing highway traffic volume and, hence, the level of pollution. Other types of transportation planning with noticeable impacts on mitigating runoff pollution are urbanization control policies, land use monitoring, public education, increasing the percentage of permeable areas in watersheds to reduce flow, erosion and sediment control to reduce suspended solids in runoff, and building lanes restricted to car pools (Burch et al., 1985). Lanes restricted to car pools are widely used in the U.S.A., especially on California and Washington State highways. In B.C., this restricted lane practice is being considered in some areas, is under implementation along the Trans-Canada Highway and is fully implemented on the 83 Barnet Highway. No data are available yet on its effectiveness in both traffic and pollution reduction. Of all the transportation planning measures, erosion and sediment control programs are the most widely implemented. The most frequently used erosion and sediment control practices, according to Lorant (1992a), are road stabilization, sediment fences, dikes, diversions, sediment banks, waterway protection, vegetation, mulches, and temporary channel lining. Proper implementation of any of the above-mentioned erosion and sediment control practices will significantly reduce the level of suspended solids in stormwater runoff (Burch et al., 1985; B.C. Research Corporation, 1991; Lorant, 1992a). Apart from transportation planning, design and operation of highways can affect traffic volume, density and duration, which will influence the amount of pollutants on highway surfaces. According to Kobriger and Geinepolos (1984) and Lorant (1992a), design features such as barrier elimination (tends to trap pollutants) and installing signal synchronization lights (to control traffic and minimize stop-and-go traffic patterns) help in reducing the level of traffic-generated pollutants. The impact of curbing on pollution is still, debatable. Even though Kobriger and Geinepolos (1984a) found curbed highways to trap sediments and associated pollutants, their finding contradicted an earlier study by Chui et al. (1981) in Washington State. In an exploratory study on over-the-shoulder runoff sampling, Chui et al. (1981) found higher concentrations of pollutants in highway runoff without curbs than with curbs. They attributed this higher pollution in non-curbed highways to the ability of runoff to carry practically all of the solids from the highway, whereas in curbed highways flow along the curb results in sedimentation, especially of the larger particles. However, earlier research studies indicated that most pollutants are associated with finer particles rather than the larger particles indicated in the study by Chui et al. (1981). The impact of curbed versus non-curbed highways on pollution 84 generation requires further study to clear this apparent contradiction. On the other hand, it is very important to control traffic flow, since the idling of vehicles in traffic congested highways results in incomplete fuel combustion and more pollution. Apart from the highway design, some of the operational features that can help in highway pollution reduction are applying the right amount of fertilizers and pesticides/herbicides to maintain the grasses in drainage ditches at the right time, use of biodegradable, less toxic and water insoluble pesticides/herbicides, timing application to coincide with periods of low air movement, and proper maintenance of the grasses (mowing and debris control/removal) to enhance their pollutant removal capacity (Burch et al., 1985; B.C. Research Corporation, 1991; B.C. Environment, 1992). Finally, traffic laws and regulations can be used to control speed limits, traffic volume control and vehicle mix, vehicle emission standards like the introduction of vehicle AirCare. emission policy in G.V.R.D., litter control by-laws as seen on B.C. and most U.S.A. highways, and mandatory vehicle inspection. These source management measures can be used to reduce the pollution levels in urban highways directly or indirectly (Burch et al., 1985). Traffic laws and regulations such as vehicle emission controls and leaded gasoline restrictions have contributed to air pollution reduction as well as reductions in highway pollutant deposition. For example, the projected 33 percent decline in carbon monoxide and smog-forming contaminants by the year 2000 has been attributed to better engine technology and the enforcement of the current AirCare vehicle emission standards in the G.V.R.D. (G.V.R.D., 1993a). Other measures that can be used to reduce both air and water pollution are: Mandatory inspection of all vehicles; Encourage cycling and walking; Improvement in mass transit systems; Exclusive car pool and bus lanes and area-wide car pool programs; 85 Development of parking facilities for park-and-ride users; Encourage employers to stagger working hours; User-pay charges such as toll booths and higher parking rates for single-occupancy vehicles; Burning of cleaner fuels; Requiring car makers to increase fuel efficiency of automobiles; Enforcement of litter control by-laws; Above all, public education on the impacts of automobiles on the environment (Kobriger and Geinepolos, 1984a; Lorant, 1992; G.V.R.D., 1993a). Where the enforcement of these measures proves to be inadequate, the law makers could perhaps introduce the alternate-day odd and even number license plate driving as in Lagos, Nigeria and Mexico City, Mexico. (These measures were used more to reduce traffic congestion than automobile associated pollution.) By reducing traffic volume, however, the regulation is also reducing the level of pollutants deposited on highways. 2.7.2 Post-Deposition Measures Applied Prior to Runoff According to B.C. Research Corporation (1991) and Lorant (1992), most of the current street cleaning operations are largely to improve aesthetic appearances of the streets through removal of litter, debris and, to some extent, dirt. As a result, current street cleaning operations have been found to be inadequate in removing pollutant-laden fine particles. Like the street cleaning operations, debris removal and accident clean-up operations are usually for aesthetic and safety purposes, but some pollution reduction benefits result. Most transportation departments have emergency clean-up operators that will perform thorough cleaning, thereby reducing any polluting effect of the hazardous material to the environment. 86 Ditch maintenance is not performed as a direct pollutant reduction measure, but to restore proper side slope configuration and grade to ensure adequate drainage through cleaning and shaping of roadside ditches. The advantages of removing sediment deposits and vegetation from ditches are removal of contaminated sediment deposits and pollutant-laden vegetation from the ditches and prevention of pollutants from re-suspension by later storm events, while providing a proper flow channel. The only disadvantage is stripping the ditch of its vegetative cover, thereby increasing both flow velocities and erosion potential while decreasing the pollutant removal capacity of the ditch (Kobriger et al, 1981; Wang et al., 1982; Little et al., 1983; Kobriger and Geinepolos, 1984a). Consequently, ditch maintenance that removes polluted sediments and pollutant-laden vegetation without stripping the vegetative grass cover is the preferred choice. Properly designed and well maintained grass drainage ditches have been found to be. effective in highway stormwater runoff pollution reduction (Wang et al., 1982; Little et al., 1983; Kobriger and Geinepolos, 1984a; Yousef et al., 1985), unlike the post-deposition measures applied prior to runoff described above, to which Sartor et al. (1974), Kobriger et al. (1981) and Burch et al. (1985) have given an overall performance rating of "low to moderate" in their pollutant removal capacity. 2.7.3 Post-Runoff Measures As in the previous two sections, post-runoff measures have overall performance efficiencies in pollution reduction which range from low to high. The effectiveness of each post-runoff measure in pollutant removal depends on its practical application to the environment as well as the type and form of pollutant. A brief review of the post-runoff measures will be undertaken in the following sub-sections. 87 Infiltration Systems. Infiltration systems usually involve temporary storage of surface runoff and the eventual percolation of the runoff into the ground. Infiltration basins can be used (Lorant, 1992): to control or reduce the amount (or rate) of stormwater runoff; to reduce the level of pollutants or completely remove them from stormwater runoff; to help recharge aquifers and other ground water resources. Infiltration systems can only be recommended in areas where the soil is relatively uncompacted, permeable and the ground water table is at least 3 m below the basin bed (Becker et al., 1973). Infiltration basins have been used in Long Island, New York and the Fresno area in California, primarily for highway stormwater runoff (Kobriger and Geinepolos 1984a; Lorant, 1992a). Some researchers like Tourbier and Westmacott (1980) recommend basins be designed to store and recharge most of the annual rainfall, whereas the designers in the Fresno area used a design storage capacity of two consecutive 24-hour storms with a 10-year frequency as part of their design criteria. The effectiveness of infiltration systems in pollutant removal is still questionable. Even though little data is available on the pollutant removal ability of infiltration systems, Burch et al. (1985) hypothesized their overall performance rating to be high. But Lindsey et al. (1992) found infiltration basin performances in Maryland to be poor, with the percentage of infiltration basins functioning as designed declining from 48 percent at the beginning to 38 percent after four years. In some cases, only about one-fourth of the basins were working for all of the 4-year period of study, and nearly half were not working either at the beginning or at the end of the research period. They attributed this dismal performance to poor maintenance. On the other hand, Whipple and Hunter (1981) questioned the common assumption that pollutants settle out in amounts proportionate to their respective particulate concentrations. 88 The pollutant removal processes associated with infiltration basins are physical, chemical and biological in nature and include adsorption, filtration, chemical precipitation, decomposition, chemical adsorption and biological transformation/degradation. As a result, the effectiveness of an infiltration basin is site-specific. One of the disadvantages of infiltration basins is sediment clogging, but with proper maintenance it can be avoided. The advantages are: ground water recharge; runoff attenuation; erosion control; less degradation of water quality; reduced need for curbs and storm sewers (Harwood, 1986; Lindsey et al., 1992). These make infiltration basins very attractive for stormwater management even though their efficiencies vary. The consensus is that when infiltration basins are properly maintained, they can be one of the most economically feasible and efficient means of implementing pollution reduction in highway runoff (Harwood, 1986; Lindsey et al., 1992; Lorant, 1992a). Wetland. Wetland is the terminology given to low-lying lands where the water table is at or near the surface or the ground is covered with water. Biological processes are involved in pollution removal but not all processes are equally effective. According to Burch et al. (1985) and Lorant (1992a), there are many factors that limit wetland effectiveness in pollution reduction namely: the seasonal and sporadic nature of stormwater runoff does not provide a reliable water supply required for wetland vegetation maintenance; flushing and shock loading effects of runoff with its associated pollutants; seasonal temperature variations which affect plant activities. 89 Burch et al. (1985) concluded that wetland, under certain conditions, can be appropriate for treating highway runoff, but that only under certain conditions can their overall performance rating be high. There is almost no research data on wetland pollutant removal efficiency. However, it seems that the types of wetland most suitable for highway pollution reduction are the moss-lichen type (organic substrate/surface saturated), the emergent type (organic substrate/shallow or surface saturated), the scrub-shrub type (organic substrate/shallow or surface saturated), and the forest type (organic substrate/shallow or surface saturated), according to Lorant (1992a). Detention Ponds. Detention ponds (wet and dry) are used for stormwater quantity and quality management. The design and the basin size relative to the drainage area are the key parameters affecting performance. Dry basins are relatively ineffective for reducing pollutant concentrations, whereas wet ponds have great pollutant removal potential (Burch et al., 1985). Wet basins use biological processes to remove soluble nutrients and physical sedimentation of the particles to remove the solids (U. S. EPA., 1983; Lorant, 1992a). In the Maryland studies, sediment traps were found to capture at least 92 percent of the sediments and Lynard et al. (1980) found that smaller storms produced better removal efficiency. Whipple and Hunter (1981) found that a 1.8 m deep pool removed substantial amounts of the common pollutants after 32 hours of settling. They found that suspended solids, lead and hydrocarbons were almost completely removed, whereas BOD, copper, nickel and zinc removal was more variable. Detention ponds remove most of the pollutants, especially solids/pollutant associated solids through sedimentation, though the wet basins also have nutrient removal ability through biological processes. According to Taylor et al. (1980), the addition of alum coagulation 90 improved total suspended solids removal from about 50 percent to 90 percent. However, in many situations, detention ponds may not be suitable for highways due to cost and lack of room. Filtration Systems, Flow Alteration and Porous Pavements. Filtration systems are not very effective for highway stormwater runoff pollution mitigation. They are usually temporary measures used for sediment control purposes, especially larger particle size suspended solids. The advantages are coarse sediment removal and runoff flow attenuation. The disadvantages are poor structural strength, frequent undercutting, end flow and constant maintenance to remove trapped sediments (Tourbier and Westmacott, 1980; Lorant, 1992a). Even though little research data is available, Burch et al. (1985) hypothesized their overall performance rating to be low. Flow alteration systems work by altering the hydraulic flow through diversion, drop structures, slope drains and level spreaders, and subsequently causing sediments to be deposited. Flow alteration systems are mainly used for erosion control purposes rather than for highway runoff pollution reduction (Lorant, 1992a). They were given the same low overall performance rating by Burch et al., (1985). There is no performance data on permeable or porous pavements. A major problem associated with porous pavements is the clogging of the pores which reduces their effectiveness. Further research is required on the use of porous pavements in highway stormwater runoff pollution reduction. Catchbasins. Catchbasins can be effective in preventing clogging of sewers by trapping coarse solids and debris, however they are very ineffective in preventing pollutant-laden fine particles from entering the sewer or receiving water bodies. Another disadvantage is the tendency of some of the materials to be re-suspended at a low to moderate flow rate. Similarly, catchbasins are small in size and require constant maintenance, usually at high cost, to be effective. Their 91 efficiency tends to improve with increasing depth, but Burch et al. (1985) and Lorant (1992a) rated catchbasin pollution removal efficiency to be low. Vegetative Controls. Vegetative controls involve using grass surfaces to reduce pollutant levels in stormwater runoff flowing over the grass. Vegetated surfaces reduce runoff velocity and enhance sedimentation, filter the suspended solids, increase infiltration rates and remove runoff contaminants. The research reports of Mar et al. (1981), Wang et al. (1982), Little et al. (1983), Horner (1988), and Khan et al. (1992) in Washington State and Yousef et al. (1985) in Central Florida on the performance of grass drainage ditches showed varying removal efficiencies. The pollutant removal efficiency of the grass drainage ditches depends on factors such as the residence time, type of vegetation, soil type, type and nature of the pollutant, the slope of the ditch channel,, width and length of the grass drainage ditch, flow depth, ground water depth and the runoff volumetric flow rate, as well as the type of maintenance (Wang et al., 1982; Yousef et al., 1985; Horner, 1985/88; B.C. Research Corporation, 1991; Khan et al., 1992). In the past, most highway drainage ditches were designed for speedy removal of runoff from highway surfaces as well as from ditches. But with increasing realization of the pollution potential from highway runoff, new drainage ditch designs aim at efficient runoff removal as well as pollution mitigation. Of all the factors affecting drainage ditch pollutant removal effectiveness, drainage ditch length has been studied the most by Wang et al. (1982) in Washington State. They found that channelling runoff water through grass drainage ditches of length 30 metres or longer resulted in a 60 to 80 percent reduction in the amount of lead, zinc, copper and suspended solids. Lead was removed more effectively than other metals due to low solubility and strong association with solids. Several years of related studies resulted in the conclusion that grass drainage ditch pollution reduction was relatively high and increased with 92 ditch length. According to Wang et al. (1982), Little et al. (1983) and Horner (1985), a 60 m long slightly sloped grassed drainage ditch channel can remove 80 percent or more of the pollutants in highway runoff. What was not explained was why at some sites the same efficiency in pollutant removal was achieved at shorter distances (less than 60 m). Similarly, nothing was said about the grass conditions at these sites, the type of maintenance operations such as mowing or debris removal involved, and the frequency of maintenance while the study was in progress. But Horner (1985) reported in the Environmental Criteria Manual - Volume 2 that, in the Seattle area, vegetated channels in previous studies had not been mowed or scraped. As a result, a more detailed evaluation may be required to accurately assess the effectiveness of the grass drainage ditches. For example, Khan et al. (1992) realized the importance of regular maintenance at the 48th Avenue W. biofiltration swale in Seattle and suggested the following maintenance operations be implemented: Keep flow spreaders free of obstruction. Sediment removal to avoid covering the grasses. Litter/debris removal to avoid blockage. Regular mowing to maintain proper height and general health of the grass. Fertilizer application to encourage grass growth, where applicable. Reseeding of bare soils/poor growth grass areas to avoid exposure to erosion. Yousef et al. (1985) conducted studies on the removal of highway contaminants by roadside swales in two locations in Florida. They sampled individual events followed by another set of tests with continuous flows. Their results tend to support the work of Wang et al. (1982) in confirming that vegetated channels are effective in pollution reduction. However, the grass channels' pollution removal efficiencies obtained in the study by Yousef et al. (1985) were not as good. Also, the lack of runoff at the end of their Maitland site drainage pathway suggested that 93 not enough stormwater runoff volume was used in the analysis. The selected site should generate enough rainfall-runoff volume to be considered "appropriate" for this type of research. Insufficient rainfall-runoff volume will result in limited amount of pollutants being washed'off the highway surfaces. Similarly, they failed to address the impact of maintenance on the grassed drainage ditches during the duration of the experiment. Grass conditions in drainage ditches affect their pollutant removal ability. Finally, the level of traffic affects the quality of highway stormwater runoff. There was nothing said in their report as to the A D T values along the highways where the research was conducted. In Canada, the only known research conducted by Lorant (1992a) in Ontario was also lacking detailed evaluation. Apart from sampling only a small data set (8 samples), the integrity of the grass channel was seriously affected when the grassed medium was reduced from 4.5m to 0.8 m. As a result, the research study was unable to address the pollutant removal effectiveness of the grass drainage ditches. In summary, based on known studies of grass drainage ditch pollution removal in Washington State by Wang et al. (1982); Little et al. (1983); Mar et al. (1981); Horner (1985/1988); Khan et al. (1992) and in Florida by Yousef et al. (1985) of the three vegetative controls—namely vegetated waterways, filter strips, and overland flow-vegetated waterways which include a ditch, channel or swale are the most efficient in pollution reduction. Nevertheless, obvious differences exist between the known research studies as to what played the dominant role in pollution removal. According to the Washington State studies, the length of the grass drainage ditch tends to be the main focus of pollution removal. Yousef et al. (1985) in their research study in Florida found a thin grass cover of 20 percent or less to be effective in decreasing some of the contaminants such as soluble N0 3 " and N H 4 + than a thick grass cover of 80 percent or more. They attributed the decline in pollutant removal by thick 94 grass cover to increase in organic debris such as grass clippings, debris and litter as well as to possible reduction in soil sorption capacity due to organic deposition. If enough organic matter is able to cover the bottom soil of the grass drainage ditch, upon decomposition it can contribute to nitrogen and phosphorous loads. Similarly, grass clippings and debris can clog outlets of channels, collect in piles and encourage channelization which may lead to potential erosion of the bottom of the ditches (Khan et al., 1992). Based on the Washington and Florida State research studies, Burch et al. (1985) were able to give vegetated ditches an overall performance rating of high in their pollution removal ability. But their effectiveness in pollution reduction varies from site to site and depends on the quality of design and installation as well as subsequent maintenance. Design Characteristics. The design of grass drainage ditches is usually treated as an open . channel flow problem and uses Manning's equation to compute velocities and flow capacity. According to Khan et al. (1992), the key advantage in using Manning's equation in grass drainage ditch design is that it is easy to use and it incorporates the key design performance variables with the exception of ditch length. According to Khan et al. (1992), some of the hydraulic design variables that affect grass drainage ditch performances are: maximum velocity, for which Horner (1988) recommended a velocity of 1.5 ft/sec (or 0.5 m/s), even though the previous research study suggested that velocity in excess of 0.9 ft/sec (or 0.3 m/s) can reduce performance; design flow rate for a 2-year, 24-hour storm event; water depth not more than one-third of the depth for tall grass height of 9 to 12 inches (23 to 31 cm) and one-half for mowed grass height with a maximum water depth of 3 inches (or 8 cm); and, finally, the channel roughness factor "n". Selection of the proper Manning's coefficient "n" is difficult and controversial among design engineers (Horner, 1988). Although a design value was not recommended, it was suggested that a minimum Manning's 95 coefficient of 0.2 be used for grass drainage ditch design when it is to be used for water pollution reduction (Khan et al., 1992). Based on several studies conducted in Washington State, Homer (1988) and Khan et al. (1992) made the following recommendations for the hydraulic design of grass drainage ditches: Installation of flow spreading to reduce channelization, excessive bottom velocity, channel bottom scouring and increase contact with vegetation. Installation of bypass channels for flows greater than design flow rate to avoid excessive channel erosion, vegetation destruction and channel foundation destabilization. Design grass channels to have a hydraulic retention time of at least 9 minutes for effective pollution reduction. Use of wetland in areas with high ground water tables, slight slope or high winter base flows. Use trapezoidal design, even though rectangular, semi-parabolic, parabolic and v-shaped designs are also possible. Poland (1975) preferred a parabolic design because it is commonly found in nature. Avoid grass channels designs with width greater than 7 ft. (or 2.1 m). Use small design widths to encourage flow spreading and discourage channelization. Longitudinal slopes of between 2 and 4 percent with a maximum of 6 percent are recommended to provide adequate velocity, retention time and reduce erosion potential. Use design side slopes of 3 horizontal to 1 vertical to avoid erosion and provide more stormwater detention. 96 Install energy dissipating device such as "rip-rap" pads at the inflow point to avoid scouring. Proper vegetation types depending on vegetated channel slope, purpose of the grass channel and the soil characteristics. When a grass drainage ditch is properly designed and adequately maintained, some of the potential benefits to be expected, according to Lorant (1992) and Khan et al. (1992), are: • Nutrient concentrations reduced by soil and vegetative processes; Bacterial breakdown of hydrocarbons through degradation; • Vegetative filtration/sedimentation of solids; • Heavy metal complexing by soil adsorption and biological assimilation; • Reduction in runoff velocity/energy; • Not too expensive to construct and maintain; • Increase infiltration and soil moisture availability to vegetation; • Tends to fit well into the highway design configuration; • Potential erosion reduction. Also, part of the design criteria should include proper selection of vegetative cover depending on the site, climate, soil type/conditions and geographic location. According to Horner (1988), Lorant (1992/1992a) and Khan et al. (1992), vegetative cover selection criteria should include: Erosion control effectiveness of the vegetation; Commercial availability of vegetative cover materials; Installation and maintenance costs; Tolerance to frost and drought; Potential fire hazard; 97 Site and climatic conditions adaptability; Required maintenance; Longevity; Runoff control and pollution reduction suitability; Vegetative cover compatibility with the local/surrounding landscape; Adequacy of vegetative cover to accommodate design slope and maximum design velocity. Although grass drainage ditches are generally effective in highway pollution reduction, there are definite difficulties associated with them. According to Lorant (1992a) and Khan et al. (1992), some of these are: • Lack of agreement among designers on Manning's coefficient "n"; • Grass ditches can be damaged by snow plowing and off-street vehicular parking; Ditches require constant maintenance such as mowing, seeding, debris and litter removal, erosion and sediment control and fertilization; • High cost of land acquisition in some areas; • Occasional channel clogging by fine particles. Despite the above disadvantages, a reasonable conclusion is that grass drainage ditches should be an integral part of highway design. Having said that, more research is needed to evaluate grass drainage ditch design variables and their pollution reduction effectiveness. Table 2-3 summarizes the pollutant removal effectiveness of all these pollution management measures. The extensive literature review conducted on grass ditches was done in order to have a basis for comparing experimental obtained data later. The present dissertation takes, as part of its focus, a further examination of these questions. 98 2.8 Summary Highway surface pollutants such as particulate matter, heavy metals, PCB/pesticides, inorganic de-icing salts, organic matter, nutrients, pathogenic bacteria, asbestos, rubber, oil and grease and special vehicle additives are present in highway runoff at some concentrations. Most of these highway runoff pollutants result from vehicular traffic and are of concern to water quality. The most common sources of these highway runoff pollutants are from engine wear-and-tear associated with traffic, wind-blown materials from surrounding land use, sanding and de-icing operations in temperate regions, dustfall or atmospheric depositions, highway maintenance operations or lack thereof, and pavement deterioration. On reaching the highway surface, pollutants are removed by two principal mechanisms depending on the road conditions. The primary removal mechanism during dry periods is a. combination of traffic-generated turbulence and naturally occurring winds in conjunction with the mechanical scrubbing actions of automobile tires, which continuously sweep off pollutants from the travelled lanes. During wet periods, pollutant transportation by stormwater runoff is the main removal mechanism. Some of the factors that affect the amount of pollutants on highway surfaces and their potential impact on nearby water bodies are the effect of surrounding land use, ADT/number of lanes, climatic effects, parameter concentration, time of year, drainage type, pavement/design characteristics, and antecedent number of dry days. The potential impact of highway pollutants on receiving water varies from site to site and depends on the sensitivity of the receiving water. They range from aesthetic deterioration, dissolved oxygen depletion, accelerated pathogenic bacteria concentrations and increase in suspended solid concentrations to accelerated eutrophication (due to high nutrient) and metal/pesticide toxicity. Both planners and design engineers have tried to use different types of models, such as the Envirex Model, regression forecasting equations and the pollutant loading 99 model to estimate the extent of water pollution problems. Computer models are, however, only as good as the data used in developing/testing the model and the experience of the user, and there are very little or no highway data in Canada. The planning phase for stormwater pollution control includes stormwater management measures such as source control, post-deposition measures applied prior to runoff and post-runoff measures. The effectiveness of these stormwater management measures in pollution removal ranges from low/moderate for source control measures to high for post-runoff measures, especially for grass drainage ditches, as can be seen in Table 2-3. The advantages of post-runoff measures for pollution reduction generally outweigh their disadvantages. But to achieve effective pollution removal, management programs have to encompass source management measures, post-deposition measures applied prior to runoff and post-runoff measures. A l l of these measures require proper planning, accurate design and installation and adequate maintenance to ensure optimal performance and longevity. Similarly, periodic evaluations are required to check on the effectiveness of the management programs being used and updating them as necessary. In the next chapter, the methodology used in this research project to assess the effectiveness of grass drainage ditches will be described. 100 Table 2-3. Summary of Highway Stormwater Management Measures Pollutant Removal Effectiveness Management Measure Particulate Metals Pesticides Organic Matter Hydrocarbon Nutrients Pathogenic Bacteria Overall Rating Source Management • Planning M M M M M M P M • Design & Operation M M M M M M P M • Regulations H H H H H H PH Post-Deposition • Street Cleaning L / M L L L L L L • Debris Removal L / H N A L / H L / H N A N A M • Ditch Maintenance L / M L L L L L L Post Runoff • Infiltration Systems H H H H H H H • Detention - wet basin H H N A H M / H M / H H • Detention - dry basin L / H L / H N A L L L L • Vegetative Controls - Grass channels H H M H H H H - Filter strips H H M H H / M H / M H - Overland flow H H M H H H H • Wetland L / H H N A M / H H M H • Catchbasins L L L M / L L L L • Filtration Systems M / L L L M / L L L L • Flow Attenuation M L L L L L L Source: Summarized from Burch et al. (1985) Note: 1) L = Low removal efficiency. 2) M = Moderate removal efficiency. 3) H = High removal efficiency. 4) PM, PH = Probably moderate and probably high removal efficiency. 5) NA = Not applicable. 101 Chapter 3 MATERIALS A N D METHODS 3.1 Experimental Overview As described in the previous chapter, many of the pollutants commonly detected in urban/highway stormwater runoff are considered to be harmful to receiving water bodies. However, the impact of these pollutants varies from site to site and with pollutant concentrations and loadings. To gain information on typical highway stormwater runoff in the G.V.R.D., samples were collected from two highway sites in the area. The two highway sites have different rainfall patterns, traffic volumes, surrounding land use activities and highway design configuration, even though the maintenance practices are • similar. Basic storm event information including total rainfall, duration of the sample event and time since the last precipitation, as well as runoff samples at pre-determined time intervals, were collected at both sites from April 1995 to September 1996. The collected rainfall and runoff samples were analyzed at the Environmental Engineering Laboratory at the University of British Columbia (U.B.C.). The parameters analyzed in this research project were total suspended solids (T.S.S.), trace metals, oil and grease. The trace metals of particular concern were cadmium (Cd), chromium (Cr), copper (Cu), iron (Fe), lead (Pb), zinc (Zn), manganese (Mn), nickel (Ni) and calcium (Ca). These metals were chosen because of their common association with vehicular-related urban/highway stormwater runoff pollution. 3.2 Site Selection Criteria The two sites selected for this research project were the Trans-Canada Highway (#1) near 102 the Cariboo exit on the Brunette River bridge deck in Burnaby and the New Westminster Highway (#91, the east-west connector) at the Canadian National Railway overpass deck near No. 9 Road in Richmond, as can be seen from Figure 3-1. The sites were selected mainly on the basis of convenience, but it is helpful to see how they compare to the following site selection criteria suggested by Aye (1979), Gupta et al. (1981a) and Environment Canada (1993): • Traffic characteristics; • Surrounding land use activities; • Precipitation characteristics / geographical location; • Pavement type/condition; • Drainage area and highway design characteristics; • Proximity to a receiving water body; • Highway maintenance practices; • Logistical considerations such as safety, easy accessibility, power availability, and any future construction/improvement plan activities. Surrounding Land Use/Traffic Characteristics. According to Aye (1979) and Gupta et al. (1981a), comparative sites for monitoring purposes should have similar traffic characteristics such as vehicular mix, number of exit/entrance ramps, number of lanes, acceleration and braking patterns, as well as adequate average daily traffic on the entire stretch of highway under consideration. With surrounding land use, on the other hand, it is harder to obtain similarities between two different sites. The watershed within which the Trans-Canada Highway site is located is 42 percent residential, 31 percent open/green area, 15 percent commercial/institutional and 5.5 percent industrial, according to Hall and Anderson (1988). Of the 5.5 percent industrial, Lawson et al. (1985) found truck-related operations to be the dominant business. The Trans-Canada site at the 103 Figure: 3-1. Burnaby And Richmond Site Locations 104 Brunette River bridge deck has a typical mix of trucks and cars, adequate exit/entrance ramps along the stretch of the highway, 2 traffic lanes with adequate travelling speed during non-rush hours, but stop-and-go traffic heading east during the evening rush hours (Onwumere and Russell, 1995). The site has more than the recommended ADT of 30,000 vehicles. In fact, the summer average daily traffic (SADT) for the Trans-Canada site for the period 1987 to 1991 was between 86,000 and 92,000 vehicles (B.C. Ministry of Transportation and Highways, 1992 and 1993), as can be seen from Figure 3-2. No ADT data was available from 1992 to the present. The New Westminster (Highway 91) Highway site in Richmond, at the Canadian National Railway overpass near No. 9 Road, has agricultural land use activities surrounding the entire stretch of highway with few industrial/residential activities. The right mix of vehicles such as trucks, light trucks and cars is provided by vehicles making a connection between Richmond in the west and Surrey, North Delta, Coquitlam, Port Coquitlam, Port Moody and New Westminster in the east. The New Westminster Highway has fewer exit/entrance ramps and less stop-and-go traffic heading west during evening rush hours than the Trans-Canada site. It has 2 traffic lanes with adequate travelling speed at all times, as well as more than the recommended ADT of 30,000 vehicles as does the Trans-Canada highway (Onwumere and Russell, 1995). In fact, the summer average daily traffic (SADT) for the Richmond site, according to the B.C. Ministry of Transportation and Highways (1992/93), for the period 1987 to 1992 was found to be between 22,000 and 50,000 vehicles, as can be seen from Figure 3-2. According to Onwumere and Russell (1995), the lower ADT at the Richmond site than at the Burnaby site may be the reason for the non-existence of stop-and-go traffic flow at the Richmond site, or it could be due to fewer exit/entrance ramps along the highway. Consequently, traffic-generated pollution at the Richmond site could be expected to be lower than at the Burnaby site, other factors being more or less equal. 