Validation of stormwater biofilters using in-situ columns

Science of the Total Environment 544 (2016) 48–55

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Validation of stormwater biofilters using in-situ columns Kefeng Zhang a,b,⁎, Valentin Valognes a,b,c, Declan Page d, Ana Deletic a,b, David McCarthy a,b,e a

Monash Water for Liveability, Department of Civil Engineering, Monash University, Wellington Rd, Clayton, VIC 3800, Australia CRC for Water Sensitive Cities, Melbourne, VIC 3800, Australia c URGC Hydrologie Urbaine, INSA Lyon, Bâtiment Coulomb, 34 avenue des Arts, 69621 Villeurbanne Cedex, France d CSIRO Land and Water Research Flagship, Waite Laboratories, Waite Rd., Urrbrae, SA 5064, Australia e Environmental and Public Health Microbiology Laboratory, Department of Civil Engineering, Monash University, VIC 3800, Australia b

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• A novel in-situ column was proposed as an alternative validation monitoring tool. • The tool can reproduce field results for fluorescein removal over different conditions. • When using the tool for studying herbicide removal, some differences were observed. • The in-situ column is a promising tool to study the field performance of biofilter.

a r t i c l e

i n f o

Article history: Received 12 October 2015 Received in revised form 27 November 2015 Accepted 27 November 2015 Available online xxxx Editor: D. Barcelo Keywords: Treatment validation Stormwater biofilters Potable end-use Stormwater harvesting In-situ column Herbicides

a b s t r a c t Stormwater harvesting biofilters need to be validated if the treatment is to be relied upon. Currently, full-scale challenge tests (FCTs), performed in the field, are required for their validation. This is impractical for stormwater biofilters because of their size and flow capacity. Hence, for these natural treatment systems, new tools are required as alternatives to FCT. This study describes a novel in-situ method that consists of a thin stainless steel column which can be inserted into constructed biofilters in a non-destructive manner. The in-situ columns (ISCs) were tested using a controlled field-scale biofilter where FCT is possible. Fluorescein was initially used for testing through a series of continuous applications. The results from the ISC were compared to FCT conducted under similar operational conditions. Excellent agreement was obtained for the series of continuous fluorescein experiments, demonstrating that the ISC was able to reproduce FCT results even after extended drying periods (Nash–Sutcliffe coefficient between the two data sets was 0.83–0.88), with similar plateaus, flush peaks, slopes and treatment capacities. The ISCs were then tested for three herbicides: atrazine, simazine and prometryn. While the ISC herbicide data and the FCT data typically matched well, some differences observed were linked to the different climatic conditions during the ISC (winter) and FCT tests (summer). The work showed that ISC is a promising tool to study the field performance of biofilters and could be a potential alternative to full scale challenge tests for validation of stormwater biofilters when taking into account the same inherent boundary conditions. © 2015 Elsevier B.V. All rights reserved.

⁎ Corresponding author. E-mail address: [email protected] (K. Zhang).

http://dx.doi.org/10.1016/j.scitotenv.2015.11.150 0048-9697/© 2015 Elsevier B.V. All rights reserved.

