<%BANNER%>

Analysis of Particulate Matter Generated from Vehicle Traffic during and Prior to Rainfall Events

Permanent Link: http://ufdc.ufl.edu/UFE0041278/00001

Material Information

Title: Analysis of Particulate Matter Generated from Vehicle Traffic during and Prior to Rainfall Events
Physical Description: 1 online resource (58 p.)
Language: english
Creator: Grigsby, Melanie
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: matter, particulate, pollutant, runoff, storm, traffic, water
Environmental Engineering Sciences -- Dissertations, Academic -- UF
Genre: Environmental Engineering Sciences thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Particulate matter (PM) generated from vehicle traffic and roadways accumulate on the roadway surface and are washed off during a rainfall-runoff event. PM is a significant source of pollutant loadings to receiving waters since metals can be associated with the material transported by rainfall-runoff. The deposition of these PM from vehicles occurs prior to and during rainfall events. Conventionally PM that is generated from an event is attributed to the accumulation on urban surfaces during antecedent dry days, not particles generated during an event. The availability of PM on urban surfaces prior to rainfall events is an important factor when establishing a street sweeping program to mitigate stormwater pollution by determining frequency. This thesis investigated the effects of vehicles travelling prior to and during rainfall events to determine their impact on stormwater quality for runoff from a bridge over City Park Lake on Interstate 10 in Baton Rouge, Louisiana. Interstate 10 has an approximate average annual daily traffic (AADT) for east and westbound of 148,000. Stormwater samples examined for a Louisiana Department of Transportation (LaDOTD) study for twenty-seven rainfall-runoff events of varying duration, intensity and volume, with resulting PM from March 14, 2004 through August 9, 2006 were analyzed. The results of the rainfall-runoff events were compared with traffic data that was recorded by LaDOTD for this section of roadway from radar vehicle detectors from May 5, 2004 through August 31, 2006, which was used to estimate the traffic volume for all event storms. The results of the analysis indicated that the runoff volume of PM was significantly related to total runoff volume, previous dry hours and the density of vehicles passing during the event.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Melanie Grigsby.
Thesis: Thesis (M.E.)--University of Florida, 2010.
Local: Adviser: Sansalone, John.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041278:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041278/00001

Material Information

Title: Analysis of Particulate Matter Generated from Vehicle Traffic during and Prior to Rainfall Events
Physical Description: 1 online resource (58 p.)
Language: english
Creator: Grigsby, Melanie
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: matter, particulate, pollutant, runoff, storm, traffic, water
Environmental Engineering Sciences -- Dissertations, Academic -- UF
Genre: Environmental Engineering Sciences thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Particulate matter (PM) generated from vehicle traffic and roadways accumulate on the roadway surface and are washed off during a rainfall-runoff event. PM is a significant source of pollutant loadings to receiving waters since metals can be associated with the material transported by rainfall-runoff. The deposition of these PM from vehicles occurs prior to and during rainfall events. Conventionally PM that is generated from an event is attributed to the accumulation on urban surfaces during antecedent dry days, not particles generated during an event. The availability of PM on urban surfaces prior to rainfall events is an important factor when establishing a street sweeping program to mitigate stormwater pollution by determining frequency. This thesis investigated the effects of vehicles travelling prior to and during rainfall events to determine their impact on stormwater quality for runoff from a bridge over City Park Lake on Interstate 10 in Baton Rouge, Louisiana. Interstate 10 has an approximate average annual daily traffic (AADT) for east and westbound of 148,000. Stormwater samples examined for a Louisiana Department of Transportation (LaDOTD) study for twenty-seven rainfall-runoff events of varying duration, intensity and volume, with resulting PM from March 14, 2004 through August 9, 2006 were analyzed. The results of the rainfall-runoff events were compared with traffic data that was recorded by LaDOTD for this section of roadway from radar vehicle detectors from May 5, 2004 through August 31, 2006, which was used to estimate the traffic volume for all event storms. The results of the analysis indicated that the runoff volume of PM was significantly related to total runoff volume, previous dry hours and the density of vehicles passing during the event.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Melanie Grigsby.
Thesis: Thesis (M.E.)--University of Florida, 2010.
Local: Adviser: Sansalone, John.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041278:00001


This item has the following downloads:


Full Text

PAGE 1

1 ANALYSIS OF PARTICULATE MATTER GENERATED FROM VEHICLE TRAFFIC DURING AND PRIOR TO RAINFALL EVENTS By MELANIE R. GRIGSBY A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE RE QUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Melanie R. Grigsby

PAGE 3

3 ACKNOWLEDGMENTS I would like to sincerely thank Dr. John Sansalone, Dr. Chester Wilmot and Dr. Ben Koo pman for helping me throughout the analysis and the writing process of this thesis. I would like to especially thank Dr. Chester Wilmot for his continuous encouragement, professional feedback, and suggestions that greatly improved the quality of my researc h.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................3 LIST OF TABLES ...........................................................................................................................6 LIST OF FIGURES .........................................................................................................................7 ABSTRACT .....................................................................................................................................7 CHAPTER 1 INTRODUCTION .............................................................................................................10 Problem Statement .............................................................................................................12 Purpose ...............................................................................................................................13 Scope ..................................................................................................................................14 2 BACKGROUND ...............................................................................................................15 National Pollution Discharge Elimination System ............................................................15 Impaired Water Body Identification 303(d) ..........................................................16 Total Maximum Daily Loads .................................................................................16 Effects of Urbanization on Stormwater .............................................................................16 Traffic Generated Constituents in Stormwater ......................................................17 Typical Stormwater Contaminant Concentrations .................................................19 Best Management Practices for Roadways ........................................................................21 Street Sweepers ......................................................................................................21 Previous Studies .....................................................................................................23 3 OBSERVATION SITE ......................................................................................................24 Amite Watershed ...............................................................................................................24 Characteristics ........................................................................................................24 Typical Contaminant Concentrations for Observation Site ...................................27 Regulatory Requirements ...................................................................................................27 Acute Discharge Limits .........................................................................................29 Monitoring Frequency ...........................................................................................31 Sample Collection ..................................................................................................31 4 METHODOLOGY ............................................................................................................33 Traffic Data ........................................................................................................................33 Background ............................................................................................................33 Analysis of Data .....................................................................................................36 Multiple Regression Model of Total Daily Traffic ................................................36 Regression Model of Cumulative Traffic Volume ................................................40 Rainfall Data ......................................................................................................................42 Water Chemistry Analysis .................................................................................................43

PAGE 5

5 5 DISCUSSION ....................................................................................................................50 6 CONCLUSIONS ................................................................................................................52 APPENDIX A GRAPHS OF LOGISTIC REGRESSION FOR CUMULATIVE DAILY TRAFFIC ......53 LIST OF REFERENCES ...............................................................................................................56 BIOGRAPHICAL SKETCH .........................................................................................................58

PAGE 6

6 LIST OF TABLES Table P age 21 Traffic generated stormwater contaminates sources ..........................................................19 22 Summary of Caltrans statewide highway stormwater runoff characteristics compared to historical data .............................................................20 23 Summary of Caltrans statewide highway stormwater runoff characteristics compared to domestic wastewater characteristics .....................................21 24 Removal rates for stre et cleaning various particles ...........................................................22 31 Leading sources of water quality impairment for Amite watershed ..................................26 32 Leading pollutants of surface water for Amite watershed .................................................26 33 Beneficial use most frequently impaired for Amite watershed .........................................26 34 Summary of Caltran s statewide highway stormwater runoff characteristics compared to observation site in Baton Rouge, LA ....................................27 35 Nonpoint source discharge limits for obs ervation site .......................................................29 41 Results of variables from multiple regression analysis ......................................................39 42 Groups according to the associated variables ....................................................................41 43 Summary of logistic regression with resulting linear equation .........................................42 44 Summary of hydrologic, sampling based indices and traffic data for 27 events analyzed for the I 10 observation site ..................................................44 45 Summary of multiple regression analysis for nine scenarios based on twentys even observed rainfall runoff events. ..............................................................48

PAGE 7

7 LIST OF FIGURES Figure P age 21 Traffic Generated Stormwater Contaminates Sources .......................................................18 41 Location of traffic monitors RVD 43/45 (links 80043, 800045) over observation site ......................................................................35 42 Total daily traffic as observed by RVD 43/45 ...................................................................35 43 Observed total daily traffic with respect to the associated variable. ..................................38 44 Observed t otal daily traffic with respect to corresponding dummy variable, or degree of variation from the base. ..................................................................38 45 Observed and predicted total daily traffic from multiple regression analysis and monitors .......................................................................................39 46 Average rainfall intensity, total runoff volume, previous dry hours (PDH), vehicles prior to storm (VPS) and vehicles during the storm (VDS) graphed against s uspended sediment concentration, PM, for twentyseven rainfall runoff events in Baton Rouge, Louisiana. ....................................................................................................45 47 Vehicles during storm as density of time and area graphed against suspended sedim ent concentration, PM, for twentyseven rainfall runoff events in Baton Rouge, Louisiana. ...........................................................................................................................46 48 Multiple regression model equation for five scenarios against measured PM for twenty seven rain fall runoff events in Baton Rouge, Louisiana. .......................................48 A 1 Graph of cumulative daily traffic volume and s curve from logistic regression (groups 15a) ......................................................................................................................53 A 2 Graph of cumulative daily traffic volume and s curve from logistic regression (groups 613) ......................................................................................................................54 A 3 Graph of cumulative daily traffic volume and s curve from logistic regression (groups 1415) ....................................................................................................................55

PAGE 8

8 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering ANALYSIS OF P ARTICULATE MATTER GENERATED FROM VEHICLE TRAFFIC DURING AND PRIOR TO RAINFALL EVENTS By Melanie R. Grigsby May 2010 Chair: John J. Sansalone Major: Environmental Engineering Science s P articulate matter (PM) generated from vehicle traffic and roadways accumulate on the roadway surf ace and are washed off during a rainfall runoff event P M is a significant source of pollutant loadings to receiving waters since metals can be associated with the material transported by rainfall runoff. The deposition of these PM from vehicles occurs pri or to and during rainfall events. Conventionally PM that is generated from an event is attributed to the accumulation on urban surfaces during antecedent dry days, not particles generated during an event. The availability of PM on urban surfaces prior to r ainfall events is an important factor when establishing a street sweeping program to mitigate stormwater pollution by determining frequency. This thesis investigated the effects of vehicles travelling prior to and during rainfall events to determine their impact on stormwater quality for runoff from a bridge over City Park Lake on Interstate 10 in Baton Rouge, Louisiana. Interstate 10 has an approximate average annual daily traffic (AADT) for east and westbound of 148,000. Stormwater samples examined for a Louisiana Department of Transportation (LaDOTD) study for twenty seven rainfall runoff events of varying duration, intensity and volume, with resulting PM from March 14 2004 through August 9, 2006 were analyzed The results of the rainfallrunoff events were compared with traffic data that was recorded by L aDOTD for this section of roadway from radar vehicle

PAGE 9

9 detectors from May 5, 2004 through August 31, 2006, which was used to estimate the traffic volume for all event storms. The results of the analysis i ndicated that t he runoff volume of PM was significantly related to total runoff volume, previous dry hours and the density of vehicles passing during the event.

