1 Introduction

Most of the environmental management in the past few decades has focused largely on point-source pollution of industrial and municipal effluents. Not much comparable effort has been made to restrict the input of nitrogen (N) and phosphorous (P) from dispersed or nonpoint sources such as agricultural and urban runoff. As a result, anthropogenic inputs of nonpoint pollutants, particularly N and P, have increased dramatically. Elevated nutrient levels in surface and ground water may cause human health problems, such as blue baby syndrome (Crittenden 2005), and may impair or destroy environmentally sensitive habitat through algal blooms and eutrophication (Allen and Kramer 1972).

Many surface waters in Central Florida, such as Lake Jesup where nitrogen and phosphorus are considered the limiting nutrients for primary production (Allen and Kramer 1972), currently experience eutrophication problems caused by high nutrient loading from stormwater detention ponds (i.e., wet ponds). Stormwater runoff is just one possible source of nitrogen; others include septic tanks and land-based applications of reclaimed wastewater or fertilizer, which can elevate nutrient concentrations. In a total maximum daily load (TMDL) report for water quality improvement proposed by the Florida Department of Environmental Protection (FDEP), the St. Johns River Water Management District (SJRWMD) examined several approaches to find a target nutrient concentration for Lake Jesup, which ranged from 0.04 to 0.08 mg/L for total phosphorus (TP) and 0.61 to 2.40 mg/L for total nitrogen (TN). The SJRWMD found concentrations of TN and TP that provide sufficient water clarity for growth of submerged aquatic vegetation (SAV) over 25% of Lake Jesup. SAV growth should enhance fisheries and provide wildlife habitat, as well as reduce the resuspension of flocculent organic sediments. The TMDL report for Lake Jesup also shows a current annual load entering the lake of 559,500 kg/year of TN and 36,000 kg/year of TP. Surface runoff accounts for 42% and 48% of the TN and TP loadings into the lake, respectively. To meet the TMDL standards, the loading into the lake should decrease 52% for TN and 37% for TP (Gao 2005a). The TN and TP target concentrations that allow the 25% SAV criteria are 0.61 and 0.04 mg/L, respectively (Gao 2005b). These concentrations were used as the standards for Lake Jesup with regard to this research project. No point sources currently discharge into the lake, so these goals must result from reduced nutrient concentrations in stormwater runoff.

Nitrogen-containing and phosphorous-containing compounds are found in urban stormwater runoff, primarily that from highways (United States Environmental Protection Agency 1999). Nitrates normally result from vehicular exhaust on the roadway itself and are also contributed from fertilized landscaped areas and residential areas alongside the road (German 1989; Vitousek et al. 1997). Considered one of the most efficient best management practices (BMPs), a wet detention pond removes contaminants through physical, biological, and chemical processes (United States Environmental Protection Agency 1999). This practice is used to treat stormwater runoff before it enters a surface water body. According to Chapter 62-40 of the Florida Administration Code, a stormwater pond shall achieve an 80% average annual load reduction of pollutants from the influent stormwater. The current law refers to the removal of solids only. However, the pond can only remove a certain percentage of a contaminant, and the discharged pollution, although significantly less than in stormwater runoff, may still damage fragile ecosystems in the receiving water body. The data compiled by Harper and Baker (2007a) from previous research studies suggest that detention ponds do not achieve this 80% goal for the nutrient pollutants of concern. The averages of the removal efficiencies from these studies show a 37% removal of TN, 79% for orthophosphorus (OP), and 69% for TP (Harper and Baker 2007a). There is an acute need to provide innovative methods, systems, apparatus, and devices for nutrient control and management for sources of nitrogen and phosphorus including stormwater runoff.