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55

1. Introduction Many studies demonstrated that urban stormwater contributes to the deterioration of water quality in receiving bodies (Jeng et al., 2005; Brown and Peake, 2006) and cities are experiencing water stress (Fletcher et al., 2007). While stormwater harvesting is becoming more common and encouraged for non-potable uses (Hatt et al., 2006), potable uses currently have only limited uptake. There are scant examples of such systems, with the most notable being Singapore which has had indirect stormwater harvesting for potable end-uses since the 1960s (Philp et al., 2008). Stormwater biofilters, which are established under Water Sensitive Urban Design (WSUD) principles, are used for stormwater harvesting, in particular for lower exposure end-uses such as irrigation and toilet flushing (Hatt et al., 2006). They have proven to be effective in dampening the high variability of stormwater quality (Zhang and Guo, 2014). Meanwhile, they are also effective in treating pollutants (sediments, nutrients, heavy metals, microorganisms and micropollutants (Bratieres et al., 2008; Chandrasena et al., 2012). Due to the natural components of these systems, the pollutants are removed through combined physical, chemical and biological processes. Moreover, stormwater biofilters experience both wetting (during rainfall events) and dry periods (when the systems are idle). The main removal processes are different in different periods. For example, micropollutants, the major removal processes in these systems involve adsorption (mainly during wet events) and biodegradation (mainly over the dry periods) (Zhang et al., 2015a). However, to date the treatment performance of stormwater biofilter is not recognized as reliable because it lacks a validation protocol for treatment validation (Zhang et al., 2015b).Validation protocols are common for other engineered systems, such as membrane filtration, and provide robust evidence as to their treatment performance (DHV, 2013). Treatment validation provides scientific evidence that the treatment process produces water of the required quality and that water quality objectives are continuously met (DHV, 2013). Treatment validation can be completed through three stages: (1) Pre-validation, which entails gathering necessary information for the following stages, including target pollutants, operational/challenging conditions, potential removal mechanisms and surrogates; (2) validation monitoring, which determines the system performance under challenging conditions; and (3) operational monitoring, which ensures the long term performance of the system during normal operation (Zhang et al., 2015b). This study focused on the validation monitoring stage. As per current validation procedures developed for engineered systems (USEPA, 2005; DHV, 2013), validation monitoring should be performed at full-scale, using challenge tests. Challenge tests are expected to confirm the maximum removal credit that a treatment system is eligible to receive. This is achieved by dosing challenging concentrations of target pollutants and measuring the removal under challenging hydraulic conditions (USEPA, 2005). However, this is difficult to apply to stormwater biofilters because they are usually very large and the operation of fullscale testing is difficult as large, uncontrolled volumes of urban stormwater will enter the system during short periods of time (i.e. b3 h). In order to support the validation of stormwater treatment by biofilters, alternative validation monitoring methods are needed instead of the traditional challenge tests. Laboratory batch and column tests are widely used to assess the removal processes of pollutants in stormwater biofilters (Bratieres et al., 2008; Chandrasena et al., 2012) and similar soil-based systems, e.g. wetlands (Chevron Cottin and Merlin, 2007) and aquifers (Ying et al., 2008). However, although ex-situ types of studies can gain insights in the underlying removal mechanisms, they have also received criticism for not being able to represent the natural conditions of these systems. This has led to the development of in-situ based techniques, e.g. in-situ microcosms/columns, which could provide more-convincing evidence of the results with conditions closer to field

49

tests (Nielsen et al., 1995; Mandelbaum et al., 1997). For example, Nielsen et al. (1995) reported that in-situ and laboratory studies on the fate of specific organic compounds in an anaerobic landfill leachate plume showed good concordance, but some transformations, for phenol particularly, were observed only in in-situ experiments. Similar in-situ style tools have been used with success to study the biodegradation of various chemical compounds in aquifer systems and wetlands (Geyer et al., 2005; Stelzer et al., 2006; Braeckevelt et al., 2007). Currently there is no in-situ technique specifically designed for stormwater biofilters. The objective of the current study is to develop an in-situ tool to validate the treatment processes within stormwater biofilters. The specific aims include: • Test the in-situ column (ISC) tool for fluorescein by comparing its performance with field challenge test (FCT) fluorescein results (both the ISC and FCT were conducted on a small field-scale biofiltration system under similar conditions); and, • Test the ISC tool for three common herbicides (atrazine, simazine and prometryn) by comparing its performance with the results of the FCT tests performed on the same facility in the past work (Zhang et al., 2014; Zhang et al., 2015a). These herbicides were chosen because they are commonly detected in urban stormwater (Becouze et al., 2009; Zgheib et al., 2012) and can represent a human health risk if consumed over the long term (which would be the case for indirect potable uses) (NHMRC-NRMMC, 2011).