PAGE 10

10 CHAPTER 1 I NTRODUCTION Urban stormwater from highways transports PM and metals that collect on the surface prior to and during rainfall events. These heavy metals, principally Pb, Cd, Cu, Ni, Zn, and Cr, can pose acute and chronic threats to receiving water bodies and soils. In receiving water, the dissolved fraction of these heavy metals has the potential for acute and longterm chr onic toxicity for aquatic life (Glen et al. 2002) These contaminants have been deposited on roadways by atmospheric deposition, vehicular traffic, litter debris, pavement deterioration, exhaust, tire wear and engine w ear (Sansalone 2003) As the rain falls on these surfaces, runoff from these areas enter into a separate storm sewer and carries all the pollutants that were on the surface at the time of the event and deposits them into the nearby water bodies. Collection systems have historically been designed to collect deposit the water into a nearby water body without any type of treatment occurring. Until the 1970s this standard practice was considered acceptable and the impacts to surface waters did not yield any concern. As metropolitan areas continually develop they are increasingly impacting the hydrological cycle of the area they occupy by reducing time of concentration, infiltration, evaporation and increasing runoff and velocity. As the volume runoff increases so do the pollutants they carry, which results in an increase of adverse effects to the environment from the continual exposure. As resources are being stretched further to met the demand of increasing population, water quality and abundance is just as muc h a priority for developing areas as economic growth. Since hydrologic events have varying characteristics as do the traffic that deposits the contaminants, modeling pollutant concentrations from events poses many challenges. Although modeling pollutant concentrations and identifying an appropriate best management plan to

PAGE 11

11 mitigate the pollution is difficult, the issue needs to be addressed to slow or stop the continual degradation of the quality of the natural environment that these metropolitan areas surround. Environmental regulatory agencies are setting limitations on the volume of pollutants that are deposited into surface waters to improve water quality. In the 1960s and 1970s the United States Congress passed a series of laws to protect the environment for pollution and prevent the degradation of natural resources as the result of actions by environmental lobbyists. The laws defined the governments responsibility as it related to the quality of the environment. The first law that was passed to protect waters of the U.S. occurred in 1972 with the passage of an amendment to the Federal Water Pollution Control Act, also known as the Clean Water Act. The act regulated pollution discharges from point sources to waters and established the statutory basis fo r the National Pollution Discharge Elimination System (NPDES), which is a system that requires point sources to obtain a permit from the Environmental Protection Agency (EPA) prior to discharging to regulated waters. The act defined point sources act as in dustrial facilities, municipal government and some agricultural facilities. The goal of the act was to eliminate additional water pollution by 1985, eliminate releases to waters that contained high amounts of toxic substances and all surface waters would m eet standards necessary for human sports and recr eation by 1983 (United States Senate 2002) In response to the need for a more comprehensive act that regulates all pollution sources to regulated waters, the Water Quality Act of 1987 was passed. The act addressed nonpoint pollution sources, which were defined as sources that could not be tracked back to a single origin or source. With the passing of this act, industrial stormwater dischargers and municipal separate storm sewer systems, referred to as MS4s, were required to obtain NPDES permits by specific deadlines.

PAGE 12

12 Although entities associated with non point source discharges were required to obtain an NPDES permit, the original goal of the Clean Water Act of 1972 was not attained for all surface waters meeting a recreational service level by 1983. In efforts to comply with this requirement, the EPA placed water bodies that do not meet applicable water quality standards with technology based controls on the section 303(d) list of water bodies not meeting st andards. Water bodies on this list require the development of a Total Maximum Daily Load (TMDL), which is the maximum pollutant concentration that any source can discharge to the water body at any given time. TMDLs are established by undergoing extensive m onitoring of the water body to establish a baseline of the current condition, monitoring of similar unimpaired water bodies to establish a background limit and modeling pollutant limitations so that the background limit is not exceed outside of the recover able range. When the EPA sets a TMDL on a regulated water body, all permitted entities that discharge to the water body through point and/or non point sources must develop a basin management action plan (BMAP), point source pollution is regulated separatel y. A BMAP is a plan of action that the permittees shall execute to reduce pollution that is discharged to a regulated water body. Items proposed in the BMAP to address water pollution would be either structural (retention/detention of stormwater, treatment facilities, and utility improvement projects) or nonstructural controls (removal of sedimentation, education, enforcement, illicit discharge identification and elimination). Problem Statement Stormwater runoff from roadways is classified as non point source pollution and operators are required to obtain a permit under the NPDES program. Curr ently, the EPA recognizes that streets, roads, highways and parking lots accumulate significant amounts of pollutants that contribute to stormwater pollutant runoff to surface waters ( EPA 1997). As a method of controlling nonpoint so urce pollution that is generated by vehicles, the EPA encourages street

PAGE 13

13 sweeping activities on a regular basis for best management practices (BMP). Street sweeping is the most economical ch oice for older highways where no other structural BMPs, such as filters or retention/detention systems exist. Retro fitting existing drainage systems to incorporate stormwater treatment is extremely expensive and in some cases impossible in urban environme nts due to space limitations. An effective street sweeping program can improve water quality and allow for the operator to meet the requirements of the TMDLs that have been set for regulated waters. If inefficiencies exist with equipment, frequency or time with respect to rainfall events, street sweeping can be ineffective in addressing water quality problems. Some previous studies on this topic have determined that there is not a strong correlation between particle buildup on roadway surfaces and previous dry days, but these studies did not utilize real time data to accurately conclude this factor. This link is critical in setting up an effective street sweeping program by allowing the operator to estimate what sweeping frequency or timing is needed to remove the particles from the surface. If particles that impair stormwater runoff continually build up on the roadway surface until they are washed away by the next rainfall event, then the operator would need to focus on sweeping the roadways prior to all eve nts. If there is a limited amount of particle buildup on the roadway surface prior to each event regardless of the previous dry days, then the operator would need to focus on sweeping more frequently. Purpose The effectiveness of a street sweeping program is dependent on the assumption that the particles collecting on the roadway surface from vehicles during antecedent dry days are a significant factor in particulate matter generated from this source. If this statement is correct, the number of vehicles tha t travel over a roadway during dry days should be directly related to particulate matter concentrations in the rainfall runoff. The intent of this paper is to determine if

PAGE 14

14 traffic volume during the event and during dry days affect the resulting particulate matter from hydrologic events. Scope The scope of this project involves analyzing data that was obtained from an observation site in Baton Rouge, Louisiana at an elevated section of urban highway on Interstate 10 that crosses City Park Lake. The data tha t shall be analyzed is: traffic volume and speed, hydrologic events (duration, intensity, date/tim e and volume), and water chemistry data (particulate matter). The observation period is March 14, 2004 to August 9, 2006, which contained 27 observed rainfall runoff events with recorded traffic data.

PAGE 15

15 CHAPTER 2 BACKGROUND The passage of the Water Quality Act of 1987 changed the method government entities regulated and managed stormwater, development and industrial industries. The government was finally accou ntable for the degradation of the nations waters and required to increase regulation to prevent further degradation and restore impaired waters. The EPA regulated governing bodies to become compliant with the original requirements of the Clean Water Act through the NPDES program and establishment of TMDLs. After the establishment of TMDLs, the complicated process of determining an appropriate BMAP to comply with the requirements has challenged most permittees. National Pollution Discharge Elimination Syst em This program was developed to require a phased NPDES permit for water discharges from point and nonpoint pollution sources, which is administered by the EPA. There are two phases through this program for discharges through MS4 discharges located in municipalities with a population of 100,000 or more, Phase I, or with a population of less than 100,000, Phase II. The permits typically last for five years, with either phase reporting to the EPA on an annual basis what actions have taken place to prevent or eliminate pollution to surface waters. The following actions are classified as preventive or rehabilitative measures by the EPA to improve water quality: removal of sediment from stormwater management facilities (ditch cleaning, cleaning stormwater pipes and dredging waterbodies), street sweeping and cleaning, litter collection, education, enforcement, permits for applicable construction and industrial activities, and inspections.

