Citation
Nutrient (N, P) Transport, Partitioning and Fate in Urban Runoff

Material Information

Title:
Nutrient (N, P) Transport, Partitioning and Fate in Urban Runoff
Creator:
Zhang, Hao
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Environmental Engineering Sciences
Committee Chair:
Sansalone, John Joseph
Committee Members:
Bonzongo, Jean-Claude J
Koopman, Ben L
Mclamore, Eric
Graduation Date:
5/4/2013

Subjects

Subjects / Keywords:
bmp
nitrogen
partition
runoff
speciation
transport
uop
Rain ( jstor )
Nutrients ( jstor )
Watersheds ( jstor )
Genre:
Unknown ( sobekcm )

Notes

General Note:
High levels of nutrients (N, P) from anthropogenic and biogenic sources through urban stormwater transport can cause eutrophication in the discharging areas. In this dissertation, the intra-event and inter-eventpartitioning, speciation, transport and transformation of N and P were examinedacross 40 runoff events and 13 retention periods on a paved urban source area. Aradial cartridge filtration (RCF) system and a volumetric clarifying filtration (VCF) system were tested for the performance of nutrient reduction. Results found that PM deposited in the watershed is the main contributor of nutrientsin runoff.  TN in dry deposition is stronglycorrelated with the organic content of PM. Results found that dissolved N (DN) is the predominant fraction, while settleable N (25~75 µm) held the lowest portion. Finer particles are preferred to be transported by runoff because it requires less streampower.  Mass-limited (first-order) modelis the predominant transport model for N fractions. Finer PM fractions are moreabundant in runoff, while coarser PM fractions show more significantfirst-flush effect.  Among DN species, NOx-and ON are the two predominant species. Mass transport behavior of DN species depends on the volume and duration of rainfall-runoff events. DP follows a mass-limitedmodel for each event due to the limited sources of phosphorus. The VCF system can effectively reduce TN and TPconcentrations by removing PM-associated nutrients. Higher removal efficiency was observed for coarser PM fraction. The RCF system showed 70% removal efficiency on P while the N removal efficiency is not significant.  N is the controlling nutrient in runoff, therefore, reducing N loadings could effectively reduce the potential of eutrophication. During retentionperiods, an aerobic/anoxic zone is quickly created inside the RCF system. Thevariation of Water chemistry indices is related to the microbial respirations and decomposition of biogenic materials. DN increased over the retention time, indicating the decomposition of biogenic materials is the prevalent process.TAN and DON increased while NOx- decreased over time,where multiple pathways could be involved.  Pourbaix diagram indicates that the dominant Nspecies in rainfall, runoff and effluent is N2 via biologica ldenitrification, while NH4+ is predominant in retained stormwater. Result found the aerobic/anoxic cycle could promote N removal as the microorganisms has grown and evolved over the extended retention periods.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Zhang, Hao. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
5/31/2015

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 NUTRIENT (N, P) TRANSPORT, PARTITIONING AND FATE IN URBAN RUNOFF By HAO ZHANG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCT OR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Hao Zhang

PAGE 3

3 To my grandmother, Tang, Yun Fen

PAGE 4

4 ACKNOWLEDGMENTS I would like to give my sincerely gratitude and deep appreciation to my advisor, Dr. John J. Sansalone, who consistently guided, encouraged and supported me throughout the journey of my Ph.D. program. His patient elucidation, enlightening ideas and precious comments have contributed a lot to my understanding of this research area and shaping m y concept of scientist and engineer. my confidence and capability on a professional level have been greatly strengthened. It was and will be great fortune and enormous inspiration for me in my life. I also extend my great appreciation to the distinguish professors on my committee for their helpful advice on the dissertation work : Dr. Ben Koopman, Dr. JeanClaude J. Bonzongo, and Dr. Eric S. Mc L amore. I express my thanks to my colleagues: Dr Gaoxiang Ying, Dr. Christian Berretta Dr. Joshua Dickenson, Dr. Hwan Chul Cho, Dr. Tingting Wu, and Dr. Natalie M. Winberry, who shared me with their knowledge and helpful discussion. My appreciation also ext ends to my colleagues including D r. Saurabh N. Raje, Mrs. Christina H. Joiner, and Dr. Ruben A. Keztesz for thei r valuable assistance and help. Their friendships have been one of my important accomplishments in the past six years. I appreciate my parents and my grandmother and they are always behind me in spite of the great distance separating us. Finally, I would l ike to express my deepest gratitude to my wife, Dr. Ting Cheng. H er continuous support of love, patience and encouragement has given me the strength to accomplish this Ph.D degree.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES .......................................................................................................... 9 LIST OF ABBREVIATIONS ........................................................................................... 11 ABSTRACT ................................................................................................................... 13 CHAPTER 1 GLOBAL INTRODUCTION ..................................................................................... 15 2 PARTITIONING AND FIRST FLUSH OF NITROGEN IN SOURCE AREA URBAN RAINFALL RUNOFF ................................................................................. 21 Abstract ................................................................................................................... 21 Introduction ............................................................................................................. 22 Objectives ............................................................................................................... 24 Methods and Materials ............................................................................................ 25 Watershed Description ..................................................................................... 25 Sampling Methodology ..................................................................................... 25 TN Analysis Method ......................................................................................... 26 Hydrologic Transport Variables ........................................................................ 27 TN Partitioning between Dissolved and PM Fractions ...................................... 28 Mass Transport of N Fractions ......................................................................... 29 Logistic Regression .......................................................................................... 29 First Flush Transport ........................................................................................ 30 Dry Deposition TN Analysis .............................................................................. 31 Results and Discussion ........................................................................................... 31 RainfallRunoff Events ...................................................................................... 31 Partitioning of TN between Dissolved and PM Fractions .................................. 32 Categorical Analysis fo r Transport of N Fractions ............................................ 34 Mass Transport of N Fractions ......................................................................... 34 Distribution of N over Particle Size Gradation .................................................. 35 Conclusion .............................................................................................................. 37 3 VOLUMETRIC CLARIFYING FILTRATION OF NUTRIENT FRACTIONS IN URBAN SOURCE AREA RUNOFF ........................................................................ 54 Abstract ................................................................................................................... 54 Introduction ............................................................................................................. 55 Objectives ............................................................................................................... 59

PAGE 6

6 Methodology ........................................................................................................... 60 Site Description ................................................................................................ 60 VCF Configuration ............................................................................................ 61 Samplin g Protocol ............................................................................................ 62 Laboratory Analyses Methods .......................................................................... 62 N Partitioning in Runoff .................................................................................... 64 Mass Transport Behavior ................................................................................. 65 Sequestration of Mass during the Monitoring Campaign .................................. 65 Results and Discussion ........................................................................................... 66 Hydrology of Monitored Rainfall Runoff Events ................................................ 66 Distribution and Concentration of N Fractions .................................................. 66 Nutrient (N, P) Partitioning on Siteand Event Basis ....................................... 69 TN Mass Transport Behavior ............................................................................ 70 VCF Performance i n Nutrient (N, P) Removal .................................................. 71 The C: N: P Stoichiometry in Rainfall, Runoff and Effluent ............................... 71 Conclusion .............................................................................................................. 72 4 NUTRIENT (N, P) SPECIATION, TRANSPORT AND TREATMENT IN RUNOFF ON A SOURCE AREA ............................................................................ 87 Abstract ................................................................................................................... 87 Introduction ............................................................................................................. 87 Objective ................................................................................................................. 89 Methodology ........................................................................................................... 90 Source Area Watershed Description ................................................................ 90 Treatment System Design ................................................................................ 91 Sampling Protocol ............................................................................................ 92 Monitored Hydrologic Parameters .................................................................... 93 Nutrient Speciation ........................................................................................... 94 Data Elaboration ............................................................................................... 94 Results and Discussions ......................................................................................... 96 Hydrologic Profiles and Water Chemistry Characterization .............................. 96 Nutrient Pathways during Rainfall Runoff Events ............................................. 97 Intra Event Nutrient Transport Behavior ........................................................... 99 Correlation between CODd and DOC ............................................................. 100 Conclusion ............................................................................................................ 100 5 NUTRIENTS TRANSFORMATION FOR URBAN STORMWATER RUNOFF RETAINED IN A RADIAL CARTRIDGE FILTRATION (RCF) SYSTEM ................ 116 Abstract ................................................................................................................. 116 Introduction ........................................................................................................... 117 Objective ............................................................................................................... 120 Methods and Materials .......................................................................................... 121 Source Area Description ................................................................................. 121 RCF System Setup ......................................................................................... 121 Sampling Protocols ........................................................................................ 122

PAGE 7

7 Retention Water Chemistry Indices ................................................................ 123 Nutrient Speciation ......................................................................................... 124 Data Elaboration ............................................................................................. 124 Limited Exponential Growth/Decay Model ...................................................... 125 P ourbaix (EhpH) Diagram ............................................................................. 126 Results and Discussion ......................................................................................... 128 Water chemistry indices variations during retention periods ........................... 128 Nutrients Variation during Retention Periods .................................................. 129 Nutrient Pathways and Species Transformations ........................................... 131 Longterm performance shift for the RCF system ........................................... 133 Conclusion ............................................................................................................ 133 6 GLOBAL CONCLUSION ....................................................................................... 149 LIST OF REFERENCES ............................................................................................. 156 BIOGRAPHICAL SKETCH .......................................................................................... 168

PAGE 8

8 LIST OF TABLES Table page 2 1 Hydrologic parameters of 14 rainfall runoff events in Gainesville, Florida. ......... 39 2 2 Event based concentration (EMC) of N fractions and TN .................................. 40 2 3 Hydrologic indices for logistic regression (LR) of the 14 rainfall runoff events ... 41 2 4 Differentiation of the mass transport behavi or of each TN fraction. .................... 42 2 5 Watershed geometric characteristics, affiliated to Figure 21 ............................ 43 3 1 Mass transport models of flow limited and mass limited behavior ..................... 74 3 2 Event based hydrology parameters for 25 rainfall runoff events ....................... 75 3 3 Event based N fr actions and Total N EMCs and discharge limits ...................... 76 3 4 Event based P fractions and Total P EMCs and discharge limits ...................... 77 4 1 Event based hydrology parameters for the 15 rainfall runoff events ............... 102 4 2 Event based statistics of rainfall chemistry indices .......................................... 103 4 3 Sample based Statistics of major water chemistry indices for runoff ............... 104 4 4 Event mean concentration (EMC) of dissolved nutrients. ................................ 105 5 1 Summary of water chemistry i ndices for each retention periods ...................... 136 5 2 Summary of water chemistry indices for each retention periods ....................... 137 5 3 Model parameters of water chemistry indices affiliated to Figure 55 and 6. ... 138 5 4 Summary of influent and effluent EMC concentrations of 2008 storm events .. 139

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Plan view of the studied catchment in Gainesville, FL. ....................................... 44 2 2 Rainfall intensity and runoff flow rate frequency distribution for Gainesville, FL. with the 14 monitored rainfall runoff events su perimposed. ......................... 45 2 3 PM based N mass fraction for each event and nonparametric plots for N fractions as EMCs for the 14 monitored rainfall runoff events ........................... 46 2 4 Probability density functions (pdf) of N fractions, TN and the runoff flow rate of the 14 runoff events ....................................................................................... 47 2 5 fd and Kd values as nonparametric plots for the 14 rainfall runoff events .......... 48 2 6 the dissolved fraction (fd), equilibrium coefficient (Kd) and cumulative volume ........ 49 2 7 Cumulative mass volume curve and mass transport model for N fract ions ...... 50 2 8 Dimensionless cumulative mass volume curves for N fractions ...................... 51 2 9 Distributions of N and volatile fraction over t he PSD for dry deposition PM. and PSDs of dry deposition (DD) and runoff (Q) PM ......................................... 52 2 10 Probability density functions(pdf) of the PM based N (mg/kg) fractions for the 14 monitored rainfall runoff events .................................................................... 53 3 1 The site drawing of the studied watershed in UF Reitz Union parking lot and t he side view of the volumetric clarifying filtration (VCF) system ....................... 78 3 2 Probability density function (pdf) of N fractions and TN based on 25 monitored storm events between 2010 and 2011. .............................................. 79 3 3 Probability density funct ion (pdf) of P fractions and TP based on 25 monitored storm events between 2010 and 2011. .............................................. 80 3 4 Probability density function (pdf) of fd, Kd and PM concentration based on 25 monitored storm events between 2010 and 2011. .............................................. 81 3 5 Temporal variation of fd, Kd and PM concentration during the 15 July 2010 event .................................................................................................................. 82 3 6 Intra event TVT) plots for each N fraction and TN with mass transport model fitted. ............................................... 83

PAGE 10

10 3 7 Event based cumulative performance (% removal) of the VCF for N and P fractions TN and TP as a function of cumulative treated volume. ...................... 84 3 8 Probability density function (pdf) of Total Inorganic Carbon (TIC), Total Organic Carbon (TOC), and the ratio of TOC/TIC ............................................. 85 3 9 Probability density function (pdf) of C:N and N:P ratio for 25 monitored storm events, including rainfall, runoff and effluent ..................................................... 86 4 1 plan view of the studied watershed and Locations on a Florida map ................ 106 4 2 pathways of nutrient from rainfall to treated effluent ......................................... 107 4 3 Probability density function (pdf) of pH, redox, TDS, alkalinity, CODd and DOC concentrations. ........................................................................................ 109 4 4 Probability density function (pdf) of NOx -N, TAN, ON, DN, DP and flow rate .. 110 4 5 Temporal variation of major N and P species for the August 12, 2008 event ... 111 4 6 Temporal variation of major N and P species for the July 08, 2008 event. ...... 112 4 7 Measured and modeled NOx -N and TAN concentration for the July 08, 2008 event with cumulative mass superimposed as lines ........................................ 113 4 8 Temporal variation and cumulative M V plot of pH, redox, TDS, alkalinity, CODd and DOC concentration for the July 08, 2008 event. .............................. 114 4 9 Correlation betw een CODd and DOC. .............................................................. 115 5 1 Plan view of the source area and t he approximate drainage area. ................... 140 5 2 Design of the radial cartridge f iltration (RCF) system. ...................................... 141 5 3 Water chemistry indices variation through the entire monitoring program. ....... 142 5 4 N, P species and CODd varation through the entire monitoring program. ........ 143 5 5 Temporal variation of water chemistry indices, and probability density function of t urbidity and DOC. ........................................................................... 144 5 6 Temporal variation of dissolved N, P species and CODd, ................................. 145 5 7 Transformation of dissolved N species based on Pourbaix ( Eh pH) diagram.. 146 5 8 Transformation of dissolved P species based on Pourbaix ( Eh pH) diagram.. 147 5 9 Comparison of water chemistry indices of retained storage in filter between the first three events (initial events) and the last three events (final events). ... 148

PAGE 11

11 LIST OF ABBREVIATIONS C Volumetric Runoff Coefficient cdf Cumulative Density Function CODd Dissolved Chemical Oxyge n Demand CODT Total Chemical Oxygen Demand DN Dissolved Nitrogen DO Dissolved Oxygen DOC Dissolved Organic Carbon DP Dissolved Phosphorus drain Durations Eff. Effluent EMC Event Mean Concentration ICPMS Inductively Coupled PlasmaMass Spectrometry Imax Pe ak rainfall intensity Inf. Influent (Runoff) IPRT Initial Pavement Residence Time k Scale a nd Shape Factors o f Gamma Distribution Max Maximum Min Minimum NOx -N Nitrate Nitrite N NTU Nephelometric Turbidity Units ON Organic Nitrogen pdf Probability D ensity Function PDH Previous Dry Hours PM Particulate Matter

PAGE 12

12 PR Percent Removal Prain Rainfall Depth PSD Particle Size Distribution Q50 Median Flow Rates Qmax Maximum Flow Rates Qmean M ean runoff flow rate RCF Radial Cartridge Filtration RPD Relative Perce nt Difference S. M. Standard Method s.u. Standard Unit SSC Suspended Sediment Concentration TAN Total Ammonia Nitrogen TDS Total Dissolved Solids TN Total Nitrogen TP Total Phosphorus Vrunoff Influent (runoff) volume Unsteadiness Mean 50 Median Cumulative Mass during a storm event Cumulative Volume during a storm event Standard Deviation

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Pa rtial Fulfillment of the Requirements for the Degree of Doctor of Philosophy NUTRIENT (N, P) TRANSPORT, PARTITIONING AND FATE IN URBAN RUNOFF By Hao Zhang May 2013 Chair: John J. Sansalone Major: Environmental Engineering Sciences High levels of nutr ients (N, P) from anthropogenic and biogenic sources through urban stormwater transport can cause eutrophication in the discharging areas In this dissertation the intra event and inter event partitioning, speciation, transport and transformation of N and P were examined across 4 0 runoff events and 13 retention periods on a paved urban source area. A radial cartridge filtration (RCF) system and a volumetric clarifying filtration (VCF) system were tested for the performance of nutrient reduction. Results fo und that PM deposited in the watershed is the main contributor of nutrients in runoff T N in dry deposition is strongly correl ated with the organic content of PM. Results found that dissolved N (DN) is the predominant fraction, while settleable N (25~75 m) held the lowest portion. F iner particles are preferred to be transported by runoff because it requires less stream power. M ass limited (firstorder) model is the predominant transport model for N fractions. F iner PM fractions are more abundant in runoff, w hile coarser PM fractions show more significant firstflush effect. Among DN species, NOx and ON are the two predominant species Mass transport behavior of D N

PAGE 14

14 species depends on the volume and duration of rainfall runoff event s. DP follows a mass l imited model for each event due to the limited sources of phosphorus T he VCF system can effectively reduce TN and TP concentrations by removing PM associated nutrients Higher removal efficiency was observed for coarser PM fraction. The RCF system showed 70% removal efficiency on P while the N removal efficiency is not significant N is the controlling nutrient in runoff therefore, reducing N loadings could effectively reduce the potential of eutrophication. During retention periods, an aerobic/anoxic zone is quickly created inside the RCF system. The variation of Water chemistry indices is related to the microbial respirations and decomposition of biogenic materials DN increased over the retention time indicating t he decomposition of biogenic materials is the prevalent process. TAN and DON increased while NOx decreased over time where multiple pathways could be involved. Pourbaix diagram indicates that t he dominant N species in rainfall, runoff and effluent is N2 via biological denitrification, while NH4 + is predominant in retained stormwater. Result found t he aerobic/anoxic cycle could promote N removal as the microorganisms has grown and evolved over the extended retention periods.

PAGE 15

15 CHAPTER 1 GLOBAL INTRODUCTION Clean water is critical for drinking, irrigation, transportation, recreation, fishing, and esthetic enjoyment. However, the increase of pollutant inputs in recent decades has degraded water quality of many nature water resources, causing many adverse effects such as loss of nature systems, th eir component species, and the amenities that they provide (U.S. EPA 1996). Nitrogen (N) is critical for the growth of terrestrial and vascular plants (Vitousek and Howarth 1991). However, eutrophication driven by N and phosphorous (P) from relatively wel l defined point and highly variable and unsteady nonpoint (diffuse) dis charges can lead to environment concerns that are no longer limiting for native vegetated communities. A significant load of N can be transported by stormwater runoff. For example, inputs of N from fertilizer to urban landscaped areas, biogenic loadings in runo ff from these areas to impervious surfaces and fossil fuel combustion in urban environs in conjunction with altered and unsteady rainfall runoff responses from these impervious urban systems are driving eutrophication in terrestrial and aquatic ecosystems (Vitousek et al. 1997a ). Several other origins of N in runoff are known as rainfall, atmosphere deposition, biogenic materials (leaves and branches) landscape maintenance byproducts, agriculture activities, animal manure, leachate from waste disposal sites, mines, oil fields, and construction sites (Pucket t 1994; Carpenter et al. 1998). These pollutants can be mobilized by runoff and effectively conveyed across urban surfaces as a function of flow rate during a storm. Urban land uses and design/construction practices such as the use of impervious pavement have significantly altered the relationsh ip between rainfall and runoff. The

PAGE 16

16 alteration of runoff response parameters by imperv ious surfaces leads to increased peak flow, increased volume and reduced response time that drives increased chemical and PM interactions and transport to receiving waters (USEPA 1993). N loadings above certain threshold values result in algal and nuisance plant blooms, fish mortality as a result of hypoxia, altered water chemistry, increased health risks for water supplies and loss of biodiversity and habitat (Smith et al. 1999; Taylor et al. 2005). Excess N loadings have become a leading cause of impairment in the USA (Carpenter et al. 1998; USEPA 1983; USEPA 2010). For over a decade, eutrophication has been identified with over 50% of impaired lakes and over 60% of impaired rivers in the USA (USEPA 1996b ; NRC 1993). Previous research has documented nutrient levels in runoff. Carpenter et al. (1998) reported that a total of 8.16 x106 Mg/yr of nitrogen were discharged into surface water based on two individual studies conducted in the United States. A nationwide study in the USA by Pitt et al. (2004) reported a median TN of 2.0 mg/L for runoff based on a dataset of 3770 individual events. In Florida, where this study is conducted, USEPA established a recommended TN criterion of 0.52 mg/L for runoff into lakes and reservoirs and 0.36 mg/L into rivers and streams (2000a, b). Criteria were based on the 25th percentile of a pool of randomly selected sites in the designated ecoregions. In late 2012 the promulgated numeric nutrient criteria (NNC) adjusted TN value to 1.27 mg/L for colored lakes and 1.87 mg/L for s treams in North Florida ( FDEP 2012 ). Urban runoff is a complex heterogeneous mixture that includes a wide gradation of particulate matter (PM), metals, nutrients, petroleum related organic compounds (Buckler and Granato 1999; Marsalek 1999). Rainfall runof f can mobilize and transport

PAGE 17

17 et al. 1998). PM is a potential concern not only because of the environmental and ecological issues related to solids, but also because many contaminants (including N) bind to the surfaces of these particles and are then transported into and through aquatic environments (Sansalone 2002). For a water quality perspective, PM with reactiv e sites and large surface to volume ratios can mediate partitioning and transport of chemical species while also serving as reservoirs for many reactive constituents (Sansalone et al. 1998). Therefore, it is important to understand that particulate deliver y and granulometry which play an important role in partitioning and separation of PM (Marsalek 1999). While the chemical species partition and distribute across the entire PM size gradation, the suspended PM is potentially more mobile and acutely bioavaila ble than coarser PM fractions that settle more easily (Cristina and Sansalone 2003). In runoff N partitions into DN and PN phases (Taylor et al. 2005). TN partitioning depends on adsorptiondesorption phenomena, microbial activity, leaching from biogenic m aterials, and water chemistry parameters including the concentration and distribution of N and PM (Shinya et al. 2003). Previous research has reported that the DN can represent 20% to 80% of TN (Vaze and Chiew 2004; HvitvedJacobsen and Yousef 1991; Hvitved Jacobsen et al. 1994; Taylor et al. 2005; Shinya et al. 2003). It is necessary to understand the partitioning of N in urban runoff to maximize the removal efficiency by targeting the dominant fraction or fraction that is most concern to receiving environm ents (Taylor et al. 2005).

PAGE 18

18 Besides, N also distributes across the particulate size gradation in rainfall runoff (Vaze and Chiew 2004). Knowledge of PN concentrations as a function of PM size fractions (suspended, settleable and sediment) or across the part icle size distribution (PSD) allows for the prediction of PN fate based on physical PM separation processes such as filtration and clarification. PM is a major source of N and a study by Vaze and Chiew (2004) characterized PN as a function of PM fractions The Vaze and Chiew study was based on a paved source area in Melbourne, Australia and reported that less than 15% of TN (by mass) was bound to PM with particle sizes greater than 300 m, eir results also indicated that DN ranged from 20 to 50% of the TN in stormwater and that most N was rainfall duration, intensity, antecedent dry weather periods, land use, season, vegetation, average daily traffic (ADT), atmospheric deposition, watershed area and surface, slope, PM granulometry and soil parameters (Sansalone and Buchberger 1997; Pitt et al. 2004; Huber et al. 2008). N is introduced into the aquatic environment in a number of different chemical forms. Total Nitrogen (TN) refers to the summation of inorganic N (IN) and organic N ( ON). The TIN includes three species: NitrateN (NO3 -N), Nitrite N (NO2 -N), Total Ammonia N (TAN, which is the summation of NH4 +N and NH3N). Colorimetric method is usually used to determine the concentration of inorganic nitrogen species. ON consists of a complex mixture of compounds including amino acids, amino sugars and proteins. Total Kjeldahl Nitrogen (TKN) is a commonly used for determining ON, where a persulfate digestion is performed to oxidize all nitrogen form. TKN includes ON and

PAGE 19

19 TAN and hence ON can be calculated by the difference of TKN and TAN. While the proportion of IN over ON is varied by land uses and hydrologic conditions (Harris et al. 1996; Tufford et al. 2003; Kaushal et al. 2006; Pellerin et al. 2006), high level of ON is often observed in the first flush of runoff from highly urbanized areas (Flint and Davis 2007). The speciation of N in runoff is a highly dynamic process that is driven by hydrology, and coupled to water chemistry. Many p hysicochemical parameters such as temperature, dissolved oxygen (DO), pH, redox potential, conductivity, BOD and COD are involved (Petaloti et al. 2004). In order to addres s those environmental and ecological issues related to the pollutants transported in urban runoff, several best management practices (BMPs) have been developed in the past few decades to control the quality and quantity of stormwater in urban areas. Three major pathways in N cycle commonly utilized for N removal are assimilation, adsorption and denitrification. Coupling with these mechanisms, conventional stormwater control measures (SCM) include dry ponds and wet ponds were developed, where complex microbi ological reactions and change of oxidation state were involved. The hydrologic residence time is critical for these SCMs (Kaushal et al. 2008; Klocker et al. 2009).Alternative SCMs such as filters, green roofs, bioretention, vegetated channels, constructed wetlands, permeable pavements that utilized similar concept were also developed (Collins et al. 2010). Each SCMs has its advantage and shortage, therefore, various unit operations and processes (UOP) that integrated adsorption, filtration, vegetation uptake and low impact development (LID) were developed to remove N.

PAGE 20

20 A volumetric clarifying filtration (VCF) system is an UOP that integrates filtration, clarification and retention together and hence can be utilized in particulate matter (PM) separation and hydrologic attenuation. Since the hydrologic condition is unsteady during each storm, the treatment behavior is also considered nonstationary. Study the short term treatment behavior (intraevent) can assist the design such as the size of the treatment sy stem. In addition, studying the longterm treatment behavior can help people better understand the mechanisms such as the filter ripened as a function of treated volume. This study focuses on the TN treatment efficiency of a VCF system to direct rainfall r unoff loadings from an urban paved source area watershed with biogenic material as the major nutrient sources.

PAGE 21

21 CHAPTER 2 PARTITIONING AND FIRST FLUSH OF NITROGEN IN SOURCE AREA URBAN RAINFALL RUNOFF A bstract The urban surface is a largely impervious inter face across which nitrogen (N) in runoff is transported and partitions between the dissolved and particulate matter (PM) phases. Dissolved nitrogen (DN) is often the dominant phase during runoff transport impacting acute phenomena such as bioavailability. In contrast, the transported PM bound fractions of particulate nitrogen (PN) impact chronic phenomena such as accretion of N in drainage systems and leaching thereof. Intraevent partitioning and transport is examined across 14 runoff events loading a paved urban source area. The source area is representative of developed land that supports motor vehicle activities. N fractions are categorized as suspended PN for particle sizes between 0.45 ~ 25 m, settleable PN in the range of 25 ~ 75 m, sediment for sizes greater than 75 m; and dissolved (DN) is the filtrate passing through a 0.45 m membrane. The median event based total nitrogen (TN) as the sum of DN and PN is 4.5 mg/L. For TN the DN phase has a median dissolved fraction ( fd) of 62%.For PN, the s uspended fraction contained 0.72 mg/L of N, the settleable had 0.29 mg/L and the sediment had 0.78 mg/L. Logistic regression is utilized to differentiate the transport behavior of N fractions utilizing hydrologic parameters. The transport of N fractions i s masslimited (a firstflush) for most events as TN exhibited a large initial flush. While distribution of N with PM size was complex, N was linearly correlated to the PM volatile fraction. N of dry deposition PM was higher and the size gradation coarser than for runoff PM due to intraevent leaching and redeposition of PM.

PAGE 22

22 Introduction N is a limiting nutrient for terrestrial and vascular plants (Vitousek and Howarth 1991). However, eutrophication driven by N and phosphorous (P) from relatively well def ined point and highly variable and unsteady nonpoint (diffuse) discharges can lead to environments that are no longer limiting for native vegetated communities. For example, inputs of N from fertilizer to urban landscaped areas, biogenic loadings in runo n from these areas to impervious surfaces and fossil fuel combustion in urban environs in conjunction with altered and unsteady rainfall runoff responses from these impervious urban systems are driving eutrophication in terrestrial and aquatic ecosystems ( Vitousek et al. 1997).These sources of N in runoff also include nitrates in rainfall, leaf fall, landscape maintenance byproducts such as grass clippings, soil or anthropogenic PM which can be mobilized by runoff and effectively conveyed across urban surfaces as a function of flow rate during a storm (Puckett 1994). Urban land uses and design/construction practices such as the use of impervious pavement have significantly altered the relationship between rainfall and runoff. Furthermore, the alteration of r unoff response parameters by impervious surfaces leads to increased peak flow, increase volume and reduced response time that drives increased chemical and PM interactions and transport to receiving waters (USEPA 1993). N loadings above certain threshold values result in algal and nuisance plant blooms, fish mortality as a result of hypoxia, altered water chemistry, increased health risks for water supplies and loss of biodiversity and habitat (Smith et al. 1999; Taylor et al. 2005). Excess N loadings have become a leading cause of impairment in the USA(Carpenter et al. 1998; USEPA 2010). For over a decade, eutrophication has been

PAGE 23

23 identified with over 50% of impaired lakes and over 60% of impaired rivers in the USA(USEPA 1996). Previous research has documented nutrient levels in runoff. Carpenter et al. (1998) reported that a total of 8.16 x106 Mg/yr of nitrogen were discharged into surface water based on two individual studies conducted in the United States. A nationwide study in the USA by Pitt et al. (2004) reported a median TN of 2.0 mg/L for runoff based on a dataset of 3770 individual events. In Florida, where this study is conducted, USEPA established are commended TN criteria of 0.52 mg/L for runoff into lakes and reservoirs and 0.36 mg/L into rivers and streams (2000a, b). Criteria were based on the 25th percentile of a pool of randomly selected sites in the designated ecoregions. In late 2010, the promulgated numeric nutrient criteria (NNC) adjusted TN value to 1.27mg/L for colored lakes and 1.87 mg/ L for streams in North Florida (USEPA 2010). In runoff N partitions into DN and PN phases (Taylor et al. 2005). TN partitioning depends on adsorptiondesorption phenomena, microbial activity, leaching from biogenic materials, and water chemistry parameters including the concentration and distribution of N and PM (Shinya et al. 2003). Previous research has reported that the DN can represent 20% to 80% of TN (Vaze and Chiew2004; HvitvedJacobsen and Yousef 1991; HvitvedJacobsen et al. 1994; Taylor et al. 2005; Shinya et al. 2003). Knowledge of PN concentrations as a function of PM size fractions (suspended, settleable and sediment) or across the particle size distribution (PSD) allows for the prediction of PN fate based on physical PM separation processes such as filtration and clarification. PM is a major source of N and a study by Vaze and Chiew (2004) characterized PN as a function of PM fractions. The Vaze and Chiew study was based

PAGE 24

24 on a paved source area in Melbourne, Australia and reported that less than 15% of TN (by mass) was bound to PM with particle sizes greater than 300 m, while approximately 50% of indicated that DN ranged from 20 to 50% of the TN in stormwater and that most N was rainfall duration, intensity, antecedent dry weather periods, land use, season, vegetation, average daily traffic (ADT), atmospheric deposition, watershed area and surface, slope, PM granulometry and soil parameters (Sansalone and Buchberger1997; Pitt et al. 2004; Huber et al. 2008). Res ults of these studies have provided knowledge of specific N indices. Building on these results, this study investigates the first flush transport, the distribution and partitioning of N phases; each as a function of the variable and unsteady intraevent as well as event based hydrology. Objectives There were four major objectives of this study. The first objective was to evaluate event based DN and PN concentrations and variability thereof for a source area representative of small paved urban watershed subject to biogenic loadings from landscapebased vegetation. The second objective was to examine partitioning of TN between DN and PN as well as the distribution of PN across PM size classes and the PSD. The third objective was to investigate the transport of TN and TN fractions with respect to a first flush phenomenon for each of these fractions. Finally, the fourth objective was to develop the TN distribution as a function of granulometry and compare these distributions between dry deposition PM and PM transported in runoff.