The use of sorption media, such as compost, to capture pollutants from stormwater runoff started in the late 1990s (Richman 1997; DeBusk et al. 1997). Stormwater infiltration systems were then widely used to address the quality issue of stormwater runoff through the use of either infiltration (Birch et al. 2005; Hatt et al. 2007) or exfiltration (Sansalone and Teng 2004). Engineered soil mix that provides stormwater treatment through filtration has been deemed as a sustainable source-control option (Ellis 2007; Chang et al. 2009), and various types of applications have been promoted recently in the context of green infrastructure systems (Kim et al. 2000; Hsieh and Davis 2005; Seelsaen et al. 2006; Hossain et al. 2009). Sanz et al. (1996) discussed the simultaneous removal of nitrogen and solids in continuous upflow filters and a computer simulation of the nitrification process based on the Activated Sludge Model No. 1 developed by the International Association of Water Quality. Yet, the use of upflow filtration for stormwater treatment is a relatively new idea to remove pollutants from contaminated stormwater runoff. Upflow filters have the advantage of longer run times and less maintenance than traditional downflow filters due to the design of the filter. Khambhammettu et al. (2006) used an upflow filter to treat runoff from highly contaminated critical source areas before it mixed with runoff from less contaminated areas. They studied a field application of the upflow filter inserted into a catch basin that achieved reductions of 70% for suspended solids, 65% for turbidity, and 18% for phosphorus (Khambhammettu et al. 2006). There are upflow filters commercialized for runoff treatment in stormwater inlets, and the successful integration of an upflow filter filled with the green sorption medium connected with surface skimmer could provide a new BMP to improve the quality of stormwater runoff. Green sorption medium consists of several recycled and natural media types that provide a favorable environment for pollutant removal to occur. Each type of media in the medium assists in the removal of specific nutrients. Phosphorus sorbs to one type of media, while another media is utilized as a carbon source for nitrate removal under anoxic conditions. Anoxic water has no free oxygen but does contain nitrate as electron acceptor for denitrification.

The combination of these elements provides a cost-effective treatment option to reduce nutrients traveling from wet detention ponds to surface waters. This paper presents the use of a chamber upflow filter and skimmer (CUFS) filled with a specific green sorption medium as process modification of stormwater retention ponds, which can increase the removal of nitrogen and phosphorus in the stormwater runoff. A similar study has already proved the sorption medium effective for phosphorus removal from irrigation water in green roof chambers (Hardin 2006). Yet this study particularly evaluates the performance of a CUFS in terms of water quality, water quantity, and overall operation and maintenance. Specifically, the objectives of this study include: (1) estimate the head loss through an upflow filter with a chosen media mix; (2) test the applicability of a surface skimmer; and (3) assess nitrogen and phosphorus concentrations leaving a detention pond using a CUFS setup with a specific sorption media for pollution control.

2 Materials and Methods

2.1 Study Site

The wet detention pond used for this research is located in the Lake Jesup Watershed in Central Florida and discharges to Howell Creek that flows into Lake Jesup (Fig. 1). The Lake Jesup Watershed extends into Seminole and Orange counties and covers more than 35,222 ha (87,000 ac), and the lake itself has a surface area of 4,316 ha (10,660 ac; Gao 2005c). Lake Jesup has been identified as one of the most hypereutrophic lakes in Central Florida, as displayed by often fish kills and pea-green-colored water (Florida Department of Environmental Protection (FDEP) 1997). Figure 1 shows the BMPs that have been applied to the Lake Jesup Watershed.

Fig. 1
figure 1

Lake Jesup Watershed and the study area

Full size image

The CUFS receives water from Red Bug Stormwater Pond B in Seminole County which will be referred to as Red Bug Pond herein. This pond was constructed in 1990 as an improvement to stormwater control systems for the Lake Jesup Watershed in Seminole County. Red Bug Pond receives stormwater runoff from 3.65 ha (9.02 ac) of impervious surfaces, 2.72 ha (6.72 ac) of open-spaced grassed areas, and 12.61 ha (31.15 ac) of residential development (Daugherty Consulting Engineers (DCE) 1990).

