University of Florida Thesis Or Dissertation

NITROGEN FLUXES AND DYNAMICS IN AN URBAN WATERSHED

JIEXUAN LUO

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2015

To my family

ACKNOWLEDGMENTS

I would like to express my sincere gratitude to Dr. George J. Hochmuth, who is

the chair of my supervisory committee. His strong support and patient guidance

provided me courage, energy and enthusiasm to complete this dissertation. I also

would like to acknowledge the members of my supervisory committee, Dr. Mark W.

Clark, Dr. Matthew J. Cohen and Dr. John J. Sansalone for their inputs throughout my study.

Special thanks go to Dawn Lucas and Greg Means for providing me with great

field support and laboratory support. I would like to acknowledge Larry Korhnak and

Ray Thomas for giving me help in deployment of SUNAs and Bobby Hensley for guiding

me with tracer studies. Thanks also go to Hao Zhang for his valuable suggestions and

information regarding storm chasing. I also would like to thank my lab mates, Rishi

Prasad, Rajendra Gautam, Amanda Desormeaux and Ann Couch for their kindly help in

the field.

At last, I owe my deepest gratitude to my dearest parents, Mr. Huanming Luo

and Mrs. Ruiyun Jiang, and my beloved husband, Yunhan Wang. They provided more

than one hundred percent support for my study. Words cannot express how much their

love and understanding meant to me.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 8

LIST OF FIGURES ...... 10

ABSTRACT ...... 10

CHAPTER

1 LITERATURE REVIEW: MAJOR NITROGEN FLUXES FROM A GLOBAL AND REGIONAL URBAN VIEW AND URBAN NITROGEN MANAGEMENT PRACTICES ...... 14

Introduction ...... 14 Overview of Major Global N Fluxes ...... 14 Two Natural Processes to Produce Reactive Nitrogen ...... 14 Human Alterations of N Cycling ...... 16 Consequences of the Alterations to the N Cycle ...... 18 N Fluxes In Urban Ecosystems ...... 19 The Introduction of a Nutrient Mass Budget Concept ...... 19 Major N Fluxes in Urban Ecosystems ...... 21 Urban N Management Practices ...... 29 Turfgrass N Management ...... 29 Stormwater Runoff Management ...... 33 Introduction Of High Resolution In Situ Nitrate Sensor ...... 36 Sampling Techniques and Strategies-Manual Sampling Vs Autosampler Sampling ...... 37 Conventional Analysis Methods Vs In Situ Continuous N Sensors ...... 39 Conclusions ...... 41

2 NITROGEN FLUXES FROM THREE DIFFERENT SMALL URBAN CATCHMENTS ...... 44

Introduction ...... 44 Materials And Methods ...... 46 Study Area ...... 46 General Method ...... 48 Flow Measurement ...... 49 Precipitation Measurement ...... 51 Baseflow and Stormflow Sampling ...... 51 Data Analysis ...... 53 Results And Discussion ...... 54 Hydrological Factors in the Three Catchments ...... 54

Rainfall-discharge Relationship ...... 56 N Loads Estimation ...... 57 Conclusions ...... 60

3 NITROGEN BUDGETS FOR A SMALL URBAN WATERSHED AND THE NITROGEN FLUXES FOR THE DRAINAGE TO THE WATERSHED ...... 62

Introduction ...... 62 Materials and Methods...... 65 Study Area ...... 65 General Research Approach ...... 68 Flow Measurement/Estimation ...... 69 Baseflow Sampling ...... 70 Data Analysis ...... 71 Results and Discussions ...... 72 Discharge for Each Catchment ...... 72 Water Budgets for Lake Alice Watershed ...... 75 Lake Alice N Budget Estimation ...... 77 Spatial Changes in NO3-N Concentrations ...... 82 TKN Concentrations ...... 85 Conclusions ...... 86

4 URBAN STORMFLOW NITRATE-NITROGEN CONCENTRATIONS AND LOADS DETERMINED BY HIGH RESOLUTION IN SITU NITRATE SENSOR AND AUTOSAMPLERS ...... 88

Introduction ...... 88 Materials and Methods...... 91 Study Area ...... 91 Flow Measurement ...... 92 Autosampler Stormflow Sampling ...... 93 SUNA Stormflow Sampling ...... 93 Data Analysis ...... 97 Results and Discussion...... 98 Comparisons Of Autosampler/SUNA NO3-N Concentrations in Storm Events ...... 98 Event Mean Concentrations from Autosamplers and SUNAs...... 105 Event NO3-N Mass ...... 106 Conclusions ...... 108

5 CHARACTERIZATION OF NITRATE NITROGEN CONCENTRATION AND DISCHARGE RELATIONSHIP IN A SMALL URBAN CATCHMENT MONITORED BY HIGH RESOLUTION NITRATE SENSOR ...... 111

Introduction ...... 111 Materials and Methods...... 114 Study Area ...... 114

Flow Measurement ...... 115 Precipitation Measurement ...... 116 High Resolution Nutrient Analysis ...... 118 Conventional Rating Curves Between Discharge And NO3-N Concentration . 118 Characterization of Hysteresis Trajectories ...... 119 Results and Discussions ...... 121 Hydrological Factors for Storms ...... 121 NO3-N Concentration-Discharge Relationship ...... 123 Changes in NO3-N Concentrations through Individual Storms ...... 126 Conclusions ...... 136

6 SYNTHESIS ...... 138

LIST OF REFERENCES ...... 141

BIOGRAPHICAL SKETCH ...... 162

LIST OF TABLES

Table page

1-1 Fertilization rates in Florida (Cohen et al., 2007) ...... 25

1-2 Summary of stormwater N datasets included in National Stormwater Quality Database(Pitt et al., 2004) ...... 28

2-1 Pervious/impervious ratio in each Lake Alice catchment ...... 48

2-2 Completeness statistics for discharge records from each catchment through the monitoring period with annual totals for the hydrological year 2013-2014 ...... 51

2-3 Hydrological and chemical factors of the storms ...... 55

2-4 Description of N concentrations and discharge (Q) in each catchment ...... 58

3-1 Proportion of pervious/impervious surface in each stream basin ...... 68

3-2 Description of water quality monitoring sites. See Figure 3-1 for exact location of water sampling sites (blue dots on Figure 3-1)...... 71

3-3 Water budget for Lake Alice ...... 76

3-4 Summary of N budget in Lake Alice watershed ...... 78

3-5 N loads from catchments in Hume Creek Basin ...... 79

4-1 Pervious/impervious ratio in each catchment in Lake Alice watershed ...... 92

4-2 Descriptions of measurements and samples of SUNA and autosamplers for storm events ...... 97

4-3 Descriptions of storm events recorded by both SUNA and autosamplers in the two catchments. NO3-N concentrations were obtained from rain samples captured by buckets ...... 97

4-4 Comparison of NO3-N concentrations from SUNA and the autosampler in RWC and SFC ...... 99

5-1 Hydrological factors for storms in Sports Field Catchment (SFC) ...... 122

5-2 Hydrological factors for storms in Reclaimed Water Irrigated Catchment (RWC) . 123

5-3 A log-log regression between discharge (Q) and NO3-N concentration (C) and hysteresis patterns in SFC in each storm ...... 128

5-4 A log-log regression between discharge (Q) and NO3-N concentration (C) and hysteresis patterns in RWC in each storm ...... 129

LIST OF FIGURES

Figure page

1-1 Estimated changes in land use from 1700 to 1995 (Klein Goldewijk and Battjes, 1997) ...... 18

2-1 The locations of three catchments in the Lake Alice Watershed on the University of Florida Campus in Gainesville, Florida ...... 47

2-2 Flow measurement devices in sites ...... 50

2-3 The distribution of total daily precipitation in Lake Alice Watershed on the University of Florida Campus in Gainesville, Florida...... 54

2-4 Rainfall-discharge relationship in all catchments ...... 57

2-5 Discharge-NO3-N load relationship in all catchments ...... 58

2-6 Discharge-TKN load relationship in all catchments ...... 58

3-1 Major stream basins in Lake Alice watershed ...... 67

3-2 Monthly discharge from each catchment and Lake Alice well ...... 74

3-3 Boxplots representing the distribution of monthly discharge per unit ha in each catchment...... 75

3-4 TKN concentrations with standard errors in all the water quality monitoring sites ... 81

3-5 Boxplots representing the distribution of NO3-N concentration in Fraternity Stream...... 82

3-6 Boxplots representing the distribution of NO3-N concentration in Hume Creek...... 84

3-7 Boxplots representing the distribution of NO3-N concentration in Diamond Stream...... 85

4-1 Sports Field Catchment and Reclaimed Water Irrigated Catchment in Lake Alice watershed ...... 92

4-2 The look for SUNA PVC tube ...... 95

4-3 SUNA 238 and SUNA 239 in baseflow and stormflow in Sports Field Catchment (SFC) ...... 95

4-4 Grab samples vs SUNA measurements in Sports Field Catchment (SFC) and Reclaimed Water Irrigated Catchement (RWC) from February 2014 to May 2015...... 96

4-5 Changes in NO3-N monitored by SUNA and autosampler in Reclaimed Water Irrigated Catchment (RWC)...... 100

4-6 Changes in NO3-N monitored by SUNA and autosampler in Sports Field Catchment (SFC)...... 102

4-7 Event mean concentrations (NO3-N) determined by the autosampler and SUNA . 105

4-8 Event NO3-N mass determined by autosamplers and SUNA ...... 108

5-1 Sports Field Catchment (SFC) and Reclaimed Water Irrigated Catchment (RWC) in Lake Alice watershed ...... 114

5-2 High frequency monitoring data for precipitation. NO3-N concentration and discharge in Sports Field Catchment (SFC) and Reclaimed Water Irrigated Catchment (RWC)...... 117

5-3 Schematic illustration of the hysteresis index (HImid)...... 120

5-4 NO3-N concentration (C)-discharge (Q) relationships in Sports Field Catchment (SFC)...... 125

5-5 NO3-N concentration (C)-discharge (Q) relationships in Reclaimed Water Irrigated Catchment (RWC)...... 126

5-6 NO3-N hysteresis loops in individual storm events in SFC ...... 133

5-7 NO3-N hysteresis loops in individual storm events in RWC ...... 135

5-8 An example of complex storms with multiple loops...... 135

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

NITROGEN FLUXES AND DYNAMICS IN AN URBAN WATERSHED

Jiexuan Luo

August 2015

Chair: George J. Hochmuth Major: Interdisciplinary Ecology

Nitrogen fluxes have been reported in urban ecosystems in recent years. Little is known about N fluxes from areas with intense fertilization/irrigation, reclaimed water irrigation, or no irrigation, which are also common to an urban city. The main purpose of this dissertation was to identify major N sources in an urban watershed, the Lake Alice watershed on the campus of the University of Florida, and how N fluxes changed before they reached the receiving water, as well as the flow path in which the source N reached to the headwater. It can help land managers to prioritize the controlling N sources and make decisions for N management. This dissertation was comprised of five parts. First, a literature review provided background information about N cycling in global and regional (urban) scale, urban N management practices and introduction to in situ nitrate monitoring devices. Then N fluxes from three small urban catchments of different land uses were determined and compared regarding different management practices from each catchment. Sports Field Catchment (SFC) with intense fertilization/irrigation was found to produce the greatest N load (37 kg yr-1) compared to

Reclaimed Water Irrigated Catchment and Control Catchment (with no irrigation). In addition, baseflow N was dominating in the N fluxes from SFC, indicating regular

fertilization/irrigation practices were the major drivers for the great N load. Thirdly, spatial changes in N fluxes from catchments to basins in urban streams were investigated to determine the major N contributor to Lake Alice watershed, and how hydrology drove the spatial changes of NO3-N concentrations was discussed. The

results suggested that sports fields could be the largest N contributor to Lake Alice

watershed and the streams delivering flow from sports fields could produce the greatest

N load to Lake Alice. A dilution effect happened along the flow paths of urban streams

to Lake Alice, which was a sink for N although the amount of N load may be insufficient

to cause eutrophication or limit the reproduction of biomass. Fourthly, a comparison

between high resolution in situ Submersible Ultraviolet Nitrate Analyzers (SUNAs) and

autosamplers was made to determine the more effective approach in characterization

storm events. It showed that SUNAs exhibited a better performance than autosamplers

as they could capture more information during the storms than autosamplers, providing

immediate signals for NO3-N response to hydrological changes. Finally the relationship

between NO3-N concentrations and discharge was examined by SUNAs. There was a

significant relationship between NO3-N concentrations and discharge. NO3-N

concentration changes during storms were clockwise loops, suggesting immediate

dilution effect. It can also be inferred from SUNA’s curves that the increasing NO3-N concentrations after storms were attributed to NO3-N-rich groundwater rather than surface runoff.

CHAPTER 1 LITERATURE REVIEW: MAJOR NITROGEN FLUXES FROM A GLOBAL AND REGIONAL URBAN VIEW AND URBAN NITROGEN MANAGEMENT PRACTICES

Introduction

A significant change in nitrogen (N) cycling has occurred for the past few decades compared to pre-industrial times. Global N cycling has been altered by

humans through various means. This dissertation literature review starts with the review

on the major global N fluxes created by human beings, then discusses different N fluxes

nowadays in urban ecosystems, followed by urban N management practices. Finally the

review introduces the novel technology of in situ nitrate sensors and reveals the

advantage over conventional sampling methods and conventional nitrate analysis

methods.

Overview of Major Global N Fluxes

Two Natural Processes to Produce Reactive Nitrogen

Nitrogen (N) is one of the necessary elements every living organism needs for

growth and reproduction. Although inert N2 gas makes up 78% of the modern

atmosphere, reactive N is required directly or indirectly for growth and survival of all

organisms. Inert N2 gas can only be converted into reactive forms by limited means, for

example, the two major natural processes are lightning and biological N fixation (BNF).

Lightning is an important source of biologically available N in the atmosphere

through the formation of NOx-N, especially in the free troposphere (where it is the only natural source) (Galloway et al., 2004). Shepon et al. (2007) claimed that the deposition of NOx from lightning in terrestrial and oceanic ecosystems is approximately equal. In

oceans, lightning accounts for almost the entire source of NOx (Bond et al., 2002).

Reeve and Toumi (1999) indicated that as the average global temperature increases, the global lightning activity would increase and this relationship was most evident in the northern hemisphere. Schumann and Huntrieser (2007) reviewed more than 3 decades of related research and concluded that a typical thunderstorm flash produced 3.5 kg of

N mass per flash with an uncertainty factor from 0.13 to 2.7 kg, and that the estimated global annual N production from lightning was 5±3 Tg per year. They also suggested that the variability in estimates was linked with the differences in measurements and estimation methods applied, as well as the lightning’s natural variability in frequency and

energy.

On the other hand, BNF creates a significantly larger amount of reactive N

relative to lightning, but it only happens in cyanobacteria and certain higher plants that

can form symbiosis with a group of soil bacteria collectively called rhizobia. The

microorganisms can break the strong N-N triple bond and then convert it into the reactive N form. Since N fixation is an energetically expensive process (Vitousek and

Howarth, 1991), it is most important in systems where there are large amounts of energy available such as carbon source for N-fixing organisms. Those systems usually are characterized with at least seasonally high solar radiation and precipitation, therefore high net primary production (NPP). Cleveland et al. (1999) indicated that the

patterns of BNF were similar to patterns of NPP. Their data-based estimates also

showed a strong correlation between BNF and evapotranspiration. In addition,

Cleveland et al. (1999) reported that their best estimate of potential global N fixation by

natural ecosystems was approximately 195 Tg N yr-1 ranging from 100 to 290 Tg N yr-1,

which falls within the range of 44 to 220 Tg N yr-1 reported by Schlesinger (1997).

Human Alterations of N Cycling

In general, human activities have enormously altered the N cycling, even though the natural N cycling was approximately in balance during the preindustrial period with little accumulation of reactive N relative to the amount transferred to reactive N (Odum,

1971). The two influential innovations, artificial N fixation and the combustion engine, led to rapid increases in fixed N in the N cycle. The Haber-Bosch process ushered in a new era of industrially produced N-rich fertilizers for crop production. During the 20th

century, farmers gradually replaced the conventional sources of N (animal manures and

green manures) with synthetic N fertilizers. Even though the application of fertilizers in

agricultural practices doubled the agricultural food production by the end of 20th century

compared to 1960s, the magnitude of N fertilizers increased 6.87 fold (Tilman, 1999).

Meanwhile, the invention of internal combustion engines and other industrial burning

processes brought about an additional release of oxidized N (NOx) to the atmosphere

which can be deposited back to the land in the wet and dry forms. The reactive N

emitted from the fossil fuel combustion may also be transported geographically. Levy

and Moxim (1989) simulated the transportation of the reactive N in the continents and

concluded that the major sources of NOx in the North Pacific and Arctic haze were Asia

and Europe respectively. Galloway et al. (1995) concluded that human activities at

least doubled the N fixation by energy production, fertilizer production, and cultivation of

crops.

The alterations in N cycling are closely associated with the tremendous land use

changes from natural grasslands and forests to pasture, croplands, and urban area over

the past few decades (Goldewijk, 2001; Smil, 1999) (Figure 1-1, cited from Klein

Goldewijk and Battjes (1997)). Moreover, these changes have been considered to contribute to the reduced biodiversity (Dale et al., 1994; Reidsma et al., 2006), climate change (Kalnay and Cai, 2003; Searchinger et al., 2008) and soil degradation (Zhao et al., 2005). The effects of land use changes on water pollution proved particularly evident (Boesch et al., 2001; Downing et al., 1999; Qin et al., 2007). Khare et al. (2012) tested the trends between land use change (urbanization) and water quality by using the historical data from 1974-2007 in Alafia and Hillsborough River watersheds in

Florida. The results demonstrated that the majority of the water quality parameters except total N and total Kjeldahl nitrogen decreased with land use changes characterized with the loss of agricultural lands and corresponding reduction in application of inorganic fertilizers. Those results were consistent with the research conducted by using models incorporated with hydrology, chemistry and or Geographic

Information System (GIS). Tong and Chen (2002) used the Better Assessment Science

Integrating Point and Nonpoint Source (BASINS) to model the effects of land uses on total N in a local watershed in Little Miami River Basin. Their results demonstrated a significant relationship between land use type and total N, indicating total N concentration increased as residential/commercial/agricultural land use area increased and forested land use decreased. Moreover, a non linear relationship between a mixed land use watershed and N loading can be established (Nikolaidis et al., 1998).

Figure 1-1. Estimated changes in land use from 1700 to 1995 (Klein Goldewijk and Battjes, 1997)

Consequences of the Alterations to the N Cycle

Fossil fuel burning and increasing industrial N fixation have led to alterations to the N cycle by increasing the amounts of fixed N. The consequences associated with the alterations are substantial and manifold. The large quantity of residual N in soil due to fertilizer application practices can lead to a number of environmental and health problems. The microbial denitrification and nitrification in the agricultural soil are the major contributors to the emissions of N2O and NOx (Socolow, 1999). Agricultural

activities accounted for about 60% of N2O emissions, which is identified as a

greenhouse gas with a Global Warming Potential (GWP) 296 times greater than that of

CO2 (Solomon et al., 2007). Moreover, N2O is the most important ozone-depleting

substance and expected to remain the largest throughout the 21st century

(Ravishankara et al., 2009). Similarly, NOx also plays a significant role in stratospheric

ozone depletion, and it can be converted into nitric acid by reacting with the water

molecule in the rain and then do harm to plants and animals through the acid rain.

Fertilizer N that has not been taken up by plants nor released to the atmosphere

can rapidly enter the surface waterbodies and groundwater systems through surface

runoff or leaching, respectively. Excessive NO3-N in surface runoff is linked with the

development of large algal biomass blooms, leading to anoxia and even toxic or harmful

impacts on fisheries, ecosystems, and human health or recreation (Anderson et al.,

2002). Eutrophication occurs all over the world ranging from the estuary and coastal

areas (Boesch et al., 2001; Chai et al., 2006; Imai et al., 2006; Moncheva et al., 2001)

to inland lakes (Canfield Jr and Hoyer, 1988; Nagdali and Gupta, 2002; Qin et al.,

2007). Excessive NO3-N can also leach through the soil and potentially contaminate

the groundwater used for drinking water, threatening human health, reported as “blue

baby” syndrome (Kaye et al., 2006) and gastric cancer (Oenema et al., 2003).

N Fluxes In Urban Ecosystems

The Introduction of a Nutrient Mass Budget Concept

One challenge for managing N in the ecosystem is determining spatially and temporally the amounts and locations of the various forms of N. Odum (1971) stated that urban metabolism depends on external energy and matter. This dependence upon the external energy can be characterized by two approaches, one is ecological footprint and the other one is mass balance (Kaye et al., 2006). Compared with the ecological footprint approach which excludes the realistic facts such as internal interactions in the ecosystems, the mass balance approach is preferred in ecosystem studies. A mass balance is based on a closed mass budget, which was defined by Nixon et al. (1995) as one in which all of the inputs to, and all of the outputs from, a system are measured independently over the same period of time and found to balance. If the inputs and outputs do not agree, the budget can only be recognized with the independent measurements of net changes in components of interest, and the conservation of mass

provides a constraint against the ignored power, error or omission in the system.

Nutrient mass budgets are usually used to identify the sources of nutrients, widely applied in the world enabling researchers to study human impacts on the environment

(de Vries et al., 2011; Kim et al., 2008; Pathak et al., 2010). For example, the global budget for Chlorine (Cl) allowed researchers to observe how the small amounts of chlorofluorocarbons released by human activities were the source of nearly all the Cl that mixes into the stratosphere (Graedel and Keene, 1996), where it destroyed ozone

(Rowland and Molina, 1975).

The approach of a nutrient mass budget usually starts with delineated geographic boundaries, and then to identify the inputs and outputs inside the boundaries in the next step. Depending on how the boundaries are delineated, the nutrient mass budget approach for a study area can be diverse. For example, there are three types of nutrient budgets for agroecosystems: farm-gate, soil surface, and soil system budgets (Oenema et al., 2003). The accuracy and precision of the nutrient budget depend on budgeting approach, data acquisition strategy, and type of the studied ecosystems. In addition, a nutrient budget is often accompanied by a considerable amount of uncertainty, due to various possible biases and errors, particularly in the partitioning of nutrient losses (Oenema et al., 2003). The relative uncertainties are much larger in natural systems than in human-interrupted systems.

Therefore, regular uncertainty analyses are highly recommended in nutrient budget studies.

Major N Fluxes in Urban Ecosystems

N mass budget has been widely applied to studies in agricultural and forested systems (Bormann et al., 1977; Gentry et al., 2009; Jacob and Wofsy, 1990; Korsaeth and Eltun, 2000; McClaugherty et al., 1982; Meisinger and Randall, 1991; Watson and

Atkinson, 1999) where the geographical boundaries are well delineated and landscape components are monotonous, and has been introduced to urban ecosystems in the past decade (Baker et al., 2001; Groffman et al., 2004; Savanick et al., 2007). Recent watershed-level studies conducted by using N mass budget approach suggested that both natural and anthropogenic sources of N contributed to the enrichment of N in the urban ecosystems. For example, Raciti et al. (2008) indicated that the major N sources for urban lawns were lawn fertilizers and atmospheric deposition. Furthermore, they suggested that urban lawns might be sinks, rather than sources because N outputs were relatively small compared with the inputs. There are several major N sources and losses in urban ecosystems including atmospheric deposition, reclaimed water, fertilizers, biological N fixation, septic tanks, and storm water. Those N sources and losses can directly affect the water quality of urban surface and ground water

(Badruzzaman et al., 2012) and will be reviewed in detail below. The urban ecosystems referred to herein are the ecosystems that humans dominate and integrate with.

Atmospheric deposition. N from atmospheric deposition can be divided into two groups, wet deposition and dry deposition. The common forms of reactive N compounds from atmospheric deposition are nitric acid vapor (HNO3), nitrogen dioxide

- (NO2), nitric oxide (NO), ammonia (NH3), particulate nitrate (NO3 ) and particulate

+ ammonium (NH4 ). N deposition datasets have been well established nationwide. For

example, USEPA monitors the dry deposition of NOx at different locations across the

country with the Clean Air Status and Trends Network (CASTNET), and wet deposition

can be monitored in National Atmospheric Deposition Program/National Trends

Network. However, those networks avoid sampling in urban areas because they seek to

quantify major regional and national patterns rather than local influences. Therefore,

studies on local urban N depositions are also important, especially considering the

potential environmental impacts urban N depositions have in and around cities (Aber et

al., 1998; Stevens et al., 2004).

N from atmospheric deposition is considered to be one of the major contributors

to total N loads in urban areas. Fisher and Oppenheimer (1991) indicated atmospheric

nitrate deposition accounted for approximately 25% of the anthropogenic N loading to

the Chesapeake Bay, the largest US east coast estuary and the nitrate deposition arose

almost entirely from anthropogenic emissions of nitrogen oxides. The concentration and

mass of reactive N species from atmospheric deposition can have a huge variation

spatially and temporally. For example, Rao et al. (1992) reported that the atmospheric

N deposition from Pune city, an urban city in India was approximately 6 kg N ha-1 yr-1; an estimation of atmospheric N deposition in Tampa Bay, a large urban area ranged from 6 to 8.6 kg (Poor et al., 2001). (Lü and Tian, 2007) mapped N deposition rate in

China and discovered that the peak of total deposition rates of dry and wet deposition was located in central south China with 63.53 kg N ha-1 yr-1 and the overall average deposition rate was 12.89 kg N ha-1 yr-1. Another finding regarding the spatial pattern of

N deposition was that N deposition declined with the distance from urban cities (Lovett

et al., 2000). N deposition also changes with seasons. Shen et al. (2009) discovered

that the N seasonal pattern in dry deposition was consistent with anthropogenic activities. For example, fertilization and coal-fueled home heating gave rise to the highest NH3 concentrations in summer and greater NO2 concentrations in winter.

Reclaimed water. Reclaimed water, either treated from the wastewater

treatment plants or from the wetlands, has been widely used to irrigate golf courses in

urban and suburban areas due to its availability and low costs (Swancar and County,

1996). Moreover, it has become a preferred water resource for an increasing number of

universities and colleges for landscape irrigation (Hamilton et al., 2005; Ou et al., 2006).

Reclaimed water or reused water from wastewater treatment plants usually

contains significant concentrations of organic and inorganic nutrients such as N

(Badruzzaman et al., 2012; Shen et al., 2009). The reclaimed water with nutrients was

reported to benefit citrus tree growth and fruit production (Parsons et al., 2001). In

addition, owing to the nutrients in the reclaimed water, soil microorganisms have been

reported to have increased metabolic activities after the irrigation and reclaimed water

provided greater availability of a readily metabolized nutritive substrate to plants (Meli et

al., 2002; Ramirez-Fuentes et al., 2002). A survey conducted among 50 water reuse

facilities in Florida revealed that the total N concentrations from the effluent can range

from 0.13 mg L-1 to 29 mg L-1, suggesting the estimated total N released to the

environment through reclaimed water in the state of Florida can range from 1.2 x 105 to

2.6 x 107 kg yr-1 (Badruzzaman et al., 2012). However, the increasing reuse capacity in

Florida also indicates that the possible nutrient enrichment from using reclaimed water

for irrigation cannot be overlooked (Badruzzaman et al., 2012).

Fertilizers. Fertilizers have become one of the major sources of N in urban areas due to the maintenance of residential lawns and landscape and fertilizers are typically the major source of N in agricultural land use. Groffman et al. (2004) calculated a N budget for a suburban area in Baltimore Ecosystem Study Area and found that fertilizer input to residential lawn was about 55% of the total N input, the remaining N was from atmospheric deposition. In urban landscape, turfgrass is the largest irrigated crop and a major component in home lawns, parks and athletic fields, and golf courses (Blanco-Montero et al., 1995). Moreover, turfgrass area is expected to expand rapidly in conjunction with the rapid urbanization (Milesi et al., 2005).

Urban N fertilizer application rates vary greatly depending on plant species, the geographic locations, temperature, soil type, irrigation management and other factors

(Morton et al., 1988). Cohen et al. (2007) summarized the N fertilizer application rates in Florida (Table 1-1), demonstrating that N fertilizer input to residential lawns ranged from 80 to 240 kg N ha-1 yr-1, which was close to landscape plants. Also they showed

that athletic fields had a high fertilizer application rate (200 to 280 kg N ha-1 yr-1) similar

to pastures (240 to 360 kg N ha-1 yr-1), which was consistent with the recommended

fertilizer application for athletic fields (Miller and Cisar, 2001) and the optimum N level for maximum sports wear tolerance and recovery (Hoffman et al., 2010).