105 D < L U 50000 ^ 40000 W 30000 20000 10000 1987 1988 1989 1990 1991 1992 YEAR FIGURE 3-2. S U M M E R A V E R A G E D A I L Y TRAFFIC A T BOTH SITES. (DATA F R O M B.C MINISTRY OF TRANSPORTATION & HIGHWAYS, 1992/93). 106 Precipitation Characteristics/Geographical Location. The amount or rate of highway runoff obtained depends on the amount and form of precipitation—whether as rainfall or snow. For the Burnaby site, the area's average annual precipitation range is from 1,600 to 1,700 mm per year, most of which occurs as rainfall from October through March (Hay and Oke, 1976). The highest mean monthly precipitation of over 300 mm occurs in November at the Burnaby Mountain (Environment Canada, 1995), as can be seen in Figures 1-3 and 1-4. The Richmond site follows the same precipitation trend as that observed at the Trans-Canada site. The highest mean monthly precipitation occurs in November at close to 200 mm at the Nature Park, as can be seen in Figures 1-3 and 1-4. The Richmond site has about 30 percent less precipitation than the Burnaby site with an average annual precipitation range of 1,100 to 1,200 mm per year (Hay and Oke, 1976). Pavement Type/Condition. The percentage of pervious and impervious areas within a highway drainage area affects the potential impacts of pollutants from highway stormwater runoff. Both the Burnaby and Richmond sites are completely paved (impervious). The two sites fit the Type 1 site criteria established by Kobriger et al. (1981) which include urban, elevated bridge deck, and 100 percent paved lanes characteristic of urban freeway systems. At both sites, the highway stormwater runoff is channelled into grass drainage ditches. Other pavement characteristics that can affect the quality of highway stormwater runoff are pavement age and the type of pavement surface. Both sites have bridge deck pavement surfaces of concrete. The rest of the highway contributing surfaces are made of asphalt. The impact of wear and tear from pavement deterioration can have an influence on runoff quality. The C N Railway overpass in Richmond was opened in 1989 while the Brunette River bridge deck in Burnaby was opened in 1964. If the impact from pavement wear and tear on runoff quality could be isolated from other factors, highway stormwater runoff pollution from pavement 107 wear and tear is probably greater at the Brunette River bridge deck in Burnaby since it is 25 years older than the Richmond site and has double the traffic volume. Drainage Area and Highway Design Characteristics. Defining the drainage area under study is probably one of the most important site selection criteria. According to Aye (1979) and Gupta et al. (1981a), a typical highway drainage area will range from 610 to 915 m (2,000 to 3,000 ft) in length with an approximate area of 3.2 to 4.0 ha (or 8 to 10 acres). Since large highway drainage areas may be difficult to isolate because of stormwater runoff from surrounding lands, smaller sites are more suitable for detailed investigative analysis. The drainage areas of the two sites for this research project are well defined with minimal runoff contribution from the surrounding non-highway lands. Highway design characteristics can be important in choosing an appropriate site. According to Gupta et al. (1981a), some of these are vertical alignment of the freeway (elevation above ground level), type of drainage design (curb versus non-curb, gutter or flush shoulder design), type of highway section (straight or intersection) and presence or absence of median barriers. Both highway sites selected for this research project have median barriers and curbs and are straight and elevated above ground level. This makes equipment installation and sample collection easier (Onwumere and Russell, 1995). A more detailed analysis of the curbs and barriers is provided in a later section. Proximity to a Receiving Water Body. Ideally, stormwater runoff sites should be located near a receiving water body (Gupta et al., 1981a) i f detailed evaluations of the impacts of highway runoff pollutants on water quality are to be conducted. The monitoring site should be located such that contaminants from other non-monitored discharges are not introduced upstream of the highway discharge point. But this is not always possible. The Trans-Canada site in Burnaby is 108 located near the Brunette River which eventually drains into the Fraser River at New Westminster, as indicated in Figure 3-1. In this case, two receiving water bodies are affected with much more of an impact on the Brunette River than on the much larger Fraser River. Since the impact of highway contaminants on the rivers could not be easily assessed because of other larger discharges from street and agricultural runoff, daphnia bioassays were used as an indicator of the potential impacts of the highway runoff on aquatic organisms. The Richmond site is not located near a river, but the stormwater runoff eventually ends up at the Fraser River, as shown in Figure 3-1. Stormwater runoff from the Richmond site was also subjected to daphnia bioassay. Highway Maintenance Practices. Highway maintenance practices such as grass mowing, herbicide/fertilizer use, litter pickup, de-icing operations and irrigation practices are important. The two sites selected for this research project are both affected by the maintenance practices/guidelines provided by the B.C. Ministry of Transportation and Highways. Highway maintenance at the two sites is contracted out to the same maintenance company. Both sites are classified as Class 1 highway (ADT of over 10,000 vehicles per day) by the B.C. Ministry of Transportation and Highways (MoTH). This means that the two sites are subject to the same maintenance standards as well as the same frequency of maintenance. For example, the pavement surface is to be cleaned or swept every 120 days, fallen rockside debris or spilled materials of 1,000 cc on travelled lanes are to be removed within one hour of detection, and grass mowing is to be done when grass height is at 30 cm or over on all Class 1 highways. In the G.V.R.D., the frequency of litter collection on highways with over 50,000 vehicles per day is once every 7 days and once every 14 days for Highway 1 and other multi-lane urban freeways. Winter abrasive and de-icing agents applied to maintain safe driving conditions can 109 also contribute to highway stormwater runoff pollution, as described in section 2.2 - Sources of Highway Contaminants. Logistical Considerations. These include miscellaneous items such as accessibility and vandal-proofing, power availability and potential construction activity. According to Gupta et al. (1981a), monitoring sites should be easily accessible and safe for equipment installation, servicing and monitoring, as well as for data and sample collection. Both the Burnaby and the Richmond sites are easily accessible and vandalism proof. At each site, weir boxes were buried in about 30.5 cm (or 1 ft) of ground cover with the Starflow measuring devices securely mounted inside them. Power to the Starflow was provided by an external 12 volt DC source (12V battery) contained in a weatherproof enclosure with half of it buried in the ground. Finally, any construction activity during the sample collection period is likely to significantly affect the type and level of pollutants discharged from the freeway. Therefore, it is highly desirable that no major construction activity take place during the study period within the vicinity of or on the site drainage area. There was no construction activity along the New Westminster (Highway 91) Highway site in Richmond. The Trans-Canada (Highway 1) site in Burnaby, on the other hand, was slated for possible expansion and this actually started January 1996. This expansion of the freeway did not affect this research project since sample collections were completed before the expansion activity began. No single site can meet all of the above-mentioned criteria required for site selection. However, the sites chosen met most of the requirements for suitable site selection as summarized in Table 3-1. Detailed description of each site is provided in the next section. 3.3 Site Description Burnaby Site. The site selected for this research project was the eastbound Brunette River bridge 110 deck section of the Trans-Canada Highway #1 near the Cariboo exit in Burnaby, British Columbia, Canada, as shown in Figure 3-1. It is a two-lane elevated section with highway stormwater runoff draining via a grass drainage ditch into the Brunette River, which eventually drains into the Fraser River in New Westminster. The highway surface is sloped slightly southward which enables stormwater runoff to drain along the south highway curbs. The highway section contributing runoff to the weir has dimensions of 163 m (or 535 ft) in length, 8.7 m (or 29 ft) in width, a total area of 1,420 m 2 (or 15,300 ft2), as indicated in Figure 3-5, and a watershed drainage area of 5.8 ha according to Lawson et al. (1985). Between 1987 and 1991, the mean summer average daily traffic (SADT) at this location, according to B.C. Ministry of Transportation and Highways (1992/93), was approximately 89,200 vehicles with a yearly increase in SADT, as shown in Figure 3-2. The total annual precipitation range for this area was estimated to be between 1,000 to 2,000 mm (Bertrand et al., 1985). More specifically, a mean annual total precipitation range of 1,600 mm to 1,700 mm was reported for this site by Hay and Oke(1976). The entire stormwater runoff contributing area is 100 percent paved. This site was considered to be ideal because it is impossible for drainage discharges from non-highway areas to affect the runoff from highway paved surfaces due to the elevated bridge deck and the location of the sample/monitoring site. The two lane highway is bounded by a median barrier which is about 0.76 m (or 2.5 ft) high and a curb on the south side which is about 0.52 m (or 1.7 ft) high, as indicated in Figure 3-3. As a result, the runoff from the. 1,420 m 2 (or 15,300 ft2) completely paved highway surface, minus evaporation and splash-off from vehicles, passes through the sampling location at the southwest end of the Brunette River bridge deck into a grassed drainage ditch. I l l Table 3-1. Burnaby and Richmond Sites Selection Criteria Location Criteria Burnaby Richmond Type Urban Urban SADT 89,200 Vehicles 42,000 Vehicles Total Precipitation per Year 1,600-1,700 mm 1,100-1,200 mm Drainage Area % Paved 100% 100% Surface Pavement/Deck Type Concrete/Asphalt Concrete/Asphalt Contributing Highway Surface Area 1349-1,491 m 2 866-958 m 2 Number of Lanes 2 2 Curb/Barrier in Place Yes Yes Type of Section Elevated Elevated Surrounding Land Use Residential/Commercial/Industrial Agricultural/Industrial Regular Highway Maintenance Yes Yes Ditch Drainage Surface Type Grass Grass Date Highway Opened 1964 1989 Source: Onwumere and Russell (1995). Rainfall Data from Hay and Oke (1976). 112 Richmond Site. The site selected in Richmond was the westbound Canadian National Railway overpass section of the New Westminster Highway #91 near No. 9 Road, as can be see in Figure 3-1. This is a two-lane elevated highway section sloping towards the northwest. The highway slope enables the stormwater runoff to drain to a monitoring/sample collection point at the northwest corner of the railway overpass. The two-lane highway is bounded by both a median barrier which is about 0.82 m (or 2.7 ft) high and a curb on the north side, along which the stormwater runoff drains, which is about 0.72 m (or 2.3 ft) high. The dimensions of the highway section contributing stormwater runoff to the weir box located on the northwest corner are 91.2 m (or 299 ft) long and 10 m or (32.8 ft) wide with a corresponding area of approximately 912 m 2 (or 9,820 ft2), as indicated in Figure 3-4. The watershed drainage area is 0.5 ha according to the City of Richmond Sewer Engineering Division. According to B.C. Ministry of Transportation and Highways (1992 and 1993), the mean summer average daily traffic (SADT) at this location is about 42,000 vehicles between 1987 and 1992 with the summer A D T count increasing yearly except for 1992, as can be seen in Figure 3-2. Hay and Oke (1976) also reported a mean annual total precipitation range of 1,100 mm to 1,200 mm for this site. This is about 30 percent less precipitation than at the Burnaby site. The entire highway stormwater contributing area is 100 percent paved. Like the Burnaby site, there is no possibility of stormwater runoff from non-highway areas contributing stormwater discharge to the elevated C N Railway overpass deck. As a result, all the highway stormwater runoff from the contributing area under study passes through a weir box at the northwest corner of the overpass. Runoff from the Richmond site leaves the weir box through a short grass pathway to a non-paved gutter that also drains the adjoining agricultural lands. As a result, it is not possible to evaluate the effectiveness of a grassed drainage ditch in pollution reduction at this site. However, the study was able to evaluate the quantity and quality of highway stormwater 113 runoff at this site. 3.4 Experimental Equipment and Procedures Each site was instrumented for the monitoring of precipitation, runoff rates and manual collection of stormwater runoff for quality analysis. In the following subsections, details of the monitoring instrumentation, equipment installation, operation and maintenance are provided. Precipitation Measurement. Due to the limited capital available for equipment procurement, standard rain gauges were obtained and installed beside the weirs at each of the sites, as shown in Figures 3-3 and 3-4. After rainstorms the gauges were read and, from the readings, runoff coefficients/volume were estimated. This on-site precipitation data, together with the "official" climatological data from Environment Canada stations (Burnaby Mountain and Richmond . Nature Park), which recorded rainfall at hourly intervals, was compared in later tables. The standard simple gauge was developed by the Air Resources Branch, Assessment and Planning Division of B.C. Ministry of Environment, Lands and Park in Victoria, British Columbia. The rain gauge is made of two components, namely a galvanized tin funnel and a plastic storage tube that is made from P.V.C. class 160 material. The funnel channels rain water into the P.V.C. storage tube. The funnel also has a slight notch which aids in pouring off the contents of the rain gauge. ' As suggested by Gupta et al. (1981a), the precipitation measuring equipment was located in the immediate vicinity of the site. Along the Trans-Canada and New Westminster monitoring sites, the P.V.C. rain measuring devices were installed close to the weirs. The rain gauge was only about 6.1 m (or 20 ft.) away from the slow lane and 0.5 m (or 1.6 ft) above the ground along the Trans-Canada Highway site, as shown in Figure 3-3. As at the Trans-Canada Highway site, the rain gauge was located approximately 6.5 m (or 21.3 ft.) away from the slow lane and 0.53 m 114 (or 1.7 ft.) above the ground along the New Westminster Highway site in Richmond, as indicated in Figure 3-4. The Environment Canada weather stations nearest to the Burnaby and Richmond sites are located at Burnaby Mountain and Richmond Nature Park respectively. Unfortunately, it was not possible to get hourly rainfall data from the Environment Canada stations for the same periods, as those for which rainfall was recorded at the two sites. However, daily rainfall values were obtained from the two stations for comparison purposes. But the mean annual rainfall ranges from Hay and Oke (1976) were used in the loading calculations. Runoff Measurement. Runoff monitoring, which involved flow measurement and recording, was achieved at the two sites using a v-notch weir in a "weir box" and a water level recorder. A right-angle v-notch weir, constructed from aluminum, was used at the two sites. The dimensions of the weir box were 1.22 m (4 ft.) in length, 0.61 m (2 ft.) in width, and 0.31 m-(l ft.) deep. The notch was 0.091 m (0.3 ft.) from the bottom of the weir box and the head "h" was measured from the bottom of the notch to the water surface elevation, as can be seen in Figures 3-3 and 3-4. The flow rate or discharge in the channel was determined using the following equation (with SI Units) for a v-notch weir: Q = 1.38hn where Q = discharge in m3/s, h = head above the weir crest in m, and n = the exponent in the weir equation is 2.5 for v-notch weirs (Granet, 1989). See Appendix C for the flow sample calculation. 115 The second component required for measuring flow, in conjunction with the hydraulic structure described above, was the water level recording instrumentation. In this case, a Unidata Starflow Ultrasonic Doppler Instrument with Micrologger was used. The Starflow Micrologger measures both water depth and temperature. The Starflow Micrologger uses a solid state pressure sensor mounted underneath the Starflow and vented to atmospheric pressure through a vent tube inside the signal cable to measure water depth. The vent tube is filled with a desiccant to prevent moisture from entering (Unidata Australia, 1994). The Starflow was mounted on a 35 mm thick piece of wood before being securely mounted at the bottom of the weir box. The Starflow mount on top of the wood was to avoid the sensor being covered by sediment build-up even though the weir box was cleaned regularly as part of the maintenance operations. The vent tube desiccant canister and an external 12 volt DC sealed lead acid battery that powered the Starflow were located in a weatherproof enclosure. The weatherproof enclosure was mounted on two iron bars that were buried in the ground to avoid vandalism. Using an adaptor cable provided by the Unidata for connecting the Starflow to a computer and battery, the Starflow was programmed with the Version 3 support software. The Starflow Support Software complements Starflow instrument operations by providing an easy-to-use data collection, analysis and management system (Unidata Australia, 1994). The Starflow was programmed to collect water depth data. The scan rate, which measures the frequency of reading transducer signals, was set at 2 minutes. Similarly, the log interval, which measures the period between logging data, was set at 10 minutes. The Starflow was programmed to start logging when the water depth was greater than 45 mm, which was the height slightly above the notch. The logged values were stored in the Starflow's Micrologger and down-loaded periodically. The Starflow Ultrasonic Doppler sensing instrumentation was 118 installed in accordance with the manufacturer's instructions to aid in the water quality sampling analysis. The only drawback encountered with the Starflow was that the sensors were easily covered with sediment/debris material in the weir box. This could affect data accuracy, so the sensors were cleaned off on every visit to the sites as well as before every sample collection. Water Quality Sampling. In this research project, sample collection was done manually which was inconvenient at times and labour intensive. But some of the rewards associated with manual sampling were: potentially higher quality of sampling; ability to adapt the sample collection methodology to changing conditions; and lack of vulnerability to equipment failure. The following pre-storm operations were performed prior to any rainfall event: • Sample containers were pre-labelled. • Cooler was packed with ice for sample preservation. • Weirs were cleaned of any sediment/debris. • Regular contact was made with the local weather forecast office. • Two field operators were put on high alert when probability of precipitation was During the rainfall events, disposable gloves were worn to collect grab samples, employing the following procedures: 80 percent or higher. The laboratory was notified in advance of sample arrival. Sample was collected where turbulence was at a maximum within the channel. Sample was collected with the container facing upstream. Sample was collected away from floating debris. Sample collection was done every 10 minutes in a 250 mL plastic bottle before the runoff enters the weir and along the grass drainage ditch 30 m away from weir. 119 • Sample collection was performed for 3 hours. When the rainfall stopped before the 3 hour period had lapsed and 6 samples had been collected (1 h), the collected samples were composited in proportion to the flow. • Samples were placed in a cooler packed with ice and transported to the laboratory. • The oil and grease sample was collected in a 1 litre amber glass bottle before the runoff entered the weir. • Field blanks were collected and analyzed. Distilled water was poured into sampling bottles in the field, transported to the lab with other the samples and analyzed. • A total of 22 and 16 runoff events were collected and analyzed at the Burnaby and Richmond sites respectively. Once in the laboratory, the discrete samples collected were composited to obtain a 1 litre volume of liquid required for laboratory analysis. Discrete sampling during a runoff event provides information on the time distribution of pollutant loadings, while samples composited using discrete samples provides data on total pollutant load as well as on the average concentration of the pollutants. Even though Gupta et al. (1981a) and Environment Canada (1993) highly recommended compositing of highway stormwater runoff, there are disadvantages to this approach. Composite sampling does not provide any time-varying pollutant characteristics such as the minimum or maximum parameter concentration during a runoff event, and some of the sample parameters may be undetected due to dilution effects. But there are also several advantages to using composite sampling to estimate pollutant loadings: Composite sampling provides a pragmatic summary of variable stormwater runoff characteristics; Reduction in laboratory analysis costs due to lower number of samples; 120 More events can be analyzed due to fewer number of samples (Environment Canada, 1993). Similarly, there are advantages to using a discrete or grab sampling technique in highway stormwater runoff analysis: • Changes in concentrations during storm events can be better analyzed. • Discrete sampling is better for determining changes in stormwater runoff characteristics such as bacterial levels and volatile compounds during storage. • Discrete sampling provides a better single stormwater runoff characteristic profile at minimal equipment cost/labour. • Data generated from discrete sampling can be used for pollutographs (Environment Canada, 1993). Consequently, both methods of collecting highway stormwater runoff were used in this research project. Discrete samples were collected every 10 minutes from both sites and the composite samples were made up in the laboratory. Of the four basic types of composite samples, both a constant time/constant volume technique and a constant time/volume proportional to the flow-rate method were used. The constant time/volume proportional to the flow-rate method, recommended by Gupta et al. (1981a), depends heavily on accurate volumetric flow rate measurements which were obtained using the discrete sampling data. Eight rainfall events were analyzed at the Burnaby site using flow-composite technique while at the Richmond site only six events were analyzed by the same method. The remaining events were analyzed using discrete technique. Once the compositing of the samples was completed, the samples were submitted to the laboratory for analysis along with the sediment, road dirt and grass clipping samples. 121 3.5 Road Dirt, Soil Sediment and Grass Clipping Samples According to Environment Canada (1993), there are three methods used in the collection of road dirt or street sediment: hand-sweeping, vacuuming, and flushing with water, and they recommended a combination of these methods. However, lack of power near the two sites reduced the available options to hand-sweeping. Following Wullschleger et al. (1976), the following sampling procedure was used in road dirt sample collection: • . Sampling areas were located in highway sections that were in relatively good condition. • Sample collection length was 30 metres (or 100 feet) at both the Burnaby and Richmond sites. • A curb length of 3 to 5 metres (or 10 to 15 feet) was swept with a hand bristle broom. • The broom was knocked clean and road dirt samples were collected and emptied into a zip-lock plastic bag. In this research project, great care was taken during the hand-sweeping operations to pick up as much fine particle matter as possible. Unlike the road dirt sample collection, grab surface soil sediment collection within the drainage ditches was easier. At the Trans-Canada Highway site in Burnaby, the first set of samples, labelled TCH A, were collected at approximately 0.5 m from the weir box and the second set of samples, labelled TCH B, were collected about 30 m downstream from the weir within the drainage ditch channel. Only one set of samples was collected along the New Westminster Highway in Richmond at about 0.5 m away from the weir labelled N W H . A second set of samples could not be collected due to the layout of the drainage ditch channel. The 122 collected surface soil sediment grab samples were emptied into zip-lock plastic bags. Similarly, grass clipping samples were collected at both sites at the same distances as the soil sediment samples. The grass clippings were obtained by clipping the top edge of the grasses and emptying the clipped grasses into zip-lock plastic bags. The eleven road dirt, soil sediment and grass clipping samples were transported to the laboratory where they were subjected to metal analysis for Cr, Cd, N i , Cu, Zn, Pb, Fe, Mn and Ca. This laboratory analysis, like the other analyses in the next section, was in accordance with the methodology recommended by A P H A et al. (1992). 3.6 Sample Custody and Laboratory Analysis The analyses of samples for pH, electrical conductivity, suspended solids, metals and oil and grease were performed as soon as possible. Where parameter analysis could not be performed immediately, the samples were refrigerated in a walk-in cooler (at 4°C) and analyses were done within the next few days. Where sample preservation was required for both liquid and sediment samples, the samples were preserved by the laboratory personnel according to the methods recommended by American Public Health Association (APHA), American Waterworks Association (AWWA), and Water Pollution Control Federation (WPCF) (1989). On arrival in the laboratory, oil and grease samples were acidified by adding 1:1 H C L to pH 2 and stored at 4°C. Samples for metals and suspended solids were composited into a single bottle and no acid was added to preserve the metals until after the T.S.S. sample was taken. T.S.S. was analyzed as soon as possible. Once the T.S.S. sample was taken, the metal samples were preserved by adding nitric acid and stored at 4°C. Sediment and grass samples were preserved by storing at 4°C. The laboratory analyses for all constituents except daphnia bioassay were done at the 123 Environmental Engineering Laboratory, Department of Civil Engineering at the University of British Columbia, Canada. The runoff and sediment samples were analyzed in accordance with the methods approved by the A P H A et al. (1989/95). Stormwater Runoff Metal Laboratory Analysis. 250 mL of well-mixed, acid-preserved runoff sample was transferred to a flask. 5 mL of Cone. H N 0 3 and a few boiling chips were added to the flask and brought to a slow boil. The sample was evaporated to a volume of about 20 - 25 mL before precipitation occurred. Heating was continued, as well as the adding of Cone. H N 0 3 until digestion was completed. Completion of digestion was indicated by a light-coloured, clear solution. The flask walls were washed down with water and filtered when necessary. Filtrate was transferred to a 100-mL volumetric flask with two 5 mL portions of water, cooled, diluted to the mark and mixed thoroughly. Portions of this solution were used for Cd, N i , Cu, Zn, Pb,Fe, Mn and Ca determinations using a Thermo Jarrell Ash Video 22 aa/ae spectrophotometer (APHA et al., 1995) with direct air-acetylene flame method. Cr determination was performed using the direct nitrous oxide-acetylene flame method. Oil and Grease Laboratory Analysis. The acidified sample from preservation was transferred to a separatory funnel. The sample bottle was rinsed with 30 mL of 80:20 n-hexane/methyl-t-butyl ether (extracting solvent) and added to the separatory funnel. The mixture was shaken vigorously for two minutes after which the layers were allowed to separate. The extracting solvent takes the upper layer. The solvent layer was drained through a funnel containing a filter paper and 10 g Na^C^ , both of which were previously solvent-rinsed into a clean, tared distilling flask. Where a clear solvent was not obtained due to emulsion formation, the emulsion and solvent layers were drained into a glass and subjected to 5 minutes of centrifuge at about 2,400 rpm. The draining process, through a funnel with filter paper and 10 g NajSO^ was repeated. Aqueous layers and any remaining emulsion or solids were recombined in the separatory funnel 124 and extracted twice more with 30 mL solvent each time. Centrifugation was repeated when necessary. Otherwise, the filter and Na2S04 were rinsed with 10 to 20 mL solvent and combined with the extracts in a flask before distilling in a water bath at 85 °C. The clear solvent was received in an ice-bath cooled receiver and the flask was dried on top of the cover at 85 °C for 15 minutes. Vacuum was used to draw air through the flask for 1 minute before cooling in a desiccator for at least 30 minutes. The tared distilling flask was weighed. The gain in weight of the tared distilling flask is due to oil and grease, assuming that the organic solvent is free of residue. The oil and grease in the runoff was determined using the following equation (APHA et al., 1995): mg oil and grease/L = (A-B) x 1000 mL sample where A = total gain in weight B = tared flask less calculated residue from solvent blank, and mL = millilitre of initial sample volume. T.S.S. Laboratory Analysis. A 100 to 250 mL sample was filtered through a filter with a filtering apparatus under suction. The filter was wet with a small volume of reagent-grade water to seat it properly. The well mixed sample was transferred onto the seated glass-fibre filter, washed with three successive 10 mL volumes of reagent-grade water, drained completely between washes and subjected to suction for 3 minutes after complete filtration. According to A P H A et al. (1995), samples with high dissolved solids may need more washes. The filter was removed from the filtration apparatus, transferred to an aluminum weighing dish, dried for at least one hour at 103 to 105° C in an oven. The aluminum weighing dish was cooled in a desiccator to balance temperature and then weighed. The cycle of filtering, drying, cooling, desiccating and weighing was repeated until a constant weight was obtained or the weight change 125 was less than 4% of the previous weight or 0.5 mg, whichever was less. According to A P H A et al. (1995), duplicate determinations should be within 5 % of their average. The mg/L of T.S.S. was calculated using the following equation (APHA et al., 1995): mgT.S.S./L = (A-B)x 1000 sample volume (mL) where A = weight of filter + dried residue (mg), and B = weight of filter (mg). Sediment, Road Dirt and Grass Sample Analysis. Sediment, road dirt and grass samples were transferred to an evaporating dish and the dry-weight determined by drying overnight at 103 °C. On average, about 10 g, 5 g and 2 g of sediment, road dirt and grass samples respectively in a dish were placed in a muffle furnace and heated to 450°C for one hour. The sample was cooled to room temperature before adding 5 mL of cone, nitric acid. The dish sides were washed down with deionized-distilled water and the sample was gently heated on a hot plate to just below boiling for one half hour. Deionized-distilled water was added periodically to keep the volume of liquid at approximately 20 mL. The sample was cooled and filtered through Whatman No. 541 filter paper and the volume made up to 50 mL with deionized-distilled water. According to A P H A et al. (1989), the adjusted known volume should have a final H N 0 3 concentration of about 1%. Portions of this solution were taken for metal determination using a Thermo Jarrell Ash Video 22 aa/ae spectrophotometer. 48-h LC50 Bioassayfor Daphnia Magna. This 48-hour static test with no replacement of solutions was conducted in glass beakers by B.C. Research Corporation. 200 mL of well mixed runoff samples with varying test concentrations (ranging from 10 to 100 (% V/V)) were introduced into beakers. Controls were also set up with 0 (% V/V) test concentration. Each beaker contained 10 neonates less than 24 hours old which were obtained from 2-4 week old 126 females. This ensured uniform sensitive organism for the acute toxicity tests according to Hall and Anderson (1988). The tests were conducted at a temperature of 20 ± 1 °C and under the same lighting conditions as during culturing. Both the pH and dissolved oxygen concentration were measured in each test concentration including the controls at the beginning of the test, and at the end when biological observations were completed. Similarly, the percentage of the survived organism was enumerated at 24 and 48 h (Environment Canada, 1990). Also recorded were the mortalities of daphids at 48 h in each test solution which was used to calculate the 48-h LC50 and its confidence limits. B.C. Research Corporation used their own modified computer program to calculate the 48-h L C 50 based on earlier program developed by Stephan (1977). The program estimates L C 50s using probit, moving averages and binomial methods. 