NITROGEN FLUXES AND DYNAMICS IN AN URBAN WATERSHED By 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
2015 Jiexuan Luo
To my family
4 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 ac knowledge the members of my supervisory committee, Dr. Mark W. Clark, Dr. Matthew J. Cohen and Dr. John J. Sansalone for their input s throughout my study. Special thanks go to D awn 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 g uiding me with tracer studies. Thanks also go to Hao Zhang for his valuable suggestions and inf ormation 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 M rs. 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.
5 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 Stormw ater 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 Precip itation Measurement ............................................................................... 51 Baseflow and Stormflow Sampling ................................................................... 51 Data Analysis ................................................................................................... 53 Results And Discussion .......................................................................................... 54 Hydrol ogical Factors in the Three Catchments ................................................. 54
6 Rainfalldischarge 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 A pproach ............................................................................ 68 Flow Measurement/Estimation ......................................................................... 69 Baseflow Sampling ........................................................................................... 70 Data Analysis ................................................................................................... 71 Results and Di scussions ......................................................................................... 72 Discharge for Each Catchment ......................................................................... 72 Water Budgets for Lake Alice Watershed ......................................................... 75 Lake Alice N Budget Estimation ....................................................................... 77 Spatial Changes in NO3N 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 Compar isons Of Autosampler/SUNA NO3N Concentrations in Storm Events ........................................................................................................... 98 Event Mean Concentrations from Autosamplers and SUNAs ......................... 105 Event NO3N Mass ......................................................................................... 106 Conclusions .......................................................................................................... 108 5 CHARACTERIZ ATION 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
7 Flow Measurement ......................................................................................... 115 Precipitation Measurement ............................................................................. 116 High Resolution Nutrient Analysis .................................................................. 118 Conventional Rating Curves Between Discharge And NO3N Conce ntration . 118 Characterization of Hysteresis Trajectories .................................................... 119 Results and Discussions ....................................................................................... 121 Hydrological Factors for Storms ..................................................................... 121 NO3N ConcentrationDischarge Relationship ................................................ 123 Changes in NO3N Concentrations through Individual Storms ....................... 126 Conclusions .......................................................................................................... 136 6 SYNTHESIS .......................................................................................................... 138 LIST OF REFE RENCES ............................................................................................. 141 BIOGRAPHICAL SKETCH .......................................................................................... 162
8 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 20132014 .......... 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 31 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 Al ice 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. NO3N concentrations were obtai ned 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 NO3N concentration (C) and hysteresis patterns in SFC in each storm ......................................................... 128
9 5 4 A log log regression between discharge (Q) and NO3N concentration (C) and hysteresis patterns in RWC in each storm ........................................................ 129
10 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 Rainfalldischarge relationship in all catchments .................................................... 57 2 5 Discharge NO3N load relationship in all catchments .............................................. 58 2 6 Discharge TKN load relationship in all catchments ................................................. 58 3 1 Major stream basins in Lake Alice waters hed ......................................................... 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 NO3N concentration in Fraternity Stream. ............................................................................................................... 82 3 6 Boxplots representing the distribution of NO3N concentration in Hume Creek. ...... 84 3 7 Boxplots representing the distribution of NO3N 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
11 4 5 Changes in NO3N monitored by SUNA and autosampler in Reclaimed Water Irrigated Catchment (RWC). ............................................................................. 100 4 6 Changes in NO3N monitored by SUNA and autosampler in Sports Field Catchment (SFC). ............................................................................................. 102 4 7 Event mean concentrations (NO3N) determined by the autosampler and SUNA . 105 4 8 Event NO3N 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. NO3N 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 NO3N concentration (C) discharge (Q) relationships in Sports Field Catchment (SFC). ............................................................................................................... 125 5 5 NO3N concentration (C) discharge (Q) relationships in Reclaimed Water Irrigated Catchment (RWC). ............................................................................. 126 5 6 NO3N hysteresis loops in individual storm events in SFC .................................... 133 5 7 NO3N hysteresis loops in individual storm events in RWC .................................. 135 5 8 An example of complex storms with multiple loops. .............................................. 135
12 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 By 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, t he 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 contr olling 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 ur ban 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 yr1) 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
13 fertilization/irrigation practices were the major drivers for the great N load. Thir dly, 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 NO3N concentrations was discussed. The results sugg ested 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 Lak e 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 imme diate signals for NO3N response to hydrological changes. Finally the relationship between NO3N concentrations and discharge was examined by SUNAs. There was a significant relationship between NO3N concentrations and discharge. NO3N concentration changes during storms were clockwise loops, suggesting immediate dilution effect. It can also be inferred from SUNAâ€™s curves that the increasing NO3N concentrations after storms were attributed to NO3N rich groundwater rather than surface runoff.
14 CHAPTER 1 LIT ERATURE 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 preindustrial times. Global N c ycling 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 ma nagement 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 o f Major Global N Fluxes Two Natural Processes t o 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 NOxN, 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) .
15 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 53 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 lightningâ€™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 l east 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 databased 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 yr1 ran ging from 100 to 290 Tg N yr1, which falls within the range of 44 to 220 Tg N yr1 reported by Schlesinger (1997) .
16 Human Alterations o f 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 cy cle. 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 ferti lizers. 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) con cluded 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
17 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 a l., 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 19742007 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 Bet ter 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) .
18 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 nitrif ication in the agricultural soil are the major contributors to the emissions of N2O and NOx ( Socolow, 1999) . Agric ultural 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 ozonedepleting 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
19 runoff or leaching, respectively. Excessive N O3N 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 NO3N can also leach through the soil and potentially contaminate the groundwater used for drinking water, thr eatening human health, reported as â€œblue babyâ€ 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 a pproach 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 a nd 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
20 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 agroecosyst ems: 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 humaninterrupted systems. Therefore, regular uncertainty analyses are highly recommen ded in nutrient budget studies.
21 Major N Fluxes i n 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 watershedlevel 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 domin ate 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 di oxide (NO2), nitric oxide (NO), ammonia (NH3), particulate nitrate (NO3 -) and particulate ammonium (NH4 +). N deposition datasets have been well established nationwide. For
22 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 maj or 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 en tirely 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 urb an city in India was approximately 6 kg N ha1 yr1; an estimation of atmospheric N deposition in Tampa Bay, a large urban area ranged from 6 to 8.6 kg ( Poor et al., 2001) . ( L 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 ha1 yr1 and the overall average deposition rate was 12.89 kg N ha1 yr1. 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
23 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 avai lability 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 L1 to 29 mg L1, 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 yr1 ( 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) .
24 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 foun d 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) summar ized the N fertilizer application rates in Florida (Table 11), demonstrating that N fertilizer input to residential lawns ranged from 80 to 240 kg N ha1 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 ha1 yr1) similar to pastures (240 to 360 kg N ha1 yr1), 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) .
25 Table 1 1 . Fertilization rates in Florida ( Cohen et al., 2007) Type of Application N application rates (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, wheat 150 210, 90 and 80 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 (NO3N, 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 soilwater 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 L1) compared to other seasons (e.g. 0.2 mg L1 in the fall) in the same year. Other studies i n Rhode Island and Michigan reported higher concentrations of nitrateN in soil water in the fall compared to spring ( Liu et al., 1997; Miltner et al., 1996) . Overall annual N leaching rates for turfgrass
26 ranged from 0 to160 kg N ha1, 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 ur ban 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 m2 yr -1) but high in eutrophic lakes (0.2 to 9.2 g N m2 yr1) ( 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 m2 yr1 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
27 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 fer tilizer and/or soil N. This result was partly in agreement with Kaushal et al. (2006) who foun d 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 greenhousegas 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 m ay 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 nonpoint source pollution.
28 Table 1 2 . Summary of stormwater N datasets included in National Stormwater Quality Database( Pitt et al., 2004 ) Land use NH 3 N (mg L 1 ) NO 2,3 N (mg L 1 ) TKN (mg L 1 ) TN (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 12 ( Pitt et al., 2004 ) . Flint and Davis (2007) also claimed that NO3N concentration ranged
29 from 0.15 to 4.32 mg L1 and TKN concentration ranged from 0.8 to 10 mg L1 in the highway from an ultraurban 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 runoff with NO3N concentrations up to 3.57 mg L1 and TKN concentrations up to 19.5 mg L1 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
30 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 turfg rass 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 ha1 yr1) when clippings were recycled. Recently, Kopp and Guillard (2002) reported that turf plots receiving 0 to 98 kg N ha1 yr1 with clippings returned produced a comparable quality and clippings yields to plots receiving 390 kg N ha1 yr1 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
31 management scenarios of low N fertilization (150 kg N ha1 yr1) with clippings returned based on the results from CENTURY model. The frequency of fertil izer 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, suc h 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 w ater to move beyond the active rooting zone ( Al Kaisi and Yin, 2003; Gheysari et al., 2009; Nakamura et al., 2004; Vzquez et al., 2006) . In a 2year 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 avoi d 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 w as 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.
32 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 sandbased root zones cannot retain enough water and nutrients to support turfgrass vigor. Organic amendments like sphagnum peat moss are often incorporated into sandbased 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 â€“, 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 perform ance 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
33 area of biochar ( Glaser et al., 2002 ; Laird et al., 2010; Lehmann et al., 2003) . The only concern for soil amendments is t hat 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. Stormwa ter 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 treatm ent 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 ammoniumâ€™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 6hour 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 pract ice ( 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. .
34 Bioretention is also associated with high TKN removal. Wu et al. (1996) reported a removal effic iency 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. W etlands 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 nitrateâ€“ N in Putrajaya Wetlands where plenty of vegetation was planted in Malaysia, indicating that vegetation uptake had a huge impact on N removal.
35 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 hor izontal 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 ( Khler, 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, som e 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 greenroofsâ€™ 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 L1 with average about 3.6 mg L1). Berndtsson et al. (2009) compared the performance of nutrient removal in an intensive vegetated greenroof
36 (maintainancerequired with deep soil layers) and an extensive vegetated greenroof (maintenanc e 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 o f High Resolution In Situ Nitrate Sensor The summary of N sources and N m anagement above shows the various pathways that N flows through in urban ecosystems. To understand complex nutrient processes on watershedscale 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 fieldbased automated analyzer technologies begin to be deployed, to allow for routine highresolution 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 i n situ nitrate concentrations is introduced here with comparison with conventional sampling approaches and conventional analysis methods.