PAGE 16

16 Impaired Water Body Identification 303(d) Section 303(d) of the Clean Wat er Act requires that states develop a list of deficient waters that fail to meet the water quality standards and provide a priority ranking. A state 303(d) list identifies the basic information regarding the impaired waterbody and the observed impairment, usually including the waterbody characteristics (e.g., name, location, size), the water quality standard that was violated, the pollutant of concern (if known), and the suspected causes and sources contributing to the impairment. It is usually necessary to analyze available monitoring data to further characterize and understand the impairments ( EPA 2008). Th e state must submit updates to the 303(d) list every two years along with a priority ranking for the establishment of TMDLS within the next two year lis ting cycle. After a water body has received the 303(d) listing status, the state must identify the pollutant causing the impairment and the potential sources. Total Maximum Daily Loads Total maximum daily loads (TMDLs) are established through extensive mo nitoring of the characteristics of the impaired water body and unimpaired waters within the same region with similar recreation requirements. TMDLs that are set on a water body are intended to limit the pollutant loading that the system can take to recover to the required service level with some degree of safety. TMDLs are typically concentration based limits and do not account for buildup of pollution within the water body. Effects of Urbanization on Stormwater Prior to urbanization, fifty percent of all rainfall that fell on land was infiltrated into the subsurface and used to recharge underground aquifers, while only ten percent flowed over the lands surface and was deposited into surface waters. After urbanization, the opposite occurs with only fifteen percent of the rainfall infiltrating the lands surface and fifty five percent being deposited into surface waters. The impacts of urbanization upon peak stream discharges are even

PAGE 17

17 more pronounced in areas serviced by storm sewers since the rapid conveyan ce of storm waters to nearby streams dramatically increases mean annual floods. The development of urbanized areas also impacts water quality by increasing: nutrients, pathogens, sediment, toxic contaminants, debris and thermal stresses in the surface waters. With these changes to the hydrologic characteristics also come changes to quality of stormwater. The reduction in detention time results in more contaminants being carried by the storm water and not settling back onto the lands surface and the increas ed likelihood that they will be deposited into water bodies. The decrease to the infiltration and evaporation rates results in a larger volume of surface runoff, increasing the particulate matter size that can be transported by the event. During a weather event, particles are rapidly washed from the surface of the roadway and transported with the runoff. Th is shall result in an increase in contaminant concentration in the runoff that is being deposited into the water bodies and increase to the risk of adver se effects from the aquatic organisms that are exposed to the storm water. Traffic Generated Constituents in Stormwater Sources of potential pollutants from traffic generated stormwater are from the fuel system, body of the vehicle, engine, brakes, exhaust tires and the pavement (Sansalone 2003). T he contaminants of each source have been shown in Table 21 and the proportion of particles generated by sources is shown in Figure 21. The sources of greatest concern for water quality is the body/frame, engine brakes, tires and pavement since these are the contributors of the contaminants of greatest concern for adverse health effects to humans: zinc, cadmium, lead and copper. The contaminants lead and cadmium are listed as carcinogens with the Environmental P rotection Agency. Cadmium can accumulate in the kidneys, liver and bloodstream and can potentially cause effects on the kidneys, liver, lung bone, immune system, blood and nervous system. Prolonged

PAGE 18

18 oral inhalation of lead can effect the neurological, physi cal and reproductive development and also cause bone deterioration in humans. Exposure to copper can cause hemolytic anemia as well as liver and kidney damage. Zinc has been known to decrease higher density lipoprotein cholesterol and upset the bodys natu ral copper balance. Stormwater is composed of many more pollutants than are listed here. Lead, Copper, Cadmium and Zinc will be a focal point due to their potential health effects on humans and the environment as well as the high concentrations that are fo und in stormwater and pose the greatest risk. Although pavement is not a source of metals contaminants, the thermal pollution is significant enough to hinder the growth and development of the aquatic environment that the stormwater is deposited into. Tempe rature data from North America indicates that there is a 2 degrees Celsius temperature rise in surface waters for every pop ulation increa se of 10^3 (Sansalone 2003). Figure 21. Traffic Generated Stormwater Contaminates Sources

PAGE 19

19 Table 21. T raffic generated s tormwater c ontaminates s ources Source Contaminants Fuel Systems Volatile Organic Compounds, Petroleum Body/Frame Zinc, Chromium, Iron, Aluminum Engine Zinc, Copper, Chromium, Oil/grease, Manganese Brakes Copper, Lead Exhaust Particulates Ties Zinc, Cadmium, Solids Pavement Solids, Polycyclic Aromatic Hydrocarbons, Phenols, Thermal Typical Stormwater Contaminant Concentrations It has been observed that it is difficult to quantify generalized concentrations of each contaminant for every storm event, since the concentrations are dependant upon number of days between storms, traffic loadings, vehicle characteristics, volume of rainfall, duration of rainfall, geographical location and features, etc. A study has been completed by the Californ ia Department of Transportation to quantify the average concentrations of the major contaminants that is found in stormwater, the results are shown in Table 2 2 (CalTrans 2002) Some of the observed parameter averages for stormwater were compared with the averages for domestic wastewater according to Metcalf and Eddy and shown in T able 2 3 (Metcalf and Eddy 2003; CalTrans 2002) Stormwater averages for total solids are the same as that of domestic stormwater, but the concentrations of lead and zinc are higher, as would be expected considering the contributing sources. The observations from the CalTran sites may not be able to be applied to other locations throughout the nation. Contaminant concentrations are a factor of the geographical area such as: types of industries that surround the area of concern, land usage, population density, highway speed, average daily traffic, average annual rainfall, vehicle condition, local regulations concerning litter and vehicle maintenance, etc. Although CalTran had numerous observation sites throughout the state, thirty one, it would be difficult to apply these observations to a

PAGE 20

20 location in Louisiana. CalTran even concluded from there study that all the observations indicated that the concentrations were lower for the 20002001 observations than the 20001997 observations and the data demonstrated no trend except that concentrations will be higher in areas of acceleration/deceleration on the highway. Table 22. Summary of Caltrans statewide h ighway s tormwater r unoff c haract eristics compared to historical da ta Parameter Reporting Limit Unit Statewide Monitoring (2000 01) Min Max Mean Median CV Conventionals pH +/ 0.1 pH units 5.1 10.1 7.2 7.2 0.1 Conductivity +/ 1.0 Umhos/cm 7.0 1285 95.8 65 1.3 TSS TDS 1.0 1.0 mg/L mg/L 2.0 5.0 1373 724 94.4 84.8 55 57 1.8 1.1 Hardness 1.0 mg/L 3.0 400 36.8 26 1.1 DOC 1.0 mg/L 1.3 155 14.7 9.8 1.2 TOC 1.0 mg/L 1.4 137 17.7 13 1.0 Nutrients Nitrate as N TKN Total Phosphorus Orthophosphate Total Metals Arsenic Cadmium Chromium Copper Lead Nickel Zinc Dissolved Metals Arsenic Cadmium Chromium Copper Lead Nickel Zinc 0.1 0.1 0.03 0.03 1.0 0.2 1.0 1.0 1.0 2.0 5.0 1.0 0.2 1.0 1.0 1.0 2.0 5.0 mg/L mg/L mg/L mg/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L ug/L 0.1 0.1 0.03 0.04 0.5 0.2 1.0 1.2 1.0 2.0 7.5 0.6 0.2 1.0 1.1 1.0 1.1 3.0 48 14.5 4.7 2.3 8.6 5.0 98 230 327 208 1245 4.8 4.7 19 121 143 52 1017 1.2 1.8 0.3 0.2 1.4 0.7 7.8 22.3 21.9 10.9 129.8 0.9 0.4 2.6 11.4 3.2 4.4 59.4 NA 1.4 NA NA NA NA 5.0 16.8 6.1 6.9 81 0.8 0.4 1.5 8.5 1.1 2.9 28.0 3.5 1.0 1.8 1.3 0.9 0.9 1.6 1.2 2.0 1.8 1.3 0.7 0.9 1.2 1.1 4.0 1.3 2.0

PAGE 21

21 Table 23. Summar y of Caltrans statewide highway s tormwater r unoff characteristics compared to domestic wastewater characteristics Parameter Units Domestic Wastewater Urban storm water Mean Range Mean Range Solids, Total (TS) mg/L 720 350 1200 750 150 22000 D issolved, Total (TDS) mg/L 500 250 850 84.8 5 724 D issolved, Volatile (VDS) mg/L 300 145 525 100 50 1000 Suspended, Total (TSS) mg/L 220 100 400 94.4 2 1373 Suspended Volatile (VSS) mg/L 165 80 275 75 30 800 TOC mg/L 160 8 0 290 17.7 1.4 137 Nitrogen (Total as N) mg/L 40 20 85 1.2 0.1 48 Total Phosphorus (as P) mg/L 8 4 15 0.3 0.03 4.7 Total Lead ug/L 10 2 15 21.9 1.0 327 Total Cadmium ug/L 1 N/D 3 0.7 0.2 5.0 Total Zinc ug/L 75 40 120 129.8 7.5 1245 Total Copper ug/L 35 20 45 22.3 1.2 230 Best Management Practices for Roadways Best management practices (BMPs) to reduce pollutant loadings to surface waters are a necessity to meet total maximum daily loading requirements. Structural and nonstructural BMPs are effective in meeting TMDL requirements if properly applied. Examples of structural BMPs for urban roadways are: filtration, retention, detention and sedimentation. Examples of nonstructural BMPs for urban roadways are: litter and debris removal, education and training, landscaping and vegetation practices, pollution prevention and identification and sediment and erosion control. Nonstructural BMPs are more cost effective to incorporate into older areas that were developed prior to rec ent regulatory actions. Street sweeping is typically the BMP of choice to collect particles that are deposited on the roadway surfaces. Street Sweepers There are four different sweeper types that are currently used throughout the United States: mechanical regenerative air, vacuum filter and tandem sweeping. Mechanical typically uses a broom type sweeper to pick up debris, are the least expensive and make up about 90% of sweepers currently in service. Regenerative Air is a type of sweeper that blows air on to the road surface causing fine particles and sediments within pavement crevices to rise and then vacuums

PAGE 22

22 them up. Vacuum filter is a small micron particle sweeper, either wet or dry, that combines mechanical broom sweeping with a vacuum to capture small particles. Tandem sweeping are two machines used in process with the first pass by a mechanical sweeper followed by a second pass with a vacuum machine. The vacuum assist sweeper shows notably higher percent reduction for both total suspended solids and n itrogen than the other types. Monthly sweeping can decrease pollutants reductions to 60% of the weekly sweeping. For major roadways, vacuum assisted sweepers were able to reduce total suspended solids by 79% and nitrogen by 53%. The removal efficiencies for residential roadways are very similar if no obstructions such as parked cars or garbage cans exist ( C WP 2008). Particles less than 100 um contribute most significantly to stormwater pollution. Typical street sweepers are limited in removing particles of this size, as they are most efficient in respect to courser material, as shown in T able 24 (Pitt et al. 2004) Factors that affect the performance of street sweeping include: frequency, particle loadings, street texture, moisture, parked car conditions, a nd equipment operating conditions. Sweepers that utilize a vacuum mechanism are more efficient at lower loadings, then when particles are abundant. As moisture increases, the removal rates of the smaller particles increases. Table 24. Removal r ates for s treet cleaning various particles Particle Size (um) Removal Efficiency (%) 0 40 16 40 100 0 100 250 48 250 850 60 850 2000 67 > 2000 79