3.7 Quality Control/Quality Assurance The integrity of the sample collection was assured by the collection of representative samples by using as much as possible the methods outlined by Gupta et al. (1981a) as well as Environment Canada (1993). The integrity of the samples was assured by placing the stormwater runoff and sediment samples in specially washed plastic and glass bottles/containers as well as the use of bottle and field blanks. The plastic bottles used for suspended solids and heavy metal contaminants sampling were washed in accordance with the methods recommended by Marsalek and Greek (1984) which included: Washing with detergent and tap water; Rinsing the bottles 2 to 3 times with tap water; Rinsing the bottles with 10 percent nitric acid; Rinsing the bottles 2 to 3 times with distilled/deionized water; 127 Air-drying the containers and capping them; Minimum volume of each rinse was 2 to 3 percent of the container volume. The oil and grease glass sampling bottles were supplied by the laboratory and were washed in accordance with the method outlined by A P H A et al. (1992) to ensure their sampling integrity. Bottle blanks, containing distilled water, were submitted periodically to determine whether sample bottles were sources of pollutant contamination. The sampling bottles used in this research project were recycled periodically, therefore they were cleaned according to the methods outlined above. One bottle blank was submitted for each batch of sample bottles used for each site. Field blanks were also submitted to determine whether sample contamination occurred during field sample collection, handling, transportation to the laboratory, storage and sample analysis. According to Environment Canada (1993), field blanks should be filled with deionized-distilled water in the field and the handling procedures should be the same as with the stormwater runoff handling procedures. Due to budget constraints, other quality assurance/quality control methodologies such as field replication, method blanks and duplicate sampling could only be done occasionally. However, all necessary precautions were taken by field personnel to ensure the following of methods outlined by Gupta et al. (1981a) and Environment Canada (1993) for field sample collection, sample handling, transportation to the laboratory, compositing, storage and laboratory analysis. Hopefully, the quality assurance/quality control procedures used in this research project, as well as the precautions taken by the field and laboratory personnel, reduced any sample collection/analytical anomalies and errors to the minimum practical. Some errors in field and laboratory work are inevitable, but they can be reduced through proper sample collection and handling, suitable transportation and storage and, above all, through the diligent efforts of 128 experienced laboratory analysts. Where anomalies and errors were found despite the above measures, explanations were developed and are presented in the next chapter to accommodate the problem(s). Also, recommendations are made in the last chapter for future research work. The appropriate metal detection levels are provided in Appendix A as well as the summary of the QA/QC program determined using blanks, replicates and spike recovery. Also, precision, which is a measure of the variability of the individual sample measurements, was tabulated in Table A - l . Similarly, Table A-2 in Appendix A provides the appropriate metal detection levels for the sediment, road dirt and grass samples. 3.8 Statistical Analysis Correlation of the trace metal concentration results with the selected influential environmental variables was performed using regression analysis to obtain "best fit" linear correlations. The goodness of fit of the data to the regression line was given by r2 with a perfect fit resulting in r^=l. Less than perfect fit has r2 < ±1 indicating negative and positive correlations (Mansfield, 1986). Multiple regression analysis was also performed to determine relationships between the dependent variables and more than one independent variable with an r2 value to measure goodness of fit (Mansfield, 1986). Similarly, correlation matrix analysis was performed between environmental variables to determine the variables that were dominant in influencing highway runoff pollution. Annual average loading estimates were calculated using log normal transformed data. According to Macdonald et al. (1996), when estimating means and confidence limits, log transformed statistics should be used to obtain the least biased estimates of the untransformed statistics, especially for limited data set. This supported the previous studies by U.S. EPA 129 (1983), Di Torro (1984), Harremoes (1988) and Van Buren et al. (1995), which stated that many pollutants commonly detected in urban runoff have event mean concentrations that are not normally distributed. Most times, certain constituents have values that are below the detection limit, therefore it may be appropriate to use a log normal distribution to make a statistical inference about samples. Similarly, according to Sherwani and Moreau (1975), normal, log normal, and gamma distributions are useful in modeling water quality random variables. Loftis et al. (1983), E l -Shaarawi (1989), and U.S. EPA (1983) also further suggested that normal and log normal distributions are the most widely applicable to water quality and many environmental applications. According to Loftis et al. (1983), normal and log normal distributions are of most interest to regulatory agencies because of their acceptability in modeling water quality parameters and they are fairly easy to use in modeling means and variances. But for log normal distribution, the geometric mean is preferred as a measure of central location than the arithmetic mean (Little et al., 1983). However, in this study, the differences between the geometric and arithmetic means are very small, especially for the metal pollutants. A l l the statistical analyses were performed by computer using the Lotus 1-2-3 and Systat statistical packages. The same computer statistical packages were used for the multiple-regression forecasting model. Loading Calculations Seasonal/annual pollutant loadings were computed using seasonal/annual volumes of runoff and mean constituent concentrations estimated from field sampling. According to Marsalek and Ng (1989), the annual runoff volume can be adequately estimated from the runoff coefficient by approximation method. However, the main difficulty is in estimating seasonal/annual mean concentration due to variability during individual runoff events, between 130 events, and among sites. According to Marsalek and Ng (1989), these variabilities are corrected by using flow-proportional composite samples, sampling a number of events at each site, and sampling in areas with various land use/rates of export of pollutants. They suggested that this methodology yields results consistent with those from general load functions, hence the reason for its application to this research as shown in appendix C. Uncertainty Analysis According to previous research studies by Gupta et al. (1981a) and Kobriger et al. (1981a), a measurement error in the range of 1% to 20% is common with highway stormwater runoff studies. In this research study, a measurement error of 5% was presumed with the contributing highway surface area at both sites. The error was attributed to taking only one width measurement rather than taking several highway width measurements every one metre length or so. The errors associated with the estimated total runoff, using the 5% error in contributing surface area, were calculated and shown in the sample calculation subsection in Appendix C. The errors were estimated by the square root of the sum of the squares of the coefficient of variation (CV) of the variables in the function, where the coefficient of variation is the standard deviation divided by the mean, as shown in Appendix C. 131 Chapter 4 RESULTS A N D DISCUSSION This chapter presents the results from the research study, compares them with the findings from other similar studies, discusses the results and attempts to explain the differences. Since many parameters were investigated, it was considered better to provide both the results and discussion for each parameter, one at a time, rather than presenting all the results first and then discussing all of them in a separate chapter. 4.1 Rainfall-Runoff Relationships Tables 4-1 and 4-2 show a complete listing of the hydrological data for all monitored storm events at both the Burnaby and Richmond sites for which runoff quality data were collected and analyzed. The data analyzed are only for selected rainfall events at both sites. The total daily rainfall amounts at nearby Environment Canada weather stations during the storm event days are shown in Tables 4-1 and 4-2 for comparison. As explained in Chapter 3 section 3.4, it was not possible to obtain hourly rainfall data at these stations for the same storm periods. From the analysis of the data presented in Tables 4-1 and 4-2, it was found that the highest amount of rainfall fell during the winter months (November - March) at both sites using the sampled rainfall data. The average winter rainfall (for all sampled events) at the Burnaby site was found to be 5.89 mm compared to 4.20 mm at the Richmond site. There is about 29% less rainfall at the Richmond site than the Burnaby site during the winter sample collection period. The average winter rainfall intensity at the Burnaby site was 2.49 mm/h, with a range from 1.33 to 4.17 mm/h compared to 2.01 mm/h at the Richmond site, although the highest rainfall amount and intensity of 12.6 mm and 4.20 mm/h respectively were obtained at the Richmond site, with a 132 Table 4-1. Burnaby Hydrological Data for Sampled Events Date Antecedent # of Dry Days Site Rainfall (mm) Duration (min.) Average Intensity (mm/h) Daily Rainfall (mm) 11/04/95 8 10.0 180 3.33 9.60 01/06/96 0 12.5 180 4.17 32.0 02/07/96 1 5.10 100 3.05 13.0 02/17/96 0 3.40 90 2.27 11.5 02/19/96 1 1.80 60 1.80 7.25 03/08/96 1 4.00 180 1.33 9.25 03/09/96 0.5 4.40 180 1.47 8.90 03/31/96 0 N /A 60 N /A 5.00 04/01/96 0 7.00 180 2.33 21.0 04/05/96 3 3.40 180 1.13 22.0 04/09/96 4 0.70 60 0.70 12.0 04/11/96 1 0.20 60 0.20 3.00 04/22/96 2 1.20 100 0.72 33.0 04/23/96 0 2.50 100 1.50 21.0 04/25/96 1 1.20 90 0.80 22.0 05/07/96 0 1.80 90 1.20 5.00 05/17/96 2 0.80 60 0.80 12.0 05/18/96 0 5.70 90 3.80 16.0 06/10/96 8 2.80 120 1.40 N /A 07/02/96 7 0.85 60 0.85 3.00 07/17/96 14 1.75 90 1.17 8.00 08/02/96 13 7.80 100 4.67 10.0 Notes: 1) N / A - Not available. 2) Dates with bold indicate flow-composite sampling days. 3) Daily Rainfall Estimate - Total Daily Rainfall near the Burnaby site from the Environment Canada Burnaby Mountain weather station (hourly data not available, only morning and afternoon rainfall totals are available). 133 Table 4-2. Richmond Hydrological Data for Sampled Events Date Antecedent # of Dry Days Site Rainfall (mm) Duration (min.) Average Intensity (mm/h) Daily Rainfall (mm) 11/04/95 8 N / A N / A N / A 4.00 01/06/96 0 7.20 180 2.40 20.6 02/07/96 1 12.6 180 4.20 5.40 02/17/96 0 0.80 90 0.53 4.20 02/19/96 1 N / A 60 N / A 3.80 03/08/96 1 3.40 180 1.13 6.40 03/09/96 0.5 5.40 180 1.80 9.80 03/31/96 0 N / A N / A N / A 3.00 04/01/96 0 6.30 120 3.15 16.4 04/05/96 3 1.00 120 0.50 16.4 04/09/96 4 N / A N / A N / A 7.00 04/11/96 1 N / A N / A N / A 3.80 04/22/96 2 2.50 90 1.67 32.6 04/23/96 0 1.20 90 0.80 21.0 04/25/96 1 0.70 80 0.53 9.60 05/07/96 0 N / A N / A N / A 4.20 05/17/96 2 1.10 100 0.66 11.4 05/18/96 0 1.20 100 0.72 8.40 06/10/96 8 N / A N / A N / A 2.40 07/02/96 7 N / A N / A N / A 2.60 07/17/96 14 0.35 60 .35 2.60 08/02/96 13 N / A 60 N / A 21.4 Notes: 1) N / A - Not Available. 2) Dates with bold indicate flow-composite sampling days. 3) Daily Rainfall Estimate -Total Daily Rainfall near the Richmond site from the Environment Canada Richmond Nature Park weather station (hourly data not available, only morning and afternoon rainfall totals are available). 134 range from 0.53 to 4.20 mm/h, as can be seen from Table 4-2. Unlike the winter rainfall pattern, the highest non-winter (April - October) rainfall amount of 7.8 mm was obtained at the Burnaby site as well as the highest rainfall intensity of 4.67 mm/h for the same time period, as can be seen from Tables 4-1 and 4-2. But the average non-winter rainfall amount followed the same trend as the winter, with the Burnaby site recording 2.69 mm compared to 1.79 mm obtained at the Pvichmond site. Again, the Richmond site has about 33% less precipitation than the Burnaby site. Similarly, the average non-winter rainfall intensity at the Burnaby site was higher at 1.52 mm/h than 1.05 mm/h observed at the Richmond site. The rainfall amount and the associated runoff volume for each storm were used to calculate the range and average volumetric runoff coefficients tabulated in Tables 4-3 and 4-4. The runoff coefficients relate the total volume of runoff from the paved highway drainage area to the amount of rainfall causing the runoff. The volumetric runoff coefficients were calculated using the following formula: Runoff Coefficient (RC) = Total Runoff (m3) Total Rainfall (m) x Catchment Area (m2) The Burnaby mean runoff coefficients for individual storms ranged from 0.51 to 1.18 and the Richmond ones from 0.26 to 0.86. See sample calculation in Appendix C. Of course, it is impossible to have a runoff coefficient value that is greater than 1.0 unless there is an external flow source that is contributing runoff to the defined monitoring drainage area. There are many possible sources of error, such as in the estimated drainage area, but in this case, the main sources of error are probably from inaccuracies in rainfall and flow measuring devices. Nevertheless, all the runoff coefficients were used in the data analysis. At the Burnaby and Richmond sites, the mean winter runoff coefficients were found to be 0.82 and 0.65, as can be seen in Tables 4-3 and 4-4 respectively. This means that, on average, close to 74% of the rain 135 Table 4-3. Burnaby Runoff Coefficients for Sampled Storm Events Date Site Rainfall (mm) Duration (min.) Total Runoff Volume (m3) Average Flow (xlO4 m3/s) Runoff Coefficient (calculated) 0.028947 10.0 180 9.34 8.65 0.72-0.80 (0.76) 01/06/96 12.5 180 11.5 10.6 0.71-0.78 (0.75) 02/07/96 5.10 100 4.37 7.28 0.66-0.73 (0.70) 02/17/96 3.40 90 2.34 4.33 0.53-0.59 (0.56) 02/19/96 1.80 60 1.92 5.32 0.82-0.91 (0.87) 03/08/96 4.00 180 5.84 5.41 1.13-1.24(1.18) 03/09/96 4.40 180 5.03 4.66 0.88-0.97 (0.93) 03/31/96 N / A N / A N / A N / A N / A 04/01/96 7.00 180 6.54 6.06 0.72-0.82 (0.76) 04/05/96 3.40 180 3.94 3.68 0.89-0.99 (0.94) 04/09/96 0.70 60 0.55 1.53 0.61-0.67 (0.64) 04/11/96 0.20 60 0.14 0.40 0.54-0.60 (0.57) 04/22/96 1.20 100 0.97 1.61 0.62-0.69 (0.66) 04/23/96 2.50 100 1.60 2.67 0.49-0.55 (0.52) 04/25/96 1.20 90 0.88 1.63 0.57-0.62 (0.60) 05/07/96 1.80 90 1.34 2.49 0.57-0.63 (0.660) 05/17/96 0.80 60 0.50 1.38 0.48-0.53 (0.51) 05/18/96 5.70 90 3.79 7.01 0.51-0.57(0.54) 06/10/96 " ' 2.80 120 2.07 2.87 0.57-0.63 (0.60) 07/02/96 0.85 60 N / A N / A N / A 07/17/96 1.75 90 1.34 2.49 0.59-0.65 (0.62) 08/02/96 7.80 100 6.75 11.3 0.67-0.74 (0.70) Notes: 1) Mean Winter Runoff Coefficient = 0.82 2) Mean Non-Winter Runoff Coefficient = 0.63 3) Overall Runoff Coefficient = 0.70 4) N / A ~ Not available, and numbers in bracket are mean values. 136 Table 4-4. Richmond Runoff Coefficients for Samples Storm Events Date Site Rainfall (mm) Duration (min.) Total Runoff Volume (m3) Average Flow (xlO4 m3/s) Runoff Coefficient (calculated) 11/04/95 N / A N / A N / A N / A N / A 01/06/96 7.20 180 2.96 2.74 0.40-0.44 (0.42) 02/07/96 12.6 N / A N / A N / A N / A 02/17/96 0.80 90 0.64 1.18 0.78-0.86 (0.82) 02/19/96 N / A 60 N / A N / A N / A 03/08/96 3.40 180 2.84 2.63 0.81-0.90 (0.86) 03/09/96 5.40 180 2.61 2.42 0.47-0.52 (0.50) 03/31/96 N / A N / A N / A N / A N / A 04/01/96 6.30 120 3.38 4.69 0.52-0.58 (0.55) 04/05/96 1.00 120 0.53 0.73 0.52-0.57 (0.54) 04/09/96 N / A N / A N / A N / A N / A 04/11/96 N / A N / A N / A . N / A N / A 04/22/96 2.50 90 1.44 2.67 0.56-0.62 (0.59) 04/23/96 1.20 90 0.66 1.22 0.54-0.59 (0.57) 04/25/96 0.70 80 0.48 1.01 0.67-0.74 (0.70) 05/07/96 N / A N / A N / A N / A N / A 05/17/96 1.10 100 0.28 0.47 0.25-0.27 (0.26) 05/18/96 1.20 100 0.62 1.04 0.50-0.56 (0.53) 06/10/96 N / A N / A N / A N / A N / A 07/02/96 N / A N / A N /A N / A N / A 07/17/96 0.35 60 0.10 0.29 0.28-0.31 (0.29) Notes: 1) Mean Winter Runoff Coefficient = 0.65 2) Mean Non-Winter Runoff Coefficient = 0.51 3) Overall Runoff Coefficient = 0.55 4) N / A ~ Not available, and numbers in bracket are mean values. 137 that fell at both sites was measured at the drainage area outlet as stormwater runoff during the winter period. The non-winter runoff coefficient was also higher at the Burnaby site with an average of 0.63 and a range from 0.51 to 0.94, as can be seen in Table 4-3. At the Richmond site, the average coefficient was 0.51 and a range of 0.26 to 0.70, as shown in Table 4-4. The runoff coefficients obtained at both sites were used to calculate the seasonal runoff volumes at both sites and the total annual runoff volume, as can be seen in Appendix Table A-3 and Appendix C. These total volumes were used later to calculate annual pollutant loadings at both sites. Thus, the determination of runoff coefficients played a key role in estimating the total annual pollutant loadings generated at both sites. Discussion According to Kobriger et al. (1981), the average runoff coefficients are directly related to the percentage of paved or unpaved drainage area. In this case, both sites were 100% paved, as can be seen in Table 3-1, and the overall mean runoff coefficients for both Burnaby and Richmond were found to be 0.70 and 0.55 respectively. Between the two sites, the Burnaby site generated higher runoff coefficients than the Richmond site. A categorical variable analysis of variance test (ANOVA) at 95% confidence level showed the differences in the runoff coefficients between the two sites to be statistically significant. The F-ratio was 5.38 with a P-value of 0.03 at 1 and 30 degrees of freedom indicating that site did influence the runoff coefficients. Table 4-5 shows a comparison of the measured runoff coefficients with those obtained by other researchers. The overall mean runoff coefficient of 0.70 obtained for the Burnaby site was within the runoff coefficients obtained in the Washington State research which ranged from 0.52 138 Table 4-5. Comparison of Runoff Coefficients Runoff Coefficient Range Mean Site Type Location % Paved Reference 0.51-1.18 0.70 Highway #1 Burnaby, Canada 100% This Study 0.26-0.86 0.55 Highway #91 Richmond, Canada 100% This Study N / A 0.72 1-5 Highway Seattle, U S A N / A Chui et al.(1981) N / A 0.70 SR-520 Highway Seattle, U S A N / A Chui et al.(1981) N / A 0.75 Highway Montesand, U S A N / A Chui etal.(1981) N / A 0.77 Highway Pasco, U S A N / A Chuietal.(1981) N / A 0.69 Highway Spokane, U S A N / A Chuietal.(1981) N / A 0.52 Highway Pullman, U S A N / A Chui etal.(1981) N / A 0.80 Highway (Control) Pullman, U S A N / A Chui et al.(1981) 0.40-1.11 0.92 1-794 Highway Milwaukee, U S A 100% Gupta et al.(1981) 0.10-1.42 0.30 Highway 45 Milwaukee, U S A 31% Gupta et al.(1981) 0.04-0.95 0.43 Highway Harrisburg, U S A 27% Gupta et al.(1981) 0.10-1.00 0.43 Highway Nashville, U S A 37% Gupta et al.(1981) 0.12-0.66 0.41 Highway Denver, U S A 37% Gupta et al.(1981) 0.41-0.91 N / A Highway Orange County, U S A 100% Yousef et al. (1985) 0.03-0.94 0.40 Residential Vancouver, Canada N / A Swain. (1983) 0.03-0.80 0.31 Industrial Burnaby, Canada N / A Lawson et al. (1985) Note: 1) N / A - Not available. 139 in Pullman to 0.72 for 1-5 in Seattle. But the Washington State research did not specify the percentage of the highway surface that was paved. 1-794 in Milwaukee, with a 100% paved drainage area, has a mean runoff coefficient of 0.92 which is higher than the 0.70 obtained at the Burnaby site with similar percentage of paved area. Other sites, with a range of 27 to 37% paved surface, have runoff coefficients considerably lower, as can be seen in Table 4-5. On the other hand, the Richmond site with a 100% paved surface has a runoff coefficient of 0.55 which is higher than that at Pullman, Washington but was within the runoff coefficient range of 0.41 to 0.91 reported in Orange County, Florida by Yousef et al. (1985). The report on the research in Florida did not specify the percent of the highway surface that was paved. Both Burnaby and Richmond site runoff coefficients were higher than the residential runoff coefficient obtained in Vancouver by Swain (1983) and the Burnaby industrial runoff coefficient obtained by Lawson et al. (1985), as indicated in Table 4-5. This is expected, as the road research sites are 100% paved. Runoff coefficients obviously vary from site to site and depend on differences in prestorm history, rainfall amounts and durations, monitoring instrumentation and differences in vehicular splash-off. Also, while evaporation from pavement might possibly contribute to runoff coefficient variation, in the temperate regions, the effect is likely to be minimal, especially during the winter period. These factors are discussed below: Prestorm History. Prestorm history affects the state of the drainage area at the beginning of the storm and can be very important, particularly i f some of the drainage area is unpaved. Even when the basin is 100% paved, as was the case in this study, the amount of runoff is affected by whether the surface is wet or dry. This is relevant when the rain begins, since the surface has to be wet before there can be any runoff. In this study, the highway surfaces were already wet before sample collection was initiated. The researchers waited until rainfall began before going out to the two sites. Hence, it is difficult to account as to what percentage of the rainfall went to depression 140 storages at the two sites. Also, the researchers could not account for how long it rained or how much rain fell before overland flow started. But, since there were overland stormwater flows at the weirs each time, one would assume that all the surface depression storage requirements were met. More rainfall will probably be required to satisfy surface depression storages during the non-wniter than the winter period. This factor probably explains the higher runoff coefficients found in the winter than in the non-winter periods at both sites. Rainfall Amount and Duration. At a particular site, two storm events with the same total runoff amount but different durations can produce quite different runoff amounts. For example, the storms of 02/17/96 and 04/05/96 at the Burnaby site had the same rainfall amount of 3.4 mm but lasted 90 and 180 minutes and had runoff coefficients of 0.56 and 0.94 respectively. Monitoring Instrumentation. The accuracy of the rain gauges and flow monitoring devices can lead to variation in the runoff coefficient data obtained. According to Gupta et al. (1981a), runoff coefficient variation caused by inaccuracy in monitoring instrumentation can be as high as 20%, especially when small rainfall and flow volumes approach the instrument's detectable limits. The flow device used for this project has an accuracy of ± 0.25% of calibrated range (Unidata Australia, 1994), which corresponds to a variation in flow of ± 0.6%. No such accuracy calibration range was issued by the rain gauge manufacturer, hence the rain gauge measurements could be a potential cause of runoff coefficient variations in this research. Vehicular Splash-Off. Vehicular splash-off, which varies with traffic volume and speed during the storm, can be a source of runoff coefficient variation. Usually daylight hours are characterized by higher traffic volumes and faster mean speed except during the "rush hour" traffic, when traffic slows down dramatically at the Burnaby site. As a result, storms that occur during the daylight hours, with the exception of the "rush hour" periods, are more likely to produce less runoff due to more vehicular splash-off. Night or "rush hour" traffic storms are more likely to produce more 141 runoff (less vehicular splash-off) due to lower vehicle speed and less night vehicular traffic. There was no way to establish the amount of runoff lost through vehicular splash-off, however, visual observations made during sample collection indicated enough loss for vehicular splash-off to be considered a reasonable source of variation in runoff coefficient data analysis. At both sites, all the samples were collected during the daylight hours, however, the impact of vehicular splash-off on runoff coefficient variation is expected to be highest at the Richmond site. Although the Richmond site has one-half the traffic volume of the Burnaby site, it has no traffic congestion, unlike the eastbound Burnaby traffic. Therefore, the speeding vehicles are more likely to cause more splash-off, thereby producing less runoff at the Richmond site. Tables 4-6 and 4-7 tend to indicate similar vehicular splash-off patterns at both sites. Analysis of the difference between the measured total runoff volume and estimated total runoff volume shows that the overall percent volume difference was found to be 30% at the Burnaby site compared to 45% at the Richmond site. This indicates that similar runoff was lost at both sites. This was probably largely due to vehicular splash-off. An A N O V A analysis of the runoff volumes at the Burnaby and Richmond sites indicates that there is no significant difference between measured and estimated volumes at 95% confidence interval with F-ratio values 1.24 and 0.95, and P-values of 0.27 (DF 1,38) and 0.34 (DF 1, 22) respectively. However, there is runoff loss due to vehicular splash-off. The runoff loss is likely to affect the amount of pollutants generated at both sites, but it is difficult to estimate by how much. Perhaps the splash-off is just as polluted as the runoff and adds to the total pollutant load to the surrounding environment. Between the two sites the difference in measured runoff volume is statistically significant at 95% confidence level with an F-ratio value of 5.15 and a P-value of 0.03 (DF 1,30). 142 Table 4-6. Measured and Estimated Total Runoff Data at the Burnaby Site Date Site Rainfall (mm) Duration (min.) Measured Total Runoff Volume (m3) Estimated Total Runoff Volume (m3) % Volume Difference (%) 11/04/95 10.0 180 9.34 11.4-13.0(12.4) 24.4 01/06/96 12.5 180 11.5 14.7-16.2 (15.4) 25.5 02/07/96 5.10 100 4.37 5.99-6.62 (6.30) 30.6 02/17/96 3.40 90 2.34 3.99-4.41 (4.20) 44.3 02/19/96 1.80 60 1.92 2.11-2.33 (2.22) 13.7 03/08/96 4.00 180 5.84 4.69-5.19(4.94) -18.2 03/09/96 4.40 180 5.03 5.16-5.71 (5.44) 7.46 03/31/96 N / A N / A N / A N / A N / A 04/01/96 7.00 180 6.54 8.22-9.08 (8.65) 24.4 04/05/96 3.40 180 3.97 3.99-4.41 (4.20) 6.2 04/09/96 0.70 60 0.55 0.82-0.91 (0.86) 36.4 04/11/96 0.7 60 0.14 0.23-0.26 (0.25) 43.3 04/22/96 1.20 100 0.97 1.41-1.56 (1.48) 34.6 04/23/96 2.50 100 1.60 2.93-3.24 (3.09) 48.2 04/25/96 1.20 90 0.88 1.41-1.56(1.48) 40.6 05/07/96 1.80 90 1.34 2.11-2.33 (2.22) 39.7 05/17/96 0.80 60 0.50 0.94-1.04 (0.99) 49.4 05/18/96 5.70 90 3.79 6.69-7.39 (7.04) 46.2 06/10/96 2.80 120 2.07 3.29-3.63 (3.46) 40.2 07/02/96 0.85 60 N / A N / A N / A 07/17/96 1.75 90 1.34 2.05-2.27 (2.16) 38 08/02/96 7.80 100 6.75 9.15-10.1 (9.64) 30 Note: 1) Mean Winter Vol . % Difference = 18% 3) Overall Percent Difference = 30% 2) Mean Non-Winter Vol . % Difference = 37% 4) N / A - Not available. 143 Table 4-7. Measured and Estimated Total Runoff Data at the Richmond Site Date Site Rainfall (mm) Duration (min.) Measured Total Runoff Volume (m3) Estimated Total Runoff Volume (m3) % Volume Difference (%) 11/04/95 N / A N / A N / A N / A N / A 01/06/96 7.20 180 2.96 6.67-7.38 (7.03) 57.9 02/07/96 12.6 N / A N / A N / A N / A 02/17/96 0.80 90 0.64 0.74-0.82 (0.78) 18 02/19/96 N / A 60 N / A N / A N/A 03/08/96 3.40 180 2.84 3.15-3.49 (3.32) 14.4 03/09/96 5.40 180 2.61 5.00-5.54 (5.27) 50.5 03/31/96 N / A N / A N / A N / A N / A 04/01/96 6.30 120 3.38 5.84-6.46 (6.15) 45 04/05/96 1.00 120 0.53 0.93-1.03 (0.98) 45.7 04/09/96 N / A N / A N / A N / A N / A 04/11/96 N / A N / A N / A N / A N / A 04/22/96 2.50 90 1.44 2.32-2.56(2.44) 41 04/23/96 1.20 90 0.66 1.11-1.23 (1.17) 43.6 04/25/96 0.70 80 0.48 0.65-0.72 (0.68) 29.7 05/07/96 N / A N / A N / A N / A N / A 05/17/96 1.10 100 0.28 1.02-1.13 (1.07) 73.9 05/18/96 1.20 100 0.62 1.11-1.23 (1.17) 47.1 06/10/96 N / A N / A N / A N / A N / A 07/02/96 N / A N / A N / A N / A N / A 07/17/96 0.35 60 0.10 0.32-0.36 (0.34) . 70.7 08/02/96 N/A. N / A N / A N / A N / A Notes: 1) Mean Winter Vol . % Difference = 35% 3) Overall Percent Difference =45% 2) Mean Non-Winter Vol . % Difference = 50% 4) N / A - Not available. 144 Summary Both sites have higher average rainfall amounts and intensities in the winter than the non-winter. This also resulted in higher winter runoff volumes at both sites. The runoff coefficients (which relate the amount of runoff to the amount of rainfall), measured at both research stations are in the range of values measured at other sites in other research studies. Since the drainage areas at both sites are 100% paved, one would expect the coefficients to be almost 1.0 in both cases, instead of the 0.70 and 0.55 found. The most likely explanation seems to be vehicle splash-off. 4.2 Composite Data Analyses Composite data analyses refer to total suspended solids (T.S.S.) and trace metal concentrations/loadings. A l l the concentration and loading calculations were done using flow composite samples. The winter period is November to March and non-winter is April to October. 4.2.1. Total Suspended Solids (T.S.S.) Concentrations/Loadings T.S.S. Concentrations The total suspended solids (T.S.S.) concentrations and loadings obtained in this study are presented in Table 4-8 for both sites. The mean winter flow composite T.S.S. concentration at the Burnaby site was 261 mg/L, compared to a non-winter mean T.S.S. concentration of 257 mg/L. The overall mean concentration of T.S.S. at the Burnaby site was found to be 259 mg/L with a standard deviation of 86.4. The T.S.S. concentration range was found to be from 122 mg/L to 369 mg/L. The mean winter flow composite T.S.S. concentration at the Richmond site was 96.3 mg/L, compared to a non-winter mean T.S.S. concentration of 100 mg/L. 145 The overall mean T.S.S. concentration at the Richmond site was 98.2 mg/L with a standard deviation of 48.4. The T.S.S. concentration range was found to be from 38.0 mg/L to 154 mg/L, as can be seen in Table 4-8. At both Burnaby and Richmond sites, the differences in T.S.S. concentrations between winter and non-winter were not statistically significant at 95% confidence level. At the Burnaby site, the F-ratio was found to be 0.003 (P=0.96) with an r2 value of 1.00x10"3 at 1 and 6 degrees of freedom compared to the Richmond site with an F-ratio of 0.007 (P=0.94) with an r2 value of 2.00xl0"3 at 1 and 4 degrees of freedom. Between the two sites, the Burnaby site consistently had higher T.S.S. concentration than the Richmond site. For example, the overall mean T.S.S. concentration at the Burnaby site was 259 mg/L , almost triple the 98.2 mg/L obtained at the Richmond site. An analysis of variance (ANOVA) on T.S.S. data indicates a significant difference between the two sites with an F-ratio of 16.7, P-value of 2.00xl0"3 and an r2 value of 0.58 at 1 and 12 degrees of freedom. T.S.S. Concentration Discussion The concentration differences between the two sites can be attributed to precipitation percentage differences of 29% in the winter and 34% in the non-winter, higher rainfall intensity at the Burnaby site (more ability to wash off T.S.S.), higher runoff volume generation, more daily traffic (higher traffic density) and the age (Table 3-1) of the Burnaby site. The T.S.S. concentration ranges obtained at the two sites were within the ranges obtained in other highway stormwater studies, as can be seen in Table 4-9. The T.S-S. concentration range and mean were comparable to those obtained by Chui et al. (1981) in Seattle and Gupta et al. 146 Table 4-8. Concentrations and Loadings of T.S.S. at Burnaby and Richmond Sites Burnaby Richmond Units T.S.S. T.S.S. Number of Samples - 8 6.00 Minimum mg/L 122 38.0 Maximum mg/L 369 154 Average mg/L 259 98.2 Standard Deviation mg/L 86.4 48.4 Mean Winter Cone. mg/L 261 96.3 Total Winter Loading kg/yr 305-337(321) 39.9-44.1 (42.0) Mean Non-Winter Concentration mg/L 257 100 Total Non-Winter Loading kg/yr 130-143 (136) 18.3-20.2(19.3) Total Annual Loading kg/yr 458 61.3 Mean Export Coefficient kg/ha/yr 32.7xl0 2 681 Notes: 1) Winter Period - November to March. 2) Non-Winter Period - April to October. 3) Total Annual Stormwater Runoff: Burnaby = 1,760 (mVyr) Richmond = 629 (mVyr) 4) Total Annual Load (kg/yr) = (Total Volume (mVyr) (Approximate Mean Conc.Value (mg/L)) 1,000 5) Export Coefficient (kg/ha/yr) = Total Load (kg/yr) Catchment Area (ha) 6) Area: Burnaby = 0.14 ha Richmond = 0.09 ha 7) Numbers in bracket are mean values. 147 (1981) in Harrisburg, Nashville and Denver, USA, but were lower than those obtained in Milwaukee (Gupta et al., 1981). The higher T.S.S. values obtained at the Milwaukee and Denver sites can be attributed to more severe winter weather conditions which required more sanding/salt applications. The similarity in the T.S.S. concentration range between Burnaby, B C and Seattle, Washington, is probably due to similar climatic conditions. For example, 1-5 near Seattle, Washington, generated an overall mean T.S.S. concentration of 168 mg/L, which is in between the concentrations at the Burnaby and Richmond sites with mean T.S.S. concentrations of 259 mg/L and 98.2 mg/L respectively. The differences in T.S.S. concentration between the G.V.R.D. sites and Seattle could be due to differences in rainfall amount, traffic volume, number of lanes, maintenance practices and age of the highways. Asplund et al. (1980) stated neither the age of the highways nor the type of maintenance operations accorded to these highways with regards to their impacts on pollutant concentrations. Age and highway maintenance data were also lacking from all the other studies. As a result, T.S.S. concentration comparison between the sites mentioned in Table 4-9 can only take into account available factors such as average daily traffic, number of highway lanes and rainfall amount. In this case, 1-5 near Seattle with higher ADT (105,000 vehicles) and more lanes (4) is expected to generate higher T.S.S. concentrations than the Trans Canada Highway site in Burnaby and the New Westminster Highway site in Richmond with lower ADT (89,000 and 42,000 respectively) and a reduced number of lanes (2). Instead, the Burnaby site generated more T.S.S. than Seattle highways, although the Richmond site with a value of 98 mg/L and an annual rainfall of 1,230 mm has lower T.S.S. compared to the 119 mg/L with an annual rainfall of 1,130 mm obtained in Seattle. 148 Table 4-9. T.S.S. Concentration Comparison With Other Highway Studies T.S.S. (mg/L) Site Type Location Reference Range Mean 122-369 259 Urban Highways Burnaby, Canada This Study 38-154 98 Urban Highways Richmond, Canada This Study 14-552 119 A l l Highways Seattle, U S A Chui et al. (1982) 1.24a 26b Rural Highways U S A Rural Areas Shelley and Gaboury (1986) 0.36a 220b Urban Highways U S A Cities Shelley and Gaboury (1986) 1.78a 108b A l l Highways USA Cities Shelley and Gaboury (1986) 26- 1,576 268 Urban Highways Milwaukee, USA Gupta etal. (1981) 146-1,656 445 Urban Highways Milwaukee, U S A Gupta etal. (1981) 25 - 938 303 Rural Highways Milwaukee, U S A Gupta et al. (1981) 4-163 53 Urban Highways Harrisburg, U S A Gupta etal. (1981) 13-478 209 Urban Highways Nashville, U S A Gupta et al. (1981) 118-1,029 259 Urban Highways Denver, U S A Gupta etal. (1981) 2-342 23 Residential Vancouver, Canada Swain (1983) 6-1,520 242 Light Industrial Bumaby, Canada Lawson etal. (1985) 15 - 1,868 256 Urban Runoff Burnaby, Canada Hall and Anderson (1988) Notes: 1) a--Coefficient of Variation (Standard Deviation/Mean) 2) b - Median of US FHWA Data Base 149 Both the Bumaby and the Richmond sites highway T.S.S. concentration ranges are within the ranges reported by Lawson et al. (1985) and Hall and Anderson (1988) in their Burnaby stormwater analyses. A recent stormwater runoff analysis by Macdonald et al. (1996) reported mean T.S.S. concentrations of 54.8, 97.9 and 38.6 mg/L for Renfrew, Gilmore and Eagle stations in the Brunette River Watershed in Burnaby. These T.S.S. values are considerably lower than the Burnaby T.S.S. value but comparable to the Richmond site T.S.S. value. But the Richmond and Macdonald et al. (1996) reported mean T.S.S. values were lower than the previous T.S.S. values by Lawson et al. (1985) and Hall and Anderson (1988). Macdonald et al. (1996) did not give any reason for the deviation in their data from previous studies. Similarly, Swain (1983) reported a lower T.S.S. value (23 mg/L) in his stormwater runoff study in Vancouver. These differences in T.S.S. concentrations are probably attributable to differences in the catchment surfaces and rainfall amounts. These differences in mean T.S.S. concentrations also apply to T.S.S. loadings. T.S.S. Loadings The higher winter T.S.S. loading at both sites can be attributed to higher winter T.S.S. concentrations discussed earlier and to the higher winter rainfall. Also factors such as more road surface frost (reduce saltation), more sanding/de-icing agent application, less street sweeping and probably higher atmospheric fallout rates during the winter months, due to higher rainfall, can lead to higher T.S.S. loadings, as discussed earlier. Using the winter and non-winter runoff volume ranges calculated in Table A-3 in the Appendix, total winter and non-winter loading ranges were calculated, as indicated in Table 4-8. See Appendix C for sample calculations. The T.S.S. loadings followed the same pattern for both sites. The total winter and non-winter T.S.S. loading ranges at the Burnaby site were found to be 150 305-337 and 130-143 kg/yr respectively, with overall mean loadings of 321 and 136 kg/yr respectively. There was 57% more T.SiS. loading during the winter than the non-winter. The total annual T.S.S. loading at the Burnaby site was found to be 458 kg/yr with an export coefficient 32.7xl0 2 kg/ha/yr, as can be seen in Table 4-8. The export coefficient is the annual loading per unit area. The Richmond winter and non-winter T.S.S. loading ranges were found to be 39.9-44.1 and 18.3-20.2 kg/yr respectively, with overall mean loadings of 42.0 and 19.3 kg/yr respectively. There was approximately 54% more T.S.S. loading during the winter than non-winter at the Richmond site. The total annual T.S.S. loading at the site was found to be 61.3 kg/yr with an export coefficient of 681 kg/ha/yr. T.S.S. Loading Discussion In other research studies in the G.V.R.D., annual T.S.S. loading ranged from 224 to 2,010 kg/ha/yr., as shown in Table 4-10. The total annual loading in kg/ha/yr is also known as the export coefficient. The Richmond site export coefficient value of 681 kg/ha/yr falls within this range, but the Burnaby site value of 32.7x102 kg/ha/yr was significantly higher. The differences in export coefficients between sites are due to different reasons of which one of them is the percent impervious area. For example, the research report of Macdonald et al. (1996) in Burnaby had percent impervious surface areas that ranged from 24 to 50% compared to this study with 100% impervious highway surface area. As a result, one would expect the export coefficients from their study sites, from Table 4-10, to be less than those from this study. The higher unit area loading at the Burnaby site than at the Richmond site is probably due to the higher rainfall amount/intensity, more traffic and more wear and tear due to highway age, as discussed below. 