Table 1-1. Fertilization rates in Florida (Cohen et al., 2007)

N application rates Type of Application (kg N ha-1 yr-1) Residential lawns 80-240 Landscape plants 80-260 Athletic fields 200-280 Vegetable 180-200 Citrus 140-200 Corn, sugarcane, 150-210, 90 and 80 wheat respectively Hay 140-300 Fruit trees 140-200 Pastures 240-360

High N fertilizer application rate is usually accompanied by high N concentrations

in surface runoff or high leaching. For example, high nitrate concentrations in

stormwater runoff from residential lawns were found after fertilizer application (Kelling and Peterson, 1975); highest N concentrations (NO3-N, TKN, and NH3-N) were reported from a fertilized golf course compared to other land uses such as residential, pasture, industrial and wooded (Line et al., 2002). The nitrate from residential lawns in runoff is also associated with seasons. Line et al. (2002) found N concentrations were

four times higher in run-off from single family residential land uses during February

compared to fall or winter concentrations in suburban sites in North Carolina. Seasonal

variability has also been observed for soil water nitrate concentrations from fertilized

lawns. For example, Gold et al. (1990) claimed that measured soil-water nitrate-N

concentrations in fertilized residential lawns in Rhode Island were approximately three

to thirteen times greater in the spring (e.g. 2.6 mg L-1) compared to other seasons (e.g.

0.2 mg L-1 in the fall) in the same year. Other studies in Rhode Island and Michigan

reported higher concentrations of nitrate-N in soil water in the fall compared to spring

(Liu et al., 1997; Miltner et al., 1996). Overall annual N leaching rates for turfgrass

ranged from 0 to160 kg N ha-1, and represented up to 30% of the fertilizer applied N

(Barton and Colmer, 2006).

Biological N fixation. Biological N fixation is controlled by a variety of

autotrophic and heterotrophic bacteria. It usually happens in aquatic ecosystems such

as scenic lakes or ponds in urban areas. However, the N fixation rate is much lower

than legumes in agriculture which has been discussed above. Turner et al. (2002),

working on a N budget for Lake Pontchartrain watershed located north of New Orleans,

suggested that biological N fixation from the lake was relatively minor compared to

atmospheric deposition and urban runoff. Cyanobacteria appear responsible for most

planktonic fixation in aquatic ecosystems. Planktonic N fixation rates in lakes were

strongly related to lake trophic status, with low in oligotrophic and mesotrophic lakes

(generally <0.1 g N m-2 yr-1) but high in eutrophic lakes (0.2 to 9.2 g N m-2 yr-1)(Howarth

et al., 1988). Benthic N fixation rate may account for the remaining fixation, which was

also consistent with the trophic status of the waterbodies and reported to range from 0.4

to 1.6 g N m-2 yr-1 in estuaries rich in organic sediment (Nadelhoffer et al., 1999).

Septic tanks. In the U.S., approximately 20% of the population makes use of septic systems to treat their wastewater (EPA, 2011). Such systems usually consist of

a septic tank and a soil filtration system for effluent dispersal. Septic tank systems can

pose potential hazards to the aquatic environment, especially in rural areas where no

centralized sewage system is connected to the wastewater treatment plant. There is a

growing body of evidence demonstrating one of the major N sources to surface

waterbodies or groundwater was from septic tank systems. Albertin et al. (2012) used

isotope tracers of N to determine the N sources within karst springs in Florida, finding

that in the five springs they sampled, 10% to 20% N came from the manure/septic tanks in residential land use while 3% to 9% came from an inorganic N source such as oxidation of ammonium fertilizer and/or soil N. This result was partly in agreement with

Kaushal et al. (2006) who found 19% to 23 % of annual N in Rocky Mountain watershed tributaries was derived from septic tanks in residential land use. However, septic systems are not usually considered as a sufficiently important source of water pollution for policy-driven regulatory controls, or for adequate policing where these controls already exist (Withers et al., 2013). This might be because the systems have been long considered as an effective solution for wastewater treatment since they are usually buried underground and hidden from view. In addition, they have the advantages in energy and cost efficiency in more densely populated areas as well as low greenhouse- gas emissions (Diaz-Valbuena et al., 2011; Weiss et al., 2008).

Storm water. Tremendous changes in land use over the past few decades have substantially influenced the urban hydrology by altering the water supply and drainage.

For example, the increase in amounts of impervious surfaces changes the way in which water moves through urban ecosystems, and that results in shorter lag time between precipitation and discharge and higher flood peak discharges (Paul and Meyer, 2001).

In addition, the large amounts of stormwater runoff may also be accompanied by nutrients that can lead to water quality deterioration in surface waterbodies, groundwater, estuaries as well as oceans, which is identified as non-point source pollution.

Table 1-2. Summary of stormwater N datasets included in National Stormwater Quality Database(Pitt et al., 2004)

NH3-N NO2,3-N TKN TN Land use (mg L-1) (mg L-1) (mg L-1) (mg L-1) Overall summary (n = 3765) Median 0.44 0.6 1.4 2 Residential (n = 1042) Median 0.31 0.6 1.5 2.1 Mixed residential (n = 611) Median 0.39 0.57 1.4 2 Commercial (n = 527) Median 0.5 0.6 1.5 2.1 Mixed commercial (n = 324) Median 0.6 0.58 1.4 2 Industrial (n = 566) Median 0.42 0.69 1.4 2.1 Mixed industrial (n = 218) Median 0.58 0.59 1.1 1.7 Institutional (n = 18) Median 0.31 0.6 1.35 2 Freeways (n = 185) Median 1.07 0.28 2 2.3 Mixed freeways (n = 26) Median 0.9 2.3 3.2 Open space (n = 68) Median 0.18 0.59 0.74 1.3 Mixed open space (n = 168) Median 0.51 0.7 1.1 1.8

N loads from urban stormwater runoff are variable depending on the dominating

land use type, but are greater than those found in undisturbed natural areas (Dodd et

al., 1992; Groffman et al., 2004; Line et al., 2002). In an analysis of urban stormwater

runoff from different land use types, freeways were shown to have the highest N

concentrations as documented in National Stormwater Quality Database in Table 1-2

(Pitt et al., 2004). Flint and Davis (2007) also claimed that NO3-N concentration ranged

from 0.15 to 4.32 mg L-1 and TKN concentration ranged from 0.8 to 10 mg L-1 in the

highway from an ultra-urban area in Mount Rainier, MD.

Stormwater systems are combined with sewer systems in some of urban areas in

UK, transporting both untreated wastewater and surface runoff to surrounding waterbodies and are considered as one of the principal contributors to water impairment in UK (Balmforth, 1990). The combined sewer-storm systems were reported to deliver

-1 runoff with NO3-N concentrations up to 3.57 mg L and TKN concentrations up to 19.5

mg L-1 in storms in high density residential areas with commercial activity (Lee and

Bang, 2000). The high N concentrations in storms from sewer-storm systems were close to the agricultural runoff observed by David et al. (1997).

Urban N Management Practices

As discussed above, fertilizers to turfgrass in lawns, golf courses as well as

athletic fields and N from urban stormwater runoff are the two major N contributors to

urban ecosystems. The following discussion focused on the management practices with

regards to those concerns.

Turfgrass N Management

Fertilizer management. N fertilizer management can help to decrease N

leaching or N runoff from a turfgrass area. Ideally, N should be applied at a rate and

frequency that matches turfgrass demand, and if possible should not be applied

immediately before heavy rainfall (Miltner et al., 1996; Shuman, 2004; Smith et al.,

2007; Snyder et al., 1984; Zhao et al., 2012). The amounts of N required by turfgrasses

were reported to decline over time after turfgrass establishment. Estimates based on

historical data and simulation modeling suggested that N requirements for cool-season

turfgrass would be maintained for the first 10 years after establishment, and then continue to decline for up to 60 years (Petrovic, 1990; Qian et al., 2003). Therefore, adjusting the fertilizer regimes to match N removal rates (plus atmospheric losses) has been proposed as an approach to minimize N leaching from older turfgrass stands

(Petrovic, 1990).

Clippings from turfgrass can also help to adjust the N fertilizer amount applied to turfgrass. Turfgrass produces a large amount of clippings every year (Harivandi et al.,

2001). Clippings are often removed from lawns because clippings can be unsightly and may contribute to disease and thatch build up (Murray and Juska, 1977). However, clipping removal from turf represents a major N loss. With increasing landfill restrictions on yard waste and efforts to minimize resource input to turfgrass systems, recycling clippings through a mulching mower is becoming a common practice (Heckman et al.,

2000; Qian et al., 2003). Several studies have been conducted to determine the impacts of clippings management on turf quality and N requirements. Starr and DeRoo

(1981) claimed that 30% of the total N applied could be saved through clipping return.

Heckman et al. (2000) also demonstrated that the amount of applied N could be cut in half (from 195 to 98 kg N ha-1 yr-1) when clippings were recycled. Recently, Kopp and

Guillard (2002) reported that turf plots receiving 0 to 98 kg N ha-1 yr-1 with clippings

returned produced a comparable quality and clippings yields to plots receiving 390 kg N

ha-1 yr-1 with clippings removed. This result suggested that returning clippings could

reduce N fertilization by 75% or more without reducing turf quality. The reduction in

fertilization application rates is associated with reduction in leaching. Qian et al. (2003)

found that the predicted N leaching would be minimal (close to zero) under

management scenarios of low N fertilization (150 kg N ha-1 yr-1) with clippings returned

based on the results from CENTURY model.

The frequency of fertilizer application also has a large impact on N leaching and varies depending on the fertilizer type. For more water-soluble N fertilizers, lower rates and more frequent applications should be used to minimize N leaching. Less water- soluble fertilizers, such as slow or control-release fertilizers, can be applied at higher, less frequent rates than water-soluble fertilizers, without increasing N leaching (Shoji and Kanno, 1994; Snyder et al., 1984). Also applying fertilizers at times when the turfgrass is actively growing will minimize N leaching (Brown et al., 1977; Turner and

Hummel, 1992).

Irrigation management. In addition to the fertilizer management, N leaching from turfgrass can also be attributed to irrigation rates and frequencies that cause water

to move beyond the active rooting zone (Al-Kaisi and Yin, 2003; Gheysari et al., 2009;

Nakamura et al., 2004; Vázquez et al., 2006). In a 2-year field study, Morton et al.

(1988) demonstrated that over irrigation could increase drainage and N leaching from a

sandy loam with Kentucky bluegrass compared with those irrigated to avoid drought

stress but prevent percolation. In addition, N leaching was equivalent to 14% of the

applied N from the over-watered treatment and <3% from the scheduled irrigation

treatment, which was not different to losses from plots to which no fertilizer was added.

Minimizing soil water movement by using a soil tensiometer-controlled irrigation system

also decreased mineral-N leaching from 22% to 7.5% of applied N over 6 months

(spring and summer) as Snyder et al. (1984) demonstrated.

In some soil types with coarse soil structure, the irrigation rate and frequency may cause water and N to move unevenly through the top soil via large cracks, worm holes, root channels and water-repellent zones (e.g., preferential flow) (Bauters et al.,

1998; Brady and Weil, 1996; McLeod et al., 2001). Preferential flow causes dissolved nutrients to move quickly through the top soil, minimizing the opportunity for plant roots and soil microbes to utilize applied water and N (McLeod et al., 1998; Starrett et al.,

1995). One of the most efficient and practical approaches to reduce preferential flow in soils and to decrease N leaching is to decrease the irrigation rates per application and increase the irrigation frequency (McLeod et al., 1998).

Soil amendments. The use of soil amendments has been proposed to reduce N

leaching by improving soil water-holding capacity. Turfgrass grown on sandy soils tends

to have high leaching capacity because the sand-based root zones cannot retain

enough water and nutrients to support turfgrass vigor. Organic amendments like

sphagnum peat moss are often incorporated into sand-based turfgrass root zones to

promote water and nutrient retention without adversely affecting water infiltration and

drainage (Beard, 2002; Bigelow et al., 2001). Fly ash was also suggested as a soil

– + amendment to retard NO3 , NH4 , and P leaching in the sandy soil (Adriano et al., 1980;

Pathan et al., 2002) . However, organic materials can be prone to decomposition by

microbial activity thus reducing their overall effectiveness in the sand profile (Bigelow et

al., 2004). Other materials also have been evaluated for the performance as soil

amendments. For example, biochar was reported to increase nutrient retention when

mixed with soils and have a better attraction for cations than do other forms of organic

matter due to its relatively large surface area and large negative surface charge per unit

area of biochar (Glaser et al., 2002; Laird et al., 2010; Lehmann et al., 2003). The only

concern for soil amendments is that the growth and color of turfgrass can be affected if

the portion of soil amendments is too high (Engelsjord and Singh, 1997). Also adding

relatively large quantities of materials to soils presents practical challenges for

incorporation to reasonable depths and most amendments can be costly.

Stormwater Runoff Management

Bioretention. Bioretention facilities consist of a layer of engineered

soil/sand/organic media that can support a variety of different plant species (Hsieh et

al., 2007). The primary treatment plan of bioretention for stormwater runoff focused on the physical separation processes such as filtration, sorption and ion exchange for suspended solids removal. But a great number of studies proved that bioretention was efficient in ammonium removal due to ammonium’s cationic status in aqueous solution with sorptive interactions with the soil media (Brady and Weil, 1996; Davis et al., 2001;

Juang et al., 2001). Nitrate, on the contrary, is minimally held by bioretention media and, as an anion, is very mobile in soils and will not be adsorbed to soil media to any significant extent. Therefore, removal of nitrate was poor, at 1 to 24% in the 18 6-hour bioretention column studies (Hsieh and Davis, 2005). Still improved reengineered

concept for of bioretention for nitrate removal via microbial denitrification was introduced

and has the potential for future successful application as an urban stormwater treatment

practice (Kim et al., 2003). It is noted that nitrification processes occur during wetting- drying cycles when ammonium in water that is held in the bioretention column may be transformed to nitrate. Hsieh et al. (2007) reported a high ammonium removal efficiency during the process when water was drained through the bioretention columns. .

Bioretention is also associated with high TKN removal. Wu et al. (1996) reported a

removal efficiency of 60% for TKN in an urban retention pond in Charlotte, North

Carolina. Overall, it is suggested that utilizing a small percentage of the watershed area

for the development of wet detention ponds at strategic locations could reduce the

pollutant loadings to meet targeted or regulatory requirements of water quality

improvement (Wu et al., 1996)

Wetlands. Wetlands have been recently recognized as an ecologically

sustainable option for water pollution control such as removal of pollutants from

stormwater runoff. Wetlands are characterized as a biologically diverse ecosystem with

the capacity to provide physical, biological and chemical processes to facilitate the

removal, recycling, transformation or immobilization of sediment and nutrients. Most of

these processes are undergone by the wetland vegetation and associated

microorganisms.

The N removal rates in wetlands vary depending on the plant species, hydraulic

residence time, temperature, and dissolved oxygen content (Hammer and Knight, 1994;

Lee et al., 2009). Birch et al. (2004) reported that the average removal efficiency of

NOx, TKN and TN in stormwater runoff was 22%, 9% and 16%, respectively in a

constructed wetlands. The low N removal rate might be associated with the critical C:N

ratio in denitrification, with ratio > 5:1, resulting in 90% nitrate removal efficiencies

(Baker, 1998). Sim et al. (2008) claimed that the nutrient removal performance was

82.11% for total N, 70.73% for nitrate–N in Putrajaya Wetlands where plenty of

vegetation was planted in Malaysia, indicating that vegetation uptake had a huge impact

on N removal.

Green roofs. The increasing area of impervious surfaces such as buildings and streets in urban areas has dramatically changed the urban hydrology of water flow.

Roofs can represent up to 32% of the horizontal surface of built-up areas and play a significant role in transporting the water flow particularly in stormflow (Oberndorfer et al.,

2007). The addition of vegetation and soil to roof surfaces can reduce several negative effects of buildings on local ecosystems and can decrease buildings' energy consumption and provide increased sound insulation (Dunnett and Kingsbury, 2004), fire resistance (Köhler, 2003) and cooling potential to buildings (Del Barrio, 1998).

In addition, greenroofs have proved to reduce the urban stormwater quantity by absorbing and filtering pollutants and become one of the alternatives for stormwater runoff management. For example, Hathaway et al. (2008) investigated the hydrologic

performance of greenroofs in North Carolina, finding greenroofs retained approximately

64% of the total rainfall received and reduced average peak flow by 75%. The results

were consistent with other studies in Germany (Mentens et al., 2006) and close to the retention ranges summarized by Berndtsson (2010). However, some studies reported that greenroofs may not make any improvements in water quality such as reduction of N concentrations. For example, Hathaway et al. (2008) indicated that greenroofs increased the nutrient concentration (TN) in the outflow, which was presumably associated with initial media composition in the greenroofs’ structure. The findings were consistent with Moran et al. (2004) in North Carolina, USA, who showed that compost in the substrate layer may cause high concentrations of N in greenroof runoff (TN ranged from 0.8 to 6.9 mg L-1 with average about 3.6 mg L-1). Berndtsson et al. (2009) compared the performance of nutrient removal in an intensive vegetated greenroof

(maintainance-required with deep soil layers) and an extensive vegetated greenroof

(maintenance-free with thin soil substrate), and drew a different conclusion. They demonstrated that both greenroofs behaved as a sink for nitrate N and reduced ammonium N and total N losses. The difference in findings between Moran et al. (2004) and Berndtsson et al. (2009) may be attributed to the difference in vegetation as

Berndtsson et al. (2009) suggested vegetation uptake was very important in nutrient removal in greenroofs.

Greenroofs overall proved to be a good alternative in stormwater runoff management although the performance in improving water quality may vary depending on the greenroof types and structures.

Introduction of High Resolution In Situ Nitrate Sensor

The summary of N sources and N management above shows the various

pathways that N flows through in urban ecosystems. To understand complex nutrient

processes on watershed-scale or catchment-scale, accurately quantify nutrient exports

under different hydrology (e.g. baseflow vs stormflow), and observe changes in water

quality as a result of nutrient mitigation efforts over time, it is vital that the newly

emerging field-based automated analyzer technologies begin to be deployed, to allow

for routine high-resolution monitoring of our waterbodies in the future. By accurately

tracking and determining concentrations of pollutants both in baseflow and stormflow

from potential sources, urban land managers can make decisions about controlling the

pollutant loads. A novel technology to measure in situ nitrate concentrations is

introduced here with comparison with conventional sampling approaches and

conventional analysis methods.

Sampling Techniques and Strategies-Manual Sampling Vs Autosampler Sampling

Because of the critical nature of nutrient pollution, diverse N transport scenarios

in the environment, and difficulty in managing many nutrient sources, technologies are

needed to address nutrient inputs to the water environment from a variety of source

pathways. One important aspect for validity of water quality parameter measurements

is the sampling frequency which can reflect the variability of the systems under study,

and the sampling frequency is determined by sampling approaches.

Conventional water sampling is generally performed by two methods, grab

sampling and autosampler sampling. Grab sampling is the most common technique in

most of the fieldwork. Basically, a grab sample refers to the sample the technician

takes at a specific time and location, which means the sample only reflects the

characteristics of the water at that moment. In addition, the person who takes the

sample has to be very careful in order to get a representative sample for the real

scenario. The slight difference in time of day and location between the first time

sampling and the next time sampling may cause the difference in the final results.

Autosampler sampling is usually applied in stormwater studies which involve severe

weather, long sampling time and high sampling frequency. Autosampler sampling has a

couple advantages over grab sampling. Firstly, the autosamplers can be programmed,

therefore they can be activated at regular time intervals. Also they can be triggered by a

rise of the water stage or the onset of rain as indicated by signals from a tipping bucket

rain gage. Then they provide the confidence of sampling in the severe weather under

condition of normal functioning since they are always at the site

(http://www.fs.fed.us/waterdata/PDFfiles/FieldGuide_Turk.pdf). However, in research

requiring sampling of first flush, manual sampling is usually preferred since it has greater flexibility and allows larger sample volumes to be collected as well as special samples using different bottles (Han et al., 2006; Stenstrom and Kayhanian, 2005).

The autosampler sampling can be programmed in many ways, which determines the accuracy of the samples. The frequency with which samples are typically collected by the autosmapler can be defined by three approaches: (1) flow-weighted, (2) time- weighted, and (3) user-defined. Once it is determined how samples will be collected, the next step is to determine whether to collect discrete sample samples or a composite sample. A few studies have been conducted to compare different sampling strategies with autosamplers. It turns out that different sampling strategies are preferred based on different research purpose. For example, for the purpose of nutrient loads estimation,

Shih et al. (1994) studied the accuracy of nutrient load calculations using time- composite sampling. They found that when flow and concentrations were positively correlated, the time-composited sampling tended to underestimate the stream loading.

Swistock et al. (1997) compared six methods for calculating annual stream exports of several water quality parameters and finally suggested more intensive sampling for solutes that correlated strongly with stream flow. Coincidentally, Stone et al. (2000) drew the same conclusion after the comparison of four sampling methods and found that flow-proportional composite sampling method predicted significantly greater mass loading rates compared with other methods. Even though flow-proportional sampling method is preferred in many studies, it still has the drawbacks such as high costs for the sample analyses due to high frequency sampling, and some uncertainties regarding the

results. For the purpose of characterization of storms, discrete samples are preferred because they capture the individual samples at specific moments during the storm.

Conventional Analysis Methods Vs In Situ Continuous N Sensors

Conventional nitrate/nitrite analysis methods. The chemistry between nitrate and nitrite makes them inextricably linked because the majority of the detection strategies for nitrate relies on the detection of nitrite (Badruzzaman et al., 2012; Meli et al., 2002). Spectroscopic methods are possibly the most popular one for nitrate/nitrite analysis due to the excellent detection limits and explicit assay-type protocols. The

Griess Assay, first developed in 1879, is considered to be the most common approach that appears in most of the applications (Ramirez-Fuentes et al., 2002). This assay mainly relies on the diazotization of acidified nitrite and sulphanilamide with the subsequent coupling reaction with N-(1-naphthyl) ethylenediamine providing a highly colored azo dye which can be measured colorimetrically. Nowadays, it is equivalent to

U.S Environmental Protection Agency method 353.2 (U.S. Environmental Protection

Agency, 1993) and U.S. Geological Survey (USGS) method I-2545-90 (Fishman, 1993, p. 157) except that sulfanilamide and N-(1-naphthy) ethylenediamine reagents were separate in Griess Assay rather than combined. Nitrate does not undergo a diazotization reaction with sulfanilamide and therefore must first be reduced quantitatively to nitrite to react with those reagents. Generally the two methods used to reduce nitrate to nitrite are cadmium metal (Milesi et al., 2005) and bacterial nitrate reductase (Morton et al., 1988).

More sensitive techniques such as chemiluminescence can determine the nanomolar quantities of nitrate/nitrite with a precision of ±2 nM, however, it requires

further reduction to nitric oxide (Morton et al., 1988). Photo-induction reduction is novel in nitrate-nitrite conversion. Takeda and Fujiwara (1993) reported the formation of nitrite and oxygen from nitrate as shown below by ultraviolet (UV) light irradiation in the wavelength between 200 – 300 nm. This approach avoids the chemical reduction reactions which may involve the use of toxic metals such as cadmium, and provides a clean and efficient alternative (Guillard and Kopp, 2004).

1 3 2 + 2 푈푈 200−300 푛푛 2 푁푁 −�⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯� 푁푁 − � The advantages of In situ nitrate sensors. Long-time water quality monitoring

needs a lot of labor and time for sample collection and water chemistry analyses, also it

is especially impractical for the studies requiring the spatial and temporal resolution.

Besides, sample handling and storage can introduce contamination and analysis errors.

The high-resolution in situ nitrate sensor can overcome all the problems above. Firstly,

in situ nitrate sensors were reported to be as accurate as grab samples and can be

deployed in various aquatic environment such as seawater (Bijoor et al., 2008; EPA,

2011; Weiss et al., 2008; Withers et al., 2013), freshwater (Diaz-Valbuena et al., 2011)

and wastewater (Weiss et al., 2008); Secondly, in situ nitrate sensors can be set up to make measurements at high resolution (e.g. every second at maximum for Submersible

Ultraviolet Nitrate Analyzer (SUNA) from Satlantic Inc. (Halifax, Canada)), and the deployment can last as long as the battery has power (Bowes et al., 2015; Gilbert et al.,

2013). The capability for long term field deployment suggests the datasets from in situ nitrate sensors can contain thousands of data points and are not limited by sources of funding as other sampling methods are. Thirdly, the fact that in situ nitrate sensors are

more portable than autosamplers and not limited to numbers of samples enables the

capture of N dynamics both temporally and spatially. Hensley et al. (2014) estimated N removal in large rivers, mapped the nitrate concentrations on high-resolution longitudinal profiling by using in situ nitrate sensor, and concluded that the N removal was controlled by spatial variation in channel hydraulics and vegetation.

Even though the in situ sensors have many advantages over the conventional measurement, the in situ sensors can have problems. Nitrate absorbs significantly at

wavelengths up to 230 nm, while some interferences in the seawater may be introduced

by bromide or dissolved organic matter and carbonate which exhibit close band

absorption with nitrate (EPA, 2011). Thomas et al. (2010) suggested that shortening

the path length of the light beam in the sensor could overcome the attenuation caused

by the dissolved organic carbon, but it resulted in reduced sensitivity as well (Paul and

Meyer, 2001).

Currently, the representatives of in situ nitrate sensors are Submersible

Ultraviolet Nitrate Analyzer (SUNA) from Satlantic Inc. (Halifax, Canada), Digital

NitraVis UV/Visible sensor from YSI Inc. (Yellow Springs, Ohio), and Ion-Selective

Electrode sensor from In-Situ Inc. (Bingen, Washington). With all the advantages of in

situ continuous sensors, an increasing number of studies have been conducted

producing more details about N biogeochemical processes under different aquatic

environment. The new technology may open a new era for nutrient control and

management studies.

Conclusions

The development of human society has made the world awash with N. Human

activities have enormously altered the N cycling mainly with the additions of artificial N

fixation (fertilizer) and combustion engine. What is new about the alteration is the magnitude. In 1970, anthropogenic activities including fertilizer production, fossil fuel combustion and legume and rice production mobilized about 70 Tg N yr-1. By the end of

20th century, the number had doubled to about 140 Tg N yr-1. Furthermore, based on the estimated anthropogenic N inputs, the reactive N input rate worldwide had been doubled in the terrestrial cycling compared to preindustrial times.

Excessive reactive N can cause a series of environmental problems such as lake eutrophication and even pose hazards to human health such as “blue-baby” syndrome and gastric cancer if the groundwater is polluted. Sources that lead to excessive N in the environment include atmospheric deposition, reclaimed water, fertilizers, biological

N fixation, septic tanks, and storm water.

To manage the two major N sources-urban fertilizers and stormwater runoff, different management practices were proposed. Proper fertilizer management and irrigation management can help to reduce N leaching/runoff from lawns. Bioretention, wetlands and greenroofs are suggested to reduce the N in stormwater runoff, meanwhile, water quantity from stormwater runoff should be reduced. Managing these

N sources is important to reduce risks to the environment, but management depends on understanding the quantities of the various pools of N in the environment.

The validity of the water quality samples is closely associated with the proper sampling methods and strategies. Grab sampling is usually used in baseflow sampling and some stormwater first flush sampling, while autosampler sampling is usually applied in the stormwater studies with the options of flow-proportion and time-proportion. With the advancement of technology, there is a trend to apply in situ sensors in the field to

obtain continuous information regarding nutrient dynamics. In situ sensors can provide a new interpretation of hydrological and biogeochemical processes through the continuous data, therefore it may open a new era for future nutrient control and management studies.

CHAPTER 2 NITROGEN FLUXES FROM THREE DIFFERENT SMALL URBAN CATCHMENTS

Introduction

Nitrogen (N) is one of the necessary elements every living organism needs for

growth and reproduction. In general, lightning and biological nitrogen fixation are the

two major sources for N enrichment in the natural environment. The low conversion

rate from inert N2 to active N makes N the limiting factor in many plant systems

(Vitousek and Howarth, 1991). The Haber-Bosch process and internal combustion

engine resulted in a doubling of the N inputs compared to preindustrial times (Galloway

et al., 1996). Dissolved inorganic N such as nitrate/nitrite is the major N form

associated with anthropogenic inputs. The leaching of nitrate to the groundwater can

lead to human health issues such as blue baby syndrome (Kaye et al., 2006) and gastric cancer (Forman et al., 1984). Moreover, the excess N runoff has been closely linked with eutrophication in surface water bodies which may give rise to the depletion of oxygen and the suffocation of fish (Anderson et al., 2002; Glibert et al., 2002).

Researchers have focused on the watershed scale to address water quality

protection and regulation (Brezonik and Stadelmann, 2002). Watershed studies can be

used to identify sources and fates of nutrients facilitating mass balance calculations.