PAGE 23

23 Previous Studies One of the first studies that were conducted on the subject matter was published by the American Public Works Association in 1969. The study measured the volume of particles on a roadway surface with varying watersheds characteristics, traffic loadings and geometries and analyzed different areas based on the volume collected, land uses and previous dry days. It was concluded that the volume of collected particles on the roadway surfaces was linearly related to time. (APWA 1969) This study was conducted at the same time the EPA was developing the Stormwater Management Model (SWMM) and the find ings of this research was incorporated into the model. In 1983 the U.S. Geological Survey released Report 834153, Water Quality Assessment of Stormwater Runoff from a Heavily Used Urban Highway Bridge in Miami, Florida, which concluded that the number of dry days and traffic volume did not indicate a strong relation to pollutant concentrations or loads for rainfall events. The observation site consisted of a 1.43 acre watershed that contained a bridge on I 95 which had 70,000 vehicles per day. All of the observed events were a short duration (less than 30 minutes) with less than 0.5 of total rainfall. The pollutants that were analyzed during this study included turbidity and solids (total and dissolved). Since these studies, new theories have been intro duced to recognize that PM can not continually accumulate on the roadway surface, as the build up is limited due to vehicle induced winds. The most recent theory states that pollutant buildup outside of a rainfall event occurs when the surface is still wet and traffic travels over the area. When the pavement dries, the accumulated particles are transported by vehicle induced winds, volatilization, biodegradation and chemical decay until it reaches a steady state (Li et al. 2008).

PAGE 24

24 CHAPTER 3 OBSERVATION SI TE The observation site is located in Baton Rouge, Louisiana on Interstate 10 on an elevated section of highway that crosses over City Park Lake. The catchment area consist ed of two 544 m2 well defined hydraulically parallel concrete paved catchments. The bridge contains curb and gutter for east and westbound traffic (Kim et al. 2008) The average traffic on this section of roadway is 74,000 average vehicles per day per direction, with the average speed of 58 miles per hour. The interstate, which is managed by LaDOTD, does not sweep this section of roadway on a regular basis. Records obtained from LaDOTD indicated that sweeping activities were completed as follows: eastbound bridge on 06/28/05 and westbound bridge on 10/26/04 and 11/20/05. Amite Watershed T he observation site is located within the Amite Watershed that is monitored and regulated by Louisiana Department of Environmental Quality (LDEQ). An NPDES permit was issued for nonpoint source discharge by LDEQ under Permit No. LAS000101, and has associated TMDL limitations within the permit. Characteristics In order to quantify the environmental impacts of stormwater pollution, the watershed for Baton Rouge, Louisiana was analyzed. The watershed for this area is the Amite Watershed which includes the following parishes: Amite, Ascension, East Baton Rouge, East Feliciana, Iberville, Livingston, West Baton Rouge, Wilkinson, Lincoln, St. Helena and Franklin. There are nine water bodies listed by the Environmental Protection Agency, EPA, for this watershed: Amite River (Upper and lower), Bayou Manchac, Comite Creek, Comite River, East Fork Amite River, Little Beaver Creek, Mississippi River and West Fork Amite River. Table 3 1 ( LaDEQ 2007) lists the leading sources of water quality impairment by the EPA in th is area. It is shown

PAGE 25

25 that urban runoff alone accounts for the impairment of 33% of surface water in this area, and ranks among the top three pollution contributors. Stormwater runoff is a contributing factor to: municipal point sources, construction and c ombined sewer overflows. Comparing with national averages of watersheds, the Amite watershed is ranked by the EPA to be in the dirtiest/worst watershed category. Urban stormwater is a contributor for: sediment, low dissolved oxygen, metals, and pH. For all these impairments, the Louisiana Departm ent of Environmental Quality, La DEQ, has determined that only five water bodies have a regulatory priority of low, meaning that remediation of these contaminated waters will not receive remediation unless there prio rity level is increased. Table 3 2 shows that metals account for 56% percent of all impairments to surface waters in this watershed ( LaDEQ 2007). The EPA has stated that 43% of water bodies within this watershed are impaired. Table 33 shows the beneficial uses of the water bodies within the watershed that are impaired as a result of pollution (EPA 2007) The use of concern for human consumption is shellfish (including fish) and drinking water. These are the two exposure pathways in which human oral consumption of these contaminants can occur. Aquatic and terrestrial habitats can be impaired from the exposure to urban runoff contaminants by means of degradation, loss and fragmentation, which occur as a direct result of the increase of impervious surfaces. Ha bitat degradation is the diminishment of habitat quality and its ability to support biological communities. Habitat loss is the result of the destruction of habitats by altering the watershed such as: filling in wetlands, clearing the channel of a stream, removal of trees/vegetation within an aquatic environment. The effects can be seen immediately in the aquatic / terrestrial organisms populations. Habitat fragmentation is the result of development in which a natural habitat is broken up and

PAGE 26

26 some areas ar e left isolated. Aquatic and bay habitats are adversely impacted by the nonpoint pollutants and the higher volumes of runoff issuing from urbanized lands. Observable declines in the biological integrity of streams and the quality of stream habitats occur w hen watershed imperviousness reaches 10 15% (Chesapeake Bay Program 2003) Table 31. Leading s ources of w ater quality i mpairment for A mite w atershed Ranking Pollution Source Percent of All Impairments 1 U nknown Sources 56% 2 Municipal Point Sources 44% 3 Agricultural, Industrial Point Sources, Urban Runoff/Storm Sewers 33% 4 Construction, Hydromodification/Habitat Modification, Land Disposal, Resource Extraction 22% 5 Combined Sewer Overflows, Other So urces, Silviculture 11% Table 32. Leading pollutants of surface w ater for Amite w atershed Ranking Pollutant Percent of All Impairments 1 Pathogens 89% 2 Nutrients, Sediments 78% 3 Low Dissolved Oxygen 67% 4 Mercury, Metals, Pesticides 56% 5 Aesthe tics, Flow Alterations, Organic Compounds, Other Habitat Alterations 22% 6 Ammonia, pH 11% Table 3 3. Beneficial use most f requently i mpaired for Amite w atershed Ranking Beneficial Use Percent of All Impairments 1 Aquatic Life Support 100% 2 Secondary Contract Recreation (Boating) 89% 3 Primary Contact Recreation (Swimming) 67% 4 Aesthetic/Scenic, Wildlife Support, Shellfish Consumption, Agriculture 56% 5 Drinking Water Supply 44%

PAGE 27

27 Typical Contaminant Concentrations for Observation Site The observation site is at the outflow to a separate storm sewer pipe that drains multiple catch basins on the bridge that crosses City Park Lake near Louisiana State University. This location has an average of 1 48,000 vehicles per day with an average operating spee d of 57 miles per hour. After comparing the data, the solids concentrations seem to fall within the range of the CalTran observations which is shown in T able 3 4, but the contaminants that pose a greater environmental risk and are more heavily regulated: L ead, Cadmium, Zinc and Copper, are well above the CalTran average, but the averages do fall within the CalTran ranges (CalTran 2002) Due to the variations in these observations, it would be too difficult to apply an average of typical concentrations for s tormwater. Table 34. Summary of Caltrans statewide highway stormwater r unoff c haracteristics c ompared to observation s ite in Baton Rouge, LA Parameter Units Baton Rouge, LA CalTran Observations Mean Range Mean Range Solids, Total (TS) mg/L 482.9 750 150 22000 Dissolved, Total (TDS) mg/L 100.42 84.8 5 724 Suspended, Total (TSS) mg/L 78.54 94.4 2 1373 Suspended Volatile (VSS) mg/L 24.91 75 30 800 TOC mg/L 99 66 138 17.7 1.4 137 Nitrogen (Total as N) mg/L 8.5 1.2 0.1 48 Total Phosphorus (as P) mg/L 1.63 0.3 0.03 4.7 Total Lead ug/L 125 100 1000 21.9 1.0 327 Total Cadmium ug/L 9.9 6 13 0.7 0.2 5.0 Total Zinc ug/L 540 108 1075 129.8 7.5 1245 Total Copper ug/L 85 120 1245 22.3 1.2 230 Regulator y Requirements The Environmental Protection Agency has mandated that Louisiana begin to issue discharge permits under the Louisiana Pollution Discharge Elimination System to comply with the National Pollution Discharge Elimination System, NPDES, a federall y supported program to limit the amount of contaminants that are being discharged into surface waters. The permit

PAGE 28

28 (La DEQ Permit No. LAS000101) which has been signed by: City of Baton Rouge/Parish of East Baton Rouge, Louisiana Department of Transportation and Development, Louisiana State University, Southern University, City of Baker and City of Zachary, is intended to reduce the contaminants that are discharged into the Amite Watershed. The current regulatory limits, shown in T able 3 5 are event based max imum concentrations; this means that for any rainfall event, these maximum limits can not be exceeded. The permit states that locations throughout the city will be monitored twice per year, once in November April and again in May October. During these bi yearly monitoring periods, one sample will be taken from the determined location during the first two hours of the storm. There has been some concern that the concentration limit is too strict while the monitoring and sampling criterion is too vague. As a rule of thumb for water treatment, assuming ideal conditions, one treatment process will yield at the most an 80% reduction in mass, or suspended sediment concentration however, in real world conditions this hardly ever observed. These current limits w ould be expensive to meet and the treatment system would be complicated to design and maintain. A more environmentally beneficial program would place limits on the surface water that is receiving the discharge. The limits should be based on the threshold c ontaminant exposure of the aquatic life within that surface water as opposed to placing a generalized storm event on all discharge which will be deposited into surface waters of varying size, limits, aquatic life and beneficial uses.