PAGE 25

25 Methods and Materials Watershed Description The watershed is an asphalt Gainesville (GNV), FL. The source area is approximately 500 m2 with 76% of the watershed consisting of asphalt pavement while 24% is raised vegetated islands. The plan view of the wat ershed is illustrated in Figure 2 1.The watershed includes a northsouth pavement with 1.5% slope and a 3% east west pavement slope. Flows are intercepted by a catch basin that is modified so all flows are piped through the catch basin and into a Parshall flume at a 6% slope. The Parshall flume has a 25.4 mm throat and a calibrated ultrasonic sensor (Shuttle Ultrasonic level transmitter, MJK Inc.) for flow monitoring at a monitoring frequency of 1Hz (1 s1). Rainfall depth is measured near the catchment by a tipping bucket rain gauge that has an accuracy of 0.254 mm (0.01 inch). The average daily traffic loading (ADT) to the watershed is low (approximately 700 vehicles/day) and traffic generated N and PM are low as compared to biogenic N and PM that is transported from the raised vegetated islands onto the impervious pavement. The age of the pavement is approximately 10 years, and although the pavement remains serviceable, it is oxidized and cracked. The climate of Gainesville is humid and subtropical with a distinctive dry season (October ~ May) and wet season (June ~ September). The mean annual rainfall is 1130mm (NCDC, 2009). Sampling Methodology Samples were manually taken from a sampling drop box which collected the freefalling discharge of the Parshall flume. Previous studies (Liu et al. 2010; Clark et al. 2009) have demonstrated that automated samplers do not representatively sample

PAGE 26

26 coarser PM and biogenic material. The manual sampling method was utilized to capture the biogenic material such as leaves and biogenic detritus. To demonstrate the representativeness of this sampling method, a mass balance of PM (all PM > 0.45 m) was conducted by quantifying influent PM, effluent PM and PM separated by a unit operation. Replicated runoff samples were c ollected manually at closely spaced intervals ranging from two to ten minutes throughout each event. Approximately 10 sets of samples were collected for each event. At each sampling time, one replicated 4 L, one replicated 1 L and two replicated 0.5 L samp les were collected using widemouth polypropylene (PP) containers. Samples were analyzed for PSDs or fractionated and preserved within 6 hours. The through a #200 sieve. Detailed procedures of the physical granulometric separation and laser diffraction analysis were shown in Sansalone and Kim (2008) and Kim and Sansalone (2008). The 0.5 L samples were used for water chemistry analysis and PSDs. Finally,1 L rainfall samples were also collected in elevated Pyrex trays placed adjacent to the rain gau ge located 180 meter southwest of the watershed. TN A nalysis M ethod TN analysis was performed through persulfate digestion and spectroscopic absorbance measured at 410 nm with a Hach DR5000 spectrophotometer. Calibration curves relating absorbance to TN concentrations were made from serial dilutions of a TN standard solution for each reagent lot with a coefficient of determination (R2) larger than 0.99. Validation was based on a soil standard (Nutrients in Soil, No. D061 542, Environmental Resource Associates) and was found to be within a 95% confidence interval. TN recovery based on the st udy results and the certified standard was 96%.

PAGE 27

27 TN concentration often varies by several magnitudes during a runoff event. In order to compare different events using a single index, an event mean concentration (EMC) is commonly applied (Huber 1993). The E MC is defined as the total constituent mass load divided by total volume. Concentrations of a constituent are most commonly log normal distributed and the median concentration is benchmarked in this study, the commonly utilized EMC is also tabulated (Van B uren et al. 1997; Berretta and Sansalone 2011; Liu and Sansalone 2010;Strecker et al. 2001;Kim and Sansalone 2010). Hydrologic Transport Variables Transport of N fractions in runoff is influenced by hydrologic and watershed parameters including total volum e (TV), runoff duration (DT), shape and scale of the hydrograph, previous dry hours (PDH), and unit stream power ( P u ) (Sheng et al. 2008). A hydrograph can be modeled as a cumulative gamma distribution with the shape ( k) and scale ( ) gamma parameters being utilized as physical indices of the hydrograph. Parameters are estimated by minimizing the sum of squared errors (SSE), while maximizing the R2 between modeled and measured data. The incremental and cumulative gamma distributions are shown below, where x represents the elapsed time. ) ( ) / ( ) () / ( 1k e x x fx k ( 2 1) xdx x f x F0) ( ) ( ( 2 2) The stream power ( ) is defined as the rate of energy dissipation resulting from friction between the flow and the channel (Bagnold, 1966). When the flow has reached

PAGE 28

28 equilibrium and is moving at constant velocity, the unit stream power ( P u ), which is the stream power for unit width of flow. WgQSWpu ( 2 3) In this expression, Pu is the unit stream power (W/m2 (W/m), is the density of water (1000 kg/m3g is acceleration due to gravity (9.81 m/s2), Q is runoff hydraulic discharge (m3/s), S is the pavement slope as shown in Figure 2 1 and W is the width of the watershed (6.0 m for the studied watershed). TN P artitioning between D issolved and PM F ractions TN partitioning between dissolved and particulate phases is characterized by a dissolved fraction, fd a nd partitioning coefficient, Kd. The dissolved fraction is expressed as: p d d dM M M f ( 2 4) In this expression, Md is the dissolved mass, and Mp is the PM mass of TN. The conventional equilibrium partitioning coefficie nt ( Kd) assumes a linear isotherm between dissolved and PM phases. dsdCCK ( 2 5) In this expression, Cs is the PM based TN concentration (mg/kg of dry mass); and Cd is the dissolved TN concentration (mg/L). There is an inverse relationship between the dissolved fraction ( fd) and the partition coefficient ( Kd) as shown in Equation 2 6 Here, Cp is the PM fraction concentration, which is commonly taken as the suspended PM fraction:

PAGE 29

29 pddCkf11 ( 2 6) Mass T ransport of N F ractions N mass transport in this system is categorized as either mass limited (e.g., a classical mass first flush) or flow limited (Sheng et al. 2008). With flow limitations, mass delivery is proportional to runoff volume. This model is expressed in Equation 2 7 and 2 8 : ( 2 7) ( 2 8) In this expression, M is TN mass, K is a washoff coefficient [M/T ] t is time [T] and Q is flow [L3/T] MT is mass delivered [M], k0 is the washoff constant [M/L3], and VT is the volume [L3]. In contrast, mass li mited transport occurs when the available mass is the limiting factor, producing an initially disproportionate mass transport with respect to runoff volume. ( 2 9) )1(10tVkTeMM ( 2 10) In this expression, k1 is the firsto rder washoff coefficient [ L3] M0 is the available mass of TN [M] on the pavement, MT is mass delivered [M] and VT is volume [L3] Logistic Regression Categorical analysis by logistic regression is utilized to classify transport of TN fractions as flow or mass limited. Logistic regression is utilized to predict the probability

PAGE 30

30 of occurrence of an outcome (mass or flow limited) by fitting the transport parameters to a logistic function. A logit variable ( z) is defined as the total contribution of all variables used in the model. As suggested by Sheng et al. 2008 variables are runoff duration ( DT ), total volume ( TV ), hydrograph parameters ( and ), previous dry hour ( PDH) and stream power ( Pu). The form of the logit variable z is shown in Equation 2 11 up PDH TV DT z6 5 4 3 2 1 0 ( 2 11) In this expression 0 is the intercept and 126 are the regression coefficients. Then, the logistic function, f (z), is applied to calculate the probability of occurrence for a specific transport behavior. In this study, an f (z) of 1 is defined as mass limited whereas an f (z) of 0 is a flow limited event. The logistic function is expressed in Equation 2 12. 1 ) ( z ze e z f ( 2 12) Each of the model coefficients are calibrated by a number of events with a definitive value of f (z) (1 or 0). Once calibrated with the watershed data, the model can be utilized solely with hydrologic data to differentiat e mass limited and flow limited transport. First Flush T ransport A first flush model is defined by a disproportionate mass delivery during the initial stages of a runoff event (Sansalone and Cristina, 2004). The first flush phenomenon has been studied in urban runoff for PM fractions (Sansalone et al. 1998, Sansalone and Kim 2008), metals (Sansalone et al. 1997), phosphorus (Ma et al. 2010) and chemical oxygen demand, COD (Kim and Sansalone 2010). This study examines the firstflush of TN and fractions th

PAGE 31

31 first flush (Larsen et al. 1998). Dry D eposi tion TN A nalysis The granulometric distribution of TN mass is examined as a function of PM size for dry deposition samples collected from the watershed before rainfall and after runoff. Samples are dried at 40 C and sieved into 15 gradations ranging from 2000 m to < 25m. The sieving protocol follows ASTM Standard Method D42263 with additional sieve sizes (ASTM 1998 ; Sansalone et al. 1998). TN analysis is performed for each gradation to determine the PM based TN concentration. The volatile fraction (Standard Method 2540E, APHA 1992) as an index of organic content (biogenic loadings) is measured for each gradation in order to correlate the volatile fraction and PM based TN concentration. Results and Discussion Rainfall Runoff E vents Fourteen runoff events between 16 May and 23 October were captured and hydrologic parameters are summarized in Table 2 1. A mass balance on PM generated a 95% recovery across the monitoring campaign. Rainfall depths of monitored events are compared with the rainfall depth distr ibution for a 10year period in GNV (NCDC, 2009). Figure 2 2 illustrate that the captured events largely occurred within the upper 50% of single event rainfall depths for GNV, ranging from the 45th to 94th percentile of the historical data. As shown in Figure 2 2, the relationship between single event rainfall depth and intensity is exponential. The flow rates for the 14 events fit a lognormal distribution (p < 0.05), with a median value of 0.51 L/s. The events are classified as low,

PAGE 32

32 medium and high flow r ate events based on the historical frequency distribution for flows from the watershed. Each category has equal odds of occurrence, as shown in the bottom plot in Figure 2 2. As tabulated in Table 2 1, the event based median value of the volumetric rainfal l runoff coefficient (C) is 0.49. Similar values have been observed for other small urban watersheds (Sansalone et al. 2009).The initial pavement residence time (IPRT), i.e., the lag time between the initial wetting of the pavement surface and the generat ion of runoff (Sansalone et al. 1997; Sansalone et al. 1998), ranges from 2 to 12 minutes with a median value of 4 minutes. The event based concentrations (EMCs) and range of intraevent concentrations for each TN fractions are summarized in Table 2 2 fo r the 14 events. The median concentration of TN is 4.52 mg/L, which is approximately four times higher than the discharge limit of TN for Florida colored lakes (1.27mg/L) (USEPA, 2010). Rainfall samples are also listed in Table 2 2, with a median TN concentration of 0.58mg/L. TN in the atmospheric nitrogen that contributes to TN runoff is relatively constant. Partitioning of TN between D issolved and PM F ractions As shown in Table 2 2, DN a ccounts for over 50% of TN. In the particulate phase the suspended and sediment PM fractions have higher TN concentrations (median value: 716 and 778 g/L, respectively) than the settleable fraction (median value: 298 g/L). The concentration rat io of TN in suspended, settleable and sediment fractions was approximately 2:1:3. The PM based TN mass fractions are plotted as percentages on an event basis, and the EMCs of each fraction are expressed as box whisker plots,

PAGE 33

33 as shown in Figure 2 3. The res ults indicate that the proportion of each TN fraction is not constant and changes significantly from one event to the next. The probability density function (pdf) of each TN fractions and the flow rates for the 14 events was modeled. As shown in Figure 2 4, each TN fraction and the flow rates followed lognormal distributions, with model parameters and measured values not showing significant difference at the 95% confidence level. The pdf results also demonstrate that the dissolved fraction contained the l argest portion of the TN value, while the settleable fraction comprised the smallest portion. For the same watershed, the N concentrations of each fraction which varied by at least 3 to 4 magnitudes illustrates the high variability and size dispersivity of N fractions in rainfall runoff. The dissolved fractions, fd, and equilibrium distributions, Kd, for the partitioning of TN between dissolved and solid phases have also been evaluated. As shown in Figure 2 5, fd and Kd values are plotted as nonparametric results. The nonparametric results are organized in descending order of the median flow rate (Q50) for each event. The median fd of each category is designated with dotted lines. The results show that fd values tend to be higher for higher flow rate stor ms. According to Figure 2 5, high flow rate storms had a median fd value of 0.73, while that of medium and low flow rate storms were 0.56 and 0.52, respectively. Kruskal Wallis oneway ANOVA analysis indicates that the difference in fd values for each category was statistically significant (p < 0.05). A similar test was performed for Kd values, though, Kd values did not show significant differences between categories (p> 0.05). In contrast to an event basis, during the passage of the hydrograph peak, fd res ults shown in Figure 2 6 for the 10 June 2008 event illustrate an inverse relationship

PAGE 34

34 with flow rate. For the falling limb of the hydrograph fd generally increases since less coarse PM is transported and longer flow residence times increase dissolution. Beyond the response to the hydrograph peak the Kd values gradually increase during the event regardless of the flow rate. The cumulative volume distribution for the 10 June 2008 event fits a gamma distribution with an R2 of 0.97. The gamma distribution pa rameters physically represent the shape of the intraevent flow distributions and the scale of flow rates as an index for low to high flow. Categorical A nalysis for T ransport of N F ractions A logistic regression was applied to differentiate transport of N fractions utilizing hydrologic parameters. Values of the six selected hydrologic parameters are listed in Table 2 3. Four logistic regression models were calibrated for each N fractions based on 9 selected events. The remaining 5 events were then calculate d by the calibrated models. The overall accuracy of each model is larger than 90%. The results are shown in Table 2 4, with each event designated as either mass limited or flowlimited. The results indicate that logistic regression can statistically differ entiate transport of N fractions for the urban source area watershed under consideration. Mass Transport of N F ractions The modeled cumulative N mass verse volume plots for each fraction are shown in Figure 2 7. Exponential growth and linear models were f it to each fraction with a 95% confident interval as shown by the shaded area. The model parameters and R2 are displayed in each plot. Mass limited events were more frequent than flow limited events. Sediment and settleable N had smaller K1 values than eit her suspended or dissolved N, indicating that the first flush effect was more significant for coarser PM fractions, which corroborates previous research (Sheng et al. 2008). For flow limited events, K0

PAGE 35

35 obtained a higher value in the dissolved fraction than in any particulate fraction, indicating a faster transport rate of N in the dissolved phase than in the particulate phase. Figure 2 7 also indicates that for the larger diameter PM fractions (sediment and settleable N), mass limited behavior generally occurred in high intensity storms, while for smaller diameter PM fractions (suspended and dissolved N), mass limited behavior was more frequently observed in low intensity storms. The reason for the difference is that transport of N for large diameter PM de pends more on the hydraulic power of runoff, while for small diameter PM, dissolution or disaggregation during the residence time on the catchment are mechanistically involved. In order to examine the first flush of N fractions, dimensionless relationships between cumulative mass and cumulative volume were investigated for all 14 events and plotted in Figure 2 8. All the fractions demonstrated a clear first flush behavior since all the correlation curves were above the 1:1 bisector line. According to Figure 2 8, the first flush behavior for coarser PM fractions is more significant than for finer PM fractions since the curves of coarser PM are further away from the 1:1 line. Distribution of N over P article S ize G radation The granulometric distribution of N m ass concentration for dry deposition PM is shown in Figure 2 9. As particle diameter decreased from 2000 to 200m, N concentrations decreased from approximately 8000 to a local minimum of 2000 mg/kg, after which N concentration increased to a plateau of 2500 mg/kg as particle diameter further decreased. This result is related to the composition of each PM gradation. The volatile fraction, which is indicative of organic content, has the same trend as N. N concentration for different PM gradations showed a st rong correlation with the volatile

PAGE 36

36 fraction suggesting that N concentration is highly correlated to the organic content of dry deposition PM. As shown in the middle plot the linear relationship had a R2 of 0.96. For the sake of comparison, runoff N mass concentrations were elaborated on an event basis for this watershed by the probability density function (pdf) for each PM fraction, as shown in Figure 2 10. Each N fraction was fitted to a lognormal distribution. The watershedmedian value of suspended N wa s higher than that of settleable and sediment fractions and had a larger standard deviation. This result indicates that the composition of suspended PM is more heterogeneous. Due to the larger specific surface area of suspended PM these particles are more reactive and subject to greater re partitioning and dissolution. Particle size distributions (PSDs) for both dry deposition samples and watershedmedian runoff PM are plotted in Figure 2 9.The d50 decreased from 280 to 154 m going from dry deposition to runoff, indicating that runoff did not deliver the coarsest fraction of the PSD. This accreted PM represents a potential and leachable source for N during a subsequent event. A portion of the coarse gradation is trapped at the paved watershed despite direct conveyance from the watershed. The results indicate that only 10% of the dry deposition PM is finer than 100 m, while 40% of runoff PM is finer than 100 m. Ying et al. demonstrated similar result in which 10% of dry deposition was finer than 100 m, while over 60% of runoff PM was finer than 100 m. Although PM larger than 400 m has a higher N concentration, t he relative mass fraction transported in runoff is lower which attenuates the intraevent N contribution from this coarse PM fraction.

PAGE 37

37 Conclusion This study investigated the partitioning, transport and distributions of N in urban runoff based on a vegetated 500m2 paved parking facility in Gainesville, FL (GNV).The major sources of N were identified as biogenic materials. Intra event samples from 14 rainfall runoff events in 2008 were collected and analyzed. Each sample was fractionated into four categories: median TN concentration in the studied watershed was 4.52 mg/L, which is about4 times higher than Floridas fresh water lakes recommended discharge level (USEPA, 2010). Among each fraction, dissolved N occupied the highest percentage (> 50%) while N bound to settleable particles held the lowest. Dissolved fractions ( fd) tend to be higher for higher flow rate events. The median value of fd for the 14 events was 0.62. N mass transport behavior in runoff is either flow limited (zero order) or mass limited (first order). A categorical analysis, logistic regression, is utilized to differentiate the transport behavior utilizing hydrologic parameters. For the 14 events, TN transport was predominantly mass limited for each N fraction, while a few flowlimited events were also observed. For larger PM (sediment N), mass limited behavior generally occurred in high intensity storms, while for su spended and dissolved N mass limited behavior was more frequently observed in low intensity storms. Normalized cumulative mass verses volume curves showed clear first flush (disproportional mass delivery during the initial stages of a runoff event) effects for every N fraction, and N for coarser PM showed stronger first flush effects than the finer fractions. The granulometric distribution of N mass concentration for dry deposition samples showed a trend of first decrease and then increase as particle siz e

PAGE 38

38 decreased. The median value of TN in dry deposition was 2648 mg/kg. While N illustrated a complex distribution with PM size, N associated with PM was strongly correlated with the volatile fraction, and index of organic content. Coarser PM in runoff tend to have lower N mass concentrations relative to dry deposition PM as a result of leaching to the dissolved phase. PM in runoff was also finer than dry deposition as a result of intraevent deposition along the flow path. Suspended N in runoff showed larger variations due to the higher surficial reactivity of suspended PM. Although PM > low compared to inter event dry deposition PM, and as a result attenuates the N contribution from this coarse PM. PM deposited in the watershed and drainage system is a source of N and inter event practices of pavement and drainage system cleaning provide PM and N source control.

PAGE 39

39 Table 2 1 H ydrologic parameters of 14 rainfall runoff events in Gainesville, Florida. 2008 Events IPRT PDH dra in Imax Prain Vrunoff Qmax Qavg C (min) (hr) (min) (mm/hr) (mm) (L) (L/s) (L/s) 16 May 12 960 62 91.4 10.7 2213 8.75 0.90 0.41 03 Jun 8 432 69 15.2 2.0 308 0.75 0.11 0.30 10 Jun 2 168 94 152.4 30.7 8002 10.96 1.61 0.52 21 Jun 6 144 127 61.0 6.6 1132 3.98 0.34 0.34 8 Jul 4 141 197 167.6 74.2 31268 13.17 2.14 0.84 15 Jul 9 66 79 167.6 62.2 22438 13.07 3.60 0.72 29 Jul 6 309 35 30.5 5.6 1411 3.69 0.55 0.50 8 Aug 3 191 42 30.5 3.0 474 2.16 0.14 0.31 12 Aug 7 93 90 45.7 16.3 3866 3.92 0.62 0.48 19 A ug 4 117 57 45.7 4.3 1461 6.78 0.81 0.68 10 Sept 4 264 56 30.5 8.1 1569 1.96 0.47 0.39 20 Sept 3 234 36 45.7 2.8 509 1.09 0.24 0.36 8 Oct 3 431 180 30.5 5.8 1557 2.75 0.14 0.53 23 Oct 3 334 56 30.5 3.6 978 1.46 0.29 0.55 50 4 213 66 45.7 6.2 1509 3.81 0.51 0.49 5 277 84 67.5 16.9 5513 5.32 0.85 0.50 3 229 51 54.7 23.1 9411 4.43 0.98 0.16

PAGE 40

40 Table 2 2 Event based N concentrations of suspended, settleable, sediment, dissolved fractions and the sum of each fraction expressed as an EMC 2008 Events Sediment N [g/L] Settleable N [g/L] Suspended N [g/L] Dissolved N [g/L] TN Rain TN [g/L] 16 May 913 (2116 30) 653 (2338 49) 717 (3667 46) 3193 (13316 1362) 5476 (19450 1487) 770 03 Jun 5608 (19532 5) 689 (2086 2) 715 (971 324) 3955 (4853 1529) 10967 (27442 2041) 540 10 Jun 2566 (7879 53) 477 (2463 33) 1074 (1808 250) 4001 (10098 1276) 8119 (22055 2851) 570 21 Jun 1034 (1911 39) 232 (345 18) 614 (1293 144) 3993 (5098 2799) 5 873 (8554 3357) 710 08 Jul 172 (1077 21) 65 (316 15) 1011 (3462 128) 4469 (32885 801) 5704 (33419 1358) 580 15 Jul 120 (528 10) 53 (274 12) 246 (545 48) 968 (2821 64) 1386 (3847 342) 740 29 Jul 998 (2056 14) 1129 (5090 18) 1538 (2318 455) 2577 (4909 112 1) 5481 (13974 2032) 820 08 Aug 816 (3083 27) 89 (529 6) 2000 (4344 1090) 2150 (4790 944) 5054 (12747 2530) 960 12 Aug 672 (2106 72) 160 (422 16) 461 (1078 60) 937 (6078 240) 2231 (8498 1165) 570 19 Aug 1263 (4999 37) 644 (956 2) 197 (419 12) 1886 (3683 868) 3989 (9208 1032) 470 10 Sept 740 (221255) 254 (48226) 195 (313814) 1063 (4719561) 2252 (8704853) 480 20 Sept 383 (1655 90) 341 (756 56) 832 (1294 272) 1560 (336 745) 3117 (6759 1690) 570 08 Oct 468 (710 52) 450 (893 59) 1268 (4668 102) 988 (4 362 740) 3174 (10131 1263) 600 23 Oct 496 (2054 73) 127 (512 26) 157 (494 26) 1660 (5581 988) 2440 (8641 1208) 520 50 778 298 716 2150 4522 575 1201 329 788 2433 4703 646 1414 225 550 1267 2623 143

PAGE 41

41 Table 2 3 Hydrologi c indices for logistic regression (LR) of the 14 rainfall runoff events monitored in Gainesville, FL. 2008 Events DT TV Hydrograph Gamma Parameters PDH pu (min) (L) k (hr) (W/m 2 ) 16 May 62 2213 37.7 0.5 960 0.34 03 Jun e 69 308 4.6 2.3 432 0.03 10 Ju n e 94 8002 0.7 22.2 168 0.43 21 Jun e 127 1132 8.1 0.8 144 0.16 8 Jul y 197 31268 3.6 8.5 141 0.52 15 Jul y 79 22438 12.6 3.2 66 0.51 29 Jul y 35 1411 2.0 6.7 309 0.14 8 Aug ust 42 474 3.1 2.4 191 0.08 12 Aug ust 90 3866 14.1 1.6 93 0.15 19 Aug ust 57 1461 9.7 0.6 117 0.27 10 Sept ember 56 1569 10.5 2.9 264 0.08 20 Sept ember 36 509 1.6 4.5 234 0.04 8 Oct ober 180 1557 9.2 0.7 431 0.11 23 Oct ober 56 978 1.6 13.6 334 0.06 84 5513 8.0 5.8 277 0.21 50 66 1509 4.1 2.8 213 0.15 51 9411 9.6 6.5 229 0.17

PAGE 42

42 Table 2 4 Differentiation of the mass transport behavior of each TN fraction for the 14 monitored rainfall runoff events The differentiation is based on the physical modeling and logistic regression (LR) of the measured data. PTC: physical transport criteria. 2008 Events Sediment N Settleable N Suspended N Dissolved N PTC LR PTC LR PTC LR PTC LR Calibration 03 Jun 10 Jun 21 Jun 8 Jul 8 Aug 8 Oct 23 Oct 10 Sept 20 Sept Calculation 16 May 15 Jul 29 Jul 12 Aug 19 Aug Classification Accuracy 93% 93% 93% 100% limited event limited event

PAGE 43

43 Table 2 5 Watershed geometric characteristics, affiliated to Figure 2 1. Area N S slope E W slope tc ADT Pavement Condition C ~500 m2 1.50% 3.00% 2 12 min ~700 Asphalt and Vegetated 0.3 0.7

PAGE 44

44 Figure 2 1 Plan view of the studied catchment in Gainesville, FL with approximate w atershed area superimposed. The geometry of the studied catchment is summarized as well, where tc, ADT, and C are the time of concentration, average daily traffic, and volumetric rainfall runoff coefficient, respectively.

PAGE 45

45 Rainfall Depth, d (mm) 020406080100 Rainfall Intensity (mm/hr) 04080120160200 Measured Modeled f = a (1 e b [d] )a = 169.8b = 0.07r2 = 0.98 Rainfall Depth, d (mm) 0.010.11101001000 Cumulative Frequency distribution, % 0.00.20.40.60.81.01.2 a.b.08 July15 July10 June12 August16 May29 July10 September08 August03 June23 October21 June20 September08 October Runoff, Q50 (L/s) 0.1110 Cumulative Frequency Distribution (%) 020406080100 6733 HighMediumLow08 July15 July10 June12 August16 May29 July10 September08 August03 June23 October21 June20 September08 October19 August19 Augustc. Figure 2 2 a ) Rainfall frequency distribution for Gainesville, FL based on 1999 2008 rainfall data with the monitored rainfall runoff events superimposed. b ) Relationship between rainfall intensity and depth. c ) Runof f flow rate frequency distribution of 14 monitored rainfall runoff events. Q50 is the median flow rate calculated on an event basis.

PAGE 46

46 15 Jul. 08 Jul. 10 Jun. 16 May 19 Aug. 12 Aug. 29 Jul. 10 Sept. 21 Jun. 23 Oct. 20 Sept. 08 Oct. 08 Aug. 03 Jun. PM-based N mass fractions (%) 0 20 40 60 80 100 Suspended Settleable Sediment Sediment Settleable Suspended Dissolved N fractions on an event basis [mg/L] 0.01 0.1 1 10 TN (mg/kg) 900 1000 1100 1200 1300 1400 95% CI (upper limit: 1329 mg/kg) 95% CI (lower limit: 990mg/kg) soil standard (1160 mg/kg) Figure 2 3 PM based N mass fraction for each event and nonparametric plots for N fraction s as EMCs for the 14 monitored rainfall runoff events

PAGE 47

47 N [mg/L] 10-310-210-1100101102 pdf (%) 0 4 8 12 16 n = 180 = 3.37 = 2.19 pdf (%) 0 3 6 9 12 n = 180 = 0.11 = 3.80 N [mg/L] 0.001 0.01 0.1 1 10 100 pdf (%) 0 3 6 9 12 pdf (%) 0 3 6 9 12 n = 180 = 0.50 = 3.38 n = 180 = 0.26 = 4.09 N [mg/L] 10-310-210-1100101102 pdf (%) 0 3 6 9 12 n = 180 m = 1.89 = 2.27 Flow rate, L/s 10-310-210-1100101102 pdf (%) 0 2 4 6 8 Measured Modeled n = 180 = 0.50 = 7.49 Flow rate Dissolved N Settleable N Sediment N Suspended N TN Figure 2 4 Probability density functions (pdf) of N concentrations for sediment, settleable, s uspended, and dissolved fractions, the sum of each fraction and the runoff flow rate of the 14 runoff events. Each pdf was fit to a log normal distribution (p < 0.05).

PAGE 48

48 fd 0.0 0.2 0.4 0.6 0.8 1.0 High Medium Low 15 Jul 08 Jul 10 Jun 16 May 19 Aug 12 Aug 29 Jul 10 Sept 21 Jun 23 Oct 20 Sept 08 Oct 08 Aug 03 Jun kd [L/kg] 102103104105 Q50 Intensity fd 0.1 1 pdf (%) 0 5 10 15 20 25 kd [L/kg] 102103104105106 pdf (%) 0 10 20 30 = 0.62 = 0.20 n = 180 = 5524 = 56373 n = 180 Figure 2 5 fd and Kd values as nonparametric plots for the 14 monitored rainfall runoff events, with me dian values marked as a dotted line. The order of storms is sorted by median flow rate of runoff, with the maximum value on the left The probability density functions (pdf) of fd and Kd are shown on the right.

PAGE 49

49 Runoff Rate (L/s) 0 2 4 6 8 10 fd 0.00 0.25 0.50 0.75 1.00 10 June event Flow fd Runoff Rate (L/s) 0 2 4 6 8 10 kd (L/kg) 102103104105 kd 0 250 350 1150 1500 1550 Cumulative Volume (L) t/t max 0.0 0.2 0.4 0.6 0.8 1.0 Runoff Rate (L/s) 0 2 4 6 8 10 Cumulative Volume (L) 0 2000 4000 6000 8000 10000 measured data modeled data R2 = 0.97 Figure 2 6 B ehavior of the dissolved fraction ( fd), equilibrium coefficient ( Kd) and cumulative 10 June event. A cumulative gamma distribution was fit to the cumulative volume.