Understanding nutrient balance in a watershed will help develop best management

practices for reducing nutrient losses. For example, in the Baltimore Ecosystem Study

funded by US National Science Foundation (NSF), Groffman et al. (2004) suggested

that urban and suburban watersheds had a higher N flux than the completely-forested

watershed. In addition, the suburban watershed had a greater retention of N compared

to the urban watershed; the sources were mainly home lawn fertilizer and atmospheric

deposition. It is important to understand the various land uses that comprise the watershed. The results from a study of nutrient flux at the watershed level are comprised of the integrated results from those land uses in the watershed. Therefore characterization of a single land use can help build an understanding of environmental

phenomena under the watershed scale. So far only a few research studies have been

conducted on the N losses from specific urban land uses. Those studies included impervious road surfaces (Gilbert and Clausen, 2006; Wu et al., 1998), residential areas

(Line et al., 2002), golf courses (Wong et al., 1998), and woodlands (Ávila et al., 1992).

Little is known about the nutrient imports and exports in an urban watershed comprised of areas of land receiving no irrigation or fertilization, urban land which involves intensive fertilization/irrigation and high leaching capacity, or urban land which is only

irrigated by reclaimed water from wastewater treatment plants.

In order to better estimate the nutrient loads from various land uses, the

hydrological factors should be considered. Johnes (2007) stated that catchments with a

high Baseflow Index (BFI) had a lower Root Mean Square Error (RMSE) on load

estimates when sampled infrequently than those with a low BFI. Infrequent sampling

can have a reduced chance of capturing extreme flow conditions, reducing the

confidence in the load estimation. Therefore, nutrient dynamics in both baseflow and

stormflow are usually included in load estimations (Robertson and Roerish, 1999). The

goal of this research was to understand flow of N from three small urban catchments

having different nutrient management practices. The objectives of this study were: 1. to

establish a relationship between rainfall and runoff in each catchment; 2. to establish a

relationship between discharge and N load in each catchment; 3. to estimate the N load

from each catchment. A number of studies have been conducted to determine the relationship between rainfall and runoff from different land uses (Garcia-Martino et al.,

1996; Liu et al., 2008). Also among all the urban land uses in this study, the sports field

catchment with intensive fertilization management was likely to deliver larger N loads

(Miller and Cisar, 2001) than other catchments. Therefore, my hypotheses were: 1.

There would be a relationship between rainfall and discharge in each catchment; 2.

Sports field catchment would make the greatest N load among all the catchments. This

study is needed by regulators and land managers to identify land use practices that may

be associated with greater N loads and to guide decisions about improved management

of N imports and exports from different land uses. Study results will enrich the database

of N mass in various land uses in Florida. Also once the relationships between rainfall

and runoff, and between discharge and N load are established, future environmental

monitoring work can be improved with less water quality sampling, replaced with

continuous hydrology monitoring instead (monitoring rainfall and flow).

Materials and Methods

Study Area

The study was carried out in three small urban catchments on the campus of the

University of Florida in Gainesville, Florida (Figure 2-1) from September 2013 to August

2014. Gainesville is located in North Central peninsular Florida with a humid subtropical

climate. It has very distinct dry season (October to April) with an average monthly

precipitation of 75 mm and a wet season (May to September) with an average monthly

precipitation of 153 mm (Zhang and Sansalone, 2014). The annual precipitation from

September 2013 to August 2014 was 1600 mm.

Figure 2-1. The locations of three catchments in the Lake Alice Watershed on the University of Florida Campus in Gainesville, Florida

The University of Florida (UF) is one of the largest universities in the nation with an area of 800 hectare (ha) and more than 900 buildings (including 170 with classrooms and laboratories). It has nearly 50,000 students. The UF residence halls have a total capacity of 7,500 students and the five family housing villages house more than 1,000 married students and graduate students. The campus includes a large quantity of parking spaces in the outdoor parking lots or the parking garages. The campus land area also contains many natural areas and areas receiving irrigation and fertilization.

This study used three catchments that represented a cross-section of the variable land uses on the campus. The urban Sports Field Catchment (SFC) with intense fertilization is in the northern part of the campus, operated by University Athletic

Association, and it mainly consists of the football practice field and the baseball field

with an approximate total area of 6.7 ha. The pervious surface makes up 95% of the

entire area (Table 2-1). The urban Reclaimed Water Irrigated Catchment (RWC) is in

the center eastern part of the campus with dormitories and lab/classroom buildings

including a large number of landscaped sites with an approximate area of 1.5 ha with

76% pervious surface. The irrigation with reclaimed water is applied twice a week in

this catchment except on the days when air temperature drops below 0 Celsius degree.

The Control Catchment (without irrigation/fertilization) (CC) is located in the

southeastern corner of the campus where the apartments of graduate students and their

families are located with an approximate area of 2.6 ha. This area is comprised of 76%

pervious surface. Runoff in all three catchments eventually drains to Lake Alice, a large

retention pond with an open area of 35 ha on the campus. Water in Lake Alice

overflows to two injection wells connected with the groundwater system.

Table 2-1. Pervious/impervious ratio in each Lake Alice catchment Land Use1 CC RWC SFC

Impervious 24.5% 24.5% 5.4%

Pervious 75.5% 75.5% 94.6% 1 Control catchment (CC), Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

General Method

A linear regression relationship was established between daily precipitation and

daily discharge in each catchment. Based on this relationship, the missing daily

discharge data due to severe weather, used-up batteries and other reasons were

calculated. Grab samples from baseflow sampling and composite samples from autosamplers during storm events were analyzed for NO3-N and TKN concentrations. A

log-log relationship was established between the N loads and the simultaneous

recording flow rates. Then the total N loads were calculated based on recorded flow

rates and summed up eventually to obtain the annual N load.

Flow Measurement

A naturally formed channel conveyed runoff from the drainage pipe of SFC in the

woods at University of Florida. A compound aluminum weir (diameter=91.5cm, 2-2 a)

was installed at sufficient horizontal distance from the pipe culvert to avoid submerged

conditions at the weir. The weir was exposed to an open environment surrounded with trees and other plants. Two Thel-Mar volumetric weirs (Thel-Mar, LLC, Brevard, NC)

were deployed separately at the end of stormwater culverts in the other two catchments

(the diameters of the pipes are 45.7cm for RWC and 38.1cm for CC, Figure 2-2 b and

Figure 2-2 c) (DeBusk and Wynn, 2011; Fowler et al., 2009). The Thel-Mar weir for

RWC was installed in an underground concrete stormwater collection area covered by a

cast-iron grate. There were two in-coming pipes delivering stormwater from the

surrounding area and a single pipe transporting the stormwater from the collection area.

The weir was installed in the outflow pipe. All of the runoff from the two incoming pipes

was compiled in the collection area and went over the Thel-Mar weir before flowing to

Diamond Stream, a stream located 200 m from this collection area and draining to Lake

Alice. The Thel-Mar weir in CC was installed at an outlet pipe where the runoff drained to Diamond Stream (Figure 2-1). All three sites were equipped with a pressure

transducer WL16 (Xylem Inc., Sacramento, California) which recorded the water stage

every 5 minutes. The discharge in SFC was calculated based on the stage-discharge rating curve determined by Kindsvater-Shen equation (USBR, 1997); the discharge in

two Thel-Mar weirs was calculated based on the rating curve provided by Thel-Mar,

LLC. The pipe in RWC was full of runoff several times, the scenario was beyond calculation provided by Thel-Mar LLC, therefore, the full-pipe discharge was calculated based on Manning’s equation.

The calculated discharge from all the sites was validated with manual measurement, using a bucket to measure the volume of runoff within a recorded time.

a b c

Figure 2-2. Flow measurement devices in sites (a. Sports Field Catchment (SFC) with an aluminum weir; b. Reclaimed Water Irrigated Catchment (RWC) with a Thel-Mar weir (in-flow end of pipe) and a pressure transducer on the inflow side of the weir; c. Control Catchment (CC) with a Thel-Mar weir in the outflow end of the pipe and a pressure transducer on the inflow side of the weir)

The completeness of the time series discharge datasets in each catchment is

calculated as the ratio of the number of obtained measurements, nm, to the number of

expected measurements, ne, the numbers of measurements estimated based on the

frequency if the measurements were recorded continuously for the duration of the

record. Table 2-2 provides the statistics for each site for the entire study period.

Table 2-2. Completeness statistics for discharge records from each catchment through the monitoring period with annual totals for the hydrological year 2013-2014 No. of Land Start of End of No. of expected Completeness2 use1 record record measurements measurements % CC 9/1/13 0:00 8/31/14 23:55 85590 105120 81.4 RWC 9/1/13 0:00 8/31/14 23:55 97950 105120 93.2 SFC 9/1/13 0:00 8/31/14 23:55 95592 105120 90.9 1 Control catchment (CC), Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC). 2Completeness is calculated as the number of measurements divided by the number of expected measurements.

% Completeness = × 100 푛푛 푛푛 Precipitation Measurement

Two tipping bucket rain gauges (Blue Siren Inc., Melbourne, FL) which tipped

after every 0.254 mm of rainfall were placed in an open area on the top of a utility

building on campus and used for measurement of precipitation. The building was

approximately central to all three catchments. The rain gauges were operational from

November 21st 2013. The precipitation data before that date was retrieved from the UF

Physics Department which was also recorded by a tipping bucket. Three 4.73-L plastic

buckets were put near the study areas to catch the rainfall water for TKN and NO3-N

analyses.

Baseflow and Stormflow Sampling

Baseflow water samples were collected for one hydrological year from

September 2013 to August 2014. 100 ml water samples with duplicates were manually

collected by hand using individual sample bottles near the V-notch in weirs which water

flowed over in both SFC and CC (Stone et al., 2000). Clear plastic tubing (Thermo

Fisher Scientific Inc., Waltham, MA) was used to pump the water samples before the

water flowing over the Thel-Mar weir from the stormwater collection area in RWC under

the cast-iron grate. Baseflow collection was carried out at least twice a month in dry

days, except in February 2014 when it kept raining around the expected second sampling time of that month. The specific day of sampling was determined one week earlier. Dry days are defined as being dry for the past 24 hours.

Stormwater samples were collected automatically using ISCO 6700 auto sampler

(Teledyne Technologies Inc., Lincoln, Nebraska) during 15 storms from February 2014 to August 2014. The sampling strategy was to collect time-proportional composite samples. Storm satellite images were tracked before the storms, autosamplers were triggered by the estimated start time of the storm and programmed to take samples every 5 to10 minutes. Every 3 to 4 samples comprised a 500-ml composite sample depending on the duration of the storms. After the storms, the discharge peaks in all hydrographs were checked to ensure they were captured by the samples. Composite samples with 15 minutes or 20 minutes duration (3 to 4 5-min samples) and averaged discharge during the composite time were used to establish the relationship between N concentrations and discharge together with the samples and discharge under baseflow scenario. This composite method during high flow combined a few water samples (3 to

4 samples) during a relatively short time (15 min to 20 min), it can well mix the sampled water during high flow and therefore reduce the potential errors from turbulent flow. It is assumed that the difference in sampling between low flow (baseflow) and high flow

(stormflow) would not make significant changes to the relationship between N concentration and discharge.

Baseflow water samples (100 ml) were stored on ice once collected and then were subsampled in the laboratory. Stormflow water samples (500 ml) were taken directly from the field to the laboratory within 12 hours after the storm for subsampling.

Sub-samples (two 20-mls duplicates) in new plastic scintillation vials were sent to UF

Analytical Research Laboratory (ARL) at University of Florida for analyses on the same day or the next day. All sample bottles (100 ml and 500 ml) were pre-cleaned by acid washing followed by distilled-deionized water rinses. NO3-N subsamples were filtered

before submitting for analyses. Sulfuric acid was added to each subsample to ensure a

pH<2 before they were sent to ARL. NO3-N was measured in every baseflow

subsample. Total Kjehldahl Nitrogen (TKN) was measured twice in 2013, and then

twice a month since February 2014 in baseflow subsamples. Both NO3-N and TKN

were measured in every stormwater subsample. Total N concentration was calculated

as the sum of TKN concentration and NO3-N concentration. The Method Detection

-1 Limits (MDLs) in Analytical Research Laboratory (ARL) were 0.148 mg L for NO3-N,

and 0.125 mg L-1 for TKN. Concentrations below MDLs were assumed to have imputed values of the MDL divided by 2.

Data Analysis

A linear regression relationship was established between rainfall (P) and average daily discharge (Qd) in each catchment:

= + (2-1)

푄푄 훼Where∗ 푃 훽α is the slope and β is the intercept.

A log-log transformed linear relationship was established between N loads

(MNO3-N for the mass of NO3-N, MTKN for the mass of TKN) and discharge (Q) in each

catchment:

ln = ln + ln (2-2)

푀 Where푎 푏 a and푄 b are regression constants.

N loads (NO3-N load and TKN load) were calculated based on Equation 2-2 with known discharge records. The missing discharge was calculated from Equation 2-1 and then used to calculate the N loads based on Equation 2-2.

Results and Discussion

Hydrological Factors in the Three Catchments

Precipitation. The dry season and wet season in Florida are very distinct. The

precipitation increased in each catchment as the dry season moved towards the wet

season. The maximum daily precipitation was 67 mm on March 17th 2014 and the

minimum daily precipitation was 0.254 mm which was the minimum detection limit for

the rain gauges. The distribution of total daily precipitation (Figure 2-3) demonstrated

that about 50% of the total daily precipitation was between 0 mm and 5 mm, indicating

small storms (<5mm) usually happen through the year.

Figure 2-3. The distribution of total daily precipitation in Lake Alice Watershed on the University of Florida Campus in Gainesville, Florida. Top and bottom edge of each box represents the 75th percentile (20 mm) and 25th percentile (1.3 mm), respectively, the line bisecting the box represents the median (5.09 mm), points are outliers, and the ends of the whiskers represent the 90th and 10th percentile.

Storm descriptions. 15 storms were sampled from February 2014 to August

2014 with precipitation ranging from 1.5 mm to 46 mm (Table 2-3). The precipitation of

6 storms fell in the range between 25% (1.3 mm) and 50% (5.09 mm) of the distribution

of total daily precipitation (Figure 2-3), 4 fell the range between 50% (5.09 mm) and

75% (20 mm), and 5 fell in the range between 75% (20 mm) to 100% (67 mm).

Sufficient runoff may not be produced for storm studies when the precipitation is less

than 1.3 mm, therefore storms below 1.3 mm were not considered in this study. In this

case, the storms collected in this study were almost evenly distributed in the three

ranges above, hence the samples collected during storms can be considered to

represent the N concentrations during peak flow through the hydrological year.

Table 2-3. Hydrological and chemical factors of the storms 1 Precipitation Previous Dry Imax NO3-N TKN (mg TN Date -1 -1 -1 -1 (mm) hours(hr) (mm hr ) (mg L ) L ) (mg L ) 2/5/14 5.6 114 91.4 0.2 0.3 0.5 2/21/14 3.8 148 45.7 0.12 0.2 0.3 2/21/14 32.5 2 106.7 0.12 0.2 0.3 3/17/14 46.01 238 76.2 0.2 0.12 0.2 4/30/14 2.8 19 45.7 0.4 0.8 1.1 5/1/14 1.5 23 30.5 0.3 0.9 1.2 5/25/14 29.0 233 121.9 0.3 1.1 1.4 5/29/14 4.6 24 76.2 0.9 1.2 2.1 5/30/14 21.3 24 121.9 0.2 0.5 0.7 7/22/14 8.6 49 91.4 0.3 0.3 0.6 8/2/14 7.9 263 45.7 0.4 0.3 0.7 8/8/14 2.0 42 15.2 0.5 0.3 0.8 8/11/14 4.6 47 30.5 0.4 0.3 0.6 8/29/14 19.8 120 61.0 0.3 0.1 0.4 2 8/30/14 41.7 12 91.4 0.1 0.2 0.3 Median 7.9 47.0 76.2 0.3 0.3 0.6 Mean 15.4 90.5 70.1 0.3 0.5 0.7 SEM 3.9 23.4 8.8 0.1 0.1 0.1 1The precipitation was not the total daily precipitation, it refers to event precipitation during which we collected samples. 2 -1 -1 Values with subscripted 2 were below Method Detection Limit (NO3-N=0.148 mg L , TKN=0.125 mg L ) in UF Analytical Research Laboratory, and were substituted with half of the MDL for statistical analysis -1 -1 (NO3-N=0.074 mg L , TKN=0.063 mg L ).

The maximum intensity occurred in two storms (5/25/14 and 5/30/14) with a value of 121.9 mm hr-1. The previous dry hours varied from 2 hours (within a day) to

263 hours (11 days). During the second storm on February 21st 2014 and the storm on

August 29th 2014, the equipment in Control Catchment (CC) did not properly function in

the field, therefore, only 13 storms were recorded for CC.

Rainfall-discharge Relationship

There was a strong positive linear relationship between precipitation and average

daily discharge in the Reclaimed Water Irrigated Catchment (RWC) (R2=0.77) and

Sports Field Catchment (SFC) (R2=0.87) (Figure 2-4) whereas there was a weak

positive linear relationship in CC. This suggested that most of the precipitation water

became runoff in RWC and SFC, but precipitation water may have been stored in soil or

lost in other N forms in the CC. The slopes in the relationship between precipitation and

average daily discharge in CC and RWC were very small (0.05 for CC and 0.06 for

RWC) relative to the intercepts, indicating the power from precipitation was not great.

Instead the baseflow line which remained in all hydrographs may play a more important

role or make a greater contribution in total discharge. On the other hand, SFC had a

slope of 0.22, suggesting stormflow may play a significant role in total discharge. This

can be associated with the well-designed drainage system for the sports fields (Adams,

1986). The system drained the rainfall water to the drainage pipe so fast that the

pervious field surface can function equal to an impervious surface.

Land use =CC Land use =RWC Land use =SFC Qd=0.61+0.05*P Qd=0.32+0.06*P Qd=0.47+0.22*P 2 2 R =0.12, p<0.0001* R2=0.77, p<0.0001* R =0.87, p<0.0001*

Average daily discharge (Qd), L/s discharge daily Average Precipitation, mm (P) Figure 2-4. Rainfall-discharge relationship in all catchments (CC=Control Catchment, RWC=Reclaimed Water Irrigated Catchment, SFC= Sports Field Catchment). * Significance level p=0.05

N Loads Estimation

A log-log transformed linear regression relationship was established between

discharge (Q) and NO3-N load (MNO3-N) (Figure 2-5), and between discharge (Q) and

TKN load (MTKN) (Figure 2-6) in each catchment. The results showed a strong positive

relationship between discharge and NO3-N load, suggesting the NO3-N loads increased

as the discharge increased in each catchment. The discharge-NO3-N load relationship was weaker in SFC (R2=0.64) compared with CC and RWC. This relationship may be

associated with high NO3-N concentration in baseflow (Table 2-4) which possibly

created a larger NO3-N load during low flow. Similarly, a strong positive relationship

was established between discharge and TKN load. The slopes in the discharge-TKN

load relationship indicated that the changes in discharge may possess a greater power

to the variation of TKN load than what stayed in the baseflow line.

) N - Land use =CC Land use =RWC Land use =SFC Ln(MNO3-N)=-1.52+0.77*Ln(Q) Ln(MNO3-N)=-0.52+0.68*Ln(Q) Ln(MNO3-N)=1.98+0.52*Ln(Q)

NO3 2 2 R =0.84, p<0.0001* R2=0.79, p<0.0001* R =0.64, p<0.0001*

N load, mg (M mg load, N - 3 NO Discharge, L s-1 (Q) Figure 2-5. Discharge-NO3-N load relationship in all catchments (CC=Control Catchment, RWC=Reclaimed Water Irrigated Catchment, SFC= Sports Field Catchment). * Significance level p=0.05

Land use =CC Land use =RWC Land use =SFC ) Ln(MTKN)=-0.40+1.15*Ln(Q) Ln(MTKN)=-1.02+1.19*Ln(Q) Ln(MTKN)=-0.11+1.13*Ln(Q) R2=0.89, p<0.0001* R2=0.89, p<0.0001* R2=0.86, p<0.0001* TKN TKN load, mg (M mg TKN load,

Discharge, L s-1 (Q)

Figure 2-6. Discharge-TKN load relationship in all catchments (CC=Control Catchment, RWC=Reclaimed Water Irrigated Catchment, SFC= Sports Field Catchment). * Significance level p=0.05

Table 2-4. Description of N concentrations and discharge (Q) in each catchment Land use1 CC RWC SFC -1 NO3-Nmin, mg L 0.07 0.07 0.17 -1 NO3-Nmax, mg L 2.59 1.48 15.61 -1 TKNmin, mg L 0.06 0.06 0.06 -1 TKNmax, mg L 4.75 3.22 2.92 -1 Qmin, L s 0.01 0.06 0.14 -1 Qmax L s 34.66 60.89 251.98 1 Control catchment (CC), Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

Based on the discharge-N load relationship and precipitation-discharge relationship, N loads from all catchments were calculated and listed in Table 2-5.

The results (Table 2-5) illustrated that the total N loads from intensely-fertilized catchment (SFC) were substantially greater than loads from the other catchments which represented the common urban (RWC) and suburban catchments (CC). These results supported one of the hypotheses that SFC contributed greater N loads than other catchments. The N yields from SFC were even greater than from agricultural watersheds according to some reported scientific literature(Groffman et al., 2004;

McMahon and Woodside, 1997), but similar to the agricultural watershed reported by

Peterjohn and Correll (1984) and fell within the range of N export from an agricultural

watershed reported by Schilling and Zhang (2004). The recommended annual

fertilization rate for sports fields in Florida is 200 to 300 kg ha-1 depending on the

sources of fertilizers and the frequency of application (Miller and Cisar, 2001).

Calculations from Table 2-5 indicated that up to 20% of the total annual N applications

could be lost in SFC, assuming a fertilizer application rate of 200 kg ha-1. CC and RWC

produced a greater TN load than the suburban/urban catchments in Baltimore, MD

(Groffman et al., 2004). Nevertheless, they fell within the urban N load range reported

by Young et al. (1996). The TN yields from RWC are close to CC with only a difference

of 3 kg ha-1 yr-1 despite reclaimed water irrigation was practiced in RWC.

Table 2-5. Annual N loads estimation for each catchment from 2013 to 2014

Land use1 CC RWC SFC

NO3-N TKN TN NO3-N TKN TN NO3-N TKN TN (kg ha-1) (kg ha-1) (kg ha-1) (kg ha-1) (kg ha-1) (kg ha-1) (kg ha-1) (kg ha-1) (kg ha-1) 1.87 7.54 9.41 6.81 5.49 12.31 28.54 8.58 37.12 1 Control catchment (CC), Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

Conclusions

A number of studies have been conducted on urban catchments such as golf

courses, parking lots, and residential lawns, but very few on a reclaimed water irrigated

catchment or for an intensely-fertilized sports field catchment. TKN made up a larger

portion than NO3-N in TN contribution in CC, but NO3-N contribution was greater in

RWC. In the annual N contribution, CC and RWC only had 3 kg ha-1 yr-1 difference.

Therefore, it can be inferred that using reclaimed water as an irrigation and fertilizer

source is an economical and green way to save drinking water in the condition that the

irrigation frequency is low. However, more concerns can be caused when the irrigation

frequency/intensity increases. More studies also need to be conducted with regards to

the environmental hazards it might pose from phosphorus and other chemicals from

reclaimed water.

Like agricultural watersheds/catchments, Sports Field Catchment is heavily fertilized and irrigated, also it has a high capability of leaching losses. This study

illustrated that SFC had the highest NO3-N concentration among all catchments, falling

within the range of the runoff from agricultural watersheds in other studies. The annual

N contribution from SFC was as great as 40 kg ha-1 yr-1, which was about the same or

even larger than agricultural watersheds. Catchments like SFC are usually overlooked

because they are located in the urban metropolitan area and do not take as much area as agricultural watersheds/catchments. Considering the fact that a great number of

universities, colleges and metropolises have sports fields similar to this study, the long-

term environmental consequences can be substantial. Hence, more attention is needed

to explore and develop management plans or best management practices for nutrient

control in those catchments/watersheds.

This study provided relationship between precipitation and discharge as well as

discharge and N loads, indicating that future studies in the same catchments can rely

much on the hydrology records than water quality sampling. The similar relationships

can also be used in other catchment study.

CHAPTER 3 NITROGEN BUDGETS FOR A SMALL URBAN WATERSHED AND THE NITROGEN FLUXES FOR THE DRAINAGE TO THE WATERSHED

Introduction

The nutrient mass budget approach has proven to effectively identify the sources and fates of nutrients. For example, the global budget for Chlorine (Cl) allows researchers to observe how the small amounts of chloroflurocarbons released by human activities are the source of nearly all the Cl that mixes into the stratosphere

(Graedel and Keene, 1996), where it destroys ozone (Rowland and Molina, 1975).

Moreover, this approach has been widely applied in evaluating and improving agricultural management practices. Hiscock et al. (2003) estimated the annual phosphorus (P) budget in the northern Lake Okeechobee watershed, FL, and indicated that the adoption of Best Management Practices (BMPs) since 1991 substantially decreased P runoff concentration for truck crops from 6 to 0.55 mg L-1. Alva et al.

(2006) applied a nitrogen (N) mass budget approach to a citrus tree plot experiment.

They developed improved N fertilization and irrigation management practices for

improved N uptake efficiency and N loss reduction. With the increasing interests in

human-dominated urban ecosystems, the mass budget approach has also been carried

out to evaluate urban ecosystems’ function and structure. For example, Baker et al.

(2001) constructed a N budget for different subsystems under the Central Arizona-

Phoenix (CAP) ecosystem and suggested that anthropogenic N inputs to urban and

agricultural subsystems were at a greater magnitude than inputs to the desert. Similarly,

Groffman et al. (2004) applied the mass budget approach to study different N fluxes in

the urban, suburban, agricultural and forested watersheds in Baltimore, MD and

discovered that the urban watershed and suburban watershed (mainly lawns) had much higher N yields than a forested watershed. An isotopic tracer study suggested that the urban and suburban lawns under low to moderate management intensities were an important sink for atmospheric N deposition (Raciti et al., 2008).

Even though the mass budget approach has become preferred over the

ecological footprint approach in ecological studies (Kaye et al., 2006), it has some

requirements and limits. Firstly, it requires delineated geographic boundaries.

Depending on how the boundaries are delineated, the nutrient mass budgets for a study

area can be diverse. Then, there are often considerable amounts of uncertainties in the

nutrient budgets, due to various possible biases and errors, particularly in the

partitioning of nutrient losses (Fowler et al., 2009; Harmel and King, 2005; Johnes,

2007). In addition, the relative uncertainties are much larger in natural systems than in human-interrupted systems (Johnes, 2007) although human-interrupted systems have more complex nutrient fluxes than natural systems.

Research focuses have been placed on nutrient fluxes quantified in urban water flow (e.g. urban streams ) to understand human-dominated urban ecosystems

(Groffman et al., 2004). Urban streams play a significant role in the urban ecosystems because of their unique functions. For example, their position in urban landscape makes the natural native ecosystems particularly vulnerable to impacts associated with land use change, however, their feature as habitats for a potentially diverse and productive biota enables them to process different materials/pollutants (Paul and Meyer,

2001). In N dynamics studies, smallest streams are considered to be places where the most rapid uptake and transformations of inorganic N occur. Peterson et al. (2001) used

15N to measure uptake of inorganic N in small streams throughout North America. Their results indicated that small streams were important sites of N retention because their large surface-to-volume ratio favored rapid N uptake and processing, suggesting that small stream management may reduce N loading of downstream rivers, lakes, estuaries and oceans. Small streams are an important component of many urban environments.

Large university campuses, like large urban areas, may have several streams flowing through the campus.

Chapter 2 determined the N fluxes from catchments with various land uses and discussed about the potential sources. It demonstrated that Sports Fields Catchment produced the greatest N load among all catchments, equivalent to 20% of the total annual N applications in SFC, assuming a fertilizer application rate of 200 kg ha-1.

However, little is known about the spatial changes of the N concentrations as they are delivered to downstream, and the influence of catchments/basins with those land uses on the urban streams as well as Lake Alice in Lake Alice watershed in terms of N loads.

In this study, three small urban streams on the campus of the University of Florida were targeted for investigation of the spatial changes of N fluxes as they pass from upstream to downstream in urban ecosystems. The objectives were: 1. to make a N mass budget for Lake Alice watershed; 2. to identify the major N sources associated with land uses and possible losses in Lake Alice watershed; 3. to describe the spatial changes of N fluxes and N concentrations in urban streams in Lake Alice Watershed;. My hypotheses were: 1. The major N source in Lake Alice Watershed would be associated with fertilizer from sports fields, the streams closely related to delivery of flow from sports fields produced most N; 2. nitrate-N concentrations decreased as the flow reached to

downstream from upstream; 3. The hydrological change from upstream to downstream attributed to the decrease of nitrate-N concentrations. There were some uncertainties in this study which can affect the accuracy of the final N budget estimation; however, the results can provide the land managers and researchers who conduct similar studies an understanding of the magnitude of the N transport in a small urban watershed and a method to estimate the N budget under some uncertain circumstances.