2. Materials and methods 2.1. Site description The stormwater biofilter system selected in this study is located at Monash University, Melbourne, Australia. Full details regarding characteristics and configurations of the studied biofilter have been previously reported (Hatt et al., 2009; Zhang et al., 2014). The selected biofilter has a submerged zone and uses sand (sand 96.0%, silt 0.8%, clay 3.2% — by weight; soil organic matter 0.4%) as filter media, containing 0.35% soil organic matter, 30 mg/kg total phosphorus and 300 mg/kg total nitrogen content. The length and width of the biofilter are 9.6 and 1.4 m, respectively. The design maximum ponding depth, filter media depth and submerged zone depth are 410 mm, 500 mm and 200 mm, respectively. This biofilter is predominantly planted with Melaleuca ericifolia, which has been previously reported to be efficient in removing nutrients (Read et al., 2008). This biofilter was used to perform both fluorescein and herbicides field challenge tests (FCTs).The results from FCTs were compared with those of the in-situ column (ISC) tests; this allowed the evaluation of the new proposed tool (i.e. the ISC) against the industry standard (i.e. the FCT). The following sections describe these tests. 2.2. Full scale experiment: field challenge tests (FCTs) Two FCTs were conducted: (1) Fluorescein FCT, which was a short study undertaken to gain initial insights into the behavior of the biofilter using a fluorescein as a tracer; this is a cheap, commonly used model micropollutant which can degrade in sunlight, adsorb to soil and be degraded by microbes (Smart and Laidlaw, 1977; Sabatini, 2000), and (2) Herbicide FCT, which was a stand-alone study used to challenge the system for the removal of three herbicides (fully reported in Zhang et al., 2014). The Fluorescein FCT (August 2013) was used to study breakthrough of fluorescein via spiking and flushing events; in-between the events, different lengths of dry periods were maintained (Fig. 1 left). Water was pumped from an adjacent stormwater pond, and if it was a spiking event, it was loaded with fluorescein to a concentration of 120 ± 5.0 μg/L. The inflow concentration of 120 μg/L was selected because it

50

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55

Fig. 1. Dosing regimes followed for both the field challenge tests (FCTs) and the in-situ column (ISC) tests. Left — fluorescein experiments which include spiking, flushing without pollutants and exposure to dry weather periods (N.B. the dry weather period between the 1st spiking and flushing was ~54 h for the ISC test, while for the FCT it was ~62 h — these are typical antecedent dry periods in Melbourne). Right — herbicide removal experiments which include two adjacent challenging spiking events followed by a dry weather period (~21 days — challenging dry conditions in Melbourne) and an extreme spiking event (4 PVs).

was best suited to the detection range of the measurement device — AquaFluor® Handheld Fluorometer (Turner) (0.4–200 μg/L), and it allowed for visualization of fluorescein in water. The first spiking test was conducted with a total of 2.5 PVs (1 PV = 3.5 m3) inflow dosed into the biofilter, which permitted examination of performance of the system during unsaturated conditions. The second spiking test (again 2.5 PVs) was conducted immediately after a flushing event (1.5 PVs), allowing examination of system behavior during challenging wetted conditions. During the Fluorescein FCT, the infiltration rate experienced fluctuations and presented an average hydraulic conductivity of 150 mm/h, with average retention time of ~3.3 h. The Herbicide FCT (November–December 2012) was used to examine the biofilter during challenging conditions (Fig. 1 left). Two 3 PV events were dosed with very short dry period in-between; analysis of the long term climatic data for Melbourne suggests that this

combination (3 PVs for consecutive events, 10 h drying) represents the 95th percentile of what is expected in the region (Zhang et al., 2014). The system was left to dry for a period of 21 days (again, representing the 95th percentile), which was then followed by a 4 PV event (the 95th percentile for single event). Inflow was generated by pumping water from the local stormwater pond, and spiking it with various chemicals to generate a semi-synthetic stormwater containing defined concentrations of the herbicides for the FCT. The target concentrations used for the chemicals were derived from a review of chemicals in urban stormwater (Zhang et al., 2014). The average inflow concentrations of triazines were 39 ± 4.0 μg/L (n = 3) for atrazine, 38 ± 5.0 μg/L (n = 3) for simazine and 35 ± 3.0 μg/L for prometryn (n = 3). Similar hydraulic conditions in Herbicides FCT were maintained as to the Fluorescein FCT, i.e. average hydraulic conductivity of 150 mm/h (average retention time of ~ 3.3 h). The detailed flow

Fig. 2. Left — ISC main dimensions. Top right — ISC before covering intake holes with the suction chamber. Bottom right — ISC with suction chamber sealed around intake holes.