PAGE 50

50 0 5 10 15 20 25 30 (g) 0 2 4 6 8 0 5 10 15 20 25 30 (g) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 M (g) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 M (g) 0 1 2 3 4 M (g) M (g) V (m 3 ) 0 5 10 15 20 25 30 (g) 0 5 10 15 20 25 V (m 3 ) (g) Sediment N Settleable N Suspended N Dissolved N Suspended N Dissolved N M T = M 0 (1-e -K 1 V ) M 0 = 9.12 K 1 = 0.22 R2 = 0.90 M T = M 0 (1-e -K 1 V ) M 0 = 2.06 K 1 = 0.80 R2 = 0.74 M T = M 0 (1-e -K 1 V ) M 0 = 1.89 K 1 = 0.17 R2 = 0.89 M T = K 0 V K 0 = 0.85 R2 = 0.81 M T = K 0 V K 0 = 3.90 R2 = 0.98 95% C. I. Flow-limited Mass-limited Mass-limited Mass-limited Mass-limited Flow-limited M T = M 0 (1-e -K 1 V ) M 0 = 5.00 K 1 = 0.22 R2 = 0.90 Figure 2 7 Cumulative mass versus volume curve and mass transport m odel for each N fraction. The shaded area represents the 95% confidence interval.

PAGE 51

51 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative load 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative load 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative load 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative volume 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative load 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative volume 0.0 0.2 0.4 0.6 0.8 1.0 Normalized cumulative load 0.0 0.2 0.4 0.6 0.8 1.0 Sediment N Settleable N Suspended N Dissolved N TN 10 September 20 September 16 May 03 June 10 June 21 June 08 July 15 July 29 July 08 August 12 August 19 August 08 October 23 October Symbol legend: Figure 2 8 Dimensionless cumulative mass and volume curves for sediment, settle able, suspended and dissolved N fractions for the 14 monitored rainfall runoff events. T he 1:1 diagonal line is plotted in order to evaluate firstflush transport.

PAGE 52

52 Particle diameter [ m] 10 100 1000 Volatile fraction (%) 0 9 18 27 36 45 TN [mg/kg] 0 2000 4000 6000 8000 10000 Volatile fraction TN Volatile fraction (%) 05101520253035 TN [mg/kg] 0150030004500600075009000 TN = 20564 x [% volatile] r2 = 0.96 PM diameter [m] 0.11101001000 % finer by mass 0255075100 Dry depostion (DD) Runoff (Q) (site median) (DD), d50 = 280 m(Q), d50 = 154 m Figure 2 9 a ) Distributions of N and volatile fraction over the PSD for dry deposition PM. b ) Relationship between N and v olatile fraction over the PSD. c ) Comparison of PSDs for dry deposition (DD) sa mples and runoff ( Q ) from the watershed. The d50 ( m) w ere also shown for both PM. a. b. c.

PAGE 53

53 PM-based N [mg/kg] 102103104105106 Frequency (%) 0 5 10 15 20 25 30 Suspended N Sediment N Settleable N n = 180 Sediment N: = 2.91 = 1.89 Settleable N: = 4.41 = 1.94 Suspended N: = 14.39 = 3.70 Figure 2 10 Probability density functions (pdf) of the PM based N (mg/kg) fractions for the 14 monitored rainfall runoff events.

PAGE 54

54 CHAPTER 3 VOLUMETRIC CLARIFYING FILTRATION OF NUTR IENT FRACTIONS IN URBAN SOURCE AREA RUNOFF Abstract Urban rainfall runoff (stormwater) transports nitrogen (N) and phosphorus (P) loads at levels that impact receiving water eutrophication. This study examines a volumetric c larifying filt er (VCF 2.3 m3, 106.2 m2 of filter surface area, SA ) which integrates filtration, clarification and retention to sequester nutrients associated with particulate matter ( suspended, settleable and s ediment PM) and in solution. Results are based on a monitoring campaign (rainfall: 375 mm, runoff: 117 m3) that encompassed 25 hydrologic events generated from a 500 m2 surface parking facility located in Gainesville, FL. Median concentration of influent (runoff) and effluent total nitrogen; TN as the sum of dissolved and PM associated N, were 3.00 mg/L (2.57 mg/L for TP) and 1.74 mg/L (1.47 mg/L for TP). Median N dissolve d fraction ( fd) were 0.42 (0.25 for P) and 0.76 (0.61 for P) for influent and effluen t. The lowest PM based N (and P) was the settleable fraction for the influent and sediment for the effluent. The e quilibrium partitioning coefficient ( Kd) was in the range of 102 to 105 L/kg for N (103 to 106 L/kg for P). N fractions and TN in influent fol lows a mass limited model while in effluent the transport behavior becomes flow limited. The VCF separated 99% of sediment PM, 88% of settleable PM and 33% of suspended PM. Separation of TN and TP was 50% and 75%, and separation of total dissolved N (TDN) and total dissolved P (TDP) was 0 % and 0 %. The total organic: inorganic carbon (TOC: TIC) ratio was 3.78 for the influent. The median C: N: P stoichiometry of rainfall, influent and effluent were 94:32:1, 62:9:1, and 348:9:1, respectively.

PAGE 55

55 Introduction S ince the end of World War II, urban development, redevelopment and infill in the United States has resulted in a significant increase in urban imperviousness generating increased runoff volumes, peak flows, and commensurately higher constituent loadings ( Line and White, 2007; Walsh et al., 2004, Sansalone and Glenn 2007). Urban runoff transports significant loads of constituents, including PM, metals, nutrients (N and P), and polycyclic aromatic hydrocarbons (PAHs) that impact receiving water bodies (Davis et al. 2001; Hoffman et al. 1984; Sample et al. 2012, Sansalone and Buchberger 1997; Kim and Sansalone 2010; Barrett et al. 1998; Wu et al. 1998; USEPA, 1996; and USEPA 1998). Increased nutrient loading can cause excessive growth of water borne plants and algae. Effects on receiving aquatic ecosystems may include oxygen depletion, loss of biodiversity and toxicity (USEPA 1993; Smith et al 1999; Taylor et al 2005; Glasgow et al. 2001; Liu and Sansalone 2007). In 2012, Florida DEP promulgated the new numeri c nutrient criteria that adjusted the discharge limit of TN and TP and for colored lake to be 1.272.23 mg/L and 0.050.16 mg/L (FDEP, 2012). N transported in runoff exists in a particulate (inorganic or organic PM or detritus) or aqueous phases. The aqueous phase can be further classified as inorganic N (ammonium N, nitrate N, and nitriteN) and organic N. The partitioning to each phase, the distribution within a phase or the aqueous species is a function, in part, of watershed and hydrologic parameters (Col lins et al., 2010). For example, Vaze and Chiew (2004) found that 20 to 50% of TN was TDN in roadway runoff from an urban Melbourne, Australia watershed. Based on work in Denmark HvitvedJacobsen et al. (1994) reported that 50 to 60% of TN in highway runof f is associated with PM. Potential

PAGE 56

56 nonpoint sources of N in urban areas are airborne constituents such as N aerosols and NOX due to fossil fuel combustion, surface deposition (road dust and PM), biogenic PM and detritus from urban landscapes, fertilizer, reclaimed wastewater for urban irrigation, eroded soils, leaf fall and landscape maintenance byproducts such as grass clippings, and uncollected pet wastes (Carpenter et al. 1998, Carey et al. 2013, Scudlark et al. 1998, Davidson et al. 2010, Kaushal et al. 2011). Dry deposition whether as dust fall or biogenic detritus, is considered as a major contributor of N in runoff through leaching and dissolution processes. Research conducted in urban areas of Asahimachi, Japan indicated that over 50% of N in runoff came from road dust (Kojima et al., 2011). Aqueous N species such as nitrate and ammonia are of critical concern with respect to water chemistry of the aquatic ecosystem as well as groundwater protection (Kim et al. 2003). High loading of TN coupled to runoff volume, high TN concentration has been identified in the first flush in roadway runoff (Flint and Davis 2007, Line et al. 1997), which could became a potential N sources due to their bioavailability for water borne microorganisms (Berman and Bronk, 2003). In addition, nutrients associated with PM will repartition between the PM and aqueous phase when contacted with rainfall (Makepeace et al., 1995). This is a dynamic process and the equilibrium partitioning is a function of water chemistry, PM g ranuometry, hydrology, microbial activities, and residence time (Berretta and Sansalone 2011a; Sheng et al. 2008). Phosphorus has been identified as a primary pollutant for surface and ground water deterioration, such as eutrophication (USGS 1999). P in urban runoff is a typical nonpoint source that transports high loadings of P (USEPA 1993). Flint (2007) reported that the EMC of TP in runoff from a parking facility ranges from 0.03 to 1.9 mg/L, and for

PAGE 57

57 highway runoff the median TP ranges from 0.19 to 1.8 mg/L (Ma et al. 2010, Drapper et al. 2000). Known sources that contribute to the P in urban runoff include biogenic materials, discharges from automobile exhaust and other combustion processes that re deposited to the watershed surface, mineral liberated from pavement surface during weathering (Strecker 1994). P is introduced to the aquatic environment in multiple chemical forms, which partitions between aqueous and solid phases, based on a dynamic equilibrium (Compton et al. 2000). Besides, P also distributes across the particle size distribution (Sansalone et al. 1998; Sansalone and Kim 2008b). A previous research on an urban land use shows a median dissolved fraction of 0.32, indicating that P is predominately bound to particulate matter fractions (Berr etta and Sansalone 2011b). The ratio between N and P in runoff and the treatment thereof can also provide metrics to quantify the potential impact of runoff on an ecosystem and the benefit provided by control of N and P. Measuring the stoichiometry of C, N and P provides an index of nutrient limitation in ecosystems. For example, in rivers, streams, estuaries and lakes when the N/P mass ratio is less than 10 (or 5 considering the variability in plant stoichiometry); N controls; while when N/P > 10 to 20, P controls (Thomann and Mueller 1987). The Redfield ratio is the atomic ratio of C, N, and P (CNP) Redfield (1958). Redfield showed that the C:N:P stoichiometry is 106:16:1 on average for plankton in global marine ecosystem which is also similar to the ratio of C:N:P in marine water (Redfield 1958). For the terrestrial ecosystem, the mean atomic C: N: P ratios in soil was found to be 186:13:1 and the value was 60:7:1 in the soil microbial biomass; and these ratios are well constrained on a global scale (C leveland and Liptzin 2007).

PAGE 58

58 While N/P and CNP ratios for runoff are potentially useful metrics for runoff, most characterization studies have been conducted to quantify nutrients level in runoff primarily using TN, TP and partitioning metrics. For example, the Nationwide Urban Runoff Program (NURP) identified an median event mean concentration (EMC) of TN of 2.2 mg/L, based on 28 projects including several industrial sites, facilities, roadways, and other source areas in the United States (USEPA, 1983). Li ne et al. (1997) measured the first flush (< 30 minutes of runoff) for 20 industrial sites in North Carolina, reporting a mean value of TN of 3.6 mg/L. As a comparison, Flint and Davis (2007) reported that for a commercial/residential area in Maryland the EMC of TN was 3.8 mg/L. N management in the urban environment is challenged by multiple and abundant sources of N as well as the efficient conveyance provided by urban infrastructure and pavement (Hsieh et al., 2007, Bean et al. 2007). While P management c an be less complex than N since P is predominately PM bound, the distribution of P across the particle size distribution (PSD) and potential leaching of P from PM challenges the management of P (Berretta and Sansalone 2012, Vaze and Chiew 2004). N and P m anagement through treatment such as sequestration is also challenged due to the stochastic nature of rainfall intensity, duration, and available mass on the watershed (Liu et al. 2001). Insitu treatment units have been developed to manage PM, nutrient and metal loads transported in urban runoff; sometimes coupled with also managing hydrologic parameters such as volume and peak flow. Typical mechanisms provided by these units include: gravitational separation operations for PM (Kalainesan et al 2009), volum etric and hydraulic flow control (Pathapati and Sansalone 2009), adsorption (Liu

PAGE 59

59 et al. 2005); biological and microbial processes (Davis et al., 2006; Hsieh et al., 2007; Johengen and LaRock, 1993); filtration (Berretta and Sansalone 2012); low impact desi gn systems such as permeable pavement (Collins et al., 2010, Kuang et al. 2007; Bean et al 2007). While volumetric clarification (retention or detention) is a common unit operation for PM gravitational sedimentation coupled to hydrodynamic control, volumet ric clarifying filtration (VCF) integrates hydrodynamic control, sedimentation and filtration in the same physical unit (Pathapati and Sansalone, 2011; Liu et al., 2010; Sansalone et al., 2009). This class of a combined unit operations have been deployed with either granular media to provide adsorption and filtration, or as in this study, cartridges to provide filtration after sedimentation. Objectives This study examines the transport, phase transformation and fate of urban source area N and P as modifi ed by an insitu unit operations; a VCF. Therefore the first objective quantified the transformation of N and P concentrations from rainfall to source area runoff and as effluent from the VCF after any sequestration. As part of this objective the transport of TN in runoff and effluent is modeled. The second objective examined the change in phase partitioning between runoff and effluent flows as a function of hydrology. The third objective was to examine the cumulative sequestration for N and P fractions by the VCF across the monitoring campaign as a function of cumulative influent volume. The fourth objective was to examine the C: N: P stoichiometry in rainfall, runoff and effluent, with transformation from rainfall to runoff by the source area and transformation from runoff to effluent by the VCF.

PAGE 60

60 Methodology Site D escription The s ource area is an asphalt paved surface parking facility, locat ed in Gainesville, FL. A plan view is shown in Figure 31 The total drainage area is ~ 500 m2, with a slope of 3% in the East West (E W) direction and 1.5% in the NorthSouth (N S) direction; conveying runoff during rainfall runoff events. Runoff drains as gravitationally driven sheet flow to a curb and gutter section that flows in an E W direction to a catch basin, whic h is directly connected to a storm sewer pipe system at a 6% slope. Approximately 24% of the source area is raised vegetated islands that drain directly to the asphalt paved parking aisles. The balance of the source area was impervious asphalt pavement that was oxidized and had cracking, ruts and deterioration from lack of regular maintenance. In addition to a complete grass cover, the vegetation in the islands is magnolia, pine, oak and sycamore trees, which shed leaves to the island and pavement surfaces. The source area combination of raised vegetated islands draining to impervious asphalt pavement resulted in significant loads of biogenic material available for transport and leaching. In order to determine the intraevent transport and partitioning of nutrients, at least 10 sample sets of influent and effluent were taken manually with duplicates at one to ten minute interval s throughout each entire storm event providing sufficient resolution of the temporal variation of nutrient s and PM concentration. A total of 25 storm events were fully captured from 24 May 2010 through 27 June 2011. Rainfall depth and intensity were monitored by a rain gage located within 180 m of the source area. Rainfall samples were collected in acidwashed and de ionized water rinsed elevated Pyrex trays placed adjacent to the rain gage. The following event based hydrologic

PAGE 61

61 indices were monitored or calculated for each event: previous dry hours (PDH), event duration, peak flow rate, median and mean flow rate s, total runoff volume, rainfall depth, i nitial p avement r esidence t ime (IPRT) and the volumetric runoff coefficient (C). VCF Configuration Relative to the hydrographs from the source area the VCF was sized as a relatively a small sedimentation tank with filter cartridges that provided relatively high SA for filtration. The in situ VCF transformed source area runoff into clarified effluent, through physical mechanisms of primary sedimentation and secondary filtration with only nominal flow attenuation. The side and top view of t he VCF are shown in Figure 31 The VCF was comprised of 3 radial filter cartridges (length of 1.4 m and total SA of 106.2 m2). These 3 cartridges were submerged and vertically oriented in a cylindrical tank with a total height of 2.6 m and a diameter of 1.2 m. The 1.0 turnover volume is 2313 L. The filtration cartridges were mounted through the tank deck; the upper side of the deck contoured to create a separate backwash pool weir that extended 0.15 m above the tank deck. Two cartridges were mounted insi de the weir and 1 cartridge were mounted outside the weir. Influent head allowed the filtered effluent to overtop the weir and be conveyed to the effluent piping through the deck that was contoured to the effluent piping. When the driving head decreased, w ater in the backwash pool reversed flow direction and drained downward through the cartridges back into the tank; thereby passively backwashing the cartridges inside the weir. As the runoff flow rate increases and decreases during a storm event, this self cleaning mechanism can occur multiple times during an event. The design hydraulic capacity of the VCF was (12.6 L/s).

PAGE 62

62 Sampling P rotocol The influent and effluent conveyance and sampling collection system is shown in the lower plot of Figure 31 The catch basin that collected gutter and sheet flow was located at the lowest elevation of the source area. The catch basin collected and conveyed all flow through a 152 mm diameter PVC pipe to a calibrated Parshall flume where a calibrated 30 kHz ultrasonic sensor monitored the depth of flow. Runoff samples were taken at the drop box that collected the outflow from the Parshall flume. Effluent samples were taken at the drop box below the VCF effluent piping. A set of sample replicates included two 0.5 L bottles for particle size distribution ( PSD ) analysis, two 1 L bottles for granul ometric analyses, and two 1 L bottles for water chemistry analyses including nutrient s. All the sample bottles were wide mouth polypropylene. Laboratory A nalyses M ethods A fter each event rainfall, runoff an d effluent samples were transported directly to the laboratory for analysis of water chemistry including N, P, PM fractions {sediment ( > 75m), settleable (25 ~ 75m), suspended (0.45 ~ 25 m) as TSS}, and dissolved ( < 0.45 m), using Standard Method 2540D, 2540E (APHA, 1992) and ASTM D397797(ASTM, 1997). Samples were fractionated within 2 hours from the time of sampling to minimize time dependent changes in partitioning. D etailed procedures of phase separation and PM size fractionation can be found in Sansalone and Kim (2008a) and Kim and Sansalone (2008) N concentration was determined by a digestion method where potassium persulfate (K2S2O8) convert ed all forms of N to nitrate. Nitrate then react ed with chromotropic acid and the absorbance measured at 410 nm with a spectrophotometer. Lot specific calibration curves between absorbance and concentration were generated

PAGE 63

63 based on a series of NH3N standards. Each calibration curve must have a coefficient of determination (R2) larger t han 0.99. Validation was performed on a standard soil sample (Nutrients in Soil, No. D061542, Environmental Resource Associates ) and was found to be within a 95% confidence interval. The TN recovery based on the study results and the certified standard was 96%. P concentration was based on the ascorbic acid method (APHA Standard Method 4500P E) and absorbance was measured at 880 nm with a spectrophotometer. Lot specific calibration curves converted absorbance to PO4 3-concentration (R2> 0.99). Water chemi stry indices were measured for all samples, including pH (APHA Standard Method 4500H+B), redox potential (APHA Standard Method 2580), conductivity/total dissolved solids (TDS) (APHA Standard Method 2510),temperature (APHA Standard Method 2550) and dissol ved oxygen (DO) (APHA Standard Method 4500O).Total Organic Carbon (TOC) was measured by a DOHRMANN Phoenix 8000 UV persulfate TOC analyzer following Standard Method 5310C ( APHA, 1992) and D483903 ( ASTM 2011) Alkalinity is determined by the acid titr ation method (APHA Standard Method 2320b) Total Inorganic Carbon (TIC) was determined by summation of bicarbonate and dissolvedCO2 concentration. Since runoff and effluent pH is in the range of 6 to 8, bicarbonate is the predominant species, and hence the alkalinity is e ssentially represented by the bicarbonate concentration. Dissolved CO2was determined by the nomographic method (APHA Standard Method 4500CO2) where pH, temperature, TDS and bicarbonate alkalinity were applied to the model. Event mean concentrations (EMC) were calculated as an event based index (Flint and Davis, 2007)

PAGE 64

64 N Partitioning in R unoff Considering a twophase partition, inclusive of the nominal colloidal fraction (<1 m), partitioning is the interaction between the aqueous and PM bound fractions (Zgheib et al. 2011). Quantifying N and P partitioning in runoff is useful to identify the predominant phase for transport, transformation or sequestration; for example as might be provided by the VCF of this study through physical separat ion mechanisms. The dissolved fraction (fd) is defined in the following equation. pddpdddmmmcccf (3 1 ) In this expression, md represents the dissolved mass of a constituent (N or P) (mg) and mp denotes the PM associated mass of a constituent (m g) (Sansalone and Buchberger 1997). Another commonly used parameter is the equilibrium partitioning coefficient (Kd), which is by definition the ratio of the constituent mass normalized to the dry mass of PM to the concentration in solution (Thomann and Mueller 1987). d s dc c K (3 2 ) In this equation cd is the dissolved mass concentration (mg/L) and cs is the PM based mass concentration (mg/kg of dry mass) of a constituent. The unit of Kd are L/kg. As with fd, Kd can also be used to examine the ratio between the dissolved and PM associated constituent. Generally, larger Kd values indicates greater partitioning to the PM phase (Mishra et al, 2004). There is an inverse relation between fd and Kd as shown in the following equation.

PAGE 65

65 mkcccccccccfdspdsdppddd111111 (3 3 ) In this equation, spccm/ which is the PM mass concentration (kg/L). Mass Transport B ehavior The transport behavior of a constituent mass with respective to runoff volume could be classified as either mass limited or flow lim ited (Sheng et al. 2008). A mass limited behavior is often characterized as a mass based first flush and refers to the disproportionate mass transport of a constituent with respect to runoff volume during the initial stage of a runoff hydrograph; as a f irst order declining profile. By contrast, flow limited behavior refers to a zero order relationship between mass and runoff volume. The models for mass and flow limited events are shown in Table 31. S equestration of Mass during the Monitoring C ampaign The sequestration for a constituent or constituent phase is quantified as a percent removal (PR) on an event basi s using the influent and effluent EMC values for the VCF. %100)()()((%)251251251iINFiINFiiiEFFiEFFiINFiINFiINFiEFFiINFiEMCVEMCVEMCVMMMPR (3 4) In this equation, Mi INF and Mi EFF are the mass of runoff and effluent for storm i [L3]; Vi INF and Vi EFF are the volume of runoff and effluent for storm i [L3]; EMCi INF and EMCi EFF are the event mean concentration of runoff and effluent in storm i [M/L3].

PAGE 66

66 Results and Discussion Hydrology of Monitored Rai nfall R unoff Events Twenty five runoff events were monitored and sampled over a period from 28 May 2010 to 27 June 2011. The monitored hydrologic indices are tabulated in Table 32 on an event basis. The previous dry hours (PDH) ranges from 10 to 910 hours event durations ranges from 26 to 691 minutes, peak runoff flow rate ranges from 0.5 to 14.3 L/s, median flow rate ranges from 0.04 to 5.5 L/s, rainfall depth ranges from 3 to 50 mm, runoff volume ranges from 206 to 13,229 L, peak rainfall intensity ranges from 5 to 137 mm/hr, and IPRT ranges from 1 to 34 minutes. The 25 captured events provide a representative distribution of rainfall depth that falls within 25th to 85th percentile of the longterm (10 years) historical data for Gainesville, FL (NCDC 201 2). No bypass was observed during the entire monitoring campaign. Distribution and Concentration of N Fractions The probability density function (pdf) of TN and N concentration for each PM fraction are plotted as shown in Figure 32. Influent and effluent results for the VCF are illustrated for measured data as histograms and are modeled as lognormal distributions (at = 0.05). The nonparametric statistics of plot are also compared for the N fractions of influent and effluent. With the lognormal distri bution of TN and each N fraction the median is used as a representative index of each distribution. The median TN concentration for runoff and effluent are 3.00 mg/L and 1.74 mg/L. With respect to the influent the highest median concentration is the dissolved N fraction at 1.10 mg/L while the lowest is the settleable N fraction value of 0.09 mg/L. For the suspended and dissolved fractions in the influent, N concentrations were approximately an order of magnitude higher than for the coarser settleable and s ediment fractions. The variability

PAGE 67

67 of N associated with each influent fraction decreased based on the PM size fraction with the sediment fraction having the greatest variability and the dissolved fraction the lowest variability. The larger variability of N associated with coarse PM suggests that transport of the sediment size PM was driven by the previous dry deposition of this fraction and the transport hydraulics as compared to the suspended and dissolved N fractions. By comparison for the effluent the highest median concentration was the dissolved fraction at 1.22 mg/L while the concentration consistently decreased for coarser size PM fractions with the lowest value of 0.01 mg/L for the suspended and dissolved fractions. The nominal increase in the di ssolved N concentration through the VCF, from 1.10 to 1.22 mg/L was not significant (at = 0.05). This result is reasonable given that the VCF predominately facilitates physical sequestration of N fractions, potential leaching during storage, and only nominal chemical or biological conversions such as nitrification and denitrification. Exc ept for the dissolved fraction for which the variability did not change the VCF reduced the variability of all other fractions. The MannWhitney rank sum test (Devore, 1990) was applied in this study for test for differences between corresponding influent and effluent N fractions. Test results indicated a significant (at = 0.05) difference between influent and effluent for sediment, settleable and suspended N and no significant difference for dissolved N. As a comparison to the N results in Figure 32, TP and P fraction results are summarized in the same manner in Figure 33 as histograms, modeled probability distributions and as nonparametric results. Similar to the lognormal N distributions, TP and all P fractions were lognormal distributions (at = 0.05). The median TP concentration for runoff and effluent are 2.57 mg/L and 1.47 mg/L. With respect to the

PAGE 68

68 influent the highest median concentration is the suspended P fraction at 0.74 mg/L while the lowest is the settleable P fraction value of 0.06 mg/L. For the suspended and dissolved fractions in the influent, P c oncentrations were approximately an order of magnitude higher than for the coarser settleable fraction and a factor of two higher for the sediment fraction. The variability of P associated with each influent fraction was greatest for the sediment fraction and significantly lower for all other fractions; with the variability of TP reflecting that of the sediment fraction. Similar to N, the much larger variability of P associated with coarse PM suggests that transport of the sediment size PM was driven by t he previous dry deposition of this fraction and the transport hydraulics as compared to the other P fractions. Similar to N, the dissolved fraction of P was highest in the effluent at 0.90 mg/L while the concentration consistently decreased for coarser si ze PM fractions with the lowest value of 0.01 mg/L for the suspended and dissolved fractions. The increase in the dissolved P concentration through the VCF, from 0.65 to 0.90 mg/L was not significant (at = 0.05). This result is reasonable given that th e VCF predominately facilitates physical sequestration of P fractions, potential leaching during storage, with only nominal chemical or biological conversions such as nitrification and denitrification. The VCF reduced the variability of all P fractions. The MannWhitney rank sum test (Devore, 1990) was applied in this study for test for differences between corresponding influent and effluent N fractions. Test results indicated a significant (at = 0.05) difference (decrease) between influent and effluent for all P fractions except dissolved P which generated a significant increase. The EMCs of N and P fractions and TN, TP were tabulated in Table 33 and 4.

PAGE 69

69 N utrient (N, P) Partitioning on Site and Event B asis Histograms of measured data and modeled probability density function (pdf) for fd, Kd, suspended PM (as SSC ) based on the 25 monitored storm events are plotted in Figure 34. The fd pdf of runoff and effluent are modeled as lognormal dist ributions (at = 0.05) with median values of 0.42 and 0.76, respectively. The higher median value of the effluent fd indicates that PM associated N was sequestered by the VCF while the dissolved (including the leached N during storage) fraction were eluted from the VCF. The variance of fd was smaller for the effluent samples due to attenuation of the VCF system. Kd value of runoff and effluent also modeled as lognormal distributions, with median values of 1.78 x 104 and 3.24 x 104 L/kg. Kd values of N in the effluent had a smaller variance than that of runoff. PM concentration ranges from 3.0 to 908.3 mg/L for runoff and from 0.1 to 38.8 mg/L for effluent. The median concentration of PM in runoff and effluent are 54.0 and 9.3 mg/L. In contrast to the sitebased part itioning results just summarized, the variability of intra event partitioning results for N are illustrated for the 15 July 2010 event of high unsteadiness and flow rate. As shown in Figure 35, fd, Kd and PM for runoff and effluent are plotted as a functi on of elapsed time with the runoff hydrograph embedded in the background. During the event, fd values in runoff varied from 0.1 to 0.7 and effluent fd variability was reduced to the range of 0.2 to 0.6 noting that over 98% of influent PM had been sequester ed by the VCF unit. While there is no clear trend between fd and Kd on the rising limb of the hydrograph there is a clear inverse trend on the falling limb illustrating a decreasing fd and a corresponding increase in Kd. The Kd values varied significantly during the event, ranging from 1.0103 to 3.9105 L/kg for runoff and

PAGE 70

70 9.2104 to 6.4105 L/kg for effluent. The Kd of the runoff increased on the falling limb of the runoff hydrograph and gradually decreased for the effluent. In many events, Kd trended in the range of 104 to 105 L/kg, approaching higher values towards the end of the event. This range is similar for rivers, lakes and watersheds with longer residence times with equilibrium conditio ns (Thomann and Mueller 1987). SSC shows a classical concentrationbased first order washoff (firstflush) and an inverse correlation with Kd for the influent. There is no clear trend for the effluent given the VCF functions as a relatively constant resistor for PM transport. In this event, PM concentration ranges from 14 to 630 mg/L for runoff and 4 to 12 mg/L for effluent. The event mean removal efficiency of PM is 99% in this event. TN Mass Transport Behavior Cumulative mass volume curves of N fractions and TN were plotted for both runoff and effluent, as s hown in Figure 36. Results from all 25 events were plotted to create a siteaveraged N mass delivery model Regression curves suggest that N fractions and TN in influent (runoff) fit in a first order exponential growth model, while in effluent these spec ies fit in a zero order linear model., which means that the VCF unit has altered the N mass transport behavior from mass limited to flowlimited. This result could be possibly related to the decomposition of biogenic materials in the VCF during retention periods that lead to less restriction of N sources in the effluent. In runoff, the scale factor, M0, follows a descending order of suspended, dissolved, sediment, and settleable N, indicating that suspended N is most abundant (or available) on the watershed. Similarly, the shape factor, k1, follows a descending order of sediment, dissolved, suspended, and settleable N, indicating that sediment N is the first depleted fraction during a storm. In effluent, the rate factor, k0, follows an order of

PAGE 71

71 dissolved, suspended, settleable, and sediment N, indicating that N associated with finer PM is easier to be transported. Result also shows that the difference between runoff and effluent is progressively larger as the PM size becomes larger, which is due to the bett er performance on coarser PM separation of the VCF. VCF Performance in Nutrient (N, P) Removal Across the monitoring campaign the sequestration of TN, TP as well as N and P fractions by the VCF was evaluated as the cumulative removal efficiency as a funct ion of influent volume. Results are plotted in Figure 37. Cumulatively, sequestration by the VCF for the sediment fraction of N and P resulted in a 99% separation of these fractions compared to the influent. Results for N and P settleable fractions were 88 and 92%. For these coarser fractions the primary mechanism was gravitational settling. Following the initial series of event loadings that provided the onset of filter cartridge ripening the event based sequestration for suspended N and P increased from 42 to 31%, and 21 to 74%. There was a net elution of dissolved N and P (including leached N and P) since the VCF did not provide adsorption or chemical precipitation. Cumulatively, sequestration by the VCF for TN and TP resulted in a 50 and 75% separation compared to the influent. The C : N: P Stoichiometry in Rainfall, Runoff and Effluent The total organic carbon (TOC), total inorganic carbon (TIC) transported in runoff and their ratio were plotted as histograms in Figure 38. TOC, TIC and the TO C: TIC ratio were modeled as lognormal distributions with median values of 3.7 mg/L, 13.8 mg/L and 3.78. The organic fraction is nearly 4 times higher than inorganic carbon. The C: N: P stoichiometry in rainfall, runoff and effluent for the source area wer e compared to the Redfield ratio, as shown in Figure 39. The C: N: P stoichiometry in

PAGE 72

72 rainfall is approximately 94:32:1, in runoff is 62:9:1, and in VCF effluent is 348:9:1. The median C: N ratio of rainfall, runoff and effluent are 2.9, 7.0 and 38.2 respectively. The C: N ratio in runoff approaches that of the Redfield ratio (C: N = 106:16 = 6.8:1), indicating runoff would promote growth of plankton and algae. The C: N stoichiometry for rainfall is much less than the Redfield ratio. Results also show an enrichment of carbon in effluent due to leaching of biogenic materials and PM sequestered by the VCF unit during the retention periods. The median N:P ratio in rainfall, runoff and effluent are 32.3, 8.8 and 9.1 respectively. Compared to the Redfield ratio (N: P = 16:1), the ratio in rainfall was much higher while runoff and effluent were both lower. This result indicates that N is the controlling nutrient in both runoff and effluent for the source area. Effluent N:P ratios are slightly higher than runoff due to improved removal efficiency for P (75% for P versus 50% for N) by the VCF unit. C onclusion Based on a 13 month monitoring campaign encompassing 25 fully captured runoff events from an urban source area, this study examined the transport, phase transf ormation and fate of N and P as modified by an insitu unit operation; a volumetric clarifying filter (VCF). The source area was a Gainesville, FL surface parking facility with a drainage area of ~500 m2 of which 76% was asphalt paved and the remaining 24% of the drainage area was raised vegetated islands that drain to the pavement. Across the monitoring campaign, 58% of N was PM bound and 75% of P was PM bound with the dissolved fraction representing 42% of N and 25% of P. The median concentrations of the sediment fraction for N and P were 1.04 and 1.28 mg/L, the settleable fraction were 0.20 and 0.16 mg/L, the suspended fraction were 0.56 and 0.71 mg/L and the dissolved fraction were 1.04 and 0.67 mg/L. Across the monitoring campaign the equilibrium

PAGE 73

73 parti tioning coefficient ( Kd) of N was modeled as a lognormal distribution with a median value in runoff is 1.78x104 L/kg and a standard deviation of 4.77x105 L/kg. By comparison the Kd varied significantly during single storm events as a result of highly uns teady hydraulics and therefore transport of PM; and values range from 104 to 106 L/kg across an event. For this source area, the median Kd for P was 3.62x104 L/kg and a standard deviation of 3.66x105 with a log normal distribution. To modify the fate of N and P fractions in the influent runoff an insitu physical unit operation, a volumetric clarifying filter (VCF) combined sedimentation, filtration, and nominal volumetric retention was monitored as part of this study. Across the monitoring campaign the c umulative sequestration (removal efficiencies based on influent) for both N and P fractions were 99 and 99% for sediment, 88 and 95% for settleable, 29 and 28% for suspended, 0 and 0% for dissolved, and 50 and 75% for TN and TP. The VCF system has altered the N mass transport behavior from mass limited to flowlimited due to the decomposition of biogenic materials. The Regression param e ter suggests that suspended N is most abundant i n runoff, while sediment N has the highest depletion rate. In effluent, f iner fraction has the higher N transport rate given that less stream power is required. Site specific median C: N: P stoichiometry of rainfall, runoff and effluent were 94:32:1, 62:9:1, and 348:9:1 respectively. Compared to the Redfield ratio (106:16:1), results indicate that runoff is suitable for the growth of plankton and algae, and that N is the controlling nutrient in the effluent discharge.