Materials and Methods

Study Area

Gainesville is located in North Central peninsular Florida with a humid subtropical

climate. The seasonal and annual precipitation was described in more detail in Chapter

2. The University of Florida (UF) is one of the largest universities in the nation. It has

various land uses (described in Chapter 2) which make the campus form a small urban

“city” and function similar to a city. The Lake Alice watershed covers an area of 426

hectare (ha), approximately 55% of the total campus area. There are three major

streams in the watershed, they are named in this study by the author as Hume Creek,

Fraternity Stream and Diamond Stream. Hume Creek is comprised of two branches

(Graham Woods Stream and Reitz Union Stream) which flow through the most popular

and centralized areas on campus including the academic buildings, residential halls,

and sports fields. Reitz Union Stream originates from academic buildings in the center

of the campus. Graham Woods Stream and Fraternity Stream originate from the

students’ dormitories and sports fields in the Northern boundary of the campus.

Diamond Stream originates from the East boundary of the campus, which passes

through the residential housing areas, Shands Hospital (university medical center), and

the UF Wastewater Treatment Plant. Eventually the three streams discharge to Lake

Alice, a scenic 35-ha retention pond on campus (Figure 3-1) after they pass through the wetlands surrounding the lake. The outfall of the lake is connected to the groundwater system directly with two underground siphon pipes or injection wells. Only the larger siphon pipe was monitored in this study since the other pipe does not transport lake discharge except under extreme storm events. It is assumed that the annual discharge from the smaller injection well does not significantly affect the annual N load discharge from Lake Alice.

The composition relationship in this study is defined as following: a watershed is comprised of basins and a basin is comprised of catchments. Lake Alice watershed has

three major stream basins which the three major streams pass through and were

named as Hume Creek Basin, Fraternity Stream Basin and Diamond Stream Basin,

respectively (Figure 3-1). Each stream basin consists of catchments with different land

uses. There are five catchments in the study. Sports Field Catchment (SFC), Control

Catchment (CC) and Reclaimed Water Irrigated Catchment (RWC) are single land-use

catchments (more details in Chapter 2). Fraternity Stream Upstream Catchment and

Graham Woods Stream Catchment are mixed land-use catchments. The Hume Creek

Basin includes the Graham Woods Stream Catchment and several small urban

catchments that Reitz Union Stream passed through (not shown in Figure 3-1). The

Graham Woods Stream Catchment includes the SFC (Chapter 2), Graham Woods and

some residential areas. Similar to Graham Woods Catchment, Fraternity Stream

Upstream Catchment is also comprised of sports fields, forests and residential

buildings, and becomes part of Fraternity Stream Basin. Both CC and RWC are small

catchments in Diamond Stream Basin. The area with white color in Figure 3-1 is the area in Lake Alice watershed except the three major basins. It is referred to as “Outside

Basins” in this study. Outside Basins are mainly composed of dormitory buildings, academic buildings, landscape and parking lots like Diamond Stream basin. Table 3-1 shows the ratio of pervious area to impervious area in each basin.

Figure 3-1. Major stream basins in Lake Alice watershed

Table 3-1. Proportion of pervious/impervious surface in each stream basin Fraternity Hume Diamond Outside Stream Creek Stream Basins Area, ha 9.5 65.6 156.5 194.5 Impervious 15.4% 7.97% 19.8% Pervious 84.6% 92.03% 80.2%

General Research Approach

N exports from three streams were calculated from measured water quality and

discharge data from catchments with single land use in Chapter 2 and from catchments

with mixed land uses in this Chapter. N exports from Outside Basins were estimated

from obtained data from the stream basins. The general idea was quantify the N inputs

and outputs to Lake Alice. First, to calculate the N exports from all basins to Lake Alice,

and then to quantify N efflux from Lake Alice. A number of sampling sites were chosen

to describe the changes in N in each stream as part of description of the entire N budget

along a stream. Water quality data from those sampling sites were collected for a

hydrological year. The N export sites to Lake Alice from each stream (Site No.3 for

Fraternity Stream Basin, Site No.9 for Hume Creek Basin, Site No.12 for Diamond

Stream Basin) had no discharge records, therefore the discharge from those export

sites were extrapolated from catchments which had discharge data under some

assumptions. The N export from Outside Basins had neither water quality nor

discharge data, therefore the data were referenced from measurements from stream

basins.

Flow Measurement/Estimation

Ultrasonic water level and velocity sensors (Bluesiren Inc., Melbourne, FL) were

deployed in the Lake Alice injection well (Site No.1, Figure 3-1), Fraternity Stream

upstream (Site No.2, Figure 3-1), and Graham Woods Stream downstream (Site No.7,

Figure 3-1). The sensors were set up to measure the water level and velocity every 5

minutes. The data were uploaded to the virtual system online (www.ziscape.com,

user’s information needed for data management) through a SIM card every day. The

discharge data recording began from September 2013 in Fraternity Stream upstream

and Graham Woods Stream downstream, and from November 2013 in Lake Alice

injection well. Recording ended in August 2014. It was assumed that the missing discharge data for the study period (less than 10% of total annual flow for Fraternity

Stream upstream and Graham Woods Stream downstream, approximately 22% for

Lake Alice injection well) did not affect the monthly average discharge from measured data. Monthly average discharge data from Fraternity Stream downstream (Site No.2,

Figure 3-1) and Graham Woods Stream downstream (Site No.7, Figure 3-1) were calculated accompanied by the discharge data from single land use catchments

(Control Catchment, Reclaimed Water Irrigated Catchment and Sports Field Catchment) in Chapter 2. The calculated monthly average discharge for the five catchments was then divided by the area of each catchment respectively, to obtain the data of discharge per unit area for each catchment. The data of discharge per unit area in each catchment was used to extrapolate the discharge for each stream basin.

Baseflow Sampling

The research lasted for one hydrological year from August 2013 to July 2014.

One of the purposes of this study was to examine the spatial changes of N

concentrations in three urban streams in Lake Alice watershed. Baseflow sampling

were conducted in 12 sites (Figure 3-1, Table 3-2), but stormflow sampling was

restricted to the catchments of single land use (described in Chapter 2) due to limited

number of stormflow sampling devices and flow measurement devices as well as

funding. In baseflow sampling, two 100 ml water samples were collected in each water

quality monitoring site except in Diamond Stream downstream (Site No.12) where the

data were obtained from UF Clean Water Campaign dataset in 2014. The sampling frequency was at least twice a month for baseflow NO3-N analysis in dry days, except in

February 2014 when it kept raining around the expected second sampling time of that month. The specific day of sampling was determined one week earlier. Dry days were

defined as being dry for the past 24 hours. Total Kjehldahl N (TKN) was determined

three times in all the sites during the hydrological year, the months were 08/2013,

11/2013, 02/2014, respectively.

Table 3-2. Description of water quality monitoring sites. See Figure 3-1 for exact location of water sampling sites (blue dots on Figure 3-1). Site ID Site name Catchments/Basins 1 Lake Alice well N/A

2 Fraternity Stream upstream Fraternity Stream Upstream Catchment/Fraternity Stream Basin Fraternity Stream 3 downstream Fraternity Stream Basin

4 Sports fields pipe Sports Field Catchment/Graham Woods Stream Catchment/Hume Creek Basin 5 O'Connell Center pipe Graham Woods Stream Catchment/ Hume Creek Basin 6 Retention pond pipe Graham Woods Stream Catchment/ Hume Creek Basin 7 Graham Woods Stream Graham Woods Stream Catchment/ downstream Hume Creek Basin 8 Reitz Union stream Hume Creek Basin

9 Hume creek Hume Creek Basin

10 Diamond family housing pipe Control Catchment/Diamond Stream Basin 11 Diamond Stream upstream Diamond Stream Basin

12 Diamond Stream downstream Diamond Stream Basin

Storage and processing of water samples are described in more detail in Chapter

Data Analysis

The Kruskal-Wallis test (a nonparametric analog of one-way ANOVA) was used

to determine whether differences existed at p=0.05 among median monthly discharge

from all the catchments, also the difference in N concentration (NO3-N) in the water

quality monitoring sites from upstream to downstream in each stream (JMP Pro 10.0,

SAS Institute, Cary, North Carolina). As mentioned earlier, each urban stream has a N

export location to Lake Alice. In order to calculate the N fluxes from the three streams to

Lake Alice, baseflow N concentrations and stormflow N concentrations were needed.

However, only baseflow TN concentrations were retrieved for the three streams, it was assumed that the baseflow TN concentration and the stormflow TN concentration did not change substantially. The assumption was consistent with findings from other studies (Taylor et al., 2005). In all, the N budget in the Lake Alice watershed was calculated based on directly or indirectly measured data, and extrapolated data based

on several assumptions regarding missing data, stormflow N concentrations and

discharge for each stream basin. The related calculation was listed as below:

Discharge per unit area (catchment) = total annual discharge (catchment) / catchment area

Total annual discharge (basin) = representative discharge per unit area (catchment) *basin

area

TN export (stream basin) = CTN *Total annual discharge (basin)

where CTN referred to the TN concentration.

It was noted that discharge per unit area was calculated in all catchments and

the representative discharge per unit area was the value which was close to the

discharge from all basins and used to estimate the total discharge for each stream

basin.

Results and Discussions

Discharge for Each Catchment

The monthly discharge data from September 2013 to July 2014 from all the flow-

monitored stations are displayed in Figure 3-2. The discharge rate from all the single

land-use catchments and Fraternity Stream Upstream Catchment was below 4 L s-1.

Graham Woods Stream downstream exhibited a much higher monthly discharge than the upstream SFC. The huge difference may come from the subsurface flow from

groundwater due to changes in topography as reported by Harvey and Bencala (1993).

SFC was located in an upland position compared with Graham Woods Stream

downstream site (Site No.7). The groundwater in adjacent aquifers could recharge to

the stream and form part of the downstream water as the stream water went through a

steep hillslope from upstream to downstream in Graham Woods Stream. Another

source of discharge to Graham Woods Stream downstream site was the drainage pipes

from nearby dormitories. There were at least 12 drainage pipes emptying into the

stream along the path in the woods although they only discharge during storms and

barely discharge in baseflow. It is also noted that the monthly discharge from both

stations followed a similar pattern with drops in October 2013 and April 2014 when very

little rainfall occurred, and peaks in March 2014 and May 2014 when the monthly

precipitation exceeded 120 mm. The pattern from Lake Alice coincided with that from

Graham Woods Stream downstream and SFC except in May 2014 and June 2014. This

can be attributed to the delay of the inflow since half the precipitation in May 2014

occurred at the end of the month.

Fraternity stream upstream catchment 1 16 120 1 - Reclaimed water irrigated catchment - Sports fields catchment s

14 L Control catchment 100 Graham woods stream downstream basin 12 Lake Alice well rate, 80 10 flow rate, L s flow 8 60 well

6 40 4 20 Alice 2

0 0 Lake Catchment/basin

Months Figure 3-2. Monthly discharge from each catchment and Lake Alice well Discharge per ha was calculated in each catchment (Figure 3-3). The discharge from the Reclaimed Water Irrigated Catchment (RWC) ranked the highest in discharge per ha among all the catchments, significantly different (p<0.05) from the Control

Catchment (CC) (non-irrigated catchment), Sports Field Catchment (SFC) and

Fraternity Stream Upstream Catchment. This may be explained by the high ratio of irrigated area to unirrigated area in such a small catchment and the recharge from groundwater through leaking pipes (Ellis, 2001; Karpf and Krebs, 2004). SFC also has

a large portion of pervious area with regular irrigation practices, but had lower discharge

per ha than RWC. In catchments with lower portion of irrigated area, CC had the lowest

discharge due to non-irrigation. Discharge from Graham Woods Stream Catchment

was significantly greater than that from Fraternity Stream Upstream Catchment

(p<0.05). This may be because of the difference in topography and multiple pipes in

Graham Woods Stream Catchment as in the comparison between Graham Woods

Stream Catchment and SFC.

Figure 3-3. Boxplots representing the distribution of monthly discharge per unit ha in each catchment. Top and bottom edge of each box represents the 75th and 25th percentile, respectively, the line bisecting the box represents the median, points are outliers, and the ends of the whiskers represent the 90th and 10th percentile. (catchments not connected by the same letter are significantly different at the 5% level of probability).

Water Budgets for Lake Alice Watershed

All the N export sites (Site No.3, No.9 and No.12) close to Lake Alice had N concentration data but had no direct discharge records, Outside Basins had neither N concentration data nor discharge data. In order to estimate the N exports from each basin, discharge from each basin had to be extrapolated from the obtained discharge data from single land-use catchments and mixed land-use catchments. In this study, median discharge per ha (in L s-1 ha-1) was 0.14, 0.20, 0.21, 0.31 and 0.36 for CC,

Fraternity Stream Upstream Catchment, SFC, Graham Woods Stream Catchment and

RWC, respectively, during the study period. Graham Woods Stream Catchment was the largest catchment with various land uses, therefore, it is likely to have similar land use composition as other basins. It was assumed that the discharge per ha for Graham

Woods Stream Catchment can represent the discharge from all basins (including

Outside Basins), hence the discharge per ha used for water budget calculation was 0.31

L s-1 ha-1. The estimated water yield from each stream basin is illustrated in Table 3-3.

Table 3-3. Water budget for Lake Alice Annual Precipitation water yield, Area, ha water, m3 Influx/efflux m3 Fraternity stream 9.5 152,000 Influx 93,000 basin Hume creek basin 65.6 1,049,000 Influx 641,000 Diamond stream 156.5 2,504,000 Influx 1,530,000 basin Outside Basins 194.5 3,111,000 Influx 1,900,000 Lake Alice 35.1 562,000 Influx 562,000 Lake Alice well N/A N/A Efflux 1,548,000 Water loss 3,178,000

The Lake Alice injection well exported approximately 1.5 million m3 water to

groundwater from 2013 to 2014, which was approximately the amount Diamond Stream

exported to Lake Alice. However, the amount of outflow was lower than the amount

Outside Basins exported (Table 3-3). The total annual inflow to Lake Alice from the

three major streams and Outside Basins (4.2 million m3) was almost three times as

much as Lake Alice outflow (1.5 million m3). Evaporation from Lake Alice can be a critical factor for the difference. Sacks et al. (1994) compared the evaporation rate from two morphometrically different Florida seepage lakes, and concluded that the shallow lake (Lake Barco in north-central Florida) had a higher evaporation rate than the deeper lake (Lake Five-O in Florida panhandle) in winter and spring but a lower rate in late summer and autumn. Lake Alice is also a shallow lake. If the evaporation rate for Lake

Alice is 151 cm yr-1 based on the data from Sacks et al. (1994), then the annual water loss from evaporation would be 0.53 million m3 (35.1 ha*151 cm), which can explain

part of the difference between the import and export for the lake. The effect from

evaporation in water budget was also reported by Owen (1995). He suggested that

evaporation was the largest factor for water loss for an urban streamside wetland.

Transpiration from vegetation can also account for part of the water loss. Jasechko et al. (2013) suggested that the largest water fluxes in terrestrial land was transpiration which made two-thirds of the total surface water evaportranspiration, and the transpiration rates ranged from less than 0.01 m yr-1 to approximately 1.3 m yr-1.

Another reason for the difference in incoming and outgoing discharge in Lake Alice can

be attributed to vertical seepage to the upper Floridan aquifer, which was commonly

found in lakes that interact with both surface-water and groundwater systems (Motz et

al., 2001) with the rate ranging from 0.12 to 4.27 m yr-1 (Motz, 1998).

The annual water yield from each stream basin accounted for 60% of the total

precipitation water (Table 3-3). The remainder of water may be lost through

evaporation, transpiration (as discussed previously) or become soil water (Brye et al.,

2000; Wilson et al., 2001).

Lake Alice N Budget Estimation

N concentration data from all fluxes to and from Lake Alice (Site 1 for Lake Alice

injection well, Site No.3 for Fraternity Stream Basin, Site No.9 for Hume Creek Basin

and Site No.12 for Diamond Stream Basin) were obtained from the baseflow sampling

described in Materials and Methods except for Diamond Stream downstream site (Site

No.12, Figure 3-1) where the data were taken from UF Clean Water Campaign dataset

in 2014 (http://soils.ifas.ufl.edu/campuswaterquality/). Outside Basins were mainly

composed of residential buildings, academic buildings, landscape, parking lots and so

on, which was similar to Diamond Stream basin, therefore it was assumed that the N

concentrations from Outside Basins were the same as Diamond Stream basin. The N

budget for Lake Alice watershed is summarized in Table 3-4.

Hume creek delivered 49% of the total N loads (mainly NO3-N) to Lake Alice

(Table 3-4) while the entire area of Hume creek basin did not produce the largest of the basin flows measured. The great N loads from Hume Creek Basin came from N loads from catchments in the basin. The spatial changes in N loads from SFC in Graham

Woods Stream upstream to downstream in Hume Creek Basin were illustrated in Table

3-5. Graham Woods Stream delivered most of N to Hume Creek Basin. This can be explained by the large area of sports fields in the upstream. Reitz Union Stream delivered the remainder of N to Hume Creek, part of which came from the Ben Hill

Griffin Stadium, the 12th largest college stadium is located in that basin.

Table 3-4. Summary of N budget in Lake Alice watershed Annual Annual Annual TN Median Median NO3-N TKN loads, kg NO3-N, TKN, Water loads, loads, (% in total mg L-1 mg L-1 yield, m3 kg kg influx) Basins Influx Fraternity 4.63 0.56 92,000 430 52 482 (11%) stream Hume creek 2.80 0.67 641,000 1,796 430 2,225 (49%) Diamond 0.02 0.43 1,530,000 24 658 682 (15%) stream Outside 0.02 0.43 1,900,000 38 817 855 (19%) Basins Lake Alice 0.29 0.28 562,000 163 157 320 (7%) Sum 4,726,000 2,451 2,114 4,565 Efflux Lake Alice 0.07 0.83 1,548,000 115 1,285 1,400 well Sum 1,400 Retention 3,165

Table 3-5. N loads from catchments in Hume Creek Basin Median Total Discharge annual Median Median Annual Area, per area, discharge, NO3-N, TKN, TN Catchment ha L s-1 ha-1 m3 mg L-1 mg L-1 load, kg Sports Field Catchment 6.7 0.21 45,000 12.21a 0.75a 583b Graham Woods Stream Catchment 28.7 0.31 281,000 5.41 0.81 1,748 Hume Creek Basin 65.6 0.31c 641,000 2.8 0.67 2,225 Values with letter a refer to Chapter 2; Value with letter b was different from the calculation based on Chapter 2 because the calculation method in Chapter 2 included stormflow N and baseflow N, but the calculation here only used baseflow N concentration assuming baseflow N does not differ from stormflow N. value with letter c was extrapolated from the discharge data from all catchments

Fraternity Stream contributed 11% of the total N load in Lake Alice watershed

while Diamond Stream contributed 15%. The relatively low contribution was presumably associated with the land uses comprised of the basins, mainly academic buildings, residential dormitories and reclaimed water irrigated landscape. The N yield from those land uses was very likely to be close to the N load reported for CC and RWC (Chapter

2).

The only N load from Lake Alice itself (7%) was precipitation water, which was

912 mg m-2 (calculated from annual precipitation amount and measured N concentration

in precipitation). This value was similar to other studies in Florida (Hendry and Brezonik,

1980; Zhang and Sansalone, 2014), falling in the range of atmospheric N loading to

Harp Lake, Ontario (Nicholls and Cox, 1978) but much less than the wet deposition to

Lake Taihu, China (Luo et al., 2007).

Lake Alice had a very low NO3-N concentration but very high TKN concentration

compared to the three streams (Figure 3-4). This could be the result of the high

plankton production in the lake which transformed (plant uptake) the imported inorganic

N to organic N (Paloheimo and Fulthorpe, 1987). However, chlorophyll a and Secchi

disc depth’s data from Florida LAKEWATCH program, a water quality monitoring

program in Florida (http://lakewatch.ifas.ufl.edu/Lakewatch_County_Data.HTM), suggested low phytoplankton biomass in Lake Alice. This result indicated that N may not be excessive in this system, instead, existing N may be not enough for the reproduction of phytoplankton. The high TKN can be attributed to the detrital materials from decomposition of vegetation which took up NO3-N and reduced the NO3-N

concentration in the water. In addition, Reddy et al. (1996) revealed that high TKN

concentrations could come from wind induced sediment resuspension or by constant

flux due to diffusion. Lake Alice injection well had constant flux to groundwater system

and therefore the sediment could become suspended and make high TKN

concentration.

1 2.5 -

1.5

0.5

TKN concentrations,TKN mg L 0

Water quality monitoring sites

Figure 3-4. TKN concentrations with standard errors in all the water quality monitoring sites

Results in Table 3-4 showed that about 60% of the imported N from the three

streams (2300 kg) was retained or lost in the Lake Alice watershed, suggesting Lake

Alice was functioning as a net sink. The fate of the retained/lost N is likely to be

associated with three processes: denitrification, sedimentation and uptake by aquatic

plants (Jansson et al., 1994; Saunders and Kalff, 2001). Saunders and Kalff (2001) compared the total N retention and loading data from 23 wetlands, 23 lakes and 5 rivers, and concluded that wetlands retained the highest N loading and denitrification was the primary mechanism of N retention in his study. This finding was consistent with the conclusion from Jansson et al. (1994) after they studied different types of wetlands

in Sweden. However, Braskerud (2002) investigated four constructed wetlands aged 3 to 7 years in Norway and suggested N retention decreased as the wetlands aged, presumably because trapped organic N was converted to inorganic forms that were exported from the wetlands. Therefore he concluded that sedimentation of N in organic particles was the main retention process in his study. Plant uptake was also considered as an important process for N retention in lakes. Reddy (1983) used labeled 15N to

15 + 15 − differentiate preferential uptake of NH4 and NO3 for different kinds of vascular

aquatic macrophytes. His research showed that 34 to 40% of the added inorganic 15N

15 + 15 − ( NH4 + NO3 ) was removed through plant uptake, the rest presumably lost through

NH3 volatilization and nitrification-denitrification processes.

Spatial Changes in NO3-N Concentrations

Figure 3-5. Boxplots representing the distribution of NO3-N concentration in Fraternity Stream. Top and bottom edge of each box represents the 75th and 25th percentile, respectively, the line bisecting the box represents the median, points are outliers, and the ends of the whiskers represent the 90th and 10th percentile. (Sampling sites not connected by the same letter are significantly different at the 5% level of probability)

The changes in N fluxes in Lake Alice watershed can also be observed based on

the spatial changes in NO3-N concentrations. NO3-N concentrations were recorded in

both upstream and downstream in each major stream. There was no significant

difference in NO3-N concentration between Fraternity Stream upstream and Fraternity

Stream downstream (Figure 3-5). This might be because that this stream has shorter

total length compared with the other streams, and also it has fewer contributing pipes

than other streams. On the contrary, the NO3-N concentrations were significantly

different among all the sampling sites in Graham Woods Stream. SFC in the upstream

of Graham Woods Stream delivered a very high NO3-N concentration with an annual

median of 12.21 mg L-1 (Chapter 2), however, the concentration was reduced

significantly as it approached the downstream sampling sites (Figure 3-6). This

reduction in NO3-N concentration might be attributed to dilution from the seepage or

drainage into the stream along its path considering the annual discharge from Graham

woods Stream downstream was approximately three to four times of the upstream SFC

(Figure 3-2, Table 3-5). In addition, flow in Graham Woods Stream passed through two

small retention ponds before it reached the discharge monitoring station (Site No.7,

Figure 3-1). Retention ponds may help to reduce NO3-N concentration by vegetation uptake although the efficiency of nitrate removal was reported to be very low (Hsieh and

Davis, 2005; Silva et al., 1995). After the stream water passed the discharge monitoring station, it merged with the branch from the urban academic area (sampling site No.8,

Figure 3-1) where reclaimed water irrigation was regularly practiced. When the two branches combined in Hume creek, the NO3-N concentrations were reduced even more

(Figure 3-6). In the Diamond Stream, the NO3-N concentrations showed significant difference between the upstream (Site No.11, Figure 3-1) to the downstream (Site

No.12, Figure 3-1) (Figure 3-7), which is largely attributed to dilution because the downstream discharge was much higher than upstream discharge (visual observation

based on the changes in depth and width of the stream). It is noted that the upstream

NO3-N concentrations in the Diamond Stream were approximately 10 times the NO3-N

concentrations in the downstream portion of the stream.

Figure 3-6. Boxplots representing the distribution of NO3-N concentration in Hume Creek. Top and bottom edge of each box represents the 75th and 25th percentile, respectively, the line bisecting the box represents the median, points are outliers, and the ends of the whiskers represent the 90th and 10th percentile. (Sampling sites not connected by the same letter are significantly different at the 5% level of probability)

The types of spatial changes in NO3-N concentrations in Hume creek and

Diamond Stream were also demonstrated in several studies as an indication or

explanation for N mass changes. For example, Hager and Schemel (1992) found that

the dissolved inorganic nitrogen (DIN) in the upstream increased by four times after the

water passed through agricultural drains and a wastewater treatment plant in the

downstream. After they calculated the change in N load, they concluded that the

agricultural drains and the wastewater treatment plant contributed 70% DIN to the

Northern San Francisco Bay. Kemp and Dodds (2001) also observed a huge increase

in NO3-N concentrations in the downstream sampling well and in stream branches from

surrounding croplands compared with the upland watershed characterized with tallgrass prairies, indicating there was an increase in N load in the downstream.

Figure 3-7. Boxplots representing the distribution of NO3-N concentration in Diamond Stream. Top and bottom edge of each box represents the 75th and 25th percentile, respectively, the line bisecting the box represents the median, points are outliers, and the ends of the whiskers represent the 90th and 10th percentile. (Sampling sites not connected by the same letter are significantly different at the 5% level of probability)

TKN Concentrations

TKN concentrations (sum of organic-N plus ammonium-N) were estimated from

limited sampling times (3 times) during the hydrological year. Unlike the wide range in

NO3-N concentrations among all water quality monitoring sites, TKN concentrations

ranged from 0.25 to 1.34 mg L-1 (Figure 3-4). Fraternity Stream upstream had the

lowest TKN concentration and increased in the downstream. The increase in TKN

concentration in downstream also occurred in Diamond Stream and Hume Creek if the

SFC (Sampling site No.4, Table 3-2) was excluded, indicating that surface

runoff/erosion processes played an important role in TKN addition at the steep hill sites

(Lake Alice was the lowest point on UF campus). Regeneration of N, by plant decay

within the stream channel, can also result in export of dissolved organic N and particulate N from the catchments (Cooper and Cooke, 1984).

Conclusions

Urban stream ecosystem functions provide ecosystem services. For example,

biomass community respiration transforms organic matter to CO2, an essential service

for streams receiving effluent from wastewater treatment; and removal of water-column nutrients from point and non-point sources by plant uptake may improve water quality in downstream waterbodies. These stream services are essential to human-dominated urban ecosystems. Understanding sources of nutrient additions to a stream can help explore nutrient management practices that could reduce nutrient loading to a stream.

This paper investigated the N exports from three small urban streams to Lake

Alice and established a N budget for Lake Alice watershed. It was determined that

Hume Creek Basin contributed the greatest N load to Lake Alice (60% of the total annual N import) compared to other streams, presumably because it had a large coverage of sports fields including the Ben Hill Griffin Stadium, the 12th largest college

stadium. This result was consistent with the first hypothesis that the major N source in

Lake Alice Watershed was from sports fields, and streams delivering the flow from

sports field produced the greatest N load. The NO3-N concentration decreased as the flow approached to the downstream in Hume Creek Basin, this can be associated with the dilution from subsurface groundwater and other sources of runoff. This finding agreed with the second hypothesis that NO3-N concentrations declined from upstream to downstream in urban streams in Lake Alice watershed. Most of the N imported to

Lake Alice from the three streams remained in the lake or the surrounding wetland with

only 40% discharged to the groundwater system. This result suggested that Lake Alice

was a sink for N. The N may be stored in phytoplankton, wetland plants or sediments,

but the amount of N was not excessive to cause eutrophication, instead, it may be

limiting the reproduction of biomass. The internal N pool can be released by

resuspension, regeneration of N by plant decay and other paths, which can potentially

provide a continuous N source for eutrophication with sufficient phosphorus. This

research identified a need to carefully manage sports fields because the intense

fertilization in sports fields could possibly be the greatest source of N in Lake Alice

watershed and create a large N pool in the watershed. More studies involving Best

Management Practices (BMPs) are recommended to reduce the N runoff from sports

fields.

The water budget and the spatial changes in N concentrations and N loads in

Lake Alice demonstrated that dilution played an important role in reduction of NO3-N concentrations and N yield in downstream, which met the third hypothesis that spatial changes in NO3-N concentrations were attributed to hydrological changes in Lake Alice

watershed. Processes such as denitrification/nitrification and plant uptake are also very

important in N reduction during transport, but more studies needed to understand

the N transformation during transport.