151 Table 4-10. Comparison of T.S.S. Loadings in the G.V.R.D. Site Annual Loading (kg/yr) Export Coefficient (kg/ha/yr) Drainage Area (ha) Author Source Burnaby 458 32.7xl0 2 0.14 This report Highway Pvichmond 61.3 681 0.09 This report Highway Renfrew -Burnaby 357,878 473 756 Macdonald et al. (1996) Urban Runoff Gilmore -Burnaby 908,414 865 1,050 Macdonald et al. (1996) Urban Runoff Eagle - Burnaby 207,050 327 634 Macdonald et al. (1996) Urban Runoff Vancouver 2,898 224 12.95 Swain (1983) Urban Runoff Burnaby 11,658 2010 5.8 Lawson et al. (1985) Urban Runoff 152 Rainfall As shown in Table 3-1, the long term average annual rainfall is about 33% higher at the Burnaby site, than at the Richmond site, although, during the research period it was 39%. The Burnaby site thus has a higher potential for generating pollution, as suggested by Gupta et al. (1981), Chui et al. (1981), Wang et al. (1982) and Kobriger and Geinepolos (1984) in the USA and Colwill et al. (1984) in the U K . Colwill et al. (1984) suggested rainfall and the associated runoff events to be the dominant factor in determining the concentration and loading of T.S.S. discharged to receiving waters. ADT The Burnaby site has higher ADT at 89,200 vehicles compared to the Richmond site with an ADT of 41,967 vehicles, as indicated in Table 3-1. Based on the ADT data and overall T.S.S. mean loadings between the two sites, traffic volume among other factors seems to contribute to the higher T.S.S. loadings at the Burnaby site, contrary to the finding of Gupta et al. (1981) at the 1-794 site in Milwaukee. Wear arid Tear The wear and tear due to highway age is not conspicuously related to T.S.S. loadings, however, the higher T.S.S. loading at the Burnaby site may have been, partially, due to this factor. Since it is impossible to isolate the impact of highway age on T.S.S. loading, one can only infer that the Burnaby site, which was opened in 1964, will be subjected to more wear and tear due to age than the Richmond site which was opened in 1989, as shown in Table 3-1. Atmospheric Dustfall According to dustfall data collected from the G.V.R.D. Air Monitoring and Assessment Division, both sites have similar mean annual total suspended particulate (TSP) fallout rates at monitoring stations near the sites, as shown in Table 4-11. Burnaby and Richmond sites have overall annual mean TSP values of 42 and 40 ug/m3 respectively. Since the two sites have similar and very small atmospheric deposition rates of 5.1xl0"3 and 4.7xl0"3 kg/m for 153 Table 4-11. Total Suspended Particulate for Burnaby and Richmond Sites from 1994 - 1996 (pig/m3) Month of the Year BURNABY RICHMOND Year Monthly Average Year Monthly Average 1994 1995 1996 1994 1995 1996 January 51 49 27 42 32 34 24 30 February 53 51 60 55 40 44 103 62 March 61 45 35 47 45 40 34 40 April 46 50 25 40 32 87 24 48 May 47 45 N / A 46 38 46 N / A 42 June 31 44 N / A 38 22 43 N / A 33 July 54 49 N / A 52 40 45 N / A 43 August 48 40 N / A 44 71 31 N / A 51 September 56 46 N / A 51 46 48 N / A 47 October 40 36 N / A 38 35 31 N / A 33 November 34 19 N / A 27 25 16 N / A 21 December 27 23 N / A 25 29 27 N / A 28 Annual Mean 47 41 37 42 38 41 46 40 Data Source: G.V.R.D. Air Monitoring and Assessment Division (T19 & #32). Notes: 1) Winter Total Suspended Particulate (or Dustfall) - Burnaby = 39 //g/m3 (196 Aig/m3) - Richmond = 36 pig/m3 (180 /ug/m3) 2) Non-Winter Total Suspended Particulate (or Dustfall)-Bumaby = 44 fxg/m3 (309 Aig/m3) - Richmond = 42 /ug/m3 (294 Mg/m 3) 3) N / A - - N o t available. 4) Annual TSP mean - Burnaby = 42 Aig/m3 -Total (505 /ug/m3) - Richmond = 40 fxg/m3 - Total (474 /ug/m3) 5) Bold numbers in brackets are overall seasonal and annual TSP 154 Burnaby and Richmond sites respectively, their impact on T.S.S. concentrations and loading differences at the two sites is negligible. Summary Similar T.S.S. concentrations were obtained during the winter than the non-winter period at both sites, however the loadings were different. The differences in T.S.S. concentration were not statistically significant at both sites. The Burnaby site has significantly higher T.S.S. concentrations and loadings than the Richmond site and the difference is statistically significant with regards to T.S.S. concentration. The reasons for the T.S.S. concentration/loading difference between the two sites can be attributed to higher rainfall amounts, more traffic and probably more wear and tear due to highway age at the Burnaby site. The T.S.S. concentrations and loadings obtained in this study are within the range of and comparable to values obtained in other studies. 4.2.2 Heavy Metal Concentrations and Loadings Heavy Metal Concentrations Seasonal mean metal concentrations for both sites are shown in Table 4-12 and more detailed statistics are given for the Burnaby site in Table 4-13 and for the Richmond site in Table 4-14. The metals examined included Cd, Cr, N i , Cu, Fe, Pb, Zn, M n and Ca. Cd, Cr, and Ni were also considered but the concentrations were below the instrument's detection limits of 0.008, 0.05 and 0.04 mg/L respectively. The graphite furnace method can detect lower metal concentrations, but it was not used because the project funding agency was not interested in concentrations at such low levels. Only two measurable concentrations of Pb were obtained at the Burnaby site during 155 Table 4-12. Seasonal Mean Metal Concentrations at Both Sites (mg/L) Burnaby Richmond Metal Winter Mean Non-Winter Mean Overall Mean Winter Mean Non-Winter Mean Overall Mean Cu 0.07 0.08 0.08 0.04 0.05 0.05 Fe 6.74 5.56 6.15 3.26 3.87 3.57 Pb 0.122 ND 0.122 ND 0.482 0.482 Zn 0.26 0.30 0.28 0.16 0.13 0.15 Mn 0.14 0.15 0.15 0.09 0.10 0.10 Ca 3.81 5.63 4.72 3.89 4.77 4.33 Notes: 1) ND~Not detectable. 2) 2 -- Only two detectable concentrations were obtained. 3) Cr, N i and Cd concentrations were below detection. 156 Table 4-13. Burnaby Site Stormwater Runoff Statistics Parameter N Min. Max. Ave. Std. Cu 8 0.04 0.10 0.07 0.02 Fe 8 3.94 9.05 6.15 1.90 Mn 8 0.10 0.19 0.15 0.04 Zn 8 0.17 0.40 0.28 0.13 Ca 8 1.74 6.61 4.72 1.68 Notes: 1) Average is arithmetic mean. 2) N is total number of samples analysed. 3) A l l units in mg/L. Table 4-14. Richmond Site Stormwater Runoff Statistics Parameter N Min. Max. Ave. Std. Cu 6 0.03 0.07 0.04 0.01 Fe 6 2.19 5.28 3.57 1.35 Mn 6 0.05 0.13 0.09 0.03 Zn 6 0.01 0.24 0.14 0.09 Ca 6 1.66 6.50 4.32 1.64 Notes: 1) Average is arithmetic mean. 2) N is total number of samples analysed. 3) A l l units in mg/L. 157 the winter and two at the Richmond site during the non-winter periods. The mean winter Pb value of 0.12 mg/L at the Burnaby site and mean non-winter Pb value of 0.48 mg/L at the Richmond may not reflect the actual Pb levels around the two sites, since the rest of the Pb concentrations at both sites were below the instrument's detection limit, as can be seen from Table 4-12. The mean winter Fe concentration was higher than the corresponding non-winter mean value at the Burnaby site. At the Richmond site, the mean non-winter concentration of Fe was higher than the winter value. The winter mean Ca concentrations were lower than the non-winter values at both sites. Cu and Mn concentrations did not show much seasonal variation at either sites. Zn had higher non-winter mean concentration at the Burnaby site, whereas at the Richmond site the mean concentration of Zn was higher during the winter than the non-winter period. At both sites, the differences between winter and non-winter concentrations were less than 20% and not statistically significant at 95% confidence level for any of the metals. The rest of the other statistical analyses using the data sets for both sites are listed in Tables 4-13 and 4-14. Between the two sites, the Burnaby site consistently generated higher overall mean concentrations of Cu, Fe, Mn, Zn and Ca than the Richmond site. Analysis of variance (ANOVA) performed to test statistical significance show the differences in Cu, Fe, Mn, and Zn concentrations between the two sites to be significant with F-ratios that ranged from 7.50 (P=1.80xl02) and r2 of 0.39 for M n to 10.5 (P=7.00xl03) and r2 of 0.47 for Cu at 1 and 12 degrees of freedom. Ca, however, did not show any significant difference between the two sites. 158 Heavy Metal Concentration Discussion Although there was some seasonal variation in metal concentrations, the differences were small and not statistically significant. This is in line with the findings of other researchers, who reported similar trends in their observations. The reason for the lower metal concentration during the non-winter could be due to lack of high intensity rainstorms to wash off pollutants from road surfaces. Winter Fe concentration obtained in this research follows this trend, supporting the finding of Gupta et al. (1981) and Kobriger and Geinepolos (1984) in the USA. Zn, on the other hand, contradicted the above trend. The lack of seasonal variation found in this research with Cu was also reported in the U S A studies. However, the seasonal variation in Ca concentrations. obtained at both the Burnaby and Richmond sites did not support Kobriger and Geinepolos (1984) who reported Ca to show no seasonal trends. The seasonal pattern of pollutant concentration and loading seem to depend on the geographical location of the site. For example, there were no observable winter/non-winter variations with the data obtained in Sacramento, California by Kerri et al. (1985) or Orange County and Miami, Florida by Yousef et al. (1985), and McKenzie and Irwin (1983) respectively. The reasons given for higher winter pollutant concentrations and loadings where they occur are increased surface loading due to lack of sweeping, less atmospheric wind blow-off, more stop-and-go traffic due to dangerous winter conditions, reduced runoff caused by freezing winter roadway conditions, increased automobile body rusting due to winter de-icing agents, lack of regular highway maintenance to stop surface deterioration (Kobriger and Geinepolos, 1984; Lorant, 1992), and the use of studded tires which generated an estimated 20 to 50 g/km/vehicle of asphalt wear on Norwegian roadways (Lygren et al., 1984). 159 Metal concentrations were higher at the Burnaby site than the Richmond site. This can be attributed to higher ADT, surrounding land use activities and greater rainfall amount/intensity. Generally, Pb and Zn levels are highly influenced by traffic volume. Zn concentrations at the Burnaby site with higher vehicular traffic reflected this pattern. Kobriger and Geinepolos (1984) reported that the two sites, Milwaukee and Sacramento with high ADT values, have higher concentrations of metals, especially Pb and Zn. The lack of correlation between ADT and Pb concentrations observed between the Burnaby and Richmond sites can be attributed to limited detectable Pb data. The low levels of Pb were probably caused by the banning of leaded gasoline in Canada. Comparisons of the highway mean metal concentrations at both sites to the urban stormwater runoff data of Lawson et al. (1985) and Macdonald et al. (1996) in Burnaby for Cu, Zn and M n are discussed briefly. Urban stormwater runoff Cu concentrations observed at both Renfrew and Gilmore (Macdonald et al. 1996) were higher than the highway Cu concentration at the Richmond site (this study), but lower than that at the Burnaby site. The Renfrew, Gilmore and Eagle stations' Zn concentrations in the Burnaby urban stormwater runoff study of Macdonald et al. (1996) were lower than those at both the Burnaby and Richmond sites (this study) and the urban stormwater runoff Zn concentration found in the study of Lawson et al. (1985). Higher Zn concentration was observed with the highway runoff in Burnaby (this study) than urban stormwater runoff by Lawson et al. (1985), also in Burnaby. The Zn concentration reported in residential stormwater runoff in Burnaby by Swain (1983) was comparable to those reported at Gilmore and Renfrew by Macdonald et al. (1996), but they are all lower than highway Zn concentrations reported at both sites in this study. The Mn concentrations observed in all three urban stormwater stations in Burnaby by Macdonald et al. (1996) were higher than the highway 160 Mn concentration observed in Richmond (this study), but were similar to the Burnaby highway stormwater runoff M n concentration (this study). The Eagle station had higher Mn concentration. The Eagle station urban stormwater Mn concentration of Macdonald et al. (1996) was also higher than that of Lawson et al. (1985) in another section of Burnaby. Finally, the Ca concentrations reported in the highway stormwater runoff of the Burnaby and Richmond sites are within the ranges of Ca concentrations reported in urban stormwater residential runoff of 7 - 313 mg/L in Burnaby (Lawson et al., 1985) and 2 - 19 mg/L in Vancouver (Swain, 1983). Comparison of highway metal concentrations found by other researchers are given in Table 4-15. Also, Table 4-15 shows that the highest overall Fe mean concentration of 16.5 mg/L was obtained in Denver, whereas the lowest was 2.00 mg/L at the Harrisburg site (Gupta et al., 1981). The overall mean Fe concentration values of 6.15 and 3.57 mg/L for both the Burnaby and Richmond sites respectively are within this Fe concentration range. Like Fe, the overall mean Pb, Cu and Zn concentrations of 0.12, 0.08 and 0.28 mg/L, and 0.48, 0.05 and 0.15 mg/L found at Burnaby and Richmond sites respectively are within the ranges for these metals found at other sites, as can be seen in Table 4-15. The two measurable Pb levels could be due to lead oxide filler material from tire wear and tear, lubricating oil and grease or bearing wear-since leaded gasoline has been banned in Canada. Mn and Ca values obtained at both Burnaby and Richmond sites cannot be compared to other highway sites due to lack of other published M n and Ca concentration data. The mean Cr and Ni concentrations of 0.03 and 0.03 mg/L respectively reported by Gupta et al. (1981) and Yousef et al. (1985) were below the instrument's Cr and Ni detection limits of 0.05 and 0.04 mg/L respectively set for this research project. However, Cd measured by Yousef et al. (1985) had a mean overall concentration of 0.04 mg/L which is considerably higher than the instrument's detection limit of 0.008 mg/L set in this project. 161 Table 4-15. Comparison of Metal Concentrations in Highway Runoff Metal Concentrations (mg/L) Location Reference Fe Pb Cu Range Mean Range Mean Range Mean 3.94-9.05 5.90 0.01-0.14 0.12 0.04-0.10 0.07 Burnaby This Study 2.19-5.28 3.36 0.10-0.86 0.48 -0.04 0 Richmond This Study N / A N / A 0.08-1.17 0.47 0.01-0.28 0.04 Seattle Chui et al. (1982) N / A N / A 0.30-3.78 0.76 0.02-0.11 0.04 Seattle Chui et al. (1982) N / A N / A N / A 1.56 N / A 0.05 Seattle Chui et al. (1982) N / A N / A N / A 0.62 N / A 0.05 Florida Yousef et al. (1985) 2.5-4.30 3.5 0.80-13.1 2.90 0.01-0.66 0.16 Milwaukee Gupta et al. (1981) 5.6-45.0 14.6 0.40-6.6 1.20 0.01-0.88 0.14 Milwaukee Gupta et al. (1981) 1.1-43.6 14.9 0.05-0.70 0.21 0.01-0.23 0.08 Milwaukee Gupta et al. (1981) 0.1-6.6 2.0 .05-0.20 0.10 0.01-0.10 0.05 Harrisburg Gupta et al. (1981) 1.5-12.0 5.5 .02-1.70 0.50 0.01-0.20 0.07 Nashville Gupta et al. (1981) 6.5-37.0 16.5 0.30-1.80 0.45 0.03-0.26 0.11 Denver Gupta et al. (1981) N / A N / A 0.09-1.26 0.33 0.01-0.29 0.06 England Harrison & Wilson (1985) N / A N / A 0.01-2.00 0.19 N / A N / A France Balades et al. (1984) N / A N / A 0.01-1.50 0.24 N / A N / A France Balades et al. (1984) Notes: 1) N / A - - N o t available. 2) Cd, Cr and Ni were below detection most of the time. 162 Table 4-15. Comparison of Metal Concentrations in Highway Runoff (cont'd) Metal Concentrations (mg/L) Location Reference Zn Mn Ca Range Mean Range Mean Range Mean 0.17-0.40 0.35 0.10-0.19 0.14 1.74-6.61 4.39 Bumaby This Study 0.01-0.24 0.10 0.05-0.13 0.09 1.66-6.50 4.00 Richmond This Study 0.11-3.09 0.64 N / A N / A N / A N / A Seattle Chui etal. (1982) 0.20-4.77 0.97 N / A N / A N / A N / A Seattle Chuietal. (1982) N / A 0.50 N / A N / A N / A N / A Seattle Chui etal. (1982) N / A 0.3 N / A N / A N / A N / A Florida Yousef et al. (1985) N / A N / A N / A N / A N / A N / A Milwaukee Gupta et al. (1981) 0.14-3.40 0.69 N / A N / A N / A N / A Milwaukee Gupta et al. (1981) 0.20-1.90 0.55 N / A N / A N / A N / A Milwaukee Gupta et al. (1981) 0.07-0.34 0.18 N / A N / A N / A N / A Harrisburg Gupta et al. (1981) 0.01-0.23 0.08 N / A N / A N / A N / A Nashville Gupta et al. (1981) 0.33-1.50 0.72 N / A N / A N / A N / A Denver Gupta et al. (1981) N / A N / A N / A N / A N / A N / A England Harrison & Wilson (1985) .04-25.50 0.85 N / A N / A N / A N / A France Balades et al. (1984) 0.13-1.60 0.39 N / A N / A N / A N / A France Balades et al. (1984) Note: 1) N / A - - N o t available. 2) Cd, Cr and N i were below detection most of the time. 163 Heavy Metal Loadings Tables 4-16 and 4-17 show winter, non-winter metal and total annual loadings at both sites. Just like the metal concentrations, no loading data was obtained for Cd, Cr, N i and Pb. The winter loadings were consistently higher than the corresponding non-winter loadings at both sites, as presented in Table 4-16. For example, Cu winter loadings at both the Burnaby and Richmond sites were found to be 0.09 and 0.02 kg/yr, compared to non-winter Cu loadings of 0.04 and 0.01 kg/yr respectively. The loading trends in descending order for both winter and non-winter loadings at both sites are: Burnaby Winter: Fe > Ca > Zn > M n > Cu Burnaby Non-Winter: Ca > Fe > Zn > Mn > Cu Richmond Winter: Ca > Fe > Zn > Mn > Cu Richmond Non-Winter: Ca > Fe > Zn > Mn > Cu The non-winter loading pattern at the Burnaby site is exactly the same as both winter and non-winter loading patterns observed at the Richmond site and slightly different from the Burnaby winter loading pattern when the Fe loading is the highest. The other patterns show Ca loadings to be the highest and they all show Cu to have the least loading, irrespective of the season. Table 4-17 shows the total annual metal loadings and their corresponding export coefficients for both sites. The export coefficients express the annual loading per unit area (kg/ha/yr). The descending order of the total annual metal loading per unit area at both sites is as follow: Burnaby: Fe > Ca > Zn > Mn > Cu Richmond: Ca > Fe > Zn > Mn > Cu 164 Table 4-16. Seasonal Metal Loadings in Highway Runoff at Both Sites Metal Loadings (kg/yr) Burnaby Richmond Parameter Winter Non-Winter Winter Non-Winter Cu 0.08-0.09 (0.09) 0.04-0.04 (0.04) 0.02-0.02 (0.02) 0.01-0.01 (0.01) Fe 7.89-8.69 (8.29) 2.81-3.10(2.96) 1.35-1.49(1.42) 0.71-0.78 (0.75) M n 0.16-0.18 (0.17) 0.08-0.08 (0.08) 0.04-0.04 (0.04) 0.02-0.02 (0.02) Zn 0.31-0.34 (0.32) 0.15-0.18(0.16) 0.07-0.07 (0.07) 0.02-0.03 (0.02) Ca 4.53-4.91 (4.72) 2.84-3.14(2.99) 1.61-1.78 (1.70) 0.87-0.96 (0.92) Note:l) Seasonal metal loadings (kg/yr) =(Seasona1 Vo1ume(m3/yr))(Seasona1 Mean Conc.(mg/L)) 1,000 Table 4-17. Highway Runoff Metal Concentrations and Loadings at Both Sites Burnaby Richmond Parameter Total Annual Loading (kg/yr) Mean Export Coefficient (kg/ha/yr) Total Annual Loading (kg/yr) Mean Export Coefficient (kg/ha/yr) Cu 0.13 0.93 0.03 0.34 Fe 11.3 80.4 2.17 24.1 Mn 0.25 1.79 0.06 0.67 Zn 0.48 3.42 0.09 .1 Ca 7.71 55.1 2.62 29.1 Notes: 1) Drainage Area: Burnaby = 0.14 ha Richmond = 0.09 ha 2) Total Annual Stormwater Runoff: Burnaby = 1,760 (m3/yr) Richmond - 629 (m3/yr) 3) Total Annual Load (kg/yr) = (Total Volume (m3/yr) (Approximate Mean Conc.Value (mg/L)) 1,000 4) Export Coefficient (kg/ha/yr) = Total Load (kg/yr) Catchment Area (ha) 165 At the Burnaby site, Fe annual loading was the highest with a value of 11.3 kg/yr and an export coefficient of 80.4 kg/ha/yr, whereas Ca annual loading was the highest at the Richmond site with a value of 2.62 kg/yr and an export coefficient of 29.1 kg/ha/yr, as indicated in Table 4-17. Cu was the lowest with annual loadings of 0.13 and 0.03 kg/yr for both the Burnaby and Richmond sites respectively with corresponding export coefficients of 0.93 and 0.34 kg/ha/yr. Heavy Metal Loading Discussion Between the two sites, the Burnaby site consistently generated more Cu, Fe, Zn, Mn and Ca loadings than the Richmond site. The presumed reasons for higher metal loading at the Burnaby site than the Richmond site are the same as mentioned earlier for the higher metal concentrations, namely traffic volume and rainfall amount/intensity which are both higher at the Burnaby site. A comparison of the highway metal loadings obtained in this study with those from other urban stormwater studies in the G.V.R.D. is shown in Table 4-18. The two highway stormwater runoff metal loadings are lower than the residential metal loadings of Lawson et al. (1985) in Burnaby and two residential metal loadings reported by Swain (1983) in Vancouver. However, the metal export coefficients from this study, from Swain (1983) and Lawson et al. (1985) studies are higher than the residential/commercial metal export coefficients reported by Macdonald et al. (1996) in all three stations in Burnaby. For example, the export coefficients of Zn in highway runoff from Burnaby and Richmond were found to be 3.42 and 1.00 kg/ha/yr respectively in this study, compared to residential Zn values of 1.53 kg/ha/yr in Burnaby (Lawson et al., 1985), 1.13 kg/ha/yr in Vancouver (Swain, 1983), 0.87, 1.08 and 0.37 kg/ha/yr reported by Macdonald et al. (1996) at Renfrew, Gilmore and Eagle stations respectively in their Burnaby study. However, the 166 Table 4-18. Comparison of Metal Loadings in the G.V.R.D. Metal Loadings (kg/yr) Source Type Cu Fe Mn Zn Ca Area (ha) Location References Highway 0.13 11.3 0.25 0.48 7.71 0.14 Burnaby This Study Highway • 0.03 2.17 0.06 0.09 2.62 0.09 Richmond This Study Residential 1.40 325.00 7.90 8.90 1856.00 5.8 Burnaby Lawson et al. (1985) Residential 4.75 - - 14.60 - 12.95 Vancouver Swain (1983) Residential/ Industrial 432.00 - 907.00 656.00 - 756 Renfrew-Burnaby Macdonald etal. (1996) Residential/ Commercial 556.00 - 2324.00 1130.00 - 1050 Gilmore-Burnaby Macdonald et al. (1996) Residential/ Commercial 109.00 - 1447.00 232.00 - 634 Eagle-Burnaby Macdonald etal. (1996) Note: 1) Area = Highway Drainage Area 167 percent impervious surface areas in Macdonald et al. (1996) study were reported to range from 24 to 50% compared to 100% in this study. Consequently, metal loadings in their report are expected to be lower than those reported in this study as indicated above. Gupta et al. (1981) and Kobriger and Geinepolos (1984) observed similar environmental influential factors on metal loadings at Milwaukee, Harrisburg, Efland, Denver, Nashville and Sacramento. They attributed the similarities in seasonal metal loading trends to similar site characteristics such as percent paved and urban environment. However, the differences in metal loading between sites were attributed to traffic volume. Gupta et al. (1981) reported that since metals originate primarily from vehicles, the impact of ADT on metal loadings would seem to be more dominant than other environmental factors. However, an attempt to relate ADT to metal loadings at Denver, Nashville and Milwaukee did not show a clear relationship. The Denver site, with double the traffic volume of Nashville and Milwaukee Highway 45 sites, showed similar metal loadings. They attributed this poor correlation between A D T and metal loadings to an underestimation of measured metal loadings at the Denver site. Further ADT/ metal loading analyses using correlation between the sites indicated metal loadings to be more related to percent of the drainage area paved than the ADT. At the Burnaby and Richmond sites, the percentage of the drainage area paved is the same. Therefore, ADT and rainfall amount/intensity have more of an influence than other environmental variables when the two sites are compared. The Burnaby site with double the traffic volume, higher rainfall amounts/intensities and associated runoff volumes, higher duration and older in age generated more metal loadings than the Richmond site. 168 Summary The heavy metal concentrations/loadings observed in this research are comparable to measured values obtained in other research studies. The Burnaby site generated higher heavy metal concentrations/loadings than the Richmond site and this was probably due to more rainfall amount/runoff volume, higher ADT, more wear and tear due to highway age. Although there were seasonal differences in the concentrations of most metals with the exception of Mn and Cu, these differences were not statistically significant at either sites. 4.3 Discrete Data Analyses 4.3.1 Pollutographs Discrete samples, collected at constant time intervals throughout the runoff event, defined the pattern of pollutant discharge. Six sets of discrete samples were collected and analysed for the pollutographs, but only three sets are shown in these graphs to show pollutant patterns. The three event graphs shown are one winter storm event for each site (11/04/95 and 01/06/96) and one non-winter event (04/01/96) collected at both sites. T.S.S. pollutographs for the remaining three sets of discrete samples and other pollutant data are shown in Figures C1-C3 and Tables C1-C3 respectively in Appendix C. A l l the unused remaining three sets of discrete samples are for non-winter pollutant data collected only at the Burnaby site. The pollutant and flow measurements were used to show the concentration variation with time for T.S.S., Cu, Fe, Mn, Zn, Pb and Ca at both Burnaby and Richmond sites, as shown in Figures 4-1 to 4-20. The storm event of 11-04-95 was only monitored at the Burnaby site, as indicated in Figures 4-1 to 4-7. The rainfall began about 30 minutes before the first runoff sample was collected. The highest flow peak occurred 90 minutes into the sample collection period with a peak flow of close to 1.40xl0"3 m3/s. The T.S.S. concentration displayed three distinct peaks (tri-169 modal), while the flow showed multiple peaks, as indicated in Figure 4-1. The rainfall lasted for three hours, generating flow and T.S.S. data for the entire time. Also indicated in Figures 4-1 to 4-7 is the pattern of pollutants remaining at the sampling location downstream of the 30 m grassed channel below the point of discharge from the road with respect to time. The removal efficiencies of these pollutant parameters will be further discussed in a later section. Nevertheless, all the figures suggest definite reduction in pollutants after going through the 30 m grassed drainage ditch with the exception of Ca, in Figure 4-7. With the exception of Ca, all the metals, including T.S.S. in Figure 4-1, seem to show some kind of a first flush effect in the storm event of 11/04/95. The parameters responded to the initial phase of the storm. Also, all the parameters responded to the highest peak flow which occurred 90 minutes into the sample collection, with the exception of Pb. Also, there seems to be a relationship between T.S.S. and the metals in Figures 4-1 to 4-7. A l l the metal peak patterns are similar to the pattern observed with T.S.S. in Figure 4-1, indicating an association. Flow and parameter concentration measurements taken in Richmond during the storm event of 01-06-96 show the patterns in Figures 4-8 to 4-10. This flow also has multiple peaks. Most of the higher pollutant concentrations were observed later in the storm event. The initial phase of the storm probably soaked the highway surface with the middle part, with a second to highest flow peak of close to 5.00xl0"4 m3/s, dislodging and beginning the transportation of the pollutants. The highest flow peak of close to 6.00x10'4 m3/s occurred within the first 50 minutes of the storm, however, it failed to carry much of the pollutant parameters in the flow, as shown in Figures 4-8 to 4-10. The storm event of 01/06/96 did not show any first flush effect with all the metals including T.S.S. But the relationship between T.S.S. and metals was observed. A l l the metals seem to 170 follow the same concentration pattern observed with T.S.S. The metal concentration peaks seem to increase during the second phase of the storm following the T.S.S. concentration pattern shown in Figure 4-10. Since the storm event of 04-01-96 was monitored at each of the two sites, it provides an excellent comparison of flow and pollutant concentration patterns at both sites. Figures 4-11 and 4-20 illustrate typical monitoring results for the Richmond and Burnaby sites. Runoff monitoring was initiated at both sites at the same time. The highest peak flow of close to l.OOxlO"3 m3/s occurred at the Richmond site 60 minutes into the storm, whereas the highest concentration of most pollutants such as T.S.S., Zn, Fe, Ca and Mn occurred about 70-80 minutes into the storm, as shown in Figures 4-11 to 4-13. Like the storm of 01-06-96, the initial part of this storm probably dislodged the pollutants while the actual transportation was provided by subsequent lower flows. Some of the pollutants like T.S.S., Zn, Fe and Ca have three distinct concentration peaks with the higher concentration peaks occurring in the middle of the storm event, as can be seen from the figures. These concentration patterns lack the "first flush" effect claimed by other researchers. At the Burnaby site, the highest peak flow of 1.40x10"3 m3/s occurred 70 minutes after the storm began, as indicated in Figures 4-14 to 4-20. Also, higher pollutant peak concentrations occurred after the first 60 minutes into the storm, with the exception of Cr which showed two distinct peaks during the initial storm phase, as shown in Figure 4-15. Also, for the same storm event at the Burnaby site, most of the metals followed the T.S.S. concentration pattern with exception of Cr, as shown in Figure 4-15. Figures 4-14 to 4-20 show reduction in pollutant parameters after going through the grassed ditch. But Cr, Figure 4-15, and Ca, Figure 4-20, concentrations were removed the least at the Burnaby site. 171 INITIAL CONC. CY1) FLOW (Y2) CONC. A F T E R 30m ( Y l ) Figure 4-1. T.S.S. Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) - T 1 1 1 1 1 1 1 1 1 1 r 12.4 13 13.2 13.4 14 14.2 14.4 TIME OF THE D A Y (Hosn) OOOS 2 T V 1S IS.2 INITIAL CONC. CY1) F L O W (Y2) - C O N C . A F T E R 30m CY1) Figure 4-2. Copper Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) 172 0.0014 o -j 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ro 12.4 13 13.2 13.4 14 14.2 14.4 IS 15.2 T I M I OF T H E D A Y ( B u n ) INITIAL CONC. ( Y l ) - CONC. AFTER 30m ( Y l ) PLOW (Yl) Figure 4-3. Iron Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) 0.2 - | ; 0.0014 o H 1 1 1 1 i 1 1 1 1 1 1 1 1 1 i 1 ro 12.4 13 13.2 13.4 14 14.2 14.4 IS 1S.2 TIME OF THE D A Y (HODlf) INITIAL CONC. ( Y l ) - CONC. AFTER 30m ( Y l ) FLOW (Y2)' gure 4-4. Manganese Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) 173 INITIAL CONC. CY1) PLOW (Y2) CONC. AFTER 30m (Yl) Figure 4-5. Zinc Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) O.OOOB O INITIAL CONC, CV1) FLOW (Yl ) - CONC. AFTER. 30m CY1) Figure 4-6. Lead Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) 174 3r " Y V - O.OOOg O - o.oooe ir «— 0.0004 — 1 1 12.4 13 — i 1—•—i 1 1 1 1 1 1 r 13.2 13.4 14 14.2 14.4 TIME OF THE D A Y (Mann) —I 1 1 IS IS.2 INITIAL CONC. CY1) FLOW (Y2) CONC. AFTER 30m (Y 1) Figure 4-7. Calcium Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #1 (11-04-95) Co CONC. (Yl) FLOW (Y2) Zn CONC. (Yl) Mn CONC. (Yl) Figure 4-8. Copper, Zinc, and Manganese Concentration Patterns at the New Westminster Highway, Richmond, Event #2 (01-06-96) 175 — 0.0003 — O . O O O l Tm CONC. <Y1> - Ca CONC. CY1) FLOW CV2> Figure 4-9. Iron and Calcium Concentration Patterns at the New Westminster Highway, Richmond, Event #2 (01-06-96) 3 o T.S.8. CONC. (Yl) FLOW fY2) Figure 4-10. T.S.S. Concentration Pattern at the New Westminster Highway, Richmond, Event #2 (01 -06-96) 176 — 0.0004 T . S . S . C O N C , C V I ) Figure 4-11. T.S.S. Concentration Pattern at the New Westminster Highway, Richmond, Event #9 (04-01-96) x \ . - - A \ A 7 — O . O O O g — 0.0006 —i 1 1 1 1 1 1 1 1 1 1 1 1 1 — 9.3 9.5 l O . l 10.3 lO.S 11.X 11.3 11.S T I M E O P T H E D A Y (Honn) Co C O N C . <V 1> F L O W CV2) - - Z n C O N C . ori> - - M D C O N C . ( Y 1 } Figure 4-12. Copper Zinc, and Manganese Concentration Patterns, New Westminster Highway, Richmond, Event #9 (04-01-96) 177 s <3 F « C O N C . CY 1> Cm C O N C . ( Y l ) T 1 " O P L O W ( Y 2 ) Figure 4-13. Iron and Calcium Concentration Patterns at the New Westminster Highway, Richmond Event #9 (04-01-96) — 0.0014 , — O . O O l O.OOOB — 0 . 0 0 0 2 10.3 l O . S 11.1 11.3 T I M B O F T H E D A . Y ( H o u n ) X . S . S . C O N C . ( Y l ) F L O W ( Y 2 ) — X . S . S . C O N C . A F T E R 3 0 m ( Y l ) Figure 4-14. T.S.S. Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) 178 s £ 0.O3 ' 10.3 10. S 11.1 11.3 T I M E O F T H E D A Y ( H o n n ) C r C O N C . <Y1> F L O W ( Y 2 ) C r C O N C . A F T B K l O n CY1> Figure 4-15. Chromium Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) Figure 4-16. Copper Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) 179 Figure 4-17. Zinc Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) Figure 4-18. Iron Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) 180 Figure 4-19. Manganese Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) Figure 4-20. Calcium Concentration Patterns at the Trans-Canada Highway, Burnaby, Event #9 (04-01-96) 181 Discussion At the Burnaby site, the concentration of pollutants were consistently lower after it had gone through a 30 m grass drainage ditch, with the exception of Ca in Figure 4-7. T.S.S. seems to be the parameter that was most removed as shown in Figures 4-1 and 4-14, whereas Ca, as indicated in Figures 4-7 and 4-20, was removed the least. The increase in Ca after the runoff had gone through the drainage ditch, in Figure 4-7, could be due to N a + replacing C a 2 + ions at the cation exchange sites in the soil. The Burnaby site has higher rainfall amounts/intensities and this has translated into higher flow and pollutant concentrations, as can be seen from these pollutographs. Also evident in these graphs is the effectiveness of the 30 m grass drainage ditch at the Burnaby site in pollutant removal. Pollutants were consistently and continuously removed during the entire storm events with the exception of Cr and Ca, as shown in Figures 4-15 and 4-20. Earlier analyses of pollutant composite samples indicated that Cr levels were below instrument's detection limit. However, some of the discrete samples from the storm of 04-01-96 at the Burnaby site had measurable levels of Cr with the highest peak concentration of over 0.07 mg/L occurring 80 minutes after the storm began. This Cr concentration peak occurred ten minutes after the highest flow peak, as shown in Figure 4-15. The above pollutographs show one indication of the "first flush" phenomenon in which the initial portion of the runoff contains the highest loading of pollutants as reported by many other researchers. First flush effect was observed for T.S.S., Cu, Fe, Mn and Zn for the storm event of 11/04/95. The other events did not show any first effect. For the other rainfall events, the peak flow and pollutant concentration measurements obtained tend to occur considerably into the storm most of the time. In addition, the highest peak flow and pollutant concentration do not generally occur concurrently. Thus, the data is inconclusive about the occurrence of the "first flush" effect. 