PAGE 74

74 Table 3 1 Mass transport models of flow limited and mass limited behavior. M represent s mass of nutrient [M]; K is a wash off coefficient; t is time, [T]; Q is flow rate [L3/T]; MT is mass delivered at time T, [M]; is the zeroorder wash off constant, [M/L3]; is the volume at time T, [L3]; 1k is the firstorder wash off coefficient, [L3]; and M0 is the initial avail able mass of nutrient on the pavement [M]. Flow limited (zero order) Mass limited (first order) Differential form QkKdtdM0 KMdtdM Integrating form TTVkM0 ) 1 (10tV k Te M M

PAGE 75

75 Table 3 2 Event based hydrology parameters for 25 rainfall runoff events and summary of those parameters (Mean, Median, Standard Deviation). Event date I max (mm/hr) PDH (hr) IPRT (min) P rain (mm) d rain (min) V runoff (L) Q max (L/s) Q 50 (L/s) Q m ean (L/s) 28 May 10 76 96 10 21 112 7465 4.30 0.98 1.12 0.066 16 Jun 61 288 18 16 61 5006 5.36 0.65 2.21 0.089 21 Jun 122 96 6 23 43 8695 7.46 5.47 5.09 0.085 30 Jun 76 288 8 13 50 5459 9.13 3.30 3.95 0.128 15 Jul 91 96 8 10 28 3608 13.26 1.4 4 3.12 1.085 1 Aug 127 24 5 30 36 11973 14.25 4.74 5.47 0.119 6 Aug 51 120 5 4 104 1395 6.80 0.01 0.27 0.012 7 Aug 61 24 7 9 48 2622 8.24 0.43 0.90 0.178 23 Aug 5 48 20 3 42 312 1.25 0.01 0.12 0.138 12 Sep 51 172 18 7 52 1643 3.85 0.10 0.53 0.253 26 Sep 5 40 1 4 78 1129 0.45 0.26 0.24 0.110 27 Sep 91 10 20 15 388 3841 10.94 0.04 0.16 0.404 4 Nov 46 910 5 5 43 994 3.53 0.12 0.38 0.303 16 Nov 25 286 8 3 34 305 1.75 0.02 0.13 0.418 5 Jan 11 107 72 3 21 125 5800 7.36 0.16 1.14 0.143 10 Jan 91 106 4 5 26 1129 3.32 0.01 0.41 0.138 25 Jan 18 365 5 44 389 12387 4.09 0.39 0.53 0.067 7 Feb 30 12 8 33 306 13229 2.22 0.77 0.71 0.065 9 Mar 15 79 10 29 691 10051 3.13 0.10 0.24 0.108 28 Mar 33 438 7 3 66 522 1.03 0.06 0.13 0.039 30 Mar 76 48 34 15 179 3707 5.61 0.10 0.29 0.061 20 Apr 15 196 9 4 61 206 3.28 0.01 0.06 0.252 14 May 137 188 5 50 295 11256 7.53 0.02 0.63 4.665 6 Jun 23 541 4 4 69 960 1.55 0.01 0.23 0.653 27 Jun 43 88 2 11 50 3383 3.35 0.12 0.64 0.008 59 185 9 15 135 4683 5.3 0.82 1.12 0.384 50 51 96 7 11 61 3608 4.1 0.13 0.53 0.128 39 206 7 13 160 4284 3.7 1.50 1.61 0.932

PAGE 76

76 Table 3 3 Event based N fractions and Total N EMCs for 25 rainfall runoff events with mean, median, and standard deviation calculated. Median EMCs of TN from the National wide Urban Runoff Program (NURP) and discharge limits for TN based on Florida Numeric Nutrient Criteria (FDEP, 2010) are also tabulated at the bottom Event Date Dissolved N Suspende d N Settleable N Sediment N Total N EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e 28 May 10 1.56 2.94 0.35 0.35 0.65 0.04 3.88 0.04 4.91 3.38 16 Jun 1.34 1.30 0.11 0.28 0.27 0.01 1.39 0.01 3.11 1.61 21 Jun 1.46 1.47 0.23 0.40 0.10 0.01 3.03 0.01 4 .82 1.89 30 Jun 1.13 1.28 0.62 0.46 0.07 0.01 0.07 0.00 1.89 1.75 15 Jul 1.16 0.67 0.59 1.52 0.28 0.01 0.68 0.00 2.72 2.20 1 Aug 1.04 0.93 0.24 0.24 0.15 0.06 0.61 0.01 2.03 1.23 6 Aug 2.02 0.90 0.60 0.56 0.26 0.03 2.63 0.09 5.50 1.57 7 Aug 0.70 0.59 0.12 0.14 0.18 0.03 0.17 0.00 1.17 0.76 23 Aug 1.22 1.59 0.39 0.50 0.21 0.02 1.61 0.01 3.42 2.11 12 Sep 0.83 1.84 0.56 0.74 0.28 0.03 0.86 0.02 2.52 2.63 26 Sep 0.52 1.12 0.84 0.50 0.03 0.02 1.33 0.01 2.72 1.65 27 Sep 0.30 0.65 0.24 0.09 0.27 0.01 1.45 0.01 2.27 0.76 4 Nov 1.00 0.88 1.44 0.21 0.18 0.01 0.78 0.02 3.40 1.12 16 Nov 1.31 0.80 2.14 0.36 0.17 0.01 2.08 0.08 5.70 1.25 5 Jan 11 0.38 0.29 0.26 0.26 0.20 0.01 1.04 0.00 1.88 0.55 10 Jan 0.60 0.88 0.13 0.23 0.16 0.01 0.35 0.00 1.24 1.12 25 Jan 0.47 0.55 0.55 0.17 0.11 0.01 0.27 0.00 1.40 0.73 7 Feb 0.34 0.34 0.63 0.48 0.05 0.00 0.16 0.00 1.18 0.82 9 Mar 0.80 0.77 0.20 0.44 0.05 0.01 0.24 0.01 1.30 1.20 28 Mar 3.88 2.36 2.02 0.55 0.21 0.03 0.40 0.01 6.51 2.96 30 Mar 0.75 1.02 0.96 0.35 0.33 0.01 1.99 0.00 4.02 1.35 20 Apr 3.15 5.93 0.52 0.55 0.29 0.01 6.52 0.01 10.48 6.50 14 May 0.76 2.12 0.98 0.07 0.22 0.01 1.98 0.01 3.94 2.20 6 Jun 1.83 3.57 1.41 0.81 0.16 0.01 0.90 0.00 4.31 4.39 27 Jun 1.32 5.22 1.25 1.33 0.39 0.01 2.60 0.02 5.56 6.58 1.19 1.60 0.69 0.46 0.21 0.02 1.48 0.01 3.52 2.09 50 1.04 1.02 0.56 0.40 0.20 0.01 1.04 0.01 3.11 1.61 0.84 1.44 0.57 0.35 0.13 0.01 1.45 0.02 2.16 1.61 FDEP NNC b Colored lake 1.27 2.23 Stream 0.67 1.87 Estuary 0.17 1.29 NURP EMC c 2.20 a. All values are in the unit of mg/L; b. F.A.C. CHAPTER 62 302: Surface water Quality Standards (FDEP 2010); c. Median EMC from the Nationwide Urban Runoff Program (U.S. EPA 1983).

PAGE 77

77 Table 3 4 Event based P fr actions and Total P EMCs for 25 rainfall runoff events with mean, median, and standard deviation calculated. Median EMCs of TN from the National wide Urban Runoff Program (NURP) (USEPA 1983) and discharge limits for TP based on Florida Numeric Nutrient Cr i teria (FDEP, 2010) are also tabulated at the bottom Event Date Dissolved P Suspended P Settleable P Sediment P Total P EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e 28 May 10 0.50 0.36 0.71 0.36 0.18 0.03 1.03 0.02 2.41 0.76 16 Jun 0.33 0.32 0.44 0.53 0.27 0.01 2.22 0.02 3.26 0.88 21 Jun 0.18 0.23 0.12 0.22 0.10 0.01 5.49 0.01 5.88 0.47 30 Jun 0.22 0.25 0.27 0.35 0.08 0.01 0.66 0.02 1.22 0.62 15 Jul 0.64 0.45 0.32 0.26 0.36 0.01 2.23 0.01 3.55 0.73 1 Aug 0.19 0.40 0.57 0.46 0.41 0.03 1.18 0.02 2.34 0.92 6 Aug 0.67 0.52 0.34 0.36 0.01 0.00 1.02 0.04 2.04 0.92 7 Aug 0.20 0.32 0.59 0.61 0.17 0.03 0.45 0.00 1.41 0.96 23 Aug 0.53 0.66 0.43 0.22 0.02 0.00 0.59 0.00 1.57 0.88 12 Sep 0.69 1.05 0.75 0.47 0.02 0.00 0.67 0.02 2.14 1.54 26 Sep 0.67 0.8 5 0.53 0.61 0.01 0.01 1.83 0.01 3.04 1.49 27 Sep 0.50 0.71 1.22 1.01 0.07 0.00 1.28 0.01 3.06 1.73 4 Nov 1.44 1.03 1.98 1.35 0.36 0.02 1.24 0.01 5.01 2.41 16 Nov 1.92 1.60 2.51 0.79 1.47 0.03 2.89 0.17 8.79 2.57 5 Jan 11 1.06 1.36 0.77 0.73 0.17 0.01 1 .94 0.01 3.95 2.10 10 Jan 1.04 1.39 1.11 1.09 0.33 0.02 1.38 0.00 3.85 2.50 25 Jan 0.45 0.51 1.80 0.36 0.13 0.01 2.12 0.01 4.50 1.15 7 Feb 0.68 0.63 0.70 0.53 0.01 0.00 1.55 0.01 2.95 1.18 9 Mar 0.39 0.63 0.38 0.38 0.04 0.01 0.18 0.01 1.00 1.02 28 Mar 4.67 2.56 2.04 1.17 0.11 0.02 0.23 0.01 7.06 3.75 30 Mar 0.88 1.11 1.36 1.35 0.26 0.01 1.87 0.00 4.36 2.47 20 Apr 2.05 4.00 0.79 0.76 0.16 0.01 3.51 0.00 6.50 4.77 14 May 0.74 0.79 0.62 0.69 0.13 0.00 1.51 0.00 2.99 1.48 6 Jun 1.18 1.88 0.74 0.47 0.18 0.01 0.67 0.00 2.77 2.37 27 Jun 0.79 2.18 0.83 0.56 0.37 0.01 1.24 0.01 3.23 2.76 Mean a 0.90 1.03 0.88 0.63 0.22 0.01 1.56 0.02 3.55 1.70 Median 0.67 0.71 0.71 0.53 0.16 0.01 1.28 0.01 3.06 1.48 Std. Dev. 0.92 0.88 0.62 0.33 0.29 0.01 1.15 0.03 1.91 1 .05 FDEP NNC b Colored lake 0.05 0.16 Stream 0.06 0.49 Canal 0.05 1.28 NURP EMC c 0.33 a. All values are in the unit of mg/L; b. F.A.C. CHAPTER 62 302: Surface water Quality Standards (FDEP 2010); c. Median EMC from the Nat ionwide Urban Runoff Program (U.S. EPA 1983).

PAGE 78

78 Center Drive Museum Road Slope = 1.5% 65.3 m Slope = 3% Treatmentsystem 2 15.7 m5 m91.6 mDrainage area (~500 m ) InfluentEffluentN Parshall flume w/ u ltrasonic sensor Drop box ( Influent ) Drop box ( Effluent ) Pressure transducers Siphon break 10 m 0.6 m Discharge pipes Bypass pipes Note:Influent and effluent conveyance piping are 15.24 cm (6 inch) PVC Catch basin (CB) weir Pavement drainage area2.0 m 1.2 m Filters 2.6 m Q CB High-flo cartridges Maintainance pipe Weir Skirt Backwash cartridge Pressure relief Inlet Outlet Deck 0.3 m 0.4 m Figure 3 1 a ) S ite drawing of the studied watershed, located in UF Reitz Union b ) S ide and topview of the volumetric clarifying filtration (VCF) system. a. b.

PAGE 79

79 Total N N [mg/L] 10-410-310-210-1100101102 10-310-210-1100101102 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 1.10 mg/L (i) = 2.45 mg/L 50 (e) = 1.22 mg/L (e) = 2.44 mg/LDissolved N (< 0.45 m) 10-310-210-1100101102 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 0.52 mg/L (i) = 3.53 mg/L 50 (e) = 0.33 mg/L (e) = 2.53 mg/LSuspended N (0.45 25 m) N [mg/L] 10-610-410-2100102104 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 0.09 mg/L (i) = 5.00 mg/L 50 (e) = 0.01 mg/L (e) = 2.29 mg/LSettleable N (25 75 m) N [mg/L] 10-610-410-2100102104 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 0.34 mg/L (i) = 9.17 mg/L 50 (e) = 0.01 mg/L (e) = 4.25 mg/LSediment N (> 75 m) N [mg/L] 10-210-1100101102103 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 3.00 mg/L (i) = 2.91 mg/L 50 (e) = 1.74 mg/L (e) = 2.22 mg/LTotal N Dissolved N Suspended N Settleable N Sediment N Influent (i) Effluent (e) Effluent (e) Influent (i) Influent-model Effluent-model Figure 3 2 Probability density function (pdf) of N fractions and TN based on 25 monitored storm events between 2010 and 2011, including bot h runoff and effluent. Each pdf fits to a lognormal distribution. A box whisker plot is also showed to demonstrate the range and median values in each N fraction.

PAGE 80

80 Total P P (as PO4 3-) [mg/L] 10-310-210-1100101102 10-310-210-1100101102 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 0.65 mg/L (i) = 2.34 mg/L 50 (e) = 0.90 mg/L (e) = 1.01 mg/LDissolved P (< 0.45 m) 10-310-210-1100101102 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 0.74 mg/L (i) = 1.00 mg/L 50 (e) = 0.56 mg/L (e) = 0.36 mg/LSuspended P (0.45 25 m) P as (PO4 3-) [mg/L] 10-610-410-2100102104 pdf 0.00 0.05 0.10 0.15 0.20 0.25 0.30 50 (i) = 0.06 mg/L (i) = 0.89 mg/L 50 (e) = 0.01 mg/L (e) = 0.01 mg/LSettleable P (25 75 m) P (as PO4 3-) [mg/L] 10-610-410-2100102104 pdf 0.00 0.05 0.10 0.15 0.20 0.25 0.30 50 (i) = 0.31 mg/L (i) = 18.56 mg/L 50 (e) = 0.01 mg/L (e) = 0.07 mg/LSediment P (> 75 m) P (as PO4 3-) [mg/L] 10-210-1100101102 pdf 0.00 0.05 0.10 0.15 0.20 50 (i) = 2.57 mg/L (i) = 19.95 mg/L 50 (e) = 1.47 mg/L (e) = 1.19 mg/LTotal P Dissolved P Suspended P Settleable P Sediment P Influent (i) Effluent (e) Effluent (e) Influent (i) Influent-model Effluent-model Figure 3 3 Probability density function (pdf) of P fractions and TP (as PO4 3-) based on 25 monitored storm events between 2010 and 2011, including both runoff and effluent. Each pdf fits to a lognormal distribution. A box whisker plot is also showed to demonstrate the range and m edian values in each N fraction.

PAGE 81

81 Figure 3 4 Probability density function (pdf) of fd, Kd and PM concentration based on 25 monitored storm events between 2010 and 2011, including both runoff and effluent. Each pdf fits to a l ognormal distribution. A box whisker plot is also plotted to demonstrate the range and median values. Dissolved fraction, fd 0.010.11 pdf 0.000.050.100.150.200.25 50 (i) = 0.42(i) =0.25 50 (e) = 0.76 (e) = 0.18(i)(e)Effluent Influent Influent-model Partitioning coefficient, Kd [L/kg] 102103104105106107 pdf 0.000.050.100.150.200.25 50 (i) = 1.78x104 L/kg(i) = 4.77x105 L/kg50 (e) = 3.24x104 L/kg(e) = 1.91x105 L/kg fd 0.00.20.40.60.81.0 Kd [L/kg] 103104105106 Effluent-model PM (as SSC) [mg/L] 10-1100101102103104 pdf 0.000.050.100.150.200.25 50 (i) = 54.0 mg/L (i) = 140.1 mg/L50 (e) = 9.3 mg/L (e) = 6.1 mg/L SSC [mg/L] 100101102103 = 0.05

PAGE 82

82 051015202530Flow, Q (L/s) 02468101214 fd 0.00.20.40.60.81.01.2 Flow, Q Influent Effluent 051015202530Flow, Q (L/s) 02468101214 051015202530 Kd [L/kg] 102103104105106107 Flow, Q Influent Effluent Elapsed Time (min) Flow, Q (L/s) 02468101214 Elapsed Time (min) 051015202530 PM [mg/L] 100101102103104105 Flow, Q Influent Effluent 15 July 201015 July 201015 July 2010 Figure 3 5 Temporal variation of fd, Kd and PM concentration during the 15 July 2010 e vent including both runoff and effluent with flow rate embedded in the background.

PAGE 83

83 Intra-event volume, VT (m3) 02468101214 Cumulative Mass of N, MT (g) 10-410-310-210-1100101 05101520 Intra-event volume, VT (m3) 02468101214 10-410-310-210-1100101 Cumulative Mass of N, MT (g) 05101520 Dissolved N Intra-event volume, VT (m3) 02468101214 10-410-310-210-1100101 Cumulative Mass of N, MT (g) 05101520 Suspended N Intra-event volume, VT (m3) 02468101214 10-410-310-210-1100101 05101520 Sediment N Intra-event volume, VT (m3) 02468101214 10-510-410-310-210-1100101102 051015202530 Total NSettleable NEffluentRunoff EffluentRunoff EffluentRunoff EffluentRunoff EffluentRunoff Runoff (n = 299) Effluent (n = 280) Effluent-modeled Runoff-modeled Figure 3 6 Intra TVT) plots for each N fraction and TN. Runoff and effluent data from different storms were superimposed in one plot and fitted with mass transport models. The modeled parameters were tabulated on the bottom right. Fractions of Nitrogen (N) Runoff Effluent )011(TVkTeMM T TV k M0 M 0 (g) k 1 (m 3 ) k 0 (g/m 3 ) Dissolved 9.78 0.10 0.700 Suspended 15.70 0.04 0.300 Settlea ble 1.45 0.04 0.011 Sediment 4.12 0.52 0.005 Total N 18.50 0.20 1.165

PAGE 84

84 0 25 50 75 100 125 -1.0 -0.5 0.0 0.5 1.0 % removal -1.0 -0.5 0.0 0.5 1.0 Suspended N Treated Volume (m3) 0 25 50 75 100 125 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 Total N Dissolved N Sediment N Settleable N Suspended N% removal % removal % removalTreated Volume (m3) % removal 0 25 50 75 100 125 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 Suspended N Treated Volume (m3) 0 25 50 75 100 125 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0 Total P Dissolved P Sediment P Settleable P Suspended P% removal % removal % removalTreated Volume (m3) % removal % removal Figure 3 7 Event based cumulative performance ( % removal) of the VCF for N and P fractions, TN and TP as a function of cumulative treated volume.

PAGE 85

85 TOC/TIC 0.1110100 TOC:TIC ratio 0.1110100 pdf 0.000.050.100.150.20 TC [mg/L] 0.11101001000 pdf 0.000.050.100.150.20 TICTOC TC [mg/L] 0.11101001000 TIC TOC TOC:TIC ratio TICmg/LTICmg/Lp < 0.05TOCmg/LTOCmg/Lp < 0.05mg/Lmg/Lp< 0.05 Figure 3 8 Probability density function (pdf) of Total Inorganic Carbon (TIC), Total Organic Carbon (TOC), and the ratio of TOC/TIC based on 25 monitored storm events. Each pdf fits to a lognormal distribution. A box whisker pl ot is also showed to demonstrate the range and median values.

PAGE 86

86 RainfallRunoffEffluent C:N ratio 0.11101001000 C:N ratio 0.11101001000 pdf 0.00.10.20.30.40.5 50 (i) = 7.0 (i) = 16.8 C:N ratio 0.11101001000 pdf 0.00.10.20.30.40.5 Rainfall Model Effluent (e) Influent (i) Influent-Model Effluent-Model 50 (e) = 38.2 (e) = 36.1 50 = 2.9 = 2.5 (Redfield) C:N = 6.8 RainfallRunoffEffluent N:P ratio 110100 N:P ratio 0.11101001000 pdf 0.00.10.20.30.40.5 Runoff (i) Effluent (e) Runoff-Model Effluent-Model 50 (i) = 6.8i N:P ratio 1101001000 pdf 0.00.10.20.30.40.5 Rainfall Model 50 (e) = 7.0(e) = 8.5 50 = 11.6 = 28.5 (Redfield) N:P = 16 Figure 3 9 Probability density function (pdf) of C : N and N : P ratio for 25 monitored storm events, including rainfall, runoff and effluent. Each pdf was fitted to a log normal distribution. A box whisker plot is also showed to demonstrate the range and median values with the Redfield ratio marked as dotted lines.

PAGE 87

87 CHAPTER 4 NUTRIENT (N, P) SPECIATION, TRANSPORT AND TREATMENT IN RUNOFF ON A SOURCE AREA Abstract Urban runoff transports nutrients (N and P) in both aqueous and particulate phases that encompass multi ple species transformations. This study examined the transport of dissolved N and P species in runoff and treated effluent via a radial cartridge filtration (RCF) system R esult s are based on a monitoring campaign (rainfall: 236 mm, runoff: 72.5 m3) including 15 hydrologic events in 2008 generated from an urban watershed located in Gainesville, FL, which is loaded with biogenic materials. Median concentration of influent and effluent DN were 1.95 mg/L (0.64 mg/L for DP) and 2.58 mg/L (0.16 mg/L for DP). The removal efficiency of DP reaches 70% while an export of DN (32%) is observed. Nitrate Nitrite N ( NOx -) and Organic N ( ON ) in runoff occupies 41% (36% in effluent) and 54% (60% in effluent) of DN, while TAN occupies 4% of DN in both influent and effluent. DP exists primarily in the form of H2PO4 and HPO4 2-, while the proportion is determined by the pH of the environment D N species follows a first order or zeroorder model depending on the hydrologic condition of the storm, while P species tend to follow a first order (mass limited) model for all events. The mass volume correlation of TDS, alkalinity, CODd and DOC follows a linear relationship CODd and DOC shows a linear relationship with a slope of 0.355 and 0.375 for influent and effluent. Introduction Urban storm water runoff from impervious watersheds is a heterogeneous mixture that contai ns dissolved, colloidal and particulate constituents (Sansalone and Buchberger 1997). High load ing s of Metals, nutrients ( phosphorus and nitrogen),

PAGE 88

88 Particulate Matter (PM), and other anthropogenic materials have been found in urban runoff, which could cause severe environment problem s if discharged to receiving water bodies without further treatments (Ma et al. 2010; Sansalone et al. 1998; Sansalone and Kim 2008). Although n utrients (N and P) are of fundamental importance for plant growth, the excessive use of man made fertilizers and fossil fuel combustions have resulted in serious issues for the health of sur face and groundwater ecosystems ( USEPA 1998, Correll 1998). Dissolved species account for substantial proportion of the total nutrients in runoff P revious studies on roadway watersheds indicate that dissolved species could take over 50% of T N and over 30% of T P in runoff ( Vaze et al. 2004; Berretta and Sansalone 2011). In addition, the dissolved nutrients could be directly uptake by plants and microorganisms and therefore has more acute impacts to the receiving ecosystems. The toxicity, bioavailability, and transport properties of an element are determined by its chemical form. To completely understand the ways in which particular elements will affect living organisms, it is necessary to determine and to quantify the chemical form of such elements in the sample by their inorganic or organic species Analysis performed to identify and quantify one or more distinct chemical species in a sample is known as speciation anal ysis (Lund 1990) Nutrients in runoff exist in multiple forms and each species has different toxicity and bioavailability therefore, it is important to understand the speciation of nutrients in terms of treatment (Cristina and Sansalone 2003). Due to the high mobility and acute bioavailability, nitrogen in dissolved phase is more difficult to be reduced than the PM bound fraction. Opposite to the aerobic anaerobic processes that were widely used in wastewater treatment plants,

PAGE 89

89 the stochastic natural of s tormwater requires a different mechanism for nitrogen control. A variety of Best Management Practices (BMPs) have been developed that generally incorporates sedimentation, filtration, and adsorption for TN management however, most of them could not show effective removal of DN (Collins et al. 2010; Stanley 1994; Schueler and Holland, 2000) Also, conventional BMPs requires sufficient surface area and volume to avoid resuspension of P M, short circuiting or scouring, which has become a concern for ultrau rban areas (Liu et al. 2010). In order to enhance the N removal performance, m odifications includ ing bio retention, extended detention time, extended flow path, adding sufficient carbon source for denitrification, and creating aerobic/anaerobic environment s in the floor of the basin are attempted, while the performances are varied by sites (Collins et al. 2010). Due to the complex composition, stochastic and highly variable loadings of urban runoff la rge unit surface areas and volumes are required. Sorpt ive filtration is an effective way in removing particulate matters (PM), PM bound constituents and dissolved chemical species such as DP in urban runoff (Reed and Arunachalam 1994; Theis et al. 1992; Liu et al. 2004; Liu et al. 2005; Sansalone and Ma 2009). E ngineered media coated with Fe or Al oxide demonstrate high P adsorptive capacity at higher surface loading rates than nonengineered substrates (Ayoub et al. 2001; Arias et al. 2006; Sansalone and Ma 2009 ) DP adsorption depends on media surface area, pore size, surface charging, and runoff characteristics, such as DP concentration, surface loading rate, pH, and competitive ions (Sansalone and Ma 2009). Objective This study f ocus ed on the transport and fate of dissolved N and P in runoff as well as t he treatment performance of those constituents through a RCF system. The

PAGE 90

90 first objective is to assess water chemistry, hydrology characteristics, and the nutrient level of the studied watershed. The second objective is to study the speciation of dissolved N and P for influent (runoff) and effluent as a function of time, and compare the model the result s by Visual MINTEQ The third objective is to identify the pathway of n utrients from rainfall to effluent and provide indications on the nutrient fate The fourth objective is to develop a correlation between CODd and DOC for influent and effluent Methodology Source A rea W atershed D escription The studied watershed is part of a surface parking facility located in Gainesville, FL, USA. The location and plan view of the site are shown in Figure 41 .The drainage area is approximately 500m2, with an approximate 3% E W and 1.5% N S slope. About 76% of the watershed is paved with asphalt and the rest 24% were planted with landscaped vegetation ( mainly magnolia and syc amore trees ) which shed biogenic materials directly to the pavement throughout the year. This kind of parking facility that combines asphalt pavement and landscaped vegetation is commonly designed in the United States and therefore the site is representati ve to similar infrastructures in the area. The treatment system is located 9.6 m west and 2.0 m down to the sampling point During a rainfall runoff event influent (runoff) is collected as sheet flow from the catch basin to the RCF system by gravity throu gh a set of 15.2 cm diameter PVC pipes. A Parshall flume with 2.54 cm throat width and a 30 kHz MJK ultrasonic sensor are installed between catch basin and the RCF system for influent flow rate measurement. The MJK ultrasonic sensor measures the water depth at the throat of Parshall flume, which is then converted to flow rate by a previously determined calibration curve. Influent samples are sampled at the end of Parshall flume. Duplicate samples are taken

PAGE 91

91 at each sampling time. Effluent samples are taken wi th duplicates at the outlet of RCF system. Rainfall is collected before reaching the ground to characterize the atmospheric constituents washedoff by rainfall. Four acidwashed and DI water rinsed Pyrex troughs are placed at about 180 m southwest of the s tudied catchment for the duration of an event. Rainfall depth is monitored by t wo tipping bucket rain gauges located at the same position. Treatment System D esign T he d esign of radial cartridge filtration (RCF) system is shown in Figure 42 The RCF system is a passive, radial flow cylindrical cartridge deployed with Aluminum Oxide Coated Media (AOCM) which is a clay based porous sorptive media with large surface area. The AOCM media provides multiple functions including PM separation, pH buffer, and P ads orption. P revious benchscale experiments with respect to isotherms, kinetic, breakthrough and desorption evaluations prove d that t he AOCM can effectively reduce both dissolved and PM bound phosphorus (Liu et al. 2005, Sansalone and Ma 2009). The RCF syst em is comprised of a polypropylene (PP) tank (0.55 m in diameter, 0.91 m in h e igh t ) and a single cartridge (0.47 m in diameter, 0.56 m in h e igh t ) installed inside the PP tank. Two gate valves (76.2 mm in diameter) one for inlet and one for drainage are installed at 0.15 m above the bottom of the PP tank A bypass outlet of a 7 6 2 m m diameter pipe and a gate valve is designed at the top of treatment system i n case of unexpected high flow events A monometer (2 5 4 m m diameter clear PVC pipe) at the side of t ank is used to indicate water level. The media is held between an outer screen and inner mesh with plastic covers on the top and bottom. Inflow passing through the media comes into a 76.2 m m diameter perforated PVC pipe that connects to the effluent outlet, located at 1.1 m above the bottom of treatment system. The radial

PAGE 92

92 flow design maximized the contact area between runoff and media, and provides better sedimentation performance. Effluent sample is collected at the outlet of treatment system with duplicates. After each event, about 60 L iters of runoff is stored inside the RCF, which provides adequate matrix for micro organism growth, and generating an aerobic/ an ox ic environment inside the RCF that could promote denitrification. S ampling P rotocol The field test was conducted in 2008 with a total of 15 rainfall runoff events captured. About 10 f ull cross sectional samples were taken manually at influent and effluent sampling point at 2 5 minutes intervals throughout the entire event to avoid unrepresentative sampling (Sansalone and Cristina 2004; Appel and Hudak 2001) T hree sets of sampling containers including four 0.5 L, two 1.0 L and two 4.0 L (or 5 gallon bucket for effluent) wide mouth polypropylene bottles are used for each influent sample. The four 0. 5 L replicated samples are used for particle size distribution (PSD) and suspended sediment concentration (SSC) analysis. The 1.0 L replicate samples are immediately fractionated into sediment (larger than 75 m), settleable (between 75 and 25 m), suspended (between 25 and 0.45 m) and dissolved (smaller than 0.45 m) fraction, followed by Standard Method 2540D, 2540E (APHA, 1992) and ASTM D397797 (ASTM, 1997). I n order to minimize time dependent changes in speciation, t his process must be done within 2 hours from the time of sampling. Detailed procedures of fractionation could be found in Sansalone and Kim (2008) and Kim and Sansalone (2008). Each PM fraction is carefully recovered, dried and measured for PM bound nutrient analys e s. The 4.0 L influent and the 5 gallons effluent sampl es are used to recover sediment siz ed PM from each sample.

