Article pubs.acs.org/est
Urban Stormwater Runoff Nitrogen Composition and Fate in Bioretention Systems Liqing Li† and Allen P. Davis*,‡ †
School of Environmental Studies, China University of Geosciences, Wuhan, Hubei, 430074, P. R. China Department of Civil and Environmental Engineering, University of Maryland, College Park, Maryland 20742, United States
‡
ABSTRACT: Multiple chemical forms of nitrogen in urban stormwater make its management challenging. Sixteen storm events were monitored and analyzed for total nitrogen (TN), particulate organic nitrogen (PON), nitrate (NO3-N), nitrite (NO2-N), ammonium (NH3-N), and dissolved organic nitrogen (DON) in stormwater runoff and in treated discharge through a conventional bioretention cell. Influent PON can be effectively removed via bioretention sedimentation/filtration, NH3-N by ion exchange/ sorption, and NO2-N by oxidation. However, significant DON and NO3-N leached from the bioretention cell, resulting in only 9% net overall TN concentration reduction. Captured PON and vegetation detritus in the bioretention cell can be leached as DON or mineralized into NO3-N. The effluent N is dominated by NO3-N (46%) and DON (42%). Therefore, in addition to creating denitrification conditions for NO3-N, preventing DON leaching is also critical for effective nitrogen removal though bioretention systems. The bioretention cell exhibited a moderate mass load reduction for TN (41%), which mainly results from runoff volume reduction.
1. INTRODUCTION Urban stormwater has become an increasingly important source of nitrogen (N) to receiving waters.1 Excess N inputs to aquatic ecosystems cause eutrophication, which leads to alterations in community structure, degradation of habitat quality, and increased incidences and duration of harmful algal blooms. For example, a Chesapeake Bay study showed that increases in reactive nitrogen contributed to increased anoxic and hypoxic waters within the Bay.2 Inputs of N to the coastal waters of the United States are projected to continue to increase over future decades, in part due to rapid population growth in the coastal zone.3,4 Appropriate management strategies are therefore required to reduce the nitrogen loads entering these waters. Various stormwater control measures (SCMs) are being employed in watersheds to reduce pollutant loads from stormwater runoff, with bioretention as one of the more effective urban SCMs.5 Bioretention is an infiltration-based SCM that is widely used because of its ability to improve water quality and improve the hydrologic condition of the developed landscape.6 Bioretention systems are effective at removing a range of pollutants, including suspended solids, heavy metals, phosphorus, oil and grease, and fecal coliform.7−13 Nitrogen removal performance, however, has been highly variable, with reported results ranging from as high as 60% removal to net nitrogen export.7−9,11,12 Stormwater nitrogen is present in a number of chemical forms, including ammonium (NH3-N), nitrate (NO3-N), nitrite (NO2-N), dissolved organic N (DON), and particulate organic N (PON).14 The nitrogen composition varies with land use and hydrologic conditions.14,15 Nitrogen behavior in bioretention systems is therefore complex because of the biogeochemical © 2014 American Chemical Society
complexity of the nitrogen species and the numerous treatment mechanisms inherent to bioretention, including sedimentation/ filtration, adsorption, mineralization, and biological transformations.8,16−18 The conversion of captured nitrogen forms, and ammonium to nitrate and eventual nitrate export, appear to be the main reasons for the highly variable and generally poor performance. Laboratory19−21 and field22−25 studies have shown that the creation of a saturated zone at the bottom of bioretention systems can promote conditions for favorable denitrification to increase nitrogen removal. In addition, appropriate vegetation within a bioretention system can also enhance nitrogen removal by plant uptake and developing localized favorable conditions for denitrification.26−29 However, no significant improvement in total nitrogen removal due to the presence of a saturated zone for nitrate denitrification also has been reported.9 In some cases, organic nitrogen (DON) in the bioretention underdrain effluent significantly increased;9,12 its removal is variable and depends on system design and operational conditions. Therefore, nitrate leaching in bioretention systems may not be the only reason for poor nitrogen performance. More information on nitrogen species concentrations in bioretention systems is needed to provide better fundamental understanding of N behavior and fate. This information can lead to enhanced N removal through improved bioretention design. With this consideration, the objectives of this study are Received: Revised: Accepted: Published: 3403
December 12, 2013 February 24, 2014 February 26, 2014 February 26, 2014 dx.doi.org/10.1021/es4055302 | Environ. Sci. Technol. 2014, 48, 3403−3410
Environmental Science & Technology
Article
to (1) characterize the composition of nitrogen species in urban stormwater runoff, (2) quantify the nitrogen fate and identify the contribution of organic nitrogen and nitrate leaching to the poor performance of a bioretention system, and finally (3) provide recommendations for improving nitrogen removal via bioretention.