2.2 Experimental Setup

The setup of the CUFS consists of a floating pond skimmer connected by a pipe to the bottom of a precast concrete chamber. The chamber, a Department of Transportation (DOT) inlet, houses the filtering media which serves as the main nutrient removal mechanism in the setup. The floating pond skimmer is the inlet that directs water from the surface of the pond through the filter. The inlet at the water surface allows heavier particles to settle in the pond, and the water has fewer particles that will travel to the filter. The skimmer helps provide the power required to push the pond water through the filtering media and out to the pond effluent. This happens as a result of the difference in water elevations between the pond and upflow filter. The actual water surface elevation in the filter will be lower than that of the pond due to the head loss of the filtering media. At a time when no inflow to the pond is encountered (i.e., no stormwater runoff), the water level in the stormwater pond will equal the head difference of the upflow filter and pollution control media so that the filter media cannot treat any water. When water enters the pond (during and after a storm event), the pond water surface elevation rises, along with the floating skimmer. The rise in the skimmer provides a difference in water surface elevation between the pond and the surface of the upflow filter. The elevation head differential supplies the power required to push the pond water through the upflow filter and out to the pond effluent pipe as shown in Fig. 2.

Fig. 2
figure 2

Red Bug Pond CUFS with example elevations (cleanout pipe not pictured)

Full size image

Because water flows through the filtering media causing head loss, a bench study was performed over target surface loading rates to conclude the head loss of the chosen pollution control media, shown in Table 1. For flow conditions, ordinary filtration velocities are considered for the design of the filter (Cleasby and Logsdon 1999). To confirm the head loss and functionality of the CUFS, a pilot-scale design was built on a smaller detention pond than the target pond in Seminole County. As a pilot test, the Arboretum chamber structure on campus at the University of Central Florida (UCF) was installed adjacent to the rectangular weir outflow. The skimmer connects to the piping into the bottom of the chamber, and the filter effluent flows into the pond effluent downstream of the weir, as shown in Fig. 3a. This arrangement allows the CUFS to function in parallel with the pond effluent structure to directly compare the water quantity and quality data for storm events. The pilot study confirmed the head loss obtained at the laboratory-scale testing for the CUFS.

Table 1 Sizing specifics of the CUFS
Full size table
Fig. 3
figure 3

a Plan view of the Arboretum pond at UCF (N.T.S.), b plan view of Red Bug Pond B in Seminole County (N.T.S.)

Full size image

The full-scale CUFS is installed in parallel with the detention pond outlet in Seminole County, as shown in Fig. 3b. Stormwater from the detention pond flows down the inlet pipe, up through the filtering media, and out the filter outlet pipe. The outlet from the filter connects to the concrete outlet pipe of the pond, and the filtered water and effluent pond water mix and travel to Howell Creek. The first field application of the CUFS occurred on site at Red Bug Stormwater Pond B. At the maximum possible flow from the skimmer (Faircloth and Son 2005), the pollution control media treats the stormwater at a surface loading rate within the target range. The outflow pipe in the upflow filter must also be large enough to handle the inflow. The filter outflow pipe is sized using the orifice equation (Finnemore and Franzini 2002):

$$ Q = {C_{\text{d}}}A\sqrt {2gH} $$
(1)

where Q is the flow (in cubic meters per second), C d is the coefficient of discharge, A is the area of orifice (in square meters), g is the acceleration from gravity (9.81 m/s2), and H is the head acting on the top of the CUFS (in meters). The result of this equation yields the minimum size outflow pipe required to achieve the target flow.

2.3 Material Characterization

The upflow filter uses several types of green sorption media to improve water quality in stormwater runoff. This mix consists of 45% expanded clay, 45% recycled tire crumb, and 10% saw dust. The mix is poorly graded with an uneven curve over the range of consideration. More details of the mix are shown in Table 2 and Fig. 4 (Penn State Agricultural Analytical Services Laboratory 2006). The water permeability is good for operation though.

Table 2 Media characteristics
Full size table
Fig. 4
figure 4

Graduation curve for particle size distribution

Full size image

2.4 Reaction Mechanism

The adsorption, absorption, ion exchange, and precipitation processes are intertwined with the overall physicochemical process. Some nutrients such as phosphorus, removed by inorganic media, are likely a sorption/precipitation complex. The distinction between adsorption and precipitation is the nature of the chemical bond that might form between the pollutant and sorption media (Cucarella and Renman 2009). The attraction of a sorption surface between the pollutant and the sorption media causes the pollutants to leave the aqueous solution and simply adhere to the sorption media. Ammonia, nitrite, nitrate, and phosphorus may be sorbed in the CUFS between sequential storm events.