PAGE 93

93 Monitored H ydrologic P arameters Hydrological parameters includes rainfall intensity ( Imax) rainfall depth ( Prain) influent (runoff) flow rates ( Q50 and Qmax) even t durations ( drain) previous dry deposition ( PDH) volumetric runoff coefficient (C ) and unsteadiness ( ) as shown in Table 41. The volumetric runoff coefficient ( C ) is a site specific parameter that represents the fraction of runoff that will result f rom a given volume of rainfall ( Chow et al. 1988) with the formula shown as follow nwnnAIQC (4 1) In this expression, Qn represents volumetric flow rate of the n th interval observed at the outlet in [L3/T]; In is rainfall intensity of the n th interval in [L/T]; and Aw is cont ributing area of the watershed in [L2]. The unsteadiness ( U ) is a dimensionless factor that characterizes the variation of flow rate during a storm. For a typical storm, the range usually goes from 0~1 while higher value indicates the flow is changing more dramatically during the event. The formula is shown as follow (Garofalo and Sansalone 2011) tdQd50 ( 4 2) In this expression 50 represents the median flow, Q represent s the normalized flow rate with respect to peak flow rate ( Qp) t represents the normalized value of t with respect to duration ( tp). The Summary of analyses methods are tabulated in Table 42 pH, redox potential, conductivity /salinit y/ total dissolved solids (TDS), dissolved oxygen ( DO ) and temperature are measured by a thermo 5star multiple probe meter with designated

PAGE 94

94 electrodes attached. Each electrode is calibrated before each storm event by following standard calibration procedur es A lkalinity is measured by the titration method and results are reported in mg/L as CaCO3. Turbidity is measured by a Hach 2100 Turbidimeter Ion concentrations includ ing PO4 3-, NO3 -N, NO2 -N TAN (NH3+NH4 +N ), SO4 2and Clare determined by colorimet ric methods via designated Hach test kits, and the absorbance is measured by a Hach DR/5000 spectrophotometer. Metals including Fe3+, Al3+, Cu2+, Zn2+, Pb2+, Mg2+, Ca2+ and K+ are measured by an Inductively Coupled Plasma Mass Spectrometry (ICP MS) while Na+ is measured by an ion selective electrode (ISE) connected with a thermo probe meter. Dissolved Organic Carbon (DOC) is measured through an UV Persulfate TOC analyzer following standard method 5310C (APHA 1992 ) Nutrient S peciation The Mineral thermody namic equilibrium model Visual MINTE Q v3.0, is used for nutrient speciation in this study. The MINTEQ utilizes temperature, pH, redox potential alkalinity, aqueous concentrations of cations, anions, ligands dissolved organic matter (DOM) and PM along wi th a charge balance to determine the concentration of each aqueous species formed under specified conditions (Allison et al. 1991; Dean et al. 2005) The model is based on an iterative method to solve reaction equations and mass action expressions using log equilibrium constants to find a numerical solution to equilibrium concentrations (Allison et al. 1991). Data E laboration Due to the stochastic nature of stormwater, the concentrations are often very unsteady during a rainfall runoff event. Therefore, a single index that represent s a large number of analyses i s required to obtain event based estimates and to compare results

PAGE 95

95 with other literatures. Event mean concentration ( EMC) is often used that represent s a flowaveraged concentration of a storm event ( Huber 1993, Sansalone and Buchburger 1997). The formula of EMC is shown as follow rrttdttqdttqtcEMC00 (4 3) In this expression, c(t) is the time variable nutrients(N or P) fraction [M/L3] ; q(t) is the time variable flow [L3/T] ; tr is the duration of the event [T], and t is time [T] Another way to obtain a sitebased distribution is to perform a probability density function (pdf), where samples falls within a particular range is represented by the frequency of occurrence. The pdf of runoff constituents are often observed following lognormal distribution (Buren et al. 1996; Berretta and Sansalone 2012), and therefore the median, which is the peak of distribution, is representative to the entire data set. Percent removal (PR) is commonly used to quantify the removal efficiency for a constituent on an event basis the formula is shown as follow 100)()()((%)111INiniINiEFFjmjEFFjINiniINiCVCVCVPR % (4 4) In this expression Vi IN and Vj EFF are the volume of influent flow and effluent flow during sampling period i and j ; Ci IN and Cj EFF are the mean concentration associated with period i and j ; and n and m are the total number of influent and effluent measurements taken during event, respectively.

PAGE 96

96 Results and Discussions Hydrologic P rofiles and W ater C hemistry C haracterization T he event based hydrological parameters are summarized in Table 41. The source area shows obvious seasonal change where higher rainfall intensity and shorter PDH events are more common during wet season (June September), while lower rainfall intensity and longer PDH events are more often observed during dry season (October May). previous dry hours (PDH) of the 15 rainfall runoff events captured in 2008 ranges from 66 to 960 hours with a median of 230 hours. Rainfall depth ranges from 0.3 to 74.2 mm with a median of 5.8 mm. Duration ranges from 35 to 197 minutes with a median of 62 minutes. Median runoff flow rate ranges from 0.11 to 3.60 L/s with a median of 0.55 L/s, and the total treated runoff volume is 72.5 m3. Site based mean and median values as w ell as the range of water chemistry indices of influent (runoff) and effluent are summarized in Table 43 and 4 with the probability density functions (pdf) plotted in Figure 43 and 4. Each pdf fits a lognormal distribution. In general, both TN and TP fr om the source area exceed the Florida numeric nutrient criteria (FDEP 2012), requiring load control before discharge. The median DN for influent and effluent is 1.95 (0.64 for DP) and 2.58 (0.19 for DP) mg/L, and the values in PM N is 1.22 (1.82 for PM P) and 0.80 (0.75 for PM P) mg/L. The median dissolved fraction of N are 62% and 76% for influent and effluent, and the dissolved fraction of P are 26% and 20% in influent and effluent. The elevated nutrient levels are generally due to the biogenic material s from the catchment area, such as grass clipping and leaf fall. In influent median NOx -N and DON which accounts for 36% (30% for effluent) and 46% (48% for effluent) of DN are the predominant DN species The increase of effluent ON could be caused by the decomposition of biogenic materials

PAGE 97

97 during retention periods or microbial assimilation of inorganic N. The median TAN in influent and effluent is 0.07 and 0.05 mg/L, which is about a magnitude smaller than the NOx -N and ON. Median DP in influent and eff luent are 0.64 and 0.19 mg/L, Orthophosphates (PO4 3-, HPO4 2and HPO4 -) accounts for more than 90% DP mass, and their speciation is determined by the pH of environment. With respect to the pH of runoff and effluent, which ranges from 5.4 to 8.5, the order of species predominance is consistently HPO4 2H2PO4 ->>CaHPO4>MgHPO4 (Berretta and Sansalone 2011). Nutrient Pathways during Rainfall Runoff Events The pathways of PM from the source area to the discharging effluent are show n in Figure 42. Rainfall water chemistry indices, which is summarized in Table 42 show that rainfall is slightly ac idic (median pH: 4.3), aerobic (median redox: 495 mV) low conductivity (median conductivity: 16.6 S/cm) and low turbidity (median turbidity: 0.9 NTU). Therefore, the PM fraction is negligible in rainfall while al l constituents are considered dissolved. The median concentrations of DN and DP in rainfall are 0.65 mg/L and 0.11 mg/L, which accounts for 15 % and 4 % of TN and TP in runoff. Dry deposition on the catchment is the major nutrient source, which is a heterogenic mixture of inorganic and organic materials with a wide particle size gradation. PSD result shows that the median diameter (d50) of dry deposition is 280 m. When rainfall precipitates on the catchment and generates runoff a complex process including dissolution, partitioning, ion exchange, suspension, precipitation, and adsorption occurs between the solidaqueous interfac es resulting in a significant change in water chemistry indices. In i nfluent (runoff) t he pH ranges from 5.4 to 8.5 with a median of 7.4, redox potential ranges from 217 to 495 mV with a median of 353

PAGE 98

98 mV, conductivity ranges from 13 to 214 with a median of 84 S/cm which is much higher than the rainfall. The d50 of PM transported by runoff is 154 m, w hich is about half of the dry deposition, meaning that finer particles are more likely to be transported by runoff than coarser fractions. When influent is conveyed to the RCF system, the media function as an adsorptive substrate for P and a granular filter for PM separation. In addition, the cylinder tank provides certain sedimentation and retention capacity. The removal efficiency for DP is 70 % while D N species do not show any removal efficiency. In fact, an elevation of ON is observed in some events due to the decomposition of biogenic materials during retention time. The PM separation efficiency is over 90%. Influent (runoff) PM ranges from 89 to 2049 mg/L with a median of 399 mg/L, while effluent PM ranges from 15 to 117 mg/L with a median of 42 mg/L. Results on the PM based N, P fractions also showed an 35% removal efficiency via the filtration and sedimentation function of the RCF system. The d50 of effluent is 25 m which is even smaller than that of influent, indicating the RCF system can effectively separate large PM fraction. After each rainfall runoff event, about 60 liters of runoff was stored in the RCF system to monitor the interevent transformations of nutrients. Due to the microbial activities, the stored runoff is a low pH (median: 6.5), low redox (median: 220mV), high conductivity (median: 381 S/cm ), high alkalinity (median: 150 mg/L as CaCO3) and high nutrient concentration (median DN: 2.63 mg/L, median DP: 0.97 mg/L) solution. However, the difference between influent and effluent in pH, redox, conductivity, TDS, alkalinity CODd and DOC are not significantly different, indicating that the impact of

PAGE 99

99 stored runoff on effluent is limited due to the rel ative small volume compared to the volume of runoff ( median: 1586 L ). Intra E vent N utrient T ransport B ehavior The temporal variation of concentration for DN and DP species and the mass transport behavior as cumulative mass versus volume for the and 08 Jul y 2008 events as shown in Figure 45 and 4 6 where the 12 August 2008 event is a low volume (3161 L) and short duration (90 minutes) event and the 08 July 2008 event is a high volume ( 28482 L) and lo ng duration (157 minutes) event. The concentration of D N species and DP are measured by analyses, and the water chemistry indices and major ions concentrations are inserted to Visual Minteq to calculate the equilibrium concentration of each DN and DP species. The charge balance error must be less than 10% for accurate modeled results. Result found that the major DP species are HPO4 2and H2PO4 because of the pH, and the RPD between measured and modeled NOx -N and TAN are less than 2%, as shown in Figure 47 which is because of the high mobility and stability of NOx -N and TAN. The cumulative mass versus volume plot indicates that DN species in low volume, short duration events follows a zeroorder (flow limited) model, where flow is the limiting factor for mass transport; while for high volume, long duration events, DN species follows a first order exponential model (mass limited) in that most DN has been washed out during the "first flush". The transport of DP species follows a first order (mass limited) model for all events, which is due to the limited sourc es of DP in the source area. The intraevent variation of nutrient is influenced by many factors, such as water chemistry conditions and hydrologic characteristics. The temporal variation, pH, redox,

PAGE 100

100 TDS, alkalinity, CODd and DOC for t he July 08, 2008 event are shown in Figure 48 pH showed a maximum value while redox showed a minimum value at the beginning of the event due to the dissolution and repartitioning between rainfall and dry deposition. TDS and alkalinity both decreases along with the time, whi le the initial values in effluent is higher than influent, which could be related to the stored runoff during retention periods. The cumulative mass versus volume plot s of TDS, alkalinity, CODd and DOC in influent and effluent follow zeroorder model s. The trans port rate ( k0) of CODd in influent is much higher than the effluent while the difference in alkalinity TDS and DOC between influent and effluent are not significant C orrelation between CODd and DOC U nderstanding the correlation between CODd and DOC in runoff it is possible to use either one parameter as an surrogates of the other parameter. build a model. The correlation between CODd and DOC are plotted based on 192 data points from 15 rainfall runoff events as shown in Figure 49. Result s found that CODd and DOC follow linear relationship with a slope of 0.355 and 0.375 for i nfluent and effluent with a coefficient of determination (R2) of 0.73 and 0.87. The slope for effluent is slightly higher than the influent could possibility due to the decompos ition of biogenic materials inside the RCF during retention periods The ranges of CODd and DOC in effluent are smaller than the influent Conclusion This research focuses on the intra event speciation, transport and fate of dissolved nutrients (N and P ) i n urban runoff from a biogenic loaded catchment, where a radial cartridge filtration (RCF) system is instrumented for treatment purpose. Fifteen rainfall runoff events were monitored in 2008. Result found that dissolved nitrogen (DN)

PAGE 101

101 and dissolved phosphor us (DP) in rainfall represents 15% and 4% of TN and TP in runoff indicating the dry deposition is the major nutrient source to the runoff Result found that both TN and TP from the source area exceed the Florida numeric nutrient criteria (FDEP 2012), and therefore, load control before discharge is required. Results indicate that NOx and organic nitrogen (ON) account for over 90% of the D N species while HPO4 2and H2PO4 are the major DP species and their predominance is depending on the pH of runoff Speciation results found that NOx and NH4 + can be assumed equilibrated due to the high mobility of these species. For high volum e and long duration event s, D N species tend to follow a first order (mass limited) transport model, while for low volume and short duration event s, D N species tend to follow a zeroorder (flow limited) transport model. Dissolved P species tend to follow a first order (mass limited) transport model for all event s due to the limitation of phosphorus sources. The impact of stored runoff on effluent is limited due to the relative small volume compared to the rainfall runoff event. TDS, alkalinity, CODd and DOC follow zeroorder model s, while t he CODd and DOC showed a linear relationship with a slope of 0.355 and 0.375 for influent and effluent.

PAGE 102

102 Table 4 1 Event based hydrology parameters for the 15 rainfall runoff events in 2008 and statistics shown as mean ( ), median( 50), standard deviation( ),minimum ( Min ), and maximum (Max ) 2008 Events IPRT PDH drain Imax Prain Vrunoff Qmax Q50 C Hydrograph Gamma Parameters (min) (hr) (min) (mm/hr) (mm) (L) (L/s) (L/s) k 20 Mar 2 230 36 45.7 0. 3 1838 3.4 0 0.88 0.6 0.8 11.1 0.184 16 May 12 960 62 91.4 10.7 2142 8.75 0.9 0 0.4 37.7 0.5 0.223 3 Jun 8 432 69 15.2 2 .0 394 0.75 0.11 0.3 4 .6 2.3 0.096 10 Jun 2 168 94 152.4 30.7 7129 11 .0 1.61 0.5 0.7 22.2 0.037 21 Jun 6 144 127 61 .0 6.6 1260 3.98 0.34 0.3 8.1 0.8 0.135 8 Jul 4 141 157 167.6 74.2 28482 13.2 2.14 0.8 3.6 8.5 0.126 15 Jul 9 66 79 167.6 62.2 20037 13.1 3.6 0 0.7 12.6 3.2 0.0 88 29 Jul 6 309 35 30.5 5.6 1522 3.69 0.55 0.5 2 .0 6.7 0.274 8 Aug 3 191 42 30.5 3 .0 402 2.16 0.14 0.3 3.1 2.4 0.163 12 Aug 7 93 90 45.7 16.3 3161 3.92 0.62 0.5 14.1 1.6 0.072 19 Aug 4 117 57 45.7 4.3 1664 6.78 0.81 0.7 9.7 0.6 0.712 10 Sep 4 264 56 3 0.5 8.1 1237 1.96 0.47 0.4 10.5 2.9 0.201 20 Sep 3 234 36 45.7 2.8 540 1.09 0.24 0.4 1.6 4.5 0.219 8 Oct 3 431 180 30.5 5.8 1586 2.75 0.14 0.5 9.2 0.7 0.233 23 Oct 3 334 56 30.5 3.6 1051 1.46 0.29 0.6 1.6 13.6 0.190 50 4 230 62 45.7 5.8 1586 3.69 0.55 0.5 4.6 2.9 0.184 5 274 81 66.0 15.7 4830 5.19 0.86 0.5 8.0 5.4 0.197 3 221 50 53.0 22.7 8211 4.30 0.95 0.2 9.4 6.1 0.158 Min 2 66 35 15.2 0.3 394 0.75 0.11 0.3 0.7 0.5 0.037 Max 12 960 157 167.6 74.2 28482 13.17 3.60 0.8 37.7 22.2 0.712

PAGE 103

103 Table 4 2 Event based statistic of rainfall chemistry, shown as mean ( ), median( 50), standard deviation( ), minimum ( Min ), maximum ( Max ), number of events ( n ), and standard method. Note: the unit of pH is standard unit (s. u.), and the unit of alkalinity is m g/L as CaCO3. Rainfall Water Chemistry Unit 50 Min Max n Method pH s.u. 4.3 4.4 0.2 3.8 4.9 49 S.M. 4500 H + B Redox mV 495 477 89 267 638 49 S. M. 2580 DO mg/L 7.6 7.0 1.5 4.3 8.8 49 S. M. 4500 O Temperature C 23.7 23.7 2.1 17.1 28.0 49 S. M. 2 550 Conductivity S/cm 16.6 18.4 12.1 4.4 69.7 49 S. M. 2510 TDS mg/L 8 9 6 2 35 49 S. M. 2510 Alkalinity mg/L 0.0 0.6 1.4 0.0 6.0 49 S. M. 2320 B COD d mg/L 5.7 9.0 8.5 1.3 40.0 49 Reactor Digestion Turbidity NTU 0.9 1.4 1.2 0.2 5.1 49 S. M. 2130 B TN mg/L 0.65 0.78 0.53 0.05 2.97 49 Persulfate Digestion PO 4 3 mg/L 0.11 0.14 0.10 0.01 0.50 49 Ascorbic Acid SO 4 2 mg/L 0.84 0.95 0.64 0.30 2.14 15 SulfaVer 4 Cl mg/L 0.49 1.05 0.94 0.14 2.95 15 Mercuric Thiocyanate K + mg/L 1.9 2.7 2.3 0.0 7.8 15 S M. 3030B Mn 2+ g/L 1.6 2.1 1.7 0.1 4.5 15 S. M. 3030B Fe 3+ g/L 16.1 18.1 11.2 2.2 50.2 15 S. M. 3030B Al 3+ g/L 10.3 9.8 3.8 2.8 16.8 15 S. M. 3030B Cu 2+ g/L 0.2 0.3 0.2 0.1 0.7 15 S. M. 3030B Zn 2+ g/L 8.8 8.1 4.3 1.1 13.8 15 S. M. 3030B Cd 2+ g /L 0.1 0.1 0.0 0.0 0.2 15 S. M. 3030B Pb 2+ g/L 0.1 0.2 0.2 0.0 0.6 15 S. M. 3030B Mg 2+ g/L 0.8 1.4 1.7 0.3 6.8 15 S. M. 3030B Ca 2+ g/L 34.7 43.7 44.6 0.9 185.5 15 S. M. 3030B Ni 2+ g/L 11.1 12.7 7.5 5.1 35.1 15 S. M. 3030B Cr 2+ g/L 0.3 0.5 0.6 0.0 2.3 15 S. M. 3030B Ba 2+ g/L 6.6 10.9 11.8 1.4 46.5 15 S. M. 3030B As 2+ g/L 0.3 0.4 0.5 0.0 2.0 15 S. M. 3030B

PAGE 104

104 Table 4 3 Sample based Statistics of major water chemistry indices and other dissolved ions other than nutrients for the 15 rainfall runoff events in 2008, shown as mean ( ), median( 50), standard deviation( ) minimum (Min ), and maximum (Max ) for both influent (runoff) and effluent. Note: HCO3 is calculated by alkalinity and pH. Water chemistry Units 50 Min Max Inf. E ff. Inf. E ff. Inf. E ff. Inf. E ff. Inf. E ff. pH s.u. 7.4 6.9 7.4 6.9 0.4 0.2 5.4 6.2 8.5 7.3 R edox mV 353 364 352 360 59 88 217 100 495 581 Cond. S/cm 84 94 87 121 58 87 13 20 214 330 TDS mg/L 36 38 49 72 45 78 4 5 272 597 COD d mg/L 67 46 104 7 1 116 62 6 0 727 310 COD t mg/L 115 61 179 94 200 81 11 0 1253 408 Alkalinity mg/L 29 28 33 47 22 49 5 7 171 271 DOC mg/L 21 15 39 26 46 25 1 1 243 121 PM mg/L 399 42 594 4 8 490 25 89 15 2049 117 SO4 2 mg/L 5.18 6.47 5.42 9.34 4.38 11.54 0.03 1.60 14.15 48.33 Cl mg/L 2. 20 2.71 6.56 6.58 8.29 8.62 0.48 0.47 25.86 29.48 HCO 3 mg/L 0.36 0.37 0.33 0.40 0.16 0.23 0.09 0.11 0.66 0.79 Mg 2+ mg/L 1.67 2.25 1.98 2.69 1.25 1.89 0.33 0.59 5.11 7.13 Ca 2+ mg/L 12.61 16.19 13.73 20.18 7.60 13.46 2.89 4.86 31.00 49.67 K + mg/L 2.10 2 .91 3.56 2.92 3.16 2.29 0.08 0.25 11.36 8.02 Na + mg/L 1.54 1.77 3.03 2.28 5.65 2.28 0.44 0.41 23.22 9.42 Mn 2+ 12 18 17 24 16 21 2 0 64 64 Fe 3+ 46 61 71 103 66 91 5 15 254 296 Al 3+ 167 168 174 182 33 64 111 108 247 370 Cu 2+ 26 15 25 16 17 10 3 1 53 38 Zn 2+ 40 53 56 53 31 23 17 17 119 101 Cd 2+ 2 2 20 3 69 2 1 0 271 8 Pb 2+ 20 21 31 31 31 33 2 4 102 109 Ni 2+ 6 7 6 6 3 4 2 2 11 14 Cr 2+ 5 5 4 5 2 2 2 3 7 8 Ba 2+ 7 15 9 15 5 9 2 5 19 32 Charge Balance 10% 9% 10% 10% 6% 6% 1% 1% 20% 18%

PAGE 105

105 Table 4 4 Event mean concentration (EMC) of dissolved nutrients, including dissolved phosphate ( PO4 3-), nitratenitrite N (NOx -N), total ammonia N (TAN), organic N (ON) and dissolved N (DN) for the 15 rainfall runoff events in 2008, both influent (runoff) and effluent are tabulated with statistics shown as mean ( ), median( 50), standard deviation( ), minimum (Min ), and maximum (Max ) 2008 Events PO4 3 [mg/L] NOx N [mg/L] TAN [mg/L] ON [mg/L] DN [mg/L] Inf. Eff. Inf. Eff. Inf. Eff. Inf. Eff. Inf. Eff. 28 Mar 0.64 0.16 1.72 3.38 0.23 0.26 0.90 1.24 2.85 4.88 16 May 0.52 0.12 0.71 2.29 0.07 0.07 2.34 1.11 3.12 3.47 3 Jun 1.42 0.07 2.13 0.71 0.21 0.03 1.78 3.44 4.11 4.18 10 Jun 0.76 0.23 0.67 0.80 0.23 0.15 2.71 3.68 3.61 4.63 21 Jun 0.24 0.13 1.12 0.80 0.13 0.16 2.73 2.54 3.98 3.50 8 Jul 0.21 0.15 2.31 0.63 0.02 0.02 1.87 0.19 4.20 0.84 15 Jul 0.47 0.19 0.59 0.46 0.05 0.05 0.34 0.24 0.98 0.75 29 Jul 2.56 0.51 0.54 0.78 0.06 0.04 1.94 1.76 2.53 2.58 8 Aug 0.53 0.11 1.01 1.08 0.11 0.04 0.81 2.93 1.93 4.05 12 Aug 0.93 0.33 0.29 0.73 0.02 0.00 0.60 0.38 0.91 1.12 19 Aug 0.43 0.30 0.69 0.71 0.04 0.11 1.22 4.49 1.95 5.31 10 Sep 0.68 0.28 0.56 0.53 0.00 0.03 0.39 0.81 0.96 1.37 20 Sep 0.66 0.22 1.05 0.89 0.08 0.27 0.59 0.34 1.72 1.50 8 Oct 0.94 0.22 0.52 0.80 0.03 0.05 0.33 0.51 0.88 1.35 23 Oct 0.45 0.08 0 .73 0.28 0.17 0.19 0.66 1.25 1.57 1.72 50 0.64 0.19 0.71 0.78 0.07 0.05 0.90 1.24 1.95 2.58 0.76 0.21 0.97 0.99 0.10 0.10 1.28 1.66 2.35 2.75 0.58 0.12 0.61 0.80 0.08 0.09 0.87 1.41 1.23 1.61 Min 0.21 0.07 0.29 0.28 0.00 0.00 0.33 0.19 0.88 0.75 Max 2.56 0.51 2.31 3.38 0.23 0.27 2.73 4 .49 4.20 5.31

PAGE 106

106 Figure 4 1 Location of the studied watershed. The location of Gainesville, FL is marked on the Florida county map, and the studied watershed is circled in the University of Florida Campus map. The approximate drainage area for the sa mpling catch basin is shown as a shaded area on the site map (not draw n to scale) N Alachua County, FL

PAGE 107

107 Figure 4 2. The granulometric distribution of PM in dry deposition, influent and effluent as a flow chart to show the pathway of PM during a storm event. The Design of the radial cartridge filtration (RCF) system is also included with arrows showing the flow direction.

PAGE 108

108 Rainfall depth (mm) 0.01 0.1 1 10 100cdf 0.0 0.5 1.0 Rainfall : 50 = 2.1 mm = 4.50 mm Particle diameter ( m) 10-1100101102103PM mass fraction 0.0 0.5 1.0 Filter Effluent : d 50 = 25 m = 0.84 = 45.8 Gamma distribution Total Vrunoff :72.5 m3 Qmax : 0.75~13.17 L/s Effluent was directly discharged to Lake Alice [PM]50 = 34 mg/L [PM] 50 = 180 mg/L Filter S ystem Particle diameter ( m) 10-1100101102103PM mass fraction 0.0 0.5 1.0 Watershed : Dry deposition d 50 = 280 m = 1.79 = 194 Area : ~500 m 2 N S slope : 1.5% E W slope : 3 .0 % PDH : 66~960 hrs tc : 2 12 min S urfac e : Asphalt and Vegetated C : 0.3 0.7 log normal distribution Distribution based on 19982008 rainfall d epth of GNV airport (NCDC,2009) drain:35197 minutes Particle diameter (m) 10-1100101102103PM mass fraction 0.00.51.0 d50 = 154 m = 0.80 = 144 0.54 m 0.15 m UnderdrainEffluentBypass ManometerInfluent Runoff 0.56 m 0.47 m 0.91 m 0.56 m Sludge Zone 1.68 m ValveValveValve Q (L/s) 10-310-210-1100101102cdf 0.0 0.5 1.0 Runoff (Influent): 50 = 0.62 L/s = 3.08 L/s

PAGE 109

109 pH 5 6 7 8 9 10pdf 0.00 0.25 0.50 5 6 7 8 9 10pdf 0.00 0.25 0.50 Effluent Influent Redox (mV) 200 400 600 100 1000 pdf 0.00 0.25 200 400 600 100 1000 pdf 0.00 0.25 Effluent Influent TDS [mg/L] 1 10 100 1000pdf 0.0 0.1 0.2 1 10 100 1000pdf 0.0 0.1 0.2 Effluent Influent Alkalinity as CaCO3 [mg/L] 1 10 100 1000 pdf 0.0 0.1 0.2 1 10 100 1000 pdf 0.0 0.1 0.2 Effluent Influent CODd [mg/L] 1 10 100 1000pdf 0.0 0.1 0.2 1 10 100 1000pdf 0.0 0.1 0.2 Effluent Influent CODt [mg/L] 1 10 100 1000 pdf 0.0 0.1 0.2 1 10 100 1000 pdf 0.0 0.1 0.2 Effluent Influent PM [mg/L] 1 10 100 1000 10000pdf 0.0 0.1 0.2 1 10 100 1000 10000pdf 0.0 0.1 0.2 Effluent Influent DOC [mg/L] 1 10 100 1000 pdf 0.0 0.1 0.2 1 10 100 1000 pdf 0.0 0.1 0.2 Effluent Influent Figure 4 3 Probability density function (pdf) of pH, redox, TDS, alkalinity, CODd and DOC con centrations based on 189 influent and 185 effluent samples captured in 2008.