182 At the Richmond site, the flows were considerably lower, thereby resulting in decrease in pollutant transportation. Although the metal concentration patterns tend to follow the T.S.S. concentration patterns, showing some association, the storms did not appear to show any first flush effect, as can be seen from Figures 4-8 to 4-13. The amount of rainfall, intensity, antecedent number of dry days, and runoff volume are important factors that affect the concentration of contaminants in highway runoff. According to McKenzie and Irwin (1983), the influence of these factors on contaminant concentration/loading depends to a large extent on the parameter concentration (the amount of pollutants available for washoff). For example, none of the storm events observed at both sites generated the same flow patterns and amount of pollutants, as indicated in all the figures. Similarly, a comparison of all the storm events shows different flow patterns and amounts of pollutant between different storms at the two sites. Between the two sites, the Burnaby site generated higher flows and pollutant concentrations than the Richmond site, as shown in Figures 4-1 to 4-10. The storm event of 04/01/96 collected and analyzed at both sites, indicated similar highest peak flows at both sites, as shown in Figures 4-11 to 4-20. The factors that affect rainfall and pollution differences between the two sites were discussed earlier. The same factors such as climate, rainfall amounts/intensities, basin drainage area, percent of area paved and the configuration of the drainage area probably also affect the pollutant concentration patterns. The percent of the area paved, climate and the configuration of the drainage systems are almost the same, however, the rainfall amount/intensity and the area of the drainage basin are higher at the Burnaby site, which resulted in a higher runoff coefficient at the site, as indicated earlier. Differences in flow and pollutant concentration patterns similar to those observed between 183 the Bumaby and Richmond sites were also observed by Gupta et al.(1981) in their Milwaukee study. This suggests that there is considerable variability in flow and pollutant concentration even between sites that are relatively close together. Gupta et al. (1981) did not mention the "first flush" phenomenon in their Milwaukee study, suggesting that it probably was not detected there. Summary The flow and pollutant concentration patterns at both sites indicated multiple peaks with the highest peaks occurring several minutes to sometimes close to an hour into the storm. This observed patterns at the Burnaby site tend to indicate that the earlier flows will soak and simultaneously dislodge the pollutants while the actual transport is provided by the middle peak flows. The pollutograph data is comparable to other research studies, and does confirm the . existence of at least one "first flush" phenomenon as reported by some of the other studies. 4.3.2 Oil and Grease Oil and Grease Concentration Flow composite samples were used to determine oil and grease concentrations as well as loadings at both sites. Table 4-19 presents the ranges and means for both winter and non-winter concentrations at the two sites. At the Burnaby site, the mean winter concentration was 21.1 mg/L, with a range from 10.7 to 46.8 mg/L, while for the non-winter period, the mean concentration was 28.6 mg/L, with a range from 19.6 to 65.0 mg/L. Corresponding values for the Richmond site were 6.17, with a range from 3.60 to 10.2 mg/L in winter and 8.81, with a range from 6.02 to 12.2 mg/L in the non-winter period. The lower winter concentrations are probably due to more antecedent dry days during the non-winter than the winter months. Table 4-20 shows the overall means to be 24.6 and 7.38 mg/L with standard deviations of 1.84 and 1.56 mg/L for 184 Table 4-19. Highway Runoff Oil and Grease Concentrations in the G.V.R.D., USA and UK (mg/L) Location Winter Range Winter Mean Non-Winter Range Non-Winter Mean Reference Burnaby, G.V.R.D. 10.7-46.8 21.1 19.6-65.0 28.6 This Study Richmond, G.V.R.D. 3.60-10.2 6.17 6.02 -12.2 8.81 This Study Milwaukee, USA N / A 13.00 N / A 5.00 Kobriger & Geinepolos (1984) Sacramento, USA N / A N / A N / A 11.0 Kobriger & Geinepolos (1984) Denver, USA N / A N / A 3.00- 55.0 14.0 Gupta et al. (1981) Nashville, USA 11.0-57.0 27.0 2.00 - 9.00 4.00 Gupta et al. (1981) Toddington, U K 8.00 - 40.0 N / A 6.00 - 23.0 N / A Colwill et al. (1984) Notes: 1) G.V.R.D. - Greater Vancouver Regional District, B C , Canada 2) N / A - Not available. 185 both Burnaby and Richmond respectively. Also shown in the table are the oil and grease concentration ranges for both sites. Between the two sites, the Burnaby site with double the traffic generated more oil and grease. In fact, the mean oil and grease concentration at the Burnaby site is three times the level observed at the Richmond site, as indicated in Table 4-20. The difference in oil and grease concentrations between the two sites was statistically significant with an F-ratio of 16.6, a P-value of 2.00xl0' 3, and an r 2 value of 0.58 at 1 and 12 degrees of freedom. Oil and Grease Concentration Discussion During the winter period in the G.V.R.D., the nearly continual precipitation does not allow time for the oil and grease to accumulate on the road surfaces. This leads to lower oil/grease concentration per storm event than in the summer. Similarly, there may be a dilution effect due to the higher rainfall amounts during the winter period. Thus, the potential for complete oil/grease wash-off is higher during the winter which results in little road surface accumulation, i f any. Both winter and non-winter oil/grease concentrations observed in this study are comparable to those found in other research studies. The Richmond mean oil and grease concentration of 7.38 mg/L is comparable to Lawson et al. (1985) oil/grease value of 7.80 mg/L in a residential/industrial site in Bumaby. However, both the Richmond site oil/grease value (this study) and Lawson et al. (1985) oil/grease value were higher than Swain's (1983) oil/grease value of 3.00 mg/L obtained in a residential runoff in Vancouver. A l l these other oil/grease values were lower than the Bumaby highway stormwater oil/grease concentration obtained in this study. This might have been due to more vehicles passing through the Bumaby site than these other sites. The research studies of Swain (1983/85) and 186 Lawson et al. (1985) did not evaluate any variation in oil/grease concentrations due to seasons. At the Milwaukee and Nashville sites in the USA, the findings of both Kobriger and Geinepolos (1984) and Gupta et al. (1981) contradicted the higher non-winter oil and grease concentration patterns observed in the G.V.R.D. study. The higher winter oil and grease levels observed at the Milwaukee site could have been due to the limited oil and grease data set used in their analysis, or it could have been due to the severity of the winter in Milwaukee which can limit and affect oil and grease sample collection. The winter conditions allow more time for accumulation which would result in higher oil and grease levels when the conditions are conducive for sample collection. Oil and Grease Loading The oil and grease loadings at both sites were higher in the winter than the non-winter period, as shown in Table 4-20. The winter oil and grease loadings at the Burnaby and Richmond sites were 26.0 and 2.69 kg/yr, compared to non-winter values of 14.4 and 1.69 kg/yr respectively. The total annual oil and grease loadings were found to be 41.2 and 4.38 kg/yr with corresponding export coefficients of 294 and 48.7 kg/ha/yr for Burnaby and Richmond sites respectively. In terms of loading per unit area of the road surface, the Burnaby site had a value of 0.03 kg/m2/yr compared to the Richmond site with a value of 0.01 kg/m2/yr. Consistently, the Burnaby site generated both more oil and grease concentrations and loadings than the Richmond site. Oil and Grease Loading Discussion The higher winter oil and grease loadings at both sites could be due to higher rainfall amounts/intensities. The higher non-winter loadings, as indicated in Table 4-21, at the sites studied by Gupta et al. (1981) could have been due to not enough winter oil and grease data 187 Table 4-20. Seasonal Oil and Grease Loadings and Statistics at Both Sites Units Burnaby Richmond N - 8 8 Min. mg/L 10.7 3.60 Max. mg/L 65.0 12.2 Mean Cone. mg/L 24.6 7.38 Std. Deviation mg/L 1.84 1.56 Winter Loading kg/yr 24.7-27.2 (26.0) 2.55-2.83 (2.69) Non-Winter Loading kg/yr 14.4-16.0(15.2) 1.61-1.78 (1.69) Total Loading kg/yr 41.2 4.38 Export Coefficient kg/ha/yr 294 48.7 188 Table 4-21. Overall Loading of Oil and Grease in the G.V.R.D., USA and UK Location Winter (kg/ha) Non-Winter (kg/ha) ADT Reference Burnaby, G.V.R.D. 0.51 0.39 89,000 This Study Richmond, G.V.R.D. 0.09 0.05 42,000 This Study 1-794, Milwaukee, USA 1.16 2.20 53,000 Gupta etal. (1981) Hwy. 45, Milwaukee, U S A 0.27 1.71 85,000 Gupta etal. (1981) Grassy Site, Milwaukee, USA 0.03 0.38 85,000 Gupta et al. (1981) Nashville, U S A 0.58 1.14 88,000 Gupta etal. (1981) Denver, U S A 0.39 3.49 149,000 Gupta et al. (1981) Toddington, U K N / A N / A 33,000 Colwill et al. (1984) Note: 1) N / A - Not available. 189 due to frost effect, lower rainfall amounts/intensities to wash-off oil and grease from highway surfaces, and reduced traffic volume during the winter months. According to Gupta et al. (1981), the unusually high oil and grease concentration/loading at the Nashville site was probably due to the possible oil and grease contribution from a truck maintenance company with a small drainage that discharges into the study site drainage system. The Burnaby site has mild winters, much more trucking activities, higher rainfall amounts/intensities and higher ADT, thus generating more oil and grease than the Richmond site. Both Burnaby and Richmond sites have oil and grease loadings (kg/ha) comparable to the USA sites, as indicated in Table 4-21. Comparison of these oil and grease loading values to the only other known oil and grease loading value in the G.V.R.D. shows that the Burnaby residential oil/grease loading of 279 kg/yr (48.1 kg/ha/yr) obtained by Lawson et al. (1985) is higher than the Burnaby highway oil/grease loading of 41.1 kg/yr (294 kg/ha/yr) and the Richmond value of 4.38 kg/yr (48.7 kg/ha/yr). But in terms of the export coefficients, the Richmond site (this study) and Lawson's et al. (1983) Burnaby residential site values were similar. As mentioned earlier, the reason for the higher oil/grease concentration, and now loading, at the Burnaby site is probably due to the traffic volume. Summary Even though the oil/grease concentrations did not show definite seasonal patterns at both sites, the loadings did. The winter oil/grease loadings were higher than the corresponding non-winter values at both sites. Between the two sites, the Burnaby site generated higher oil/grease concentration and loading values than the Richmond site, however the difference is only statistically significant for oil/grease concentration. Both oil/grease concentration and loading values are comparable to those found in other studies. The oil/grease data seem to indicate that 190 rainfall amount and associated runoff volume, ADT, and drainage area size are some of the environmental factors that influence pollution potential at both sites. 4.4 T.S.S. and Metal Removal Efficiency Study 4.4.1 T.S.S. and Metal Removal Efficiencies The pollutant removal efficiency was calculated using the mean concentration differences, since flow measurement within the grass drainage ditch was unattainable. Pollutant removal efficiencies could not be performed at the Richmond site due to the structure of the grass drainage ditch. At the Burnaby site, pollutant removal efficiencies for T.S.S., Cu, Fe, Mn, Zn and Pb are tabulated on Tables 4-22 and 4-23. Cr, Cd and Ni were below the instrument's detection limits. More Cu and Pb are removed during the winter months. In fact, no Pb was detected at the Burnaby site during the non-winter period, as shown in Table 4-22. T.S.S., Fe, Mn and Zn removal efficiencies are higher during the non-winter months than in winter. The higher mean removal efficiency for M n during the non-winter period was probably due to the fact that only two of the fourteen storm events analysed produced detectable M n levels. The descending order winter and non-winter pollutant removal efficiencies at the Burnaby site are: Winter: T.S.S. > Fe > Zn/Pb > Cu > M n Non-Winter: T. S. S. > Fe > Zn/Mn > Cu (no Pb detected) The pollutant removal efficiency patterns at the site for both winter and non-winter are almost the same. The overall mean pollutant removal efficiencies at the Burnaby site are 77, 64, 56, 53, 49 and 48% for T.S.S., Fe, Zn, Pb, Mn and Cu respectively. The overall pollutant removal 191 Table 4-22. Seasonal Grass Drainage Ditch Pollutant Removal Efficiencies at the Burnaby Site. Pollutant Removal Efficiencies (%) Pollutant Winter Winter Non-Winter Non-Winter Parameters Range Mean Range Mean T.S.S. 58-83 72 70-90 81 Cu 41-66 49 9-68 46 Fe 50-72 63 34-86 65 M n 13-75 38 56-63 59a Zn 40-61 53 46-80 59 Pb 39-71 53a ND ND Notes: 1) a - Only two samples detected. 2) ND ~ Not detectable. Table 4-23. Summary of Mean Pollutant Removal Efficiencies at Other Sites Pollutant Removal Efficiencies (%) Site Ditch Length T.S.S. Cu Fe Mn Zn Pb Cd #of Samples Burnaby, G.V.R.D. 30 m 77 48 64 49 56 53* N / A 20 Seattle, Washington 31 m N / A 54 53 51 31 70 60 1 Seattle, Washington 30-40 m N / A 51 60 49 55 52 N / A 2 Epcot, Florida 170 m N / A N / A 0 N / A 62 57 43 N / A Maitland, Florida 53 m N / A N / A 69 N / A 85 N / A N / A N / A Notes: 1) N / A — Not available. 2) 0 - No pollutant removal. 3) * - Only winter value obtained. 192 efficiencies in descending order are: T.S.S. > Fe > Zn > Pb > M n > Cu The differences in pollutant removal efficiencies between the weir and 30 m-mark concentrations were statistically significant at 95% confidence level at the Burnaby site except for Mn. Mn has an F-ratio of 0.90 (P=0.42) with r2 value of 0.05 at 2 and 37 degrees of freedom. The rest of the parameters have F-ratios that range from 11.7 (P=1.12xl0"4) for Fe to 16.5 (P=7.00xl06) for Zn at 2 and 37 degrees of freedom. Similarly, the r2 values range from 0.39 for Fe to 0.47 for Zn. Discussion The overall indication is that the grass drainage ditch was effective in pollution reduction at the Burnaby site. Pb has low solubility and strong association with particulate matter (Wang et al., 1982), therefore the increase in vehicular generated turbulence and wind action during the non-winter months can enhance its removal from highway surfaces. Hence, the undetectable Pb levels observed at the Burnaby site during the non-winter period could have been due to saltation, which refers to injection of sand-sized particles into the atmosphere by vehicular turbulence (Kobriger and Geinepolos, 1984) as well as to the banning of leaded gasoline in Canada (Onwumere and Russell, 1996). Contrary to this study, Pb was removed more effectively than other metals, in the Washington State study, due to its low solubility and strong association with solids. In the Burnaby study, T.S.S., Fe, and Zn were removed the most with Cu and M n being the least removed. The reasons for differences in metal removal efficiencies at the same site and between sites are probably due to grass/drainage conditions, flow rate, metal species, pH and pollutant availability. Although the chemistry of heavy metals in natural water is fairly complex and site-specific, the type of metal species, for example, is an important influence in metal removal 193 efficiencies. According to Yousef et al. (1985), metal species that are present as charged ions like Fe and Zn are removed more effectively. They concluded that metal removal efficiencies are governed by the predominant ionic species and complexes, since charged species are retained by sorption processes. This explains why, apart from the T.S.S., Fe and Zn consistently show higher removal efficiencies at the Burnaby site. Also, the grass ditch at the Burnaby site was well maintained and therefore yielded good pollutant removal capacity compared with the sub-optimal pollutant removal efficiencies at the sparsely vegetated and less maintained Epcot site. Other factors such as flow rate and pollutant availability are influenced by climate, ADT, surrounding land use, dustfall, highway configuration and maintenance practices. However, the higher the flow rate the lower the pollutant removal potential (reduced retention time). pH, on the other hand, influences metal bioavailability. The lower the pH (more acidic), the higher the potential for metal pollution. pH at the Burnaby site was not very acidic. Overall, removal of pollutants by grass drainage ditch may occur through sorption, sedimentation, precipitation, co-precipitation and biological uptake processes. Several years of related studies resulted in the conclusion that grass drainage ditch pollution reduction was relatively high and increased with ditch length. According to Wang et al. (1982), Little et al. (1983) and Horner, (1985), a 60 m long slightly sloped grassed drainage ditch channel can remove 60 to 80 percent of the pollutants such as Pb, Zn, Cu, and T.S.S. from highway runoff. What was not explained was why at some sites the same efficiency in pollutant removal was achieved at shorter distances (less than 60 m). Similarly, nothing was said about the grass conditions at these sites, the type of maintenance operations such as mowing or debris removal involved, the slope of the ditch and the frequency of maintenance while the study was in progress. 194 The removal efficiency values reported at the Burnaby site were achieved with a 30 m grass drainage ditch. The values are better than the removal efficiencies reported by Yousef et al. (1985) at their Epcot and Maitland sites. For example, they reported Zn and Pb removal efficiencies of 62 and 57% respectively at their Epcot site over a grass drainage ditch length 170 m, compared to the Burnaby site with Zn and Pb removal efficiencies of 56 and 53% respectively over a 30 m grass drainage ditch, as shown in Table 4-23. Yousef et al. (1985) attributed their low pollutant removal efficiencies with the 170 m grass drainage ditch at the Epcot site to high ground water table, little vegetation and constant saturation of the soils with water during most of the study period. But they reported improved pollutant removal efficiencies with a 53 m grass drainage ditch at the Maitland site. Nevertheless, considering that the Epcot and Maitland sites' grass drainage ditch lengths were 5 and 1 lA times the Burnaby site, the Burnaby site outperformed the Florida sites in its pollution reduction. Unlike the Florida research, the Washington state report by Wang et al. (1982) yielded comparable metal removal efficiencies at the distance of 30 to 40 m, as can be seen in Table 4-23. But the data set used for their analysis was very small. Summary The Burnaby site grass drainage ditch study was effective in pollutant parameter reduction and the results were comparable to or better than those in other similar research studies. There was a statistically significant difference in pollutant removal efficiencies between the weir and 30 m-mark concentrations for all the parameters except for Mn. The non-winter pollutant removal efficiency values seem to be better than the corresponding winter values. This may have been due to lower rainfall amounts/intensities and their associated runoff volumes during the non-winter months which will create more opportunities for the sedimentation process (lower flow rates) in 195 the channel. For the rest of the year, the reasons for the effective pollutant removal range from well maintained grass conditions and their influence on flow rate to availability of pollutants, their basic chemistry and antecedent number of dry days. 4.5 Water Quality Assessment Over the years, the Canadian Council of Resource and Environment Ministers has developed a set of official Canadian Water Quality Guidelines to help assess the water quality of lakes and rivers, evaluate the effectiveness of clean up programs/operations and guide pollution control measures (Environment Canada, 1987/95). Most of the water quality parameters of interest in this research have guidelines set for drinking water purposes and freshwater aquatic life protection. 4.5.1 Water Quality Data Comparison Table 4-24 presents the mean water quality parameter concentrations observed in this study at both sites and their respective guidelines set for maximum acceptable concentration (MAC). These guidelines were set for health reasons, aesthetic objectives and freshwater aquatic life protection. Cr, Cd and Ni were found to be below the instrument's detection limits of 0.05, 5.00x10"3 and 0.04 mg/L, however, these are higher than the M A C values established for freshwater aquatic life protection. At the Burnaby site, both Cu and Zn initial concentrations and the concentrations after the highway runoff had gone through the 30 m long grass drainage ditch were below the drinking water M A C values of 1.00 and 5.00 mg/L respectively, however, they were higher than the M A C 196 Table 4-24. Average Water Quality Parameter Concentrations and Their Guidelines Burnaby Richmond Established Guidelines Parameter Initial Concentration Cone, at 30 m Initial Concentration Drinking Water (MAC) Freshwater Aquatic (MAC) Cr (total) — — — 0.05 0.002-0.02 Cd (total) — — — 0.005 0.002-0.0018 Ni (total) — — — 0.04 0.025-0.15 Cu 0.07 0.04 0.04 1 0.002-0.004 Fe (total) 5.55 2.12 3.37 0.30 0.30 Zn 0.25 0.12 0.16 5.00 0.03 Mn 0.13 0.11 0.12 0.05 N / A Pb (total) 0.12a ~ 0.48a 0.05 0.001-0.007 Ca 5.91 8.27 6.15 N / A N / A Oil/Grease 24.9 N / A 9.25 N / A N / A T.S.S. 155 50.6 74.8 N / A <10% Background Levels > 100 mg/L pH 6.82 6.56 6.49 6.5-8.5 6.5-9.0 EC (//S/cm) 190 281 125 N / A N / A Notes: 1) Initial concentration is taken right off the highway. 2) Concentration at 30 m is taken after going through grass ditch. 3) — Undetectable limits. 4) N / A - Not Available. 5) A l l concentrations in mg/L except for pH and EC. 6) a — Only two detectable concentrations were observed. 7) Average values are for composite samples. 197 values set for freshwater aquatic life protection. Initial concentrations of Fe, M n and Pb at both the Burnaby and Richmond sites, and the concentrations after the runoff had gone through the 30 m grass drainage ditch in length, were consistently higher than the established M A C values for drinking water and freshwater aquatic life protection, as shown in Table 4-24. Unlike the above metals, Ca and oil/grease do not have any set M A C values for either drinking water or freshwater aquatic life protection. The highway stormwater runoff pH values were within the pH ranges established for both drinking water and freshwater aquatic life, hence do not pose a threat to water quality. Finally, there are no guidelines recommended for T.S.S. concentration in drinking water, although turbidity, which can be used as a measure of T.S.S., is set at 5 N T U (nephelometric turbidity unit) for drinking water. Discussion According to Environment Canada (1987/95), the drinking water guidelines for Cu and Zn were set to avoid unpleasant tastes and laundry staining. In the freshwater environment, the guidelines were set for aquatic health reasons. Both ionic Cu and Zn have toxic effects on aquatic organisms including fish (Yousef et al., 1985a; Kushner, 1993). Therefore, there is reason to take further mitigative actions to reduce the concentrations of Cu and Zn in highway runoff at both sites before they enter a receiving water body (Onwumere, 1996a). Dilution from general urban runoff may further reduce Cu and Zn concentrations, however, more research is needed to determine the extent of the dilution effect, i f any. Fe and M n M A C values were established for aesthetic considerations while the established M A C value for Pb was for health reasons. Even though Fe and Mn are considered essential elements, their presence in water can lead to unpleasant taste, laundry staining and 198 undesirable growth, especially Fe at elevated concentrations (Environment Canada, 1987/95). Measures should be taken to further reduce the concentrations of Fe and M n in highway runoff before they enter receiving waters. Measures should also be taken to reduce Pb levels when detected in highway stormwater runoff. However, Pb levels at both sites were below the instrument's detection limit. Ca is a component of water hardness. According to Davies (1986), the presence of Ca and Mg ions results in hardness antagonism of metals uptake at the gill surface of the fish. Ca concentrations at both sites are in the excess of 5.00 mg/L, as shown in Table 4-24, and concentrations increased after going through the grass drainage ditch. The concentration is low enough not to require special treatment. Oil and grease, which do not have any guidelines, may require reduction. There was no maximum acceptable value (MAV) set for EC even though their values in highway runoff exceeded 100 yuS/cm at both sites. This is within the specific conductance ranges reported by Lawson et al. (1984) in Burnaby, BC and McKenzie and Irwin (1983) in Miami, Florida, and the EC range reported in U K by Colwill et al. (1984). The established pH ranges were set up to minimize corrosion, encrustation and toxic metals bioavailability in drinking water and for freshwater aquatic life protection (Environment Canada, 1987/95; Wang et al., 1982; Gupta et al., 1981b). According to the B.C. Research Corporation (1991), aquatic biota are known to be sensitive to extremes of pH. These pH values were within and comparable to the pH ranges of 5.6 - 7.5 and 6.2 - 8.7 reported in Vancouver and Burnaby residential sites by Swain (1983) and Lawson et al. (1985) respectively in the G.V.R.D. The pH values obtained at both sites are also within the pH ranges of 5.7 - 7.6 reported in U K (Colwill et al., 1984) and 6.4 - 8.1 reported in the USA (Gupta et al., 1981; Kobriger and Geinepolos, 1984). 199 Turbidity control is considered essential for both aesthetic and health reasons. Some of the health considerations include disinfection efficiency, biological nutrient availability, trihalomethane formation and concentrations of heavy metals and biocides. But for freshwater aquatic life, it was recommended that T.S.S. concentration be less than 10% for water with background T.S.S. levels of more than 100 mg/L (Environment Canada, 1987/95). The T.S.S. concentration after going through the 30 m grass drainage ditch was considerably reduced, as indicated in Table 4-24. Therefore, any impact T.S.S. concentration from highway runoff is going to have on freshwater aquatic life, especially fish, depends on the receiving water background T.S.S. concentrations. Summary Some of the parameters analysed such as Fe, Mn, Pb, and Zn exceeded the established M A C guidelines for either the drinking water or for freshwater aquatic life protection. Others, such as Cr, Cd and Ni were below detection limits; and for others, such as Ca and oil/grease, M A C guidelines have not been established. There may be further dilution from the general urban stormwater runoff below the measurement sites, i f urban runoff pollutant concentrations are lower, however the extent of this dilution is not known. Hence, further mitigation might be required for the highway runoff, before discharging to a nearby receiving water body. The impacts of these pollutants depend on a number of variables which range from their basic chemistry and bioavailability to their interaction in natural water environment. 4.5.2 Daphnia Magna Bioassay Data In order to investigate the impact of highway stormwater runoff on aquatic biota directly, a series of Daphnia Magna bioassays was conducted. Table 4-25 shows the percent survival of these 200 organisms and their corresponding 48h - LC50 (median lethal concentration) values for both winter and non-winter at the two sites. A l l the winter and Comp. "B" organisms' survival rate was 100% after 24 and 48 hours time lapses respectively. The Comp. "B" samples were taken after the highway runoff had gone through the 30 m long grass drainage ditch at the Burnaby site. The organisms' survival rate was less than 100%) for most of the non-winter Comp. "A" samples at the Burnaby site and for the only Comp. "A" sample at the Richmond site after 24 and 48 hour time lapses respectively. However, the 48h - LC50 values for all the composite samples were greater than 100 with the exception of the storm event of 07/02/96. This storm had the lowest organism survival rates of 20 and 0 after 24 and 48 hours respectively, generating a 48h - LC50 value of 60.3. This storm event may have been an isolated event since it was the only storm that generated a different 48h - LC50 value. Other non-winter storm events have the same 48h - LC50 values as the winter storms, however, they have organisms' percent survival rate ranges of 70 - 100 and 57 -90 after 24 and 48 hour time lapses respectively, as shown in Table 4-25. Discussion At these percentages of organisms' survival rates, it is fair to infer that there is some toxic effect from non-winter highway stormwater runoff. Lack of toxicity would have resulted in 100 percent organism survival rate, as was observed with the winter and all Comp. "B" samples. The 100 percent organisms' survival rate observed with Comp. "B" samples at the Burnaby site during both seasons is probably due to the ability of the 30 m grass drainage ditch to remove most of the pollutants, especially T.S.S. According to Portele et al. (1982), the toxicity of runoff from high volume highways can be reduced considerably by well-designed grass drainage ditches. 201 Table 4-25. Daphnia Bioassay Data at Burnaby and Richmond Sites Burnaby Richmond Date #of Dry Days Season % Survival Comp. A 48h-LC50 % Survival Comp. B 48h-LC50 % Survival Comp. A 48h-LC50 01/06/96 0 Winter (100,100) >100 (100,100) >100 (100,100) >100 02/07/96 1 Winter (100,100) >100 (100,100) >100 (100,100) >100 03/08/96 1 Winter (100,100) >100 (100,100) >100 (100,100) >100 06/10/96 8 Non-Winter (70,57) >100 N / A N / A N / A N/A 07/02/96 7 Non-Winter (20,0) 60.27 N / A N / A N / A N / A 07/17/96 14 Non-Winter (100,80) >100 (100,100) >100 (100,77) >100 08/02/96 13 Non-Winter (90,90) >100 (100,100) >100 N / A N / A Notes: 1) N / A - N o t Available 2) First number in the bracket is percent survival after 24 hours. 3) Second number in the bracket is percent survival after 48 hours. 4) Comp. A - is composite sample taken right off the highway. 5) Comp. B.- is composite sample taken after the runoff had gone through 30 m long grass drainage ditch. 202 They found that suspended solids are more harmful to small trout than dissolved metals at the concentrations measured in their study, and reported the same survival rate with undiluted highway runoff after grass filtration as the controls. The Burnaby site Comp. "B" samples with 100% survival rates for highway runoff (grass-filtered) and controls support the above-mentioned research data from Seattle, Washington. Similarly, the daphnia bioassay research data of Mar et al. (1981), also in Seattle Washington, reported comparable survival percentage rates with grass-filtered 1-5 highway runoff, as was observed at the Burnaby site. However, their unfiltered and undiluted highway runoff from the same site yielded much lower survival percentage rates than the Burnaby site. They reported organism survival rates that range from 0 - 30% at 1-5 with unfiltered highway runoff samples after 96 hours. Based on other similar analyses, they concluded that unfiltered highway runoff at 1-5 was toxic to daphnia in 48, 72 and 96 hr. bioassays. A similar conclusion cannot be made for the Burnaby site since the site had much higher survival percentage rates, with the exception of the storm event of 07/02/96, as shown in Table 4-25. At the Burnaby site, the much reduced organism survival rate of the storm event of 07/02/96 had Zn, Pb, Mn, Cu, Fe and Ca concentrations of 0.90, 0.09, 0.39, 0.19,4.7 and 27.0 mg/L respectively, compared to the high organism survival rate of the storm event of 07/17/96 which generated the corresponding concentrations of 0.36, 0.00, 0.16, 0.09, 4.98 and 9.90 mg/L. The storm event of 07/02/96, with a total rainfall of 0.85 mm, generated higher concentrations of metals than the storm of 07/17/96 with a total rainfall of 1.75 mm, with the exception of Fe. No Pb was generated by the storm event of 07/17/96, which may be the reason for the higher organism mortality associated with the storm of 07/02/96. Similarly, no Pb was observed in other storms listed in Table 4-25. However, Hall and Anderson (1988) reported in their urban runoff study in 203 Burnaby that Cu and Zn showed the highest correlation with toxicity. They also reported that Pb increased the toxicity of Cu and Zn. A l l these metals (Pb, Cu and Zn) were present in appreciable concentrations to cause the lower organism survival rates associated with the storm event of 07/02/96 at the Burnaby site. The number of dry days may not have been a factor. Most of the other daphnia bioassays conducted in the G.V.R.D. are with urban stormwater runoff and have variable results. Swain (1983), Lawson et al. (1985) and Anderson (1982) all reported some toxicity in the dry weather discharges with daphnia. For example, Swain (1983) reported his 96h - LC50 to be between 32 and 56%, thus suggesting that his Vancouver residential stormwater runoff may be acutely toxic to organisms in the lower food chain. Lawson et al.(1985), on the other hand, reported a 48h - LC50 of between 26 and 56% for their July 8, 1982 samples. They found the other four stormwater runoff samples in their Burnaby residential site to be non-toxic and did not report any 48h - LC50 for those samples. Hall and Anderson (1988) reported varying 96h - LC50 values for their laboratory bioassays with synthetic stormwater, with pH and different metal combinations having the most influence on toxicity. However, in their Burnaby site research, they found the urban stormwater runoff toxicity to daphnia to follow the sequence: commercial > industrial > residential > open space. Summary Impact analysis on receiving water bodies using daphnia bioassays showed both the winter and the grass filtered highway stormwater runoff to be non-toxic based on organism survivability. The winter organisms' survival rates were the same as the controls at both sites. Also, all the Comp "B" samples are non-toxic, thus resulting in the 100% organism survival rates observed at the Burnaby site. These results are comparable to those in other highway studies and further illustrates the importance of grass drainage ditches in pollution removal. A l l the non-winter Comp 204 "A" daphnia bioassays have some toxicity especially after 48 hours time lapse at both sites. The average survival rate of organisms was 70% after 24 hours compared to 57% after 48 hours at the Burnaby site. 4.6 Road Dirt, Soil Sediment and Grass Clipping Data Road dirt, soil sediment and grass clipping samples were collected at the two sites and analysed for Cr, Cd, N i , Cu, Zn, Pb, Fe, Mn and Ca. The results obtained are tabulated in Tables 4-26 through 4-34 respectively. 