Within the microbiological process, if organic sources are present in the stormwater runoff, hydrolysis converts particulate organic nitrogen (N) to soluble organic N, and ammonification releases ammonia into the water (Metcalf and Eddy 2003). Nitrification in the detention pond occurs in the presence of oxygen in which ammonium is converted to nitrite (NO −2 ) and nitrite is converted to nitrate (NO −3 ) continuously (Metcalf and Eddy 2003). Because the filter media contain water on the top and bottom and is enclosed within a chamber, the media is not exposed to air, developing anoxic conditions. Denitrification occurs in the absence of free oxygen (under anoxic conditions) using nitrate as a final electron acceptor resulting in the stepwise reduction of NO −3 to NO −2 , nitric oxide (NO), nitrous oxide (N2O), and nitrogen gas (N2; Clark et al. 2001; Metcalf and Eddy 2003). Denitrification also requires the presence of an electron donor, which is provided in the green sorption medium by sawdust and part of the expanded clay with bioavailable organic content. Ultimately, the amount of denitrification may be limited by the frequency and duration of the oxic/anoxic fluctuations within the filter with respect to the reaction rates during the intermittent storm events.

2.5 Sampling and Analysis

The experiment was performed in a 9-month period with water quality samples taken after storm events that contribute at least 0.51 cm (0.2 in.) of rainfall. During times of no rainfall, baseflows from the detention pond were sampled. A sample was considered baseflow if no stormwater entered the pond within the past 6 days. In order to collect a range of samples for comparison, the sampling times following a rainfall event varied in the study, with no more than one sample taken per day. All storm samples were collected within 24 h of a rainfall event. Rainfall is documented using an on-site rain gage and a backup U.S. Geological Survey tipping bucket rain gage located nearby.

For sampling, this experiment compares the detention pond outflow to the CUFS outflow. Since the CUFS is installed in parallel with the detention pond, the two concentrations are directly compared. One liter of sample is taken after the water travels through the CUFS and from the surface of the detention pond near the outlet structure, in line with the skimmer. The sampling bottles are cleaned between sampling events with distilled water. A list of methods used to determine the concentrations of these chemicals is shown in Table 3.

Table 3 Methods and detection limits for each chemical species
Full size table

Quality assurance and quality control was conducted for each constituent in a sample set. To measure precision, duplicate samples are analyzed to produce a relative percent difference between the two measurements. The accuracy of the measurements is determined by spiking a sample with a known concentration of the parameter and calculating the percent recovery. Two-tailed t tests were used to find the difference between the mean values of the Red Bug Pond (referred to as “RBP” herein) and Red Bug CUFS (referred to as “RBF” herein) samples. The two random samples were collected independently of each other and have normal distributions for all parameters except pH. The normal distributions were verified via a normal probability plot and interquartile range test where the ratio of interquartile range to standard deviation equals approximately 1.3 (Mendenhall and Sincich 1995). Extreme values were not considered outliers due to the varying nature of rainfall events and pollution carried by stormwater runoff. An unusually high value may just be a characteristic of a large storm event or a recent fertilizer application in a nearby neighborhood.

3 Results and Discussion

3.1 Water Quality Samples

We compare the RBP outflow water quality to the RBF effluent water quality within a total of 35 sampling dates. These samples come from 28 storm events and seven baseflows collected over a period of 9 months from 2007 to 2008. The storm samples were taken at different time intervals following the event, with time ranges shown in Fig. 5. Ten stormwater samples collected at the pilot-study location can be used to compare the pond effluent (referred to as “AP” herein) to the pilot-scale CUFS outflow (referred to as “AF” herein). Such a sampling strategy may capture broader scenarios of nitrification versus denitrification in such a simultaneous system.