PAGE 110

110 NOx--N [mg/L] 10-310-210-1100101102pdf 0.000.25 10-310-210-1100101102pdf 0.000.25 EffluentInfluent TAN [mg/L] 10-310-210-1100101102 pdf 0.00.10.2 10-310-210-1100101102 pdf 0.00.10.2 EffluentInfluent DON [mg/L] 10-310-210-1100101102pdf 0.00.10.2 10-310-210-1100101102pdf 0.00.10.2 EffluentInfluent TDN [mg/L] 10-310-210-1100101102 pdf 0.00.10.20.3 10-310-210-1100101102 pdf 0.00.10.20.3 EffluentInfluent TDP [mg/L] 10-310-210-1100101102pdf 0.00.20.4 10-310-210-1100101102pdf 0.00.20.4 EffluentInfluent PM-P [m/L] 10-310-210-1100101102 pdf 0.00.20.4 10-310-210-1100101102 pdf 0.00.20.4 EffluentInfluent PM-N [mg/L] 10-310-210-1100101102pdf 0.00.20.4 10-310-210-1100101102pdf 0.00.20.4 EffluentInfluent Flow, Q (L/s) 10-310-210-1100101102 pdf 0.00.10.2 10-310-210-1100101102 pdf 0.00.10.2 EffluentInfluent Figure 4 4 Probability density function (pdf) of NOx -N, TAN, ON, DN, DP and flow rate based on 189 influent and 185 effluent samples captured in 2008.

PAGE 111

111 Elapsed time (min) 020406080100Flow, Q (L/s) 0.00.51.01.52.02.53.03.5 P [mg/L] 012345 Flow HPO42H2PO4TDP Elpased time (min) 020406080100Flow, Q (L/s) 0.00.51.01.52.02.53.03.5 N species [mg/L] 02468 Flow NOx--N TAN TON TDN 12 August, 2008 Elapsed time (min) 020406080100 Flow, Q (L/s) 0.00.51.01.52.02.53.0 P [mg/L] 012345 Flow HPO42H2PO4TDP Elpased time (min) 020406080100 Flow, Q (L/s) 0.00.51.01.52.02.53.0 N species [mg/L] 02468 Flow NOx--N TAN TON TDN InfluentEffluentInfluentEffluent M (g) 01234 NOx--N TAN TON TDN V (m3) 0.00.51.01.52.02.53.03.5 M (g) 01234 NOx--N TAN TON TDN M (g) 01234 HPO42H2PO4TDP V (m3) 0123 M (g) 01234 HPO42H2PO4TDP InfluentEffluentInfluentEffluent Figure 4 5 Temporal variation of major N and P species for the August 12, 2008 event (low volume, short duration event) with hydrograph embedded. The cumulative mass volume plots for each species are shown on the bottom to demonstrate the transport behavior. Each dissolved N species fits a zeroorder (flow limited) model, while dissolved P species are fitted to a firstorder (mass limited) model. a. b. c. d. e. f. g. h.

PAGE 112

112 Elapsed time (min) 0 30 60 90 120 150 180 210Flow, Q (L/s) 0 1 2 3 4 5 P [mg/L] 0.0 0.2 0.4 0.6 Flow HPO4 2H2PO4 TDP Elpased time (min) 0 30 60 90 120 150 180 210Flow, Q (L/s) 0 1 2 3 4 5 N species [mg/L] 0 10 20 30 Flow NOx --N TAN TON TDN 08 July, 2008 Elapsed time (min) 0 30 60 90 120 150 180 210 Flow, Q (L/s) 0 1 2 3 4 5 P [mg/L] 0.0 0.2 0.4 0.6 0.8 Flow HPO4 2H2PO4 TDP Elpased time (min) 0 30 60 90 120 150 180 210 Flow, Q (L/s) 0 1 2 3 4 5 N species [mg/L] 0 10 20 30 Flow NOx --N TAN TON TDN Influent Effluent Influent Effluent M (g) 0 10 20 30 40 50 V (m3) 0 2 4 6 8 10 12 M (g) 0 10 20 30 40 50 NO x -N TAN TON TDN M (g) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 HPO4 2H2PO4 TDP V (m3) 0 2 4 6 8 10 12 M (g) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 HPO4 2H2PO4 TDP Influent Effluent Influent Effluent Figure 4 6 Temporal variation of major N and P species for the July 08, 2008 event (high volume, long duration event) with hydrograph embedded. The cumulative mass volume plots for each species are shown on the bottom to demonstrate the transport behavior. Each dissolved N and species fits a first order (mass limited) model. a. b. c. d. e. f. g. h.

PAGE 113

113 Elapsed time (min) 0 20 40 60 80 100 120 140 160 Flow, Q (L/s) 0 2 4 6 8 10 12 NOx --N [mg/L] 0 3 6 9 12 15 18 21 NOx-N (mg) 103104105 Flow Measured Modeled NOx --N-measured NOx --N-modeled Elapsed time (min) 0 20 40 60 80 100 120 140 160 Flow, Q (L/s) 0 2 4 6 8 10 12 TAN [mg/L] 0.00 0.02 0.04 0.06 0.08 0.10 TAN (mg) 101102103 Flow Measured Modeled TAN-measured TAN-modeled 0 20 40 60 80 100 120 140 160 RPD (%) 0 1 Elapsed time (min) 0 20 40 60 80 100 120 140 160 RPD (%) 0.0 0.5 July 08, 2008 July 08, 2008 Figure 4 7 Measured and modeled NOx -N and TAN concentration for the July 08, 2008 event with cumulative mass superimposed as lines. T he relative percentage difference (RPD) between measured and modeled results is shown on top of each species. Note: T he difference between cumulative measured and modeled mass are indistinguishable.

PAGE 114

114 Elapsed time (min) 0 30 60 90 120 150 180 pH (s.u.) 4 5 6 7 8 9 Redox (mV) 320 340 360 380 400 420 440 Flow pH redox TDS [mg/L] 0 50 100 150 200 Alkalinity [mg/L as CaCO3] 0 50 100 150 200 250 TDS Alkalinity July 08, 2008 Influent Elapsed time (min) 0 30 60 90 120 150 180 pH (s.u.) 4 5 6 7 8 9 Redox (mV) 320 340 360 380 400 420 440 TDS [mg/L] 0 50 100 150 200 Alkalinity [mg/L as CaCO3] 0 50 100 150 200 250 Effluent Effluent Elapsed time (min) 0 30 60 90 120 150 180 CODd [mg/L] 0 20 40 60 80 100 DOC [mg/L] 0 10 20 30 40 50 CODd DOC Elapsed time (min) 0 30 60 90 120 150 180 CODd [mg/L] 0 20 40 60 80 100 DOC [mg/L] 0 10 20 30 40 50 Effluent Influent Influent Flow pH redox TDS Alkalinity Volume (m3) 0 5 10 15 20 25 30 0 200 400 600 800 1000 Mass [g] 0 200 400 600 800 1000 CODd (29.9) Alkalinity (9.1) TDS (5.9) DOC (2.2) Influent Volume (m3) 0 5 10 15 20 25 30 Mass [g] 0 200 400 600 800 1000 0 200 400 600 800 1000 CODd (16.3) Alkalinity (9.9) TDS (7.4) DOC (2.1) Effluent CODd DOC Figure 4 8 ( a f ) Temporal variations of pH, redox, TDS, alkalinity, CODd and DOC for the J uly 08, 2008 event. ( g h ) Cumulative mass volume plot for CODd, alkalinity, TDS and DOC. Both influent and effluent fit zero order (linear) transport model wi th slopes shown in parentheses. The coefficient of determination (R2) for each regression curve are larger than 0.97. a. b. c. d. e. f. g. h.

PAGE 115

115 CODd [mg/L] 0200400600800 DOC [mg/L] 050100150200250300 CODd [mg/L] 0200400600800 DOC [mg/L] 050100150200250300 slope = 0.375r2 = 0.87n = 181slope = 0.355r2 = 0.73n = 192InfluentEffluent Figure 4 9 Correlation between dissolved chemical oxygen demand (CODd) and dissolved organic car bon (DOC) I nfluent and effluent samples are plotted separately and fitted with linear regression curves The slopes and coefficient of determination ( R2) are shown in the plot. The slopes of influent and effluent are not statistically significant different (p value < 0.05).

PAGE 116

116 CHAPTER 5 NUTRIENTS TRANSFORMATION FOR URBAN STORMWATER RUNOFF RETAINED IN A RADIAL CARTRIDGE FILTRATION (RCF) SYSTEM Abstract Excessive N, P loadings in runoff as a major nonpoint nutrient source impact the ecological health of sur face and ground water. This study investigated the nutrients fate in runoff from a biogenic ally loaded source area in Gainesville (GNV), FL and the effluent modified by a Radial Cartridge Filtration (RCF) system Thirteen retention periods between rainfall runoff events were monitored in 2008, with retention time ranged from 89 to 385 hours The stored runoff showed low redox potential (Median: 227 mV), high conductivity (Median: 150 S/cm ), high alkalinity (Median:150 mg/L), and high nutrients concentrati on (Median: 2.63 mg/L for DN and 0.97 mg/L for DP). An aerobic/anoxic turnover occurred at the beginning of retention periods due to the consumption of free oxygen by microorganisms. Nitrate nitrite decreased exponentially by denitrification, while ammonium increased exponentially via ammonification. Pourbaix diagrams suggest that the most stable N species for rainfall, runoff and effluent is N2, while for stored runoff NH4 + will be the most stable form. DN and DP both showed an exponential growth due to the decomposition of biogenic materials. Longterm performance shift study indicates that final events tend to have better pH buffering capacity, more rapid turnover in redox potential, quicker denitrification and mineralization rate, a lower DN concentratio n and a higher DP concentration. Therefore, careful management of nutrient loads from the source area is needed.

PAGE 117

117 Introduction A nthropogenic activities have significantly altered the nutrient cycles in the past few decades (Vitousek et al. 1997a ). For example, the use of synthetic fertilizer has greatly increased the T N and TP loadings to global biogeochemical cycles; fossil fuels combustion by automobiles has generated a significant amount of NOx in the atmosphere; the creation of impervious surfaces (roa dway, parking facility, and rooftops) has blocked the nutrient flux that was previously mitigated by biological uptake and denitrification ( Galloway et al. 1995, Vitousek et al. 1997b, Carpenter et al. 1998) H igh loadings of nutrients are delivered to re ceiving water bodies during rainfallrunoff event s, resulting in many environmental problems including eutrophication, algae bloom, and fish kill (FDEP 2009; Vitousek et al. 1997) Research found that the nutrient level in urban and highway stormwater runoff could be as high as 18 mgN/L as Total Kjeldahl Nitrogen (TKN), 2.3 mgN/L as nitratenitrite (NOx -) and 9.4 mgP/L as Total Phosphorus (TP). (Taylor et al. 2005, Davis and McCuen 2005, Wu et al. 1996 and 1998, Barrett et al. 1998, Brezonik and Stadelmann 2002). Biogenic materials (leaf litter, grass clippings), atmospheric deposition and anthropogenic inputs (fertilizers and detergents) are the three primary nonpoint sources of nutrient in urban runoff (USEPA 1999). Nutrients and sediments are the lead ing pollutants of impaired surface waters and groundwater in the United States (USEPA 1998). For the state of Florida, surface water is more of a concern in that 38% of the drinking water supply of the state is from surface water (Marella 2010). With the promulgation of the numeric nutrient criteria (NNC) for evaluating impairment of water bodies in Florida in 2012, USEPA and FDEP requires the water bodies in Florida to be characterized and protected against a more

PAGE 118

118 strict set of nutrient limits. The numeri c criteria range from 0.51 mg/L to 1.27 mg/L for TN and 0.01 mg/L to 0.05 mg/L for TP varied by nutrient watershed region (FDEP 2012). In response to this challenge, various best management practices (BMPs) have been developed. However, due to the diverse transport pathways and numerous sources of nitrogen in the environment, the performances on nitrogen removal are still not satisfied (Collins et al. 2010) As a result, in situ BMPs has been advocated for use in metropolitan areas where expensive land pri ce is the biggest constraint (Berretta and Sansalone 2012) A radial cartridge filtration (RCF) system is an insitu BMP for PM and nutrient control that incorporates adsorption, filtration and storage via a passive, radial flow cylindrical cart ridge deployed with clayey soil adsorptivefilter media Research found that engineered substrates modified with oxide of Fe or Al could gain higher dissolved P adsorptive capacity than natural substrates at higher surface loading rate (Sansalone and Ma 2011; Arias et al. 2006; Ayoub et al. 2001; Bourjelben et al. 2008; Sansalone and Ma 2009, Erickson et al. 2012). The inter event (retention periods) behavior of nutrients is quite different from the intra event behavior (rainfall runoff events). During a rainfall ru noff event, pollutants are transported by runoff flows and passed though the filtration media. All processes must occur within the infiltration time of the runoff. Such limited contact time would only allow physical processes (filtration) and some chemical reactions (adsorption and ionic exchange), but it is too short for most biogeochemical transformations. During retention periods, the retention time could last for several days to weeks, there is sufficient time for more complex chemical and biological t ransformations to occur. Aerobic nitrification

PAGE 119

119 is a two stage processes mediated by chemoautotrophic genera Nitrosomonas and Nitrobacter which oxidizes ammonium to nitrite then nitrate (Heish et al. 2007). Nitrite is relatively toxic because of its inter action with hemoglobin, however, t he oxidation of ammonia to nitrite by Nitrosomonas is usually the rate limiting step, so nitrite is rarely present in appreciable concentrations in freshwater. The process depends on the availability of NH4 + and O2. The optimum pH for nitrification is 8.4 to 8.6 (Wild et al. 1971), and the action of nitrifying bacteria at pH 7 may be reduced by as m uch as 50% (Krenkel and Novotny 1980). The process also requires the water temperature to be above 15C. The growth rate constant in creases by about 10% per degree Celsius up to about 25C (Heathwaite 1993) Therefore, nitrification shows an seasonal variation with the maximum occurred in summer when algal utilization of nitrogen is maximized due to high water temperatures. Many microbial activities including respiration, nitrification, and denitrification would consume free oxygen and consequently lower the redox potential. As the stored runoff impedes the diffusion of oxygen, after a certain period an anoxic zone is created on the bottom of RCF system that has suitable redox potential for denitrification (Brady and Weil, 2002). Nitrate trapped in the anoxic zone would transform to gaseous nitrogen (N2) via biological denitrification. An adequate C:N is critical for denitrification (Lance et al. 1976, Burford and Bremmer 1975). Also, the process is depending on the availability of denitrifiers in the environment, such as Paracoccus denitrificans and various pseudomonads ( Carlson and Ingraham 1983) Denitrification in lakes shows seasonal variations with the maximum occurred in summer (Pi a Ochoa and lvarez Cobelas 2006). In a ddition, studies found that the creation of an aerobic/anoxic zone by

PAGE 120

120 submerging a portion of media could promote microbial denitrification reactions in bio retention columns (Dietz and Clausen 2006; Hunt et al. 2002; Kim et al. 2003, Hsieh et al. 2007). Since the media cartridge in this study is also partially saturated through retention periods, similar results are expected. Dissimilatory nitrate reduction t o ammonium (DNRA) that directly reduce nitrate to ammonium is also possible, but the process requires specific bacteria and it is quantitatively less important than denitrification. Dissolved phosphorus in primarily exists in the form of phosphates (HxPO4 x 3), and the speciation is only driven by pH (Stumm and Morgan, 1981). The accumulated phosphate species absorbed by the media may be slowly released to the stored water and into the effluent with the next rainfall runoff event. Increased N concentration in retained water could also be observed due to the decomposition of biogenic materials. Export of nitrate under aerobic conditions has been observed in other studies ( Hsieh et al. 2007, Davis et al. 2006, Carleton et al., 2001). Objective This study focuses on the transformation of nutrients (especially N) for retained runoff through dry wet weather periods in a RCF system and the impacts of loaded biogenic materials to the maintenance strategies of RCF system. Understanding the fate of nutrient species from the source area could facilitate an assessment of the relative nutrient contributions from different sources. Thirteen dry wet weather cycles (retention periods) were monitored between June 03 and October 23, 2008. Four objectives were discussed in this study. The first objective is to model the temporal variation of nutrients and other water chemistry indices during retention periods. The second objective is to investigate the transformation pathways of nutrients species via

PAGE 121

121 an Eh pH diagram approac h for rainfall, inflow, outflow and retention periods. The third objective is to differentiate the initial and final retention periods for nutrients and other water chemistry indices, thereby discuss the longterm performance change of the RCF system. The fourth objective is to discuss the temperature influence on nutrient speciation for inflow, outflow and storage runoff. Methods and Materials Source Area Description The studied catchment is part of an asphalt paved parking facility located in Gainesville FL. The catchment is built with a slope of approximate 3% on the east west direction and 1.5% on the northsouth direction for runoff conveyance purpose T he pavement area occupies 76% of the total catchment area, and the rest are landscaped vegetation ( magnolia and sycamore trees ) which shed s biogenic materials directly to the pavement throughout the year Since the daily traffic load is relatively low (about 700 vehicle/day), biogenic materials are considered as the predominant nutrient sources to the c atchment Multiple catch basins were built inside the catchment to collect runoff d uring each rainfall runoff event In this study, a single catch basin on southwest of the catchment is selected for influent (runoff) sampling, as shown in Figure 5 1. The watershed area for this catch basin is approximately 500m2. RCF System Setup The radial cartridge filtration (RCF) system is a passive, radial flow cylindrical cart ridge deployed with adsorptivefilter media The setup of RCF system is shown in Figure 52 The RCF system is comprised of a nonreactive polypropylene (PP) tank (0.55 m in internal diameter, 0.91 m in h e igh t ) and a cylinder cartridge (0.47 m in

PAGE 122

122 diameter, 0.56 m in h e igh t ) inside the PP tank where t he media is held between an outer mesh and a central outlet drainage pipe with nonreactive impervious PP covers on top and bottom of the cartridge. Two gate valves, one for inlet and one for drainage are installed at 0.15 m from the bottom of PP tank The RCF system is located 9.6 m west and down to the catch basin. Runoff is conveyed by gravity to the RCF system through a set of 0.15 m (6 inch) diameter PVC pipe from the catch basin. During rainfall runoff events, i nflow pass through the media flows into the central outlet drainage pipe and exit the VCF from the effluent outlet by gravity A porous clayey media coated with Aloxide (AOCM) was filled into a radial cartridge, which provides P adsorption as well as PM separation (Liu et al. 2005, Sansalone and Ma 2009). The media that filled in the RCF system is a 1:1 (by mass) mixture of two size ranges, 0.852 mm and 24.75 mm, with a specific gravity of 2.29. Due to its negatively charged surface of the clayedsoil media, ammonium removal was also expected (Juang et al. 2001; Brady and Weil 2002). The nitrate removal efficiency during rainfall runoff event is limited given that the media could not adsorb nitrate to any significant extent. However, nitrate reduction during retention periods may occur by the biological denitrification. Sampling Protocol s The field test was conducted in 2008 with a total of 13 retention periods monitored. T he initial time ( t0) is defined at the cessation of runoff for the last treated and captured event. In order to simulate the actual condition when the RCF system is deployed underground, approximately 60 Liters of runoff was stored inside the RCF system after each rainfall runoff event S amples are taken manually from the under

PAGE 123

123 drain pipe using a 0.5 liter polypropylene (PP) bottle 1 to 3 samples were taken daily for t he first 3 days, then the sampling frequency reduces to 1 sample per day. 7 to 17 samples were taken for each retention period depending on the length of dry weather period. Samples must be taken slowly and carefully to minimize any irritation to the syste m as well as the sample itself. The stored runoff has three effects: 1. provides sufficient carbon sources and aqueous environment for microbial growth; 2. creates an anoxic zone for denitrification; 3 maintains the porous structure of filter media from b reak down. While the retained runoff could be beneficial for nitrate reduction, there are a few drawbacks: 1. t he anoxic condition would promote the mineralization, causing an elevated ammonium level; 2. decomposition, leaching, dissolution and repartitioning could happen at the sediment water interface, resulting in an elevated concentrations in storage; 3. mosquito larvae could spawn in the stagnated water during warm seasons and becomes a potential health issue. Retention Water Chemistry Indices W ater chemistry indices including pH, r edox potential conductivity, salinity, total dissolved solids (TDS) alkalinity, turbidity, NOx -N (summation of NO3 -N and NO2 -N) total ammonia nitrogen (TAN, summation of NH3 and NH4 +) dissolved nitrogen (DN, summat ion of organic and inorganic nitrogen), diss olved phosphorus ( DP, as phosphate PO4 3-) dissolved chemical oxygen demand (CODd) and Dissolved Organic Carbon (DOC) were analyzed. R edox potential pH, conductivity/salinity/TDS and temperature are monitored by an YSI multisensor sonde (model: YSI 600XLM M ) at a 15 minutes interval. The YSI multi sensor sonde is calibrated and installed at the same height of drainage pipe, which is 0.15m from the bottom of RCF system A lkalinity is measured by titration method (APHA 1998). Turbidity is measured by a Hach 2100 Turbidimeter

PAGE 124

124 (APHA 1998). DP NOx -N TAN, DN and CODd are determined colorimetrically following standard methods for water and wastewater (APHA 1998) via designated Hach test kits. A bsorbance is measured by a Hach DR 5000 spectrophotometer. Dissolved Organic Carbon (DOC) is measured through an UV Persulfate TOC analyzer (DOHRMANN Phoenix 8000) (Standard Method 5310C/EPA 415.1 and 9060A/ASTM D4779 and D4839) Detailed procedures are provided in Dean et al. (2005) and Kim and Sansalone (2008). Nutrient S peciation The Mineral thermodynamic equilibrium model Visual MINTE Q v3.0, is used to predict phosphorous speciation in this study. The MINTEQ utilizes temperature, pH, redox potential alkalinity, aqueous concentrations of cations, anions, ligands dissolved organic matter (DOM) and PM along with a charge balance to determine the concentration of each aqueous species formed under specified conditions (Dean et al. 2005) The model is based on an iterative method to solve reaction equations and mass action expressions using log equilibrium constants to find a numerical solution to equilibrium concentrations (Allison et al. 1991). Data Elaboration Water chemistry indices and constituent concentrations can vary c onsiderably between storm events, so a large number of analyses are frequently required to produce a representative estimate of concentrations. Event mean concentration ( EMC) is often used as a single index that represent s a flow averaged concentration of a storm event (Huber 1993, Sansalone and Buchburger 1997). The formula of EMC is shown as follow.

PAGE 125

125 rrttdttqdttqtcEMC00 (5 1) In this expression, c(t) is the time variable nutrients species concentration [M/L3] ; q(t) is the time variable flow [L3/T] ; tr is the duration of the event [T], and t is time [T] Another way to obtain a siteaveraged index is to perform a probability density function, where the number of observations at a given range (called "bins") is counted and demonstrated as a frequency distribution The pdf for constituents in runoff usually follows a log normal distribution (Huber 1993), therefore, median is often used to represents the average of data set. Limited E xponential Growth/Decay Model Exponential growth models work when the population can be expected to increase without limitation. In reality, populations of an index will be limited by space and resources. As time elapses, the population growth rate will slow until the size of the population reaches equilibrium maximum population size that is sustainable in a given environment. Limited growth models apply when population growth can be described in relation to the portion of the environment that remains to be filled. The differential equation of a limited growth model is shown as follow ) ( y N k dt dy (5 2) In this expression, N is the carrying capacity, which is the maximum value of an index (e.g. alkalinity); and k is the growth constant which is the rate of growth if not limited by outside factors.

PAGE 126

126 By solving this differential equation using the initial condition y0, the integration form of a limited growth model is shown as follow. kteyNNy)(0 (5 3 ) When N is smaller than y0, Eq. (5 3 ) can also be used to express an exponential decay model, where N is the minimum value of an index (e.g pH), and k becomes the decay constant. The reciprocal of k is commonly expressed as the mean lifetime ( ), which is the time at which the population is reduced to 1/e ( 0.36) times its initial value. Another more intuitive characteristic of exponential decay is the half life ( t1/2), which is the time required for the decaying quantity to fall to one half of its initial value. The relationship between half life and mean lifetime and k is as follow. k t 2 ln 2 ln2 / 1 (5 4) Both limited growth and decay models have been widely applied in modeling microbial growth and decay, life science and Newton's law of cooling. P ourbaix (Eh pH) D iagram The effects of pH and redox potential on the form in which an element exists in aqueous phase can be summarized with Pourbaix (Eh pH) diagrams Predominant ion boundaries are calculated by Nernst Equation based on specific half reactions. For a half reaction writte n as a reduction reaction in the following format a A + b B + n e= cD + d D (5 5) the Nernst Equation is shown as follow (Langmuir 1997) : b a d cB A D C nF RT E Eh ] [ ] [ ] [ ] [ ln0 (5 6)

PAGE 127

127 In this expression, R is the universal gas constant ( 1mol1); T is absolute temperature (K); n is the number of electrons (e-); F is the Faraday constant (9 6 48 Kcal/mol); [A] to [D] are the concentration of chemical species involved in a chemical reaction, and a to d are their stoichiometric number s, respectively; E0 is the standard potential of the half reaction in volts; Eh is the Redox potential in volts with respect to the standard hydrogen electrode (SHE). In occasions, pE which is the negative common logarithm of the Eh is used instead of Eh (Drever 1988). The correlation between pE and Eh is shown as follow. Eh RT Eh nF pE 0169 0 303 2 ] [ at 25 C (5 7) In Pourbaix diagram, a horizontal line stands for the predominant species is solely depends on redox potential, while a vertical line means pH is the only factor determining the predominant species. For reactions involving H+ and electron transfer, the boundary line is a diagonal line in the Pourbaix diagram, indicating the equilibrium composition is determined by both pH and Eh. Any point on t he diagram will give the thermodynamically most stable (theoretically the most abundant) form of the element under that E h and pH conditions The diagram gives a visual representation of the oxidizing and reducing abilities of the major stable compounds of an element. Pourbaix Diagrams present the relation of redox potential, pH, and the projected element speciation. H owever Pourbaix diagrams are generated for ideal conditions which may be appropriate for drinking water but quite limited for highly complex matrices such as groundwater and runoff. Therefore, assumptions based on ideal conditions should always be confirmed by analytical evidence.