PAGE 29

29 Table 35. Non poi nt source discharge l imits for observation s ite Effluent Event Based Limit Current Conditions Percent Reduction Chemical Oxygen Demand 30 mg/L 351 mg/L 91.5% Cadmium 1 ug/L 35 ug/L 97.1% Copper 10 ug/L 200 ug/L 95.0% Lead 5 ug/L 85 ug/L 94.1% Zinc 2 0 ug/L 850 ug/L 97.6% Phosphorus 0.1 mg/L 1.63 mg/L 93.9% Turbidity 50 NTU 110.61 NTU 54.8% Mass (SSC) 80% Reduction 482.9 mg/L 80.0% After reviewing Permit No. LAS000101 there were several areas that could prove potential problems in the future if t his permit is issued. The areas of concern are as follows: A cute D ischarge L imits The maximum limits placed on the pollutants are not based on existing conditions and are too conservative. Rainfall data that was collected through Louisiana State Universit y Department of Civil and Environmental Engineering at locations: I 10 at City Park Lake and another location five miles away in a park setting will be used to compare the existing conditions to the desired. The data was gathered during the monitoring peri ods of May 2001 June 2002 and January 2002 June 2002, data from twelve storm events were recorded and tested, the results of Zinc are shown below: If this limit were regulated it would require a reduction of 97% for the maximum observed value and a 95% reduction for the mean value. The EPA states that secondary treatments could yield a reduction of approximately 85% from the existing conditions. In order for the permittees to meet the discharge limit, storm water would have to be captured and treated prior to discharging in addition to point source treatment, such as pavement cleaners and street sweepers.

PAGE 30

30 This solution is complicated and extremely expensive. These discharge limits should be based on a percent reduction of the existing conditions. Clea rly these limits will be extremely expensive to meet and are unreasonable. After reviewing the impaired waters in the East Baton Rouge Parish, it was noticed that some impairments have been excluded from the list that was given a priority rank of five or better by the EPA. The impairments: pesticides, priority/non priority organics, arsenic, nutrients, PCBs, pathogens, dissolved oxygen and siltation are among those excluded from the monitoring requirements. Also, contaminants such as: oil and grease, suspen ded solids, phosphorus and ammonia, which were prioritized by the EPA, were not given a discharge limit. EPA lists industrial, agricultural, construction, recreational activities, stream bank modifications as well as municipal activities among the sources of these contaminants. Some consideration should be given to regulating these sources instead of relying on the permittees to handle regulating these multiple sources. An alternative to a discharger bydischarger basis that is achieved through NPDES permit s is the application of the Water Effect Ratio (WER) on a watershed that is strongly encouraged by the EPA. The WER approach compares bioavailability and toxicity of a specific pollutant in receiving w aters and in laboratory waters (EPA 1983) It sets limits on metals by addressing bioavailability rather than expressing the limits in terms of dissolved metals. By using this method, the limits would be focused on the chemicals that pose a greater threat to a watershed as well as its inhabitants. In addition to using the WER method, the hardness factor should be used in determining the discharge limits. In using these methods, a better understanding will be given of the characteristics of the affected water bodies and the metal toxic limits for aquatic life an d the human factor.

PAGE 31

31 M onitoring F requency The monitoring frequency on twice a year during the annual monitoring periods of November April and May October would not yield accurate data. Results from the data of storm events can vary drastically depending on the discharge of the storm, duration, environmental conditions and when the event takes place. When the number of days between storm events (number of dry days) increases, the pollutant concentration increases. To avoid gathering data that is not repre sentative of existing conditions, increase the monitoring frequency to once a month or once every two months. By monitoring the pollutant discharge more closely a more accurate maximum could be obtained by taking into effect the deviation of these storms. Sample Collection The permit specifies that samples may be taken using the Composite or Grab Sample method. The Grab Sample method only specifies that the samples are to be taken during the first two hours of the storm. The permittee could interpret this a s meaning that a few samples could be taken at the end of this time frame, when toxic metals concentrations are much lower than in the first fifteen minutes of the storm, and they could report only the samples that were within the range specified. The Composite Sample, although it details the sampling process better than the Grab method, allows for data manipulation to report results that are within the specified parameters as well. This sampling method requires a minimum of nine samples, three per hour, during the first three hours of the storm. Previous analysis of storm events indicates that metals concentration is at the highest during the first fifteen minutes of the storm and after the first hour the concentrations begin to stabilize to show a much low er concentration than the first interval of the storm. The permittee could manipulate the data by taking multiple samples, more than the required, and only selecting the samples of favorable results since concentrations during the storm can vary depending on the characteristics of the storm.

PAGE 32

32 One solution to this potential problem is to require that the entire storm event be sampled at one minute increments. Also, a time constraint of no greater than twelve hours should be placed between the time of collect ion and the time of analysis in the laboratory. The reason for this is due to the fact that the metal concentration in the sample will start to stabilize between the particulate bound and dissolved matter, this will result in inaccurate metal concentration s. The issuing of this permit could prove to be a liability to La DEQ in future lawsuits for the enforcement of the Clean Water Act. There are many ways in which this permit could actually exacerbate existing conditions as opposed to ameliorating them. Too much responsibility is being placed on the permittees through the requirement of developing a Storm Water Management Plan that is to include: industrial and construction activity, regulating of agricultural land by means of reducing pesticide, herbicide and fertilizer application, monitoring programs, public education, pollution prevention, measurabl e goals and spill prevention. La DEQ fails to advise permittees on how to set up a Storm Water Pollution Prevention Plan, SWPP, by referring them to the proper E PA document or current SWPPs used in other cities that would meet all of these requirements. In the least, more guidance should be offered as opposed to placing all of the legal respons ibility, which is that of the La DEQs, and handing it over to universit ies, parishes, municipalities and La DOTD.

PAGE 33

33 CHAPTER 4 METHODOLOGY The evaluation of particulate matter generated for the observation site in Baton Rouge, Louisiana, was analyzed with respect to: traffic loading prior, during and after the event, rain fall intensity and total rainfall volume to determine the most influential factor(s) on the resulting particulate matter concentration. Data was collected for this location from: National Oceanic Atmospheric Administration and Louisiana Department of Trans portation. Since the datasets had some missing records and were recorded on different time intervals, modeling the data using statistical analysis methods was required to ensure an accurate comparison. Data that was collected from the traffic sensors was used to estimate to traffic that travelled over the observation site before, during and after a rainfall event. The rainfall data that was obtained from NOAA was used to determine the start/end time for each event. The traffic and rainfall data was then c ombined with the twenty seven rainfall runoff events for the observation period, March 14, 2004 August 9, 2006, that had measured PM that was completed through an LaDOTD research project to determine significant factors in PM generation for this catchmen t. Traffic Data Traffic data was obtained for this area from automated radar vehicle detectors that were installed and maintained by LaDOTD. The resulting data consisted of a recorded traffic volume, speed and occupancy every fifteen minutes for the vehic les that travelled over each catchment (east and west bound) for May 5, 2004 through August 31, 2006. Background There are 54 Radar Vehicle Detectors, RVD, stations along the Interstate in Baton Rouge. These detectors are located from Interstate 12 at Air line Highway, Interstate 10 at Bluebonnet to Interstate 110 just prior to Governors Home and I 10 just prior to Port Allen. Some detectors

PAGE 34

34 were installed in May 2004 and the latest section of the Interstate to receive these devices was I 10 after the I 12 split. The data was then collected and archived by a computer program located at LaDOTDs ITS section near the Ryan Airport called MIST, Management Information System for Transportation. This program then takes the average between two monitors, referred to as a link, for fifteen minute intervals and records the following information for each link: volume, speed and occupancy. Occupancy counts, or number of vehicles per lane per mile were not re liable therefore were not considered during the analysis. The volume and speed of each link is then collected for the purpose of creating a congestion map and displaying accurate travel times on message boards for motorists. The data that was used for this analysis was observed by monitors RVD 43/45 and recorded f or links 800043 (eastbound) and 800045 (westbound) which is shown in F igure 41. These recording devices were operational on May 6, 2004, with the last record obtained from LaDOTD ending on August 31, 2006. The monitors experienced outages that lasted for a few months during October November 2004 and July November 2005. The dataset contained over 54,564 traffic counts for east and westbound traffic which is shown in F igure 42, with a total of 26,844 counts missing that needed to be estimated. Since t he stormwater runoff from the east and westbound bridges were combined when they were collected, the data for east and west bound traffic was able to be combined and analyzed.