Fig. 5
figure 5

Number of samples taken at different time intervals

Full size image

3.2 pH and Alkalinity

Note that background concentration of the pollutant, dissolved oxygen concentration, pH values, and alkalinity may cause changes in thermodynamic equilibrium between the sorption media and aqueous solution. Table 4 shows the pH and alkalinity averages for 32 samples at the Red Bug site and ten samples at the pilot-scale site. The pollution control media in the CUFS at both locations did not alter the pH. The alkalinity increased slightly in the CUFS at the Red Bug Pond site and more substantially at the pilot-study site. However, the increases at both locations were not enough to conclude that the mean values are not equal based on the statistical analysis using a 95% confidence level. This is acceptable because denitrification recycles the alkalinity needed for nitrification.

Table 4 pH and alkalinity data summary
Full size table

3.3 Turbidity

The turbidity ranges from 8.19 to 1.94 nephelometric turbidity units (NTU) for the Red Bug Pond and 4.54 to 1.38 NTU for the CUFS based over 32 observations. Table 5 shows the sampling statistics. The average turbidity for the CUFS is lower than that of the pond outflow at both the Red Bug and pilot locations. For the turbidity measured at the Red Bug site, there is enough statistical evidence to conclude that the mean values are not equal at a 5% significance level. The 95% confidence interval (95%CI) for the Red Bug Pond samples shows that the CUFS reduces the mean detention pond turbidity value between 0.8 and 1.7 NTU.

Table 5 Turbidity data summary
Full size table

3.4 Solids

Thirty-one observations are used to compare the total suspended solids (TSS) and total dissolved solids (TDS) concentrations for the Red Bug site. As shown in Table 6, the CUFS at the Red Bug Pond reduced the TSS concentration almost in half. The mean values of the Red Bug Pond and CUFS are not equal at a 5% significance level, and the CUFS reduced the mean pond TSS concentration between 1.9 and 6.8 mg/L. Even though the TDS concentrations decreased in the CUFS for both sites, there was not enough evidence to reject the equality of the two means.

Table 6 Total solids data summary
Full size table

3.5 Phosphorus

The OP and TP concentrations are measured using 37 observations at the Red Bug site. For the majority of the sampling dates, TP consists of mostly OP and very little organic phosphorus. The OP values at the Red Bug Pond outflow are very low, but the CUFS reduces the values almost in half (Table 7). Mean hypothesis testing at a 95% confidence level confirms the reductions of both OP and TP from the Red Bug CUFS.

Table 7 Phosphorus data summary
Full size table

3.6 Nitrogen

The nitrogen forms compared in the CUFS experiment include nitrate + nitrite (NO x ) and TN. To ensure that denitrification could occur, dissolved oxygen measurements were taken periodically throughout the experiment. These measurements were taken below the filter in the CUFS, above the filter in the CUFS, and in Red Bug Pond itself. The low dissolved oxygen measurements show that anoxic conditions occur within the filtering media of the CUFS, allowing denitrifying bacteria to utilize nitrate and remove it from the water. As with OP, the Red Bug Pond discharges very low concentrations of nitrite + nitrate (Table 8). The average TN concentration of the Red Bug CUFS is relatively lower than the pond outlet, but at a 95% confidence level, there is not enough data to conclude that the two mean values are not equal. There is enough data to conclude that the pilot-study CUFS reduced the mean TN concentration from the pond when compared at a 5% significance level.

Table 8 Nitrogen data summary
Full size table

3.7 Comparative Analysis of Water Quality in Storm Events and Baseflows

The data analyzed in Tables 4, 5, 6, 7, and 8 resulted from 28 storm events and seven baseflows in the Red Bug detention pond. Separating the concentrations between storm events and baseflows results in the values shown below in Table 9. It shows that the turbidity level in storm events is higher than that in baseflows. Analysis of TN and TP has similar results, but this is not the case with TSS.

Table 9 Storm events and baseflows separated
Full size table

A graph comparing the combined storm events and baseflows to storm events only and baseflows only is shown below for TP and TN (Figs. 6 and 7). The standards chosen for Lake Jesup (0.04 mg/L TP and 0.61 mg/L TN) are shown on the graphs, as well as error bars representing 95%CIs. As shown in Fig. 6, the CUFS reduces TP below the standard concentration for all the samples (storm + base) and the storm events. The baseflow TP concentration from the Red Bug Pond does not exceed the standard concentration. The high contribution of TN from the storm events is shown in the TN graph in Fig. 7. The “baseflow only” concentration at Red Bug Pond does not exceed the standard, but the storm event concentration greatly exceeds the standard. The CUFS reduces the TN concentration, but not below the standard of 0.61 mg/LN.