37 Sampling Tec hniques a nd 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 p athways. 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 locatio n, 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
38 requiring sampling of first flush, manual sampling is usually preferred since it has greater flexibility and allows larger sample volumes to be collected as w ell 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) timeweighted, and (3) user defined. Once it is determined how samples will be collected, the n ext 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 timecomposite sampling. They found that when flow and concentrations were positively correlated, the timecomposited 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 signific antly 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 uncert ainties regarding the
39 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 C onventional ni trate/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 a ppears 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
40 further reduction to nitric oxide ( Morton et al., 1988) . Photoinduction reduction is novel in nitratenitrite conversion. Takeda and Fujiwara (1993) reported t he formation of nitrite and oxygen from nitrate as shown below by ultraviolet (UV) light irradiation in the wavelength between 200 â€“ 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 ) . 3 2 + 1 2 2 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 ( Bijoo r 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
41 capture of N dynamics both temporally and spatially. Hensley et al. (2014) estimated N removal in large rivers, mapped the nitrate concentrations on highresolution longitudinal profiling by using in situ nitrate sensor, and concluded that the N removal was controlled by spatial variation in chann el 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 IonSelective Electrode sensor from InSitu 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 h ave enormously altered the N cycling mainly with the additions of artificial N
42 fixation (fertilizer) and combustion engine. What is new about the alteration is the magnitude. In 1970, anthropogenic activities including fertilizer production, fossil fuel co mbustion and legume and rice production mobilized about 70 Tg N yr1. By the end of 20th century, the number had doubled to about 140 Tg N yr1. 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 â€œbluebabyâ€ 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, meanw hile, 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 t he 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
43 obtain continuous information regarding nutrient dynamics. In situ sensors can provide a new interpr etation of hydrological and biogeochemical processes through the continuous data, therefore it may open a new era for future nutrient control and management studies.
44 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 mak e s 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 groundwat er 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) . W atershed studies can be used to identify sources and fates of nutrient s 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 hi gher 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
45 deposition. It is important to understand t he 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 bui ld 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 ( vila et al., 1992) . Little is known about the nutrient imports and exports in an urban watershed comprised of area s 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 B aseflow I ndex (BFI) had a lower Root Mean Square Error ( RMSE ) on load estimates when sampled infrequently than those with a low BFI . I nfrequent 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 w ere: 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
46 from each catchment . A number of studies have been conducted to det ermine 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 Cis ar, 2001 ) than other catchments. T herefore, m y 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 relationship s 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 hy drology monitoring instead (monitoring rainfall and flow). Materials a nd 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 21) 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.
47 Fi gure 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 nat ion 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 mor e 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 ferti lization. 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
48 A ssociation , 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 21) . The urban Reclaimed Water Irrigated Catchment (RWC) is in the center east ern part of the campus with dormitories and lab/classroom buildings including a large number of landscape d sites with an approximate area of 1. 5 ha with 76% pervious surface. The irrigation with reclaimed water is applied twice a week in t his 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 fami lies are located with an approximate area of 2. 6 ha. This area is comprised of 76% pervious surface. R unoff 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 ove rflows to two injection well s connected with the groundwater system. Table 2 1 . Pervious/impervious ratio in each Lake Alice catchment Land Use 1 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 NO3N and TKN concentrations. A
49 log log relationship was established between the N load s 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. 5 cm , 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 T hel Mar volumetric weir s (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 22 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 incoming 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 21). All t hree 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
50 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 calc ulation provided by Thel Mar LLC, therefore, the fullpipe discharge was calculated based on Manningâ€™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. 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. a b c
51 Table 2 2 . Completeness statistics for discharge records from each catchment through the monitoring period with annual totals for the hydrological year 20132014 Land use 1 Start of record End of record No. of measurements No. of expected measurements Completeness2 % CC 9 / 1 /1 3 0 : 00 8 / 31 /14 2 3 : 5 5 85590 105120 81.4 RWC 9 / 1 /13 0 : 00 8 / 31 /14 2 3 : 5 5 97950 105120 93.2 SFC 9 /1/13 0 : 00 8 / 31 /14 2 3 : 5 5 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 n umber of expected measurements. % Completeness = 100 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 Phys ics Department which was also recorded by a tipping bucket. Three 4.73L plastic buckets were put near the study areas to catch the rainfall water for TKN and NO3N analyses. Baseflow a nd Stormflow Sampling Baseflow water samples were collected for one hy drological 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 I nc., 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
52 days, except in F ebruary 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 timeproportional 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 to 10 mi nutes . E very 3 to 4 samples comprised a 500 ml composite sample depending on the dura tion of the storms . After the storms, t he discharge peak s 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 5min samples) and averaged discharge during the compos ite 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 fro m the field to the laboratory within 12 hours after the storm for subsampling.
53 Sub samples (two 20mls 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 precleaned by acid washing followed by distilleddeionized water rinses. NO3N sub samples were filtered before submitting for analyses . Sulfuric acid was added to each subsample to ensure a pH<2 before they were sent to ARL. NO3N was measured in every baseflow sub sample. Total Kjehldahl Nitrogen (TKN) was measured twice in 2013, and then twice a month since February 2014 in baseflow sub samples. Both NO3N and TKN were measured in every stormwater sub sample . Total N concentration was calculated as the sum of TKN concentration and NO3N concentration. The Method Detection Limits (MDLs) in Analytical Research Laboratory (ARL) were 0.148 mg L1 for NO3N, and 0.125 mg L1 for TKN. Concentrations below MDL s were assumed to have i mputed 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 ) Where is the slope and is the intercept. A log log transformed linear relationship was established between N loads (MNO3 N for the mass of NO3N, MTKN for the mass of TKN) and discharge (Q) in each catchment: ln = ln + ln ( 2 2 ) Where a and b are regression constants.
54 N loads (NO3N load and TKN load) were calculated based on E quation 22 with known discharge records. The missing discharge was calculated from E quation 2 1 and then used to calculate the N loads based on E quation 2 2. Results a nd Discussion Hydrological Factors in t he 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 gaug es. The distribution of total daily precipitation (Figure 23) 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.
55 Storm descriptions . 1 5 storms were sampled from February 2014 to August 2014 with precipitation ranging from 1.5 mm to 46 mm (Table 23 ). 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 23), 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 st orms 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 Date Precipitation 1 (mm) Previous Dry hours(hr) Imax (mm hr -1 ) NO 3 N ( mg L -1 ) TKN ( mg L -1 ) TN ( mg L -1 ) 2/5/14 5.6 114 91.4 0.2 0.3 0.5 2/21/14 3.8 148 45.7 0.1 2 0.2 0.3 2/21/14 32.5 2 106.7 0.1 2 0.2 0.3 3/17/14 46.01 238 76.2 0.2 0.1 2 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 8/30/14 41.7 12 91.4 0.1 2 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. 2Values with subscripted 2 were below Method Detection Limit (NO3N=0.148 mg L-1, TKN= 0.125 mg L-1) in UF Analytical Research Laboratory, and were substituted with half of the MDL for statistical analysis (NO3N=0.074 mg L-1, TKN= 0. 063 mg L-1).
56 The maximum intensity occurred in two storms (5/25/14 and 5/30/14) with a value of 121.9 mm hr1. The previous dr y 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 w ere recorded for CC. Rainfalldischarge R elationship 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) (Fi gure 24) 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 wit h 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.
57 Figure 2 4 . Rainfalldischarge 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 NO3N load (MNO3 N) (Figure 2 5) , and between discharge (Q) and TKN load (MTKN) (Figure 26) in each catchment. The results showed a strong positive relationship between discharge and NO3N load, suggesting the NO3N loads increased as the discharge increased in each catchment. The dischargeNO3N load relationship was weaker in SFC (R2=0.64) compared with CC and RWC . This relationship may be associated with high NO3N concentration in bas eflow (Table 24) which possibly created a larger NO3N 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. Precipitation, mm (P) Average daily discharge (Qd), L/s Land use =CC Qd =0.61+0.05*P R 2 =0.12, p<0.0001* Land use =RWC Qd=0.32+0.06*P R2=0.77, p<0.0001* Land use =SFC Qd=0.47+0.22*P R 2 =0.87, p<0.0001*
58 Figure 2 5 . Discharge NO3N load relationship in all catchments (CC=Control Catchment, RWC=Reclaimed Water Irrigated Catchment, SFC= Sports Field Catchment). * Significance level p=0.05 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 use 1 CC RWC SFC NO 3 N min , mg L 1 0.07 0.07 0.17 NO 3 N max , mg L 1 2.59 1.48 15.61 TKN min , mg L 1 0.06 0.06 0.06 TKN max , mg L 1 4.75 3.22 2.92 Q min , L s 1 0.01 0.06 0.14 Q max L s 1 34.66 60.89 251.98 1 Control catchment (CC), Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC). NO 3 N load, mg (M NO3 N ) Discharge, L s 1 (Q ) Land use =CC L n (MNO3 -N)= 1. 52+0.77*Ln(Q) R 2 =0.84, p<0.0001* Land use = RWC Ln( MNO3 -N) = 0 . 52+0.68*Ln(Q) R 2 =0. 79 , p<0. 0 001* Land use = SFC Ln( MNO3 -N) = 1.98+0.52*Ln(Q) R 2 =0. 64 , p<0. 0 001* Discharge, L s 1 (Q ) TKN load, mg (M TKN ) Land use =CC L n (MTKN)= 0.40+1.15*Ln(Q) R 2 =0. 89 , p<0.0001* Land use = RWC L n (MTKN)= 1.02+1.19*Ln(Q) R 2 =0. 89 , p<0.0001* Land use = SFC L n (MTKN)= 0.11+1.13*Ln(Q) R 2 =0. 86 , p<0.0001*
59 Based on the dischargeN load relationship and precipitationdischarge relationship, N loads from all catchments were calculated and listed in Table 25. The results (Table 25 ) 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 ha1 depending on the sources of fertilizers and the frequency of application ( Miller and Cisar, 2001) . Calculations from Table 25 indicated that up to 20 % of the total annual N applications could be lost in SFC, assuming a fertilizer application rate of 200 kg ha1. 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 ha1 yr1 despite reclaimed water irrigation was practiced in RWC .
60 Table 2 5 . Annual N loads estimation for each catchment from 2013 to 2014 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 NO3N in T N contribution in CC, but NO3N contribution was greater in RWC. In the annual N contribution, CC and RWC only had 3 kg ha1 yr1 difference. Therefore, it can be inferred that using reclaimed water as an irrigation and fertilizer source is a n 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 t he environmental hazards it might pose from phosphorus and other chemicals from reclaimed water. Like agricultural watersheds/catchments, S ports F ield C atchment is heavily fertilized and irrigated, also it has a high capability of leaching losses . This study illustrated that SFC had the highest NO3N 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 ha1 yr1, which was about t he same or even larger than agricultural watersheds. Catchments like SFC are usually overlooked Land use1 CC RWC SFC NO 3 N (kg ha -1 ) TKN (kg ha -1 ) TN (kg ha -1 ) NO 3 N (kg ha -1 ) TKN (kg ha -1 ) TN (kg ha -1 ) NO 3 N (kg ha -1 ) TKN (kg ha -1 ) TN (kg ha -1 ) 1.87 7.54 9.41 6.81 5.49 12.31 28.54 8.58 37.12
61 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 n umber 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 wat er quality sampling. The similar relationships can also be used in other catchment study.