CHAPTER 4 URBAN STORMFLOW NITRATE-NITROGEN CONCENTRATIONS AND LOADS DETERMINED BY HIGH RESOLUTION IN SITU NITRATE SENSOR AND AUTOSAMPLERS

Introduction

Urban stormwater systems are designed to efficiently transport untreated surface runoff from urbanized areas. Management of urban stormwater runoff was one of the major topics included in a national assessment of urban research needs (Heaney et al.,

1999). Nonpoint pollution from urban stormwater runoff has been identified as one of the major causes of water quality impairment of receiving waters in and near urbanized areas (Carpenter et al., 1998). Excessive nutrients such as nitrogen (N) and phosphorus (P) from stormwater runoff can result in eutrophication of surface waterbodies (Conley et al., 2009; Ryther and Dunstan, 1971). A major challenge for stormwater managers is determining and tracking concentrations of pollutants in storm water so decisions can be made quickly about controlling the pollutant loads. Part of the challenge is determining the best sampling technique to meet the goals in tracking pollutant loads.

Manual sampling and mechanical sampling are typically the traditional approaches to collect samples for urban stormwater runoff. Manual sampling consists of “grab samples” collected at a pre-determined interval, and is considered to be a preferred sampling technique over mechanical sampling in first flush characterization studies (Line et al., 1997). Mechanical sampling can be accomplished by a sampler that automatically collects samples on a preset interval. The advantage of mechanical sampling is to collect samples in severe weather, especially when the weather may post hazards to the safety of personnel in the field. Mechanical sampling can also be

adopted under different strategies, for example, time composite sampling or flow- proportional sampling. However, these methods consume a lot of time and labor, and the number of samples can be limited based on funding levels for costs of the laboratory analyses. In addition, sample collection, handling, and storage can introduce contamination and analysis errors.

An alternate sampling and analytical method is the deployment of a high- resolution in situ water quality sensor. The idea of the sensor is to calculate the nitrate concentrations based on the absorption of ultraviolet (UV) light at wavelengths between

200 – 300 nm (Johnson and Coletti, 2002). This approach avoids the chemical reduction reactions in the laboratory which may involve the use of toxic metals such as cadmium, and provides a clean and efficient alternative to laboratory analyses (Guillard and Kopp, 2004).

The first deployment of in situ nitrate sensors started in the seawater research

(Chang et al., 2004; Finch et al., 1998; Johnson and Coletti, 2002) and has been utilized in freshwater systems and wastewater systems in recent years (Capelo et al., 2007;

Drolc and Vrtovšek, 2010; Gutierrez et al., 2010; Hensley et al., 2014; Weiss et al.,

2008). Regardless of the good performance in seawater and freshwater systems, few

studies have been reported for the use of nitrate sensors in stormwater systems. Even

though the in situ sensors have many advantages over the conventional measurement,

the in situ sensors can have problems. Nitrate absorbs significantly at wavelengths up

to 230 nm, while some interferences in the seawater may be introduced by bromide or

dissolved organic matter and carbonate which exhibit close band absorption with nitrate

(Johnson and Coletti, 2002). Thomas et al. (2010) suggested that shortening the path

length of the light beam in the sensor could overcome the attenuation caused by the dissolved organic carbon, but it resulted in reduced sensitivity as well (Paul and Meyer,

2001). Currently, examples of in situ nitrate sensors are the Submersible Ultraviolet

Nitrate Analyzer (SUNA) from Satlantic Inc. (Halifax, Canada), Digital NitraVis

UV/Visible sensor from YSI Inc. (Yellow Springs, Ohio), and Ion-Selective Electrode sensor from In-Situ Inc. (Bingen, Washington).

Previous chapters determined N fluxes from catchments of various land uses and basins in Lake Alice watershed, and identified sports fields as the major N contributor to

Lake Alice watershed. The N load from Sports Field Catchment was calculated based on the event mean concentrations obtained by ISCO 6700 auto sampler (Teledyne

Technologies Inc., Lincoln, Nebraska) and regular baseflow grab sampling. Both sampling approaches were conventional, time-consuming and costly, also were limited by the number of samples due to various reasons as mentioned earlier. This study introduced an in situ nitrate sensor to monitor nitrate dynamics in stormwater in comparison with conventional sampling devices-autosamplers. The objectives of this study were 1. to evaluate the performance of in situ nitrate sensor (SUNA) in an urban stormwater system; 2. to compare the nitrate-N (NO3-N) mass in stormflow determined

from SUNA and autosampler concentration data and flow data. My hypothesis was that

SUNA could capture the changes in NO3-N concentrations better than autosamplers in

storms. If the continuous nitrate sensor works effectively in stormwater systems, then

pollutant control specialists will be able to continually monitor changes in NO3-N

concentrations in baseflow and in storms. The sensor can also help land managers

adopt best management practices to reduce N losses to water bodies.

Materials and Methods

Study Area

Gainesville is located in North Central peninsular Florida with a humid subtropical

climate. The seasonal and annual precipitation was described in more detail in Chapter

The University of Florida (UF) is one of the largest universities in the nation. It

has various land uses (described in Chapter 2) which makes the campus similar to a

small urban “city” and function similar to a city. Two urban catchments used in this study

are located at University of Florida (UF) main campus. In this study, the urban reclaimed

water irrigation catchment (RWC) is in the center east part of the campus where there is

widespread landscaping around buildings with an approximate area of 1.5 hectare (ha)

with 76% pervious surface. The landscape irrigation was scheduled for application twice

a week. The irrigation is made from reclaimed water from the campus wastewater

treatment facility. The urban sports field catchment (SFC) with intense fertilization is in the northern part of the campus, operated by University Athletic Association. The SFC consists of the football practice field and the baseball field with an approximate area of

6.7 ha, the pervious surface takes up to 95% in the entire area (Table 4-1). All the runoff in the two catchments eventually drains to Lake Alice, a large retention pond with an open area of 35 ha on the campus (Figure 4-1). Water in Lake Alice overflows to two injection wells connected with the groundwater system.

Figure 4-1. Sports Field Catchment and Reclaimed Water Irrigated Catchment in Lake Alice watershed

Table 4-1. Pervious/impervious ratio in each catchment in Lake Alice watershed Land use1 RWC SFC Impervious 24.5% 5.4% Pervious 75.5% 94.6% 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

Flow Measurement

The details for flow measurement are described in Chapter 2. Briefly, flow was

determined from V-notch weirs and pressure transducers WL16 (Xylem Inc.,

Sacramento, California) measuring at 5 min intervals. The discharge was calculated

based on the stage-discharge rating curve except that the full pipe discharge in RWC was calculated based on Manning’s equation.

Autosampler Stormflow Sampling

Stormwater samples were collected by the ISCO 6700 auto samplers (Teledyne

Technologies Inc., Lincoln, Nebraska) during 16 storms from February 2014 to August

2014. The sampling strategy was described in detail in Chapter 2. Briefly, time- triggered composite samples were collected during the storm. The frequency of sampling differed depending on the duration of the storm.

Two tipping bucket rain gauges (Bluesiren Inc., Melbourne, FL) which tipped for

every 0.254 mm of rainfall were placed on the top of a utility building on campus, central

to the two instrumented catchments and used for measurement of precipitation. Three

5-quart plastic buckets were placed near the study areas to catch the rainfall water for

NO3-N analysis.

Details on water sample storage and analysis are described in Chapter 2. All the

samples were taken back to the lab within 12 hours after the storm. Samples were

filtered and acidified before being sent to UF Analytical Research Lab (ARL).

SUNA Stormflow Sampling

Two SUNAs (SUNA 238 and SUNA 239) were used in this study. Prior to field

deployment the SUNA units were tested in February 2013 in lab evaluations using

standard NO3-N solutions (1 ppm, 2 ppm, 3 ppm, 4 ppm, 5 ppm, 10 ppm, 15 ppm, 20

ppm, 25 ppm) and double deionized water as control to determine the accuracy. The

results indicated that at least 99.9% of the variation in the measurements from SUNAs

could be explained by standard NO3-N solutions. Later, two SUNAs were placed side by

side in the field in SFC (the pipe in RWC was too small to place two SUNAs) to evaluate their performance in the field. They were separately placed in padded protective PVC tubes to provide physical protection of the SUNA unit (Figure 4-2). The tube had a number of holes to facilitate the water exchange through the tube. The lenses of both

SUNAs were exposed to the water environment by a slit in the PVC tube. The tubes were anchored to the stream bed (SFC) or stormwater pipe (RWC) to keep them stable against water flow forces. The two SUNAs were set up to measure NO3-N 10 times

every 5 minutes. Field results revealed that measurements from SUNA 238 and SUNA

239 exhibited similar patterns under several small changes in baseflow and a large drop

in stormflow (Figure 4-3). The coefficient of determination (R2) between SUNA 238 and

SUNA 239 was 99.4% (p<0.0001) through simple linear regression. In addition, grab

samples of baseflow were taken in the field (both RWC and SFC) to evaluate the

SUNA’s performance from February 2013 to May 2014. Samples were stored on ice

once collected, and then filtered and acidified before sent to UF Analytical Research

Laboratory (ARL) at University of Florida. There was a high linear correlation between

grab sample results and measurements from SUNAs (R2=99%, p<0.01) (Figure 4-4).

The difference between the values from ARL and SUNAs was within the accuracy range

(15%) reported by Satlantic Inc. (Halifax, Canada). Before the storm studies, the same

lab test as described above, was performed in February 2014 to determine the accuracy

of both SUNAs. Similar results were achieved with a R2 value above 99.9%. All SUNA

readings were converted to the standard NO3-N concentrations based on the

relationship established in lab test for data analysis. These preliminary laboratory and

field studies proved that both SUNA samplers provided similar results and that the

SUNA units provided accurate results similar to lab analyses from grab samples.

Following these lab tests the SUNA units were deployed in the field in studies to compare SUNA and autosampler for monitoring NO3-N concentrations in stormwater.

Figure 4-2. The look for SUNA PVC tube

18 SUNA238 200

1 16 SUNA239 14 Flow rate 150 mg L -

12 1 -

10 100 L s

8 rate,

6 50

4 Flow N concentration,N - 2 0

NO3 0 0 1000 2000 3000 4000 -2 -50 Time, min Figure 4-3. SUNA 238 and SUNA 239 in baseflow and stormflow in Sports Field Catchment (SFC)

20 SUNA,

- 15

mg L 10 measured

N 5 - 3

NO 0 0 5 10 15 20 25 -1 Grab sample NO3-N analyzed by ARL, mg L

Figure 4-4. Grab samples vs SUNA measurements in Sports Field Catchment (SFC) and Reclaimed Water Irrigated Catchement (RWC) from February 2014 to May 2015. Grab samples were taken at or close to the time when SUNA was reading. A linear regression model (NO3-N measured by SUNA = 0.97 * Grab sample NO3-N analyzed by ARL+0.06) was used to describe the relationship between grab samples and SUNA measurements (R2=0.99, p<0.0001).

In all the storms, SUNAs and autosamplers were set up to start sampling at the

same time, therefore it is reasonably assumed that the samples simultaneously

measured by SUNAs and collected by autosamplers reflected the same water

chemistry. The descriptions of measurements in each storm together with autosampler

samples are displayed in Table 4-2. Detailed descriptions of the storms are displayed in

Table 4-3.

Table 4-2. Descriptions of measurements and samples of SUNA and autosamplers for storm events Number of samples No. of Sampling SUNA Precipitation, duration, missing Date mm min Autosampler SUNA values Land use1= RWC 4/8/14 16.0 570 20 110 5 5/25/14 29.0 100 21 18 3 5/29/14 4.6 55 12 12 0 5/30/14 21.3 160 30 33 0 Land use= SFC 3/17/14 46.0 195 40 70 2 5/1/14 1.5 85 18 18 0 5/29/14 4.6 55 12 12 0 5/30/14 21.3 170 30 35 0 8/29/14 19.8 385 57 78 0 8/30/14 41.7 355 51 72 0 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

Table 4-3. Descriptions of storm events recorded by both SUNA and autosamplers in the two catchments. NO3-N concentrations were obtained from rain samples captured by buckets Previous Precipitation Dry Imax NO3-N Date (mm) hours(hr) (mm/hr) (mg/l) 3/17/14 46.0 238 76.2 0.16 4/8/14 16.0 226 76.2 0.18 5/1/14 1.5 23 30.5 0.32 5/29/14 4.6 24 76.2 0.90 5/30/14 21.3 24 121.9 0.21 8/29/14 19.8 120 61.0 0.26 8/30/14 41.7 12 91.4 0.07 Median 19.8 24.0 76.2 0.2 Mean 21.6 95.3 76.2 0.3 SEM 6.4 37.9 10.5 0.1

Data Analysis

NO3-N concentrations from SUNAs were averaged based on the frequency of

sampling from the autosampler. For example, if the composite sample from

autosamplers was comprised of 3 samples covering 15 min period, then the NO3-N concentrations from SUNA in that 15 min period would be averaged to compare with the result from autosamplers. The t test was used to determine whether differences exist at p=0.05 between the averaged SUNA NO3-N concentrations and the composite samples

from autosamplers in each storm (JMP Pro 10.0, SAS Institute, Cary, North Carolina).

Event mean concentrations (EMCs) were used to characterize nutrient

concentrations during a storm event. The EMC is a flow-weighted concentration and

calculated as:

EMC (mg/l)= ∑ 푄푄푄푄 ∑ 푄푄 where, Qi is the time variable flow and Ci is the time variable concentration. The

The NO3-N mass for each storm is calculated as

MNO3-N (mg) = ×

퐸�퐸 푇푇푇�푇 푒푒푒푒Results� �푑푑푑ℎ푎푎푎 and� Discussion

Comparisons Of Autosampler/SUNA NO3-N Concentrations in Storm Events

Significant difference existed between the means of NO3-N concentration from

SUNA and the autosampler in RWC (Table 4-4; p<0.01). SUNA showed a higher NO3-

N concentration than the autosampler. The difference between the measurements of

SUNA and samples from the autosampler was most evident in the event on May 29th when the total precipitation was smallest (4.6 mm) among all the storms in RWC (Figure

4-5). The difference between sensor measurements and laboratory analysis was also reported by Rusjan et al. (2008) where the in situ nitrate sensor produced slightly higher nitrate concentration than laboratory analysis, but in that study the nitrate determined by the sensor was not using UV analysis like SUNA. Those researchers used Hydrolab

MiniSonde 4a water quality multi-parameter data-sonde (Campbell Scientific, Inc.,

Edmonton, Canada) instead. It is noted that some in situ sensors have a global calibration, and recalibration with local reference samples can significantly improve the performance (trueness and precision) for quantitative measurements (Langergraber et al., 2003). In this study, the local reference samples were grab samples in baseflow condition, which may have less interference than stormflow condition (e.g. less particulate matter to interfere with SUNA). Therefore, it is likely that the interference gradient in stormwater such as turbidity and particulate matter, may affect the absorption of NO3-N by UV light, therefore affected the NO3-N concentration calibrated

in SUNA, and this interference gradient had greater effect on RWC than SFC because

NO3-N concentrations in RWC was much lower than SFC. This explained the reason why there was no significant difference between the means of NO3-N concentration from SUNA and the autosampler in SFC. In addition, it was dry for several days and there was no other substantial hydrological changes before May 29th, so the extremely

low NO3-N concentrations from the autosampler was very likely to be outliers.

Table 4-4. Comparison of NO3-N concentrations from SUNA and the autosampler in RWC and SFC SUNA Autosamplers Land No. of Std Std use1 pairs Mean dev. Mean dev. p RWC 25 0.86 0.34 0.50 0.33 <0.01* SFC 69 3.92 3.78 3.88 3.90 0.37 * Significance level p=0.01 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

2 50 a. 4/8/2014 SUNA

40 - 1 1.5 Autosampler L s

mg L - Flow rate 1 20 N, - 3 10

NO 0.5 0 Discharge, 0 -10 0 100 200 300 400 500 600 2 70 b. 5/25/2014

1 60 - 1 1.5

50 L s

mg L - 40 1

N, 30 - 3 20

NO 0.5 10 Discharge, 0 0 0 20 40 60 80 100 120 2 25 c. 5/29/2014

1 20 1.5 L s 15 mg L - 1 10 N, - 3 5

NO 0.5 0 Discharge, 0 -5 0 10 20 30 40 50 60 70 80 1.5 80

d. 5/30/2014 1

- 1

- 60 L s 1 40 mg L

- 20 3 0.5 NO

0 Discharge,

0 -20 0 50 100 150 200 Time, min Figure 4-5. Changes in NO3-N monitored by SUNA and autosampler in Reclaimed Water Irrigated Catchment (RWC).

100

There may be other reasons for the difference between SUNAs and

autosamplers in low NO3-N concentration environment (RWC). NO3-N concentrations

in SFC ranged from 0.17 to 13.19 mg/l and from 0.16 to 1.41 mg/l in RWC in SUNAs.

Correspondingly, the detection range for SUNAs was from 0.007 to 28 mg/l with the

accuracy of +/- 0.028 mg/l or 10% of reading whichever is greater. The detection limit

for NO3-N in the UF Analytical Research Lab (ARL) was 0.148 mg/l. Consequently, it is

likely that SUNAs were more sensitive to the changes of NO3-N concentrations in

storms regardless of how high or low the concentrations were in RWC than the methods

used by the ARL. In addition, sampling errors can be increased in composite samples

from autosamplers (Harmel et al., 2003). The errors are more evident in narrow storm

water pipe systems (RWC) where the force from intensive flow has a more significant

impact on the sampling equipment than in the open stream channel (SFC).

There was no significant difference (Table 4-4) between the means of NO3-N

from SUNA and the autosampler in SFC (p=0.37). The changes tracked by the autosampler coincided with SUNA very well in SFC except for a slight difference in the beginning of storms (Figure 4-6). The slight difference was because the autosampler

had at least 3 subsamples in one composite sample bottle which possibly included both

baseflow (high NO3-N) and stormflow (low NO3-N). Meanwhile SUNA only read one

discrete sample at a time when the autosampler collected one subsample.

101

20 SUNA 150 1

a. 3/17/2014 -

- Autosampler L s 15 Flow rate 100 mg L 10 N, - 3 50 5 NO NDischarge, 0 0 0 100 200 300 400

14 b. 5/1/2014 6

12 1 1 -

- 5 10

4 L s

mg L 8 3

N, 6 - 3 4 2 NO 2 1 Discharge, 0 0 0 20 40 60 80 100

12 25

c. 5/29/2014 -

1 10 20 - L s 8 15 mg L 6 10 N, - 3 4 5 NO

2 0 Discharge, 0 -5 0 20 40 60 80 14 50

12 - 1 - d. 5/30/2014 40

10 L s

30 mg L 8

- 6 20 3 4

NO 10 2 Discharge, 0 0 0 50 100 150 200 Time, min Figure 4-6. Changes in NO3-N monitored by SUNA and autosampler in Sports Field Catchment (SFC).

102

14 e. 8/29/2014 80

12 - 1 - 60 10 L s

mg L 8 40

N, 6 - 3 4 20 NO

2 Discharge, 0 0 0 50 100 150 200 250 300 350

12 400 f. 8/30/2014

1 1 - - 10 300 L s

8 200 mg L 6 N,

- 100 3 4

NO 2 0 Discharge, 0 -100 0 100 200 300 400 Time, min

Figure 4-6 Continued

In addition to the proximity to laboratory results from autosamplers, SUNAs also gave a faster response to the changes of flow in the entire storms. For example, in the storm of May 30th which lasted about 180 minutes (Figure 4-5 d), SUNAs captured the sharp decrease in NO3-N concentrations in response to the peak of discharge changes while the autosampler responded later in both RWC and SFC. Still this was mainly because SUNAs were set up to measure discrete samples more frequently than the composite samples collected by autosamplers as described above. SUNAs proved to be better in characterization of storms as they provided more information about storms.

SUNAs created smooth curves as discharge curves through the storms but autosamplers collecting composite samples created stepwise curves, indicating SUNAs can have simultaneous response to the changes of discharge during the entire storm

103

events but autosamplers with composite sampling cannot. Although autosamplers can

also collect discrete samples, the collected samples may still miss the response of NO3-

N to the peak of the hydrograph. SUNAs cannot guarantee the collection of the extreme changes of NO3-N to the changes of discharge, but they can capture a closer

image for NO3-N changes than autosamplers as they can be set up to measure NO3-N

at very high resolution (up to 1 second as reported by Satlantic Inc.) during the entire

storm. This advantage is more evident as the duration of storms got longer in both sites

(Figure 4-5 a. and Figure 4-6 f). In addition, in most research cases, the number of

storm water samples collected manually or mechanically is constrained by funding

levels and the capacity of the equipment (e.g., most autosamplers only have 24 bottles),

hence only a limited number of samples can be collected and analyzed during a storm.

Even authors of a number of storm studies admitted that they only caught 80% to 90%

of the storm (Horowitz, 1995; Horowitz, 2009).

However, the drawback for SUNAs’ performance in stormwater is missing data

during the beginning of storms. In Figure 4-5 a, Figure 4-5 b and Figure 4-6 a, there was

a period when SUNA obtained no data. This might be associated with the first flush

when high initial particulate matter concentration came in the early portion of a storm

event with a subsequent rapid concentration decline (Bertrand-Krajewski et al., 1998;

Sansalone and Cristina, 2004). The particulate matter could have blocked the

ultraviolet light from SUNA therefore kept SUNA from reading NO3-N concentration.

104

1.6 1

1 Land use = RWC - 1.4 Autosampler 1.2 SUNA 1 0.8 0.6 0.4

N concentration,N mg L 0.2 - 3 0 NO 4/08/2014 5/25/2014 5/29/2014 5/30/2014

10 1 Land use= SFC 9 8 7 6 5 4 3 2 N concentration,N mg L -

- 1 3 0 NO 3/17/2014 5/01/2014 5/29/2014 5/30/2014 8/29/2014 8/30/2014 Event date

Figure 4-7. Event mean concentrations (NO3-N) determined by the autosampler and SUNA 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

Event Mean Concentrations from Autosamplers and SUNAs

Event mean concentrations (EMCs) for the RWC catchment, determined by the

autosampler were approximately half of the EMCs determined by the SUNA (Figure 4-

7). The difference was 7 times lower for the event of 5/29/2014 in RWC. On the

contrary, the EMCs determined by both methods were close to each other in SFC. The

results were consistent with the comparison between the mean SUNA NO3-N

105

concentrations and mean autosampler NO3-N concentrations (Table 4-4). Based on the fact that SUNAs provided more (continuous) information for each storm, and there was no difference between SUNA measurements and autosamplers in SFC, EMCs determined by SUNAs are more likely to reflect the changing water chemistry in SFC than the autosampler. Overall, the EMCs determined by SUNAs were consistently higher than by autosamplers in RWC (Figure 4-7), which conformed well to the pattern observed in unit concentration comparison discussed above. EMCs determined by

SUNAs were higher than by autosamplers in storm events with longer duration and great precipitation in SFC, while lower in storm events which had relatively short duration and small precipitation. The difference in EMCs can be associated with several factors such as disturbance from dissolved organic matter or particulate matter.

Event NO3-N Mass

The event NO3-N mass was calculated based on total event discharge and EMCs from autosamplers and SUNAs (Figure 4-8). The order of magnitude was the same among events in EMCs from SFC calculated by SUNAs and the autosampler. There were differences in N mass among sampling dates. For example there was at least one

th order of magnitude difference between events in NO3-N mass in SFC (May 29 vs May

th 30 ). The huge difference in NO3-N mass was associated with the duration of storms and the number of dry days before the event (Charbeneau and Barrett, 1998; Kim et al.,

2006; Whipple Jr et al., 1977). In the same storm when samples were collected in both

th sites (May 30 ), NO3-N mass was about 50 times greater in SFC than RWC determined by SUNA, and the difference in the order of magnitude was even greater when comparing the NO3-N mass determined by the autosampler in SFC with that in RWC.

106

The difference in mass between RWC and SFC was related to different management

practices in each land use (details described in Chapter 2). Overall, the total NO3-N mass for storm events ranged from 3 to 44 g in RWC based on autosampler’s results, and from 21 to 75 g based on SUNA’s results. Similarly, it ranged from 70 to 1930 g in

SFC based on autosampler’s results, and from 60 to 2030 g based on SUNA’s results. It showed that the difference in mass determined by both methods could be extremely large in some storms, suggesting accurate NO3-N concentration is very important in

NO3-N mass determination.

107

Land use1= RWC 80 75 Autosampler 72 70 SUNA

50 44 39 40 36 N g N mass, - 3 30 25

NO 21 20

10 3 0 4/08/2014 5/25/2014 5/29/2014 5/30/2014

2500 Land use= SFC 2030 2000 1930 1670 1500

N g N mass, 960 -

10003 NO 310 500 360 240 280 100 90 70 60 0 3/17/2014 5/01/2014 5/29/2014 5/30/2014 8/29/2014 8/30/2014 Event date

Figure 4-8. Event NO3-N mass determined by autosamplers and SUNA 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC).

Conclusions

Measurements from SUNAs were compared with lab-analyzed composite samples from autosamplers in two catchments with different land uses in this study.

Generally both SUNAs and autosamplers showed similar trends in storms where

108

concentration decreased in high discharge and increased in low discharge. However,

the difference between the means of NO3-N concentrations determined by the two

methods was significant (p<0.01) in RWC. The SUNA showed a higher NO3-N concentration than the results from the autosampler. This might be associated with the potential tremendous changes in hydrology in narrow stormwater pipes which affected the sample collection of the autosampler. Another possible explanation was that the minimum detection limit (MDL) from UF Analytical Research Lab (ARL) was higher than the MDL for the SUNAs. RWC had a much lower concentration (mean grab sample

-1 NO3-N concentration was 1.19 mg L ) than SFC (mean grab sample NO3-N

concentration was 12.21 mg L-1) as described in Chapter 2, which was closer to the

-1 detection limit of ARL (0.148 mg L for NO3-N). The event mean concentrations of

NO3-N determined by both methods were consistent with the mean concentrations

among all events with higher NO3-N in SFC than RWC. SUNA showed a higher EMC

than the autosampler in RWC but had no difference with the autosampler in SFC. The

difference in the NO3-N concentrations determined by both methods affected the

calculation of event NO3-N mass, suggesting difference in NO3-N concentrations caused by different methods could introduce huge difference in load calculation (at least twice or larger).

SUNAs proved to be close to the laboratory results in both laboratory and field tests, also they had no difference from the results from the autosampler (e.g. in SFC).

Overall, SUNAs showed several advantages over autosamplers. SUNAs can depict the changes of NO3-N better than autosamplers with smooth curves in response to the changes of discharge. They can provide high resolution data through the storms but

109

autosamplers may be limited to the number of bottles. Also autosampler composite

samples can introduce cross contamination while SUNAs measured the NO3-N

concentrations immediately without any physical processing; therefore, the sampling

errors can be minimized. Moreover, the accuracy for delivered volume of samples in

autosampler ISCO 6700 was +/- 10ml or +/- 10%, whichever is greater. This can cause

the difference between the concentrations from composite samples and that from real

water environment. In addition to the possible manual errors in handling and

transporting, SUNAs become a good alternative to manual and mechanical sampling.

Even though in this study SUNAs displayed their advantages over composite

sampling, SUNAs may have no readings during the beginning of storms due to first

flush phenomenon. An option to reduce the interference caused by dissolved organic

matter and other chemicals is to shorten the wavelength of SUNA, but this can cause

the reduction in the sensitivity to NO3-N changes. Meanwhile, in order to evaluate the performance of SUNAs compared with autosamplers, comparisons between SUNAs and different sampling strategies of autosamplers are needed such as flow triggered discrete sampling. Also laboratories with a lower detection limit are in need to compare lab results with SUNA results.

110

CHAPTER 5 CHARACTERIZATION OF NITRATE NITROGEN CONCENTRATION AND DISCHARGE RELATIONSHIP IN A SMALL URBAN CATCHMENT MONITORED BY HIGH RESOLUTION NITRATE SENSOR

Introduction

Nitrogen (N) is one of the necessary elements every living organism needs for growth and reproduction. The amount of reactive N in the ecosystem has been doubled

compared to preindustrial times due to human activities (Galloway et al., 1996).

Excessive N can cause a number of environmental problems such as eutrophication

(Smith et al., 1999), global warming (Smith et al., 1997) and reduced biodiversity

(Nordin et al., 2005; Xiankai et al., 2008). In addition, it can pose hazards to human

health (Kaye et al., 2006; Miller, 1971; Oenema et al., 2003).