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55

51

Fig. 3. Fluorescein outflow/inflow concentrations during series of continuous tests for both field challenge tests (FCT) and in-situ column (ISC) tests. E: Nash–Sutcliffe coefficient.

performance of the tested biofilter during Herbicide FCT is reported in (Zhang et al., 2014). During each FCT, 10–12 discrete inflow samples and 20 discrete outflow samples were collected using a flow weighted sampling regime. For the Fluorescein FCTs, all samples were analyzed in-situ using AquaFluor® Handheld Fluorometer (Turner) (detection range 0.4– 200 μg/L), while for the Herbicides FCT, all samples were analyzed for atrazine, simazine and prometryn in a NATA accredited commercial laboratory using GCMS with the limit of report of 2 μg/L. Soil moisture was measured at 250 mm from the surface of biofilter using Theta Probes (model ML2x, Delta-T Devices Cambridge, UK) over the whole series of FCT.

2.3. In-situ column tests 2.3.1. In-situ column design The in-situ column is a stainless steel column inserted into the biofilter media with minimal disturbance. The basic concept is to dose the column with inflow from the top and pump treated water that had passed through the biofilter media from the bottom. The stainless steel ISC is composed of three main elements (Fig. 2). The main body is made of a 1.2 m stainless steel (E316) column of 100 mm internal diameter. The wall thickness is 1.6 mm; thin enough to facilitate insertion into the biofilter with minimal media disturbance. The interior of the column was pre-brushed with sand paper to prevent edge effects. Water was pumped from drainage holes at 30 and 40 mm from the bottom. This position was chosen as a compromise to prevent suction from the surrounding water and to minimize the potential for a shortcircuiting through the column. Two ranges of 16 holes of 3 mm diameter each were drilled in a triangular pattern (Fig. 2, right top). These holes sizes had previously been tested and were sized to prevent sand media (d50 = 0.20 mm) passing through the holes and maintained a flow of up to 30 mL/min (corresponding hydraulic conductivity of 250 mm/h) and to prevent clogging. A suction chamber made of 1 mm thick stainless steel was sealed around the holes (Fig. 2, bottom right), creating a small 5 mL chamber. The third component is a small stainless pipe (1 mm internal diameter) that extends from the chamber to the top of the column allowing water to be pumped directly from the chamber. These ISCs were used in the biofiltration system in two phases: (1) Fluorescein ISC, which was used to compare directly to the Fluorescein FCT and (2) Herbicide ISC, which was used to compare directly with the Herbicide FCT.

2.3.2. Fluorescein ISC (July 2014) Two ISCs were challenged with a series of spiking and flushing tests following the same procedure as the Fluorescein FCT (Section 2.2). Both experiments were conducted under similar environmental conditions and the results were directly comparable. The two ISCs were installed along the middle line of the field biofilter 3 m from inlet and 3 m from outlet separately.

2.3.3. Herbicide ISC (July–August 2014) Another three ISCs were used and spiked with herbicides (atrazine, simazine and prometryn) as described above for the Herbicide FCT. The three ISCs were installed along the middle line of the field biofilter 2.0 m from inlet and 2.0 m from outlet, at the center point (4.8 m to the inlet and outlet). These results were compared to the Herbicide FCT. It should be noted that the Herbicide FCT were performed almost 2 years prior the ISC tests in slightly different conditions (as discussed in Results and discussion). It is acknowledged that ISCs were never inserted to the bottom of the biofilter, but had a gap of approximately 15 mm, enabling natural migration of water down through the columns even without pumping. In addition, after the installation of the ISCs, the filter media inside all the columns were compacted to some extent. This compaction resulted in a decrease of hydraulic conductivity from an average of 150 mm/h for the full biofilter to 143 ± 2.0 mm/h (~4.7% reduction) for the ISCs for the fluorescein test and 125 ± 5.0 mm/h (16% reduction) for the ISCs for herbicides test. For all ISCs, time-weighted composite inflow samples were collected. Two methods were trialed for the outflow sampling of the ISCs; (1) Saturated conditions — if the media was already saturated (i.e. because a spiking or flushing event had recently occurred) a constant pumping rate of 18 mL/min was applied and the ponding depth was kept constant at 200 mm, which corresponded to the average infiltration rate (143 ± 2.0 mm/h). This is 4.7% lower than the FCTs (150 ± 8.0 mm/h); (2) Unsaturated conditions — if the media had not been exposed to water recently, the pumping rate was adjusted to the real-time infiltration rates as the infiltration rate presented variations during the test. In the second method, after each 50 mL sampling, the pumps were disconnected and the ponding water infiltrated naturally permitting evaluation of the infiltration rate. The ponding oscillated between 200 and 220 mm but was always maintained above the soil surface. The pumping speed in both methods was carefully calibrated to be as close as possible to field infiltration rates and to ensure that the latter were not accelerated by pumping and that the water from the