62 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 L1. 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 practi ces for improved N uptake efficiency and N loss reduction. With the increasing interests in humandominated urban ecosystems, the mass budget approach has also been carried out to evaluate urban ecosystemsâ€™ function and structure. For example, Baker et al. (2001) constructed a N budget for different subsystems under the Central ArizonaPhoenix (CAP) ecosystem and s u ggest ed 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
63 discovered that the urban wat ershed 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 di verse. 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 humaninterrupted systems ( Johnes, 2007) although humaninterrupted 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 humando minated 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 potenti ally 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
64 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 surfaceto 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 appl ications in SFC, assuming a fertilizer application rate of 200 kg ha1. 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 f luxes 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. nitrateN concentrations decreased as the flow reached to
65 downstream from upstream; 3. The hydrological change from upstream to downstream attributed to the decrease of nitrateN 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 meth od to estimate the N budget under some uncertain circumstances. Materials a nd 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 â€œcityâ€ 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 studentsâ€™ 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 resi dential housing areas, Shands Hospital (university medical center), and
66 the UF Wastewater Treatment Plant. Eventually the three streams discharge to Lake Alice, a scenic 35ha retention pond on campus (Figure 31) 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 ex cept 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 water shed 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, respect ively (Figure 31). 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 landuse catchme nts (more details in Chapter 2). Fraternity Stream Upstream Catchment and Graham Woods Stream Catchment are mixed landuse catchments. The Hume Creek Basin includes the Graham Woods Stream Catchment and several small urban catchments that Reitz Union Str eam passed through (not shown in Figure 31). 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
67 catchments in Diamond Stream Basin. The area with white color in Figure 31 is the area in Lake Alice watershed except the three major basins. It is referred to as â€œOutside Basinsâ€ in this study. O utside Basins are mainly composed of dormitory buildings, academic buildings, landscape and parking lots like Diamond Stream basin. Table 31 shows the ratio of pervious area to impervious area in each basin. Figure 3 1 . Major stream basins in Lake Alice watershed
68 Table 3 1 . Proportion of pervious/impervious surface in each stream basin Fraternity Stream Hume Creek Diamond Stream Outside 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 O utside 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, a nd 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 Hu me 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.
69 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 31), Fraternity Stream upstream (Site No.2, Figure 31), and Graham Woods Stream downstream (Site No.7, Figure 31). The sensors were set up to measure the water level and velocity every 5 minutes. The data were uploaded to the v irtual system online ( www.ziscape.com , userâ€™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 Fr aternity Stream downstream (Site No.2, Figure 31) and Graham Woods Stream downstream (Site No.7, Figure 31) were calculated accompanied by the discharge data from single land use catchments (Control Catchment, Reclaimed Water Irrigated Catchment and Spor ts 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 uni t area in each catchment was used to extrapolate the discharge for each stream basin.
70 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 31, Table 32), 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 obtai ned from UF Clean Water Campaign dataset in 2014. The sampling frequency was at least twice a month for baseflow NO3N analysis in dry days, except in February 2014 when it kept raining around the expected second sampling time of that month. The specific d ay 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 , respecti vely.
71 Table 3 2 . Description of water quality monitoring sites . See Figure 31 for exact location of water sampling sites (blue dots on Figure 31). Site ID Site name Catchments/Basins 1 Lake Alice well N/A 2 Fraternity Stream upstream Fraternity Stream Upstream Catchment/Fraternity Stream Basin 3 Fraternity Stream 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 downstream Graham Woods Stream Catchment/ 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 2. Data Analysis The Kruskal Wallis test (a nonparametric analog of oneway ANOVA) was used to determine whether differences existed at p =0.05 among median monthly discharge f rom all the catchments, also the difference in N concentration (NO3N) 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
72 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: D ischarge per unit area (catchment) = total annual discharge (catchment) / catchment ar ea 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 calc ulated 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 a nd Discussions Discharge f or Each Catchment The monthly discharge data from September 2013 to July 2014 from all the flow monitored stations are displayed in Figure 32. The discharge rate from all the single landuse catchments and Fraternity Stream Upstream Catchment was below 4 L s1.
73 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 W oods 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 drainag e 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 Gra ham W oods S tream 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.
74 Figure 3 2 . Monthly discharge from each catchment and Lake Alice well Discharge per ha was calculated in each catchment (Figure 33). 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) (nonirrigated 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 smal l 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 nonirrigation. Discharge from Graham Woods Stream Catchment was significantly greater than that from Fraternity Stream Upstream C atchment ( 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 C atchment and SFC. 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 16 Lake Alice well flow rate, L s1 Catchment/basin flow rate, L s 1 Months Fraternity stream upstream catchment Reclaimed water irrigated catchment Sports fields catchment Control catchment Graham woods stream downstream basin Lake Alice well
75 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 signifi cantly different at the 5% level of probability). Water Budgets f or 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 conc entration 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 landuse catchments and mixed landuse catchments . In this study, median discharge per ha (in L s1 h a1) 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 w ith 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 s1 h a1. The estimated water yield from each stream basin is illustrated in Table 33 .
76 Table 3 3 . Water budget for Lake Alice Area, ha Pre cipitation water, m 3 Influx/efflux Annual water yield, m 3 Fraternity stream basin 9.5 152,000 Influx 9 3 ,000 Hume creek basin 65.6 1,049,000 Influx 641 ,000 Diamond stream basin 156.5 2,504,000 Influx 1, 5 30 ,000 Outside Basin s 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 ex ported 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 northcentral Florida) had a higher evaporation rate than the deeper lake (Lake FiveO 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 y r1 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.
77 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 twothirds of the total surface water evaportranspiration, and the transpiration rates ranged fr om less than 0.01 m yr1 to approximately 1.3 m yr1. 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 w ith both surfacewater and groundwater systems ( Motz et al., 2001) with the rate ranging from 0.12 to 4.27 m yr1 ( Motz, 1998) . The annual water yield from each stream basin accounted for 6 0% 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 31) 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 34.
78 Hume creek delivered 4 9 % of the total N loads (mainly NO3N) to Lake Alice (Table 34) 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 Hum e 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 whi ch 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 Median NO3N, mg L 1 Median TKN, mg L 1 Water yield, m 3 Annual NO3N loads, kg Annual TKN loads, kg Annual TN loads, kg (% in total influx) Basins Influx Fraternity stream 4.63 0.56 92 ,000 430 52 482 (11%) Hume creek 2.80 0.67 6 41,000 1,796 430 2,225 (49%) Diamond stream 0.02 0.43 1,530,000 24 658 682 (15%) Outside Basins 0.02 0.43 1, 900 ,000 38 817 855 (19%) 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 well 0.07 0.83 1,548,000 115 1,285 1,400 Sum 1,400 Retention 3,165
79 Table 3 5 . N loads from catchments in Hume Creek Basin Catchment Area, ha Median Discharge per area, L s 1 ha 1 Total annual discharge, m 3 Median NO3N, mg L 1 Median TKN, mg L 1 Annual TN load, kg Sports Field Catchment 6.7 0.21 45,000 12.21 a 0.75 a 583 b Graham Woods Stream Catchment 28.7 0.31 281,000 5.41 0.81 1 , 748 Hume Creek Basin 65.6 0. 3 1 c 6 41 ,000 2.8 0.67 2 , 2 25 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 whi le 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 clos e 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 m2 (calculated from annual precipitation amount and measured N concentration in precipitation) . This value was si milar to other studies in Florida ( Hendry and Brez onik, 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 NO3N concentration but very high TKN concentration compared to the three streams (Figure 34). This could be the result of the high
80 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 d isc depthâ€™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 bi omass 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 NO3N and reduced the NO3N 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 sedim ent could become suspended and make high TKN concentration.
81 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: denitrific ation, 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 0 0.5 1 1.5 2 2.5 TKN concentrations, mg L1 Water quality monitoring sites
82 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 differentiate preferential uptake of 15NH4 + and 15NO3 for different kinds of vascular aquatic macrophytes. His research showed that 34 to 40% of the added inorganic 15N (15NH4 + + 15NO3 ) was removed through plant uptake, the rest presumably lost through NH3 volatilization and nitrificationdenitrification processes. Spatial Changes i n NO3N Concentrations Figure 3 5 . Boxplots representing the distribution of NO3N 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 NO3N concentrations. NO3N concentrations were recorded in both upstream and downstream in each major stream . There was no significant
83 difference in NO3N concentration between Fraternity S tream upstream and Fraternity S tream downstream (Figure 3 5) . This might be because that this stream has shorter total length compared with the other streams, and also it has f ewer contributing pipes than other streams. On the contrary, the NO3N concentrations were significantly different among all the sampling sites in Graham W oods S tream. SFC in the upstream of Graham W oods S tream delivered a very high NO3N concentration w ith a n annual median of 12.21 mg L1 (Chapter 2), however, the concentration was reduced significantly as it approached the downstream sampling site s (Figure 3 6) . This reduction in NO3N concentration might be attributed to dilution from the seepage or dr ainage 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 35 ). In addition, flow in Graham Woods Stream passed through two small retention ponds before it reached the discharge monitoring station (Site No.7, Figure 31). Retention ponds may help to reduce NO3N concentration by vegetation uptake although the efficiency of nitrate removal was reported to be very low ( Hsieh and Davis, 2005; Silva et al., 1995) . A fter the stream water passed the discharge monitoring station, it merge d with the branch from the urban academic area (sampling site No.8, Figure 3 1) where reclaimed water irrigation w as regularly practiced. When the two branches combined in Hume creek, the NO3N concentrations were reduced even more (Figure 3 6) . In the Diamond Stream, the NO3N concentrations showed significant difference between the upstream ( Site No.11, Figure 31 ) to the downstream ( Site No.12, Figure 31 ) (Figure 37) , which is largely attributed to dilution because the downstream discharge was much higher than upstream discharge (visual observation
84 based on the changes in depth and width of the stream) . It is no ted that t he upstream NO3N concentrations in the Diamond S tream were approximately 10 times the NO3N concentrations in the downstream portion of the stream . Figure 3 6 . Boxplots representing the distribution of NO3N 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 NO3N concentrations in Hume creek and Diamond Stream were also demonstrated in several st udies as an indication or explanation for N mass changes . For example, Hager and Schemel (1992) found that the dissolved inorganic nitrogen (D IN) 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 t reatment plant contributed 70% DIN to the Northern San Francisco Bay. Kemp and Dodds (2001) also observed a huge increase in NO3N concentrations in the downstream sampling well and in stream branches from
85 surrounding croplands compared with the upland watershed characterized with tallgrass prairies , indicating there was an increase in N load in the downstrea m. Figure 3 7 . Boxplots representing the distribution of NO3N concentration in Diamond Stream. Top and bottom edge of each box represents the 75th and 25th percentile, respectively, the line bisecting th e 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 conc entrations (sum of organic N plus ammonium N) were estimated from limited sampling times (3 times) during the hydrological year. Unlike the wide range in NO3N concentrations among all water quality monitoring sites, TKN concentrations ranged from 0.25 to 1.34 mg L1 (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, indicat ing 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
86 within the stream channel, can also result in export of dissolved organic N and particulate N from the catchments ( Cooper and Cooke, 1984) . C onclusions 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 nonpoint sources by plant uptake may improve water quality in downstream waterbodies. These stream services are essential to humandominated urban ecosystems. Understanding sources of nutrient additions to a stream can help expl ore 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 t he 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 great est N load. The NO3N 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 t hat NO3N 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
87 only 40% discharged to the groundwater system. This result suggested t hat 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 sport s 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 NO3N concentrations and N yield in downstream, which met the third hypothesis that spatial changes in NO3N 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.