4.6.1 Highway Road Dirt Analysis Road Dirt Metal Concentrations There was a considerable amount of metals in road dirt samples analysed. At the Burnaby site, winter metal concentration values were higher than the non-winter values for all metals with the exception of Pb, Fe and Ca. However, in all cases, the differences between the winter and non-winter metal concentrations at the Burnaby site were not statistically significant at 95% confidence level. The F-ratio values ranged from 3.00x10"3 for Fe to 1.61 for Ca (DF 1, 9) which were far lower than the table F-ratio value of 5.12 with the same degrees of freedom. The corresponding P-values for Fe and Ca were 0.62 and 0.24 respectively. The sequences of metal concentration in road dirt samples at the Burnaby site in descending order were as follows: Winter: Fe > Ca > Zn > Mn > Pb > Cu > Cr > N i > Cd Non-winter: Fe > Ca > Pb > Zn > Mn > Cu > Cr > N i > Cd 205 Fe concentrations of over 16.5xl03 and 17.0xl03 mg/kg for both winter and non-winter respectively were the highest of all the metals at the Burnaby site, as shown in Table 4-26. Cd concentrations were the lowest for both seasons. The other metals almost follow the same pattern for both seasons with the exception of Pb, Mn and Zn which changed concentration patterns between the two seasons. This observed road dirt metal concentration pattern is similar to the metal concentration patterns observed earlier with highway stormwater runoff. Table 4-27 shows statistical analyses including the 95% confidence intervals (CI) for all the metal concentrations and loadings. For example, Fe has a mean concentration of 16.8xl03 mg/kg with lower and upper 95% CI values of 6.62xl0 3 and 27.0xl0 3 mg/kg respectively compared to Cr with a mean concentration of 31.1 mg/kg and lower and upper 95% CI values of 7.34 and 54.8 mg/kg respectively. At the Richmond site, the non-winter road dirt metal concentrations were consistently higher than the winter values and, with the exception of Zn, these differences were statistically significant at 1 and 9 degrees of freedom. An F-ratio of 4.02 with a P-value of 0.08 was obtained for Zn which was below the table F-ratio of 5.12. F-ratio values for other metals ranged from 7.30 (P=0.02) for Cr to 45.3 (P=0.00) for Cu. But like the Burnaby site, statistically significant amounts of metals were found in road dirt samples for both seasons, as indicated in Table 4-26. The descending order of metal concentration patterns for both seasons at the site is as follows: Winter: Fe > Ca > Zn > Mn > Cu > Pb > Cr > N i > Cd Non-winter: Fe > Ca > Zn > Mn > Cu > Pb > Cr > N i > Cd Unlike the Burnaby site, the order of metal concentration patterns generated at the Richmond site is the same for both seasons. Table 4-28 shows the rest of the metal concentration statistical data at the Richmond site including the 95% confidence intervals and Figure 4-21 shows 206 Table 4-26. Road Dirt Seasonal Metal Concentrations for Burnaby and Richmond Sites (mg / kg) Burnaby Richmond Metals Winter Non-Winter Winter Non-Winter Cr 34.5 28.2 13.2 32.7 Cd 1.14 0.90 0.52 2.20 Ni 33.6 22.8 12.4 27.0 Cu 246 218 52.2 147 Zn (xlO2) 3.42 3.18 2.62 11.4 Pb 324 274 46.8 143 Fe (xlO3) 16.5 17.0 9.25 22.0 M n 324 278 105 353 Ca (xlO3) 8.95 4.30 3.55 5.61 Notes: 1) Values are mean concentrations. Table 4-27. Burnaby Site Road Dirt Statistics Parameter N Min. Max. Mean STD.DEV 95% CI Lower 95% CI Upper Annual Load Cr 11 1.30 130 31.1 35.3 7.34 54.8 1.00 Cd 11 0.00 4.00 1.02 1.16 0.03 1.59 0.03 Ni 11 7.20 122 27.7 32.9 5.61 49.8 1.00 Cu 11 95.0 562 231 146 133 329 10.0 Zn 11 113 1.03xl03 329 250 161 497 14.0 Pb 11 0.00 814 296 233 140 453 13.0 Fe (xlO3) 11 .09 56.9 16.8 15.1 6.62 27.0 0.73 M n 11 147 955 299 229 146 453 13.0 Ca (xlO3) 11 2.48 23.4 6.41 6.24 2.22 10.6 0.28 Note: 1) A l l concentrations in mg/kg except for loading. 2) Loading is in kg/km/yr. 3) CI—Confidence interval. 4) Highway length: Burnaby= 0.16 km 207 Table 4-28. Richmond Site Road Dirt Statistics Parameter N M i n . Max. Mean STD.DEV 95% CI Lower 95% CI Upper Annual Load Cr 11 0.00 51.0 23.8 15.2 13.6 34.0 0.07 Cd 11 0.30 3.00 1.42 1.02 0.73 2.10 0.01 N i 11 3.60 36.0 20.4 10.2 13.5 27.2 0.06 Cu 11 29.0 169 104 54.2 67.5 140 0.30 Zn (xlO2) 11 0.91 23.7 7.40 8.23 1.87 12.9 0.02 Pb 11 25.5 243 99.4 70.0 52.3 146 0.30 Fe (xlO2) 11 0.60 313 162 92.0 100 224 0.51 Mn 11 44.0 464 240 152 138 342 0.80 Ca (xl0 2 ) 11 25.4 66.8 46.7 14.9 36.7 56.7 0.15 Note: 1) A l l concentrations in mg/kg except for loading. 2) Loading is in kg/km/yr. 3) CI—Confidence interval. 4) Highway length: Richmond= 0.09 km 208 100000 Cr Cd Ni Cu Zn Pb Fe Mn Ca Types of metal • M E A N B U R N A B Y R O A D DIRT • B U R N A B Y WINTER ROAD DIRT • B U R N A B Y NON-WINTER ROAD DIRT • M E A N RICHMOND ROAD DIRT • RICHMOND WINTER ROAD DIRT • RICHMOND NON-WINTER ROAD DIRT Figure 4-21. Road Dirt Metal Concentrations at Both Sites. 209 the road dirt metal concentration patterns at both sites. For example, the metals with the highest and lowest concentrations are Fe and Cd. The mean, lower and upper 95% confidence intervals for Fe are 162xl0 2, lOOxlO2 and 224xl0 2 mg/kg and for Cd are 1.42, 0.73 and 2.10 mg/kg. Between the two sites, the order of metals generated in road dirt is similar with Fe, Ca and Zn having higher concentrations and Cr, N i and Cd having the lowest concentrations, as displayed in Figure 4-21. Road dirt samples from the Burnaby site had higher Cr, N i , Cu, Pb, Mn, Fe and Ca concentrations, as indicated in Figure 4-21. The concentrations of Cd and Zn were higher at the Richmond site. Although the Burnaby site generally had higher metal concentrations than the Richmond site, the differences were not statistically significant, except for Cu and Pb. Both Cu and Pb have F-ratio values of 7.33 (P=0.01) and 7.19 (P=0.01) respectively with 1 and 20 degrees of freedom (DF). These F-ratio values were higher than the table F-ratio value of 4.35. A l l the other metals have F-ratio values that ranged from 0.01 (P=0.91) for Fe to 2.51 (P=0.13) for Zn (DF 1 and 20) at 95% confidence level. l-Road Dirt Metal Concentration Discussion The higher concentration of most of these metals at the Burnaby site is similar to the trend observed with the highway stormwater runoff. The same influencing factors such as ADT, dustfall, surrounding land use and highway age that affect highway stormwater runoff probably also apply to road dirt samples. The high concentration of metals in road dirt samples at both sites support the data of Sartor et al. (1974), Gupta et al. (1981/1981b), Wang et al. (1982) and Kobriger and Geinepolos (1984) who reported high metal association with dirt and dust fractions from both street and highway surfaces. In fact, Wang et al. (1982) reported that more Pb per unit 210 weight was found in smaller size particles than in the larger fractions in their research in Washington State. These larger fractions are more easily removed than the finer pollutant-laden particles. Asplund et al. (1982) estimated the pollutant removal range of street cleaning operations to be between 25 to 78 percent of the highway pollutant mass. But since the particles removed are generally the larger fractions, and since most of the metal pollutants are contained within the fine particles (U.S. EPA, 1983; Shaheen, 1975; Howell, 1978; Wilbur and Hunter, 1979), one would expect street sweeping to be relatively ineffective in removing the metals. This supports the earlier work of Sartor et al. (1974). Nevertheless, these fine particles can be removed from travelled lanes by saltation (injection of fine-sized particles into the atmosphere by vehicle-generated turbulence). For example, Kerri et al. (1985) reported that the low pollutant loading in the Redondo Beach, Walnut Creek and Sacramento, California highways study was due to traffic generated turbulence which continuously sweeps the travelled lanes and highway shoulders. However, in the present study, saltation was probably not a major factor considering the concentrations of metal pollutants in the road dirt at both sites. This is because of traffic congestions during the rush hours especially at the Burnaby site. Both Scanlon (1977) and Kobriger and Geinepolos (1984) reported high levels of other metals in road dirt and dust fractions. Based on these reported research studies and the high metal levels in road dirt samples at both the Burnaby and Richmond sites, it is fair to assume that road dirt is a significant source of metal pollution in highway stormwater runoff. Road Dirt Metal Loadings The Burnaby site generated a greater total loading of metals in kg/km/yr than the Richmond site. The Burnaby site metal loadings range from a low of 0.03 kg/km/yr for Cd to a 211 high of 726 kg/km/yr for Fe, as presented in Table 2-27 above. The descending order of metal loadings at the Burnaby site is: Fe > Ca > Zn > Mn/Pb > Cu > Ni/Cr > Cd At the Richmond site, the metal loadings range from 0.01 kg/km/yr for Cd to 51.0 kg/km/yr for Fe, as indicated in Table 4-28 above. However, the descending order of the metal loadings at the Richmond site is slightly different from the Burnaby site and is as follows: Fe > Ca > Zn > Mn > Pb/Cu > Cr > N i > Cd Between the two sites, the Burnaby site consistently generated higher metal loadings than the Richmond site. Road Dirt Metal Loading Discussion The reasons for the higher loading of these metals at the Burnaby site are the same as given previously for the higher concentrations. Similarly, the same influential environmental factors that affect highway stormwater runoff contaminants probably also apply to the road dirt samples at both sites. The Burnaby Pb and Zn loading results of 13.0 and 14.0 kg/km/yr were higher than 4.00 and 8.00 kg/km/yr respectively reported by Colwill et al. (1984) in U.K. But their U.K. road results were higher than the results obtained from the Richmond site for both Pb and Zn. Sartor et al. (1974) reported Zn, Cu, Pb, N i and Cr loadings of 0.18, 0.05, 0.16, 0.01 and 0.03 kg/km respectively in their analyses across the U.S.A. But these were one-time or momentary sample analysis and they did not collect enough data at any one particular site to forecast annual loads. The descending order of Sartor et al. (1974) metal loading is: Zn > Pb > Cu > Cr > N i 212 The observed descending order metal loading pattern in this research supports the earlier result of Sartor et al. (1974) across the U.S.A. The above observations suggest that significant amounts of heavy metals are present in road dirt on street/highway surfaces. Fe, Ca, Zn, M n and Pb were the most prevalent in this research and in both the works of Colwill et al. (1984) in U.K. and Sartor et al. (1974) across the U.S.A. But the lower Pb levels observed in Canada than the U.S.A. are due to the banning of leaded gasoline in Canada several years ago. In B.C., McCallum (1995) found a statistically significant difference in the street sediment across the Burnaby Brunette River Watershed basin. He reported a median of percentage differences between 1973 and 1993 at each station for Pb, Cu and M n to be -67, -43 and 43% respectively. He found no significant changes in the street sediment concentrations of Zn, Ni , Fe and Hg between 1973 and 1993. Since there were no road dirt results available along the two highways from past research, this study cannot validate or void the research observations made by McCallum (1995). However, this research generally supports McCallum's (1995) suggestion that street generated contaminants are at least partially responsible for stream sediments enrichment by Pb, Cu, N i and Cr. Summary Road dirt analyses showed winter Cr, Cd, Ni , Cu, Zn and M n concentrations at the Burnaby site to be higher than the non-winter values. Pb, Fe and Ca values were higher during the non-winter months. However, the differences in road dirt metal concentrations between winter and non-winter were not statistically significant at 95% confidence level for any of the metals at the Burnaby site. A l l the road dirt non-winter metal concentrations were higher than the winter concentrations at the Richmond site and the concentration differences were significant at 95% confidence level. Between the two sites, road dirt samples from the Bumaby site had higher metal 213 concentrations than the Richmond site with the exception of Ca. The higher road dirt Ca concentration at the Richmond site can be attributed to the proximity of a cement production plant to the site. The reasons for higher road dirt metal concentrations/loadings at the Burnaby site are the same as those presented in section 4.2.2 for metal contaminants in the highway runoff. At both sites, Fe and Cd were the most and the least generated pollutant loadings respectively. 4.6.2 Soil Sediment Data Analysis At the Burnaby site, sediment samples designated " A " were collected near the weir and samples designated "B" were taken from 30-32 m along the grass drainage ditch. The samples were analysed for metal content. The non-winter sediment metal concentration values for " A " and " B " were higher than their corresponding winter metal concentration values. For the winter and non-winter " A " and " B " soil sediment samples, higher concentrations of metals were obtained with the soil sediment " B " samples, as can be seen from Table 4-29. These differences between sediment " A " and " B " metal concentration samples are statistically significant for all the metals. The 95% confidence level F-ratio values for Cr, Cd, N i , Cu, Zn, Pb, Fe, Mn and Ca were 7.08 (P=0.02), 27.0 (P=0.00), 19.0 (P=0.00), 32.6 (P=0.00), 14.9 (P=0.00), 32.2 (P=0.00), 6.57 (P=0.02), 8.41 (P=0.01) and 10.2 (P=0.01) respectively with 1 and 20 degrees of freedom. This increase in sediment " B " metal concentrations is also evident in Figure 4-22. The descending order of soil sediment metal concentrations for both seasons is: Winter A : Fe > Ca > Zn > Mn > Cu > Pb > Cr > N i > Cd Non-winter A : Fe > Ca > Zn > Mn > Pb > Cu > Cr > N i > Cd Winter B : Fe > Ca > Pb > Mn >Zn > Cu > Cr > N i > Cd Non-winter B: Fe > Ca > Zn > Mn > Pb > Cu > Cr > N i > Cd 214 Table 4-29. Seasonal Soil Sediment Metal Concentrations for Burnaby and Richmond Sites (mg/kg) Burnaby Richmond Metals Winter"A" Non-Winter"A" Winter"B" Non-Winter"B" Winter Non-Winter Cr 19.2 29.0 31.5 61.0 27.7 40.0 Cd 1.48 2.00 3.64 7.00 1.68 6.00 Ni 13.1 20.0 28.4 39.0 24.0 22.0 Cu 93.8 179 323 407 120 133 Zn (xlO2) 2.78 3.67 4.41 7.06 5.10 11.2 Pb 93.3 248 562 502 142 177 Fe(xl0 3 ) 8.45 15.1 14.0 30.9 12.8 18.0 Mn 155 255 482 506 241 332 Ca (xlO3) 2.72 3.95 1.72. 2.24 1.82 2.99 Note: 1) "A"—Means sample taken near the weir ( distance of 0-2 m) 2) " B " ~ Means sample taken at 30-32 m along the grass drainage ditch. 215 For sediment " A " samples, there were no statistically significant differences in metal concentrations between the two seasons for four out of the nine metals. The exceptions are Ni , Cu, Pb and Fe at the Burnaby site. An A N O V A performed at 95% confidence level showed Ni , Cu, Pb and Fe concentrations to differ significantly between winter and non-winter for sediment " A " Burnaby samples. The F-ratio values ranged from 6.13 (P=0.04) for N i to 24.0 (P=1.00xl0'3) for Pb (DF 1 and 9). For sediment " B " Burnaby site samples, there were also no statistically significant differences in metal concentrations between winter and non-winter for most metals except for Cd, Zn and Fe and their F-ratio values were found to be 8.73 (P=0.02), 8.15 (P=0.02) and 6.68 (P=0.03) respectively with 1 and 9 degrees of freedom performed at 95% confidence level. Like the road dirt, the three main metals with higher concentrations are Fe, Ca and Zn and the ones with lower concentrations are Cr, N i and Cd. Other statistical analyses including the 95% confidence intervals are listed in Table 4-30. Higher concentrations of metals were also observed during non-winter at the Richmond site except for Ni , as indicated in Table 4-29. However, the differences in sediment metal concentrations between winter and non-winter at this site were not statistically significant for any of the metals. A l l the metals have F-ratio values lower than the table F-ratio value of 5.12 with 1 and 9 degrees of freedom performed at 95% confidence level. No grass drainage ditch study was performed at this site due to the structure of ditch design. The site also generated the highest Fe concentration for both seasons. In fact, the Fe mean concentration at the Richmond site was higher than the Burnaby sediment " A " but lower than the corresponding sediment " B " mean concentration, as displayed in Figure 4-22. The other statistical data is shown in Table 4-31. The descending order of sediment metal concentration at the Richmond site is: Winter: Fe > Ca > Zn > M n > Pb > Cu > Cr > N i > Cd Non-winter: Fe > Ca > Zn > Mn > Pb > Cu > Cr > N i > Cd 216 100000 Cr Cd Ni Cu Zn Pb Fe Mn Ca Typei of metal MEAN BURNABY SEDIMENT A BURNABY WINTER SEDIMENT A BURNABY NON-WDMTER SEDINENT A MEAN BURNABY SEDIMENT B BURNABY WINTER SEDIMENT B BURNABY NON-WINTER SEDIMENT B RICHMOND SEDIMENT RICHMOND WINTER SEDIMENT RICHMOND NON-WINTER SEDIMENT Figure 4-22. Soil Sediment Metal Concentrations at Both Sites. 217 Table 4-30. Burnaby Site Soil Sediment Statistics ("A" Samples Only) Parameter N Min. Max. Mean STD.DEV 95% CI Lower 95% CI Upper Cr 11 1.80 37.0 24.4 10.6 17.3 31.5 Cd O i l 0.00 3.00 1.60 0.87 1.02 2.18 N i 11 4.50 22.0 16.5 5.30 13.0 20.1 Cu 11 54.0 226 140 59.0 101 180 Zn 11 167 465 326 96.3 261 391 Pb 11 50.0 339 177 89.4 113 233 Fe (xlO3) 11 0.75 17.4 12.1 5.08 8.68 15.5 Mn 11 67.0 334 214 79.5 161 268 Ca (xlO3) 11 1.66 4.85 3.39 1.14 2.63 4.16 Note: 1)A1 2) CI concentrations in mg/kg. --Confidence interval. Table 4-31. Richmond Site Soil Sediment Statistics Parameter N Min. Max. Mean STD.DEV 95% CI Lower 95% CI Upper Cr 11 18.0 66.0 34.3 15.1 24.2 44.5 Cd 11 1.00 21.0 3.76 5.84 0 7.69 Ni 11 2.00 37.5 22.9 9.80 16.4 29.5 Cu 11 87.0 205 127 37.2 102 152 Zn (xlO2) 11 3.21 18.7 . 8.41 5.32 4.83 12.0 Pb 11 95.9 281 161 53.3 125 197 Fe (xlO3) 11 0.16 23.6 15.6 5.84 11.7 19.6 Mn 11 31.0 497 291 11.9 215 366 Ca(xl0 2 ) 11 4.11 49.2 24.5 13.4 15.5 33.6 Note: 1)A1 2) CI concentrations in mg/kg. -Confidence interval. 218 The sediment metal concentration patterns generated at the Richmond site are the same for both seasons. Like the road dirt metal concentrations, the three metals with the highest concentrations are Fe, Ca and M n and the three with the least concentrations are Cr, N i and Cd. Other statistical data is shown in Table 4-31 including sample number and 95% confidence intervals. Between the two sites, the Richmond site has higher soil sediment metal concentrations than the Burnaby site during the winter months with the exception of Ca. During non-winter, the Richmond site also has higher soil sediment metal concentration for most metals except for Cu, Pb and Ca. Although there were differences in sediment metal concentrations between Burnaby and Richmond sites, these differences were not statistically significant except for Zn, with an F-ratio of 9.95 (P=0.01) and an revalue of 0.33 at 1 and 20 degrees of freedom. The rest of the metals have F-ratio values below the table F-ratio value of 4.35. Discussion Higher sediment "B" metal concentrations were observed at the Burnaby site than the corresponding sediment "A" values. It could be due, in part, to flow rate. The flow near the weir, where soil sediments " A " were collected, was high and the relatively high velocities may have allowed only the coarser sediment to settle, with the finer sediments passing on downstream. Since, as discussed in section 4.6.1, metals tend to adhere more to finer sediments, the pollutant-laden fine particles are transported down the grass drainage ditch. At the 30-32 m length segment, where the sediment " B " samples were collected, the flow has decreased considerably allowing sedimentation of pollutant-laden fine particles to occur, resulting in the much higher metal concentrations observed with the soil sediment " B " samples for both seasons. 219 The higher concentration of metals in Richmond site soil sediment samples probably has more to do with the local soil metal concentrations and surrounding land use than ADT, dustfall and highway age. Recall from the previous section that the Burnaby site has consistently high metal concentrations in the road dirt samples collected. One would expect that accumulation of these road dirt samples with higher metal concentrations over the years would result in high soil sediment metal concentrations at the site. Instead, the much younger Richmond site has higher soil sediment metal concentrations and lower road dirt metal concentrations than the Burnaby site. Also, i f these metals were from any other sources, they should also have affected the road dirt and the highway runoff and not just the soil sediment samples. A similar observation with metals was made by Kobriger and Geinepolos (1984) in 1-81 in Harrisburg, Pennsylvania where they attributed the unusually higher Fe concentrations observed in their study to the high Fe content in the local soils. In other soil studies, Motto et al. (1970), Zimdahl (1972), Getz et al. (1975), Scanlon (1977), and Gupta and Kobriger (1980) found the highest metal concentrations in the soils immediately adjacent to the highways. They found the concentrations of Pb as well as Cd, Ni , and Zn to decrease rapidly with depth and distance from the highways. This decrease in metal concentrations with distance is perpendicular to the highway. Scanlon (1977) found that metal concentrations decreased to background levels within 48 m from the Virginia highways, compared to 50 m from Illinois highways (Getz et al., 1970), 30 m from Denver highways (Zimdahl, 1972), and between 30 and 35 m from Milwaukee highways (Gupta and Kobriger, 1980). This is automobile generated heavy metal pollution of local soil conditions as opposed to land use generated heavy metal pollution, as indicated at the Harrisburg site. These automobile generated heavy metals are most likely to add to the metal background levels of the surrounding land use, 220 thereby increasing the heavy metal levels on land adjacent to the highways. Some of these metals can be immobilized and confined to a certain soil layer. Generally, Pb in the soil is effectively immobilized and confined to the top 15 cm depending on the distance from the roadway according to Laxen and Harrison (1977); the top 15.2 cm at 7.6 m away from New Jersey highways, according to Motto et al. (1970); and the top 10 cm according to Getz et al. (1975). The immobilization of Pb by the soil, according to Zimdahl (1972) and Hassett (1974), depends directly on the soil cation exchange capacity (CEC) and it is inversely related to soil pH. As a result, there is a tendency for heavy metals, especially Pb, to accumulate on lands adjacent to highways. In Burnaby, B.C., McCallum (1995) found a decrease in Pb levels in stream, lake and street sediments in the Brunette River Watershed, over the last 20 years, due to the banning of leaded gasoline. On the other hand, he found Zn, Cu, Mn and Hg levels in stream sediments, during the same time period, to have increased by 45, 81, 130, and 290 percent respectively. He attributed these increases to urbanization and automobiles within the watershed. The data from this study tend to support the findings of these other research studies with the comparable soil sediment data obtained. Summary Soil sediment analyses at the Burnaby site indicated higher metal concentrations during the non-winter than the winter period. However, there were no significant differences between winter and non-winter metal concentrations at the Burnaby site for soil sediment " A " samples with the exception of N i , Cu, Pb, and Fe. Similarly, the only significant differences in seasonal metal concentrations for soil sediment " B " samples were observed with Cd, Zn, and Fe. The most interesting and unexpected finding was that the soil sediment " B " samples, collected at 30 - 32 m away from the weir, have considerably higher metal concentrations than soil sediment " A " 221 samples collected near the weir. This is believed to be due to lower flow rate at the 30-32 m mark resulting in greater sedimentation of pollutant-laden fine particles. The lower soil sediment " A " metal concentrations is probably due to the settling of coarser sediment particles resulting from the higher flow rates near the weir. The differences in metal concentrations between soil sediment " A " and " B " samples were statistically significant at 95% confidence level. The Richmond site also has higher metal concentrations during the non-winter than the winter months except for N i . Between the two sites, the Richmond site soil sediments have higher overall metal concentrations than the Burnaby site, although the differences were generally not statistically significant. This is probably due to higher metal concentrations in the local soils rather than other influencing environmental factors. 4.6.3 Grass Clipping Data Analysis Grass clipping " A " and " B " samples were collected near the weir and 30-32 m along the grass drainage ditch respectively. The results of the metals analysis are tabulated in Table 4-32. The grass clipping Cr, Cd and N i concentrations were below the instrument's detection limits of 0.5, 1.0 and 1.0 mg/kg most of the time. The grass clipping metal concentrations had winter values higher than the non-winter values for both samples " A " and " B " at the Burnaby site. For grass samples " A " , the differences in metal concentrations between winter and non-winter at the Burnaby site were statistically significant for all of the metals except for Cd and Ni . The rest of the metals have F-ratios that ranged from 6.42 (P=0.03) for M n to 24.0 (P=1.00xl03) for Cu performed at 95% confidence level with 1 and 9 degrees of freedom. These F-ratio values were 222 higher than the table F-ratio value of 5.12. Like the grass samples " A " , the grass sample " B " metal concentrations showed significant differences between winter and non-winter concentrations with the exceptions of N i , Fe, and Ca. N i , Fe, and Ca have F-ratio values of 4.92 (P=0.05), 3.23 (P=0.11), and 2.46 (P=0.15) respectively. The rest of the metals have F-ratio values that ranged from 5.41 (P=0.05) for Cu to 12.3 (P=0.01) for Cd performed at 95% confidence level with 1 and 9 degrees of freedom. Other metal statistics for grass " A " and other samples with their confidence intervals are presented in Tables 4-33 and 4-34. The descending order of metal concentration patterns at the Burnaby site is: Winter A : Ca > Fe > Zn > Mn > Cu > Pb > Cr > N i > Cd Non-winter A: Ca > Fe > Zn > Mn > Cu > Pb > Cr > N i > Cd Winter B: Fe > Ca > Mn > Zn > Cu > Pb > N i > Cr > Cd Non-winter B: Ca > Fe > Mn > Zn > Cu > Pb > N i > Cr > Cd The winter and non-winter grass clipping " A " metal concentration patterns are similar, but the winter and non-winter grass " B " samples have different concentration patterns. The three elements with higher concentrations in grass clipping " A " samples are Ca, Fe and Zn compared to Ca, Fe and M n in grass clipping " B " samples. As usual, the concentrations of Cr, N i and Cd were the least. Both winter and non-winter grass clipping " B " metal concentrations were higher than their corresponding grass clipping " A " values most of the time. Between grass clippings " A " and " B " , grass " B " samples have consistently high metal concentrations, except for Ca, as shown in Figure 4-23 for the Burnaby site. However, a significance test performed at 95% confidence level, with 1 and 20 degrees of freedom, confirmed significant differences only for Cd, Mn, and Ca, with F-ratio values of 6.92 (P=0.02), 11.9 (P=3.00xlO'3), and 9.60 (P=0.01) respectively, compared to a table F-ratio value of 4.35. 223 Table 4-32. Seasonal Grass Clipping Metal Concentrations for Burnaby and Richmond Sites (mg/kg) Burnaby Richmond Metals Winter"A" Non-Winter"A" Winter"B" Non-Winter"B" Winter Non-Winter Cr 8.86 2.17 12.8 4.00 16.8 2.00 Cd 0.72 0.33 3.10 1.00 2.30 0.83 . N i 8.12 4.5 10.9 5.00 13.3 5.5 Cu 43.4 17.0 61.9 14.0 67.6 19.0 Zn 169 92.0 274 101 448 200 Pb 30.1 4.67 47.5 13.0 47.3 9.5 Fe (xlO2) 27.4 6.25 44.7 12.6 51.0 10.2 M n 167 82.0 468 203 247 157 Ca(xl0 3 ) 3.96 6.18 3.26 3.89 5.03 5.77 Note: 1) "A"~Means sample taken near the weir ( distance of 0-2 m) 2) " B " ~ Means sample taken at 30-32 m along the grass drainage ditch. 224 Table 4-33. Burnaby Site Grass Statistics ("A" Samples Only). Parameter N Min. Max. Mean Std. 95% CI Dev. Lower Upper Cr 11 0.00 13.2 5.21 5.21 1.71 8.71 Cd 11 0.00 2.00 0.51 0.70 0.04 0.98 N i 11 0.00 14.8 6.15 3.61 3.72 8.57 Cu 11 12.0 54.0 28.7 16.3 17.8 39.6 Zn 11 60.0 201 127 48.5 94.0 159 Pb 11 0.00 45.0 16.2 17.4 4.50 27.9 Fe(xl0 2 ) 11 3.15 42.8 15.9 14.4 6.17 25.6 Mn 11 43.0 250 120 69.2 73.9 167 Ca (xlO2) 11 28.6 72.7 51.7 15.2 41.5 61.9 Note: 1) A l l concentrations in (mg/kg) 2) CI ~ confidence interval 225 The rest of the metals did not show any statistically significant difference between grass clipping " A " and " B " samples. They have F-ratio values that were below the table F-ratio value of 4.35. Like the Bumaby site, the winter grass clipping metal concentrations at the Richmond site were higher than the non-winter values with the exception of Ca, as presented in Table 4-32. The differences in grass clipping metal concentrations between winter and non-winter at the Richmond site are statistically significant for most of the metals at 95% confidence level, with 1 and 9 degrees of freedom. The only exception was Ca with an F-ratio value of 5.12. The rest of the grass clipping metal samples that show significant differences between winter and non-winter metal concentrations have F-ratio values that range from 5.54 (P=0.04) for Pb to 11.6 (P=0.01) for Ni , with 1 and 9 degrees of freedom. The other Richmond site grass clipping metal statistics are noted in Table 4-34. The Richmond site grass clipping samples have consistently higher metal concentrations than the Bumaby site grass clipping " A " samples, as noted in Figure 4-23. However, an A N O V A performed between the two sites at 95% confidence level, with 1 and 20 degrees of freedom, shows statistically significant differences only for Cd, Zn and Mn, with F-ratio values of 6.35 (P=0.02), 10.1 (P=0.01) and 6.64 (P=0.02) respectively. The rest of the metals showed no significant differences in their concentrations and have F-ratio values that are below the table F-ratio value of 4.35. These differences or lack of it in metal concentrations between the two sites are also evident in Figure 4-23. 226 Cr Cd Ni Cu Zn Pb Fe Mn Ca Type» of metal. MEAN BURNABY GRASS A BURNABY WINTER GRASS A BURNABY NON-WINTER GRASS A • MEAN BURNABY GRASS B BURNABY WINTER GRASS B BURNABY NON-WINTER GRASS B MEAN RICHMOND GRASS • RICHMOND WINTER GRASS RICHMOND NON-WINTER GRASS Figure 4-23. Grass Clipping Metal Concentrations at Both Sites. 227 Table 4-34. Richmond Site Grass Statistics. Parameter N Min. Max. Mean Std. 95% CI Dev. Lower Upper Cr 11 0.00 37.0 8.72 11.5 1.01 16.4 Cd 11 0.00 3.70 1.50 1.10 0.76 2.24 N i 11 3.00 20.2 9.06 5.44 5.40 12.7 Cu 11 12.0 143 41.2 38.9 15.1 67.3 Zn 11 138 704 313 188 186 439 Pb 11 0.00 109 26.7 32.0 5.20 48.2 Fe (xlO2) 11 5.48 11.3 28.7 32.2 7.12 50.4 Mn 11 124 340 198 72.5 150 247 Ca (xlO2) 11 38.8 69.8 54.3 8.93 48.4 60.3 Note: 1) A l l concentrations in (mg/kg) 2) CI ~ confidence interval v 228 Discussion At the Richmond site, samples were taken only near the weir. At the Burnaby site, samples were taken at two locations and as was the case with the sediments, metal concentrations in the grass clippings were higher with the "B" samples than the " A " ones. The explanation is probably the same as given for the sediment, namely that only the coarser sediments would be deposited upstream near the weir, because of the high flow velocities. At the 30-32 m mark, where flows and velocities were lower, only the finer sediments are deposited. Metals adhere more to fine sediment particles than to coarse ones, thus offering an explanation for the higher grass clipping "B" metal concentrations. For example, during the winter the mean grass clipping " A " Fe, Zn, and Mn concentrations (near the weir) were found to be 27.4x102, 169 and 167 mg/kg respectively . compared to corresponding grass clipping " B " (at 30-32 m) values of 44.7x102, 274 and 468 mg/kg for the sample time period. Between the two seasons, the grass clipping samples have higher metal concentrations during the winter than the non-winter period at both sites, This is probably due to reduce grass activities caused by less favourable environmental growing conditions. This would probably lead to lower ability for the grasses to process metals, thereby leading to metal accumulation in grass tissues during the winter months. More Ca in the non-winter soil sediment as observed in the previous section may have resulted in more Ca plant uptake observed with the grass clipping analysis in this section. Both sedimentation and plant metal uptake are two processes that are hypothesized to have impacts on grass drainage ditch metal removal effectiveness at the Burnaby site. The sedimentation process will take the metals out from the flowing water while the plant uptake will remove the metals from 229 the soils, thus leading to the higher grass clipping "B" metal concentrations as was observed at the Burnaby site. This tends to support the research observations of Wang et al. (1982) in Washington State and Yousef et al. (1985) in Florida in which they suggested that both sedimentation and plant metal uptakes play key roles in stormwater runoff metal pollution reduction. Although other researchers did not analyse grass clipping samples and hypothesized on the effect of sedimentation and plant uptake, the grass clipping samples analysed in this study supported their hypothesis. Summary The grass clipping samples have higher metal concentrations in the winter than during the non-winter at both sites. Grass clipping "A"samples showed seasonal significant differences in their concentrations except for Cd and Ni whereas grass clipping " B " samples have significant differences in most of the metal concentrations except for N i , Fe, and Ca. At the Burnaby site, the grass clipping " B " samples for both winter and non-winter were consistently higher than the corresponding grass clipping " A " values and the differences were significant for most of the metals except for Cd, Mn, and Ca. This indicates that more metals are picked up through sedimentation and plant uptake processes at the 30 - 32 m mark, along the grass drainage ditch, than near the weir. This supports the findings of other researchers. At the Richmond site, the seasonal differences in grass metal concentrations were significant except for Cd. Between the two sites, higher metal concentrations were obtained in grass clipping samples at the Richmond site than the Burnaby site, although the only significant differences in grass metal concentrations between the two sites were with Cd, Zn, and Mn. 230 4.7 Regression and Correlation Analyses 4.7.1. Pearson Correlation Matrix Between T.S.S., Metals and Oil/Grease Pearson correlation matrix analysis was performed between T.S.S., metals and oil/grease concentrations in the flow composite samples at both sites. Tables 4-35 and 4-36 show the results of the analyses with the corresponding P-values. At the Burnaby site, the only significant correlation with T.S.S. was found with oil/grease with a correlation coefficient of 0.76 (P=0.03). Between metals, there were significant correlations between Cu-Fe, Cu-Mn, Cu-Ca, Fe-Mn, Fe-Zn, Mn-Zn and Mn-Ca with correlation coefficients that ranged from 0.68 (P=0.07) for Cu-Fe to 0.96 (P=2.00xl0"4) for Cu-Zn. No other correlations were significant. At the Richmond site, the T.S.S. was significantly correlated with Cu and Fe, both with correlation coefficients of 0.74 (P=0.10 and 0.09 respectively). There were significant correlations between Cu-Fe, Cu-Zn, Cu-OG, Fe-Zn, Fe-OG, Mn-Zn, Mn-Ca, and Zn-OG with correlation coefficient values that range from 0.75 (P=0.09) for Mn-Ca to 0.98 (P=1.00xl0'3) for Cu-Fe, as noted in Table 4-36. The rest of the relationships have coefficient values of less than 0.70. Discussion Highway stormwater runoff solids chemistry is very complicated since the material is an aggregation of pollutants and dirt/dust particles. However, past research has found that T.S.S. are the main carrier of pollutants in highway runoff. According to Kobriger and Geinepolos (1984), Sartor et el. (1974), Gupta et al. (198171981a), Asplund et al. (1982) and Wang et al. (1982), 231 Table 4-35. Burnaby Site Pearson Correlation Between the Pollutants. T.S.S. Cu Fe M n Zn Ca OG T.S.S. 1 ~ — ~ — — ~ Cu 0.18(0.68) 1.00 ~ — — — ~ Fe -0.08(0.86) 0.68(0.07) 1.00 ~ — — ~ M n 0.29(0.48) 0.90(2.0x10"3) 0.72(0.04) 1.00 ~ ~ ~ Zn 0.30(0.48) 0.96(2.0x10"4) 0.67(0.07) 0.93(9.0x10"4) 1.00 — ~ Ca -0.14(0.74) 0.74(0.03) 0.51(0.20) 0.76(0.03) 0.76(0.06) 1.00 ~ OG 0.76(0.03) 0.26(0.53) -0.32(0.44) 0.23(0.59) 0.37(0.37) -0.12(0.78) 1.00 Note: 1) OG-Oil/grease 2) Numbers in the bracket are the P-values (statistical significance of the sample values) 3) Null hypothesis of no correlation between variables was rejected i f P<=0.10. Table 4-36. Richmond Site Pearson Correlation Between the Pollutants. T.S.S. Cu Fe Mn Zn Ca OG T.S.S. 1.00 — ~ — ~ — — Cu 0.74(0.10) 1.00 ~ — ~ — — Fe 0.74(0.09) 0.98(1.0xl0"3) 1.00 — — — — Mn 0.18(0.74) 0.66(0.16) 0.60(0.21) 1.00 — — — Zn 0.71(0.12) 0.87(0.02) 0.87(0.03) 0.71(0.12) 1.00 ~ — Ca -0.48(0.34) 0.17(0.75) 0.12(0.88) 0.75(0.09) 0.13(0.82) 1.00 ~ OG 0.58(0.23) 0.94(0.01) 0.96(2.0xl0"3) 0.72(0.11) 0.80(0.06) 0.35(0.50) 1.00 Note: 1) OG~Oil/grease 2) Numbers in the bracket are the P-values (statistical significance of the sample values) 3) Null hypothesis of no correlation between variables was rejected i f P<=0.10. 