52

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55

surrounding submerged zone of the whole biofilter was not sucked into the ISC. To track this, a more saline, higher electrical conductivity (EC) in the inflow was used (420 ± 5.0 μs/cm), in comparison with the biofilter background EC value (~100 μs/cm). After 1 PV (~1500 mL) of dosing, outflow EC values were stabilized at 415 ± 8.0 μs/cm, with which the maximum amount of mixing was estimated to be 4.1% of submerged zone water outside the column. This higher EC value of inflow was applied in our previous study and did not exert any impact upon the performance of biofilters (Zhang et al., 2014). All collected samples were analyzed as described for the FCT samples. It should be acknowledged that due to the limitation of using ISCs, the horizontal fluxes of the field biofilter may not be well represented in the ISCs. However, it is argued that the ISCs are still representative of field biofilters due to the following reasons: (1) the hydraulic conditions, including inflow volumes, average hydraulic conductivity and retention time, which are reported to be the most important operational conditions that affect stormwater biofilter performance (Hatt et al., 2009; Le Coustumer et al., 2009), were similar between ISC tests and FCT; (2) the two tests (ISC and FCT) were performed under same challenging conditions (e.g. wetting and drying conditions), which allowed similar conditions for two major processes for pollutant removal: adsorption (main process during wetting) and degradation (main process during dry); (3) the pre-treatment of the ISCs (i.e. interior side brushed by sand paper) and careful operation of the experiment on ISCs ensured that the edge effects and short-circuiting issues were minimized; and (4) the design/careful installation of the ISCs ensured little disturbance and the effective representation of soils in the ISCs. Indeed, the roots were also observed in the ISCs after demolishing of the ISCs. It is worth noting that the ISC installation and sampling methodologies are very important to ensure well representation of the ISCs compared with field biofilters. More details regarding column installation and sampling are given in Table S1 of the supplementary document. Moreover, in this study only three ISCs were used for tests given the relative small size of the systems (9.6 m × 1.4 m); it is recommended that more ISCs shall be used for larger systems to cover all the representative locations for the large system, and if possible three replicates shall be utilized for each location. 2.4. Data analysis To assess the similarity between FCT and ISC, the lag time, peak, slope and plateau were calculated for comparison. The plateau is defined as C∞/C0 for spiking only, while C∞ was calculated by taking the average value of last few points for each test of the spiking. All the measured points after the lag and before the plateau were used for calculation for spiking slope, whereas all the points after the peak and up to approximately ~ 1.7 PVs (after which the concentrations stabilized) were used for the flushing slope. The Nash–Sutcliffe coefficient (E), which is often applied in literature for assessment of models by comparing model prediction results against measured results (Nash and Sutcliffe, 1970; Kalin et al., 2003; Merz and Blöschl, 2004), was also used to compare the ISC to the FCT data; ISC data was used as predicted