88 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 effi ciently 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 nu trients 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 concentrati ons 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 sampli ng are typically the traditional approaches to collect samples for urban stormwater runoff. Manual sampling consists of â€œgrab samplesâ€ collected at a predetermined 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
89 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 highresolution 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 â€“ 300 nm ( Johnson and Coletti, 2002) . This approach avoids the chemical reduction reactions in the laborat ory 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 Vrtovek, 2010; Gutierrez et al., 2010; Hensley et al., 2014; Weiss et al., 2008) . Re gardless 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 s ensors 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
90 length of the light beam in the sensor could overcome the attenuation caused by the dissol ved 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 IonSelective Electrode sensor from InSitu Inc. (Bingen, Washington). Previous chapters determined N fluxes from catchments of various land us es 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 (Teledy ne Technologies Inc., Lincoln, Nebraska) and regular baseflow grab sampling. Both sampli ng approaches were conventional, timeconsuming 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 stormwate r system; 2. to compare the nitrateN (NO3N) mass in stormflow determined from SUNA and autosampler concentration data and flow data. My hypothesis was that SUNA could capture the changes in NO3N concentrations better than autosamplers in storms. If th e continuous nitrate sensor works effectively in stormwater systems, then pollutant control specialists will be able to continually monitor changes in NO3N concentrations in baseflow and in storms. The sensor can also help land managers adopt best managem ent practices to reduce N losses to water bodies.
91 Materials a nd 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 makes the campus similar to a small urban â€œcityâ€ 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 fiel d 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 41). 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 41). Water in Lake Alice overflows to two injection well s connected with the groundwater system.
92 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 use 1 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
93 based on t he stage discharge rating curve except that the full pipe discharge in RWC was calculated based on Manningâ€™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, timetriggered 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 NO3N 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 NO3N 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 NO3N solutions. Later, two SUNAs were placed side by
94 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 padd ed 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 NO3N 10 times every 5 minutes. Field results revealed that measurements from SUNA 238 and SUN A 239 exhibited similar patterns under several small changes in baseflow and a large drop in stormflow (Figure 43 ). 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 SUNAâ€™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 44 ). 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 NO3N 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
95 SUNA units provided acc urate 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 NO3N concentrations in stormwater. Figure 4 2 . The look for SUNA PVC tube Figure 4 3 . SUNA 238 and SUNA 239 in baseflow and stormflow in Sports Field Catchment (SFC) -50 0 50 100 150 200 -2 0 2 4 6 8 10 12 14 16 18 0 1000 2000 3000 4000 Flow rate, L s 1 NO3 N concentration, mg L1 Time, min SUNA238 SUNA239 Flow rate
96 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 (NO3N measured by SUNA = 0.97 * Grab sample NO3N 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 42. Detailed descriptions of the storms are displayed in Table 43. 0 5 10 15 20 25 0 5 10 15 20 25NO3N measured by SUNA, mg L1 Grab sample NO3N analyzed by ARL, mg L1
97 Table 4 2 . Descriptions of measurements and samples of SUNA and autosamplers for storm events Number of samples Date Precipitation, mm Sampling duration, min Autosampler SUNA No. of SUNA missing values Land use 1 = 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 . NO3N concentrations were obtained from rain samples captured by buckets Date Precipitation (mm) Previous Dry hours (hr) Imax (mm/hr) NO3N (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 NO3N concentrations from SUNAs were averaged based on the frequency of sampling from the autosampler. For example, if the composite sample from
98 autosamplers was comprised of 3 samples covering 15 min period, then the NO3N 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 NO3N 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)= where, Qi is the time variable flow and Ci is the time variable concentration. The The NO3N mass for each storm is calculated as MNO3 N (mg) = Results a nd Discussion Comparisons Of Autosampler/ SUNA NO3N Concentrations i n Storm Events Significant difference existed between the means of NO3N concentration from SUNA and the autosampler in RWC (Table 4 4; p <0.01). SUNA showed a higher NO3N 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 th e sensor was not using UV analysis like SUNA . Those researchers used Hydrolab
99 MiniSonde 4a water quality multi parameter datasonde (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 NO3N by UV light, therefore affected the NO3N concentrati on calibrated in SUNA, and this interference gradient had greater effect on RWC than SFC because NO3N concentrations in RWC was much lower than SFC. This explained the reason why there was no significant difference between the means of NO3N 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 NO3N concentrations from the autosampler was very likely to be outliers. Table 4 4 . Comparison of NO3N concentrations from SUNA and the autosampler in RWC and SFC SUNA Autosamplers Land use 1 No. of pairs Mean Std dev. Mean Std 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).
100 Figure 4 5 . Changes in NO3N monitored by SUNA and autosampler in Reclaimed Water Irrigated Catchment ( RWC ) . -10 0 10 20 30 40 50 0 0.5 1 1.5 2 0 100 200 300 400 500 600 Discharge, L s 1 NO3N, mg L-1 SUNA Autosampler Flow rate 0 10 20 30 40 50 60 70 0 0.5 1 1.5 2 0 20 40 60 80 100 120Discharge, L s1 NO3N, mg L-1 -5 0 5 10 15 20 25 0 0.5 1 1.5 2 0 10 20 30 40 50 60 70 80 Discharge, L s1 NO3N, mg L-1 -20 0 20 40 60 80 0 0.5 1 1.5 0 50 100 150 200Discharge, L s1 NO3N, mg L1 Time, min a. 4/8/2014 b. 5/25/2014 d. 5/30/2014 c. 5/29/2014
101 There may be other reasons for the difference between SUNAs and autosamplers in low NO3N concentration environment (RWC). NO3N concentrations in SFC ran ged 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 NO3N in t he UF Analytical Research Lab (ARL) was 0.148 mg/l. Consequently, it is likely that SUNAs were more sensitive to the changes of NO3N 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 44) between the means of NO3N 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 46 ). The slight difference was because the autosampler had at least 3 subsamples in one composite sample bottle which possibly included both baseflow (high NO3N) and stormflow (low NO3N). Meanwhile SUNA only read one discrete sample at a time when the autosampler collected one subsample.
102 Figure 4 6 . Changes in NO3N monitored by SUNA and autosampler in Sports Field Catchment (SFC). 0 50 100 150 0 5 10 15 20 0 100 200 300 400NDischarge, L s1 NO3N, mg L1 SUNA Autosampler Flow rate 0 1 2 3 4 5 6 0 2 4 6 8 10 12 14 0 20 40 60 80 100Discharge, L s1 NO3N, mg L1 -5 0 5 10 15 20 25 0 2 4 6 8 10 12 0 20 40 60 80Discharge, L s1 NO3N, mg L1 0 10 20 30 40 50 0 2 4 6 8 10 12 14 0 50 100 150 200Discharge, L s1 NO3N, mg L1 Time, min d. 5/30/2014 c. 5/29/2014 b. 5/1/2014 a. 3/17/2014
103 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 45 d), SUNAs c aptured the sharp decrease in NO3N 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 0 20 40 60 80 0 2 4 6 8 10 12 14 0 50 100 150 200 250 300 350Discharge, L s1 NO3N, mg L1 -100 0 100 200 300 400 0 2 4 6 8 10 12 0 100 200 300 400Discharge, L s1 NO3N, mg L1 Time, min e. 8/29/2014 f. 8/30/2014
104 events but autosamplers with composite sampling cannot . Although autosamplers can also collect discre te samples, the collected samples may still miss the response of NO3N to the peak of the hydrograph. SUNAs cannot guarantee the collection of the extreme changes of NO3N to the changes of discharge, but they can capture a closer image for NO3N changes than autosamplers as they can be set up to measure NO3N 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 46 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 SUNAsâ€™ performance in stormwater is missing data during the beginning of storms. In Figure 45 a, Figure 45 b and Figure 46 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 ( BertrandKrajewski et al., 1998; Sansalone and Cristina, 2004) . The particulate matter could have blocked the ultraviolet light from SUNA therefore kept SUNA from reading NO3N concentration.
105 Figure 4 7 . Event mean concentrations (NO3N) determined by the autosampler and SUNA 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC). Event Mean Concentrations from Autosamplers and SUNA s 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 NO3N 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 4/08/2014 5/25/2014 5/29/2014 5/30/2014 NO 3 N concentration, mg L 1 Autosampler SUNA 0 1 2 3 4 5 6 7 8 9 10 3/17/2014 5/01/2014 5/29/2014 5/30/2014 8/29/2014 8/30/2014NO3N concentration, mg L1 Event date Land use 1 = RWC Land use= SFC
106 concentrations and mean autosampler NO3N concentrations (Table 44). 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 47 ), 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 g reat 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. Even t NO3N Mass The event NO3N 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 differ ences in N mass among sampling dates. For example there was at least one order of magnitude difference between events in NO3N mass in SFC (May 29th vs May 30th). The huge difference in NO3N 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 sites (May 30th), NO3N 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 NO3N mass determined by the autosampler in SFC with that in RWC.
107 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 NO3N mass for storm events ranged from 3 to 44 g in RWC based on autosamplerâ€™s results, and from 21 to 75 g based on SUNAâ€™s results. Similarly, it ranged from 70 to 1930 g in SFC based on autosamplerâ€™s results, and from 60 to 2030 g based on SUNAâ€™s results. It showed that the differenc e in mass determined by both methods could be extremely large in some storms, suggesting accurate NO3N concentration is very important in NO3N mass determination.
108 Figure 4 8 . Event NO3N mass determined by autosamplers and SUNA 1 Reclaimed Water Irrigated Catchement (RWC), and Sports Field Catchment (SFC). Conclusions Measurements from SUNAs were compared with labanalyzed composite samples from autosamplers in two catchments with diff erent land uses in this study. Generally both SUNAs and autosamplers showed similar trends in storms where 36 44 3 25 72 75 21 39 0 10 20 30 40 50 60 70 80 4/08/2014 5/25/2014 5/29/2014 5/30/2014NO3N mass, g Autosampler SUNA 240 100 70 1930 280 960 360 90 60 2030 310 1670 0 500 1000 1500 2000 2500 3/17/2014 5/01/2014 5/29/2014 5/30/2014 8/29/2014 8/30/2014 NO 3 N mass, g Event date Land use 1 = RWC Land use= SFC
109 concentration decreased in high discharge and increased in low discharge. However, the difference between the means of NO3N concentrations determin ed by the two methods was significant ( p <0.01) in RWC. The SUNA showed a higher NO3N concentration than the results from the autosampler. This might be associated with the potential tremendous changes in hydrology in narrow stormwater pipes which affect ed 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 NO3N concentration was 1.19 mg L1) than SFC (mean grab sample NO3N concentration was 12.21 mg L1) as described in Chapter 2, which was closer to the detection limit of ARL (0.148 mg L1 for NO3N). The event mean concentrations of NO3N determined by both m ethods were consistent with the mean concentrations among all events with higher NO3N 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 NO3N concentrations det ermined by both methods affected the calculation of event NO3N mass, suggesting difference in NO3N 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 NO3N better than autosamplers with smooth curves in response to the changes of discharge. They can provide high resolution data through the storms but
110 autosamplers may be limited to the number of bottles. Also autosampler composite samples can introduce cross contamination while SUNAs measured the NO3N 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 s ampling. 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 NO3N changes. Meanwhile, in order to evaluate the performance of SUNAs compared with autosamplers, comparisons between SUNAs and different sam pling 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.