PAGE 35

35 Figure 41. Location of t raffic m onitors RVD 43/45 ( links 80043, 800045) over observation s ite 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 90,000 May-04 Jun-04 Aug-04 Oct-04 Nov-04 Jan-05 Mar-05 Apr-05 Jun-05 Jul-05 Sep-05 Nov-05 Dec-05 Feb-06 Apr-06 May-06 Jul-06 Date Vehicles Eastbound Westbound Figure 42. Total daily t raffic as observed by RVD 43/45 Link 800045 Data obtained from RVD 45(WB) Link 800043 Data obtained from RVD 4 3(E B)

PAGE 36

36 Analysis of Data The data that was obtained had recorded data for every fifteen minute interval that consisted of: date, time, volume (vehicles/hou r), speed (mile/hour) and occupancy. The volume of vehicles per hour was divided by 4 to determine the number of vehicles that travelled over the catchments for that time interval. Data gaps were also observed that needed to be addressed. For gaps that were relatively small, less than two hours, averages of the readings before and after the outage were used to estimate the traffic volume. Traffic volume throughout the day had three peaks that coincided with the morning, lunch and evening rush hours for the area. There was a distinctive linear increase/decrease depending on the time of the day; therefore this method to estimate small data gaps was acceptable. For data gaps that were two hours in length and greater, statistical analysis methods were used to minimize errors and closer esti mate realistic results. Figure 42 shows the vehicle volume as was recorded by the stations, prior to using analysis methods to fill the data gaps. The total daily traffic volume for weekday traffic (Monday Friday) and weekend traffic (Saturday and Sunda y) was relatively consistent. Changes in the total daily traffic at this location were observed after August 29, 2005, the date that Hurricane Katina made landfall off the coast of Louisiana and displaced thousands of residents from New Orleans to Baton Ro uge. In order to model the missing data that spanned over a period of hours to months, abnormalities in the total daily traffic caused by variables: holidays, weekend, public events, date, etc., needed to be identified and accounted for prior to estimating the total daily traffic for the missing days. Multiple Regression Model of Total Daily Traffic The traffic data for the east and west bound catchments were combined, since the stormwater was combined in piping prior to obtaining water samples. The next st ep was to identify any potential variables that could influence motorists to change traffic behavior. In order

PAGE 37

37 to do so, statistics were obtained for the Baton Rouge metropolitan area to determine what occurrences had the greatest potential to impact a lar ge amount of people. As the capital of Louisiana, it contains a large number of government offices, as this is one of the largest employers for the area. According to the 2005 US Census Bureau profile, the population total is 670,403 with the average age being 33 and a total of 24.6% being grade school age. The city also contains two state universities: Louisiana State University, LSU, (30,000 students enrolled) and Southern University, SU, (10,365 students enrolled). LSU has two large athletic stadiums f or events that consist of: a football stadium with a seating capacity of 92,400, and a basketball stadium with a seating capacity of 13,472. The area had also experienced a population influx in the aftermath of Hurricane Katrina after August 29, 2005 due t o the displaced residents of New Orleans. A multiple regression model was used to determine the total daily traffic for both east and west bound. Dummy variables were proposed to account for disruptions which resulted in a degree of variation from the ba se traffic pattern, which is defined as the observed traffic during days in which all dummy variables are equal to zero. In order to identify the elements that would cause traffic pattern disruptions t he following dummy variables were initially introduced for to determine the influence on traffic patterns: (1) Hurricane Katrina (2) Saturdays (3) Sundays (4) No classes for LSU/East Baton Rouge Parish public schools (5) State/local offices closed (6) Federal offices closed (7) Observed h oliday that no office s were closed (8) LSU/SU f ootball games (9) No school for EBR/LSU on Monday or Friday (10) State/local/federal offices closed Monday or Friday. For every dummy variable accounted for during the observed day, this represented a degree of variation of one fr om the base traffic pattern.

PAGE 38

38 After the analysis, it was determined the following dummy variables were significant and was accounted for in the multiple regression analysis : (1) holidays would only represent the days in which it was observed by both feder al and state agencies (2) colleges and public schools would both represent the days that school is out, but both would need to be out to be accounted for (3) Saturdays and Sundays would both represent the weekend variable (4) all weekdays would be grouped (5) Hurricane Katrina and (6) normal traffic is defined as the absence of variables. Figure 43 shows the observed total daily traffic and the associated variables, F igure 44 shows the total daily traffic and the total dummy variable for each day. A varia ble of zero indicates that there are no variables above the base; a variable of two indicates that the day contained two variables that resulted in a change from the base by two degrees. The results of the statistical analysis yielded the fo llowing and is shown in Table 41: (1) weekend variable was the most influential but had the largest degree of freedom indicating variations in the total daily traffic volumes for Saturday and Sunday (2) colleges and public schools being out was the least influential but also had the smallest degree of freedom indicating consistency for its affect on traffic volume (3) holidays were an important factor with a reasonable variation in the observed daily traffic (4) changes in traffic patterns due to Hurricane Katrina were less significant than was expected, the area had a noticeable influx of traffic that resulted in excessive delays, but this could be due to low service levels prior to the influx. The multiple regression analysis resulted in a coefficient of multiple dete rminates (R squared) value of 0.7126 and an equation to estimate the daily traff ic for the missing days: AADT =139,653.1125 + ( 2741.92 x LSU/EBR Out ) + ( 24343.877 x Holiday) + ( 31340.959 x Weekend) + (8505.579 x Katrina). This equation was used to model the total daily traffic for

PAGE 39

39 the missing days, as shown in Figure 45. If the variable is true, the value is set equal to 1, otherwise it shall equal zero. 40,000 60,000 80,000 100,000 120,000 140,000 160,000 180,000 May-04 Aug-04 Nov-04 Mar-05 Jun-05 Sep-05 Dec-05 Apr-06 Jul-06 Date Daily Traffic LSU/EBR OUT HOLIDAY WEEKEND Normal Katrina Figure 43. Observed total d aily t raffic with respect to the associated variable. 40,000 60,000 80,000 100,000 120,000 140,000 160,000 180,000 May-04 Aug-04 Nov-04 Mar-05 Jun-05 Sep-05 Dec-05 Apr-06 Jul-06 Date Daily Traffic Variable=0 Variable=1 Variable=2 Figure 44. Observed total d aily t raffic w ith respect to c orresponding dummy variable

PAGE 40

40 Table 4 1. Results of v ariables from m ultiple r egression a nalysis Variable Coefficient T Statistic LSU/EBR OUT 2741.92 2.81 Holiday 24343.88 10.47 Weekend 31340.96 29.90 K atrina 8505.58 10.70 40,000 60,000 80,000 100,000 120,000 140,000 160,000 180,000 May-04 Aug-04 Nov-04 Mar-05 Jun-05 Sep-05 Dec-05 Apr-06 Jul-06 Date Daily Traffic Observed Predicted Figure 45. Observed and predicted t otal daily traffic from m ultiple r egression analysis and m onitors Regression Model of Cumulative Traffic Volume The next step in estimating the missing traffic data was to apportion the daily tra ffic volume into 15 minute traffic counts, as was recorded by the monitors. When plotting the recorded traffic volume counts throughout the day that the monitors recorded three peaks were obser ved, which correlated to the morning, lunch and evening rush hours, but the data was scattered and a clear pattern was not observed. The cumulative traffic volume throughout the day was plotted which resulted in a distinct S curve pattern Due to this observation, it was determined that the cumulative daily traffic volume could be modeled using a logistic regression model. Days that had the same dummy variables, with the exception of separating Friday,

PAGE 41

41 Saturday and Sunday, as these days had different curves and this would reduce errors, were grouped together and a l ogistic regression model was completed for that set of data. The grouping of the data according to the associat ed variable is shown in Table 4 2. The graphs of the logistic regression for each group are shown in Appendix A which was used to determine the corresponding equations and R sq uare value are given in Table 4 3. For the linear equation, y = cumulative traffic volume at time x which is equal to the time or portion of the 24 hour day. The equations were used to estimate the missing traffic data f or the cumulative traffic volume throughout the day. The cumulative traffic volume was then apportioned into 15 minute traffic volume by taking the change from the previous estimated volume. Table 42. Groups according to the associated variables Group LSU/EBR O ut H oliday W eekend K atrina F riday S aturday S unday 1 0 0 0 0 0 0 0 2 0 0 0 0 1 0 0 3 0 0 0 1 0 0 0 4 0 0 0 1 1 0 0 5 0 0 1 0 0 1 0 5 a 0 0 1 0 0 0 1 6 0 0 1 1 0 1 0 7 0 0 1 1 0 0 1 8 0 1 0 0 0 0 0 9 0 1 0 0 1 0 0 10 0 1 0 1 0 0 0 11 0 1 0 1 1 0 0 12 1 0 0 0 0 0 0 13 1 0 0 0 1 0 0 14 1 0 0 1 0 0 0 15 1 0 0 1 1 0 0 16 1 1 0 0 0 0 0 17 1 1 0 0 1 0 0 ***Groups 8,9,16 and 17 were subsequently combined ***Groups 10 & 11 were subsequently combined

PAGE 42

42 Table 43. Summary of logistic regression with resulting linear equation Group Linear Equation R Squared 1 y = (295877*x^3) + (482046*x^2) (49893x) + 2238.6 0.997 2 y = (245601*x^3) + (431149*x^2) (43500x) + 3779.6 0.974 3 y = (296709*x^3) + (482141*x^2) (41081x) +2092.5 0.9 98 4 y = (276422*x^3) + (466149*x^2) (36897x) + 2165.7 0.999 5 y = (162999*x^3) + (297784*x^2) (26005x) + 4904.0 0.939 5 a y = (177592*x^3) + (329122*x^2) (51350x) +6277.7 0.916 6 y = (196759*x^3) + (359944*x^2) (43594x) + 5389.4 0.992 7 y = (190151*x^3) + (364109*x^2) (66965x) + 6901 0.990 8, 9, 16, 17 y = (155844*x^3) + (298123*x^2) (26810x) + 4214.9 0.866 10, 11 y = (206642*x^3) + (372311*x^2) (46067x) + 4296.1 0.941 12 y = (273805*x^3) + (450563*x^2) (44973x) + 2953.7 0.9 78 13 y = (278142*x^3) + (477051*x^2) (57929x) + 4042.1 0.990 14 y = (289484*x^3) + (472630*x^2) (42091x) +2372.1 0.993 15 y = (278650*x^3) + (469197*x^2) (42319x) + 2881.9 0.998 Rainfall Data Rainfall data was obtained from two monitors, one is located at the observation site and another is located 7.6 miles north at Ryan Airport and is maintained by the National Oceanic and Atmospheric Administration, NOAA, which recorded precipitation data every hour. The monitor at Ryan Airport was used to determine the start time of the rainfall events for entire analysis period (March 14, 2004 August 9, 2006), since the sensor has certified hourly precipitation data Utilizing this method is acceptable since the majority of the storm events came from th e east and both recorders should reco rd the event at relatively the same time. The monitor at the observation site was used to determine event duration, total rainfall, and previous dry hours. The hourly rainfall data from Ryan Airport was apportioned int o a rainfall series with a 15 minute time step using a stochastic disaggregation model. The precipitation analyzer, which was created by the University of Florida, evaluated the recorded data and calculated the