A large amount of attention has been placed on N from stormwater runoff in

several studies (Kim et al., 2003; Lee and Bang, 2000; Mallin et al., 2009; Taylor et al.,

2005). As an important pollutant from a non-point source, nitrate-N (NO3-N) often has

not been treated in the storwater before it reaches surface waterbodies. The

subsequent environmental impacts can be significant (Fischer et al., 2003; Lapointe and

Matzie, 1996; Massal et al., 2007). Understanding NO3-N sources and behavior in

stormwater is important for identifying means for pollution management in urban

settings (House and Warwick, 1998; McDiffett et al., 1989). Information on N sources

and fates can help land managers make plans to manage or control the nutrient

sources, and predict future trends that may arise as a consequence of changing

management practices.

In most cases, difference occurred in concentration of NO3-N and other dissolved

and particulate chemicals measured for similar discharge during the rising and falling

111

limbs of storm hydrographs. This is called hysteresis phenomenon. For example,

Webb and Walling (1985) examined NO3-N transport from a grassland catchment and

provided evidence for the hysteresis phenomenon including both dilution and

concentration effects for nitrate in stream water. The dilution effect is often observed

under increasing water discharge. It may be caused by surface runoff derived directly from rainfall to the solutes in baseflow. If solute concentrations increase in the rising or

falling stage of discharge, it is possible that sub-surface “reservoirs” are flushed with

high solute concentrations depending on the flushing rate and antecedent conditions

(Burt et al., 1983; House and Warwick, 1998). Similar hysteresis effect in NO3-N

concentration was observed in other studies using high-resolution nitrate sensors

(Bowes et al., 2015; Bowes et al., 2009). NO3-N behavior in stormwater runoff may also

associated with seasons. Webb and Walling (1985) claimed that the dilution responses

were typical for the winter wet period while the concentration effect was more

characteristic of summer dry period in a grassland catchment. The intensity of storms

can also affect the hysteresis effect on NO3-N concentration. Bowes et al. (2009)

reported a lack of hysteresis for NO3-N concentration during most of the largest storm events, suggesting the dominating source was the groundwater due to its constant NO3-

N concentration. The scenarios with no hysteresis effect can be associated with the

chemostatic characteristics of catchments, suggesting the solute concentration does not

change as discharge varies. Chemostatic behaviors have been commonly observed in

many catchments. For example, Godsey et al. (2009) examined the relationship

between solute concentrations and discharge in 59 geochemically diverse US

catchments and found that those catchments exhibited chemostatic characteristics.

112

Their study showed that the concentrations of Ca, Mg, Na, and Si varied by factors of

only 3 to 20 while discharge varied by several orders of magnitude, implying the

changes in solute fluxes driven by climatic factors are largely dependent on changes in

hydrology.

The understanding of NO3-N dynamics in storms can help to identify NO3-N

sources during NO3-N transport, thereby it is very critical to NO3-N source control and management. Several studies associated with NO3-N dynamics were conducted in

large watersheds or catchments where the samples were collected every hour or at less

frequency (House and Warwick, 1998; Webb and Walling, 1985), but few have been

conducted in small catchments where NO3-N concentrations can change more rapidly.

The flow in small catchments was suggested to reach the surrounding surface waterbodies in a shorter time than large watersheds (McGlynn et al., 2004), requiring data with higher resolution. Chapter 2 revealed the great N loads created from the intensely managed sports fields and the low N loads in the reclaimed water irrigated catchment, but the mechanisms of NO3-N transport in storms in the two small urban catchments have not been known yet. In this study, in situ high resolution continuous nitrate sensors called Submersible Ultraviolet Nitrate Analyzers (SUNAs) (Satlantic Inc.,

Halifax, Canada) were introduced to observe NO3-N dynamics for seasonal and annual patterns. The objectives of this study were: 1. To establish a relationship between NO3-

N concentration and discharge for the hydrological year and for individual storms in both

catchments; 2. to determine the hysteresis effect of NO3-N concentrations from both

catchments with regards to annual patterns; My hypotheses were: 1. There would be a clear relationship between NO3-N concentration and storm discharge in both

113

catchments; 2.there would be a dilution effect caused by surface runoff in NO3-N concentration from both catchments.

Materials and Methods

Study Area

The study was carried out in two small urban catchments on the campus of the

University of Florida in Gainesville, Florida (Figure 5-1). The annual precipitation with

distinct dry season (November to April) and wet season (May to October) was described

in more details in Chapter 2.

Figure 5-1. Sports Field Catchment (SFC) and Reclaimed Water Irrigated Catchment (RWC) in Lake Alice watershed

The University of Florida (UF) is one of the largest universities in the nation. It has various land uses (described in Chapter 2) which make the campus analogous to a

114

small urban “city” and function similar to a city. The urban Sports Field Catchment

(SFC) with intense fertilization is in the northern part of the campus, operated by

University Athletic Association, and it mainly consists of the football practice field and

the baseball field with an approximate total area of 6.7 ha. The pervious surface makes

up 95% of the entire area. The urban Reclaimed Water Irrigation Catchment (RWC) is in

the center east part of the campus where there is widespread landscaping around

buildings with an approximate area of 1.5 ha with 76% pervious surface. The landscape

irrigation was scheduled for application twice a week. The irrigation is made from

reclaimed water from the campus wastewater treatment facility. The runoff from SFC

and RWC drains to the nearby stream and flows to Lake Alice, a 35-ha retention pond

connected to groundwater system through a deep injection well. The Lake Alice

watershed covers an area of 426 hectare (ha), approximately 55% of the total campus

area.

Flow Measurement

The details for flow measurement are described in Chapter 2. Briefly, flow was

determined from a V-notch weir. A pressure transducer WL16 (Xylem Inc., Sacramento,

California) was used to record the water stage every 5 minutes. The discharge was

calculated based on the rating curve except that the full pipe discharge in RWC was

calculated based on Manning’s equation. The completeness test (more details

described in Chapter 2) showed that monitored flow datasets made up 90.9% of the

total discharge in SFC, and 93.2% in RWC from September 1st 2013 to August 31st

2014.

115

Precipitation Measurement

Two tipping bucket rain gauges (Bluesiren Inc., Melbourne, FL) which tipped for

every 0.254 mm of rainfall were placed on the top of a utility building on campus, close

to the instrumented catchments and used for measurement of precipitation. The rain

gauges started to work from November 21st 2013. The precipitation data before that

date were retrieved from the UF Physics Department by a tipping bucket. The annual precipitation and discharge was displayed in Figure 5-2 with numbered recorded storm events.

116

a 13 Land use=SFC 6 14, 15 1, 2, 3 7 5 9 8 16, 17 4 10, 11, 12

c 4 5, 6 Land use=RWC 12 7 9, 10, 11, 12 3 8

Date

Figure 5-2. High frequency monitoring data for precipitation. NO3-N concentration and discharge in Sports Field Catchment (SFC) and Reclaimed Water Irrigated Catchment (RWC). Number 1 to 17 in Figure 5-2-c and number 1 to 12 in Figure 5-2-e with arrows referred to the storm events (Table 5-1) recorded by Submersible Ultraviolet Nitrate Analyzer (SUNA) (Satlantic Inc., Halifax, Canada)

117

High Resolution Nutrient Analysis

Two Submersible Ultraviolet Nitrate Analyzers (SUNAs) (Satlantic Inc., Halifax,

Canada) were tested both in laboratory with Standard NO3-N solutions and in the field

(more details described in Chapter 4). The results from Chapter 4 demonstrated the good performance of the SUNAs over autosamplers in stormwater monitoring. SUNAs

were carefully deployed near the weir of SFC and the weir of RWC in the underground drainage basin, measuring NO3-N 10 times every 5 minutes or 15 minutes for storms.

The SUNA was deployed periodically during the year, depending on its availability and

when storms were expected. The detection range of NO3-N for SUNAs was between

0.007 and 28.0 mg L-1 (Satlantic Inc., Halifax, Canada). In this case, we excluded the

-1 -1 values above 25 mg L from the field data. 25 mg L was the highest NO3-N

concentration we tested for standard test in lab (more details described in Chapter 4).

Conventional Rating Curves Between Discharge And NO3-N Concentration

The rating curve shows a relationship between the discharge and NO3-N

concentration. Such curves have been widely used to estimate the concentrations

based on discharge (Ferguson, 1986; Lohani et al., 2007). A log-transformation for both discharge and concentration is commonly applied and a linear least squares regression is then used to determine the line of best fit between the transformed discharge and concentration. The relationship between discharge (Q) and NO3-N concentration (C) is

of the form:

= × (5-1) 푏 � 푎 Where푄 a and b are regression constants. The natural log transformation of

Equation (5-1) is:

ln = ln + ln (5-2)

� 푎 푏 푄 118

A log-log regression was used to illustrate the relationship between discharge

and SUNA NO3-N concentrations through the hydrological year. Also the regression

was used to determine the relationship between discharge and NO3-N concentration in

individual storms.

Characterization of Hysteresis Trajectories

Hysteresis phenomenon refers to the difference in concentrations of chemicals at

the same discharge during rising and falling limb of the hydrograph, and it is often

observed during storm events (Hall, 1970). When plotted, such concentration/discharge

relationships result in “loop trajectories”. A clockwise hysteresis loop is produced when

the concentration is higher on the rising limb of the hydrograph, and an anticlockwise

loop when the concentration is higher on the falling limb. The high-frequency NO3-N concentration data were plotted against stormwater discharge for each of the individual storm events, to determine the direction of the hysteresis trajectory (clockwise, anticlockwise, no hysteresis or a more complex pattern). The size of the loop trajectories (Hysteresis Index, HI) were quantified by determining the difference in concentration on the rising and falling limbs of the hydrograph, at the 50% of flow range

(from minimum discharge to peak discharge) for storms with one or two loops. The HI was averaged based on the returned values at 25%, 50% and 75% flow range (from minimum flow to maximum flow during the hydrograph) for complex storms with loops

more than 2 (Lawler et al., 2006).

119

RL C ations

FL C Concentr

Qmin Qmid Qmax Discharge (Q)

Figure 5-3 Schematic illustration of the hysteresis index (HImid). The simplest form is shown here, though several different discharges, across the full flow range, can also be selected iteratively for hysteresis determinations. CRL=Concentration on the Rising Limb; CFL= Concentration on the Falling Limb of the hydrograph.

There are two steps to calculate the hysteresis index (Lawler et al., 2006).

1) to determine the point of discharge at which the concentrations are to be compared, for

example, k=0.5 if Qmid is to be calculated here (Figure 5-3)

= ( ) +

where Qmax is the peak discharge푄푄 �of the푄푄 �event,푄 − 푄푄 Qmin𝑄 is the푄푄 𝑄starting discharge for the

event, and k represents the position at which the loop breadth is assessed relative to

the flow range. The value of k was set at 0.5 for storms with one or two loops: the

hysteresis loop was thus measured at the mid-point of the rising discharge limb; the

value of k was set to be 0.25, 0.5 and 0.75, respectively, to calculate the HI and then

averaged HI was set to be the HI for storms with multiple loops.

2) to interpolate the two concentrations at Qk, termed CRL, concentration at Qk on the rising

limb of the hydrograph, and CFL, the concentration associated with Qk on the falling limb

of the hydrograph (Figure 5-3). HIk is then determined, for clockwise hysteresis where

CRL >CFL, simply as:

120

HIk = CRL/CFL -1

Results and Discussions

Hydrological Factors for Storms

The descriptions of storms in both catchments are listed in Table 5-1 and Table

5-2. The precedent dry hours in both catchments ranged from 6 hours to 322 hours.

The runoff duration ranged from 65 minutes to 1835 minutes in SFC and from 65 minutes to 460 minutes in RWC. The time of peak discharge (Tp) occurred in the first 60 minutes in most of storms in RWC whereas it occurred relatively later in SFC depending on the duration of the storm and the intensity of the storm. The peak discharge in most of storms in RWC was 60.1 L s-1 which was calculated from Manning’s equation.

Although the different calculation approaches for discharge in RWC may result in the variability of runoff coefficient, the overall mean runoff coefficient and overall median runoff coefficient in RWC was higher than SFC. Both SFC and RWC had low runoff coefficients, suggesting the large pervious area in both catchments may store most of the rain water in the soil or the rain water may seep into the groundwater.

121

Table 5-1. Hydrological factors for storms in Sports Field Catchment (SFC)

1 Storm Dry Duration , 2 3 -1 Precipitation, Runoff Storm date Tp , min Qmax , L s number hours, h min mm coefficient

1 9/23/2013 10 160 20 122.3 19.6 0.13 2 9/24/2013 10 420 25 311.3 34.8 0.36 3 9/25/2013 6 230 100 91.3 12.4 0.25 4 11/26/2013 110 660 225 56.0 13.5 0.18 5 12/10/2013 322 195 10 44.5 5.1 0.26 6 12/15/2013 95 1125 330 91.3 61.0 0.32 7 1/28/2014 219 450 150 15.3 4.1 0.28 8 2/12/2014 102 375 105 15.3 7.6 0.15 9 3/17/2014 238 1835 260 120.0 46.0 0.33 10 5/1/2014 23 145 25 5.2 1.5 0.21 11 5/2/2014 13 295 15 91.3 17.3 0.19 12 5/3/2014 13 150 60 7.7 3.3 0.10 13 5/15/2014 77 325 120 353.2 43.9 0.34 14 5/29/2014 24 65 10 22.0 4.6 0.05 15 5/30/2014 24 140 10 39.1 21.3 0.04 16 8/29/2014 120 320 110 72.0 19.8 0.16 17 8/30/2014 12 500 125 306.8 41.7 0.41 Median 24 320 100 72.0 17.3 0.21 Mean 83 435 100 103.8 21.0 0.22 SD4 95 440 96 111.8 18.0 0.11 1 Runoff duration starting from the rising limb in the hydrograph to the end of fall limb 2 Time of peak in hydrograph referring to the time when maximum discharge occurred 3 Maximum discharge in the storm 4 Standard deviation

122

Table 5-2 Hydrological factors for storms in Reclaimed Water Irrigated Catchment (RWC)

1 Storm Dry Duration , 2 3 -1 Precipitation, Runoff Storm date Tp , min Qmax , L s number hours, h min mm coefficient

1 11/16/2013 10 240 105 4.6 6.6 0.14 2 11/21/2013 10 150 60 5 1.8 0.26 3 12/29/2013 6 180 105 60.1 22.0 0.42 4 2/21/2014 110 285 15 60.1 36.3 0.55 5 3/28/2014 322 460 50 60.1 41.4 0.40 6 3/29/2014 95 250 105 60.1 21.6 0.34 7 4/8/2014 219 330 30 60.1 16.0 0.54 8 5/25/2014 102 250 35 60.1 29.0 0.26 9 5/29/2014 238 65 5 22.1 4.6 0.22 10 5/30/2014 23 140 10 60.1 21.3 0.23 11 5/31/2014 13 195 10 60.1 54.9 0.18 12 6/1/2014 13 410 75 60.1 15.7 0.46 Median 59 245 43 60.1 21.4 0.30 Mean 97 246 50 47.7 22.6 0.33 SD4 108 113 39 22.8 15.8 0.14 1 Runoff duration starting from the rising limb in the hydrograph to the end of fall limb 2 Time of peak in hydrograph referring to the time when maximum discharge occurred 3 Maximum discharge in the storm 4 Standard deviation

NO3-N Concentration-Discharge Relationship

The NO3–N concentration monitored by the SUNA in SFC through the

hydrological year (Figure 5-2) demonstrated a clear relationship with discharge (Figure

5-4). The highest concentrations occurred at the lowest flows, with almost all NO3-N

-1 -1 concentrations greater than 10 mg L occurring at flows less than 1 L s . The NO3-N

concentrations decreased rapidly as the discharge increased. The lowest NO3-N concentrations (below 1 mg L-1) occurred at flows above 50 L s-1. Similar dilution curves

were shown in studies associated with NO3-N dynamics, but the dilution effect was shown more pronounced in this study than the reported dilution curves because the hydrobiogeochemical factors (flood duration, peak discharge) and dominating land use may have less influence on the NO3-N concentrations in this study (Bowes et al., 2015).

123

A power-law concentration-discharge relationship was shown in SFC (Figure 5-4,

R2=0.49, p<0.0001), however, the best fit ln (C)-ln (Q) slope in SFC was relatively small

(-0.35), indicating there might be some chemostatic behavior in SFC (Godsey et al.,

2009). Similarly, the NO3-N concentration and discharge had a significant relationship

in RWC (Figure 5-5, R2=0.02, p<0.0001), but the best fit ln (C)-ln (Q) slope was close to zero, suggesting an evident chemostatic phenomenon, meaning little variability in NO3-

N concentration with discharge. The relatively unchanging concentrations in RWC

across wide ranges of discharge require NO3-N production or NO3-N mobilization at rates nearly proportional to the water flux. Many researchers have documented that most water reaching a stream during a storm event was so-called ‘old’ (e.g., pre-storm) water. The ‘old’ water referred to a sizable part of the runoff within the hydrologic response of catchment transporting volumes constituted by aged water particles (e.g., by water particles injected at times preceding the event causally related to the observed runoff) (Botter et al., 2010; McDonnell, 1990; Ocampo et al., 2006). The mean transit time that water spends traveling subsurface through a catchment to a stream network ranged from 0.18 yr to 2.4 yr in catchments of area less than 10 ha according to the review from McGuire and McDonnell (2006). Therefore, it is likely that the NO3-N concentration in “old” water stayed unchanged and did not vary much with discharge. In addition, the surface runoff from residential lawns was reported to contain a median

-1 NO3-N concentration of 0.6 mg L (Pitt et al., 2004), which was close to the event mean concentrations in RWC in Chapter 4, implying little dilution effect.

Some of these dilution curves in SFC were directly related to individual large

storm events (Figure 5-4, dilution curves can be seen in storms on 9/24/13 and

124

5/15/14). These reductions in NO3-N concentration with increasing flow in SFC suggested that rainfall events were diluting a relatively constant input of NO3-N.

0.1 SUNA-NO3-N in all events, mg L-1

Storm event on 09/24/13

Storm event on 5/15/14

- 1 Ln (C) = 2.21 – 0.35 ln (Q) 2

mg L R = 0.49, p<0.0001

N concentration (C) concentration N - 3 10 NO

100 10000 1000 100 10 1 0.1 0.01 0.001 Flow rate (Q), L s-1

Figure 5-4. NO3-N concentration (C)-discharge (Q) relationships in Sports Field Catchment (SFC). The log-log relationship was established based on the data for the hydrological year. The discharge ranged from 0 to 1100 L s-1. Most of the discharges (from baseflow and small storm events) were below 200 L s-1, discharges above 200 L s-1 referred to the extreme storm events. Examples of hysteresis loops were shown with red line (storm event on 9/24/13) and green line (storm event on 5/15/14).

125

0.1

1 - Ln (C) = 0.14 – 0.04 ln (Q) R2 = 0.02, p<0.0001

1 N concentration N mg (C) concentration L - NO3 10 100 10 1 0.1 0.01 Flow rate (Q), L s-1

Figure 5-5. NO3-N concentration (C)-discharge (Q) relationships in Reclaimed Water Irrigated Catchment (RWC). The log-log relationship was established based on the data for the hydrological year.

The majority of NO3-N data points followed the well-defined pattern with flow in

SFC. However, there were some data points at higher than expected NO3-N

concentrations for a given flow. This can be associated with the complex event-scale

hydrobiogeochemical transport of NO3-N. For example, the high discharge at the

beginning of storms may not be able to dilute the baseflow to a very low concentration,

but more surface runoff later on further reduced the NO3-N concentration. This can produce a data point with high discharge but a relatively high concentration that does

not fall in the expected pattern.

Changes in NO3-N Concentrations through Individual Storms

The NO3-N concentration- discharge relationship and hysteresis patterns through

all storms in both catchments were given in Table 5-3, Table 5-4 and illustrated in

Figure 5-6 and Figure 5-7. There was a significant relationship between discharge and

NO3-N concentration in individual storms except on 01/28/14, 05/01/14 and 05/29/14

126

(p>0.05) in SFC and 05/29/14 in RWC. Those storms were relatively small with

precipitation ranging from 1.5 mm to 4.6 mm, indicating total precipitation may exert

effects on NO3-N dynamics in storms. This was in agreement with findings from Oeurng

et al. (2010) that there was a strong correlation between NO3-N transport and total

precipitation. Table 5-3 and Table 5-4 also provided evidence to show that the NO3-N

concentration- discharge relationship was most evident (high R2) in storms with high

peak discharge, which coincided with findings from Oeurng et al. (2010) that peak

discharge was also a correlated factor. During those events with high peak discharge,

the dilution effect was also distinct with high best fit Ln(C)-Ln(Q) slopes (Godsey et al.,

2009) (i.e., 09/23/13 and 09/24/13 in Figure 5-6) . The median of best fit Ln(C)-Ln(Q) slopes among all storms was -0.48 for SFC and -0.19 for RWC, indicating RWC had a more pronounced chemostatic behavior than SFC, which was consistent with the NO3-N

concentration-discharge relationship through the hydrological year. The chemostatic behavior was distinct in the storm on 05/29/14 in both catchments (Figure 5-6 and

Figure 5-7), little changes in NO3-N concentrations with discharge.

127

Table 5-3. A log-log regression between discharge (Q) and NO3-N concentration (C) and hysteresis patterns in SFC in each storm

Best fit Ln(C)-Ln(Q) relationship NO3-N Hysteresis Storm 1 1 2 No. of Hysteresis Storm date Constant Slope R p direction/ number observations Index (HI) Trajectories 1 9/23/2013 1.51 -0.53 0.90 32 <0.0001 Clockwise 8.7 2 9/24/2013 3.51 -0.84 0.84 84 <0.0001 Clockwise 3.1 3 9/25/2013 1.85 -0.47 0.63 46 <0.0001 Clockwise 1.5 4 11/26/2013 2.23 -0.52 0.33 39 <0.0001 Clockwise 1.4* 5 12/10/2013 1.56 -0.24 0.26 13 <0.0001 Clockwise 6.1 6 12/15/2013 3.00 -0.60 0.33 75 <0.0001 Clockwise 0.1* 7 1/28/2014 1.05 -0.04 0.01 30 0.69 Clockwise 2.2 8 2/12/2014 1.60 -0.31 0.39 25 <0.05 Clockwise 1.3 9 3/17/2014 2.89 -0.71 0.71 367 <0.0001 Clockwise 0.1* 10 5/1/2014 1.82 0.14 0.07 29 0.16 Clockwise 0.3 11 5/2/2014 1.85 -0.55 0.93 59 <0.0001 Clockwise 0.2 12 5/3/2014 2.36 -0.21 0.14 30 <0.05 Clockwise 1.1 13 5/15/2014 3.36 -0.78 0.94 65 <0.0001 Clockwise 0.5 No 14 5/29/2014 1.95 0.10 0.20 13 0.13 hysteresis 15 5/30/2014 1.14 -0.33 0.24 28 0.0083 Clockwise 0.8 16 8/29/2014 1.48 -0.48 0.72 64 <0.0001 Clockwise 0.6 17 8/30/2014 3.61 -0.83 0.89 100 <0.0001 Clockwise 1 Median 1.85 -0.48 0.39 39 1.1 Mean 2.16 -0.42 0.50 65 2.1 SD 0.82 0.30 0.33 82 2.5 1The regression equation refers to Equation 5-2: ln = + ln where a is the constant and b is the slope. R2 is regression coefficient. Hysteresis Index with * referred to storms that had multiple� 푎 loops푏 푄 more than two, so the HI was an averaged value based on Hysteresis Index calculated at 25%, 50% and 75% flow range.

128

Table 5-4. A log-log regression between discharge (Q) and NO3-N concentration (C) and hysteresis patterns in RWC in each storm

Best fit Ln(C)-Ln(Q) relationship NO3-N Hysteresis Storm 1 1 2 No. of Hysteresis Storm date Constant Slope R p direction/ number observations Index (HI) Trajectories 1 11/16/2013 -0.44 -0.18 0.35 17 <0.05 Clockwise 1.3 2 11/21/2013 -0.33 -0.16 0.42 10 <0..05 Clockwise 0.7 3 12/29/2013 -0.52 -0.46 0.83 16 <0.0001 Clockwise 2.1 4 2/21/2014 0.09 -0.20 0.79 20 <0.0001 Clockwise 0.6* 5 3/28/2014 -0.23 -0.16 0.41 92 <0.0001 Clockwise 0.02* 6 3/29/2014 -0.14 -0.25 0.87 51 <0.0001 Clockwise 0.1 7 4/8/2014 -0.25 -0.25 0.68 66 <0.0001 Clockwise 0.9* 8 5/25/2014 0.12 -0.19 0.51 58 <0.0001 Clockwise 1.3 9 5/29/2014 0.34 -0.01 0.03 14 0.55 Clockwise 0.1 10 5/30/2014 -0.16 -0.19 0.61 28 <0.0001 Clockwise 0.9 11 5/31/2014 -0.14 -0.18 0.84 39 <0.0001 Clockwise 1.0 12 6/1/2014 -0.09 -0.33 0.66 82 <0.0001 Clockwise 0.3 Median -0.15 -0.19 0.64 34 0.8 Mean -0.15 -0.21 0.58 41 0.8 SD 0.24 0.11 0.25 28 0.6 1The regression equation refers to Equation 5-2: ln = + ln where a is the constant and b is the slope. R2 is regression coefficient. � 푎 푏 푄

NO3-N concentrations produced clockwise hystereses for all the storm peaks

regardless of sizes: small storms or big storms except on 5/29/14 in SFC when there

was no hysteresis. The overall clockwise hystereses showed that NO3-N concentration was greater on the rising limb of hydrograph than at the same flow on the falling limb of the hydrograph. Several studies have also observed a general predominance of clockwise NO3-N hysteresis (Bowes et al., 2015; Schwientek et al., 2013; Zhang et al.,

2007). The predominantly clockwise trajectories indicated that NO3-N sources from

storms were rapidly mobilized and transported to the monitoring site during storm

events. The dominating storm-derived source NO3-N concentration was suggested to

be much lower compared to baseflow NO3-N concentration in SFC, therefore an evident

129

dilution effect was observed (higher hysteresis index). However, the storm-derived

source NO3-N concentration may not vary much with baseflow NO3-N concentration in

RWC, resulting in an indistinct dilution or no dilution (lower hysteresis index). Potential

sources would include in-channel stored NO3-N from the discharge pipe to exact NO3-N

monitoring location, NO3-N stored in the sports field drainage systems, NO3-N stored in

soil leaked to the discharge pipe as well as stormwater runoff from nearby impervious

surfaces. Another significant potential source of rapidly mobilized NO3-N would include

NO3-N in the “old” water as mentioned earlier. The NO3-N concentration increased immediately after the flow declined in SFC and RWC, suggesting the major baseflow

NO3-N source may be the “old” NO3-N-rich subsurface flow rather than the surface runoff. The dilution effect could also indicate a depletion of NO3-N supply through storm events (Bowes et al., 2009), but not for SFC.

The hysteresis index was close to zero in a couple of storms in SFC and RWC, which might be associated with the complexity of the relationship between NO3-N

concentration and discharge affected by other factors such as antecedent soil moisture,

spatial variability of land-based N source, hydrological conditions and so on (Chen et al., 2012; Evans and Davies, 1998). A complex hysteresis pattern was shown in Figure

5-8 with several hysteresis loops as an example to demonstrate the complexity in the relationship between NO3-N concentration and discharge. At the beginning of the storm

on 12/15/13, NO3-N concentration dropped with the increasing discharge and gradually

increased with the decreasing discharge, then it dropped again at the peak discharge

-1 (90 L s ) at the time of 150 min and remained low under high flow. The low NO3-N concentration at that time may be largely attributed to surface runoff which was much

130

lower than the baseflow in SFC. Later when the discharge gradually reduced to 20 L s-

1 -1 -1 , the NO3-N concentrations increased to 5 mg L and remained above 5 mg L

-1 although the discharge ascended to 35 L s . The relatively high NO3-N concentration

may be a mixture of event water and “old” water in the subsurface flow which was

stored for a certain period of time under regular fertilization practices, suggesting

subsurface flow was a major N source. As the discharge declined from 280th minute,

the NO3-N concentrations went up again and reached a higher level than precedent condition, suggesting subsurface flow was then the dominating N source.