111 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 heal th ( Kaye et al., 2006; Miller, 1971 ; Oenema et al., 2003) . A large amount of attention has been placed on N from stormwater r unoff 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 nonpoint source, nitrateN ( NO3N) 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 NO3N 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 contr ol 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 NO3N and other dissolved and particulate chemicals measured for similar discharge d uring the rising and falling
112 limbs of storm hydrographs . This is called hysteresis phenomenon. For example, Webb and Walling (1985 ) examined NO3N 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 rainfa ll to the solutes in baseflow . If solute concentrations increase in the rising or falling stage of discharge, it is possible that subsurface â€œreservoirsâ€ are flushed with high solute concentrations depending on the flushi ng rate and antecedent conditions ( Burt et al., 1983 ; House and Warwick, 1998) . Similar hysteresis effect in NO3N concentration was observed in other studies using highresolution nitrate sensors ( Bowes et al., 2015; Bowes et al., 2009) . NO3N 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 NO3N concentration. Bowes et al. (2009) reported a lack of hysteresis for NO3N concentration during most of the largest storm events, suggesting the dominating source was the groundwater due to its constant NO3N 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.
113 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 cli matic factors are largely dependent on changes in hydrology. The understanding of NO3N dynamics in storms can help to identify NO3N sources during NO3N transport, thereby it is very critical to NO3N source control and management. Several studies associated with NO3N 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 f ew have been conducted in small catchments where NO3N concentrations can change more rapidly . The flow in small catchments was suggested to reach the surrounding surface wa terbodies 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 NO3N transport in storms in the two small urban catchments have not been known yet. In this study, in situ high resolution continuous nitrate sensor s called Submersible Ultraviolet Nitrate Analyzer s (SUNA s) (Satlantic Inc., Halifax, Canada) were introduced to observe NO3N dynamic s for seasonal and annual patterns. The objectives of this study were: 1. To establish a relationship between NO3N concentration and discharge for the hydrological year and for individual storms in both catchments ; 2. to determine the hysteresis effect o f NO3N concentrations from both catchments with regards to an nual patterns; My hypotheses were: 1. There would be a clear relationship between NO3N concentration and storm discharge in both
114 catchments ; 2.there would be a dilution effect caused by surface runoff in NO3N concentration from both catchments . Materials a nd Methods Study Area The study was carried out in two small urban catchment s on the campus of the University of Florida in Gainesville, Florida (Figure 51). 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
1 15 small urban â€œcityâ€ and functio n sim ilar to a city. The urban Sports Field C atchment (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 35ha 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 me asurement 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 cur ve except that the full pipe discharge in RWC was calculated based on Manningâ€™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 Sep tember 1st 2013 to August 31st 2014.
116 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 catchment s 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 dischar ge was displayed in Figure 52 with numbered recorded storm events.
117 Figure 5 2 . High frequency monitoring data for precipitation. NO3N 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 52 e with arrows referred to the st orm events (Table 51) recorded by Submersible Ultraviolet Nitrate Analyzer (SUNA) (Satlantic Inc., Halifax, Canada) 16, 17 1 2 3 4 5, 6 7 8 9, 10, 11, 12 Land use=SFC a b c d e Land use=RWC 1, 2, 3 4 5 6 7 8 9 10, 11, 12 13 14, 15 Date
118 High Resolution Nutrient Analysis Two Submersible Ultraviolet Nitrate Analyzers (SUNAs) (Satlantic Inc., Halifax, Canada) were tested both in laboratory with Standard NO3N 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. SUNA s w ere carefully deployed near the weir of SFC and the weir of RWC in the underground drainage basin, measuring NO3N 10 times every 5 minutes or 15 minutes for storms. The SUNA was deployed periodically during the year, depending on its availabilit y and when storms were expected. The detection range of NO3N for SUNA s was between 0.007 and 28.0 mg L1 (Satlantic Inc., Halifax, Canada). In this case, we excluded the values above 25 mg L1 from the field data. 25 mg L1 was the highest NO3N concen tration we tested for standard test in lab (more details described in Chapter 4). Conventional Rating Curves Between Discharge And NO3N Concentration The rating curve shows a relationship between the discharge and NO3N 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 NO3N concentration (C) is of the form: = ( 5 1 ) Where a and b are regression constants. The natural log transformation of E quation ( 5 1) is: ln = ln + ln ( 5 2 )
119 A log log regression was used to illustrate the relationship between discharge and SUNA NO3N concentrations through the hydrological year. Also the regression was used to determine the relationship between discharge and NO3N concentration in individual storms. Characteriz ation o f 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 â€œloop trajectoriesâ€. 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 highfrequency NO3N concentration data were plotted against stormwater discharge for each of the individual storm events, to determine the direction of the hysteresis trajectory (clockwise, anticlo ckwise, 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 mini mum 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 max imum flow during the hydrograph) for complex storms with loops more than 2 ( Lawler et al., 2006) .
120 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 discharge of the event, Qmin is the starting 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 t he 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 53). HIk is then determined, for clockwise hysteresis where CRL >CFL, simply as: Concentr ations Q min Q mid Q max C FL C RL Discharge (Q)
121 HIk = CRL/CFL 1 Results a nd Discussions Hydrological Factors f or Storms The descriptions of storms in both catchments are listed in Table 51 and Table 5 2. The precedent dry hours in both catchments ranged from 6 hours to 322 hours. The runoff duration ranged from 65 mi nutes 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 int ensity of the storm. The peak discharge in most of storms in RWC was 60.1 L s 1 which was calculated from Manningâ€™s equation. Although t he different calculation approaches for discharge in RWC may result in the variability of runoff coefficient , the over all 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.
122 Table 5 1 . Hydrological factors for storms in Sports Field Catchment (SFC) Storm number Storm date Dry hours, h Duration1, min Tp2, min Qmax3, L s-1 Precipitation, mm Runoff 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 SD 4 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
123 Table 5 2 Hydrological factors for storms in Reclaimed Water Irrigated Catchment (RWC) Storm number Storm date Dry hours, h Duration1, min Tp2, min Qmax3, L s-1 Precipitation, mm Runoff 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 SD 4 108 113 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 NO3N ConcentrationDischarge Relationship The NO3â€“ N concentration monitored by the SUNA in SFC through the hydrological year (Figure 52) demonstrated a clear relatio nship with discharge (Figure 5 4 ). The highest concentrations occurred at the lowest flows, with almost all NO3N concentrations greater than 10 mg L1 occurring at flows less than 1 L s1. The NO3N concentrations decreased rapidly as the discharge increased. The lowest NO3N concentrations (below 1 mg L1) occurre d at flows above 50 L s1. Similar dilution curves were shown in studies associated with NO3N dynamics, but the dilution effect was shown more pronounced in this study than the reported dilution curves because the hydrobiogeochemical factors (flood durat ion, peak discharge) and dominating land use may have less influence on the NO3N concentrations in this study ( Bowes et al., 2015) .
124 A power law concentrationdischarge 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 NO3N concentration and discharge had a significant relationship in RWC (Figure 55, 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 NO3N concentration with discharge. The relatively unchanging concentrations in RWC across wide ranges of discharg e require NO3N production or NO3N mobilization at rates nearly proportional to the water flux. Many researchers have documented that most water reaching a stream during a storm event was socalled â€˜oldâ€™ (e.g., pre storm) water . The â€˜ old â€™ 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 McDon nell (2006) . Therefore, it is likely that the NO3N concentration in â€œoldâ€ water stayed unchanged and did not vary much with discharge. In addition, the surface runoff from residential lawns was reported to contain a median NO3N concentration of 0.6 mg L1 ( 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
125 5/15/14). These reductions in NO3N concentration with increasing flow in SFC suggest ed that rainfall events were diluting a relatively constant input of NO3N. Figure 5 4 . NO3N concentration (C) discharge (Q) relationships in Sports Field Catchment (SFC). The loglog 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). 0.1 1 10 100 0.001 0.01 0.1 1 10 100 1000 10000NO3N concentration (C) mg L1 Flow rate (Q), L s 1 SUNA-NO3-N in all events, mg L-1 Storm event on 09/24/13 Storm event on 5/15/14 Ln (C) = 2.21 â€“ 0.35 ln (Q) R2 = 0.49 , p <0.0001
126 Figure 5 5 . NO3N concentration (C) discharge (Q) relationships in Reclaimed Water Irrigated Catchment (RWC). The loglog relationship was established based on the data for the hydrological year. The majority of NO3N data points fol lowed the well defined pattern with flow in SFC . However, there were some data points at higher than expected NO3N concentrations for a given flow. This can be associated with the complex event scale hydrobiogeochemical transport of NO3N. 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 NO3N concentration. This can produce a data point with high discharge but a relatively high concentration that does not fall in the expected pattern. Changes i n NO3N Concentrations t hrough Individual Storms The NO3N concentrationdischarge relationship and hysteresis patterns through all storms in both catchments were given in Table 5 3, Table 54 and illustrated in Figure 56 and Figure 57 . There was a significant relationship between discharge and NO3N concentration in individual storms except on 01/28/14, 05/01/14 and 05/29/14 0.1 1 10 0.01 0.1 1 10 100NO3 N concentration (C) mg L 1 Flow rate (Q), L s1 Ln (C) = 0.14 â€“ 0.04 ln (Q) R2 = 0. 02, p <0.0001
127 ( 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 NO3N dynamics in storms. This was in agreement with findings from Oeurng et al. (2010) that there was a strong correlation between NO3N tra nsport and total precipitation. Table 53 and Table 54 also provided evidence to show that the NO3N concentrationdischarge relationship was most evident (high R2) in storms with high peak discharge, which coincided with findings from O eurng 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 NO3N concentrationdischarge 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 57) , little changes in NO3N concentrations with discharge.
128 Table 5 3 . A log log regression between discharge (Q) and NO3N concentration (C) and hysteresis patterns in SFC in each storm Best fit Ln(C) Ln(Q) relationship NO 3 N Storm number Storm date Constant1 Slope1 R2 No. of observations p Hysteresis direction/ Trajectories Hysteresis Index (HI) 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 14 5/29/2014 1.95 0.10 0.20 13 0.13 No 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 loops more than two, so the HI was an averaged value based on Hysteresis Index calculated at 25%, 50% and 75% flow range.