PAGE 128

128 Result s and Discussion Water chemistry indices variations during retention periods Event based water chemistry indices of the 13 retention periods are tabulated in Table 51, shown as median values (Huber 1993) and ranges. Monitored hydraulic retention time (HRT) ranged from 89 to 385 hours, with a median of 195 hours (8.1 days). The temporal variation of water chemistry indices for the entire monitoring campaign is plotted in Figure 53. Each retention periods were synchronized to an initial time t0, which is defined at the cessation of runoff for the last treated and captured event. Every four time consecutive dat a points were aggregated into one point with error bar representing their mean and standard deviation. Each index w as then fit to a 3 parameters exponential increase or decrease model, as shown in Figure 55, with the modeling parameters reported in Table 5 3. During retention periods, pH follows an exponential decay from 6.82 to 6.33, with a half life (t1/2) of 43.3 hours. The decrease in pH is caused by microbial respiration that releases CO2. The redox potential at time 0 is about 185 mV, once all fr ee oxygen is consumed, redox decreases abruptly to about 224 mV by nitrate and sulfate reduction. Redox potential remains at the same level throughout the rest of retention period, indicating anaerobic reactions such as fermentation would not be likely to occur in the RCF system. The decrease in redox potential fits to an expo nential decay model, and the conversion from aerobic to anoxic conditions would only take less than 12 hours. itioning between PM and aqueous phases (including leaching from biogenic materials, desorption from the media, and dissolution of PM) and microbial activities (mineralization and

PAGE 129

129 decomposition) during retention periods. Alkalinity also shows an exponentia l growth from 26 to 356 mg/L as CaCO3, which is caused by the production of CO2 from microbial respirations. Water temperatures range from 10 to 30 C, with the highest temperature observed in August and the lowest in October. Higher temperature would prom ote bacteria growth and respiration, resulting in an elevated CO2 level and thereby increases alkalinity and reduces pH, which is consistent to the results in Figure 55. Turbidity is fitted to a probability density function (pdf). Result suggests that tur bidity follow a log normal distribution (r2 = 0.94), ranging from 1.6 to 62.5 NTU with a median of 14.9 NTU. No significant trend was observed because turbidity is affected by multiple factors such as precipitation and coagulation of suspended PM and bacteria growth. Dissolved organic carbon (DOC) was also shown as a probability density function, which is fitted to a lognormal distribution (r2 = 0.90), ranging from 1.9 to 142.0 mg/L with a median of 17.1 mg/L. In stored runoff the site median C : N molar ratio is 7.6, which is slight higher than the Redfield ratio (C:N = 106:16 6.8), indicating that there is sufficient carbon source in stored runoff. Therefore, the biological denitrification would be promoted ( Lance et al. 1976, Burford and Bremmer 1975) Nutrients V ariation during R etention P eriods Event based concentrations of dissolved nutrient species (NOx -, TAN, DON, DN, and DP) and chemical oxygen demand (CODd) are tabulated in Table 52 with median and ranges. The temporal variations of these indices throughout the monitoring campaign are plotted in Figure 54. Each retention periods were synchronized to an initial time t0, and every four time consecutive data points were aggregated into one point with error bar representing their mean and standard deviation. A 3 parameters

PAGE 130

130 exponential growth or decrease model was then fit to each indices, as shown in Figure 5 6, with their modeling parameters reported in Table 53. During retention periods, NOx showed an exponential depletion from an initial concentration of 1.85 mg/L with a half life (t1/2) of 31 hours, which could be the result of denitrification (NO3 to N2), microbial assimilation (NO3 to organic N), or dissimilatory nitrate reduction to ammonium (DNRA) (An and Gardner 2002). Among these processes, denitrification is considered more quantitatively important (Royal S ociety 1983). TAN followed an exponential increase from 0 to 2.14 mg/L, and the mean lifetime is 125 hours, indicating that TAN increases throughout the retention period as compared to the median HRT. The increase in TAN could be a result of mineralizatio n, which is a microbial process that convert organic N into ammonium, nitrogen fixation, or dissimilatory nitrate reduction to ammonium (DNRA). Mineralization is assumed to be the primary process due to sufficient carbon sources and low redox potential of the system. Also, the decay rate of NOx is much faster than the growth rate of TAN, indicating that DNRA is unlikely to be the dominant process. As shown in Figure 54, higher TAN was observed in the August 19 and September 10 retention periods, possibly due to the higher temperature and abundant biogenic materials by leaf shedding. Dissolved Organic Nitrogen (DON) showed an exponential growth, which is largely caused by the decomposition of biogenic materials. According to Table 52, DON occupies over 50% of Dissolved Nitrogen (DN), while NOx and TAN each accounts for 25% of DN. Therefore, the DN concentration is a dynamic equilibrium of nitrification, denitrification, mineralization, and assimilation between DN species. The denitrification is not suffic ient to cope adequately with N inputs from biogenic materials,

PAGE 131

131 therefore, DN concentration are likely to increase. As shown in Figure 56, DN follows an exponential growth with the concentration increased from 1.99 to 3.09 mg/L. Dissolved Phosphorus (DP) s hows an exponential growth from 0.53 mg/L to 1.7 mg/L. Dissolved chemical oxygen demand (CODd), also increases exponentially from 22.4 to 350 mg/L, which are all caused by the decomposition of biogenic materials. The concentration during retention periods were compared to the event mean concentration (EMC) of rainfall runoff events, which were shown in Table 52 and 4. Site median NOx in stored runoff is 0.27 mg/L, which is lower than that of influent and effluent (0.71 and 0.78 mg/L, respectively). Effluent showed a slightly higher concentration in nitrate than influent could be caused by the oxidation of ammonium when the RCF were aerated during storms. The median TAN in stored runoff (0.55 mg/L) is higher than the EMC for both influent and effluent (0.07 and 0.05 mg/L, respectively). The elevated TAN in stored runoff did not increase the effluent TAN concentration, which is due to the oxidation of ammonium during storms. The DP in retained storage is 0.97 mg/L, which is slightly higher than the EMC of i nfluent and effluent, indicating that the biogenic materials retained in the RCF system must be releasing DP. These results indicates that although stored runoff showed higher concentrations than the influent, its effect on the effluent is not significant, possibly because of the relative small volume of the retained runoff as compared to the volume of a storm event. Nutrient P athways and S pecies T ransformations The Pourbaix diagram of N and P are shown in Figure 57 and 8. Each boundary line is calculated by Nernst Equation (Eq. 8) using median water chemistry conditions of runoff from the studied watershed. Five nitrogen species: NO3 -, NO2 -, N2, NH3, NH4 + and

PAGE 132

132 two phosphorus species: HPO4 2and H2PO4 are included in the Pourbaix diagram. Possible N redox reactions includes nitrification, denitrification, and fixation in the Pourbaix diagram, while a electrolysis equilibrium is involved for P. Nitrite is an intermediate product of nitrification and denitrification, however, nitrite ion is very unstable in aqueous phase and could only exist metastably, therefore it is shown as a dotted line in Pourbaix diagram. In this study, measured nitrite is significantly lower than other N species, and hence summed to nitrate, called NOx -. Event based pH and redox for ra infall, inflow, outflow and storage are marked on the Pourbaix diagrams in different symbols. Regions and arrows are added to show the area of domination and direction of transformation. As seen in Figure 57, rainfall lays in the region of N2 with a low p H (about 4) and high redox potential (about 400mV), which are typical pe and pH values for rainfall (Garrels and Christ 1965), indicating that biological denitrification that reduce nitrate to N2 would occur and the dominant N species will be N2. Runoff has a pH of 7.4 and a redox potential of 350mV, which is due to the mixing of surface pollutants and rainfall, while effluent shows a smaller pH (7.0) and similar redox potential. On Pourbaix diagram, both runoff and effluent fall in the region of N2. Water retained in the RCF system has a slight acidic pH (6.5) and low redox potential ( 224 mV), which falls in the NH4 + zone. In stored runoff where redox condition was anoxic, two major processes could be involved. First, ammonium is generated through biologi cal mineralization where organic N compounds were decomposed to ammonium. Second, nitrate is reduced to ammonium by DNRA. Based on pH of the water column, the Pourbaix diagram suggests that H2PO4 2is the predominant species in rainfall, storage, and effl uent, while HPO4 2will be the

PAGE 133

133 predominant species for influent since runoff has the highest pH, which is consistent with Berretta and Sansalone (2011). Longterm performance shift for the RCF system In order to differentiate the performance shift from t he beginning to the end of the monitoring campaign, the first and last three retention periods were grouped as "initial events" and "final events", and temporal variation of pH, redox, TAN, NOx -N, DN of each group were compared, as shown in Figure 59. Th e pH of the final events showed less variation than the initial events, which is due to the increased alkalinity provides a buffer for pH change. Redox potential in final events showed a more rapid turnover (6 hours to equilibrium) than the initial events (28 hours), indicating the population of bacteria is increased. Accordingly, TAN and NOx -N both showed a higher transformation rate in the final events. DN is lower in the final events, indicating that the creation of an aerobic/anoxic cycle could be beneficial to N removal ( Dietz and Clausen 2006) DP level is elevated in final events, indicating that prolonged use of the media could compromise the performance of P adsorption. This result indicates that routine maintenance of the filter media is necessar y for P removal. Also, the nutrient load could be significantly reduced by careful management of nutrients in the source area, such as street sweeping. C onclusion This study discussed the dissolved nitrogen and phosphorus fate on an paved surface parking f acility that is loaded with biogenic materials. A radial cartridge filtration (RCF) system is instrumented insitu for runoff treatment purposes. Approximately 60 liters of stormwater was stored in the RCF system after each rainfall runoff event. Thirteen rainfall runoff events and consequent retention periods were monitored

PAGE 134

134 between June 3 and October 23, 2008. The temporal variation of major water chemistry indices were fit to exponential growth/decay models. During retention periods, pH decreased for about half unit, while redox potential dropped abruptly from about 200 mV to 224 mV within the first 12 hours due to the consumption of oxygen by microorganisms, indicating an aerobic/anoxic zone was created at the beginning of retention periods. Conductiv ity increases in an exponential manner due to decomposition of biogenic materials and other microbial activities. Alkalinity also increases exponentially, caused by the production of CO2 from microbial respirations. Lower pH and higher alkalinity was obser ved in warm seasons, which can be related to the promoted bacteria growth due to higher temperatures and abundant carbon inputs Accordingly, TAN followed an exponential growth model while NOx followed an exponential decay model. The reduction rate of NOx is different from the accumulation rate of TAN, indicating multiple pathways could be possible. NOx is reduced by denitrification and the final product is N2, while TAN is from the biogenic materials via ammonification. DON follows an exponential growt h model, which can be related to the decomposition of biogenic materials. DN, which is the summation of NOx -, TAN, and DON, follows an exponential growth model. DP also follows an exponential growth model which is caused by the decomposition of biogenic materials. Different from the more rapid physical chemical processes during rainfall runoff events, retained stormwater has sufficient reaction time on the slower biogeochemical processes. Pourbaix diagram suggest that the dominant N species in rainfall, runoff and effluent be N2 via biological denitrification, while in storage, the dominant species is NH4 + due to the anoxic redox condition. The Pourbaix diagram also suggests that

PAGE 135

135 H2PO4 is the dominant species in rainfall, effluent, and storage, while HPO4 2is the dominant species in runoff (inflow) due to its higher pH. Results from longterm performance shift study indicates that final events tend to show a better pH buffer, a more rapid turnover in redox potential, a quicker denitrification and mineraliz ation rate, and a lower DN concentration and a higher DP concentration. Results found that high loadings of biogenic materials would elevate nutrients concentration in the stored runoff in multiple aspects. Therefore, careful management of nutrient in the source area such as street sweeping could significantly reduce nutrient load to the receiving water.

PAGE 136

136 Table 5 1 Summary of water chemistry indices for 13 retention periods, shown as median and range. Start date of retention period (2008) HRT pH Redox Temp. Cond. Alk. Turbidity (hr) (s.u.) (mV) (C) [mg/L] (NTU) 03 June 114 6.9 228 27 381 132 14.9 6.7 7.1 ( 232) (+237) 21 29 228 475 64 185 3.3 60.1 10 June 217 6.5 221 26 347 154 12.7 6.3 7.3 ( 227) (+175) 24 28 117 607 10 287 8.5 61.2 21 June 366 6.4 215 26 336 154 29.9 6.3 6.8 ( 230) (+155) 24 27 138 570 26 298 8.1 49.9 8 July 138 6.6 228 27 215 96 12.6 6.4 7.1 ( 243) (+288) 25 28 106 292 70 132 5.5 15.9 15 July 142 6.5 226 27 212 102 11.3 6.3 7.2 ( 248) (+270) 26 29 91 318 24 156 8.4 13.5 29 July 138 6.3 214 26 352 140 16.8 6.2 7.0 ( 235) (+81) 26 30 151 487 52 312 10.1 19.7 8 August 89 6.4 230 28 291 144 19.1 6.4 6.6 ( 232) (+106) 2 6 30 187 363 62 204 13.8 29.4 12 August 143 6.3 -220 26 471 150 7.0 6.1 7.2 ( 262) (+245) 25 27 106 741 40 326 5.8 21.9 19 August 385 6.3 225 26 493 235 8.5 6.3 7.0 ( 257) (+258) 25 28 117 532 60 356 5.6 12.9 10 Se ptember 213 6.3 227 27 487 196 9.8 6.3 7.0 ( 253) (+258) 26 28 256 526 69 326 1.6 62.5 20 September 192 6.6 237 25 426 182 17.1 6.5-6.8 (-244) (-214) 21 27 316 471 78 260 9.8 25.4 8 October 265 6.7 241 24 384 160 15.3 6.6 7.2 ( 255) (+213) 22 28 120 464 30 260 6.2 18.2 23 October 195 6.8 240 18 383 149 17.1 6.7 7.0 ( 252) (+199) 10 23 206 444 50 210 9.9 26.7 50 192 6.5 227 26 381 150 14.9 200 6.5 227 26 367 153 14.8 92 0.2 9 2 91 37 5.8 Note: HRT = Hydraulic Retention Time Temp = Temperature Cond.= Conductivity Alk. = Alkalinity

PAGE 137

137 Table 5 2 Summary of water chemistry indices, shown as median and range. 1 Start date of retention period (2008) DP TAN NOx -N DON DN CODd DOC [mg/L] [mg/L] [mg/L] [mg/L] [mg/L] [mg/L] [mg/L] 03 Jun e 0.36 0.19 0.86 1.97 3.14 148 21.3 0.13 0.58 0.01 1.17 0.23 1.45 1.36 3.24 2.51 4.64 133 168 16.8 39.9 10 Jun e 0.59 0.36 0.45 2 .24 4.98 135 19.5 0.39 0.70 0.00 0.67 0.00 1 .34 0.56 3 .34 2.63 5.90 50 275 6.3 65.5 21 June 0.52 0.38 0.25 2.00 2.82 90 13.0 0.17 1.78 0.00 2.47 0.05 0.55 0.71 2.57 1.22 3.53 33 138 4.1 32.8 8 Jul y 0.60 0.77 1.50 0.38 2.69 8 1 1.1 0.20 0.91 0.38 1.49 0.77 2.25 0.06 0.94 2.50 3.21 3 45 1 0.3 2 0.7 15 Jul y 0.92 0.33 0.68 1.61 2.56 43 16.6 0.36 1.25 0.02 0.48 0.00 1.77 0.14 2.72 1.92 3.21 8 78 14.0 39 .9 29 July 0.62 0.55 0.14 2.99 3.67 230 12.8 0.04 1.82 0.04 1.62 0.00 2.14 0.02 3.86 2.19 5.26 55 505 7.8 49.8 8 Aug ust 1.66 1.11 0.27 0.45 1.98 133 20.2 0.50 2.05 0.02 1.72 0.00 1.00 0.18 0.75 1.32 2.34 113 140 11.0 26.1 12 Aug ust 3.53 0.58 0.36 0.70 2.49 64 52.1 1.74 5.52 0.00 2.61 0.00 2.14 0.38 1.43 2.04 3.59 18 163 5.9 142.0 19 Aug ust 2.70 0.88 0.05 1.19 3.00 159 16.0 0.56 5.31 0.01 1 .96 0.00 2.09 0.20 2.61 1.82 5. 82 5 205 1.9 47.7 10 Sept ember 2.43 1.29 0.00 1.27 2.63 160 23.1 0.57 3.89 0.18 1 .99 0.00 2.00 0.10 1.81 1.62 6.63 40 303 9.0 46.2 20 Sept ember 1.57 0.99 0.00 1.45 2.55 203 24.6 0.79 2.01 0.25 2.08 0.00 2.18 0.27 1.62 1.73 3.71 23 313 11.4 31.4 8 Oc t ober 1.44 0.26 0.61 0.99 2.17 63 17.1 0.58 1.94 0.03 1.84 0.05 1.84 0.07 1.77 1.53 2.86 48 78 9.8 33.2 23 Oct ober 0.97 0.31 0.00 1.60 1.89 213 12.8 0.26 1.77 0.03 0.74 0.00 1.00 0.55 1.86 1.58 2.45 15 368 6.4 24.1 50 0.97 0.55 0.27 1.45 2.63 135 17 .1 1.38 0.61 0.40 1.53 2.81 127 20.0 0.98 0.36 0.43 0.87 0.81 69 10.5 2 3

PAGE 138

138 Table 5 3 Model parameters of major water chemistry indices, as shown in Figure 5 5 4 and 6. 5 Para meters pH (s.u.) Redox (mV) Cond. ** Alk. [mg/L] DON [mg/L] NO x N [mg /L] TAN [mg/L] DN [mg/L] DP [mg/L] COD d [mg/L] y 0 6.820 185.0 168.9 44.8 0.070 1.847 0.000 1.988 0.528 22.4 N 6.322 224.8 477.0 275.2 1.767 0.000 2.141 3.086 1.700 350.1 k (hr 1 ) 0.016 0.645 0.020 0.011 0.032 0.032 0.008 0.015 0.053 0.007 r 2 0.884 0.9 96 0.962 0.933 0.843 0.988 0.944 0.663 0.769 0.812 62.5 1.6 50.0 90.9 31.3 31.3 125.0 66.7 18.9 142.9 t 1/2 (hr) 43.3 1.1 34.7 63.0 21.7 21.7 86.6 46.2 13.1 99.0 *Cond. =Conductivity; **Alk. = Alkalinity 6

PAGE 139

139 Table 5 4 Summary of influent and effluent EMC concentrations of 2008 rainfall runoff events Rainfallrunoff Events ( 2008) DP [mg/L] NO x N [mg/L] TAN [mg/L] DON [mg/L] DN [mg/L] COD d [mg/L] EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e 3 June 1.42 0.07 2.13 0.71 0.21 0.03 1.78 3.44 4.11 4.18 229.5 182.3 10 June 0.76 0.23 0.67 0.8 0.23 0.15 2.71 3.68 3.61 4.63 77.5 32.2 21 June 0.24 0.13 1.12 0.8 0.13 0.16 2.73 2.54 3.98 3.50 42.7 51.5 8 July 0.21 0.15 2.31 0.63 0.02 0.02 1.87 0.19 4.20 0.84 34.2 20.0 15 July 0.47 0.19 0.59 0.46 0.05 0.05 0.34 0.24 0.98 0.75 18.3 10.2 2 9 July 2.56 0.51 0.54 0.78 0.06 0.04 1.94 1.76 2.53 2.58 87.8 65.7 8 August 0.53 0.11 1.01 1.08 0.11 0.04 0.81 2.93 1.93 4.05 114.0 114.9 12 August 0.93 0.33 0.29 0.73 0.02 0.01 0.60 0.38 0.91 1.12 35.2 26.0 19 August 0.43 0.3 0.69 0.71 0.04 0.11 1.22 4 .49 1.95 5.31 23.2 18.2 10 September 0.68 0.28 0.56 0.53 0.01 0.03 0.39 0.81 0.96 1.37 55.5 54.0 20 September 0.66 0.22 1.05 0.89 0.08 0.27 0.59 0.34 1.72 1.5 120.1 92.7 8 October 0.94 0.22 0.52 0.8 0.03 0.05 0.33 0.51 0.88 1.35 125.7 120.9 23 October 0.45 0.08 0.73 0.28 0.17 0.19 0.66 1.25 1.57 1.72 94.9 91.9 50 0.64 0.19 0.71 0.78 0.07 0.05 0.90 1.24 1.95 2.58 77.5 54.0 0.76 0.21 0.97 0.99 0.10 0.10 1.28 1.66 2.35 2.75 81.4 67.7 0.58 0.12 0.61 0.8 0.08 0.09 0.87 1.41 1.23 1.61 58.2 50.7

PAGE 140

140 Center Drive Museum Road Slope = 1.5% 65.3 m Slope = 3% 2 15.7 m5 m91.6 mDrainage area (~500 m ) N Treatmentsystem Figure 5 1 Plan view of the Gainesville, FL. source area. The approximate drainage area for the sampling catch basin is shown as a shaded area on the site map (not draw n to scale)

PAGE 141

141 Figure 5 2 Design of the radial cartri dge filtration (RCF) system with a rrows show ing the flow direction.

PAGE 142

142 pH(s.u.) 6.06.57.07.5 Redox (mV) -400-2000200400 Conductivity (S/cm) 0200400600800 Retention Time (hr) 020040060080010001200140016001800200022002400260028003000320034003600 Alkalinity [mg/L] 0200400 Turbidity (NTU) 020406080 Jun 03Jun 10Jun 21Jul 08Jul 15Jul 29Aug 08Aug 12Aug 19Sept 10Sept 20Oct 08Oct 23 Figure 5 3 Water chemistry indices throughout the entire monitoring program for the Gainesville (GNV), FL. source area. a. d b e c

PAGE 143

143 0 2 4 6 NOx-N [mg/L] 0 1 2 3 DN [mg/L] 0 5 10 15 DP[mg/L] 0 2 4 6 Retention Time (hr) 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 CODd [mg/L] 0 200 400 600 Jun 03 Jun 10 Jun 21 Jul 08 Jul 15 Jul 29 Aug 08 Aug 12 Aug 19 Sept 10 Sept 20 Oct 08 Oct 23 TAN [mg/L] Figure 5 4 N, P species and CODd throughout the entire monitoring program for the Gainesville (GNV), FL. source area. a. d b e c

PAGE 144

144 Retention Time (hr) 050100150200250300 pH 6.06.26.46.66.87.07.2 Retention Time (hr) 050100150200250300 Redox (mV) -400-2000200400 Retention Time (hr) 050100150200250300 Conductivity (S/cm) 0100200300400500600 Retention Time (hr) 050100150200250300 Alkalinity [mg/L] 050100150200250300350 Turbidity (NTU) 0.11101001000 pdf (%) 0.00.10.20.30.4 DOC [mg/L] 1101001000 pdf (%) 0.00.10.20.30.4 50= 12 hr = 8.7 hrr2 = 0.90r2 = 0.94 Figure 5 5 Temporal variation of pH, redox, conductivity and alkalini ty for the 13 retention periods during 2008, shown as medians and ranges. Regression curves were fitted to each plots with parameters tabulated in Table 5 3. Turbidity and DOC were plotted as probability density function since these indices does not show a obvious trend. a. b c d e f

PAGE 145

145 Retention Time (hr) 050100150200250300 NOx-N [mg/L] 0.00.51.01.52.0 Retention Time (hr) 050100150200250300 DON [mg/L] 0.00.51.01.52.02.53.0 TAN [mg/L] 0.00.51.01.52.02.5 DN [mg/L] 1.52.02.53.03.54.04.5 Retention Time (hr) 050100150200250300 DP [mg/L] 0.00.51.01.52.02.5 Retention Time (hr) 050100150200250300 CODd [mg/L] 0100200300400500 Figure 5 6 Temporal variation of dissolved N, P species and CODd for the 13 retention periods during 2008, shown as medians and ranges. Regression curves were fitted to each plots with modeling parameters tabulated in Table 5 3. a. b c d e f

PAGE 146

146 pH (s.u.) 4 5 6 7 8 9 10 pe -10 -5 0 5 10 Eh (mV) -600 -400 -200 0 200 400 600 800 Inflow Outflow Storage rainfall NH4+ NH3H2H2ONO3-O2H2ON2 NO2 -[N] =0.19 mmol/L pN2=0.77 atm Figure 5 7 Transformation of dissolved N species for rainfall runoff events based on Pourbaix ( Eh pH) diagram. Rainfall, influent (runoff), effluent, and storage are superimpos ed to demonstrate the predominant species.

PAGE 147

147 pH (s.u.) 4 5 6 7 8 9 10 Eh (mV) -600 -400 -200 0 200 400 600 800 pe -10 -5 0 5 10 Influent (68% HPO 4 2) Effluent (34% HPO 4 2) Storage (20% HPO 4 2) Rainfall (<1% HPO 4 2) H 2 H 2 OHPO4 2-O 2H 2 OH2PO4 Figure 5 8 Transformation of dissolved P species for rainfall runoff events based on Pourbaix ( Eh pH) diagram. Rainfall, influent (runoff), effluent, and storage are superimposed to demonstrate the predominant species.

PAGE 148

148 Retention time (hr) 0 50 100 150 200 250 300 pH 6.0 6.4 6.8 7.2 7.6 Initial events Final events Retention time (hr) 0 10 20 30 40 50 60 150 300 Redox (mV) -400 -200 0 200 400 Initial events Final events Retention time (hr) 0 50 100 150 200 250 300 TAN [mg/L] 0 1 2 3 4 Initial events Final events Retention time (hr) 0 50 100 150 200 250 300 DP [mg/L] 0 1 2 3 4 Initial events Final events Retention time (hr) 0 50 100 150 200 250 300 DN [mg/L] 0 2 4 6 8 10 Initial events Final events Retention time (hr) 0 50 100 150 200 250 300 NO x -N [mg/L] 0.0 0.5 1.0 1.5 2.0 Initial events Final events p-value < 0.05 p-value > 0.05 p-value < 0.05 p-value < 0.05 p-value < 0.05 p-value < 0.05 Figure 5 9 Comparison of water chemistry indices of retained storage in filter between the first three events ( initial events ) and the last three events ( final events ). a. b c d e f

PAGE 149

149 CHAPTER 6 GLOBAL CONCLUSION The urban surface is a largely impervious interface across which nitrogen (N) in runoff is transported and partitions between the dissolved and particulate matter (PM) phases. Dissolved nitrogen (DN) is often the dominant phase during runoff transport impacting acute phenomena such as bioavailability. In contrast, the transported PM bound fractions of particulate nitrogen (PN) impact chronic phenomena such as accretion of N in drainage systems and leaching thereof. An in depth understanding of N characteristics in urban runoff better assists system designers choosing the most effective method for nutrient reduction. This work investigated the partitioning, speciation, transport and fate of nitrogen at differ ent stages of stormwater best management practices (BMP), including dry deposition, rainfall, runoff, treated effluent, and retained stormwater. The studied catchment is part of an asphalt paved surface parking facility with approximately 500 m2 drainage areas located in Gainesville, Florida. The catchment is vegetated with landscaped vegetation; therefore, b iogenic materials are considered the major nutrient source. With respect to the promulgation of Florida numeric nutrient criteria (NNC) that requires the water bodies in Florida to be characterized and protected against a more strict set of nutrient limits two Best Management Practices (BMPs) were instrumented and tested for the studied source area between 2008 and 2011. The first BMP is a radial car tridge filtration (RCF) system filled with engineered media (AOCM). The removal mechanism and removal efficiency associated with particulate matter (PM) and nutrients were examined. Fifteen rainfall runoff events and consequent retention periods

PAGE 150

150 were captured by the RCF system in 2008. At the end of each event, a pproximately 60 liters of stormwater was stored in the RCF system that creates an anoxic zone with reduced pH and redox potential The second BMP is a volumetric clarifying filtration (VCF) system which is comprised of three high surface area membrane filter cartridges and a cylindrical glass fiber tank with a total volume of 2300 Liter s The field test for PM and nutrients removal performance were conducted in 20102011, where 25 rainfall runoff ev ents were captured and treated by the VCF system Overall, ra infall depth ranges from 2 to 50 mm, and the intensity ranges from 15 to168 mm/hr, which i s representati ve to the hydrology condition for Gainesville, FL. Runoff flow rates fit a lognormal distr ibution with a median value of 0.51 L/s. Result s found that rainfall ac counts for 15% of TN and 4% of TP in runoff therefore, PM deposited in the watershed is the main contributor of nutrient s in runoff Results found that PM based N is strongly correl a ted with the volatile fraction, since volatile fraction is an index of organic content; this result indicates PM based N is largely determined by its organic portion, which is decomposed from the biogenic materials on the catchment TN distribution over PS D shows a positive correlation with particle diameter for PM larger than 200 m ; while TN shows a negative correlation with particle diameter for PM smaller than 200 m which could be related to the different composition (organic content) of PM E levated TN and TP concentrations in runoff exceed the Florida Numeric Nutrient Criteria, indicating the biogenic m aterials on the watershed would generate high loadings of nutrient s in runoff Therefore, c areful management of nutrient s from the source area including street sweeping and drainage system cleaning would be necessary in order to provide PM and N source control

PAGE 151

151 Each sample was fractionated into four categories based on PM sizes : sediment Results found that dissolved N (DN) is the predominant fraction, which occupies 62% of the TN concentration, and settleable N held the lowest portion, which is 7% of TN concentration. D issolved fraction ( fd) of N is influenced by the hydrograph of the event, In general, fd is greater in higher flow rate events which could be related to the higher stream power of these events that improves the dissolution of N During a storm event the fd decreases initially as the dissolved fract ion is attenuated by the first flush, then the value begin to increase toward the end of event indicating that re partition and dissolution becomes more significant Compared to dry deposition PM PM transported by runoff h as lower N mass concentrations possibly due to the dissolution of N Also, the particle size in runoff is finer than that in dry deposition, indicating that finer particles are preferred to be transported by runoff because it requires less stream power. Therefore, the contribution of large PM s (>400 ) is less significant even though this fraction has higher N mass concentration. Suspended N in runoff showed larger variation than the other fractions because of its higher inter facial reactivity. Categorical analysis found that mass limited (first order) model is the predominant transport model for each N fraction For larger PM (sediment N), mass limited behavior tends to occur in high intensity storms, while for finer PM (suspended N), mass limited behavior was more frequently observed in low intens ity storms. Modeling parameters of N fractions indicate that finer PM fractions are generally more abundant in runoff ( larger M0) while coarser PM fractions showed more significant first -

PAGE 152

152 flush effect (smaller K1) For flow limited events, dissolved N frac tion showed highest transport rate (larger K0). Normalized cumulative mass versus volume curves showed clear firstflush effects for every N fraction N in coarser PM fractions showed stronger firstflush effects than the finer fractions as those fractions are more restricted by the stream power of firstflush. Dissolved nutrient species including nitrate nitrite N (NOx -N), total ammonia N (TAN), organic N (ON), dissolved N (DN) and dissolved phosphate ( PO4 3-) are fitted with a mass limited or flowlimit ed model. Among DN species, NOx and ON are the two predominant species which occup ied over 90% of the D N concentration, while HPO4 2and H2PO4 are the major DP species Regression r esult s found that D N species in high volume, long duration event s tend to follow a first order (mass limited) transport model, while for low volume, short duration event s, dissolved N species tend to follow a zeroorder (flow limited) transport model. Meanwhile, D P species tend to follow a first order (mass limited) transport m odel due to the limit ed sources of phosphorus Result found that t he VCF system can effectively reduce TN and TP concentrations in runoff by removing PM associated nutrients with overall treatment efficiencies of 50% and 75% for TN and TP respectively The VCF system showed significant removal efficiency (> 80%) on the sediment and settleable fractions for both N and P, while the treatment efficiency is low to none for suspended and dissolved fractions. After treatment, e ffluent TN from the VCF system could meet the discharging criteria while effluent TP still requires additional treatments The RCF system showed 70% removal efficiency on P, however, the N removal efficiency is not significant to any

PAGE 153

153 extent, and in some occasions an export of TN is observed due to the decomposition of biogenic materials during retention periods. N in effluent is dominated by the dissolved (75%) and suspended (22%) fraction since the coarser PM fractions were largely removed by the VCF system. The equilibrium partitioning coefficient ( Kd) in effluent is about a magnitude greater than the runoff Kd, which is caused by the higher N mass concentration ( cs) in effluent from the decomposition of biogenic materials. The mass transport behavior of N in effluent is altered by the VCF system where flowlimited events are predominant, because effluent N is less restricted by the mass transpor ted by runoff since the trapped biogenic materials could act as an additional N source. R unoff C : N ratio is closest to the Redfield ratio (106: 16:1) indicating that runoff would be most suitable for the growth of phytoplankton, and therefore could possibly cause eutrophic condition in receiving waters Runoff N : P ratio is lower than the Redfield ratio; therefore, N is the controlling nutrient in runoff. For the VCFtreated effluent the C: N ratio is higher than the Redfield ratio because N has been reduced, and t he N : P ratio is slightly closer to the Redfield ratio due to the better removal efficiency of P over N by the VCF system. Rainfal l has much lower C: N ratio due to the low solubility of CO2 in rainfall and much lower N : P ratio due to the lack of significant P sources in the atmosphere. After each rainfall event event, about 60 Liters of stormwater was stored in the RCF system to in vestigate the inter event transformations of N. Different from the more rapid physical chemical processes that occurred during rainfall runoff event s, retained stormwater has sufficient reaction time for the slower biogeochemical processes. During

PAGE 154

154 retention periods, redox potential dropped abruptly from about 200 mV to 224 mV within the first 12 hours due to the consumption of oxygen by microorganisms resulting in the creation of an aerobic/anoxic zone in side the RCF system pH decreased exponentially for a bout half a unit, while conductivity CODd, and alkalinity increased following an exponential growth model. These variations could be largely related to microbial respirations and decomposition of biogenic materials. Since the microbial activities are stro ngly correlated with temperature, the event based water chemistry indices showed a seasonal variation. For example, l ower pH and higher alkalinity were observed in warmer seasons due to the promoted bacteria growth. Results found that DN increased along wi th the retention time. Since DN is determined by the balance of TAN, NOx and ON this result indicates that the decomposition of biogenic materials is the prevalent process During a retention period, TAN and DON increased following exponential growth model s, while NOx decreased follow ing an exponential decay model. The depletion rate of NOx is much faster than the growth rate of TAN indicating that these species were not directly transformed, and m ultiple pathways from different sources could be involv ed for these transformations. According to the water chemistry conditions of the RCF system, t he variation of ON is largely related to the decomposition of biogenic materials while it could also be affected by the assimilation and fixation. The increase o f TAN is largely determined by the ammonification, while DNRA, nitrification, assimilation, and Anammox could also affect the TAN level The decrease of NOx is prevalently caused by the denitrification, although it could also be influenced by the rate of nitrification and assimilation. DP

PAGE 155

155 increased following an exponential growth model, which could also be related to the decomposition of biogenic materials which are a major phosphorous source. Pourbaix diagram s could provide indications on the fate of nutrients under specific pH and redox conditions. Based on the water chemistry conditions, Result suggest s that the dominant N species in rainfall, runoff and effluent be N2, i.e., biological denitrification would be prevalent while in retained water the dominant N species is NH4 + since the redox condition is anoxic which is consistent with the result found in this study DP speciation is only influenced by the pH, therefore, H2PO4 is the dominant species in rainfall, effluent, and retained water while H PO4 2is the dominant species in runoff due to the higher pH. Results found that final events tend to show a better pH buffer, a more rapid turnover in redox potential, a quicker denitrifi cation and mineralization rate, a lower DN and a higher DP concentr ation. This result indicates the aerobic/anoxic cycle could promote N removal as the microorganisms has grown and evolved over the extended retention periods.