131

25 30 20 09/23/13 09/24/13 15 20 10 10 5 0 0 0 50 100 150 0 200 400

25 09/25/13 30 20 11/26/13 20

15 1 - 10 10 5 0 0

, mg L 0 20 40 60 80 100 0 20 40 60

(C)

25 12/10/13 25 12/15/13 20 20 15 15 10 10 5 5 0 0

N concentrationN 0 20 40 60 0 20 40 60 80 100 -

3 NO 25 01/28/14 25 02/12/14 20 20 15 15 10 10 5 5 0 0 0 10 20 0 10 20

25 03/17/14 25 05/01/14 20 20 15 15 10 10 5 5 0 0 0 50 100 150 0 2 4 6

Flow rate (Q), L s-1 Figure 5-6. NO3-N hysteresis loops in individual storm events in SFC

132

25 05/02/14 25 05/03/14 20 20 15 15 10 10 5 5 0 0 0 20 40 60 80 100 0 5 10

1 - 25 05/15/14 25 05/29/14 20

, mg L 15 15 10 (C) 5 5 0 -5 0 100 200 300 400 0 10 20 30

25 05/30/14 25 08/29/14 20 20

N concentrationN 15 15 -

3 10 10 5 5 NO 0 0 0 20 40 60 0 50 100

25 08/30/14 20 15 10 5 0 0 200 400

Flow rate (Q), L s-1 Figure 5-6. Continued

133

2.5 11/16/13 2.5 11/21/13 2 1.5 1.5 1 0.5 0.5 0 0 2 4 6 -0.5 0 2 4 6

2.5 2.5 2 12/29/13 2 02/21/14 1.5 1.5

1 1 1 - 0.5 0.5 0 0 0 20 40 60 80 0 20 40 60 80 , mg L

(C)

2.5 2.5 2 03/28/14 2 03/29/14 1.5 1.5 1 1 0.5 0.5 0 0 0 20 40 60 80 0 50 100

N concentrationN - 3

NO 2.5 2.5 04/08/14 05/25/14 2 2 1.5 1.5 1 1 0.5 0.5 0 0

0 20 40 60 80 0 20 40 60 80

2.5 2.5 05/29/14 05/30/14 2 2 1.5 1.5 1 1 0.5 0.5 0 0 0 10 20 30 0 20 40 60 80

-1 Flow rate (Q), L s

Figure 5-7. NO3-N hysteresis loops in individual storm events in RWC

134

1 -

, mg L (C) 2.5 2.5 08/29/14 08/30/14 2 2 1.5 1.5 1 1 0.5 0.5 0 0 0 20 40 60 80 0 20 40 60 80 N concentrationN -

3 -1 NO Flow rate (Q), L s Figure 5-7. Continued

30 100

1 - NO3-N concentration

b Flow discharge 25 80

1 - 20 60 15 c a 40 10 concentration, mg L Flow rate, L s

- 20 3 5 NO 0 0 0 100 200 300 400 500

Time, min 1

20 a 2 b 30 c 20 10 1 10 0 0 0 0 50 100 0 50 100 0 50 100 N concentration (C), mg L mg (C), concentration N -

3 Flow rate (Q), L s-1 NO Figure 5-8. An example of complex storms with multiple loops. This was the storm on 12/15/13. Letter a, b, c represent single loop for a certain period of time during the storm. All the loops were clockwise, indicating dilution from the rainfall.

135

Conclusions

Urban stream ecosystem functions provide ecosystem services to human-

dominated ecosystems. Understanding NO3-N dynamics in urban streams can help to identify potential N sources and factors influencing dynamic processes and transport, therefore land managers can make plans to maintain urban stream ecology. There were a number of studies on N dynamics in watershed scale, but knowledge of how hydrological response triggers NO3-N transport at catchment level on the timescale of a single hydrological flood event is still lacking. The combined investigation of NO3-N time series information, concentration-flow relationships and hysteresis behaviors through individual storms in the Sport Field Catchement (SFC) and Reclaimed Water

Irrigated Catchment (RWC) on the campus of the University of Florida helped to identify the potential N sources through the annual cycle and during different flow regimes. The

major baseflow NO3-N source was dominated by subsurface water and responsible for

high NO3-N concentration during low flow. There was a dilution effect in stormwater

NO3-N transport associated with rapidly mobilization NO3-N sources, exhibiting a predominating clockwise hysteresis for NO3-N transport. The dilution may largely come from surface runoff. Future mitigation on NO3-N concentration in SFC should focus on the fertilizer application which may potentially increase NO3-N concentration in groundwater and increase baseflow NO3-N. Using best management practices for fertilizer management on sports fields should minimize the potential for N to be lost to the stormwater system.

This study also has important implication for high resolution sensor application.

Time scale is very important to the understanding of the process linkages between

136

catchment hydrology and streamwater chemistry. In situ high resolution nitrate sensor

in this study provided a great deal of information of how NO3-N changes with the

changes in discharge in SFC and RWC, presenting a simple, integrated methodology

that can be used to disentangle the complex nutrient signals that routinely occur in

stream systems. Determining NO3-N sources and behavior, at high temporal resolution, provides important information to allow the most appropriate mitigation options to be selected to maintain urban environment and reduce the possibility of eutrophication, thereby providing the most effective and cost-effective management of urban catchments in the future.

137

CHAPTER 6 SYNTHESIS

This dissertation focused on the urban nitrogen (N) fluxes and set the University

of Florida main Campus as the study scenario to simulate the urban environment in

metropolitan cities. Three small urban catchments were studied with various land uses

which were usually overlooked in other urban studies. The land uses were Sports Field

Catchment (SFC) with intense fertilization practice, Reclaimed Water Irrigated

Catchment (RWC) with regular reclaimed water irrigation and Control Catchment with

no irrigation, respectively. A relationship between N concentration (NO3-N and TKN) and discharge was established in each catchment based on analytical results from regular grab sample baseflow collection and autosampler stormwater collection, and corresponding discharge data. Then the relationship was used to calculate N loads based on the annual discharge datasets. The results in Chapter 2 showed that SFC created the greatest N loads (37 kg ha-1) compared with other catchments from 2013 to

2014, which were more than three times of the loads from RWC and CC, implying urban

sports fields should receive great attention in urban N management in comparison to

other urban land uses. The great N loads may be attributed to the intense fertilization

practices and high leaching capacity.

Chapter 3 attempted to make a N budget for Lake Alice watershed based on the catchment studies in Chapter 2, baseflow water quality datasets along the three major streams in the watershed and discharge datasets in basins that compose the watershed. The streams associated with the land use of sports field delivered the greatest N loads to Lake Alice. Despite the great amount of N that went to Lake Alice, only 40% of N was transported to groundwater system. This might be attributed to the

138

ecological function of wetlands near Lake Alice where denitrification was likely to occur.

It may also suggest that the N in Lake Alice was insufficient for eutrophication, instead,

it may not be enough for the reproduction of biomass, therefore it should not be an

environmental concern at the moment, meaning even though N is added from urban

runoff, the N undergoes transformations that render it harmless from an ecological

standpoint.

Chapter 4 introduced a novel technology for in situ NO3-N analysis and evaluated the performance of in situ high resolution nitrate sensors in the lab and in the field. The results showed that the readings from Submersible Ultraviolet Nitrate Analyzer (SUNAs) from Satlantic Inc. (Halifax, Canada) for standard NO3-N solution in the lab and baseflow samples in the field were very close to the results from analytical lab. In comparison with autosamplers in storms, SUNAs were able to capture more information about N dynamics with discharge than could be achieved with autosamplers. There was no significant difference between the NO3-N concentration captured in SUNA and autosamplers in SFC, however, there was a significant difference in RWC. Overall,

SUNAs performed very well in both baseflow and stormflow although it may have a slight chance to produce missing values. It can be proposed as an alternative monitoring device for N management practices.

With a good understanding of SUNAs’ performance in the field, Chapter 5 focused on the N dynamics captured by SUNAs in SFC and RWC. Understanding N dynamics in storms can help to identify the N sources, therefore to make plans for future management. In Chapter 5, SUNAs were deployed to capture N dynamics in storms and baseflow in SFC and RWC from 2013 to 2014. The results showed that RWC had

139

a more distinct chemostatic behavior than SFC, suggesting little changes in NO3-N concentrations with discharge in RWC. Clockwise hystereses were dominating in

storms in both SFC and RWC, however, the dilution effect was more evident in SFC. It

indicated that the storm-derived N source was from surface runoff in SFC and baseflow

N source with high NO3-N concentrations was from “old” water stored in subsurface

flow.

The dissertation identified the land use of greatest concern in Lake Alice

watershed regarding creating N loads among three urban land uses, and implied the N

transport flow path during storms in this land use, providing a direction for future

stormwater N control and management. The Lake Alice watershed can provide an

understanding of the magnitude of N all over the campus, which can help make plans

for mitigation of N losses in stormwater in the future.

140

LIST OF REFERENCES

Aber J., McDowell W., Nadelhoffer K., Magill A., Berntson G., Kamakea M., McNulty S., Currie W., Rustad L., Fernandez I. (1998) Nitrogen saturation in temperate forest ecosystems. BioScience:921-934.

Adams W.A. (1986) Practical aspects of sportsfield drainage. Soil Use and Management 2:51-54. DOI: 10.1111/j.1475-2743.1986.tb00679.x.

Adriano D., Page A., Elseewi A., Chang A., Straughan I. (1980) Utilization and disposal of fly ash and other coal residues in terrestrial ecosystems: A review. Journal of Environmental Quality 9:333-344.

Al-Kaisi M.M., Yin X. (2003) Effects of nitrogen rate, irrigation rate, and plant population on corn yield and water use efficiency. Agronomy journal 95:1475-1482.

Albertin A.R., Sickman J.O., Pinowska A., Stevenson R.J. (2012) Identification of nitrogen sources and transformations within karst springs using isotope tracers of nitrogen. Biogeochemistry 108:219-232.

Alva A., Paramasivam S., Fares A., Obreza T., Schumann A. (2006) Nitrogen best management practice for citrus trees: II. Nitrogen fate, transport, and components of N budget. Scientia Horticulturae 109:223-233.

Anderson D.M., Glibert P.M., Burkholder J.M. (2002) Harmful algal blooms and eutrophication: nutrient sources, composition, and consequences. Estuaries 25:704-726.

Ávila A., Piñol J., Rodà F., Neal C. (1992) Storm solute behaviour in a montane Mediterranean forested catchment. Journal of Hydrology 140:143-161.

Badruzzaman M., Pinzon J., Oppenheimer J., Jacangelo J.G. (2012) Sources of nutrients impacting surface waters in Florida: A review. Journal of Environmental Management 109:80-92.

Baker L. (1998) Design considerations and applications for wetland treatment of high- nitrate waters. Water Science and Technology 38:389-395.

Baker L.A., Hope D., Xu Y., Edmonds J., Lauver L. (2001) Nitrogen balance for the Central Arizona–Phoenix (CAP) ecosystem. Ecosystems 4:582-602.

Balmforth D. (1990) The Pollution Aspects of Storm‐Sewage Overflows. Water and Environment Journal 4:219-226.

141

Barton L., Colmer T.D. (2006) Irrigation and fertiliser strategies for minimising nitrogen leaching from turfgrass. Agricultural Water Management 80:160-175. DOI: http://dx.doi.org/10.1016/j.agwat.2005.07.011.

Bauters T.W., Steenhuis T.S., Parlange J.-Y., DiCarlo D.A. (1998) Preferential flow in water-repellent sands. Soil Science Society of America Journal 62:1185-1190.

Beard J.B. (2002) Turf management for golf courses Ann Arbor Press.

Berndtsson J.C. (2010) Green roof performance towards management of runoff water quantity and quality: A review. Ecological Engineering 36:351-360.

Berndtsson J.C., Bengtsson L., Jinno K. (2009) Runoff water quality from intensive and extensive vegetated roofs. Ecological Engineering 35:369-380.

Bertrand-Krajewski J.-L., Chebbo G., Saget A. (1998) Distribution of pollutant mass vs volume in stormwater discharges and the first flush phenomenon. Water Research 32:2341-2356. DOI: 10.1016/s0043-1354(97)00420-x.

Bigelow C.A., Bowman D.C., Cassel D.K. (2001) Nitrogen leaching in sand-based rootzones amended with inorganic soil amendments and sphagnum peat. Journal of the American Society for Horticultural Science 126:151-156.

Bigelow C.A., Bowman D.C., Cassel D.K. (2004) Physical properties of three sand size classes amended with inorganic materials or sphagnum peat moss for putting green rootzones. Crop Science 44:900-907.

Bijoor N.S., Czimczik C.I., Pataki D.E., Billings S.A. (2008) Effects of temperature and fertilization on nitrogen cycling and community composition of an urban lawn. Global Change Biology 14:2119-2131.

Birch G.F., Matthai C., Fazeli M.S., Suh J.Y. (2004) Efficiency of a constructed wetland in removing contaminants from stormwater. Wetlands 24:459-466.

Blanco-Montero C.A., Bennett T.B., Neville P., Crawford C.S., Milne B.T., Ward C.R. (1995) Potential environmental and economic impacts of turfgrass in Albuquerque, New Mexico (USA). Landscape Ecology 10:121-128.

Boesch D.F., Brinsfield R.B., Magnien R.E. (2001) Chesapeake Bay Eutrophication. Journal of Environmental Quality 30:303-320.

Bond D.W., Steiger S., Zhang R., Tie X., Orville R.E. (2002) The importance of NOx production by lightning in the tropics. Atmospheric Environment 36:1509-1519.

Bormann F., Likens G., Melillo J. (1977) Nitrogen budget for an aggrading northern hardwood forest ecosystem. Science 196:981-983.

142

Botter G., Bertuzzo E., Rinaldo A. (2010) Transport in the hydrologic response: Travel time distributions, soil moisture dynamics, and the old water paradox. Water Resources Research 46.

Bowes M., Jarvie H., Halliday S., Skeffington R., Wade A., Loewenthal M., Gozzard E., Newman J., Palmer-Felgate E. (2015) Characterising phosphorus and nitrate inputs to a rural river using high-frequency concentration–flow relationships. Science of The Total Environment 511:608-620.

Bowes M.J., Smith J.T., Neal C. (2009) The value of high-resolution nutrient monitoring: a case study of the River Frome, Dorset, UK. Journal of Hydrology 378:82-96.

Brady N.C., Weil R.R. (1996) The nature and properties of soils Prentice-Hall Inc.

Braskerud B. (2002) Factors affecting nitrogen retention in small constructed wetlands treating agricultural non-point source pollution. Ecological Engineering 18:351- 370.

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 Research 36:1743-1757.

Brown K., Duble R., Thomas J. (1977) Influence of management and season on fate of N applied to golf greens. Agronomy journal 69:667-671.

Brye K., Norman J., Bundy L., Gower S. (2000) Water-budget evaluation of prairie and maize ecosystems. Soil Science Society of America Journal 64:715-724.

Burt T., Butcher D., Coles N., Thomas A. (1983) The natural history of Slapton Ley Nature Reserve XV: hydrological processes in the Slapton Wood catchment. Field Studies 5:731-732.

Canfield Jr D.E., Hoyer M.V. (1988) The eutrophication of Lake Okeechobee. Lake and Reservoir Management 4:91-99.

Capelo S., Mira F., De Bettencourt A. (2007) In situ continuous monitoring of chloride, nitrate and ammonium in a temporary stream: comparison with standard methods. Talanta 71:1166-1171.

Carpenter S.R., Caraco N.F., Correll D.L., Howarth R.W., Sharpley A.N., Smith V.H. (1998) Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecological Applications 8:559-568.

143

Chai C., Yu Z., Song X., Cao X. (2006) The status and characteristics of eutrophication in the Yangtze River (Changjiang) Estuary and the adjacent East China Sea, China. Hydrobiologia 563:313-328.

Chang G., Mahoney K., Briggs-Whitmire A., Kohler D., Mobley C., Lewis M., Moline M., Boss E., Kim M., Philpot W. (2004) The new age of hyperspectral oceanography.

Charbeneau R.J., Barrett M.E. (1998) Evaluation of methods for estimating stormwater pollutant loads. Water Environment Research:1295-1302.

Chen N., Wu J., Hong H. (2012) Effect of storm events on riverine nitrogen dynamics in a subtropical watershed, southeastern China. Science of The Total Environment 431:357-365.

Cleveland C.C., Townsend A.R., Schimel D.S., Fisher H., Howarth R.W., Hedin L.O., Perakis S.S., Latty E.F., Von Fischer J.C., Elseroad A. (1999) Global patterns of terrestrial biological nitrogen (N2) fixation in natural ecosystems. Global Biogeochemical Cycles 13:623-645.

Cohen M., Lamsal S., Kohrnak L. (2007) Sources, transport and transformations of nitrate-N in the Florida environment. Final Report. St. Johns River Water Management District. SJ2007-SP10.

Conley D.J., Paerl H.W., Howarth R.W., Boesch D.F., Seitzinger S.P., Havens K.E., Lancelot C., Likens G.E. (2009) Controlling eutrophication: nitrogen and phosphorus. Science 323:1014-1015.

Cooper A.B., Cooke J.G. (1984) Nitrate loss and transformation in 2 vegetated headwater streams. New Zealand Journal of Marine and Freshwater Research 18:441-450.

Dale V.H., Pearson S.M., Offerman H.L., O'Neill R.V. (1994) Relating patterns of land‐ use change to faunal biodiversity in the central Amazon. Conservation Biology 8:1027-1036.

David M.B., Gentry L.E., Kovacic D.A., Smith K.M. (1997) Nitrogen balance in and export from an agricultural watershed. Journal of Environmental Quality 26:1038- 1048.

Davis A.P., Shokouhian M., Sharma H., Minami C. (2001) Laboratory study of biological retention for urban stormwater management. Water Environment Research:5-14. de Vries W., Leip A., Reinds G., Kros J., Lesschen J.P., Bouwman A. (2011) Comparison of land nitrogen budgets for European agriculture by various modeling approaches. Environmental Pollution.

144

DeBusk K., Wynn T. (2011) Storm-water bioretention for runoff quality and quantity mitigation. Journal of Environmental Engineering 137:800-808.

Del Barrio E.P. (1998) Analysis of the green roofs cooling potential in buildings. Energy and buildings 27:179-193.

Diaz-Valbuena L.R., Leverenz H.L., Cappa C.D., Tchobanoglous G., Horwath W.R., Darby J.L. (2011) Methane, carbon dioxide, and nitrous oxide emissions from septic tank systems. Environmental Science & Technology 45:2741-2747.

Dodd R., McMahon G., Stichter S. (1992) Watershed planning in the Albermarle-Pamlic Estuarine System: Average annual nutrient budgets. Research Triangle Institute, Research Triangle Park, NC.

Downing J., McClain M., Twilley R., Melack J., Elser J., Rabalais N., Lewis Jr W., Turner R., Corredor J., Soto D. (1999) The impact of accelerating land-use change on the N-cycle of tropical aquatic ecosystems: current conditions and projected changes. Biogeochemistry 46:109-148.

Drolc A., Vrtovšek J. (2010) Nitrate and nitrite nitrogen determination in waste water using on-line UV spectrometric method. Bioresource Technology 101:4228-4233.

Dunnett N., Kingsbury N. (2004) Planting green roofs and living walls Timber Press Portland, OR.

Ellis J.B. (2001) Sewer infiltration/exfiltration and interactions with sewer flows and groundwater quality, 2nd International Conference Interactions between sewers, treatment plants and receiving waters in urban areas–Interurba II. pp. 19-22.

Engelsjord M., Singh B. (1997) Effects of slow-release fertilizers on growth and on uptake and leaching of nutrients in Kentucky bluegrass turfs established on sand- based root zones. Canadian journal of plant science 77:433-444.

EPA A. (2011) Inventory of US greenhouse gas emissions and sinks: 1990-2009, EPA 430-R-11-005.

Evans C., Davies T.D. (1998) Causes of concentration/discharge hysteresis and its potential as a tool for analysis of episode hydrochemistry. Water Resources Research 34:129-137.

Ferguson R. (1986) River loads underestimated by rating curves. Water Resources Research 22:74-76.

Finch M.S., Hydes D.J., Clayson C.H., Weigl B., Dakin J., Gwilliam P. (1998) A low power ultra violet spectrophotometer for measurement of nitrate in seawater: introduction, calibration and initial sea trials. Analytica chimica acta 377:167-177.

145

Fischer D., Charles E.G., Baehr A.L. (2003) Effects of stormwater infiltration on quality of groundwater beneath retention and detention basins. Journal of Environmental Engineering 129:464-471.

Fisher D.C., Oppenheimer M. (1991) Atmospheric nitrogen deposition and the Chesapeake Bay estuary. Ambio:102-108.

Flint K.R., Davis A.P. (2007) Pollutant mass flushing characterization of highway stormwater runoff from an ultra-urban area. Journal of Environmental Engineering 133:616-626.

Forman D., Al-Dabbagh S., Doll R. (1984) Nitrates, nitrites and gastric cancer in Great Britain. Nature 313:620-625.

Fowler G.D., Roseen R.M., Ballestero T.P., Guo Q., Houle J. (2009) Sediment Monitoring Bias by Autosampler in Comparison with Whole Volume Sampling for Parking Lot Runoff. pp. 1514-1522.

Galloway J.N., Dentener F.J., Capone D.G., Boyer E.W., Howarth R.W., Seitzinger S.P., Asner G.P., Cleveland C.C., Green P.A., Holland E.A., Karl D.M., Michaels A.F., Porter J.H., Townsend A.R., Vöosmarty C.J. (2004) Nitrogen Cycles: Past, Present, and Future. Biogeochemistry 70:153-226. DOI: 10.1007/s10533-004- 0370-0.

Galloway J.N., Howarth R.W., Michaels A.F., Nixon S.W., Prospero J.M., Dentener F.J. (1996) Nitrogen and phosphorus budgets of the North Atlantic Ocean and its watershed, in: R. Howarth (Ed.), Nitrogen Cycling in the North Atlantic Ocean and its Watersheds, Springer Netherlands. pp. 3-25.

Galloway J.N., Schlesinger W.H., Levy H., Michaels A., Schnoor J.L. (1995) Nitrogen fixation: Anthropogenic enhancement‐environmental response. Global Biogeochemical Cycles 9:235-252.

Garcia-Martino A.R., Warner G.S., Scatena F.N., Civco D.L. (1996) Rainfall, runoff and elevation relationships in the Luquillo Mountains of Puerto Rico. Caribbean Journal of Science 32:413-424.

Gentry L.E., David M.B., Below F.E., Royer T.V., McIsaac G.F. (2009) Nitrogen mass balance of a tile-drained agricultural watershed in east-central Illinois. Journal of Environmental Quality 38:1841-1847.

Gheysari M., Mirlatifi S.M., Homaee M., Asadi M.E., Hoogenboom G. (2009) Nitrate leaching in a silage maize field under different irrigation and nitrogen fertilizer rates. Agricultural Water Management 96:946-954.

146

Gilbert J.K., Clausen J.C. (2006) Stormwater runoff quality and quantity from asphalt, paver, and crushed stone driveways in Connecticut. Water Research 40:826- 832. DOI: 10.1016/j.watres.2005.12.006.

Gilbert M., Needoba J., Koch C., Barnard A., Baptista A. (2013) Nutrient loading and transformations in the Columbia River Estuary determined by high-resolution in situ sensors. Estuaries and Coasts 36:708-727.

Glaser B., Lehmann J., Steiner C., Nehls T., Yousaf M., Zech W. (2002) Potential of pyrolyzed organic matter in soil amelioration, 12th ISCO conference, Beijing. pp. 421-427.

Glibert P.M., Landsberg J.H., Evans J.J., Al-Sarawi M.A., Faraj M., Al-Jarallah M.A., Haywood A., Ibrahem S., Klesius P., Powell C. (2002) A fish kill of massive proportion in Kuwait Bay, Arabian Gulf, 2001: the roles of bacterial disease, harmful algae, and eutrophication. Harmful Algae 1:215-231.

Godsey S.E., Kirchner J.W., Clow D.W. (2009) Concentration–discharge relationships reflect chemostatic characteristics of US catchments. Hydrological Processes 23:1844-1864. Gold A.J., DeRagon W.R., Sullivan W.M., Lemunyon J.L. (1990) Nitrate-nitrogen losses to groundwater from rural and suburban land uses. Journal of soil and water conservation 45:305-310.

Goldewijk K.K. (2001) Estimating global land use change over the past 300 years: the HYDE database. Global Biogeochemical Cycles 15:417-433.

Graedel T.E., Keene W. (1996) The budget and cycle of Earth's natural chlorine. Pure and applied Chemistry 68:1689-1697.

Groffman P.M., Law N.L., Belt K.T., Band L.E., Fisher G.T. (2004) Nitrogen fluxes and retention in urban watershed ecosystems. Ecosystems 7:393-403.

Guillard K., Kopp K.L. (2004) Nitrogen fertilizer form and associated nitrate leaching from cool-season lawn turf. Journal of Environmental Quality 33:1822-1827.

Gutierrez O., Sutherland-Stacey L., Yuan Z. (2010) Simultaneous online measurement of sulfide and nitrate in sewers for nitrate dosage optimisation.

Hager S.W., Schemel L.E. (1992) Sources of nitrogen and phosphorus to northern San Francisco Bay. Estuaries 15:40-52.

Hall F.R. (1970) Dissolved Solids‐Discharge Relationships: 1. Mixing Models. Water Resources Research 6:845-850.

147

Hamilton A., Stagnitti F., Boland A.-M., Premier R. (2005) Quantitative microbial risk assessment modelling for the use of reclaimed water in irrigated horticulture. Environmental health risk III:71-81.

Hammer D.A., Knight R.L. (1994) Designing constructed wetlands for nitrogen removal. Water Science & Technology 29:15-27.

Han Y., Lau S.-L., Kayhanian M., Stenstrom M.K. (2006) Characteristics of highway stormwater runoff. Water Environment Research 78:2377-2388.

Harivandi M., Hagan W., Elmore C. (2001) Recycling mower effects on biomass, nitrogen recycling, weed invasion, turf quality, and thatch. Int. Turfgrass Soc. Res. J 9:882-885.

Harmel R., King K. (2005) Uncertainty in measured sediment and nutrient flux in runoff from small agricultural watersheds. Transactions of the ASAE 48:1713-1721.

Harmel R., King K., Slade R. (2003) Automated storm water sampling on small watersheds. Applied Engineering in Agriculture 19:667-674.

Harvey J.W., Bencala K.E. (1993) The effect of streambed topography on surface‐ subsurface water exchange in mountain catchments. Water Resources Research 29:89-98.

Hathaway A.M., Hunt W.F., Jennings G.D. (2008) A field study of green roof hydrologic and water quality performance. Trans. ASABE 51:37-44.

Heaney J.P., Wright L., Sample D. (1999) Research needs in urban wet weather flows. Water Environment Research:241-250.

Heckman J., Liu H., Hill W., DeMilia M., Anastasia W. (2000) Kentucky bluegrass responses to mowing practice and nitrogen fertility management. Journal of Sustainable Agriculture 15:25-33.

Hendry C.D., Brezonik P.L. (1980) Chemistry of precipitation at Gainesville, Florida. Environmental Science & Technology 14:843-849. DOI: 10.1021/es60167a008.

Hensley R.T., Cohen M.J., Korhnak L.V. (2014) Inferring nitrogen removal in large rivers from high‐resolution longitudinal profiling. Limnology and Oceanography 59:1152-1170.

Hiscock J.G., Thourot C.S., Zhang J. (2003) Phosphorus budget—land use relationships for the northern Lake Okeechobee watershed, Florida. Ecological Engineering 21:63-74.

148

Hoffman L., Ebdon J., Dest W., DaCosta M. (2010) Effects of nitrogen and potassium on wear mechanisms in perennial ryegrass: I. wear tolerance and recovery. Crop Science 50:357-366.

Horowitz A.J. (1995) The use of suspended sediment and associated trace elements in water quality studies International Association of Hydrological Sciences.

Horowitz A.J. (2009) Monitoring suspended sediments and associated chemical constituents in urban environments: lessons from the city of Atlanta, Georgia, USA Water Quality Monitoring Program. Journal of Soils and Sediments 9:342- 363.

House W.A., Warwick M.S. (1998) Hysteresis of the solute concentration/discharge relationship in rivers during storms. Water Research 32:2279-2290.

Howarth R.W., Marino R., Lane J., Cole J.J. (1988) Nitrogen fixation in freshwater, estuarine, and marine ecosystems. 1. Rates and importance1. Limnology and Oceanography 33:669-687.

Hsieh C.-h., Davis A.P. (2005) Evaluation and optimization of bioretention media for treatment of urban storm water runoff. Journal of Environmental Engineering 131:1521-1531.

Hsieh C.H., Davis A.P., Needelman B.A. (2007) Nitrogen removal from urban stormwater runoff through layered bioretention columns. Water Environment Research 79:2404-2411. DOI: Doi 10.2175/106143007x183844.

Imai I., Yamaguchi M., Hori Y. (2006) Eutrophication and occurrences of harmful algal blooms in the Seto Inland Sea, Japan. Plankton and Benthos Research 1:71-84.

Jacob D.J., Wofsy S.C. (1990) Budgets of reactive nitrogen, hydrocarbons, and ozone over the Amazon forest during the wet season. Journal of Geophysical Research: Atmospheres (1984–2012) 95:16737-16754.

Jansson M., Andersson R., Berggren H., Leonardson L. (1994) Wetlands and Lakes as Nitrogen Traps. Ambio 23:320-325. DOI: 10.2307/4314231.