129 Table 5 4 . A log log regression between discharge (Q) and NO3N concentration (C) and hysteresis patterns in RWC in each storm Best fit Ln(C) Ln(Q) relationship NO 3 N Storm number Storm date Constant1 Slope1 R2 No. of observations p Hysteresis direction/ Trajectories Hysteresis Index (HI) 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. NO3N 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 NO3N 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 NO3N hysteresis ( Bowes et al., 20 15; Schwientek et al., 2013; Zhang et al., 2007) . The predominantly clockwise trajectories indicated that NO3N sources from storms were rapidly mobilized and transported to the monitoring site during storm events . The dominating storm derived source NO3N concentration was suggested to be much lower compared to baseflow NO3N conc entration in SFC, therefore an evident
130 dilution effect was observed (higher hysteresis index) . However, the storm derived source NO3N concentration may not vary much with baseflow NO3N concentration in RWC, resulting in an indistinct dilution or no dilution (lower hysteresis index). Potential sources would include inchannel stored NO3N from the discharge pipe to exact NO3N monitoring location, NO3N stored in the sports field drainage systems, NO3N stored in soil leaked to the discharge pipe as well as stormwater runoff from nearby impervious surfaces. Another significant potential source of rapidly mobilized NO3N would include NO3N in the â€œoldâ€ water as mentioned earlier . The NO3N concentration increased immediately after the flow declined in S FC and RWC, suggesting the major baseflow NO3N source may be the â€œoldâ€ NO3N rich subsurface flow rather than the surface runoff. The dilution effect could also indicate a depletion of NO3N 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 NO3N 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 hyster esis pattern was shown in Figure 5 8 with several hysteresis loops as an example to demonstrate the complexity in the relationship between NO3N concentration and discharge. At the beginning of the storm on 12 /1 5 /1 3 , NO3N concentration dropped with the i ncreasing discharge and gradually increased with the decreasing discharge, then it dropped again at the peak discharge (90 L s1) at the time of 150 min and remained low under high flow . The low NO3N concentration at that time may be largely attributed t o surface runoff which was much
131 lower than the baseflow in SFC. Later when the discharge gradually reduced to 20 L s1, the NO3N concentrations increased to 5 mg L1 and remained above 5 mg L1 although the discharge ascended to 35 L s1. The relatively high NO3N concentration may be a mixture of event water and â€œoldâ€ 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 di scharge declined from 280th minute, t he NO3N concentrations went up again and reached a higher level than precedent condition, suggesting subsurface flow was then the dominating N source.
132 Figure 5 6 . NO3N hysteresis loops in individual storm events in SFC 0 5 10 15 20 25 0 50 100 150 0 10 20 30 0 200 400 0 5 10 15 20 25 0 20 40 60 80 100 0 10 20 30 0 20 40 60 0 5 10 15 20 25 0 20 40 60 0 5 10 15 20 25 0 20 40 60 80 100 0 5 10 15 20 25 0 10 20 0 5 10 15 20 25 0 10 20 0 5 10 15 20 25 0 50 100 150 0 5 10 15 20 25 0 2 4 6 09/24/13 09/25/13 11/26/13 12/10/13 12/15/13 01/28/14 02/12/14 Flow rate (Q), L s 1 09/23/13 0 5 / 01 /14 0 3 / 17 /14 NO 3 N concentration (C) , mg L 1
133 Figure 5 6 . Continued 0 5 10 15 20 25 0 20 40 60 80 100 0 5 10 15 20 25 0 5 10 -5 5 15 25 0 100 200 300 400 0 5 10 15 20 25 0 10 20 30 0 5 10 15 20 25 0 20 40 60 0 5 10 15 20 25 0 50 100 0 5 10 15 20 25 0 200 400 05/02/14 05/03/14 05/15/14 05/29/14 05/30/14 08/29/14 08/30/14 Flow rate (Q), L s 1 NO 3 N concentration (C) , mg L 1
134 Figure 5 7 . NO3N hysteresis loops in individual storm events in RWC 0 0.5 1 1.5 2 2.5 0 2 4 6 -0.5 0.5 1.5 2.5 0 2 4 6 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 0.5 1 1.5 2 2.5 0 50 100 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 0.5 1 1.5 2 2.5 0 10 20 30 0 0.5 1 1.5 2 2.5 0 20 40 60 80 11/16/13 11/21/13 12/29/13 02/21/14 03/29/14 05/25/14 05/30/14 05/29/14 03/28/14 04/08/14 Flow rate (Q), L s 1 NO 3 N concentration (C) , mg L 1
135 Figure 5 7 . Continued Figure 5 8 . An example of complex storms with multiple loops. This was the storm on 12/1 5 /1 3 . Letter a, b, c represent single loop for a certain period of time during the storm. All the loops were clockwise, indicat ing dilution from the rainfall. 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 0.5 1 1.5 2 2.5 0 20 40 60 80 0 20 40 60 80 100 0 5 10 15 20 25 30 0 100 200 300 400 500Flow rate, L s1 NO3N concentration, mg L1 Time, min NO3-N concentration Flow discharge 0 10 20 0 50 100 0 1 2 0 50 100 0 10 20 30 0 50 100 a b c a b c Flow rate (Q), L s 1 NO 3 N concentration (C), mg L 1 08/29/14 08/30/14 Flow rate (Q), L s 1 NO 3 N concentration (C) , mg L 1
136 Conclusions Urban stream ecosystem functions provide ecosystem services to humandominated ecosystems. Understanding NO3N 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 kn owledge of how hydrological response triggers NO3N transport at catchment level on the timescale of a single hydrological flood event is still lacking. The combined investigation of NO3N time series information, concentrationflow relationships and hyst eresis 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 NO3N source was dominated by subsurface water and responsible for high NO3N concentration during low flow. There was a dilution effect in stormwater NO3N transport associated with rapidly mobilization NO3N sources, ex hibiting a predominating clockwise hysteresis for NO3N transport. The diluti on may largely come from surface runoff . Future mitigation on NO3N concentration in SFC should focus on the fertilizer application which may potentially increase NO3N concentr ation in groundwater and increase baseflow NO3N. 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
137 catchment hydrology and streamwater chemistry. In situ high resolution nitrate sensor in t his study provided a great deal of information of how NO3N 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 system s. Determining NO3N 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 provi ding the most effective and cost effective management of urban catchments in the future.
138 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 (NO3N 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 t he annual discharge datasets. The results in Chapter 2 showed that SFC created the greatest N loads (37 kg ha1) 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
139 ecological function of wetlands near Lake Alic e 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 envi ronmental 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 NO3N 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 NO3N solution in the lab and baseflow samples in the field w ere 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 NO3N 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 SUNAsâ€™ 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, S UNAs were deployed to capture N dynamics in storms and baseflow in SFC and RWC from 2013 to 2014. The results showed that RWC had
140 a more distinct chemostatic behavior than SFC, suggesting little changes in NO3N concentrations with discharge in RWC. C lockwise 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 NO3N concentrations was from â€œoldâ€ 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.
141 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 2:5154. DOI: 10.1111/j.14752743.1986.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 9:333344. Al Kaisi M.M., Yin X. (2003) Effects of nitrogen rate, irrigation r ate, and plant population on corn yield and water use efficiency. Agronomy journal 95:14751482. 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 108:219232. 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 109:223233. Anderson D.M., Glibert P.M., Burkholder J.M. (2002) Harmful algal blooms and eutrophication: nutrient sources, composition, and consequences. Estuaries 25:704726. vila A., Piol J., Rod F., Neal C. (1992) Storm solute behaviour in a montane Medi terranean forested catchment. Journal of Hydrology 140:143161. Badruzzaman M., Pinzon J., Oppenheimer J., Jacangelo J.G. (2012) Sources of nutrients impacting surface waters in Florida: A review. Journal of Environmental Management 109:8092. Baker L. ( 1998) Design considerations and applications for wetland treatment of highnitrate waters. Water Science and Technology 38:389395. Baker L.A., Hope D., Xu Y., Edmonds J., Lauver L. (2001) Nitrogen balance for the Central Arizonaâ€“ Phoenix (CAP) ecosystem. Ecosystems 4:582 602. Balmforth D. (1990) The Pollution Aspects of Storm Sewage Overflows. Water and Environment Journal 4:219226.
142 Barton L., Colmer T.D. (2006) Irrigation and fertiliser strategies for minimising nitrogen leaching from turfgrass. Agricultural Water Management 80:160175. 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 62:11851190. 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 Eng ineering 36:351360. Berndtsson J.C., Bengtsson L., Jinno K. (2009) Runoff water quality from intensive and extensive vegetated roofs. Ecological Engineering 35:369380. BertrandKrajewski J. L., Chebbo G., Saget A. (1998) Distribution of pollutant mass vs volume in stormwater discharges and the first flush phenomenon. Water Research 32:23412356. DOI: 10.1016/s0043 1354(97)00420x. Bigelow C.A., Bowman D.C., Cassel D.K. (2001) Nitrogen leaching in sandbased rootzones amended with inorganic soil amendments and sphagnum peat. Journal of the American Society for Horticultural Science 126:151156. 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 44:900907. 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 14:21192131. Bir ch G.F., Matthai C., Fazeli M.S., Suh J.Y. (2004) Efficiency of a constructed wetland in removing contaminants from stormwater. Wetlands 24:459466. BlancoMontero C.A., Bennett T.B., Neville P., Crawford C.S., Milne B.T., Ward C.R. (1995) Potential envir onmental and economic impacts of turfgrass in Albuquerque, New Mexico (USA). Landscape Ecology 10:121128. Boesch D.F., Brinsfield R.B., Magnien R.E. (2001) Chesapeake Bay Eutrophication. Journal of Environmental Quality 30:303320. 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 36:15091519. Bormann F., Likens G., Melillo J. (1977) Nitrogen budget for an aggrading northern hardwood forest ecosystem. Science 196:981 983.
143 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., W ade A., Loewenthal M., Gozzard E., Newman J., Palmer Felgate E. (2015) Characterising phosphorus and nitrate inputs to a rural river using highfrequency concentration â€“ flow relationships. Science of The Total Environment 511:608620. Bowes M.J., Smith J.T., Neal C. (2009) The value of highresolution nutrient monitoring: a case study of the River Frome, Dorset, UK. Journal of Hydrology 378:8296. Brady N.C., Weil R.R. (1996) The nature and properties of soils PrenticeHall Inc. Braskerud B. (2002) Fac tors affecting nitrogen retention in small constructed wetlands treating agricultural nonpoint source pollution. Ecological Engineering 18:351370. 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 64:715724. Burt T., Butcher D., Coles N., Thomas A. (1983) The natural history of Slapton Ley Natu re Reserve XV: hydrological processes in the Slapton Wood catchment. Field Studies 5:731732. Canfield Jr D.E., Hoyer M.V. (1988) The eutrophication of Lake Okeechobee. Lake and Reservoir Management 4:9199. Capelo S., Mira F., De Bettencourt A. (2007) I n situ continuous monitoring of chloride, nitrate and ammonium in a temporary stream: comparison with standard methods. Talanta 71:11661171. Carpenter S.R., Caraco N.F., Correll D.L., Howarth R.W., Sharpley A.N., Smith V.H. (1998) Nonpoint pollution of s urface waters with phosphorus and nitrogen. Ecological Applications 8:559568.
144 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 563:313328. 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 estimat ing stormwater pollutant loads. Water Environment Research:12951302. 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 431:357365. 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 13:623645. 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. SJ2007SP10. Conley D.J., Paerl H.W., Howarth R.W., Boesch D.F., S eitzinger S.P., Havens K.E., Lancelot C., Likens G.E. (2009) Controlling eutrophication: nitrogen and phosphorus. Science 323:10141015. Cooper A.B., Cooke J.G. (1984) Nitrate loss and transformation in 2 vegetated headwater streams. New Zealand Journal o f Marine and Freshwater Research 18:441450. 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 8:10271036. 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 26:10381048. Davis A.P., Shokouhian M., Sharma H., Minami C. (2001) Laboratory study of biological retention for urban stormwater management. Water Environment Research:514. 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.