Table 1. Summary of Analytical Methods for Water Quality Analysis nitrogen species total nitrogen (TN) total dissolved nitrogen (TDN) ammonia (NH3) nitrate nitrogen (NO3−-N) nitrite nitrogen (NO2−-N) particulate organic nitrogen (PON) dissolved organic nitrogen (DON)
2. MATERIALS AND METHODS 2.1. Site Description. A bioretention cell, located in College Park, Maryland, United States, was monitored to investigate stormwater nitrogen behavior and fate from January to November 2013. The cell was installed in 2004 and collects and treats stormwater runoff from 2800 m2 of asphalt parking lot, roads, and concrete. It is trapezoidal (length = 50.3 m, width = 2.4−4.8 m; area = 181 m2), with loamy−sand media from 0.5 to 0.8 m deep. The bioretention cell is conventionally drained without a saturated zone using two perforated plastic pipes. In 2011, the bioretention cell was modified by mixing water treatment residuals (alum sludge) within the top 40 cm of the media to enhance phosphorus removal. The vegetation was replanted in 2012. The organic matter content of the top 30 cm media is 3.14% (9 media samples measured via loss on ignition in July 2013). The total nitrogen content of the top 30 cm media is 0.12% (1170 mg/kg) (9 media samples measured in July 2013). The oxalate extractable phosphorus of the top 24 cm media is 3.6 mg/kg (25 media samples measured in March 2012). 2.2. Monitoring and Analysis Methodology. Stormwater monitoring began in January 2013 and concluded in November 2013, all after the bioretention media modification. The bioretention cell accepts incoming runoff through a calibrated flume and the bottom underdrain was outfitted with a plug-in weir. ISCO 6712 autosamplers were assigned to the influent flume and underdrain effluent of the bioretention cell for flow rate measurement (ISCO 730 bubble flow meter) and water sampling. One ISCO 674 Tipping Bucket Rain Gauge with 0.254-mm sensitivity was installed onsite and logged rainfall depth in 2-min increments. Water samples for influent and effluent were collected at regular intervals, resulting in up to 12 samples for each storm event. A discrete sampling program was employed to collect more samples during the beginning of each event over 6−8 h sampling periods employing different sampling intervals. Samples were analyzed for TSS and nitrogen species, including total nitrogen (TN), total dissolved nitrogen (TDN), NO3-N, NO2-N, and NH3-N according to methods presented in Table 1.30 If any nitrogen species concentration was lower than the detection limit, a value equal to 1/2 of the detection limit was used for statistical purposes. Particulate organic nitrogen (PON) was calculated as the difference between TN and TDN; DON was calculated by subtracting NO3-N, NO2-N, and NH3-N from TDN. If the sum of NO3−N, NO2−-N, and NH3-N equaled or exceeded the TDN value, DON concentration was recorded as zero. All nitrogen forms are reported as mg/L N, to allow the proportion of TN to be calculated. Media samples were collected from the bioretention cell in July 2013 to estimate the nitrogen reservoir. The cell was divided into three sections: Plots 1 (45 m2), 2 (45 m2), and 3 (90 m2) along the flow path from the inlet to the far end of the cell. Three cores for each plot were combined to form total cores for each plot. Each core was sectioned into six samples corresponding to depths of 0−10, 10−20, 20−30, 30−40, 40−
analytical method
detection limit (mg/L N)
4500-N Ca 4500-N Ca
0.10 0.10
4500-NH3 F 4500-NO3− B, 4110 B 4500-NO2− B = TN − TDN
0.05 0.10 0.01 N/A
= TDN − (NO2,3-N) − (NH3-N)
N/A
a
Determined as nitrate by ultraviolet spectrophotometric method, 4500-NO3− B.