PAGE 43

43 corresponding 15minute rainfall using a stoc hastic disaggregation model. The program utilizes an algorithm that was proposed by Ormsbee in 1989 to simulate 15 minute values from hourly data. The Ormsbee algorithm accounts for pulse depths of a specified depth to return the desired output time step. (Cowpertwait 2001 ) Events that are less than one hour in duration were not able to be apportioned and the start time was equal to the time that the NOAA sensor first recorded. The previous dry hours as recorded at both stations were compared and confirmed that the storm events reached both stations within a 30 minute time span. Water Chemistry Analysis Particulate matter (PM), which was represented by the suspended sediment concentration was measured and analyzed for 27 rainfall runoff events at the obs ervation site through a LaDOTD research project to characterize the water chemistry of the stormwater runoff. During the event s the following parameters were recorded: total runoff volume total rainfall, duration of event previous dry hours, PM and vehic les during the storm and is shown in T able 4 4 (Sansalone 2009; Kim 2008). Traffic data that was recorded by the RVD sensors were evaluated to determine the total number of vehicles that travelled over the catchment prior to and during storm events, and t he average speed of the vehicles during the event which is shown in T able 4 4. The following independent variables are to be analyzed to determine if they are a factor for the dependent variable, PM, which shall be verified through a multiple regression a nalysis: (1) previous dry hours (PDH) (2) vehicles during storm (VDS) (3) vehicles prior to storm (VPS) (4) total runoff volume (5) density of vehicles during storm and (6) average rainfall intensity. Prior to completing a regression analysis, the five var iables were plotted with a linear trend line to

PAGE 44

44 determine if a correlation could visually be confi rmed, the results are shown in F igure 46 and 47. Table 4 4. S ummary of hydrologic, sampling based indices and traffic data for 27 events analyzed for the I 10 observation site Date Runoff PM (as SSC) grams Previous Dry Hours (Hours) Total Runoff Volume (L) Average Rainfall Intensity (mm/hr) Vehicles During Storm (vehicles) Density of Vehicles During Storm (veh/min) Vehicles Prior to Storm (vehicles) Average Speed (mph) 14 Mar 2004 4,949 204 24076 0.128 41,350 100 1,074,550 24 Apr 2004 3,219 313 7288 0.023 17,646 96 1,731,836 20 Aug 2004 10,592 26 12286 0.665 4,290 143 150,252 14 Oct 2004 544 84 1672 0.031 26,136 145 467,073 5 Jun 2005 4,758 117 5 856 0.061 5,637 108 632,617 59.82 30 Jun 2005 4,044 143 15117 0.134 3,601 46 772,765 71.79 21 Aug 2005 8,838 94 50002 0.544 12,785 113 485,376 55.95 3 Oct 2005 738 217 2615 0.015 1,349 135 1,245,643 28.00 21 Apr 2006 4161.7 927 2927 0.004 6,025 123 5,3 84,253 54.90 29 Apr 2006 10466.2 84 48306 0.850 9,882 67 519,744 54.41 6 May 2006 172.2 157 495 0.006 9,620 139 898,646 62.05 7 May 2006 1630.1 21 3852 0.305 4,640 113 93,126 70.49 27 May2006 1219.1 482 2628 0.007 2,275 103 2,782,555 55.14 28 May 200 6 320.3 23 2096 0.135 2,922 108 98,512 71.25 16 Jun 2006 5363.6 451 9938 0.031 9,953 131 2,539,339 51.10 19 Jun 2006 541.4 79 1816 0.039 4,295 119 442,807 63.91 4 Jul 2006 831.8 352 2779 0.011 2,323 80 1,934,366 70.03 5 Jul 2006 1065.1 25 3838 0.224 6, 498 135 136,764 51.47 9 Jul 2006 204.7 69 674 0.026 2,999 111 452,780 70.85 10 Jul 2006 9869.2 16 25189 1.906 8,154 127 73,936 34.50 14 Jul 2006 1398.4 89 3304 0.063 6,831 118 532,962 66.19 16 Jul 2006 246.7 45 945 0.056 6,668 119 215,590 69.78 18 Jul 2006 276.1 26 1047 0.077 2,679 134 158,318 53.07 4 Aug 2006 15025 191 36990 0.274 8,735 115 2,299,270 34.87 5 Aug2006 1080.1 19 6421 0.479 5,184 104 88,508 50.85 7 Aug 2006 890 25 6022 0.356 3,153 137 191,765 69.31 9 Aug 2006 2550.2 23 12502 0.709 5, 588 82 136,848 59.62

PAGE 45

45 0 4,000 8,000 12,000 16,000 0.0 1.0 2.0 Avg. Rainfall Intensity (mm/hr) PM (grams) 0 4,000 8,000 12,000 16,000 0 25,000 50,000 Total Runoff Volume (L) PM (grams) 0 4,000 8,000 12,000 16,000 0 250 500 750 1000 Previous Dry Hours (hours) PM (grams) 0 2 4 6 0 250 500 750 1000PDH (hours)VPS (vehicles) 10^6 0 4,000 8,000 12,000 16,000 0 2 4 6 VPS (Vehicles) 10^6 PM (grams) 0 4,000 8,000 12,000 16,000 0 25,000 50,000 VDS (vehicles) PM (grams) Figure 46. Average rainfall intensity, total runoff volume, previous dry hours (PDH), vehicles prior to storm (VPS) and vehicles during the storm (VDS) graphed against suspended sediment concentration, PM, for twentyseven rainfall runoff events in Baton Rouge, Louisiana.

PAGE 46

46 0 4000 8000 12000 16000 0 250 500 VDS Density (veh/mile) PM (grams) 0 4000 8000 12000 16000 0 50 100 150 200 VDS Density (veh/min) PM (grams) Figure 47. Vehicles during storm as density of time and area graphed against suspended sediment concentration, PM, for twentyseven rainfall runoff events in Baton Rouge, Louisiana. T otal run off volume is directly correlated to the resulting PM which is demonstrated in the closeness of fit to the linear trend. Average rainfall intensity has more outliers than the total runoff volume; therefore runoff volume shall be evaluated in a regression model. VPS has a strong linear correlation with PDH, but VPS does show more of a linear trend than PDH when compar ed to PM. VDS and VDS Density (veh/min) does not have linear trend when compared to PM and the data appears to be scattered with excessive out liers VDS Density, which is a variable that accounts for VDS, event duration and weighted average vehicle speed, does have a linear trend in comparing to the PM and may be a significant factor in the resulting PM. Nine multiple regression analyses w ere completed with varying independent variables for PM to determine the significance of each factor t he results of the analysis are shown in T able 4 5. A 95% confidence level was applied to the regression analysis; a p value of less than 0.2 and t statistic > 1.5 was desired. In reviewing the results of the analysis and the plot of the resulting equations against the recorded PM, it was determined that analyses 1 4 had the greatest amount of variance and did not reject the null hypothesis. Analyses 59, whose equations are plotted in F igure 48, contained significant independent variables with 9 having the closest goodness of fit

PAGE 47

47 but 7 had the most significant variables with desired p value and t statistic. Analysis 7, which accounted for volume of runoff, PDH and VDS (veh/mil), most closely modeled the recorded PM and were the most significant variables. When the particulate matter generated form an event is greater than 5,000 grams, the regression model shows some instability and varies greatly from the recorded data. These storm events that generated greater than 5,000 grams of PM are higher volume storms that had PDH an average of 84. The results of the analysis 7 indicate that the most significant factor is total runoff volume and previous dry hours. The density of vehicles during the storm (veh/mil) was a weak indicator with a high pvalue and a low t statistic. Regardless of the lack of strong correlation of the density of VDS, this analysis had errors minimized with a reasonable R square value and overal l desired t statistic and p value. With respect to the resulting coefficients, this indicates that density of PDS and PDH are equally weighted when applied to the equation indicating equal importance. With respect to PM transported in stormwater from roadw ay surfaces, total runoff volume, PDH and density of VDS are the most significant independent variables Total runoff volume indicates the strongest correlation to the resulting PM. PDH does have a correlation when the hours are less than 100, but as that number increases the linear trend of the data is reduced and the data is scattered. The density of VDS does indicate a moderate linear trend for the resulting PM, but variations in the data increase as the density exceeds 250 vehicle/mile.

PAGE 48

48 Table 4 5. Summary of multiple regression analysis for nine scenarios based on twenty seven observed rainfall runoff events. Analysis 1 R Squared: 0.774 Coefficient t Statistic P value Intercept 851.265 1.234 0.230 Previous Dry Hours (Hours) 25.474 2.273 0 .033 Total Event Volume (L) 0.216 6.483 1.60E 06 Vehicles During Storm (vehicles) 0.020 0.379 0.708 Vehicles Prior to Storm (vehicles) 0.005 2.560 0.018 Analysis 2 R Squared: 0.742 Coefficient t Statistic P value Intercept 3292.937 1.383 0.1 80 Total Event Volume (L) 0.256 7.943 4.84E 08 Vehicles Prior to Storm (vehicles) 0.001 1.756 0.092 Vehicles During Storm (veh/min) 30.554 1.599 0.124 Analysis 3 R Squared: 0.707 Coefficient t Statistic P value Intercept 2457.189 1.010 0.323 Total Event Volume (L) 0.253 7.539 8.87E 08 Vehicles During Storm (veh/min) 28.739 1.445 0.162 Analysis 4 R Squared: 0.824 Coefficient t Statistic P value Intercept 134.325 0.223 0.826 Total Event Volume (L) 0.210 3.292 0.004 Vehicles Prior to Storm (vehicles) 0.001 2.346 0.030 Vehicles During Storm (veh/mil) 3.984 0.547 0.591 Analysis 5 R Squared: 0.713 Coefficient t Statistic P value Intercept 375.351 659.6478 0.569017 Total Event Volume (L) 0.239 0.031417 7.613395 Vehicles Prior t o Storm (vehicles) 0.001 0.000371 1.62028 Analysis 6 R Squared: 0.699 Coefficient t Statistic P value Intercept 507.236 0.734404 0.469818 Total Event Volume (L) 0.241 7.450191 1.09E 07 Previous Dry Hours (Hours) 2.593 1.148604 0.262039

PAGE 49

49 Table 4 5. Continued Analysis 7 R Squared: 0.803 Coefficient t Statistic P value Intercept 15.105 0.023 0.982 Total Event Volume (L) 0.209 3.072 0.006 Previous Dry Hours (Hours) 3.354 1.694 0.107 Vehicles During Storm (veh/mil) 4.437 0.573 0.573 Analys is 8 R Squared: 0.728 Coefficient t Statistic P value Intercept 3264.012 1.321 0.200 Total Event Volume (L) 0.259 7.767118 7.09E 08 Previous Dry Hours (Hours) 2.920 1.328 0.197 Vehicles During Storm (veh/min) 31.190 1.586 0.127 Analysis 9 R S quared: 0.824 Coefficient t Statistic P value Intercept 134.325 0.223 0.826 Total Event Volume (L) 0.210 3.292 0.004 Vehicles Prior to Storm (vehicles) 0.001 2.346 0.030 Vehicles During Storm (veh/mil) 3.984 0.547 0.591 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 0 5 10 15 20 25 30 Rainfall-Runoff Events PM (grams) Observed 5 6 7 8 9 Figure 48. Mul tiple regression model equation for five scenarios against measured PM for twenty seven rainfall runoff events in Baton Rouge, Louisiana.