145 DeBusk K., Wynn T. (2011) Storm water bioretention for runoff quality and quantity mitigation. Journal of Environmental Engineering 137:800808. Del Barrio E.P. (1998) Analysis of the green roofs cooling potential in buildings. Energy and buildings 27:179193. 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 AlbermarlePamlic 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 landuse change on the N cycle of tropical aquatic ecosystems: current conditions and projected changes. Biogeochemistry 46:109148. Drolc A., Vrtovek J. (2010) Nitrate and nitrite nitrogen determination in w aste water using online UV spectrometric method. Bioresource Technology 101:42284233. Dunnett N., Kingsbury N. (2004) Planting green roofs and living walls Timber Press Portland, OR. Ellis J.B. (2001) Sewer infiltration/exfiltration and interactions wi th sewer flows and groundwater quality, 2nd International Conference Interactions between sewers, treatment plants and receiving waters in urban areas â€“ 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 sandbased root zones. Canadian journal of plant science 77:433444. EPA A. (2011) Inventory of US greenhouse gas emissions and sinks: 19902009, EPA 430R 11005. 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 34:129137. Ferguson R. (1986) River loads underestimated by rating curves. Water Resources Research 22:7476. 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 377:167177.
146 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 129:464471. Fisher D.C., Oppenheimer M. (1991) Atmospheric nitrogen deposition and the Chesapeake Bay estuary. Ambio:102108. Flint K.R., Davis A.P. (2007) Pollutant mass flushing characterization of highway stormwater runoff from an ultraurban area. Journal of Environmental Engineering 133:616626. Forman D., Al Dabb agh S., Doll R. (1984) Nitrates, nitrites and gastric cancer in Great Britain. Nature 313:620625. 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 Park ing Lot Runoff. pp. 15141522. 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., Vosmarty C.J. (2004) Nitrogen Cycles: Past, Present, and Future. Biogeochemistry 70:153 226. DOI: 10.1007/s1053300403700. 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. 325. Galloway J.N., Schlesinger W.H., Levy H., Michaels A., Schnoor J.L. (1995) Nitrogen fixation: Anthropogenic enhancement envi ron mental response. Global Biogeochemical Cycles 9:235252. 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 32:413424. Gentry L.E., David M.B., Below F.E., Royer T.V., McIsaac G.F. (2009) Nitrogen mass balance of a tiledrained agricultural watershed in east central Illinois. Journal of Environmental Quality 38:18411847. 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 96:946954.
147 Gilbert J.K., Clausen J.C. (2006) Stormwater runoff quality and quantity from asphalt, paver, and crushed stone driveways in Connecticut. Water Research 40:826832. 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 highresolution in situ sensors. Estuaries and Coasts 36:708727. 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. 421427. 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 1:215231. Godsey S.E., Kirchner J.W., Clow D.W. (2009) Concentration â€“ discharge relationships reflect chemostatic characteristics of US catchments. Hydrological Processes 23:18441864. Gold A.J., DeRagon W.R., Sullivan W.M., Lemunyon J.L. (1990) Nitratenitrogen losses to groundwater from rural and suburban land uses. Journal of soil and water conservation 45:305310. Goldewijk K.K. (2001) Estimating global land use change over the past 300 years: the HYDE database. Global Bi ogeochemical Cycles 15:417433. Graedel T.E., Keene W. (1996) The budget and cycle of Earth's natural chlorine. Pure and applied Chemistry 68:16891697. Groffman P.M., Law N.L., Belt K.T., Band L.E., Fisher G.T. (2004) Nitrogen fluxes and retention in ur ban watershed ecosystems. Ecosystems 7:393403. Guillard K., Kopp K.L. (2004) Nitrogen fertilizer form and associated nitrate leaching from cool season lawn turf. Journal of Environmental Quality 33:18221827. 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 15:4052. Hall F.R. (1970) Dissolved Soli ds Discharge Relat ionships: 1. Mixing Models. Water Resources Research 6:845850.
148 Hamilton A., Stagnitti F., Boland A.M., Premier R. (2005) Quantitative microbial risk assessment modelling for the use of reclaimed water in irrigated horticulture. Environm ental health risk III:71 81. Hammer D.A., Knight R.L. (1994) Designing constructed wetlands for nitrogen removal. Water Science & Technology 29:1527. Han Y., Lau S. L., Kayhanian M., Stenstrom M.K. (2006) Characteristics of highway stormwater runoff. Water Environment Research 78:23772388. 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 9:882885. Harmel R., King K. (2005) Uncertainty in measured sediment and nutrient flux in runoff from small agricultural watersheds. Transactions of the ASAE 48:17131721. Harmel R., King K., Slade R. (2003) Automated storm water sampling on small watersheds. Applied Engineering in Agriculture 19:667674. Harvey J.W., Ben cala K.E. (1993) The effect of streambed topography on surface subsurface water exchange in mountain catchments. Water Resources Research 29:8998. Hathaway A.M., Hunt W.F., Jennings G.D. (2008) A field study of green roof hydrologic and water quality performance. Trans. ASABE 51:3744. Heaney J.P., Wright L., Sample D. (1999) Research needs in urban wet weather flows. Water Environment Research:241250. 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 15:2533. Hendry C.D., Brezonik P.L. (1980) Chemistry of precipitation at Gainesville, Florida. Environmental Science & Technology 14:843849. 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 59:11521170. Hiscock J.G., Thourot C.S., Zhang J. (2003) Phosphorus budget â€”lan d use relationships for the northern Lake Okeechobee watershed, Florida. Ecological Engineering 21:6374.
149 Hoffman L., Ebdon J., Dest W., DaCosta M. (2010) Effects of nitrogen and potassium on wear mechanisms in perennial ryegrass: I. wear tolerance and rec overy. Crop Science 50:357366. 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 associat ed chemical constituents in urban environments: lessons from the city of Atlanta, Georgia, USA Water Quality Monitoring Program. Journal of Soils and Sediments 9:342363. House W.A., Warwick M.S. (1998) Hysteresis of the solute concentration/discharge rel ationship in rivers during storms. Water Research 32:22792290. 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 33:669687. Hsieh C. h., Davis A.P. (2005) Evaluation and optimization of bioretention media for treatment of urban storm water runoff. Journal of Environmental Engineering 131:15211531. Hsieh C.H., Davis A.P., Needelman B.A. (2007) Nitrogen removal from urban stormwater runoff through layered bioretention columns. Water Environment Research 79:24042411. 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 a nd Benthos Research 1:7184. 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 â€“ 2012) 95:1673716754. Jansson M., Andersson R., Berggren H., Leonardson L. (1994) Wetlands and Lakes as Nitrogen Traps. Ambio 23:320325. 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 49 6:347350. 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 332:241258.
150 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 49:12911305. Juang T., Wang M., Chen H., Tan C. (2001) A mmonium fixation by surface soils and clays. Soil science 166:345352. Kalnay E., Cai M. (2003) Impact of urbanization and landuse change on climate. Nature 423:528531. Karpf C., Krebs P. (2004) Sewers as drainage systems quantification of groundwater infiltration. NOVATECH 2004 â€œSustainable Techniques and Strategies in Urban Water Managementâ€, Lyon, Proceedings Vol 2:969975. Kaushal S.S., Lewis Jr W.M., McCutchan Jr J.H. (2006) Land use change and nitrogen enrichment of a Rocky Mountain watershed. Ec ological Applications 16:299312. Kaye J.P., Groffman P.M., Grimm N.B., Baker L.A., Pouyat R.V. (2006) A distinct urban biogeochemistry? Trends in Ecology & Evolution 21:192199. Kelling K., Peterson A. (1975) Urban lawn infiltration rates and fertilizer runoff losses under simulated rainfall. Soil Science Society of America Journal 39:348352. Kemp M.J., Dodds W.K. (2001) Spatial and temporal patterns of nitrogen concentrations in pristine and agriculturally influenced prairie streams. Biogeochemistry 53:125141. 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:355367. 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 132:985993. 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 136:371378. Klein Goldewijk C., Battjes J. (1997) A hundred year (1890 â€“ 1990) database for integrated environmental assessments (HYDE, version 1.1). National institute of public health and the environment (RIVM), Bilthoven, the Netherlands.
151 Khler M. (2003) Plant survival research and biodiversity: lessons from Europe, first annual greening rooftops for sustainable communities conference, Awards and Trade show. pp. 2030. Kopp K.L., Guillard K. (2002) Clipping management and nitrogen fertilization of turfgrass. Crop Science 42:12251231. 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 8year field experiment in Norway. Agriculture, ecosystems & environment 79:199214. 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 158:443449. 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 47:6371. Lapointe B.E., Matzie W.R. (1996) Effects of stormwater nutrient discharges on eutrophication processes in nearshore waters of the Flori da Keys. Estuaries 19:422435. 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 360:109126. Lee C.g., Fletcher T.D., Sun G. (2009) Nitrogen removal in constructed wetland systems. Engineering in Life Sciences 9:1122. Lee J.H., Bang K.W. (2000) Characterization of urban stormwater runoff. Water Research 34:17731780. DOI: 10.1016/s0043 1354(99)003255. 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 41:256271. 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:100108.
152 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:305310. Liu B.M., C ollick A.S., Zeleke G., Adgo E., Ea ston Z.M., Steenhuis T.S. (2008) Rainfalldischarge relationships for a monsoonal climate in the Ethiopian highlands. Hydrological Processes 22:10591067. Liu H., Hull R.J., Duff D. (1997) Comparing Cultivars of Three Co ol Season Turfgrass 37:526534. Lohani A., Goel N., Bhatia K.S. (2007) Deriving stageâ€“ discharge â€“ sediment concentration relationships using fuzzy logic. Hydrological Sciences Journal 5 2:793807. 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 urbanrural gradient. Environmental Science & Technology 34:42944300. L C., Tian H. (2007) Spatial and tempor al patterns of nitrogen deposition in China: synthesis of observational data. Journal of Geophysical Research: Atmospheres (1984 â€“ 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. 6370. 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 159:475491. DOI: 10.1007/s1066100806444. 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 4:1929. 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:14811490. McDiffett W.F., Beidler A.W., Dominick T.F., McCrea K.D. (1989) Nutrient concentrationstream discharge relationships during storm events in a first order stream. Hydrobiologia 179:97102. McDonnell J.J. (1990) A rationale for old water discharge through macropores in a steep, humid catchment. Water Resour. Res 26:28212832.
153 McGlynn B.L., McDonnell J.J., Seibert J., Kendall C. (20 04) 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 330:543563. 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 30:21342140. McL eod 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 14:96100. McMahon G., Woodside M.D. (1997) Nutrient mass balance for the AlbemarlePa mlico drainage basin, North Carolina and Virginia, 1990. JAWRA Journal of the American Water Resources Association 33:573589. Meisinger J., Randall G. (1991) Estimating nitrogen budgets for soil crop systems. Managing nitrogen for groundwater quality and farm profitability:85124. 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 285:6977. 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 77:217226. 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 36:426438. Miller G., Cisar J. (2001) Maintaining Athletic Fields. Miller L.W. (1971) Methemoglobi nemia associated with well water. JAMA 216:16421643. 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 36:14271433. 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 53:281295.