50, and 50−60 cm. All media samples were sent to the Soil Testing Program Lab at the University of Delaware for determination of total C and N. 2.3. Data Handling and Statistical Analyses. The event mean concentration (EMC) for TN and all its species was computed to compare the input and output N concentrations.31 Annual pollutant mass loads per unit drainage area (L, kg/hayr) were calculated by
L=
MP AD
(1)
where M is overall cumulative input/output pollutant mass (kg) measured during this study, P is the average annual rainfall rate in the State of Maryland (107 cm/year; Maryland Department of the Environment (MDE) 2011), A is the site drainage area (ha), and D is the total rainfall depth (cm) measured during the study duration. Annual input (Lin) and output (Lout) pollutant mass loads were employed to summarize the stormwater nitrogen species behavior and fate in the bioretention, in combination with hydrologic volume management. Pollutant load reduction from stormwater runoff volume reduction depends on percolation, storage, and/or evapotranspiration.32 Load reduction due to volume reduction, Lv‑red, was estimated from the product of annual cumulative runoff volume reduction and output pollutant EMCs. The pollutant concentration decrease is attributed to the system retention/ adsorption/transformation in and on the media, designated as system treatment. The system net load reductions (LNR) were estimated by subtracting the output mass loads (Lout) and the runoff volume reduction loads (Lv‑red) from annual input pollutant mass loads (Lin). If the value of LNR is positive, the system retains pollutant mass; if the value is negative, the system produces/leaches pollutant loads. Exceedance probability plots were employed for performance analysis of TN and all its species, created by ranking the calculated nitrogen EMCs and plotting them on a log scale.23 Comparisons between the input and output nitrogen EMCs were evaluated for statistical significance using the student’s t test with log-transformed data; significance was accepted at p < 0.05. Pollutant−duration curves were created using individual concentration measurements for TN and all its species. These curves summarize dynamic concentrations across all monitored storm events in a single distribution to evaluate the N concentration performance with time.33 3404
dx.doi.org/10.1021/es4055302 | Environ. Sci. Technol. 2014, 48, 3403−3410
Environmental Science & Technology
Article
3. RESULTS 3.1. Nitrogen Species Variation in Influent and Effluent Stormwater. Rainfall depth for the 16 monitored events ranged from 5.6 to 34.0 mm. To evaluate the representative nature of the monitored storm events distribution, rainfall duration and frequency for 10 352 rainfall events at 15 weather stations within the state of Maryland34 were compared with the sampled events in this study. The cumulative frequency of events more than 6.36 mm in the historical study is 52.5%, whereas that for this study is 93.8%. To collect both influent and effluent samples during events, larger storms were likely targeted and the data are somewhat biased toward larger events. Influent and effluent stormwater hyetographs, hydrographs, and pollutographs for TSS and all nitrogen species on April 19, 2013 (15.5 mm) are presented in Figure 1 to show concentration variations during a storm event. The timing and shape of the influent hydrograph closely followed those of the hyetograph. However, the discharge shows a lower and delayed peak flow rate, demonstrating the ability of the bioretention cell to effectively reduce volume and flow peaks. The bioretention cell significantly reduced the stormwater TSS concentration throughout the event (Figure 1 A). Nitrogen species concentrations in influent stormwater are dynamic, demonstrating significant variation over the course of the event. The initial concentrations of all nitrogen species were substantially higher than those observed during the latter stage of the storm event (Figure 1 B and C). The effluent nitrogen species concentrations did not vary as significantly as influent, although NO3-N concentrations commonly exhibited an initial spike (Figure 1B and D). The effluent NO2-N and NH3-N concentrations were less than the corresponding detection limits. Although the range of effluent TN concentrations was narrower than that for influent stormwater, the TN EMC had an only moderate reduction (37%) from 2.9 mg/L in influent to 1.8 mg/L in effluent. For this event the bioretention cell effectively reduced PON, NH3-N, and NO2-N concentrations. However, the effluent NO3-N and DON concentrations were always higher than those of the influent (Figure 1C and D). Data from the other events were generally similar, but differ in details. 