232 T.S.S. exhibit the highest degree of association with other pollutants. This seems to be true based on the Richmond site correlation coefficients. There were significant correlations between T.S.S. and most of the other pollutants with the exception of M n and Ca. Correlations were observed to be less with the Burnaby site data. The degree of T.S.S. association with different pollutants between the Burnaby and Richmond sites may have had something to do with the local soil type. Generally, heavy metals can be associated with sediments in a number of ways, the more important processes being chemisorption of heavy metals on to clays or Mn/Fe hydrous oxides, precipitation of discrete heavy metal compounds, and flocculation/complexation of heavy elements associated with reactive organic materials (Ferguson, 1990). Chemisorption, one of the dominant processes, involves the sorption of heavy metals onto clays. This process is controlled by the number of free-sorption sites on the clay surface which depends on the free or broken bonding positions, as well as on the proportion of atoms replaced with others of different valencies in the clays. Other factors that influence the sorption process are the pH, the nature of the heavy metal species such as their charge and hydration, and the clay type. The clays with expanding ability and large surface area, such as montmorillonite, have greater capacity to accumulate heavy metals than kaolinite (Laxen, 1983; Ferguson, 1990). This may explain the differences in T.S.S. association with other pollutants between the two sites. For example, Yousef et al. (1985a), in their research at the Maitland Interchange and Interstate 4 in Florida, found that the presence of organic substances and sediments in natural waters played a key role in the detoxification of metals associated with highway runoff. According to Colwill et al. (1984), the direct impact of sediments on water quality is probably not as significant as their capacity for adsorbing, transporting and releasing contaminants, particularly metals. 233 Also there were strong associations between metals in this study agreeing with the work of Yousef et al. (1985/1985a). In fact, the activity of both essential and toxic metals depends to a large extent on the ability of their ions to combine with other atoms and molecules as well as on their speciation in aqueous solution. The relative importance of the factors affecting chemical speciation requires consideration of topics such as kinetics, thermodynamics, chemical equilibrium and stability constant data (Laxen, 1983; Morgan, 1987). Some of the factors such as pH, complexing agents and adsorption/desorption onto particulate matter are known to affect metal interactions to different degrees. However, these topics are beyond the scope of this research project. Summary Pearson correlation analysis shows the best T.S.S. correlation at the Burnaby site to be with OG followed by Mn/Zn, Cu, Fe, and Ca. At the Richmond site, the best T.S.S. correlation was obtained with Cu/Fe followed by Zn, oil/grease, Mn, and Ca. The degree of T.S.S. association with different pollutants between the Burnaby and Richmond sites may have had something to do with the local soil type. According to Ferguson (1980), heavy metals can be associated with sediments in a number of ways with the more important processes being chemisorption of heavy metals on to clays or Mn/Fe hydrous oxides, precipitation of discrete heavy metal compounds, and flocculation/complexation of heavy elements associated with reactive organic materials. There were strong associations between the metals at both sites. This agrees with the work of other researchers on metal to metal interactions. 234 4.7.2. Regression Forecasting Model In this section, correlation and regression analyses are performed with all the variables at both sites for possible use in pollutant forecasting. An attempt was made to use all the environmental variables in the multiple or step wise regression equations to predict pollutant concentrations. But single regression models showed better prediction. Data from the storm event of 03/09/96 were used to check the forecasts and thus were not used in the regression analysis to derive the predictive equations. A l l the pollutant estimates were calculated using log transformed data. According to Macdonald et al. (1996), the parent distributions of such data have log-normally distributed errors, and log transformations are necessary to meet assumptions of the general linear model. Because of limited power due to small sample size, a rejection criterion of P<=0.10 was used for all test for significance. This reduces the probability of a type 2 error-Tables 4-37 and 4-38 show the results of correlation analyses performed between all the environmental variables at both Burnaby and Richmond sites respectively. At the Burnaby site, the rainfall amounts (X,) seemed to have strong relationships with runoff volume (X 2), number of dry days (X 3), and duration (X 6) with correlation coefficients of 0.93, 0.94, and 0.67 respectively. Runoff volume (X 2), also, had a high correlation coefficient (0.73) with duration (X 6), as indicated in Table 4-37. This suggests that only one or two of these variables need to be used for predictive purposes in the models. A l l of the correlations are what one would expect, except for the relationship between rainfall amount and number of dry days before the event, which must have been just an "accident" of the data. At the Richmond site, the rainfall amounts (X,) had a strong relationship with duration (X 6) with a correlation coefficient of 0.76. Whereas the runoff volume (X 2 ) had both a negative and a positive relationship with dustfall (X 4) and duration (X 6) with correlation coefficients of -0.62 and 0.67 respectively, as shown in Table 4-38. As mentioned earlier, this suggests that only 235 Table 4-37. Correlation Matrix of Environmental Variables at the Burnaby Site x, x 2 x 3 x 4 x 5 0 x, — 0.93 0.94 -0.24 -0.16 0.67 x 2 ~ — 0.21 -0.38 -0.16 0.73 x 3 ~ — ~ -0.19 -0.17 -0.05 x 4 — — ~ — -0.24 -0.38 x 5 — — . — — -0.16 x 6 — — ~ — — ~ Note: 1) X I = Rainfall amounts 2) X2 = Runoff volume 3) X3 = # of dry days 4) X4 = Dustfall 5) X5 = ADT 6) X6 = Duration Table 4-38. Correlation Matrix of Environmental Variables at the Richmond Site x, x 2 x 3 x 4 x 5 x 6 x, — 0.41 -0.29 0.18 0.29 0.76 x 2 — — -0.31 -0.62 0.14 0.67 x 3 ~ — ~ -0.13 0.25 -0.43 x 4 ~ ~ — — -0.24 -0.19 x 5 ~ ~ — ~ — 0.42 x 6 • — — — — ~ ~ Note: 1) X I = Rainfall amounts 2) X2 = Runoff volume 3)X3 =#ofdry days 4) X4 = Dustfall 5) X5 = A D T 6) X6 = Duration 236 one or two of these variables can be used for predictive purposes in the models. The results of the rest of the correlation coefficients are so low that substitution is not possible. Due to the small number of data set at both sites, multiple regression analyses can not be used for forecasting any of the pollutants at either site. To test how each of the independent variables relate to the dependent variables, independent regression analysis was performed on each variable with respect to the pollutants. The storm event of 03/09/96 was excluded from the regression analysis at both sites, since it is be used to test how the generated models performed with predicting the pollutants. At the Burnaby site, the number of dry days had a significant correlation with T.S.S. and oil/grease with r2 values of 0.50 and 0.94 respectively whereas dustfall significantly correlated with Fe with an r2 value of 0.69. The rest of the environmental variables did not show any . correlation with any of the pollutants. However, dustfall did show certain degree of relationship with Mn with an r2 value of 0.41, thus a predictive equation was developed for Mn. Table 4-39 and Figures 4-24 to 4-27 show the rest of the regression results at the Burnaby site. Similarly, Figures 4-30 to 4-33 show the residual plots at this site. At the Richmond site, only dustfall had a significant correlation with Ca with an r2 value of 0.91. The rest of the environmental variables did not show any correlation with any of the pollutants. But like the Burnaby site, rainfall amounts did show certain degree of relationship with T.S.S. with an r2 value of 0.43, thus a predictive equation was developed for T.S.S. Table 4-40, Figures 4-28, 4-29, 4-34 and 4-35 show the rest of the regression results and residual plots at the Richmond site. The resulting forecasting equations for the pollutants at both the Burnaby and Richmond sites are: 237 Table 4-39. Results of the Single Regression Model Analyses at the Burnaby Site. Parameters Effect Coefficients Standard Error Standard Coefficient F-Ratio P-Value T.S.S (0.50) Dry Days 13.1 5.81 0.71 5.04 0.07 Fe(0.69) Dustfall 4.68 1.39 0.83 11.2 0.02 Mn (0.41) Dustfall 0.08 0.04 0.64 3.43 0.12 Oil/Grease (0.94) Dry Days 3.82 0.42 0.97 82.4 3.0x104 Note: 1) Bold numbers are the r2 and P-values. 2) Degrees of freedom (1 and 5). Table 4-40. Results of the Single Regression Model Analyses at the Richmond Site. Parameters Effect Coefficients Standard Error Standard Coefficient F-Ratio P-Value T.S.S (0.43) Rainfall 10.6 7.01 0.66 2.28 0.23 Ca (0.91) Dustfall 3.90 0.72 0.95 29.3 l.OxlO"2 Note: 1) Bold numbers are the r 2 and P-values. 2) Degrees of freedom (1 and 3). 238 Burnaby Plot of T.S.S. Vs Dry Days y = 216+13.1x R2 = 0.50 • TSS — Linear(TSS) 0 5 10 15 Number of Dry Days (days) Figure 4-24. Burnaby Regression Plot for T.S.S. Burnaby Plot of Fe Vs Dustfall c o 5 d o> c o c o o 10 5 0 0 1 -0.57+4.68x1 R2 =0.69 • F E Linear (FE) Dustfall (micrograms/cubic meter) Figure 4-25. Burnaby Regression Plot for Fe. 239 (A C o CO M V u c o o Burnaby Plot of Mn Vs Dustfall y = 0.03 + 0.08x R2 = 0.41 0.3 0.2 0.1 0 MN •Linear (MN) 0 1 2 3 DustFall (micrograms/cubic meter) Figure 4-26. Burnaby Regression Plot for Mn. Burnaby Plot of Oil & Grease Vs Number of Dry Days y = 15.6 + 3.82x ( O G ) e 0 5 10 15 Number of Dry Days (days) Figure 4-27. Burnaby Regression Plot for Oil and Grease. 240 Richmond Plot of T.S.S. Vs Rainfall C o 200 150 3 » 100 o 50 0 0 5 10 RainFall Amounts (mm) y = 55.7 + 10.6x R2 = 0.43 TSS -Linear (TSS) Figure 4-28. Richmond Regression Plot for T.S.S. Richmond Plot of Ca Vs Dustfall » c o cs = i II o o a O 8 e ^ 4 2 0 y = -1.59 + 3.90x| R2 = 0.91 CA •Linear (CA) 0 1 2 Dustfall (micrograms/cubic meter) Figure 4-29. Richmond Regression Plot for Ca. 241 Burnaby T .S .S . Plot of Residuals against Predicted Values )0 250 300 350 400 ESTIMATE Figure 4-30. Burnaby Residuals Plot for T.S.S. Burnaby Fe Plot of Residuals against Predicted Values 21 4 5 6 7 8 9 10 E S T I M A T E Figure 4-31. Burnaby Residuals Plot for Fe. 242 Burnaby M n Plot of Residuals against Predicted Values Figure 4-32. Burnaby Residuals Plot for Mn. Burnaby O i l and Grease Plot of Residuals against Predicted Value 1! I0 20 30 40 50 60 70 ESTIMATE Figure 4-33. Burnaby Residuals Plot for Oil and Grease. 243 Richmond T.S.S. Plot of Residuals against Predicted Values - J i i i i l l l 60 70 80 90 100110120130140 ESTIMATE Figure 4-34. Richmond Residuals Plot for T.S.S. Richmond Ca Plot of Residuals against Predicted Values 0. 3 4 5 6 7 ESTIMATE Figure 4-35. Richmond Residuals Plot for Ca. 244 Burnaby site predictive equations based on flow composite samples are: T.S.S. cone. = 216+ 13.1x3 Fe cone. = -0.57 + 4.68x4 Mn cone. = 0.03 + 0.08x4 Oil/grease cone. = 15.6 + 3.82x3 Richmond site predictive equations based on flow composite samples are: T.S.S. cone. = 55.7+ 10.6x, Ca cone. = -1.59 + 3.90x4 x, = rainfall amount x 2 = runoff volume x 3 = number of dry days x 4 = monthly dustfall x 5 = ADT x 6 = rainfall duration Using the above single regression predictive equations, pollutant concentrations for the storm of 03/09/96 were computed for both sites with flow composite samples, as indicated in Tables 4-41 and 4-42. The storm event of 03/09/96 was used because data on the same storm was collected at the same time at both sites. The predictions for all the pollutants, using the single regression equations for the single storm event, are quite good, as shown in Tables 4-41 and 4-42. The only exception is the oil and grease concentration at the Burnaby site which was poorly predicted. Comparison Between the Simple and Flow Composite Data An Analysis of Variance (ANOVA) was performed to test i f there is any statistical significant difference between the simple and flow composite data obtained at both sites. Only T.S.S. at the Burnaby site showed a statistically significant difference between simple and flow composite data at 95% confidence level. The F-ratio for T.S.S. was found to be 11.7 with a P-value of 2.00xl0"3 at 1 and 28 degrees of freedom. 245 Table 4-41. Comparison of Observed Pollutant Concentrations Versus those Predicted by the Single Regression Model at the Burnaby Site for the Storm Event of 03/09/96. Flow Composite Dependent Variable Observed Concentration (mg/L) Predicted Concentration (mg/L) % Difference Fe 8.26 6.45 -28.1 Mn 0.15 0.15 0.00 Oil/grease 31.2 17.5 -78.3 T.S.S. 250 223 -12.1 Note: 1) % Difference ~ Percent difference between observed and predicted data 2) - Indicates under estimation of pollutant concentration 3) + Indicates over estimation of pollutant concentration Table 4-42. Comparison of Observed Pollutant Concentrations Versus those Predicted by the Single Regression Model at the Richmond Site for the Storm Event of 03/09/96. Flow Composite Dependent Variable Observed Concentration (mg/L) Predicted Concentration (mg/L) % Difference Ca 3.50 3.48 -0.57 T.S.S. 131 113 -15.9 Note: 1) % Difference ~ Percent difference between observed and predicted data 2) - Indicates under estimation of pollutant concentration 3) + Indicates over estimation of pollutant concentration 246 The rest of the parameters like Zn, Fe, Mn, Ca and oil/grease showed no statistically significant differences at 95% confidence level between the simple and flow composite data at both sites. Discussion Kerri et al. (1985) used simple linear regression in their forecasting of pollutant concentrations/loadings from highway runoff in California. They found the forecasting equations to estimate the cumulative loads of pollutants such as COD, total Pb, T K N and total Zn to be statistically significant at 5% level when correlated with the number of vehicles during storms (VDS). When analysing cumulative pollutant loads/concentrations, they found the number of dry days and corresponding cumulative traffic before the storm not to be statistically significant. However, no attempt was made to perform any multiple regression analysis with all the potential influential environmental factors. Swain (1985), on the other hand, performed multiple regression analyses with his residential stormwater monitoring data from Vancouver, British Columbia. Using only flow, antecedent number of dry days and suspended metals, Swain (1985) gave varying r2 values for his suspended metals and T.S.S. multiple regression equations. He reported r 2 values of 0.54, 1.00, 0.43 and 0.53 for suspended Cu, Fe, Zn and T.S.S. equations respectively. The fitted equation with the best correlation depends on the independent variables used in the analysis. Using a combination of suspended metals and antecedent number of dry days as independent variables, Swain (1985) came up with several equations. For example, his best equation for Zn, with the highest correlation, was a three variable model using suspended Fe, Pb and T.S.S. as independent variables (r2= 0.53) without any influential environmental variable like the number of dry days. In this research, independent regression analysis was performed on each variable with 247 respect to the pollutants to see how they relate. The results showed that not all the environmental variables can be used in the predictive equations. An attempt to use multiple regression in developing the models did not work due to small sample size. Single regression was used and so far, has shown the best predictive results. For example, at the Burnaby site, the single regression models under estimated all the pollutants to a lesser degree, with the exception of Mn which was the same for both the observed and the predicted, as indicated in Table 4-41. The single regression model estimated the T.S.S. concentrations fairly well for the individual storm event, as indicated in Tables 4-41 and 4-42 at both sites. This agrees with the research studies of Kobriger et al. (1981a), Asplund et al. (1982), Chui et al. (1981 and 1982) and Wang et al. (1982) in which they reported T.S.S. to be the best parameter used in developing models as well as in predicting other parameters. They estimated all the other parameters from T.S.S. concentration/loading. Not enough data was available in this study to do the same estimation. In this research, an attempt was made to estimate all the other parameters independent of the T.S.S. concentration. The single regression models gave reasonable predictions for the concentrations of Fe, Mn, at the Burnaby site, and Ca at the Richmond site, but not for oil and grease. Using environmental variables in single regression models was also done in an earlier research study in California by Kerri et al. (1985). Other variables upon which no correlation analyses could be performed included surrounding land use activities and highway maintenance practices. It is difficult to measure the effect of surrounding land use activities on highway stormwater runoff quality. But indirectly, factors such as ADT, urban vs rural and industrial vs commercial vs agricultural relate to the surrounding land use activities. Both sites are located within an urban region but the Burnaby site is surrounded by residential/industrial/commercial dwelling areas compared to the Richmond site which is surrounded by mostly agricultural lands. 248 Highway maintenance activities can have quite an impact on highway pollutant loadings. But, because of lack of numerical maintenance data from the contractor, this could not be included in the correlation analysis. Summary Pollutant concentrations forecasting, using single regression equations, yielded reasonably good predictions for T.S.S., Fe, and Mn at the Burnaby site; and T.S.S., and Ca at the Richmond site. This finding that T.S.S. could be best predicted tends to agree with the research studies of Asplund et al. (1982), Chui et al. (1981 and 1982) and Wang et al. (1982) in which they reported T.S.S. to be the most predictable parameter, which could then be used in developing models to predict other parameters. Comparison between simple and flow-composite data showed a statistically significant difference only for T.S.S. concentration at 95% confidence level at the Burnaby site. The rest of the parameters like Zn, Fe, Mn, Cu, Ca, and oil/grease did not show any significant concentration differences between simple and flow-composite data at both sites. 249 Chapter 5 CONCLUSIONS A N D RECOMMENDATIONS 5.1 Objectives This study was initiated to gather information on the quality of runoff from highway surfaces in the Greater Vancouver area, by sampling and analysing data from two carefully selected sites. As outlined in section 1.4, it was hoped to answer several specific questions. These are discussed in summary form in this section. 1) Was there any pollutant concentration variation with seasons? T.S.S., and Fe showed higher concentrations during winter than during the non-winter period at the Burnaby site. Cu and Mn did not show any definite variation in seasonal concentration at either site. Ca and oil/grease, had higher concentrations during the non-winter than during the winter period at both sites as well as Zn at the Bumaby site and Fe at the Richmond site. However, none of the variations were statistically significant. 2) How effective was the grass drainage ditch in pollution reduction? There was no grass drainage ditch study at the Richmond site, but the grass drainage ditch at the Bumaby site was very effective in pollution removal. The grass ditch resulted in significant pollution reduction of all parameters except for M n and Cu, with the order of pollutant reductions as follows: T.S.S. > Fe > Zn > Pb > Mn > Cu 3) What were the primary mechanisms for trace metal removal? Both soil/sediment adsorption and biological plant uptake processes were working concurrently in trace metal removal from the highway stormwater runoff. Analyses of both soil sediment and the grass clipping samples indicated consistently higher metal concentrations in the 250 soil sediment and grass clipping samples collected at the 30 m mark along the drainage ditch, than at inlet to the ditch. 4) What will be the fate of the research data after final analysis? The data and some research results have already been forwarded to the British Columbia Ministry of Transportation and Highway. And a copy of this thesis will be provided to the office. 5) What practical recommendations can be made? The grass drainage ditch design criteria at the Burnaby site was effective in pollutant reduction. This suggests that running stormwater runoff from highways through grassed ditches could be an effective way to treat the runoff. 5.2 Conclusions Most of the detailed conclusions reached in this research are presented in the summaries at the end of each section in Chapter 4, so that specific conclusions can be related directly to the data and results from which they were derived. In this section, only the main conclusions are presented. Table 5-1 shows a summary of the data and a comparison of the two sites at which data was collected. The Burnaby site, with larger area and higher ADT, has higher mean overall runoff coefficient than the Richmond site and the differences in runoff coefficients were statistically significant at 95% confidence level between the two sites. The mean concentrations and loadings of T.S.S., and Fe were higher during the winter than during the non-winter period at both sites. Although Cu and M n did not show any seasonal variations in concentrations, there were slight variations in their seasonal loadings. At both sites, the concentrations of Cd, N i and Cr were below the instrument's detection limits. Ca and oil/grease, on the other hand, have higher mean concentrations during the non-winter than during 251 Table 5-1. Summary of Highway Runoff Data for Burnaby and Richmond Sites. Parameters Units Burnaby Richmond Catchment Area (ha) 0.14 0.09 ADT (vehicles/day) 89,000 42,000 Rainfall Mean winter Mean non-winter (mm/event) 5.89 2.69 4.20 1.79 Runoff Coefficients Mean winter Mean non-winter Mean 0.82 0.63 0.70 0.65 0.51 0.55 T.S.S. Concentrations Mean winter Mean non-winter Total (mg/L) 261 (321) 257(136) (458) 96.3 (42.0) 100(19.3) (61.3) Export Coefficients (kg/ha/yr) 32.7xl0 2 681 Cu Concentrations Mean winter Mean non-winter Total (mg/L) 0.07 (0.09) 0.08 (0.04) (0.13) 0.04 (0.02) 0.05 (0.01) (0.03) Fe Concentrations Mean winter Mean non-winter Total (mg/L) 6.94 (8.29) 5.56 (2.96) (11.3) 3.26(1.42) 3.87 (0.75) (2.17) Zn Concentrations Mean winter Mean non-winter Total (mg/L) 0.26 (0.32) 0.30(0.16) (0.48) 0.16 (0.07) 0.13 (0.02) (0.09) M n Concentrations Mean winter Mean non-winter Total (mg/L) 0.14(0.17) 0.15 (0.08) (0.25) 0.09 (0.04) 0.10(0.02) (0.06) Ca Concentrations Mean winter Mean non-winter Total (mg/L) 3.81 (4.72) 5.63 (2.99) (7.71) 3.89(1.70) 4.77 (0.92) (2.62) 252 Table 5-1. Summary of Highway Runoff Data for Burnaby and Richmond Sites (cont'd). Parameters Units Burnaby Richmond Oil/Grease Concentrations Mean winter Mean non-winter Total (mg/L) 21.1 (26.0) 28.6(15.2) (41.2) 6.17(2.69) 8.81 (1.69) (4.38) O/G Export Coefficients (kg/ha/yr) 294 48.7 Note: 1) Numbers in the bracket are parameter loadings in kg/yr. 253 the winter period at both sites as well as Zn at the Burnaby site and Fe at the Richmond site. The differences in T.S.S. and metal concentrations between winter and non-winter at both sites were not statistically significant at 95% confidence level for all the parameters analysed. But between the two sites, there were statistically significant differences in T.S.S., Cu, Fe, Mn, Zn and oil/grease concentrations at 95% confidence level. There was no grass drainage ditch study at the Richmond site, but the grass drainage ditch at the Burnaby site was very effective in pollution removal. Table 5-2 shows the seasonal and the mean overall pollutant removal efficiencies for T.S.S, Fe, Zn, Pb, Mn, and Cu. The removal efficiencies were computed after the runoff has gone through a 30 m in length grass drainage ditch. The descending order of pollutant removal efficiencies was: Overall efficiency: T.S.S. > Fe > Zn > Pb > Mn > Cu A l l the pollutants showed better removal efficiencies during the non-winter than the winter months with the exception of Cu. T.S.S. was removed the most by the grass drainage ditch. An Analysis of Variance (ANOVA) test indicated a statistically significant pollutant removal after the runoff had gone through the grass drainage ditch at 95% confidence level for all the parameters except for Mn. This indicates that probably both filtration and sedimentation processes are the primary removal mechanisms. Although these pollutants were reduced considerably after going through the grass drainage ditch, analyses of their concentrations after going through the grass ditch showed higher than normal concentrations still, at their point of collection. These reduced concentrations exceeded the established maximum acceptable concentrations (MAC) for either drinking water or the protection of freshwater aquatic life. 254 Table 5-2. Summary of the Grass Drainage Ditch Pollutant Removal Efficiencies at the Burnaby Site. Pollutant Removal Efficiencies (%) Pollutant Parameters Winter Range Winter Mean Non-Winter Range Non-Winter Mean Overall Mean T.S.S. 58-83 72 70-90 81 77 Cu 41-66 49 9-68 46 48 Fe 50-72 63 34-86 65 64 M n 13-75 38 56-63 59a 49 Zn 40-61 53 46-80 59 56 Pb 39-71 53a N D ND 53 Notes: 1 ) a - Only two samples detected. 2) N D — Not detectable. 255 Impact analysis on receiving water bodies using daphnia bioassays showed both the winter and the grass filtered highway stormwater runoff to be non-toxic. The winter organisms' survival rates were the same as the controls at both sites. A l l the non-winter Comp " A " daphnia bioassays showed some toxicity especially after 48 hours time lapse. Road dirt analyses showed winter Cr, Cd, N i , Cu, Zn and M n concentrations at the Burnaby site to be higher than the non-winter. The differences in road dirt metal concentrations between winter and non-winter were not statistically significant for any of the metals. Unlike the Burnaby site, all the road dirt non-winter metal concentrations were higher than the winter concentrations at the Richmond site and the concentration differences were significant. Between the two sites, road dirts from the Burnaby site have higher metal concentration than the Richmond site with the exception Ca. Soil sediment analyses at the Burnaby site indicated higher metal concentrations during the non-winter than the winter period. However, there were no significant differences between winter and non-winter metal concentrations at the Burnaby site for soil sediment " A " samples with the exception of N i , Cu, Pb, and Fe. Similarly, the only significant differences in seasonal metal concentrations for soil sediment " B " samples were observed with Cd, Zn, and Fe at 95% confidence level. Also soil sediment " B " samples, collected at the 30 - 32 m mark from the weir, have considerably higher metal concentrations than soil sediment " A " samples collected near the weir, suggesting that most of the metal pollutants were attached to fine sediments. The differences in metal concentrations between soil sediment " A " and " B " samples were statistically significant at 95% confidence level. Like the Burnaby site, the Richmond site also has higher metal concentrations during the non-winter than during the winter months except for N i . However, no 256 significant differences in seasonal metal concentrations were observed with any of the metals. Between the two sites, the Richmond site soil sediments have higher overall metal concentrations than the Burnaby site. However, the differences in soil sediment metal concentrations between the two sites were not statistically significant for all the metals except for Zn. The grass clipping samples showed higher winter metal concentrations than the non-winter at both sites. Grass clipping "A"samples showed seasonal significant differences in their concentrations except for Cd and N i whereas grass clipping " B " samples have significant differences in all metal concentrations except for Ni , Fe, and Ca. At the Burnaby site, the grass clipping " B " samples for both winter and non-winter were consistently higher than the corresponding grass clipping " A " values, suggesting that metal pollutants were picked up by the grass. The differences in metal concentrations between grass clipping " A " and " B " samples were significant for most of the metals except for Cd, Mn, and Ca at 95% confidence level. At the Richmond site, the seasonal differences in grass metal concentrations were significant except for Cd. Between the two sites, higher metal concentrations were obtained in grass clipping samples at the Richmond site than the Burnaby site. The only significant differences in grass metal concentrations between the two sites were with Cd, Zn, and Mn. Pearson correlation analyses were performed between T.S.S. and metals and oil/grease, and while they showed relationships, they were not very strong with most metals at both sites. Pollutant concentrations forecasting, using single regression equations with individual environmental variable, yielded reasonably good predictions for T.S.S., Fe, and M n at the Burnaby site; and T.S.S., and Ca at the Richmond site. Comparison between simple and flow-composite data showed a statistically significant difference only for T.S.S. concentration at 95% confidence level for the Burnaby site. 257 Recommendations Monitoring of all environmental variables used in the regression analysis should be continued at the two sites to generate more data as well as confirm the patterns and observations presented in this report. The research should be expanded to include other G.V.R.D. highways such as Lougheed Highway, Marine Drive and Barnett Highway with a different mix of vehicles, traffic patterns (ie. stop lights, non-curbed lanes), surrounding lane use and maintenance practices. Special studies should be designed to investigate the influence of surrounding land use activities and highway maintenance operations on stormwater runoff quality. The study should also evaluate the following: • Impact of asphalt versus concrete pavements on water quality. • Influence of curbing or lack of it on highway runoff quality. Monitor the impacts of highway runoff pollutants on receiving waters by analysing the runoff before it enters the stream as well as analysing the stream water. This will evaluate the impacts of dilution or lack of it from general urban runoff. Develop a long term special study to evaluate the fate of pollutants in soil sediments and grasses. What happens to metals that are picked up by plants upon plant death or after grass mowing? Find a better way to measure flow. The weir box used in flow measurements acted as a sedimentation basin. Approximately 474 kg (1,038 lb) and 79 kg (174 lb) of sediments were removed from the weir box at the Burnaby and Richmond sites respectively. Also installation of the weir boxes disrupted the structure of the drainage channels. 258 7) To carry out further investigations on the amount of runoff and pollutants lost from the highways due to vehicular splash-off. 8) Pollutant forecasting using single regression equations generated in this research should be extended to other highways to test their applicability. Also more research data is required to refine this model as well as test its effectiveness in pollutant concentration prediction. 259 B I B L I O G R A P H Y American Public Health Association (APHA), American Waterworks Association (A WW A), and Water Pollution Control Federation (WPCF). 1989/1995. Standard Methods for the Examination of Water and Wastewater. 16th and 18th Editions. American Public Health Association, Washington, D.C. American Public Health Association, American Waterworks Association, and Water Pollution Control Federation. 1992. Standard Methods for the Examination of Water and Wastewater. 18th Edition. American Public Health Association, Washington, D.C. Anderson, B. 1982. Toxicity of Urban Stormwater Runoff. MSc. Thesis. The University of B.C., British Columbia, Canada. ASIPWPCA. 1986. Meeting the Challenge of Nonpoint Source Control. Journal WPCF. Vol. 58. No. 7. Pp. 730-740. Asplund, R., Mar, B.W., and Ferguson, J.F. 1980. Characterization of Highway Runoff in Washington State. A Report Prepared for the Washington State Department of Transportation, Highway Runoff Water Quality Research Project, Seattle, Washington State. Asplund, R., Mar, B.W., and Ferguson, J.F. 1982. Total Suspended Solids in Highway Runoff in Washington State. J. of Environmental Engr. Division, Vol . 108. Pp. 391-404. Athayde, D.N., Shelley, P.E., Driscoll, E.D., Gaboury, D., and Boyd, G. 1983. Results of the Nationwide Urban Runoff Program, Vol . 1: Final Report. National Technical Information Service, Springfield, V A , PB 84 - 185552. Aye, R.C. 1979. Criteria and Requirements for Statewide Highway Runoff Monitoring Sites. Report to Washington State Department of Transportation by the Department of Civil Engineering, University of Washington, Seattle, Washington. B.C. Research Corporation. 1991. Urban Runoff Quality and Treatment: A Comprehensive Review, (Gibb et al.). British Columbia Research Corporation, 3650 Wesbrook Mall, Vancouver, B.C. V6S 2L2. B.C. Ministry of Transportation & Highways. 1992/3. Traffic Volumes in British Columbia 1988 - 1992. B.C. Ministry of Transportation & Highways, Planning Services Branch, 940 Blanshard Street, Victoria, B.C. B.C. Environment. 1992. Urban Runoff Quality Control Guideline for British Columbia. Prepared for B.C. Environment by B.C. Research Corporation. 3650 Wesbrook Mall , Vancouver, B.C. 260 B I B L I O G R A P H Y (cont'd) Babich, H., and Stotzky, G. 1985. Heavy Metal Toxicity to Microbe Mediated Ecologic Processes: A Review and Potential Application to Regulatory Policies. Environmental Reviews. Vol . 36. No. 1. Pp. 111-137. Balades, J.D., Cathelain, M . , Marchandise, P., Peybernard, J., and Pilloy, J.C. 1984. Chronic Pollution Intercity Motorway Runoff Waters. Water Sci. Technol. Vol . 17. Pp. 1165-1174. Bastian, R.K. 1986. Potential Impacts on Receiving Waters. In: Urban Runoff Quality: Proceedings of an Engineering Foundation Conference, New England College, Henniker, New Hampshire, June 23-27. American Society of Civi l Engineers, New York. Pp. 157-169. Becker, B.C., Char, M.L. , and Kautzman, R.R. 1973. Approaches to Stormwater Management. U.S. Department of Interior, Office of Water Resources Research, Washington, D.C. Bellinger, E.G., Jones, A.O., and Tinker, J. 1982. The Character and Dispersal of Motorway Runoff Water. Water Pollut. Control. Vol . 81 No. 3. Pp. 372-390. Bertrand, F., MacLaren, J.W., and Pearse, P.H. 1985. Canada's Freshwater Resources. Inquiry on Federal Water Policy. Government Publication. Ottawa, Canada. Black, P.E. 1980. Water Quality Patterns During a Storm on Mall Parking Lot. Water Res. Bull. Vol . 16. No. 4. Pp. 615-620. Bradley, R.W., and Sprague, J.B. 1985. The Influence of pH, Water Hardness and Alkalinity on the Acute Lethality of Zinc to Rainbow Trout, (Salmo Gairdneri). Can. J. Fish. Squat. Sci. Vol . 42. Pp. 731-736. Brief, R.S. 1962. Air Lead Concentrations from Automobile Engines. Arch. Environ. Health. Vol. 5. Pp. 527-530. Brown, L.R., Chandler, W.U., Durning, A. , Flavin, C , Heise, L. , Jacobson, J., Postel, S., Pollock-Shea, C , Stake, L. , and Wolf, E.C. 1988. State of the World. AWorldwatch Institute Report on Progress Toward a Sustainable Society. W.W. Norton & Company, New York. Burch, C.W., Johnson, F., and Maestri, B. 1985. Management Practices for the Mitigation of Highway Stormwater Runoff. Vol . 1,2,3. Prepared for Federal Highway Administration (FHWA/RD-85/001), Office of Research and Development, Environment Division, Washington, D.C. Christensen, E.R, and Guinn, V.P. 1979. Zinc from Automobile Tires in Urban Runoff. J. of Environment Engr. Division. Vol . 105 Pp. 165-168. 261 B I B L I O G R A P H Y (cont'd) Chui, D.T.W., Mar, B.W., and Horner, R.R. 1981. Highway Runoff in Washington State: Model Validation and Statistical Analysis. A Report Prepared for the Washington State Department of Transportation Highway Runoff Water Quality Research Project by Department of Civi l Engineering, University of Washington, Seattle. Chui, T.W., Mar, B.W., and Horner, R.R. 1982. Pollutant Loading Model for Highway Runoff. J. Environ. Engr. Div., Vol . 108. Pp. 1193-1210. Cole, R.H., Frederick , R.E., Healy, R.P., and Rolan, R.G. 1984. Preliminary Findings of the Priority Pollutant Monitoring Project of the Nationwide Urban Runoff Program. Journal of WPCF., Vol . 56. No. 7. Pp. 898-908. Collins, Y . , and Stotzky, G. 1989. Factors Affecting the Toxicity of Heavy Metals to Microbes. In: Metal Ions and Bacteria. Beveridge, T.J. and Doyle, R.J. Eds. Wiley. Pp. 31-90. Colwill, D .M. , Peters, C.J., and Perry, R. 1984. Water Quality of Motorway Runoff. Transport and Road Research Laboratory Supplementary Report 823. U .K. Davies, P.H. 1986. Toxicology and Chemistry of Metals in Urban Runoff. In: Urban Runoff Quality: Proceedings of an Engineering Foundation Conference, New England College, Henniker, New Hampshire, June 23-27. American Society of Civi l Engineers, New York, Pp. 60-78. Davis, G., Krannitz, S., and Goldstein, M . 