154 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 34:159167. Motz L.H., Sousa G.D., Annable M.D. (2001) Water budget and vertical conductance for Lowry (Sand Hill) Lake in northcentral Florida, USA. Journal of Hydrology 250:134148. Murray J., Juska F. (1977) Effect of management practices on thatch accumulation, turf quality, and leaf spot damage in common Kentucky bluegrass. Agronomy journal 69:365369. Nadelhof fer K.J., Emmett B.A., Gundersen P., Kjnaas 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 398:145148. Nagdali S.S., Gupta P. (2002) Im pact of mass mortality of a mosquito fish, Gambusia affinison the ecology of a fresh water eutrophic lake (Lake Naini Tal, India). Hydrobiologia 468:4551. Nakamura K., Harter T., Hirono Y., Horino H., Mitsuno T. (2004) Assessment of root zone nitrogen le aching as affected by irrigation and nutrient management practices. Vadose Zone Journal 3:13531366. Nicholls K.H., Cox C.M. (1978) Atmospheric nitrogen and phosphorus loading to Harp Lake, Ontario, Canada. Water Resources Research 14:589592. Nikolaidis N., Heng H., Semagin R., Clausen J. (1998) Nonlinear response of a mixed land use watershed to nitrogen loading. Agriculture, ecosystems & environment 67:251265. Nixon S., Granger S., Nowicki B. (1995) An assessment of the annual mass balance of carbon, nitrogen, and phosphorus in Narragansett Bay. Biogeochemistry 31:1561. Nordin A., Strengbom J., Witzell J., Nsholm 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 34:2024.
155 Oberndorfer E., Lundholm J., Bass B., Coffman R.R., Doshi H., Dunnett N., Gaffin S., Khler 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 uplandriparian zones in agricultural catchments: Implications for runoff generation and nitrate transport. Journal of Hydrology 331:643658. 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 20:316. O eurng C., Sauvage S., Snchez Prez J. M. (2010) Temporal variability of nitrate transport through hydrological response during flood events within a large agricultural catchment in southwest France. Science of The Total Environment 409:140149. Ou W.S. , Lin Y. F., Jing S.R., Lin H. T. (2006) Performance of a constructed wetlandpond system for treatment and reuse of wastewater from campus buildings. Water Environment Research 78:23692376. Owen C.R. (1995) Water budget and flow patterns in an urban wetland. Journal of Hydrology 169:171187. Paloheimo J., Fulthorpe R. (1987) Factors influencing plankton community structure and production in freshwater lakes. Canadian Journal of Fisheries and Aquatic Sciences 44:650657. Parsons L.R., Wheaton T.A., Castle W.S. (2001) High application rates of reclaimed water benefit citrus tree growth and fruit production. HortScience 36:12731277. Pathak H., Mohanty S., Jain N., Bhatia A. (2010) Nitrogen, phosphorus, and potassium budgets in Indian agriculture. Nut rient Cycling in Agroecosystems 86:287299. Pathan S., Aylmore L., Colmer T. (2002) Reduced leaching of nitrate, ammonium, and phosphorus in a sandy soil by fly ash amendment. Soil Research 40:12011211. Paul M.J., Meyer J.L. (2001) Streams in the urban landscape. Annual Review of Ecology and Systematics:333365. Peterjohn W.T., Correll D.L. (1984) Nutrient dynamics in an agricultural watershed: observations on the role of a riparian forest. Ecology 65:14661475.
156 Peterson B.J., Wollheim W.M., Mulholland P.J., Webster J.R., Meyer J.L., Tank J.L., Mart 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 292:8690. DOI: 10.1126/science.1056874. Petrovic A.M. (1990) The fate of nitrogenous fertilizers applied to turfgrass. Journal of Environmental Quality 19:114. Pitt R., Maestre A., Morquecho R. (2004) The national stormwater quality database (NSQD, version 1.1), 1st Annual Stormwater Management Research Symposium Proceedings. pp. 1351. 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 35:39473955. 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 32:16941700. Qin B., Xu P., Wu Q., Luo L., Zhang Y. (2007) Environmental issues of lake Taihu, China. Hydrobiologia 581:314. Raciti S.M., Groffman P., Fahey T. (2008) Nitrogen retention in urban lawns and forests. Ecological Applications 18:16151626. 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 85:179187. Rao P.P., Khemani L., Momin G., Safai P., Pil lai A. (1992) Measurements of wet and dry deposition at an urban location in India. Atmospheric Environment. Part B. Urban Atmosphere 26:7378. Ravishankara A., Daniel J.S., Portmann R.W. (2009) Nitrous oxide (N2O): the dominant ozonedepleting substance emitted in the 21st century. Science 326:123125. Reddy K. (1983) Fate of nitrogen and phosphorus in a wastewater retention reservoir containing aquatic macrophytes. Journal of Environmental Quality 12:137141. Reddy K., Fisher M., Ivanoff D. (1996) Res uspension and diffusive flux of nitrogen and phosphorus in a hypereutrophic lake. Journal of Environmental Quality 25:363371.
157 Reeve N., Toumi R. (1999) Lightning activity as an indicator of climate change. Quarterly Journal of the Royal Meteorological Society 125:893903. Reidsma P., Tekelenburg T., Van den Berg M., Alkemade R. (2006) Impacts of landuse change on biodiversity: an assessment of agricultural biodiversity in the European Union. Agriculture, ecosystems & environment 114:86102. Robertson D .M., Roerish E.D. (1999) Influence of various water quality sampling strategies on load estimates for small streams. Water Resources Research 35:37473759. 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., Miko M. (2008) Flushing of nitrate from a forested watershed: An insight into hydrological nitrate mobilization mechanisms through seasonal highfrequency stream nitrate dynamics. Journal of Hydrology 354:187202. Ryther J.H., Dunstan W.M. (1971) Nitrogen, phosphorus, and eutrophication in the coastal marine environment. Science 171:10081013. Sacks L., Lee T., Radell M. (1994) Comparison of energy budget evaporation losses from two morphome trically different Florida seepage lakes. Journal of Hydrology 156:311334. Sansalone J.J., Cristina C.M. (2004) First flush concepts for suspended and dissolved solids in small impervious watersheds. Journal of Environmental Engineering 130:13011314. S aunders 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 Ecosys tems 10:119137. Schilling K., Zhang Y. K. (2004) Baseflow contribution to nitratenitrogen export from a large, agricultural watershed, USA. Journal of Hydrology 295:305316. DOI: http://dx.doi.org/10.1016/j.jhydrol.2004.03.010 . Schlesinger W.H. (1997) Biogeochemistry. Geotimes 42:4444. Schumann U., Huntrieser H. (2007) The global lightning induced nitrogen oxides source. Atmospheric Chemistry and Physics 7:38233907.
158 Schwientek M., Osenbrck K., Fleischer M. (2013) Investigating hydrological drivers of nitrate export dynamics in two agricultural catchments in Germany using highfrequency data series. Environmental earth sciences 69:381393. 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 landuse change. Science 319:12381240. Shen J., Tang A., Liu X., Fangmeier A., Goulding K., Zh ang F. (2009) High concentrations and dry deposition of reactive nitrogen species at two sites in the North China Plain. Environmental Pollution 157:31063113. 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 â€“ 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 polyolefincoated fertilizers for increasing fertilizer efficiency and reducing nitrate leaching and nitrous oxide emissions. Fertilizer Research 39:147152. Shuman L. (2004) Runoff of nitrate nitrogen and phosphorus from turfgrass after watering in. Communications in soil science and plant analysis 35:924. 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 Tech nology 31:321330. 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 88:307317. Smil V. (1999) Nitrogen in crop production: An account of global flows. Global Biogeochemical Cycles 13:647662. 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 13:296304.
159 Smith V.H., Tilman G.D., Nekola J.C. (1999) Eutrophication: impacts of excess nutrient inputs on freshwater, marine, and terrestrial ecosystems. Environmental Pollution 100:179196. Snyder G., Augustin B., Davidson J. (1984) Moisture sensor controlled irrigation for reducing N leaching in bermudag rass turf. Agronomy journal 76:964969. 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 96:60016008. 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 21:531536. Starrett S., Christians N., Austin T.A. (1995) Fate of amended urea in turfgrass biosystems 1. Communications in Soil Science & Plant Analysis 26:15951606. Stenstrom M.K., Ka yhanian 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 303:18761879. Stone K., Hunt P., Novak J., Johnson M., Watts D. (2000) Flow proportional, timecomposited, and grab sample estimation of nitrogen export from an eastern Coastal Plain watershed. Transactions of the ASAE 43:281290. Sw ancar 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) Compar ison of methods for calculating annual solute exports from six forested Appalachian watersheds. Hydrological Processes 11:655669. Takeda K., Fujiwara K. (1993) Determination of nitrate in natural waters with the photoinduced 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 â€”implications for stormwater management. Water Research 39:19821989.
160 Thomas R., Foster C., Cohen M., Mar tin 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 sustai nable and efficient practices. Proceedings of the National Academy of Sciences 96:59956000. Tong S.T., Chen W. (2002) Modeling the relationship between land use and surface water quality. Journal of Environmental Management 66:377393. Turner R., Dortch Q., Justic D., Swenson E. (2002) Nitrogen loading into an urban estuary: Lake Pontchartrain (Louisiana, USA). Hydrobiologia 487:137152. Turner T.R., Hummel N.W. (1992) Nutritional requirements and fertilization. Turfgrass:385439. USBR. (1997) Water measurement manual, in: B. o. R. U. United States Department of the Interior (Ed.), Denver. Vzquez 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 112:313323. Vitousek P.M., Howarth R.W. (1991) Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry 13:87115. 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 53:259267. Webb B., Walling D. (1985) Nitrate behaviour in streamflow from a grassland catchment in Devon, UK. Water Research 19:10051016. Weiss P., Eveborn D., Krrman E., Gustafsson J.P. (2008) Environmental systems analysis of four onsite wastewater treatment options. Resources, conservation and recycling 52:11531161. Whipple Jr W., Hunter J., Yu S. (1977) Effects of storm frequency on pollution from urban runoff. Journal (Water Pollution Control Federation):22432248. 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 compo nents: sapflow, soil water budget, eddy covariance and catchment water balance. Agricultural and Forest Meteorology 106:153168.
161 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 12:123130. Wong J., Chan C., Cheung K. (1998) Nitrogen and phosphorus leaching from fertilizer applied on golf course: Lysimeter study. Water, Air, and Soil Pollution 107:335345. 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 124:584592. Wu J.S., Holman R.E., Dorney J.R. (1996) Systematic evaluation of pollutant removal by urban wet detention ponds. Journal of Environmental Engineering 122:983988. Xiankai L., Jiangming M., Shaofeng D. (2008) Effects of nitrogen deposition on forest biodiversity. Acta Ecologica Sinica 28:55325548. Young W.J., Marston F.M., Davis R.J. (1996) Nutrient exports and la nd use in Australian catchments. Journal of Environmental Management 47:165183. Zhang H., Sansalone J. (2014) Partitioning and First Flush of Nitrogen in RainfallRunoff 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 Process es 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 59:173186. Zhao X., Zhou Y., Min J., Wang S., Shi W., Xing G. (2012) Nitrogen runoff dominates water nitrogen pollution from ricewheat rotation in the Taihu Lake region of China. Agriculture, ecosystems & environment 156:111.
162 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 masterâ€™s study, she decided to apply for the doctoral program in S chool 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.