PAGE 156

156 LIST OF REFERENCES Allison, J. D., Brown D. S., and NovoGradac, K. J. ( 1991) MINTEQA2/ PRODEFA2, a geochemical assessment model for environmental sys tems: Version 3.0 user s manual. EPA/600/3 91/021, Environmental Protection Agency, Washington, D.C. American Public Health Association (APHA). (1992). "Standard Methods for the Examination of Water and Wastewater, 18th Ed. A. Greenburg, L. Clesceri, and L. Eaton, eds., American Water Works Association and Water Environment Federation, Washington, D. C. American Society for Testing and Materials ( ASTM). (1997).Standard Test Methods for Determining Sediment Concentration in Water Samples: Wet Sieving Filtration. D 397797C, ASTM International, West Conshohocken, PA. American Society for Testing and Materials ( ASTM). (1998a).Standard Test Methods for Determining Sediment Concentration in Water Sampl es:Wetsievingfiltration. D 397797C,ASTM International, West Conshohocken, PA. American Society for Testing and Materials ( ASTM). (2011).Standard Test Method for Total Carbon and Organic Carbon in Water by Ultraviolet, or Persulfate Oxidation, or Both, and Infrared Detection. D483903, ASTM International, West Conshohocken, PA. An, S., & Gardner, W. (2002). "Dissimilatory nitrate reduction to ammonium (DNRA) as a nitrogen link, versus denitrification as a sink in a shallow estuary (Laguna Madre/Baffin Bay Texas). Mar. Ecol. Prog. Ser., 237, 4150. Appel, P. L., and Hudak, P. F., (2001). Automated Sampling of Stormwater Runoff in an Urban Watershed, NorthCentral Texas. J. Environ. Sci. Health A36(6), 897907. Arias, M., Da Silva Carballal, J., Garcia Rio, L., Mejuto, J., and Nunez, A. ( 2006) "Retention of phosphorus by iron and aluminum oxide coated quartz particles. J. Colloid Interface Sci. 295, 6570. Arias, M., Da Silva Carballal, J., Garcia Rio, L., Mejuto, J., Nunez, A. (2006). Retention of phosphorus by iron and aluminum oxide coated quartz particles. J. Colloid Interface Sci. 295, 6570. ASTM. (1998b). Standard Test Method for Particle Size Analysis of Soils, D 422 63, ASTM International, West Conshohocken, PA. Ayoub, G. M., Koopman, B., Pandy, N., (2001). Coated filter media for low concentration of phosphorus removal. Water Environ. Res. 73, 478485.

PAGE 157

157 Ayoub, G.M., Koopman, B., and Pandy, N. ( 2001) "Coated filter media for low concentration of phosphorus removal. Water Environ. Res. 73, 478485. Bagnold, R. A. (1966). An approach to the sediment transport problem from general physics. Prof. Paper 422I., U.S. Geol. Surv., Reston, VA. Barrett, M. E.; Irish, L. B. Jr., Malinia, J. F. Jr., Charbeneau, R. J. (1998) "Characterization of Highway Runoff in Austin, Texas, Area." J. Environ. Eng. ASCE. 124, 131137. Bean, E. Z., Hunt, W. F., Bidelspach, D. A., (2007). "Evaluation of four permeable pavement sites in eastern north carolina for runoff reduction and water quality impacts." J. I rrig. Drain., 133, 583592. Berman, T., Bronk, D.A., ( 2003) "Dissolved organic nitrogen: a dynamic participant in aquatic ecosystems ". Aqua. Micro Ecol. 31, 279 305. Berretta, C. and Sansalone, J.J. (2011a ).Hydrologic transport and partitioning of phosphorus fractions. J. Hydro ., 403 (12), 2536. Berretta, C., and Sansalone, J. J. (2011b ). "Speciation and Transport of Phosphorus in Source Area Rainfall Runoff. Water Air Soil Pollut 222, 351365. Berretta, C., and Sansalone, J. J. (2012). "Fate of phosphorus fractions in an adsorptivefilter subject to intra and inter event runoff phenomena. J. Environ. Manage. 103, 8394. Brady, N. C., and Weil, R. R. (2002) "The Nature and Properties of Soils, 13th ed. Pearson Education Inc.: Upper Saddle River, New Jersey. Brezonik, P. L., Stadelmann, T. H. (2002) "Analysis and Predictive Models of Stormwater Runoff Volumes, Loads, and Pollutant Concentrations from Watersheds in the Twin Cities Metropolitan Area, Minnesota, USA. Water Res., 36, 17431757. Burford, J.R., Bremmer, J.M. ( 1975) "Relationship between the denitrification capabilities of soils and total, water soluble, and readily decomposable soil organic matter." Soil Bio l. and Biochem 7, 389394. Burns, D.A., Boyer, E.W., Elliott, E. M., and Kendall, C. (2009).Sources and transformationsof nitrate from streams draining varying land uses: evidence from dual isotopeanalysis. J. Environ. Qual., 38, 11491159. Carey R. O., Hochmuth, G. J., Martinez, C. J., Boyer, T. H., Dukes, M. D., Toor, G. S., Cisar, J. L., ( 2013) "Evaluating nutrient impacts in urban watersheds: Challenges and research opportunities. Environ Pollut 173, 138149.

PAGE 158

158 Carleton, J. N.; Grizzard, T. J., Godreg, A. N., Post, H. E. (2001). "Factors Affecting the Performance of Storm water Treatment Wetlands. Water Res. 35, 15521562. Carlson, C. A., and Ingraham J. L. ( 1983) "Comparison of denitrification by Pseudomonas stutzeri Pseudomonas aeruginosa, and Paracoccus denitrificans Appl. Environ. Microbiol. 45:12471253. Carpent er, S. R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N. and Smith, V.H. (1998). Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl., 8, 559568. Chow, V. T., Maidment, D. R., Mays, L. W., (1988). Applied Hydrology McGraw Hill, New York Clark, S. E.; Siu, C. Y. S.; Pitt, R.; Roenning, C. D.; Treese, D. P. (2009). Peristaltic pump autosamplers for solids measurement in stormwater runoff. Water Environ. Res., 81(2), 192200. C leveland, C. C., Li ptzin, D., ( 2007) "C:N:P stoichiometry in soil: is there a ratio for the microbial biomass?" Biogeochemistry 85, 235252. Collins, K. A., Lawrence, T. J., Stander, E. K., Jontos, R. J., Kaushal, S. S., Newcomer, T. A., Grimm, N. B., and Ekberg, M. L. C. (2010). Opportunities and challenges for managing nitrogen in urban stormwater: A review and synthesis. Ecol. Eng ., 36 (11), 15071519. Compton, J., Mallinson, D., Glenn, C. R., Filippelli, G., Follmi, K., Shields, G., Zanin, Y., ( 2000) "Variations in the global phosphorus cycle. SEPM Special Publicat ion 66, 2133. Correll, D. L. (1998). The Role of Phosphorus in the Eutrophication of Receiving Waters: A Review. J. Environ. Qual ., 27, 261266. Davidson, E. A., Savage, K. E., Bettez, N. D., Marino, R., Howarth, R. W. ( 2010) "Nitrogen in runoff from r esidential roads in a coastal area. Water, Air, Soil pollut 210, 313. Davis, A. P. Shokouhian, M. Sharma, H. and Minami, C. (2006). Water quality improvement through bioretention media: Nitrogen and phosphorus removal. Water Environ. Res. 78 (3) 284293. Davis, A. P., and McCuen, R. H. (2005) "Stormwater Management for Smart Growth ." Springer: New York. Davis, A. P., Shokouhian, M., Ni, S., ( 2001) "Loading estimates of lead, copper, cadmium, and zinc in urban runoff from specific sources. Ch emosphere 44, 9971009.

PAGE 159

159 Davis, A. P., Shokouhian, M., Sharma, H., and Minami, C. ( 2006) W ater quality improvement through bioretention media: nitrogen and phosphorus removal. Water Environ. Res 78 (3), 284293. Dean, C. M., Sansalone, J. J., Cartledge, F. K., and Pardue, J. H. (2005). Influence of hydrology on storm water metal element speciation at the upper end of the urban watershed. J. Environ. Eng. 131(4), 632642. Devore, J. L., ( 1990) "Probability and statistics for engineering and the sciences. Brooks/Cole Publishing Company, Pacific Grove, California. Dietz, M. E.; Clausen, J. C. (2006) "Saturation to Improve Pollutant Retention in a Rain Garden. Environ. Sci. Technol. 40, 13351340. Dodds, W.K., Jones, J.R., and Welch, E.B. (1998).Suggested classification of stream tropic state: distributions of temperate stream types by chlorophyll, total nitrogen, and phosphorus. Water Res ., 32, 14551462. Drapper, D., Tomlinson R., Williams, P., 2000. "Pollutant concentrations in road runoff: southeast Queensland case study. J. Environ Eng. 124, 313320. Drever, J. I. (1988). "The geochemistry of natural waters. pp. 309346. 2nd. Englewood Cliffs, NJ: Prentice Hall, Inc. Erickson, A. J., Gulliver, J. S., and Weiss, P. T. (2012). Capturing phosphates with iron enhanced sand filtration. W a t er Res. 46, 30323042. Flint, K.R., Davis, A.P., 2007. "Pollutant mass flushing characterization of highway stormwater runoff from an ultra urban area. J. Environ. Eng. 133, 616626. Florida Department of Envi ronmental Protection ( 2012). Numeric Nutrient Criteria for Florida Lakes and Streams, amendments to F.A.C. 62302 and 62303. Florida Department of Environmental Protection, Tallahassee, FL. Galloway, J.N., Schlesinger, W.H., Levy, H. II, Michaels, A, Schnoor, J.L. (1995) "Nitrogen fi xation: anthropogenic enhancement environmental response. Global Biogeochem. Cy. 9, 235252. Garofalo, G., Sansalone, J. J., (2011). Transient elution of particulate matter from hydrodynamic unit operations as a function of computational parameters and runoff hydrograph unsteadiness Chem. Eng. J. 175, 150159. Garrels, R. M., and Christ, C. L. (1965). "Minerals, solutions, and equilibria. Harper Row. Glasgow, H.B., Burkholder, J. M., Mallin, M. A., Deamer Melia, N. J ., Reed, R. E., ( 2001) "Field ecology of toxic pfiesteria complex species and a conservative

PAGE 160

160 analysis of their role in estuarine fish kills. Environ Health. Persp 109 (5), 715730. Groffman, P.M., Law, N.L., Belt, K.T., Band, L.E., and Fisher, G.T. (20 04). Nitrogen fluxes andretention in urban watershed ecosystems. Ecosystems, 7, 393403. Hach Company (2008).Total Nitrogen Test N Tube Vials, Method 10071Persulfate Digestion Method. DOC316.53.001086, Loveland, CO. Harris, G., Batley, G., Fox, D., H all, D., Jernakoff, P., and Molloy, R. (1996). Port Phillip Bay Environmental Study: Final Report. CSIRO Australia Canberra, Australia. Heathwaite, A.L., (1993). "Nitrogen cycleing in surface waters and lakes. in book: Nitrate: Processes, Patterns and Management. eds Burt, T. P., Heathwaite, A.L., Trudgill, S. T. John Wiley & Sons Ltd. West Sussex, England. pp. 99. Hoffman, E. J., Mills, G. L., Latlmer, J. S., Quinn, J. G., ( 1984) Urban runoff as a source of polycyclic aromatic hydrocarbons to coast al waters. Environ Sci. Tech. 18, 580587. Hsieh, C., Davis, A. P., and Needelman, B. A. (2007). Nitrogen Removal from Urban Stormwater Runoff Through Layered Bioretention Columns. Water Environ. Res. 79, 2404. Hsieh, C., Davis, A. P., Needelrman, B. A., ( 2007) "N itrogen removal from urban runoff through layered bioretention columns. Water Environ. Res ., 79 (12), 24042411. Huber, W. C. ( 1993) Contaminant transport in surface water. H andbook of Hydrology Maidment, D. R., eds, McGraw Hill, Inc., N ew York City, USA Huber, W.C., Heaney, J.P., Strecker, E.W., Weinstein, N. (2008). Chapter 6: influence of hydrologic characteristics. Evaluation of Best Management Practices for Highway Runoff Control. Report 565, National Cooperative Highway Research Pr ogram (NCHRP), Washington, D.C., 4143. Hunt, W. F.; Jarrett, A. R.; Smith, J. T. (2002). "Optimizing Bio Retention Design to Improve Denitrific ation in Commercial Site Runoff." American Society of Agricultural and Biological Engineers: St. Joseph, Michigan. Hvitved Jacobsen, T., and Yousef, Y. A. (1991). Highway runoff quality, environmental impacts and control. R. S. Hamilton and R. N. Harrison, eds., Highway Pollution, Studies in Environmental Science, 44, Elsevier Science Publishers, Amsterdam, 165208. Hvitved Jacobsen, T., Johansen, N. B., and Yousef, Y. A. (1994). Treatment systems for urban and highway runoff in Denmark. Sci. Total Environ., 146/147, 499 506.

PAGE 161

161 Johengen, T.H., LaRock, P.A., ( 1993) "Quantifying nutrient removal processes within a constructed wetland designed to treat urban stormwater runoff. Ecol Eng 2, 347366. Juang, T. C., Wang, M. K. Chen, H. J., and Tan, C. C. (2001). "Ammonium Fixation by Surface Soils and Clays. Soil Sci. 166, 345356. Kalainesan, J., Neufeld, R D. Qui mpo, R Yodnane, P ., 2009. "Sedimentation basin performance at highway construction sites J. Environ Manage, 90 ( 2 ), 838849. Kaushal, S. S., Groffman, P. M., Band, L. E., Elliott, E. M., Shields, C. A., Kendall, C., ( 2011) "T racking nonpoint source ni trogen pollution in humanimpacted watersheds Environ Sci. Tech 45, 82258232. Kaushal, S.S., Lewis Jr., W.M., and McCutchan Jr. (2006).Land use change and nitrogenenrichment of a Rocky Mountain watershed. Ecol. Appl., 16,299312. Kim, H.; Seagren, E. A.; Davis, A. P. (2003) Engineered Bioretention for Removal of Nitrate from Stormwater Runoff." Water Environ. Res. 75 (4), 355 367. Kim, J. Y., and Sansalone, J. J. (2008) Event based size distribution of particulate matter transported during urban rainfall runoff events. Water Res. 42 (1011), 27562768. Kim, J.Y.,and Sansalone, J. J. (2010).Representation of particulate matter COD in rainfall COD runoff from paved urban watersheds. Water Air Soil Pollut., 205, 113132. Kojima, K., Murakami, M., Yoshimizu, C., Tayasu, I., Nagata, T., and Furumai, H. ( 2011) "Evaluation of surface runoff and road dust as sources of nitrogen using nitrate isotopic composition. Chemosphere 84(11), 17161722. Krenkel, P. A. and Novotny, V. (1980) Water Quality Management Academic Press, London. Kuang, X.H., Kim J. Y., Gnecco, L., Raje, S., Garofalo, G., and Sansalone, J. J., ( 2007) "Particle separation and hydrologic control by cementitious permeable pavement Transport Res Rec 2025, 111117. Lance, J. C.; Whi sler, F. D.; Rice, R. C. (1976) "Maximizing Denitrification During Soil Filtration of Sewage Water. J. Environ. Qual., 5, 102107. Langmuir, D. (1997). Aqueous Environmental Geochemistry. pp 403. Upper Saddle River, New Jersey: PrenticeHall, Inc.. Lars en, T., Broch, K., and Andersen, M. R. (1998). First flush effects in an urban catchment area in Aalborg. Water Sci. Technol ., 37 (1), 251257.

PAGE 162

162 Line, D. E., Wu, J. S., Arnold, J. A., Jennings, G. D., Rubin, A. R., ( 1997) "Water quality of first flush ru noff from 20 industrial sites. Water Environ. Res ., 69, 305310. Line, D.E., White, N.M., 2007 "Effects of Development on Runoff and Pollutant Export. Water Environ. Res ., 79 (2), 185 190. Liu B. and Sansalone J. J., ( 2007) "Toxicity o f particulate mat ter in urban stormwater for indicator and commercial aquatic species American Society of Civil Engineers, World Water and Environmental Resour ces Congress, Tampa, Florida, 6 pages. Liu, B., Ying, G., and Sansalone, J. J. (2010). Volumetric filtration of rainfall runoff. I: event based separation of particulate matter. J. Environ. Eng., 136 (12), 13211330. Liu, D F ., Sansalone, J J., Cartledge, F. K ., ( 2005a) "Adsorption kinetics for urban rainfall runoff metals by composite oxidecoated polymeric media. J. Environ. Eng. 131 (8), 11681177. Liu, D., Sansalone, J. J., and Cartledge, F. K. (2004). Adsorption characteristics of s<1.0) for storm water treatment. II: Batch equilibria and kinetics. J. Environ. Eng. 130(4): 383390 Liu, D., Sansalone, J. J., and Cartledge, F. K. (2005b ). Comparison of Sorptive Filter Media for Treatment of Metals in Runoff. J. Environ. Eng. 131(8): 11781186 Liu, D., Teng, Z., Sansalone, J. J., Cartledge, F. K. ( 2001 ) "Su rface characteristics of sorptivefiltration stormwater media II: Higher specific gravity s> 1.0. J. Environ. Eng. 127 ( 1 0 ) 879888. Lund, W., (1990). "Speciation Analysis why and how?" Fresenius J. Anal. Chem. 337:557564. Ma, J., Ying, G., and Sansalone, J. J. (2010). Transport and Distribution of Particulate Matter Phosphorus Fractions in Rainfall Runoff from Roadway Source Areas. J. Environ. Eng., 136 (11), 11971205. Ma, J., Ying, G., and Sansalone, J. J. (2010).Transport and Distribution of Particulate Matter Phosphorus Fractions in Rainfall Runoff from Roadway Source Areas. J. Environ. Eng., 136 (11), 11971205. Makepeace, D., Smith, D., and Stanley, S., ( 1995) "Urban rainfall runoff quality: summary of contaminant data. Crit. Rev. Environ. Sci. Technol. 25 (2), 93139. Marella, R.L. ( 2010) "Water Use in Florida, 2005 and Trends 1950 2005." U.S. Geologi cal Survey Scientific Investigation Report.

PAGE 163

163 Mishra, S.K. Sansalone, J.J. G lenn, D.W., Singh, V.P., ( 2004) "PCN based metal partitioning i n urban snowmelt, rainfall/runoff, a nd river flow systems. J. Am. Water. Resour. As. 40 (5) 13151337. National Climatic Data Center (NCDC) (2008). 19982008 rainfall data series, Gainesville Regional Airport (GNV/KGNV). Asheville, NC. Natio nal Climatic Data Center (NCDC) ( 2012) "20022012 rainfall data series, Gainesville Regional Airport (GNV/KGNV). Asheville, NC. Nrnberg, G. K. (1996). Tropic state of clear and colored, soft and hardwater lakes with special consideration of nutrients, anoxia, phytoplankton and fish. Lake Reserv.Manage., 12, 432447. Pa thapati, S., Sansalone, J. J., ( 2009) "CFD Modeling of a Storm Water Hydrodynamic Separator J. Environ. Eng., 135 ( 4 ), 1 9120 2. Pa thapati, S., Sansalone, J. J., ( 2011) "M odeling volumetric clarifying filtration of particulate matter for transient rainfall runoff loadings. J. Environ. Eng., 137 (2), 119122. Pi a Ochoa, E., lvarez Cobelas, M. (2006). "Denitrification in aquatic environments: a crosssystem analysis. Biogeochemistry 81, 111130. Pitt, R., Maestre, A., and Morquecho, R. (2004). The national stormwater quality datatbase (NSQD, version 1.1) 1stAnnual Stormwater Management Research Symposium Proceedings Orlando, FL, 1351. Puckett, L. J. (1994).Nonpoint and point sources of nitrogenin major watersheds of the united states. Water Resources Investigations, Report 944001,U.S. Geol.Surv., R eston, VA. Redfield A C., ( 1958) "The biological control of chemical factors in the environment. Am Sci. 46, 205221. Reed, B. E., and Arunachalam, S. (1994). Use of granular activated carbon columns for lead removal. J. Environ. Eng., 120 (2), 416 436. Roger W. Bachmann Dana L. Bigham Mark V. Hoyer & Daniel E. Canfield Jr. (2012): "Phosphorus, nitrogen, and the designated uses of Florida lakes." Lake Reservoir Manage. 28:1, 4658 Royal Society (1983) "the Nitrogen Cycle of the United Kingdom ." The Royal Society, London. Sample, D. J., Grizzard T. J., Sansalone, J. J. Davis, A. P., Roseen, R. M., Walker J. ( 2012) "Assessing performance of manufactured treatment devices for the

PAGE 164

164 removal of phosphorus from urban stormwater. J. Environ. Manage. 113, 279 291. Sansalone, J. J. and Buchberger, S.G. (1995).An infiltration device as a best management practice for immobilizing heavy metals in urban highway runoff. Wat. Sci. Tech., 23 (1), 119125. Sansalone, J. J. and Cristina, C. M. (2004).First Flush Concepts for Suspended and Dissolved Solids in Small Impervious Watersheds. J. Environ. Eng., 130(11), 13011314. Sansalone, J. J., and Buchberger, S. G. (1997). Partitioning and First Flush of Metals in Urban Roadway Storm Water. J. Environ. Eng. 123(2), 134143. Sansalone, J. J., and Cristina, C. M. (2004). First Flush Concepts for Suspended and Dissolved Solids in Small Impervious Watersheds. J. Environ. Eng. 130(11), 13011314. Sansalone, J. J., and Kim, J. M. (2008).Transport of Particulat e Matter Fractions in Urban Source Area Pavement Surface Runoff. J. Environ. Qual ., 37, 18831893. Sansalone, J. J., and Ma, J. (2009) Parametric Evaluation of Batch Equilibria for StormWater Phosphorus Adsorption on Aluminum Oxide Media. J. Environ. Eng .,135(9):737746. Sansalone, J. J., Glenn, D. W ., ( 2007) "M etal distribution in soil receiving urban pavement runoff and snowmelt. Water Environ. Res 79 (7), 736 752. Sansalone, J. J., Kim, J. M., ( 2008a) "Transport of particulate matter fractions in urban source area pavement surface runoff. J Environ Qual 37, 18831893. Sansalone, J. J., Kim, J.M., ( 2008b) "Suspended particles destabilization in retained urban stormwater as a function of coagulant dosage and redox conditions. Water Res 42( 4 5), 909922. Sansalone, J. J., Koran, J.M., Smithson, J.A. and Buchberger, S.G. (1998).Physical Characteristics of Urban Roadway Solids Transported during Rain Events. J. Environ. Eng., 124(5), 427440. Sansalone, J. J., Liu, B., and Kim, J. Y. (2009). Volumetric Clarifying Filtration of Urban Source Area Rainfall Runoff. J. Environ. Eng., 135 (8), 609620. Sansalone, J., Ma, J. ( 2011) "Parametric analysis and breakthrough modeling of phosphorus from Al oxide filter media. J. Environ. Eng. 137 (2), 108118. Sansalone, J.J., Hird, J., Cartledge, F., Tittlebaum, M. (2005). Event Based Stormwater Quality and Quantity Loadings from Elevated Urban Infrastructure Affected by Transportation. Water Environ. Res. 77(4), 34865.

PAGE 165

165 Sartor, J. D., and Boyd, G. B (1972). Water pollution aspects of street surface contaminants. J. Water Pollut. Control Fed. 46, 45867. Sartor, J. D., and Gaboury, D. R. (1984). Street sweeping as a water pollution control measure: lessons learnt over the last ten years. Sci. Tot al Environ., 33, 171183. Schueler, T.R., Holland, H.K. (2000) The practice of watershed protection. Center for watershed Protection, Ellicott City, MD. S cudlark, J. R. R ussell, K. M G alloway, J. N C hurch, T. M Keene, W. C ( 1998) "O rganic nit rogen in precipitation at the mid Atlantic U.S. coast methods evaluation and preliminary measurements Atmos Environ ., 32 ( 10 ) 17191728. Sheng, Y., Ying, G., and Sansalone. J. J. (2008). Differentiation of transport for particulate and dissolved water chemistry load indices in rainfall runoff from urban source area watersheds. J. Hydro. 361(12), 144158. Shinya, M., Tsuruho, K., Konishi, T., and Ishikawa, M. (2003). Evaluation of factors influencing diffusion of pollutant loads in urban highway runoff. Water Sci.Technol., 47(78), 227232. Smith, V.H., Tilman, G.D., and Nekola, J.C. (1999).Eutrophication: impacts of excess nutrient inputs on freshwater, marine, and terrestrial ecosystems. Environ.Pollut., 100, 179196. Stanley, D. (1994). An Evaluation of the Pollutant Removal of a Demonstration Urban Stormwater Detention Pond. Albermarle Pamlico Estuarine Study, Report 94 07. Strecker, E. W.,(1994) "Constituents and methods for assessing BMPs. Proceedings of the Engineering Foundation Conference on Storm Water Monitoring. Crested Butte, CO. Strecker, E. W., Quigley, M. M., Urbonas, B. R., Jones, J. E., and Clary,J. K. (2001). Determining urban storm water BMP effectiveness. J. Water Resour. Plann. Manage., 127(3), 144 149. Stumm, W., and Morgan, J.J. ,(1996) "A quatic chemistry: chemical equilibria a nd rates i n n atural waters. John Wiley & Sons, Inc, New York, NY. Taylor, G.D., Fletcher, T.D., Wong, T.H.F., Breen, P.F., and Duncan, H.P. (2005) Nitrogen compostion in urban runoff implications for stormwater management. Water Res ., 39. 19821989. Theis, T. L., Iyer, R., and Ellis, S. K. (1992). Evaluating a new granular iron oxide for removing lead from drinking water. J. Am. Water Works Assoc., 84, 101105. Thomann, R., Mueller, J., ( 1987) "Principles of surface water quality modeling and control. Harper & Row, Publishers, Inc., New York, NY.

PAGE 166

166 U. S. Geological Survey, ( 1999) "The quality of our nation's waters: nutrients and pesicides. U. S. Geological Survey, Circular 1225. USGS information s ervices, Denver, CO. U.S. Environmental Protection Agency (EPA) (1998) The Quality of Our Nation's Waters: A Summary of the National Water Quality Inventory: 1998 Report to Con gress. Rep. No.EPA841/S 00001, Washington, D. C. U.S. EPA ( 1983) "Results of the nationwide urban runoff program: Volume 1Final Report. Water Planning division WH 554. Rep. No. EPA841/F96004G Washington, D.C. U S EPA ( 1993) "Urban runoff pollution prevention and control planning. Rep. No. EPA625R93004, Washington, D.C U.S. EPA ( 1996a ) "Managing Urban Runoff Rep. No. EPA841/F96004G, Washington, D.C. U.S. EPA ( 1998) "The Quality of Our Nation's Waters: A Summary of the National Water Quality Inventory: 1998 Report to Con gress." Rep. No.EPA841/S 00001, Washingto n, D.C. U.S. EPA ( 1999) "Preliminary Data Summary of Urban Storm Water Best Management Practices. U.S. Environmental Protection Agency, Washington, D.C. U.S. EPA (2010) "Water Quality Standards f or t he State o f Floridas Lakes a nd Flowing Waters." Feder al Register 75:131. 41734226. U S EPA. (1996b ). E nvironmental indicators of water quality in the united states. Rep. No. EPA841R9602, Wash., D.C. U S EPA. (2000a). Ambient water quality criteria recommendations: lakes and reservoirs in nutrient ecoregion XII. Rep. No. EPA822B00013, Wash., D.C. U S EPA. (2000b). Ambient water quality criteria recommendations: rivers and streams in nutrient ecoregion XII. Rep. No. EPA822B00021, Wash., D.C. U S EPA. (2010). Water quality standards for the state of Floridas lakes and flowing waters: final rule. Rep. No. EPAHQ OW 2009059640, CFR Part 131, Wash., D.C. Van Buren, M. A., Watt, W. E., and Marsalek, J., (1997). Application of the Lognormal and Normal distributions to Stormwater Quality Parameters. Wa t. Res. 31(1), 95104. Vaze, J., and Chiew, F.H.S. (2004). Nutrient Loads Associated with Different Sediment Sizes in Urban Stormwater and Surface Pollutants. J. Environ. Eng., 130(4), 391396.

PAGE 167

167 Vitousek, P.M., Mooney, H.A., Lubchenko, J., and Melillo, J. M. (1997a ). "Human domination of Earth s ecosystems. Science 277, 494499. Vitousek, P.M., Aber, J., Howarth, R.W., Likens, G.E., Matson, P.A., Schindler, D.W., Schlesinger, W.H., Tilman, G.D., ( 1997b) "Human alteration of the global nitrogen cycle: causes and consequences. Ecol. Appl. 7, 737750. Vitousek, P.M., and Howarth, R.W. (1991).Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry, 13, 87115. Vitousek, P.M., Mooney, H.A., Lubchenko, J., and Melillo, J. M. (1997).Hum an domination of Earths ecosystems. Science 277, 494499. Walsh, C.J., Leonard, A.W., Ladson, A.R., Fletcher, T.D. ( 2004) Urban stormwater and the ecology of streams. Cooperative research centre for freshwater ecology and cooperative research centre for catchment hydrology C anberra, A ustralia. Weber Jr., W.J., DiGiano, F.A., ( 1996) "P rocess dynamics in environmental systems. John W iley & Sons, Inc., New York, NY. Wild, H. E., Sawyer, C. N. and McMahon, J. C (1971) "Factors affecting nitrification ki netics. J. Wat. Pollut. Contro Fed. 43(9), 184554. Wu, J. S., Allan, C. J.; Saunders, W. L.; Evett, J. B. (1998). "Characterization and pollutant Loading Estimation for Highway Runoff. J. Environ. Eng 124, 584592. Wu, J. S., Holman, R. E., and Dorney, J. R. (1996). "Systematic Evaluation of Pollutant Removal by Urban Wet Detention Ponds. J. Environ. Eng. 122, 983988. Ying, G., and Sansalone, J. (2010). Transport and solubility of Heterodisperse dry deposition particulate matter subject to urban source area rainfall runoff processes. J.Hydrol. 383(34), 156166. Zgheib, S., Moilleron, R., Saad, M., Chebbo, G., ( 2011) "Partition of pollutant between dissolved and particulate phases: What abouot emerging substances in urban stormwater catchments? Water Res. 45, 913925.

PAGE 168

168 BIOGRAPHICAL SKETCH Hao Zhang was born in the historical city of Xi'an, China as the only child of the family. He grew up mostly in Xi'an and graduated from Shaanxi Normal University affiliated High School in 2002. He obtained a silver medal in the national chemistry o lympiad competition for High school students. He earned his B.E in material engineering from the Beijing Institute of Technology i n Beijing, China in 2006. In the same year, he also earned a B.E degree in bioengineering as a minor. Hao th en directly started his master's program in material engineering at University of Florida. In 2007, Hao entered the Ph.D. program in environmental engineering at University of Florida as a research assistant. His research primarily focused on the partitioning, transport behavior granulometric distribution and transformation of nutrients, specifically nitrogen, in urban runoff subject to acidic rainfall and impervious land surface. Applications of this research could assist the design of best management processes (BMP) and provide a solid database in terms of water management Hao participated in four projects that corroborate with Florida Department of Environmental Protection, Florida Stormwater Association, Federal Aviation Administr ation and I mbrium Corp Hao Zhang graduated and received his Ph.D from the University of Florida in the spring of 2013.