Fig. 6
figure 6

Storm event and baseflow comparison for TP

Full size image
Fig. 7
figure 7

Storm event and baseflow comparison for TN

Full size image

3.8 System Reliability Test

The nutrient concentrations leaving the detention pond are relatively low compared to those in typical stormwater detention ponds. To see how the CUFS performed under higher nitrogen and phosphorus loading conditions, two experiments were conducted to increase the nutrient concentrations in the pond and inflow to the CUFS. Fertilizer was scattered into the detention pond near the skimmer, and samples were taken at periodical times after the introduction of fertilizer. Flow measurements were taken from the CUFS to see how long it would take the water to cycle through the filtering system, so representative samples from the pond and CUFS could be directly compared. Figure 8 displays the average concentrations from the two experiments for phosphorus and nitrogen.

Fig. 8
figure 8

Simulated event for reliability testing

Full size image

The addition of fertilizer increased the phosphorus concentration in the pond with mainly OP as expected because fertilizer supplies phosphorus in a form readily available for plant uptake. The CUFS reduced the two phosphorus species approximately in half and increased the OP/TP ratio compared to the pond. In the analysis of nitrogen, the CUFS increased the NO x (nitrite + nitrate) concentration compared to the pond. This could be caused by nitrification of the ammonia in the fertilizer (the TN of the selected fertilizer consisted of urea/ammonia mix). The water in the CUFS begins in the skimmer and inlet pipe, which is still under aerobic conditions until it reaches the filtering media. Therefore, the water in the CUFS is exposed to aerobic conditions for a longer time before sampling than the pond, which will allow nitrifying bacteria more time to convert ammonia to nitrate. Because this experiment was performed during a period of no rainfall, the flow measurements indicate that it would take the water approximately 2 h to cycle through the CUFS. Theoretically, the NO x (nitrite + nitrate) should be removed by denitrification in the anoxic filter, but the removal rate might be lower than the formation rate due to the excessive ammonia concentrations, and the NO x concentrations may increase.

Evidence of denitrification in the filter is shown in the comparison of TN values. It is known that the NO x (nitrite + nitrate) concentrations are relatively low, but the TN concentration in the pond is close to 20 mg/LN. Since the selected fertilizer contains TN in the form of urea and ammonia, the TN concentration is composed of mainly ammonia or ammonium ion. The reduction in TN is presumably due to denitrification in the media after the ammonia is converted to nitrate before entering the filtering media. This does not remove all the TN because 4.7 mg/LN remains in the CUFS samples, which is assumed to be ammonia or organic since the NO x concentration is low. The low dissolved oxygen concentrations above and below the filter show that the chamber is anoxic, allowing denitrification to occur.

3.9 Flow Measurements and Hydraulic Loading

The flow measurement is performed by using the “bucket and stopwatch” method. This method achieves a direct measurement for the flow rate with no estimation involved. The flow is taken directly from the outlet pipe of the filter, which discharges into the stormwater effluent pipe from the pond. Since the flow through the filter increases with head (pond water elevation), a water surface measurement was taken from the outlet weir structure. The head on the filter outlet pipe was also measured.