Jasechko S., Sharp Z.D., Gibson J.J., Birks S.J., Yi Y., Fawcett P.J. (2013) Terrestrial water fluxes dominated by transpiration. Nature 496:347-350.

Johnes P. (2007) Uncertainties in annual riverine phosphorus load estimation: impact of load estimation methodology, sampling frequency, baseflow index and catchment population density. Journal of Hydrology 332:241-258.

149

Johnson K.S., Coletti L.J. (2002) In situ ultraviolet spectrophotometry for high resolution and long-term monitoring of nitrate, bromide and bisulfide in the ocean. Deep Sea Research Part I: Oceanographic Research Papers 49:1291-1305.

Juang T., Wang M., Chen H., Tan C. (2001) Ammonium fixation by surface soils and clays. Soil science 166:345-352.

Kalnay E., Cai M. (2003) Impact of urbanization and land-use change on climate. Nature 423:528-531.

Karpf C., Krebs P. (2004) Sewers as drainage systems-quantification of groundwater infiltration. NOVATECH 2004 “Sustainable Techniques and Strategies in Urban Water Management”, Lyon, Proceedings Vol 2:969-975.

Kaushal S.S., Lewis Jr W.M., McCutchan Jr J.H. (2006) Land use change and nitrogen enrichment of a Rocky Mountain watershed. Ecological Applications 16:299-312.

Kaye J.P., Groffman P.M., Grimm N.B., Baker L.A., Pouyat R.V. (2006) A distinct urban biogeochemistry? Trends in Ecology & Evolution 21:192-199.

Kelling K., Peterson A. (1975) Urban lawn infiltration rates and fertilizer runoff losses under simulated rainfall. Soil Science Society of America Journal 39:348-352.

Kemp M.J., Dodds W.K. (2001) Spatial and temporal patterns of nitrogen concentrations in pristine and agriculturally-influenced prairie streams. Biogeochemistry 53:125- 141.

Khare Y.P., Martinez C.J., Toor G.S. (2012) Water Quality and Land Use Changes in the Alafia and Hillsborough River Watersheds, Florida, USA1, Wiley Online Library.

Kim H., Seagren E.A., Davis A.P. (2003) Engineered bioretention for removal of nitrate from stormwater runoff. Water Environment Research:355-367.

Kim L.-H., Zoh K.-D., Jeong S.-m., Kayhanian M., Stenstrom M.K. (2006) Estimating pollutant mass accumulation on highways during dry periods. Journal of Environmental Engineering 132:985-993.

Kim T., Kim G., Kim S., Choi E. (2008) Estimating riverine discharge of nitrogen from the South Korea by the mass balance approach. Environmental monitoring and assessment 136:371-378.

Klein Goldewijk C., Battjes J. (1997) A hundred year (1890–1990) database for integrated environmental assessments (HYDE, version 1.1). National institute of public health and the environment (RIVM), Bilthoven, the Netherlands.

150

Köhler M. (2003) Plant survival research and biodiversity: lessons from Europe, first annual greening rooftops for sustainable communities conference, Awards and Trade show. pp. 20-30.

Kopp K.L., Guillard K. (2002) Clipping management and nitrogen fertilization of turfgrass. Crop Science 42:1225-1231.

Korsaeth A., Eltun R. (2000) Nitrogen mass balances in conventional, integrated and ecological cropping systems and the relationship between balance calculations and nitrogen runoff in an 8-year field experiment in Norway. Agriculture, ecosystems & environment 79:199-214.

Laird D.A., Fleming P., Davis D.D., Horton R., Wang B., Karlen D.L. (2010) Impact of biochar amendments on the quality of a typical Midwestern agricultural soil. Geoderma 158:443-449.

Langergraber G., Fleischmann N., Hofstaedter F. (2003) A multivariate calibration procedure for UV/VIS spectrometric quantification of organic matter and nitrate in wastewater. Water Science & Technology 47:63-71.

Lapointe B.E., Matzie W.R. (1996) Effects of stormwater nutrient discharges on eutrophication processes in nearshore waters of the Florida Keys. Estuaries 19:422-435.

Lawler D.M., Petts G.E., Foster I.D., Harper S. (2006) Turbidity dynamics during spring storm events in an urban headwater river system: The Upper Tame, West Midlands, UK. Science of The Total Environment 360:109-126.

Lee C.g., Fletcher T.D., Sun G. (2009) Nitrogen removal in constructed wetland systems. Engineering in Life Sciences 9:11-22.

Lee J.H., Bang K.W. (2000) Characterization of urban stormwater runoff. Water Research 34:1773-1780. DOI: 10.1016/s0043-1354(99)00325-5.

Lehmann J., da Silva Jr J.P., Steiner C., Nehls T., Zech W., Glaser B. (2003) Nutrient availability and leaching in an archaeological Anthrosol and a Ferralsol of the Central Amazon basin: fertilizer, manure and charcoal amendments. Plant and soil 249:343-357.

Levy H., Moxim W.J. (1989) Simulated global distribution and deposition of reactive nitrogen emitted by fossil fuel combustion. Tellus B 41:256-271.

Line D.E., White N.M., Osmond D.L., Jennings G.D., Mojonnier C.B. (2002) Pollutant export from various land uses in the Upper Neuse River Basin. Water Environment Research:100-108.

151

Line D.E., Wu J., Arnold J., Jennings G., Rubin A.R. (1997) Water quality of first flush runoff from 20 industrial sites. Water Environment Research:305-310.

Liu B.M., Collick A.S., Zeleke G., Adgo E., Easton Z.M., Steenhuis T.S. (2008) Rainfall‐discharge relationships for a monsoonal climate in the Ethiopian highlands. Hydrological Processes 22:1059-1067.

Liu H., Hull R.J., Duff D. (1997) Comparing Cultivars of Three Cool-Season Turfgrass for Soil Water NO− 3 Concentration and Leaching Potential. Crop Science 37:526-534.

Lohani A., Goel N., Bhatia K.S. (2007) Deriving stage–discharge–sediment concentration relationships using fuzzy logic. Hydrological Sciences Journal 52:793-807.

Lovett G.M., Traynor M.M., Pouyat R.V., Carreiro M.M., Zhu W.-X., Baxter J.W. (2000) Atmospheric deposition to oak forests along an urban-rural gradient. Environmental Science & Technology 34:4294-4300.

Lü C., Tian H. (2007) Spatial and temporal patterns of nitrogen deposition in China: synthesis of observational data. Journal of Geophysical Research: Atmospheres (1984–2012) 112.

Luo L., Qin B., Yang L., Song Y. (2007) Total inputs of phosphorus and nitrogen by wet deposition into Lake Taihu, China, Eutrophication of Shallow Lakes with Special Reference to Lake Taihu, China, Springer. pp. 63-70.

Mallin M., Johnson V., Ensign S. (2009) Comparative impacts of stormwater runoff on water quality of an urban, a suburban, and a rural stream. Environmental monitoring and assessment 159:475-491. DOI: 10.1007/s10661-008-0644-4.

Massal L.R., Snodgrass J.W., Casey R.E. (2007) Nitrogen pollution of stormwater ponds: potential for toxic effects on amphibian embryos and larvae. Applied Herpetology 4:19-29.

McClaugherty C.A., Aber J.D., Melillo J.M. (1982) The role of fine roots in the organic matter and nitrogen budgets of two forested ecosystems. Ecology:1481-1490.

McDiffett W.F., Beidler A.W., Dominick T.F., McCrea K.D. (1989) Nutrient concentration-stream discharge relationships during storm events in a first-order stream. Hydrobiologia 179:97-102.

McDonnell J.J. (1990) A rationale for old water discharge through macropores in a steep, humid catchment. Water Resour. Res 26:2821-2832.

152

McGlynn B.L., McDonnell J.J., Seibert J., Kendall C. (2004) Scale effects on headwater catchment runoff timing, flow sources, and groundwater‐streamflow relations. Water Resources Research 40.

McGuire K.J., McDonnell J.J. (2006) A review and evaluation of catchment transit time modeling. Journal of Hydrology 330:543-563.

McLeod M., Aislabie J., Smith J., Fraser R., Roberts A., Taylor M. (2001) Viral and chemical tracer movement through contrasting soils. Journal of Environmental Quality 30:2134-2140.

McLeod M., Schipper L., Taylor M. (1998) Preferential flow in a well drained and a poorly drained soil under different overhead irrigation regimes. Soil Use and Management 14:96-100.

McMahon G., Woodside M.D. (1997) Nutrient mass balance for the Albemarle-Pamlico drainage basin, North Carolina and Virginia, 1990. JAWRA Journal of the American Water Resources Association 33:573-589.

Meisinger J., Randall G. (1991) Estimating nitrogen budgets for soil-crop systems. Managing nitrogen for groundwater quality and farm profitability:85-124.

Meli S., Porto M., Belligno A., Bufo S.A., Mazzatura A., Scopa A. (2002) Influence of irrigation with lagooned urban wastewater on chemical and microbiological soil parameters in a citrus orchard under Mediterranean condition. Science of The Total Environment 285:69-77.

Mentens J., Raes D., Hermy M. (2006) Green roofs as a tool for solving the rainwater runoff problem in the urbanized 21st century? Landscape and urban planning 77:217-226.

Milesi C., Running S.W., Elvidge C.D., Dietz J.B., Tuttle B.T., Nemani R.R. (2005) Mapping and modeling the biogeochemical cycling of turf grasses in the United States. Environmental Management 36:426-438.

Miller G., Cisar J. (2001) Maintaining Athletic Fields. Miller L.W. (1971) Methemoglobinemia associated with well water. JAMA 216:1642- 1643.

Miltner E., Branham B., Paul E., Rieke P. (1996) Leaching and mass balance of 15N- labeled urea applied to a Kentucky bluegrass turf. Crop Science 36:1427-1433.

Moncheva S., Gotsis-Skretas O., Pagou K., Krastev A. (2001) Phytoplankton blooms in Black Sea and Mediterranean coastal ecosystems subjected to anthropogenic eutrophication: similarities and differences. Estuarine, Coastal and Shelf Science 53:281-295.

153

Moran A., Hunt B., Jennings G. (2004) A North Carolina field study to evaluate greenroof runoff quantity, runoff quality, and plant growth.

Morton T., Gold A., Sullivan W. (1988) Influence of overwatering and fertilization on nitrogen losses from home lawns. Journal of Environmental Quality 17:124-130.

Motz L. (1998) Vertical leakage and vertically averaged vertical conductance for karst lakes in Florida. Water Resources Research 34:159-167.

Motz L.H., Sousa G.D., Annable M.D. (2001) Water budget and vertical conductance for Lowry (Sand Hill) Lake in north-central Florida, USA. Journal of Hydrology 250:134-148.

Murray J., Juska F. (1977) Effect of management practices on thatch accumulation, turf quality, and leaf spot damage in common Kentucky bluegrass. Agronomy journal 69:365-369.

Nadelhoffer K.J., Emmett B.A., Gundersen P., Kjønaas O.J., Koopmans C.J., Schleppi P., Tietema A., Wright R.F. (1999) Nitrogen deposition makes a minor contribution to carbon sequestration in temperate forests. Nature 398:145-148.

Nagdali S.S., Gupta P. (2002) Impact of mass mortality of a mosquito fish, Gambusia affinison the ecology of a fresh water eutrophic lake (Lake Naini Tal, India). Hydrobiologia 468:45-51.

Nakamura K., Harter T., Hirono Y., Horino H., Mitsuno T. (2004) Assessment of root zone nitrogen leaching as affected by irrigation and nutrient management practices. Vadose Zone Journal 3:1353-1366.

Nicholls K.H., Cox C.M. (1978) Atmospheric nitrogen and phosphorus loading to Harp Lake, Ontario, Canada. Water Resources Research 14:589-592.

Nikolaidis N., Heng H., Semagin R., Clausen J. (1998) Non-linear response of a mixed land use watershed to nitrogen loading. Agriculture, ecosystems & environment 67:251-265.

Nixon S., Granger S., Nowicki B. (1995) An assessment of the annual mass balance of carbon, nitrogen, and phosphorus in Narragansett Bay. Biogeochemistry 31:15- 61.

Nordin A., Strengbom J., Witzell J., Näsholm T., Ericson L. (2005) Nitrogen deposition and the biodiversity of boreal forests: implications for the nitrogen critical load. AMBIO: A Journal of the Human Environment 34:20-24.

154

Oberndorfer E., Lundholm J., Bass B., Coffman R.R., Doshi H., Dunnett N., Gaffin S., Köhler M., Liu K.K., Rowe B. (2007) Green roofs as urban ecosystems: ecological structures, functions, and services. BioScience 57:823-833.

Ocampo C.J., Sivapalan M., Oldham C. (2006) Hydrological connectivity of upland- riparian zones in agricultural catchments: Implications for runoff generation and nitrate transport. Journal of Hydrology 331:643-658.

Odum H.T. (1971) Environment, power, and society Wiley-Interscience New York.

Oenema O., Kros H., de Vries W. (2003) Approaches and uncertainties in nutrient budgets: implications for nutrient management and environmental policies. European Journal of Agronomy 20:3-16.

Oeurng C., Sauvage S., Sánchez-Pérez J.-M. (2010) Temporal variability of nitrate transport through hydrological response during flood events within a large agricultural catchment in south-west France. Science of The Total Environment 409:140-149.

Ou W.-S., Lin Y.-F., Jing S.-R., Lin H.-T. (2006) Performance of a constructed wetland- pond system for treatment and reuse of wastewater from campus buildings. Water Environment Research 78:2369-2376.

Owen C.R. (1995) Water budget and flow patterns in an urban wetland. Journal of Hydrology 169:171-187.

Paloheimo J., Fulthorpe R. (1987) Factors influencing plankton community structure and production in freshwater lakes. Canadian Journal of Fisheries and Aquatic Sciences 44:650-657.

Parsons L.R., Wheaton T.A., Castle W.S. (2001) High application rates of reclaimed water benefit citrus tree growth and fruit production. HortScience 36:1273-1277.

Pathak H., Mohanty S., Jain N., Bhatia A. (2010) Nitrogen, phosphorus, and potassium budgets in Indian agriculture. Nutrient Cycling in Agroecosystems 86:287-299.

Pathan S., Aylmore L., Colmer T. (2002) Reduced leaching of nitrate, ammonium, and phosphorus in a sandy soil by fly ash amendment. Soil Research 40:1201-1211.

Paul M.J., Meyer J.L. (2001) Streams in the urban landscape. Annual Review of Ecology and Systematics:333-365.

Peterjohn W.T., Correll D.L. (1984) Nutrient dynamics in an agricultural watershed: observations on the role of a riparian forest. Ecology 65:1466-1475.

155

Peterson B.J., Wollheim W.M., Mulholland P.J., Webster J.R., Meyer J.L., Tank J.L., Martí E., Bowden W.B., Valett H.M., Hershey A.E., McDowell W.H., Dodds W.K., Hamilton S.K., Gregory S., Morrall D.D. (2001) Control of Nitrogen Export from Watersheds by Headwater Streams. Science 292:86-90. DOI: 10.1126/science.1056874.

Petrovic A.M. (1990) The fate of nitrogenous fertilizers applied to turfgrass. Journal of Environmental Quality 19:1-14.

Pitt R., Maestre A., Morquecho R. (2004) The national stormwater quality database (NSQD, version 1.1), 1st Annual Stormwater Management Research Symposium Proceedings. pp. 13-51.

Poor N., Pribble R., Greening H. (2001) Direct wet and dry deposition of ammonia, nitric acid, ammonium and nitrate to the Tampa Bay Estuary, FL, USA. Atmospheric Environment 35:3947-3955.

Qian Y., Bandaranayake W., Parton W., Mecham B., Harivandi M., Mosier A. (2003) Long-term effects of clipping and nitrogen management in turfgrass on soil organic carbon and nitrogen dynamics. Journal of Environmental Quality 32:1694-1700.

Qin B., Xu P., Wu Q., Luo L., Zhang Y. (2007) Environmental issues of lake Taihu, China. Hydrobiologia 581:3-14.

Raciti S.M., Groffman P., Fahey T. (2008) Nitrogen retention in urban lawns and forests. Ecological Applications 18:1615-1626.

Ramirez-Fuentes E., Lucho-Constantino C., Escamilla-Silva E., Dendooven L. (2002) Characteristics, and carbon and nitrogen dynamics in soil irrigated with wastewater for different lengths of time. Bioresource Technology 85:179-187.

Rao P.P., Khemani L., Momin G., Safai P., Pillai A. (1992) Measurements of wet and dry deposition at an urban location in India. Atmospheric Environment. Part B. Urban Atmosphere 26:73-78.

Ravishankara A., Daniel J.S., Portmann R.W. (2009) Nitrous oxide (N2O): the dominant ozone-depleting substance emitted in the 21st century. Science 326:123-125.

Reddy K. (1983) Fate of nitrogen and phosphorus in a waste-water retention reservoir containing aquatic macrophytes. Journal of Environmental Quality 12:137-141.

Reddy K., Fisher M., Ivanoff D. (1996) Resuspension and diffusive flux of nitrogen and phosphorus in a hypereutrophic lake. Journal of Environmental Quality 25:363- 371.

156

Reeve N., Toumi R. (1999) Lightning activity as an indicator of climate change. Quarterly Journal of the Royal Meteorological Society 125:893-903.

Reidsma P., Tekelenburg T., Van den Berg M., Alkemade R. (2006) Impacts of land-use change on biodiversity: an assessment of agricultural biodiversity in the European Union. Agriculture, ecosystems & environment 114:86-102.

Robertson D.M., Roerish E.D. (1999) Influence of various water quality sampling strategies on load estimates for small streams. Water Resources Research 35:3747-3759.

Rowland F.S., Molina M.J. (1975) Chlorofluoromethanes in the environment. Reviews of Geophysics 13:1-35. DOI: 10.1029/RG013i001p00001.

Rusjan S., Brilly M., Mikoš M. (2008) Flushing of nitrate from a forested watershed: An insight into hydrological nitrate mobilization mechanisms through seasonal high- frequency stream nitrate dynamics. Journal of Hydrology 354:187-202.

Ryther J.H., Dunstan W.M. (1971) Nitrogen, phosphorus, and eutrophication in the coastal marine environment. Science 171:1008-1013.

Sacks L., Lee T., Radell M. (1994) Comparison of energy-budget evaporation losses from two morphometrically different Florida seepage lakes. Journal of Hydrology 156:311-334.

Sansalone J.J., Cristina C.M. (2004) First flush concepts for suspended and dissolved solids in small impervious watersheds. Journal of Environmental Engineering 130:1301-1314.

Saunders D., Kalff J. (2001) Nitrogen retention in wetlands, lakes and rivers. Hydrobiologia 443:205-212.

Savanick S., Baker L., Perry J. (2007) Case study for evaluating campus sustainability: nitrogen balance for the University of Minnesota. Urban Ecosystems 10:119-137.

Schilling K., Zhang Y.-K. (2004) Baseflow contribution to nitrate-nitrogen export from a large, agricultural watershed, USA. Journal of Hydrology 295:305-316. DOI: http://dx.doi.org/10.1016/j.jhydrol.2004.03.010.

Schlesinger W.H. (1997) Biogeochemistry. Geotimes 42:44-44.

Schumann U., Huntrieser H. (2007) The global lightning-induced nitrogen oxides source. Atmospheric Chemistry and Physics 7:3823-3907.

157

Schwientek M., Osenbrück K., Fleischer M. (2013) Investigating hydrological drivers of nitrate export dynamics in two agricultural catchments in Germany using high- frequency data series. Environmental earth sciences 69:381-393.

Searchinger T., Heimlich R., Houghton R.A., Dong F., Elobeid A., Fabiosa J., Tokgoz S., Hayes D., Yu T.-H. (2008) Use of US croplands for biofuels increases greenhouse gases through emissions from land-use change. Science 319:1238- 1240.

Shen J., Tang A., Liu X., Fangmeier A., Goulding K., Zhang F. (2009) High concentrations and dry deposition of reactive nitrogen species at two sites in the North China Plain. Environmental Pollution 157:3106-3113.

Shepon A., Gildor H., Labrador L., Butler T., Ganzeveld L., Lawrence M. (2007) Global reactive nitrogen deposition from lightning NOx. Journal of Geophysical Research: Atmospheres (1984–2012) 112.

Shih G., Abtew W., Obeysekera J. (1994) Accuracy of nutrient runoff load calculations using time-composite sampling. Transactions of the ASAE (USA).

Shoji S., Kanno H. (1994) Use of polyolefin-coated fertilizers for increasing fertilizer efficiency and reducing nitrate leaching and nitrous oxide emissions. Fertilizer Research 39:147-152.

Shuman L. (2004) Runoff of nitrate nitrogen and phosphorus from turfgrass after watering-in. Communications in soil science and plant analysis 35:9-24.

Silva S., De Oliveira R., Soares J., Mara D.D., Pearson H. (1995) Nitrogen removal in pond systems with different configurations and geometries. Water Science and Technology 31:321-330.

Sim C.H., Yusoff M.K., Shutes B., Ho S.C., Mansor M. (2008) Nutrient removal in a pilot and full scale constructed wetland, Putrajaya city, Malaysia. Journal of Environmental Management 88:307-317.

Smil V. (1999) Nitrogen in crop production: An account of global flows. Global Biogeochemical Cycles 13:647-662.

Smith D., Owens P., Leytem A., Warnemuende E. (2007) Nutrient losses from manure and fertilizer applications as impacted by time to first runoff event. Environmental Pollution 147:131-137.

Smith K., McTaggart I., Tsuruta H. (1997) Emissions of N2O and NO associated with nitrogen fertilization in intensive agriculture, and the potential for mitigation. Soil Use and Management 13:296-304.

158

Smith V.H., Tilman G.D., Nekola J.C. (1999) Eutrophication: impacts of excess nutrient inputs on freshwater, marine, and terrestrial ecosystems. Environmental Pollution 100:179-196.

Snyder G., Augustin B., Davidson J. (1984) Moisture sensor-controlled irrigation for reducing N leaching in bermudagrass turf. Agronomy journal 76:964-969.

Socolow R.H. (1999) Nitrogen management and the future of food: lessons from the management of energy and carbon. Proceedings of the National Academy of Sciences 96:6001-6008.

Solomon S., Qin D., Manning M., Chen Z., Marquis M., Averyt K., Tignor M., Miller H. (2007) IPCC, 2007: climate change 2007: the physical science basis. Contribution of Working Group I to the fourth assessment report of the Intergovernmental Panel on Climate Change.

Starr J., DeRoo H. (1981) The fate of nitrogen fertilizer applied to turfgrass. Crop Science 21:531-536.

Starrett S., Christians N., Austin T.A. (1995) Fate of amended urea in turfgrass biosystems 1. Communications in Soil Science & Plant Analysis 26:1595-1606.

Stenstrom M.K., Kayhanian M. (2005) First flush phenomenon characterization, California Department of Transportation Division of Environmental Analysis.

Stevens C.J., Dise N.B., Mountford J.O., Gowing D.J. (2004) Impact of nitrogen deposition on the species richness of grasslands. Science 303:1876-1879.

Stone K., Hunt P., Novak J., Johnson M., Watts D. (2000) Flow-proportional, time- composited, and grab sample estimation of nitrogen export from an eastern Coastal Plain watershed. Transactions of the ASAE 43:281-290.

Swancar A., County H. (1996) Water quality, pesticide occurrence, and effects of irrigation with reclaimed water at golf courses in Florida US Department of the Interior, US Geological Survey.

Swistock B.R., Edwards P.J., Wood F., Dewalle D.R. (1997) Comparison of methods for calculating annual solute exports from six forested Appalachian watersheds. Hydrological Processes 11:655-669.

Takeda K., Fujiwara K. (1993) Determination of nitrate in natural waters with the photo- induced conversion of nitrate to nitrite. Analytica chimica acta 276:25-32.

Taylor G.D., Fletcher T.D., Wong T.H., Breen P.F., Duncan H.P. (2005) Nitrogen composition in urban runoff—implications for stormwater management. Water Research 39:1982-1989.

159

Thomas R., Foster C., Cohen M., Martin J., Delfino J. (2010) Dissolved organic carbon interferences in UV nitrate measurements and possible mitigation methods, AGU Fall Meeting Abstracts. pp. 0483.

Tilman D. (1999) Global environmental impacts of agricultural expansion: the need for sustainable and efficient practices. Proceedings of the National Academy of Sciences 96:5995-6000.

Tong S.T., Chen W. (2002) Modeling the relationship between land use and surface water quality. Journal of Environmental Management 66:377-393.

Turner R., Dortch Q., Justic D., Swenson E. (2002) Nitrogen loading into an urban estuary: Lake Pontchartrain (Louisiana, USA). Hydrobiologia 487:137-152.

Turner T.R., Hummel N.W. (1992) Nutritional requirements and fertilization. Turfgrass:385-439.

USBR. (1997) Water measurement manual, in: B. o. R. U. United States Department of the Interior (Ed.), Denver.

Vázquez N., Pardo A., Suso M., Quemada M. (2006) Drainage and nitrate leaching under processing tomato growth with drip irrigation and plastic mulching. Agriculture, ecosystems & environment 112:313-323.

Vitousek P.M., Howarth R.W. (1991) Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry 13:87-115.

Watson C., Atkinson D. (1999) Using nitrogen budgets to indicate nitrogen use efficiency and losses from whole farm systems: a comparison of three methodological approaches. Nutrient Cycling in Agroecosystems 53:259-267.

Webb B., Walling D. (1985) Nitrate behaviour in streamflow from a grassland catchment in Devon, UK. Water Research 19:1005-1016.

Weiss P., Eveborn D., Kärrman E., Gustafsson J.P. (2008) Environmental systems analysis of four on-site wastewater treatment options. Resources, conservation and recycling 52:1153-1161.

Whipple Jr W., Hunter J., Yu S. (1977) Effects of storm frequency on pollution from urban runoff. Journal (Water Pollution Control Federation):2243-2248.

Wilson K.B., Hanson P.J., Mulholland P.J., Baldocchi D.D., Wullschleger S.D. (2001) A comparison of methods for determining forest evapotranspiration and its components: sap-flow, soil water budget, eddy covariance and catchment water balance. Agricultural and Forest Meteorology 106:153-168.

160

Withers P.J., Jordan P., May L., Jarvie H.P., Deal N.E. (2013) Do septic tank systems pose a hidden threat to water quality? Frontiers in Ecology and the Environment 12:123-130.

Wong J., Chan C., Cheung K. (1998) Nitrogen and phosphorus leaching from fertilizer applied on golf course: Lysimeter study. Water, Air, and Soil Pollution 107:335- 345.

Wu J.S., Allan C.J., Saunders W.L., Evett J.B. (1998) Characterization and pollutant loading estimation for highway runoff. Journal of Environmental Engineering 124:584-592.

Wu J.S., Holman R.E., Dorney J.R. (1996) Systematic evaluation of pollutant removal by urban wet detention ponds. Journal of Environmental Engineering 122:983- 988.

Xiankai L., Jiangming M., Shaofeng D. (2008) Effects of nitrogen deposition on forest biodiversity. Acta Ecologica Sinica 28:5532-5548.

Young W.J., Marston F.M., Davis R.J. (1996) Nutrient exports and land use in Australian catchments. Journal of Environmental Management 47:165-183.

Zhang H., Sansalone J. (2014) Partitioning and First-Flush of Nitrogen in Rainfall-Runoff from an Urban Source Area. Journal of Environmental Engineering 140.

Zhang Z., Fukushima T., Onda Y., Gomi T., Fukuyama T., Sidle R., Kosugi K.i., Matsushige K. (2007) Nutrient runoff from forested watersheds in central Japan during typhoon storms: implications for understanding runoff mechanisms during storm events. Hydrological Processes 21:1167-1178.

Zhao W.Z., Xiao H.L., Liu Z.M., Li J. (2005) Soil degradation and restoration as affected by land use change in the semiarid Bashang area, northern China. CATENA 59:173-186.

Zhao X., Zhou Y., Min J., Wang S., Shi W., Xing G. (2012) Nitrogen runoff dominates water nitrogen pollution from rice-wheat rotation in the Taihu Lake region of China. Agriculture, ecosystems & environment 156:1-11.

161

BIOGRAPHICAL SKETCH

Jiexuan Luo was born in 1985 in Chenzhou, China. Her interest in environmental science developed at early stages of her life. She received a B.S. degree in Aquaculture

Technology and Science (2011) from Huazhong Agricultural University, China. Then in

2008 she joined Program of Fisheries and Aquatic Science at University of Florida,

Gainesville, FL where she gained a master degree in Fisheries and Aquatic Science.

After she developed a strong interest in water sciences during the master’s study, she decided to apply for the doctoral program in School of Natural Resources and

Environment at University of Florida under the supervision of Dr. George J. Hochmuth.

She started the pursuing of doctoral degree since 2011, now she will complete her doctoral degree in summer 2015.

162