PAGE 50

50 CHAPTER 5 DISCUSSION Total runoff volume has the strongest correlation with resulting PM, which confirms that the gre ater the capacity of the runoff to transport particles from the roadway surface the greater the PM that shall be transported. PDH does indicate a linear trend, but as the hours exceed 100 variations in the data are observed indicating that there are a limi ted amount of build up on the roadway surface during this time before these particles are transported by other means such as high velocity winds. The result of the density of VDS indicates that if vehicle speeds are maintained and the traffic volume is not congested this is a significant factor in the resulting PM. With a limited buildup of par ticles on the roadway surface, street sweeping shall need to occur on a more frequent basis in order to meet TMDL requirements and have an impact on water quality. Sweeping on monthly or annual basis is not enough to comply with TMDL requirements or improve runoff quality. Sweeping should occur in a regular basis that is less than 100 hours between each event, with more frequency occurring during times when rainfall runoff events are more frequent. Sweeping on a frequent basis with a high efficiency sweeper will increase the removal capacity for particles less than 40 um that can be dissolved in runoff and exceed surface water discharge concentration limits. Motorists can also improve stormwater runoff by being more proactive in washing their vehicles more frequently during periods of increased hydrologic events. By cleaning vehicles in carwashes that have the ability to recycle or treat the wash waters, this would red uce these particles from washing off the vehicles onto the pavement during rainfall runoff events. The results of this analysis are based only on the pollutants that are transported from the roadway surface through runoff. Pollutants that are transported by turbulent winds and deposited directly into surface waters below the bridge have not been accounted for, but do have a potential

PAGE 51

51 for impairing surface waters. Other pollution sources, such as atmospheric deposition and pavement degradation have not been accounted for in this analysis. The observation site does have asphalt approaches that are becoming a maintenance issue for LaDOTD. Although the bridge deck is constructed entirely of concrete, there is a potential for PM to become loose from the surface from the interaction of the vehicles tire with the roadway surface. If this study were performed in another location that did not have asphalt approaches to the bridge or a concentrated amount of oil refining activities occurring within the area, the results could vary, as it is unknown the effects on water quality these factors account for.

PAGE 52

52 CHAPTER 5 CONCLUSIONS This study of an urban roadway in Baton Rouge, Louisiana shows that street sweeping and washing vehicles on a more frequent basis could potent ially improve stormwater runoff by reducing the available pollutants on the roadway surface. Particles are not continually collecting on the roadway surface and if left they are deposited into the environment through means other than transported by stormwa ter runoff. By improving source control prior to rainfall events, this shall improve the stormwater quality and reduce the impacts to the environment. The TMDLs that have been set for this area shall not be able to be attained through street sweeping alone regardless of the frequency. Structural controls, such as stormwater treatment, shall need to be constructed to meet removal efficiencies of greater than 80% for metals, which is difficult for sweepers to attain as they can only remove the particulate bo und fraction, not the dissolved fraction, which is highly mobile and exceeds regulat ory limits. A control strategy intended to effectively immobilize these metals must provide for adsorption, ion exchange or precipitation in addition to a mechanism to trap particulate bound metal elements (Sansalone 1997). Additional studies are recommended to determine if increasing the frequency of washing vehicles during periods of frequent rainfall events could reduce the deposition of particles during the interaction of the rain on the vehicle. The cost of this best management practice has the potential to outweigh the benefit to water quality.

PAGE 53

53 APPENDIX G RAPHS OF LOGISTIC REGRESSION FOR CUM ULATIVE DAILY TRAFFIC Figure A 1. Graph of cumulative daily t raffic volume a nd s c urve from l ogistic regression (g roups 15a)

PAGE 54

54 Figure A 2. Graph of c umulative daily t raffic volume and s curve from l ogistic r egression (g roups 613)

PAGE 55

55 Figure A 3. G raph of c umulative daily t raffic volume and s c urve from l ogistic r egression ( groups 1415)

PAGE 56

56 LIST OF REFERENCES American Public Works Association (APWA). (1969). Water pollution aspects of urban runoff. U.S. Dept. of the Interior, Federal Water Pollution Control Administration, WP 2015. California Depar tment of Transportation. (CalTran). (2002). Stormwater program c haracteristics of s tormwater runoff from CalTrans f acilities Transportation Research Board, 81st Annual Conference, Sacramento, CA. Center for Watershed Protection (CWP). (2008). Deriving Reliable Pollutant Removal Rates for Municipal Street Sweeping and Storm Drain Cleanout Programs in the Chesapeake Bay Basin. Chesapeake Bay Program. (2003). The Changing Face of Stormwater Management, Cowpertwait, Paul (2001). A c ontinuous s tochastic disaggregation model of r ainfall for peak f low s imulation in urban hydrologic s ystems. Res. Lett. Inf. Math Sci 2, 8188. Glen, D.W., Sansalone, J.J. and Howerter, K. (2002). Heavy metal partitioning to particles in sn ow e xposed to urban t raffic distribution a cross the particle degradation. Environmental and Water Resources Institute of American Society of Civil Engineers, 10.1061/40644. Hird, J.P., Sansalone, J.J., Cartledge, F.K. and Tittlebaum, M.E. (2003). Event based s torm w ater quality and quantity loading from e levated urban i nfrastructure impacted by t ransportation Water Environment Research, 77, 348365. Kim, J.Y. and Sansalone, J.J. (2008). Event based size distributions of particulate matter transported during urban rainfall runoff events. Water Res., 42(1011):275668. Li, M. and Barrett, M. (2008). Relationship between a ntecedent dry period and h ighway pollutant: c onceptual m odels of buildup and removal p rocesses. Water Environment Research, Volum e 80, Number 8. Louisiana Department of Environmental Quality (LaDEQ). (2007). Water quality Assessment in Louisiana. < http://www.deq.louisiana.gov/portal/tabid/69/Default.aspx> Metcalf and Eddy, Inc. (2003) Wastewater Engineering: Treatment and Reuse, 4th Edition, McGraw Hill, New York. Pitt, R., Bannerman, R. and Sutherland, R. (2004). The r ole of s treet c leaning in s tormwater m anagement Environmental and Water Resources Inst itute of American Society of Civil Engineers, 10.1061/40737. Sabry, M., Abd, H., Yousef, S. and Badra, N. (2007). A time s eries f orecasting of average daily t raffic volume Australian Journal of Basic and Applied Sciences, 1(4): 386394.

PAGE 57

57 Sansalone, J. J and Buchberger, S. G. (1997) Partitioning and f irst f lush of m etals in urban r oadway s torm w ater J. Environ. Eng ., 123(2), 134143. Sansalone, J.J., Field, R., and Sullivan, D. (2003) Wet Weather Flow in the Urban Watershed, Technology and Manageme nt, Lewis Publishers, Baca Raton. Sansalone, J.J., Liu, B. and Kim, J.Y. (2009). Volume c larifying f iltration of urban s ource area rainfall r unoff ASCE J. of Environmental Engineering, 135(8), 609620 United States Environmental Protection Agency (EPA ). (1983). Title 40: Protection of Environment, Part 131Water Quality Standards, 48 FR 51405. United States Environmental Protection Agency, Office of Water (EPA). (1993). Guidance Specifying Management Measures for Sources of Nonpoint Pollution in Coas tal Waters Washington, D .C United States Environmental Protection Agency, Office of Wastewater Management (EPA). (1997). Guidance Manual for Implementing Municipal Storm Water Management Programs, Washington, D.C. United States Environmental Protection A gency (EPA). (2000). Federal Register, Part III: 40 CPR Part 131: Water Quality Standards: Establishment of Numeric Criteria for Priority Toxic Pollutants for the State of California. Washington, D.C. United States Environmental Protection Agency, Office of Water (EPA). (2008) TMDLs to Stormwater Permits Handbook, Washington, D.C. United States Geological Survey (USGS). (1983). Water Quality Assessment of Stormwater Runoff From a Heavily Used Urban Highway Bridge in Miami, Florida Tallahassee, FL. United States Senate. (2002). Federal Water Pollution Control Act 33 U.S.C. 1251 et. Seq., Washington, D.C Walker T.A. and Wong, T. (1999). Effectiveness of s treet s weeping for s tormwater pollution c ontrol Cooperative Research Center for Catchment Hydrology, Technical Report 99/8.

PAGE 58

58 BIOGRAPHICAL SKETCH Melanie R. Grigsby is a registered Professional Engineer in the state of Florida and Louisiana in the field of Civil Engineering. She completed her bachelor s degree in civil e ngineering from Louisiana State U niversity in 2002. She has practiced in her field for a total of seven years in transportation and stormwater management at state and local municipal levels. In order to focus her career in stormwater management and mitigating environmental problems from u rbanization, she earned a Master of Engineering from the University of Florida in 2010.