Calculation of a surface loading rate requires the surface area of the filter that allows water to flow through it. The Red Bug chamber measures 1.22 × 0.91 m (4 × 3 ft), equating to an inside area of 1.11 m2 (12 ft2). However, part of this area is occupied by the 15.24-cm (6-in.) cleanout pipe, which reduces the surface area to 1.09 m2 (11.80 ft2). With this surface area, the surface loading rates of the CUFS for the two observations measured are 22.90 and 31.74 m3/day/m2 (0.39 and 0.54 gpm/ft2), respectively. These surface loading rates correspond to a pond water elevation that is 40.64 cm (16 in.) and 31.75 cm (12.5 in.) below the top of the concrete overflow structure of Red Bug Pond. Although the peak flow rate through the CUFS during this experiment would be important data, obtaining a flow measurement when the water flows over the concrete structure would be too dangerous due to the large flow of water coming from the pond. The largest surface loading rate measured, 31.74 m3/day/m2 (0.54 gpm/ft2), is much smaller than the range of loading rates considered in the laboratory, 140.23 to 353.21 m3/day/m2 (2 to 5 gpm/ft2). However, in this measurement, the water level in the outlet pipe of the CUFS only occupied 70% of the outlet pipe cross-sectional area. The bench-scale study was also performed with clean tap water and new media with fewer solids to restrict the flow through the media.

3.10 Nutrient Loading Reduction

The nutrient removals from the CUFS can be expressed in the form of nutrient loading reductions by multiplying the flow and concentration. These calculations show the amount of mass per time that can be removed by the CUFS during storm events. The two flow rates measured 24.50 and 32.70 m3/day (4.5 and 6.0 gpm), the inflow to the CUFS (0.05 mg/L TP and 1.11 mg/L TN), and the outflow from the CUFS (0.04 mg/L TP and 0.92 mg/L TN) create a loading reduction range of 0.12 to 0.16 kg/year of TP and 1.70 to 2.27 kg/year of TN. For the simulated event using the same two measured flow rates, the loading reduction ranges become 5.73 to 7.64 kg/year of TP and 134 to 179 kg/year of TN. These numbers may seem quite small when compared to the target loading reductions from FDEP of 252,600 kg for TN and 21,400 kg for TP (Gao 2005c). However, these reductions occur at only one detention pond in a watershed that includes hundreds of stormwater ponds. Also, these loading reductions do not represent the maximum loading reduction possible because the maximum flow from the CUFS was not measured.

3.11 Removal Efficiency in Detention Pond

The phosphorus concentrations leaving the Red Bug Pond are lower than average for wet detention facilities in Florida. Values compiled by Harper and Baker (2007b) from previous studies show that stormwater from single-family residential developments contains average concentrations of 0.38 mg/L of TP and 2.07 mg/L of TN. According to Seminole County Engineering (2007), stormwater in the Howell Creek sub-basin contains a net concentration of 0.31 mg/L of TP and 1.72 mg/L of TN. These numbers are similar to those found by Harper and Baker (2007b). Harper and Baker (2007a) compiled several treatment efficiencies for wet detention ponds and found an average removal of 69% for TP and 37% for TN. Using these removal efficiencies for a wet detention pond and the typical stormwater concentrations for the Howell Creek sub-basin from Seminole County Engineering (2007), the expected effluent concentrations from the Red Bug Pond should be approximately 0.10 mg/L of TP and 1.09 mg/L of TN. The TN value matches the measured concentration from the Red Bug Pond of 1.11 mg/L (Table 8). This results in a TN removal efficiency of 35% in the Red Bug Pond. However, more phosphorus removal occurs in the Red Bug Pond than in an average wet detention pond. The measured TP value is 0.05 mg/L for the Red Bug Pond (Table 7), compared to the expected value of 0.10 mg/L. This shows a TP removal of 83% assuming the influent concentration is 0.31 mg/L.

The high phosphorus removal may be due to the sediment and solids removed by sedimentation in the pond. The basin that includes Red Bug Pond consists of Tavares-Millhopper, Myakka, and Eaugallie fine sands (Daugherty Consulting Engineers (DCE) 1990). To conclude if dissolved phosphorus was sorbing to the sediment, an extra sample was taken on August 7, 2007 in which the pond water was stirred by walking around the outlet area in the pond. This should physically break the bond between the sorbed phosphate and sediment, resulting in an increase in the OP and TP concentrations. The OP concentration increased from 0.04 to 0.14 mg/L P after the agitation. This indicates good removal of dissolved phosphorus by the sediment in the bottom of the pond. The pond also contains a littoral zone in which different types of aquatic grasses and plants inhabit. These plants utilize the dissolved phosphorus, removing it from the water.