1991. A Reconnaissance Study of Roadside Tree Injury and Decline at 17 Sites in Interior British Columbia. Phase II. Prepared for Roadside Tree Injury Committee. Ministry of Environment, 780 Blanshard Street, Victoria, B.C. Canada. Di Torro, D . M . 1984. Probability Model of Stream Quality due to Runoff. Journal of Environ. Engr., ASCE. Vol . 10. Pp. 607-628. Dukta, B.J., and Rybakowski, I. 1978. Microbiological Study of Some Canadian Stormwater Runoffs at Burlington and Brucewood. Bacteriological Data Collection for the Urban Drainage Subcommittee, Canada-Ontario Agreement on Great Lakes Water Quality, Ontario Ministry of Environment, Toronto, Ont. Canada, Project No. 75-3-24. Ellis, J.B., and Revitt, D . M . 1982. Incidence of Heavy Metals in Street Surface Sediments: Solubility and Grain Size Studies. Water, Air, Soil Pollut., Vol . 17. Pp. 87-100. El-Shaarawi, A . H . 1989. Inferences about the Mean from Censored Water Quality Data. Water Res. Research, Vol . 25. No. 4. Pp. 685-690. 262 B I B L I O G R A P H Y (cont'd) Environment Canada. 1993. CSO and UR Investigative Assessment Guidelines. Fraser Pollution Abatement Office, Environment Canada. 224 West Esplanade Ave., North Vancouver, B.C. Environment Canada. 1987/95. Canadian Water Quality Guidelines. Prepared by Task Force on Water Quality Guidelines of the Canadian Council of Resource and Environment Ministers. Environment Canada, Ottawa, Canada. Environment Canada. 1990. Environment Canada Biological Test Method: Acute Lethality Test Using Daphnia Species. EPS 1/RM/l 1. Environment Canada, Ottawa, Canada. Environment Canada. 1995/97/98. Canadian Climate Data. Print out from Environment Canada Climate Information Branch 1200-W.73 Avenue, Vancouver, B.C. Canada Fam, S., Stenstrom, M.K. , and Silverman, G. 1987. Hydrocarbons in Urban Runoff. J. Environ. Eng. Div. Am. Soc. Civ. Eng. Vol . 101 (EEI). Pp. 107-125. Ferguson, J.E. 1990. The Heavy Elements in Water and Sediments. The Heavy Elements: Chemistry, Environmental Impacts and Health Effects. Chapter 8: Pp. 243-327. Field, R., and Pitt, R.E. 1990. Urban Storm-Induced Discharge Impacts. Water Environ. Technol. Vol . 2(8), Pp. 64-67. Forget, E., Courchesne, F., Kennedy, G., and Zayed, J. 1994. Response of Blue Spruce (Picea Pungens) to Manganese Pollution from M M T . Water, Air and Soil Pollut. Vol . 73. Pp. 319-324. G.V.R.D. 1992. Solid Waste Management Plan Review. Stage 1 Report. Waste Management Division. 4330 Kingsway, Burnaby, B.C., Canada.. G.V.R.D. 1993. Choices: Growth Management and Transportation. G.V.R.D. Communications and Education. 4330 Kingsway, 19th Floor, Burnaby, B.C., Canada G.V.R.D. 1993a. A Long-Range Transportation Plan for Greater Vancouver: Transport 2021 Report. G.V.R.D. Communications and Education. 19th Floor, 4330, Kingsway, Burnaby, B.C., Canada. Gannon, J.J., and Busse, M . K . 1989. E. Coli and Enterococci Levels in Urban Stormwater, River Waters and Chlorinated Treatment Plant Effluent. Water Res. Vol . 23(9), Pp. 1167-1176. Getz, L .L . , Haney, A.W., Larimore, R.W., McNumey, J.W., Leland, H.V. , Price, P.W., Ralfe, G.L., Wormian, R.L., Hudson, UJ.L., Solomon, R.L., and Reinhold, K . A . 1975. Transport and Distribution in a Watershed Ecosystem, in Lead in the Environment. R. Boggess, Ed., National Science Foundation, Washington, D . C , Pp. 105-134. 263 B I B L I O G R A P H Y (cont'd) Granet, I. 1989. Fluid Mechanics for Engineering Technology. Prentice-Hall, Inc., Englewood Cliffs, N.J. Pp. 363-369. Gupta, M . K . , and Kobriger, N.P. 1980. Sources and Migration of Highway Runoff Pollutants. Envirex, Inc., Milwaukee. Gupta, M.K. , Agnew, R.W., and Kobriger, N.P. 1981. Constituents of Highway Runoff Volume I, State-of-the-Art Report. Report No. FHWA/RD-81-042. Prepared for Federal Highway Administration, Washington, D.C. Gupta, M . K . 1981. Constituents of Highway Runoff. Vol . VI-Executive Summary. Report No. FWHA/RD-81-047. Prepared for Federal Highway Administration, Washington, D.C. Gupta, M.D. , Agnew, R.W., and Kreutzberger, W. 1978. Characteristics of Runoff from Operating Highways, Vol . IV, Envirex, Inc. for F H W A Contract DOT-FH-8600. Gupta, M.K. , Agnew, R.W., and Meinholz, T.L. 1981a. Constituents of Highway Runoff Volume II, Procedural Manual for Monitoring of Highway Runoff. Report No. FHWA/RD-81/043. Final Report to Federal Highway Administration, Washington, D.C. Gupta, M.K. , Agnew, R.W., Gruber, D., and Kreutzberger, W. 1981b. Constituents of Highway Runoff Volume IV, Characteristics of Highway Runoff from Operating Highways. Research Report. No. FHWA/RD-81/045. Report to Federal Highway Administration, Washington, D.C. Hall, K . H . , and Anderson, B.C. 1988. The Toxicity and Chemical Composition of Urban Stormwater Runoff. Can. J. Civ. Eng. Vol . 15 Pp. 98-106. Hall, K . H . , Yesaki, I., and Chan, J. 1976. Trace Metals and Chlorinated Hydrocarbons in the Sediments of a Metropolitan Watershed. Westwater Research Centre, University of B.C., Vancouver, B.C., Technical Report 10. Hanes, R.E., Zelazny, L.W., and Blaser, R.E. 1970. Salt Tolerance of Trees and Shrubs to De-icing Salts. Hwy. Res. Board. Vol . 335. Pp. 17-18. Harremoes, P. 1988. Stochastic Models for Estimations of Extreme Pollution from Urban Runoff. Water Res. Vol . 22. Pp. 1017-1026. Harrison, R .M. , and Wilson, S.J. 1985. The Chemical Composition of Highway Drainage Waters. Sci. Total Environ. Vol . 45. Pp. 63-102. Harwood, E. 1986. Impact of Motorway Runoff Upon Surface Water. Chapter 3 of Effects of Land Use on Fresh Waters. Ed. Perry, R., and Mclntyre, A .E . U .K. 264 B I B L I O G R A P H Y (cont'd) Hassett, J.J. 1974. Capacity of Selected Illinois Soils to Remove Lead to Aqueous Solution. Communications in Soil Science and Plant Analysis. Vol . 5. Pp. 499-510. Hay, J.E., and Oke, T.R. 1976. The Climate of Vancouver. B.C. Geographical Series. No. 23. Pp. 27-29. Hedley, G., and Lockley, J.D. 1975. Quality of Water Discharged from an Urban Motorway. Water Pollution Control, Vol. 74. No. 6, Pp. 659-674. Helsel, D.R., Kim, J.I., Grizzard, T.J., Randall, C.W., and Hoehn, R.C. 1979. Land Use Influences on Metals in Storm Drainage. Journal WPCF, Vol . 51. No. 4, Pp. 709-717. Hoffman, E.J., Mills, J.S., Latimer, J.S., and Quinn, J.G. 1984. Urban Runoff as a Source of Polycyclic Aromatic Hydrocarbon to Coastal Waters. Environ. Sci. Technol. Vol. 18. No. 8. Pp. 580-587. Hoffman, E.J., Latimer, J.S., Hunt, C D . , Mills, G.L., and Quinn, J.G. 1984a. Stormwater Runoff from Highways. Water, Air, Soil Pollut., Vol . 25 Pp. 349-364. Hopke, P.K., Lamb, R.E., and Natusch, D.F.S. 1980. Multiclemental Characteristication of Urban Roadway Dust. Environmental Science and Technology. Vol . 14, Pp. 164. Horner, R.R. 1988. Biofiltration Systems for Storm Runoff Water Quality Control. A Report to Washington State Department of Ecology, Municipality of Metropolitant Seattle, King County, City of Bellevue, City of Moutlake Terrace, and City of Redmond by Department of Civi l Engineering, University of Washington, Seattle. Horner, R.R. 1985. Environmental Criteria Manual, Volumes 1 & 2. Basis of Criteria. Final Report to Washington State Transportation Commission and Department of Transportation/US.FHA. Research Project 4-2811, Task 11 by Department of Civil Engineering, University of Washington, Seattle. Horner, R.R., and Mar, B.W. 1985. Assessing the Impacts of Operating Highways on Aquatic Ecosystems. Transportation Research Record 1017. Transportation Research Board, Washington, D . C Pp. 47-55. Horner, R.R., Burges, S.J., Ferguson, J.F., Mar. B.W., and Welch, E.B. 1979. Highway Runoff Monitoring: The Initial Year. Dept. of Civil Engr., University of Washington, Seattle for Washington State Dept. of Transportation. Horner, R.R., Waddle, T.J., and Burges. S.J. 1977. Review of the Literature on Water Quality Impacts of Highway Operations and Maintenance. Report to Washington State Department of Transportation by the Department of Civi l Engineering, University of Washington, Seattle, Washington. 265 B I B L I O G R A P H Y (cont'd) Howell, R.P. 1978. Water Pollution Aspects of Particles Which Collect on Highway Surfaces. California Dept. of Transportation Report to FHWA-C A-TL-78-22. Huntzicker, J.J., Friedlander, S.K., and Davidson, C.I. 1975. Materials Balance for Automobile-Emitted Lead in Los Angeles Basin. Environmental Science and Technology. Vol. 9. Pp. 445-457. Jones, P.H., Jeffrey, B.A. , Walter, P.K., and Hutchon, H . 1986. Environmental Impact of Road Salting-State of the Art. Research and Development Branch, Ontario Ministry of Transportation and Communications, Ontario, Canada. Pp. 53-57. Joselow, M . M . , Tobias, E., Koehler, R., Coleman, S., Bogden, J., and Gause, D. 1978. Manganese Pollution in the City Environment and its Relationship to Traffic Density. Am. J. Public Health. Vol . 68. No. 6. Pp. 557-560. Kerri, K.D. , Baad, M . , Shirley, E.C., Howell, R.B., Torguson E., and Winters, G. 1976. Water Quality Manual, Vol. V : Chemical, Bacteriological, and Ecosystem Analysis of Water from Highway Sources for Environmental Impact Assessments. Federal Highway Administration, Washington, D.C. Kerri, K.D. , Rancin, J.A., and Howell, R.B. 1985. Forecasting Pollutant Loads from Highway Runoff. Transportation Research Record 1017, Transportation Research Board, Washington, D.C. Pp. 39-46. Khan, Z., Thrush, C , Cohen, P., Kulzer, L. , Franklin, R., Field, D., Koon, J. and Homer, R.R. 1992. Biofiltration Swale Performance, Recommendations, and Design Consideration. A Report to Washington State Department of Ecology and Municipality of Metropolitan Seattle, Seattle, Washington. Kobriger, N.P., Gupta, M.K. , and Geinepolos, A . 1981. Sources and Migration of Highway Runoff Pollutants. Report No. FHWA/RD, Federal Highway Administration, Washington, D.C. Kobriger, N.P., Meinholz, T.L., Gupta, M.K. , and Agnew, R.W. 1981a. Constituents of Highway Runoff. Vol . III. Predictive Procedure for Determining Pollutant Characteristics in Highway Runoff. Prepared for Federal Highway Administration, Office of Research and Development, Enrionment Division, Washington, D.C. Kobriger, N.P., and Geinepolos, A . 1984. Sources and Migration of Highway Runoff Pollutants. Vol . 1 Research Report for Federal Highway Administration, Office of Research and Development, Environment Division, Washington, D.C. 266 B I B L I O G R A P H Y (cont'd) Kobriger, N.P. and Geinepolos, A. 1984a "Sources and Migration of Highway Runoff Pollutants - Research Report", Report. No. FHWA/RD-84-059, Vol . I l l , Prepared for Federal Highway Administration, Office of Research and Development, Environment Division, Washington, D.C. Kushner, D.J. 1993. Effects of Speciation of Toxic Metals on Biological Activity. Water Poll. Res. J. Canada, Vol . 28 No. 1 Pp. 111-128. Larkin, G. 1995. Hydrocarbon Pollution from Urban Runoff in the Brunette Watershed. M.A.Sc. Thesis. The University of B.C., British Columbia, Canada. Lawson, E .M. , Mitchell, G.B., and Walton, D.C. 1985. Stormwater Monitoring in an Industrial Catchment Basin located in Burnaby, B.C. British Columbia Ministry of Environment Report. Laxen, D.P.H. 1983. The Chemistry of Metal Pollutants in Water. Harrison, R . M . E. In Pollution: Causes, Effects and Control: Royal Chem. Sco. Pp. 104-123. Laxen, D.P.H., and Harrison, R . M . 1977. The Highway as a Source of Water Pollution: An Appraisal with the Heavy Metal Lead. Water Res. Vol . 11. Pp. 111-118. Lee, G.F., and Jones-Lee, A . 1993. Water Quality Impacts of Stormwater-Associated Contaminants: Focus on Real Problems. Water Sci. Tech. Vol . 28. Pp. 231-240. Lee, G.F. 1970. Eutrophication: Literature Review No. 2 - Eutrophication Information Centre, University of Wisconsin, Madison. Lenz, A.T. 1943. Viscosity and Surface Tension Effects on V-notch Weirs. Transactions of ASCE, Vol . 69, Pp. 759-802. Lindsey, G., Robers, L. , and Page, W. 1992. Inspection and Maintenance of Infiltration Facilities. J. of Soil and Water Cons. Vol . 47. No. 6. Pp. 481-486. Little, P., and Wiffen, R.D. 1978. Emission and Deposition of Lead from Motor Exhaust -11: Airborne Concentrations, Particle Size, and Deposition of Lead near Motorways. Atmospheric Environ. Vol . 12. Pp. 1331-1341. Little, L . M . , Horner, R.R., and Mar, B.W. 1983. Assessment of Pollutant Loadings and Concentrations in Highway Stormwater Runoff. A Report to Washington State Department of Transportation Highway Runoff Water Quality Research Project by Department of Civil Engineering, University of Washington, Seattle. 267 B I B L I O G R A P H Y (cont'd) Livingston, E.H., 1989. State Perspectives on Water Quality. In: Design of Urban Runoff Quality Controls: Proceedings of an Engineering Foundation Conference on Current Practice and Design Criteria for Urban Quality Control, Potosi, Missouri, July 10-15. American Society of Civi l Engineers, New York. Pp. 49-67. Loehr, R.C., Jewell, W.J., Novak, J.D., Clarkson, W.W., and Friedman, G.S. 1979. Nitrogen Considerations - Land Application of Wastes Volume 11. Van Nostrand Reinhold Environmental Engineering Series. Van Nostrand Reinhold Company. New York, U.S.A. Loftis, J.C., Ward, R.C., and Smillie, G .M. 1983. Statistical Models for Water Quality Regulation. Journal WPCF. Vol . 55. No. 8. Pp. 1098-1104. Loranger, S., Zayed, J., and Forget, E. 1994. Manganese Contamination in Montreal in Relation with Traffic Density. Water, Air and Soil pollut. Vol . 74. Pp. 385-396. Lorant, F.I. 1992. Highway Runoff Water Quality - Literature Review. Research and Development Branch, Ontario Ministry of Transportation, Ontario, Canada. Lorant, F.I. 1992a. Effects of Piped Median vs. Open Ditch Drainage on Stormwater Quality. Research and Development Branch, Ontario Ministry of Transportation, Ontario Canada. Lord, B . M . 1987. Non-point Source Pollution from Highway Stormwater Runoff Sci. Total Enviorn. Vol . 54. Pp. 437-446. Lord, B .N. 1989. Program to Reduce De-icing Chemical Use. In: Design of Urban Runoff Quality Control: Proceedings of an Engineering Foundation Conference on Current Practice and Design Criteria for Urban Quality Control. Trout Lodge, Potosi, Missouri, July 10- 15, 1988. American Society of Civil Engineers, New York. Pp. 421-435. Lygren, E., Gjessing, E., and Berglind, L. 1984. Pollution Transport from a Highway. Norwegian Institute for Water Research. The Science of the Total Environment. Vol. 33. Pp. 147-159. Elseview Science Publishers B.V.,Amsterdam. Lynard, W.G., Finnemore, J., Loopp, J.A., and Finn, R . M . 1980. Urban Stormwater Management and Technology Case Histories. EPA 600/8-80-035. U.S. Environmental Protection AGency, Cincinnati, Ohio. Macdonald, R., Hall, K. , and Schreier, H . 1996. Water Quality and Stormwater Contaminants in the Brunette River Watershed, British Columbia 1994/95. Draft Report. Westwater Research Unit, Institute for Resources and Environment. The University of B.C., British Columbia, Canada. Manahan, S.E. 1984. Enviornmental Chemistry. Brooks/Cold Publishing Company, Monterey, California. 268 B I B L I O G R A P H Y (cont'd) Mansfield, E. 1986. Basic Statistics with Applications. First Edition. W.W. Norton & Co., Inc., N . Y . Mar, B.W., Ferguson, J.F., Spyridakis, D.E., Welch, E.B., and Horner, R.R. 1981. Year 4 -Runoff Water Quality August 1980 - August 1981. A Report to Washington State Department of Transportation Highway Runoff Water Quality Research Project by Department of Civi l Engineering, University of Washington, Seattle. Marsalek, J., and Greek, B. 1984. Toxic Substances in Urban Runoff in the Niagara River Area. Unpublished Report Series. Environmental Hydraulics Section, Hydraulics Division, National Water Research Institute, Burlington, Ontario. Marsalek, J. 1986. Report on N A T O Workshop on Urban Runoff Quality. In: Urban Runoff Quality: Proceedings of an Engineering Foundation Conference, New England College, Henniker, Nw Hampshire, June 23-27. American Society of Civi l Engineers, New York, Pp. 15-28. Marsalek, J., and Ng, H.Y.F. 1989. Evaluation of Pollution Loadings from Urban Nonpoint Sources: Methodology and Applications. J. Great Lakes Res. Vol . 15. No. 3. Pp. 444-451. McCallum, D.W. 1995. An Examination of Trace Metal Contamination and Land Use in an Urban Watershed. M.A.Sc. Thesis. The University of B.C., British Columbia, Canada. McKenzie, D.J. and Irwin, G.A. 1983. Water Quality Assessment of Stormwater Runoff from a Heavily Used Urban Highway Bridge in Miami, Florida. U.S. Geological Survey Water-Resources Investigations Report 83-4153. Tallahassee, Florida. Meinholz, T.L., Kobriger, N.P., Gupta, M.K. , and Agnew, R.W. 1978. Predictive Procedure for Determining Pollutant Characteristics in Highway Runoff. Vol . I l l , Final Report. Report to U.S. Department of Transportation, FHA by Envirex, Inc., Milwaukee, Wisconsin. Morgan, J.J. 1987. General Affinity Concepts, Equilibria and Kinetics in Aqueous Metals Chemistry. In: Patterson, J.W. and Pasino, R. Eds. Metal Speciation; Separation and Recovery; Lewes Pub. Co. Pp. 27-61. Motto, H.L., Daines, R.H., Chilko, D .M. , and Motto, D.K. 1970. Lead in Soils and Plants: Its Relationship to Traffic Volume and Proximity to Highways. Environ. Sci. Technol. Vol. 4. Pp. 229-231. Murray, D .M. , and Ernst, U.F.W. 1976. An Economic Analysis of the Environmental Impact of Highway De-icing. U.S. Environmental Protection Agency, Cincinnati, Ohio. EPA -600/2-76-105. Pp. 127-135. 269 B I B L I O G R A P H Y (cont'd) Novotony, V. , and Chesters, G. 1981. Handbook of Nonpont Pollution, Sources and Management. Van Nostrand Reinhold Company, New York. Pp. 322-339. OMOE (Ontario Ministry of Environment). 1980. Manual of Practice on Urban Drainage. Obtain from: Training and Technology Transfer Division (Water) Environmental Protection Service, Environment Canada. Ottawa, Canada. Onwumere, G.C. 1992. Effects of Swine Manure, Chemical Fertilizer and Drainage Control on Water Quality. M . Sc. Thesis. The University of B.C., British Columbia, Canada. Onwumere, G.C.O., and Russell, S.O. 1995. Site Selection Criteria for Highway Stormwater Monitoring in the G.V.R.D. Report No. 1. Prepared for Highway Enviroment Branch, B.C. Ministry of Transportation and Highways, 940 Blanshard Street, Victoria, B.C. Canada. Pp. 1-22. Onwumere, G.C.O., and Russell, S.O. 1996. Preliminary Investigations of the Highway Stormwater Monitoring in the Greater Vancouver Regional District. Report No. 2. Prepared for Highway Enviroment Branch, B.C. Ministry of Transportation and Highways, 940 Blanshard Street, Victoria, B.C. Canada. Pp. 1-33. Onwumere, G.C.O. (1996a) Water Quality Assessment and Toxicity of Highway Stormwater Runoff in the G.V.R.D. Report No. 3 Prepared for Highway Enviroment Branch, B.C. Ministry of Transportation and Highways, 940 Blanshard Street, Victoria, B.C. Canada. Pp. 1-16. Owe, M . , Craul, P.J., and Halverson, J.G. 1982. Contaminants Levels in Precipitation and Urban Surface Runoff. Water Res. Bull. Vol . 16. No. 4. Pp. 615-620. Pham, C.H., Halverson, H.G., and Heisler, G .M. 1978. Precipitation and Runoff Water Quality from an Urban Parking Lot and Implications for Tree Growth. U S D A Forest Service Northeast Forest Experiment Station Research Note NE-253. Poland, C. 1975. Grassed Waterway Design. In U.S. EPA Non-Point Source Pollution Seminar Section 108(a) Demonstration Projects Progress Reports. EPA 905/0-75-007 U.S. Environmental Protection Agency, Chicago, Illinois. Poon, D.T988. Annual Report: Leaded and Lead-Free Gasoline Regulations Monitoring Program. Environment Canada, 88-04. Portele, G.J., Mar, B.W., Horner, R.R., and Welch, E.G. 1982. Effects of Seattle Area Highway Stormwater Runoff on Aquatic Bioata. Report No. 11 . Prepared for the Washington State Deparment of Transportation, Highway Runoff Water Quality Research Project. Pp. 1-41. 270 B I B L I O G R A P H Y (cont'd) Randall, C , and Grizzard, T. 1983. Runoff Pollution. In Stormwater Management in Urbanizing Areas, W. Whipple, N . Grigg, T. Grizzard, C. Randall, R. Shubinski, and L. Tucker, Ed's. Pp. 58-93. Prentice-Hall, New Jersey. Russell, S.O., Kenning, B.F.I., and Sunnell, G.F. 1979. Estimating Design Flows for Urban Drainage. J. of the Hydraulic Division. Pp. 43-51. Sartor, J.D. and Boyd, G.B. 1973. Water Pollution Aspects of Street Surface Contaminants. EPA-R2-72-081. Report to U.S. Environmental Protection Agency, Washington, D.C. Sartor, J.D., Boyd, G.B., and Agardy, F.J. 1974. Water Pollution Aspects of Street Surface Contaminants. Journal of WPCF, Vol . 46, No. 3, Pp. 458-467. Scanlon, P.F. 1977. A Study of Lead, Cadmium, Nickel and Zinc Levels in Soil, Vegetation, Invertebrates and Mammals Associated with Highways of Different Traffic Densities. Report to U.S. Department of Transportation by Department of Fisheries and Wildlife Sciences, Virginia Polytechnic Institute and State University, Blacksburg. Schillinger, J.E., and Gannon, J.J. 1985. Bacterial Adsorption and Suspended Particles in Urban Stormwater. Journal WPCF, Vol . 57, No. 5., Pp. 384-389. Schueler, T.R. 1987. Controlling Urban Runoff: A Practical Manual for Planning and Designing Urban BMP's. Prepared for: Washington Metropolitan Water Resources Planning Board. Metropolitan Washington Council of Governments, Washington, D.C. Shaheen, D.G. 1975. Contributions of Urban Roadway Usage to Water Pollution. Biospherics Inc., Rockville, M D , Report for EPA Contract 86-01-0197. Shelley, P.E., and Gaboury, D.R. 1986. Estimation of Pollution from Highway Runoff - Initial Results. In: Urban Runoff Quality: Proceedings of an Engineering foundation Conference, New England College, Henniker, New Hampshire, June 23-27. American Society of Civil Engineers, New York. Pp. 459-474. Sherwani, J.K., and Moreau, P.H. 1975. Strategies for Water Quality Monitoring. Report No. 107, Water Resources Research Institute, North Carolina State Simpson, D.E., and Stone, V .C . 1988. A Case Study of Urban Runoff Pollution: 1. Data Collection, Runoff quality and Loads Water SA. Vol . 14 No. 4. Pp. 229-237. Smith, J. A . 1994. Sediment Toxicity Testing: Battery Test Evaluation of Shallow Urban Streams and the Effect of Sampling Method on Toxicity. M.A.Sc. The University of B.C., British Columbia, Canada. 271 B I B L I O G R A P H Y (cont'd) Stahre, P., and Urbonas, B. 1990. Stormwater Detention for Drainage, Water Quality and CSO Management. Prentice-Hall, Inc., Englewood Cliffs, N.J. Stephan, C.E. 1977. Methods for Calculating an LC50. In: Aquatic Toxicology and Hazard Evaluation. Eds: F X . Mayer and J.L. Hamelink. American Society Testing and Materials, ASTMS TP 634, Philadelphia. Pp. 65-84. Stotz, 1987. Investigation of the Properties of the Surface Runoff from Federal Highways in Germany. The Science of the Total Environment 59, 329-337 Elsevier Science Publishers B.C., Amsterdam. Swain, L .G. 1983. Stormwater Monitoring of a Residential Catchment Basin, Vancouver, British Columbia. B.C. Ministry of Environment, Victoria, B.C. Swain, L .G . 1983a. Evaluation of Strom Drainage from Highway 97 to Okanagan Lake at Kelowna, B.C. Working Report, B.C. Ministry of Environment. Victoria, B.C. Swain, L . 1985. Stormwater Management - The Next Step? Can. Wat. Res. J. Vol . 10., Pp. 47-68. Swain, L .G. , and Walton, D.G. 1991. Quality of Stormwater/Urban Runoff. B.C. Ministry of Environment, Victoria, B.C. Paper Presented to B.C. W N A Conference in Vancouver, B.C. Pp. 1-15. Sylvester, R.O., and De Walle, F.B. 1972. Character and Significant of Highway Runoff Waters: A Preliminary Appraisal. Report to Washington State Highway Commission by Department of Civi l Engineering, University of Washington, Seattle, Washington. Tanacredi, J.T. 1977. Petroleum Hydrocarbons from Effluents: Detection in Marine Environment. Journal WPCF. Vol . 49. No. 2. Pp. 216-226. Taylor, J.S., Moore, P., Godlewski, V. , and Snyder, B.A. 1980. Coagulation and Detention of Stormwater Runoff. Annual Meeting of the American Society of Civi l Engineers, October 27 to 31, 1980. University of Central Florida, Civi l Engineering Department, Gainesville, Florida. Terstriep, M.L . , Noel, D . C , and Bender, G . M . 1986. Sources of Urban Pollution-Do We Know Enough? In: Urban Runoff Quality: Proceedings of an Engineering Foundation Conference, New England College, Henniker, New Hampshire, June 23-27. American Society of Civi l Engineers, New York. Pp. 107-121. Tourbier, J.T., and Westmacott, R. 1980. Stormwater Managment Alternatives. University of Delaware, Water Resources Center, Newark, D.E. 272 B I B L I O G R A P H Y (cont'd) Tucker, C , and Mortimer, G. 1978. The Generation of Suspended Solids Loads in Urban Stormwater. In Urban Storm Drainage, P. Helliwell, Ed. Pp. 695-704. Pentech Press, London. U.S. FHWA. 1985. Management Practices for Mitigation of Highway Stormwater Runoff Pollution. Volume 2, Pp. 86-226107. Federal Highway Administration, Office of Research and Development, Environment Division, Washington, D.C. U.S. FHWA. 1985a. Report on Management Practices for Mitigation of Highway Stormwater Runoff Pollution. Volume 1, Guidelines. Federal Highway Administration, Office of Research and Development, Environment Division, Washington, D.C. U.S. EPA. 1983. Final Report of the Nationwide Urban Runoff Program. Volume 1. Environmental Protection Agency, Washington, D.C. U.S. EPA. 1997. Monitoring Guidance for Determining the Effectiveness of Nonpoint Source Controls. Final Report # EPA/84 l-B-96-004. Office of Water, U.S. EPA, Washington, D.C. Unidata Australia. 1994. Unidata Starflow Ultrasonic Doppler Instrument Plus Version 3 Software. User's Manual. Lynn MacLaren Publishing, Australia. Van Buren, M.A . , Watt, W.E., and Marsalek, J. 1997. Removal of Selected Urban Stormwater Constituents by an On-Stream Pond. Journal of Environmental Planning & Management. Vol . 40. No. 1. Pp. 95-104. Waller, D.H., 1972. Factors that Influence Variations in the Composition of Urban Surface Runoff. Water Pollut. Res. Ca., Vol . 7. Pp. 84-95. Wang, T.S,. Spyridakis, E.D., Mar. B.W. and Horner, R.R. 1982. Transport, Deposition and Control of Heavy Metals in Highway Runoff, F H W A WA-RD-39.10 Report to Washington State Department of Transportation. Whipple, W. Jr., and Hunter, J.V. 1981. Settleability of Urban Runoff Pollution. Journal WPCF, Vol . 53. No. 12. Pp. 1726-1731. Wilbur, W.G., and Hunter, J.V. 1979. Distribution of Metals in Street Sweepings, Stormwater Solids, and Urban Aquatic Sediments. Journal WPCF. Vol . 51. No. 12. Pp. 2810-2821. Wilbur, W.G., and Hunter J.V. 1977. Aquatic Transport of Heay Metals in the Urban Environment. Water Res. Bull. Vol . 13 No. 4. Pp. 721-734. 273 B I B L I O G R A P H Y (cont'd) Wong, W . M . 1991. Stormwater Pollution Management Primer. Drainage, Hydrology and Remote Sensing Section. Surveys and Design office, Ontario Ministry of Transportation, Ontario, Canada. WSDOE (Washington State Department of Ecology). 1990. Stormwater Management Manual for the Puget Sound Basin. Technical Review Draft. June. Olympia, WA. WSDOE. 1992. Water Quality Standards-Surface Waters. Pp. 1-12. Wullschleger, R.E., Zanoni, A.E. , and Hansen, C A . 1976. Methodology for the Study of Urban Storm Generated Pollution and Control. U.S. Environmental Protection Agency. EPA-600/2-76-145. NTIS PB-258743. Yousef, Y , . A . , Wanielista, M.P., and Harper, H.H. 1985. Removal of Highway Contaminants by Roadside Swales. Transportation Research Record 1017. Transportation Research Board, Washington, D . C , Pp. 62-68. Yousef, Y . A . , Harper, H.H., Wiseman, L.P., and Bateman, J .M. 1985a. Consequential Species of Heavy Metals in Highway Runoff. Transportation Research Record 1017. Transportation Research Board, Washington, D . C , Pp. 56-62. Zimdahl, R.L. 1972. Impact on Man of Environmental Contamination Caused by Lead. In Interim Report, National Science Foundation Grant GI-4, H.W. Edwards, Ed., Colorado State University, Fort Collins. Pp. 98-110. 274 APPENDIX A 7.1 Summary of QA/QC Data 1.) Quality Assurance/Quality Control TableA-1 includes QA/QC results from field and laboratory blanks, replicates and spike recoveries as well as metal detection limits for both highway sites. " Table A - l . Summary of QA/QC Results for the Parameters Analyzed. Sampling & Analysis precision Spike Recovery Parameter Detection Limits N % N Min. (%) Max. (%) Ave. (%) Cr 0.05 ~ — 11 80 163 110.10 Cd 0.008 — 11 75 111 98.79 Ni 0.04 ~ — 11 78 108 97.60 Cu 0.01 6 6.80 11 90 115 98.30 Zn 0.004 11 4.40 11 78 111 94.00 Pb 0.10 1 2.30 11 88 151 115.00 Fe 0.02 8 . 7.00 6 13 99 77.50 M n 0.01 7 3.80 11 50 104 92.60 Ca 0.10 8 2.40 ~ — — — A l l metal units are in mg/L - Not analyzed/not available Sampling precision: N = # of paired grab samples analyzed % = average % difference of n pairs using Pi = (Yi - Xi) *100 [0.5(Yi + Xi)] where Pi = precision of duplicate pair i Y i = concentration of primary sample i X i = concentration of duplicate sample i 4) Spike recovery: max., min. and average based on 'n' spiked samples. Note: 1) 2) 3) 275 APPENDIX A Cont'd Table A-2. Metal Detection Limits for Sediment, Road Dirt and Grass Clipping Samples at Both Sites. Metal Paramters N Detection Limits (mg/kg) Cr 11 1.25 Cd 11 0.2 Ni 11 1 Cu 11 0.25 Zn 11 0.05 Pb 11 2.5 Fe 11 0.5 M n 11 0.25 Ca 11 2.5 276 APPENDIX A Cont'd Table A-3. Annual Stormwater Flow Estimates at Burnaby and Richmond Sites. Burnaby Richmond Runoff coefficient: Non-winter Winter 0.63 0.82 0.51 0.65 Runoff Volume: Non-winter (m3/yr) Winter (m3/yr) 505-558 (532) 1,170-1,290(1,230) 183-202(193) 414-458 (436) Total volume (m3/yr) 1760 296 Drainage area (m2) 1,420 (0.14 ha) 912(0.09 ha) Note: 1) Burnaby average precipitation: non-winter = 594 mm winter = 1,056 mm 2) Richmond average precipitation: non-winter = 414 mm winter = 736 mm 3) Runoff volume = (area(m2) )* (total annual rainfall(m/yr)) * (runoff coefficient) 4) Precipitation data from Hay and Oke (1976). 5) Numbers in bracket are average values. 277 APPENDIX B 7.2 T.S.S. Concentrations for Discrete and Composite Data Table B-l. Discrete and Flow Composite Concentrations of T.S.S. at Both Sites. Date Burnaby T.S.S. (mg/L) Richmond T.S.S. (mg/L) 11/04/95 80 (365) 60 01/06/96 227(227) 37(118) 02/07/96 140 71 02/17/96 120 40 (40) 02/19/96 202(202) 80 03/08/96 345 190 03/09/96 250 (250) 131(131) 03/31/96 68 N / A 04/01/96 292 (322) 112(154) 04/05/96 158 42 04/09/96 100 N / A 04/11/96 106 N / A 04/22/96 122 (122) 38 (38) 04/23/96 108 38 04/25/96 62 83 05/07/96 104 (216) N / A 05/17/96 50 107 05/18/96 109 26 06/10/96 146 N / A 07/02/96 139 N / A 07/17/96 147 26(108) 08/02/96 191(369) N / A Note: 1) N / A - N o t available. 2) Number in bracket is flow composite concentrations. 3) Bold dates are flow composite events. 278 APPENDIX C 7.3 Sample Calculations Using the Burnaby Site Data. l )F low for (11/04/95) h = 0.06 m Q = 1.38 * h 2 5 = 1.38 * 0.06 2 5 = 1.22x1 Q-°3 m3/s 2) Average Flow for (11/04/95) = 8.65x1004 m3/s 3) Runoff Volume for (11/04/95) = (Average Flow * Runoff Duration) = 8.65X10'04 m3/s * 10800 s = 9.34 m3 4) Uncertainty (Error) Analysis for Estimated Total Runoff Contributing Surface Area at 5% error = 1349-1491 m 2 Mean area (n) = 1420 m 2 Assume normal distribution: 3o = 71 .-.o = 71/3 =212 Coefficient of variation (CV) = o/n For the area: CV1 = 23.7/1420 = 0.02 For the rainfall: CV2 = 0.87 (from statistical analysis-Lotusl23) .-. Estimated Error = Square root ((CV1) 2 + CV2) 2) = Square root( (0.02)2 + (0.87)2) = 0.87 5) Est. Total Runoff Vol . (11/04/95) = (Contributing Area (m2) * Total Rainfall (m))* Est. Error Lower = (1349 m 2 * 0.01 m* 0.87) = lXJjn3. Upper = (1491 m 2 * 0.01 m * 0.87) = LL0_m! 6) Runoff Coefficient (RC) for (11/04/95) = Total Runoff Volume (m3) Catchment Area (m2) * Total Rainfall (m) * Est. Error Lower RC = 9.34 m 3 =0.72 1491 m 2 * 0.01 m * 0.87 Upper RC = 9.34 m 3 =0.80 1349 m 2 * 0.01 m * 0.87 7) Burnaby Seasonal Runoff Volume = Catchment Area (m) * Total Annual Rainfall (m) * RC Lower Winter = 1349 m 2 * 1.056 m * 0.82 = 1,170 m3 Upper Winter = 1491 m 2 * 1.056 m * 0.82 = 1,290 m3 Average = (1170 + 1290)/2 = 1,2.10 m3 Lower Non-Winter = 1349 m 2 * 0.594 m * 0.63 = 505 m 3 Upper Non-Winter = 1491 m 2 * .56946 m * 0.63 =55S_mi Average = (505 + 558)/2 = 532 m3 279 APPENDIX C Sample Calculations Using the Burnaby Site Data (cont'd), Burnaby Total Runoff Volume =(1230 + 532) m 3 = 1/760 m3 8) Burnaby Seasonal T.S.S. Loading (kg/yr) = Total Volume (m3/yr) * Appropriate Cone. (mg/L) 1000 Lower Winter = 1,170 m 3 * 7.61 mg/T. = 305 kg/yr 1,000 Upper Winter = 1,290 m 3 * 261 mg/T. = 337 kg/yr 1,000 Average = (305 + 337)/2 m 3 = 321 kg/yr Lower Non-Winter = 505 m 3 * 257 mg/L = 130 kg/yr 1,000 Upper Non-Winter = 558 m 3 * 257 mg/L = 143 kg/yr 1,000 Average = (130 + 143)/2 m 3 = 136 kg/yr Burnaby Total T.S.S. Loading (kg/yr) = (321 + 136) kg/yr = 458 kg/yr 9) Export Coefficient = Total Loading (kg/yr) = 458 kg/yr = 32.7x102 kg/ha/yr Catchment Area (ha) 0.14 ha 280 APPENDIX C 7.4 Unused Discrete Highway Stormwater Runoff Pollutograph Data for the Burnaby Site. Table C- l . Discrete Concentrations and Flow Data for Storm Event of 05-07-96 at the Burnaby Site. Sample T.S.S. Zn Mn Cu Fe Ca Flow T C H A 376 0.41 0.19 0.09 7.46 5.8 8.51 T C H A 331 0.36 0.18 0.09 7.8 4.18 9.35 TCH A 356 0.27 0.13 0.07 5.08 3.94 12.2 T C H A 124 0.18 0.08 0.05 3.84 3.98 4.5 T C H A 64 0.15 0.06 0.05 2.94 7.02 3.9 T C H A 74 0.17 0.09 0.06 3.93 7.86 4 T C H A 150 0.24 0.1 0.06 4.77 4.67 6 TCH A 90 0.19 0.1 0.05 4.42 4.76 4 T C H A 48 0.17 0.09 0.05 4.6 6.58 3 TCH B 28 0.1 0.28 0.03 2.64 11.8 s T C H B 23 0.09 0.27 0.02 1.97 12.6 s T C H B 73 0.16 0.12 0.05 3.44 8.14 s T C H B 64 0.12 0.09 0.03 2.64 6.32 s T C H B 33 0.08 0.12 0.02 2.04 7.26 s T C H B 17 0.08 0.16 0.02 1.67 8.68 s T C H B 41 0.07 0.2 0.02 1.43 9.48 s TCH B 44 0.1 0.09 0.05 2.18 7.08 s T C H B 40 0.1 0.11 0.03 2.2 7.56 s Note: 1) Pollutant concentrations in mg/L. 2) Flow = 10-4m3/s. 3) s—means same as corresponding T C H A values above. 4) T C H A - is composite sample taken right off the highway. 5) T C H B.- is composite sample taken after the runoff had gone through a 30 m grass ditch. 281 Table C-2. Discrete Concentrations and Flow Data for Storm Event of 06-10-96 at the Burnaby Site. Sample T.S.S. Zn Mn Cu Fe Ca Flow TCH A 240 0.7 0.38 0.21 11.5 22.5 11 T C H A 60 0.31 0.19 0.08 4.63 18.2 6.97 T C H A 40 0.3 0.16 0.07 3.39 11.4 6 T C H A 140 0.56 0.4 0.13 14.8 42.4 6.26 T C H A 130 0.47 0.35 0.13 4.5 21.8 8.92 T C H A 60 0.27 0.14 0.05 5.01 8.14 8.2 TCH A 50 0.27 0.15 0.04 4.51 8.91 5 T C H A 60 0.28 0.13 0.07 3.91 8.84 4 T C H A 30 0.22 0.13 0.04 3.99 10.4 4 T C H A 30 0.21 0.12 0.05 3.3 11.5 3 T C H B 80 0.18 0.13 0.08 3.71 18.4 s T C H B 36 0.15 0.07 0.05 2.3 14.4 s T C H B 16 0.13 0.05 0.04 1.27 14.6 s T C H B 16 0.11 0.04 0.04 0.85 16.9 s T C H B 24 0.14 0.05 0.05 1.28 15.1 s T C H B 20 0.12 0.05 0.04 1.25 11.6 s TCH B 8 0.11 0.04 0.04 0.96 11.9 s T C H B 12 0.11 0.03 0.03 0.78 12.9 s TCH B 24 0.11 0.03 0.03 0.91 12.9 s T C H B 16 0.09 0.04 0.04 0.68 13.2 s Note: 1) Pollutant concentrations in mg/L. 2) Flow= 10'4m3/s. 3) s—means same as corresponding TCH A Flow values above. 4) T C H A - is composite sample taken right off the highway. 5) T C H B.- is composite sample taken after the runoff had gone through a 30 m grass ditch. 282 Table C-3. Discrete Concentrations and Flow Data for Storm Event of 08-02-96 at the Burnaby Site. Sample T.S.S. Zn Mn Cu Fe Ca Flow T C H A 270 0.25 0.14 0.08 4.3 4.56 8.51 T C H A 160 0.23 0.11 0.07 4.6 4.28 7.7 T C H A 70 0.16 0.07 0.05 2.46 3.78 5.5 T C H A 120 0.19 0.08 0.05 3 4.14 6.97 T C H A 170 0.24 0.13 0.06 4.52 4.8 6.61 T C H A 150 0.16 0.09 0.05 2.62 3.62 6.26 T C H A 60 0.11 0.05 0.03 1.42 4.44 5.5 T C H A 40 0.13 0.08 0.06 1.98 6.2 4 T C H B 60 0.12 0.1 0.05 2.4 5.16 s T C H B 50 0.08 0.06 0.04 1.34 4.4 s T C H B 10 0.07 0.04 0.04 1.06 4 s T C H B 10 0.07 0.05 0.04 0.94 4.46 s T C H B 10 0.06 0.04 0.03 0.74 4.4 s T C H B 10 0.05 0.05 0.03 1.08 3.84 s T C H B 10 0.04 0.03 0.03 0.6 3.78 s T C H B 10 0.04 0.03 0.03 0.29 4.94 s Note: 1) Pollutant concentrations in mg/L. 2) Flow = 10"4m3/s. 3) s—means same as corresponding T C H A values above. 4) T C H A - is composite sample taken right off the highway. 5) T C H B.- is composite sample taken after the runoff had gone through a 30 m grass ditch. 283 INITIAL CONC. CY1) F L O W CY2) - C O N C . A F T E R 30m ( Y l ) Figure C - l . T.S.S. Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #16 (05-07-96) 3 o — o.ooos a INITIAL C O N C . CY1) F L O W CY2) C O N C . A F T E R 30m CY1) Figure C-2. T.S.S. Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #19 (06-10-96) 284 / Figure C-3. T.S.S. Concentration Pattern at the Trans-Canada Highway, Burnaby, Event #22 (08-02-96) 285 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0063736/manifest

Comment

Related Items