3.2. Nitrogen Species Transformations and Removal. Pollutant duration curves shown in Figure 2 are created to focus on bioretention performance based on instantaneous pollutant concentrations. Probability plots for all nitrogen species based on storm event EMCs are presented in Figure 3. These plots are employed to evaluate the differences between input and output N concentrations, and characterize the output water quality for all monitored storm events.23 Total Nitrogen. The concentration duration curves for TN are shown in Figure 2A. The bioretention cell is capable of reducing the influent TN peak discharges and narrowing the range of stormwater runoff TN concentrations. However, based on TN criteria of 0.36 mg/L for Lakes & Reservoirs and 0.69 mg/L for Rivers & Streams recommended by U.S. EPA for the aggregate Ecoregion IX (Maryland),35 the bioretention discharge exceeds the 0.69 mg TN criterion for 93% (133.5 out of 143.0 h) of the discharge time compared to 83% (35.5 out of 43.0 h) for the influent. The distribution of the output TN EMCs is very similar to that of the input (Figure 3A). Both influent and effluent TN
Figure 1. Influent and effluent stormwater hydrographs, hyetographs, and pollutographs for all nitrogen species on April 19, 2013. A: rainfall, flow rate, and TSS concentrations; B: TN, PON, and TDN concentrations; C: influent dissolved N species; D: effluent dissolved N species.
EMCs all exceeded the 0.69 mg TN criterion for Rivers & Streams. The input TN EMCs ranged from 0.75 to 3.3 mg/L (median = 1.5 mg/L), and output TN EMCs ranged from 0.71 to 2.4 mg/L (median = 1.4 mg/L). Although the output TN concentration was slightly lower, it was not statistically significant (p > 0.05), indicating that this bioretention cell shows minimal capability for decreasing TN concentrations in stormwater runoff. PON, NH3-N, and NO2-N. Concentration duration curves for PON, NH3-N, and NO2-N show that the bioretention cell significantly reduced these N forms over the entire duration of the monitored period (Figure 2B and D). The duration of the 3405
dx.doi.org/10.1021/es4055302 | Environ. Sci. Technol. 2014, 48, 3403−3410
Environmental Science & Technology
Article
Figure 2. Pollutant duration curves across all monitored storm events for the bioretention cell. A: TN; B: PON; C: TDN; D: NH3-N and NO2-N; E: NO3-N; F: DON (DL = detection limit). Pollutant target concentrations shown in some plots.
mass loads, annual loads attributable to volume reduction, and the system net load reductions/productions are summarized in Table 2. The stormwater nitrogen behavior and fate in the bioretention cell is diagrammed in Figure 4. The annual input nitrogen load was 14.0 kg/ha-year, whereas the annual output and infiltration nitrogen loads were 8.2 and 4.4 kg/ha-year, respectively. The annual net nitrogen mass retained by the bioretention cell was only 1.4 kg/ha-year. The stormwater nitrogen entered the bioretention cell in the form of PON, DON, NH3-N, and NOx-N at 8.0, 2.2, 1.3, and 2.6 kg/ha-year, respectively. The annual output loads (underdrain discharge) were 1.3, 3.3, 0.15, and 3.5 kg/ha-year, respectively. From the perspective of system runoff input and output, the bioretention cell exhibited a moderate mass load reduction for TN (41%), which mainly results from runoff volume reduction; volume reduction contributed to 32% of the measured TN mass load reduction. The bioretention cell significantly reduced the PON, NH3-N, and NO2-N mass load from incoming stormwater runoff, at 83%, 89%, and 89%, respectively. NO3-N and DON were significantly leached from the bioretention system, although the volume reduction also contributed to some mass reduction; mass was increased by 45% and 50%, respectively. When the pollutant mass reduction due to volume
NH3-N concentrations below the detection limit (0.05 mg/L) was 82% of the entire monitored period (109.5 out of 133.7 h). The effluent NO2-N concentrations were lower than the detection limit (0.01 mg/L) nearly all of the monitored period (126.2 out of 133.7 h). The median EMCs for PON, NH3-N, and NO2-N were reduced from 0.57 to 0.08 mg/L, from 0.12 to