Under the normal storm events sampled, the CUFS reduced the OP concentration by 46%, TP by 25%, and TN by 17% when compared to the pond effluents. Based on the typical values for stormwater in the Howell Creek sub-basin, the Red Bug detention pond and CUFS together reduced TP by 87% and TN by 47%. The percent removals are shown below in Table 10. The lower removals with nitrogen could be due to the low values of NO x (nitrite + nitrate) in the pond, limiting denitrification in the filter. Harper and Baker (2007c) show an average TSS concentration from stormwater in single-family residential subdivisions with curb and gutter of 48.40 mg/L. Using this value as the influent TSS concentration to the detention pond, it removes 81% of the TSS.

Table 10 Percent removals in system assessment
Full size table

3.12 Operation of the CUFS

Overall, the CUFS performed with minimal maintenance throughout the life of the experiment. Early in the experiment, problems with the upflow filter lifting up due to a high flow of water occurred, but they were resolved with the installation of unistrut pieces to supply more force to keep it down. The pilot-study pond is proliferated with algae, which caused a clogging problem in the skimmer and upflow filter. Once unclogged, the intake on the skimmer at both locations was wrapped with black fabric mat to prevent the algae, small plants, or fish from entering the system. This solved the problem and both the pilot-scale and full-scale CUFS remained unclogged throughout the experiment.

The Red Bug CUFS was cleaned twice to see how the system responded. Cleaning consists of pumping out the bottom of the chamber by inserting the intake hose of a small pump down the 15.24-cm (6-in.) cleanout pipe. All the water is pumped out of the filter three times to ensure removal of the sediment. The CUFS performed with no problems following the cleanout in both cases.

4 Conclusions

This study provides unique contributions to the research endeavor of using the CUFS for nutrient removal. It is intended to facilitate greater BMPs and promote sustainable development. It is known that the concentrations of TN and TP leaving the Red Bug Pond B (wet detention pond) exceed the concentrations causing impairment to the receiving water body, Lake Jesup. To remediate this pollution impact, this paper presents laboratory-scale, pilot-scale, and full-scale testing of the CUFS that has not been tested elsewhere. Findings indicate that the desired surface loading rates for the CUFS experiment were between 140.23 and 353.21 m3/day/m2 (from 2 to 5 gpm/ft2). The head loss determined for this range of loading rates for 61.08 cm (24 in.) of the chosen pollution control media is 22.86 cm (9 in.). This value was originally determined in the laboratory with the bench-scale study and then confirmed in both the pilot-scale and full-scale field applications. With the filter outlet pipe set at an elevation 22.86 cm (9 in.) lower than the pond effluent pipe elevation, the CUFS will discharge water when the pond discharges water, making the CUFS hydraulically operational. During operation, a surface skimmer supplies a design flow of water through the upflow filter in the CUFS while improving water quality in the pond effluent. In ponds with high levels of algae, small plants, or other small debris, a layer of black fabric mat must be placed over the intake of the skimmer to prevent the debris from clogging the upflow filter after short periods of time without affecting the hydraulic performance of the CUFS. The Red Bug CUFS significantly reduced the concentrations of turbidity, OP, TP, and TSS compared to the pond effluent. The CUFS is also capable of nutrient removals in highly polluted water, as shown in the reliability analysis. The results of this research are limited to the Central Florida climate, one media mix, and a well-functioning detention pond design based on the current standards.

Future research should be conducted on adding more skimmers to possibly increase the flow of filtered water from the CUFS, but it cannot flow too fast as to inhibit adsorption of phosphorus to the media or prevent anoxic conditions in the chamber. Different sorption media mixes may be tested to improve the removal efficiencies. The addition of a fountain in the detention pond should also be investigated to aid in the reduction of nitrogen in the CUFS. The fountain may encourage more aerobic conditions to convert ammonia to nitrate, which should increase denitrification in the CUFS. To further investigate denitrification in the CUFS, samples should be analyzed to see what types of bacteria are present for denitrification. Also, the limiting variable that governs the CUFS design should be identified. The most likely options for the limiting variable are flow and oxygen content in the CUFS.