results while the FCT data was used as measured results; here linear interpolation was utilized to obtain paired data. Event mean concentrations (EMCs), which represent a flow average concentration computed as the total load (mass) divided by the total runoff volume, were calculated in each event. EMC reductions and load reductions of the micropollutants were then calculated. 3. Results and discussion 3.1. Water balances A water balance, including inflow and outflow volumes collected was produced for the Fluorescein FCTs and all ISC tests (see Table S2 of the Supplementary material for more details). The estimated loss of water in FCTs was 5.7%, while this was 5.4% in the ISC tests. The close errors of water balance indicate the hydraulic performance of ISCs was not dissimilar to field conditions. More importantly, similar hydraulic errors would not exert a different influence on treatment performance. 3.2. Testing the ISC tool for fluorescein The fluorescein removal results from the Fluorescein FCTs and the Fluorescein ISC tests are shown in Fig. 3. ISC removal was similar to the FCT with similar curves observed. The estimated Nash–Sutcliffe coefficient indicated a good agreement between the ISC and FCT results for each individual test (E = 0.38–0.89), as well as when considering the whole series together (E = 0.83 for ISC-1 and 0.88 for ISC-2). The lowest E coefficient were obtained for the first flushing test, which had relatively poor agreement with FCT (E = 0.38 for ISC-1 and 0.53 for ISC2). This could be due to the longer dry period operated in FCT (62 h) than that in ISC (52 h); this variation resulted in differences in the level of degradation occurring over the dry periods. Indeed, if first order degradation is assumed (i.e. C(t) = C0 × e−αt), the calculated degradation rates for FCT and the two ISCs were very similar (0.015 h−1, 0.012 h−1 and 0.013 h−1, respectively). This difference was not found at the beginning of the second flushing tests as the FCT and ISC all had same length of dry periods (12 h) before the test. No major differences were found with regard to lag times of FCT and ISC tests, as fluorescein breakthrough from the biofilters occurred roughly at the same time (~0.5 PV) in the 1st spiking test. This indicated that the conditions of the systems for FCT and ISC tests were very similar, e.g. both had ~3 days of dry period before test and were conducted during the wet season (winter). Most importantly, the average ISC plateaus (0.97 for 1st spiking and 0.90 for 2nd spiking) were similar to the FCT (0.94 for 1st spiking and 0.88 for 2nd spiking) (Table 1). The slightly higher plateau may be due to the shorter filtration depth than that of the field (i.e. the ISC had a slightly lower filtration depth because of the small gap of ~15 mm between the ISC and biofilter bottom). The ISC underestimated the capacity of the biofilter to treat fluorescein by 2.2%, and was hence a conservative tool for validation. The differences between the two tests

Table 1 Comparison of slope, plateau, peak, EMC reduction and load reduction between the fluorescein ISC and FCT. 1st flushing test

1st spiking test FCT Fit (Nash–Sutcliffe coefficient) Lag time (pore volume) Slope Plateau (C/C0) Peak (C/C0) EMC reduction % Load reduction %

ISC

FCT

E = 0.62 0.50 0.62 0.94 1.02 42 39

Note: Average values of the two replicate ISC were presented.

FCT

E = 0.46 0.50 0.56 0.97 1.04 48 31

0.00 −0.42 0.59 0.58 67 67

2nd flushing test

2nd spiking test ISC

ISC

FCT

E = 0.89 0.00 −0.40 0.72 0.72 57 56

0.30 0.73 0.88 0.86 49 63

ISC E = 0.81

0.30 0.68 0.90 1.00 42 51

0.00 −0.52 1.00 1.02 58 59

0.00 −0.56 0.97 0.96 50 48

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55

53

Fig. 4. Performance of biofilters in removing herbicides in the field challenge test (FCT) and in-situ columns (ISC): a) atrazine, b) simazine and c) prometryn. E: Nash–Sutcliffe coefficient.

Table 2 Comparison of slope, plateau, peak, EMC reduction and load reduction of herbicides ISC and FCT. Atrazine

Spiking Test 1

Spiking Test 2

Spiking Test 3

FCT

FCT

FCT

ISC

ISC

ISC

E = 0.75 0.00 0.32 0.60 0.56 0.94 0.98 0.98 1.12 41 47 37 48

Simazine

Spiking Test 1

Spiking Test 2

Spiking Test 3

FCT

FCT

FCT

ISC

E = 0.36 0.00 0.00 0.11 0.12 0.96 0.96 1.06 0.98 18 17 23 19

E = −0.10 0.00 0.00 0.29 0.23 1.26 1.28 1.32 1.28 −7 11 −9 11

Fit Lag time Slope Plateau Peak EMC reduction % Load reduction %

ISC

Fit Lag time Slope Plateau Peak EMC reduction % Load reduction %

E = −0.19 0.00 0.45 0.65 0.41 1.05 0.80 1.12 0.80 30 70 26 72

Prometryn

Spiking Test 1

Spiking Test 2

Spiking Test 3

FCT

FCT

FCT

Fit Lag time Slope Plateau Peak EMC reduction % Load reduction %

ISC

E = −0.22 0.20 0.70 0.41 0.33 0.59 0.41 0.62 0.42 58 83 61 83

E = −0.71 0.00 0.00 0.12 0.11 1.18 1.14 1.29 1.20 −11 9 −4 11

ISC

ISC

E = −0.71 0.00 0.00 0.14 0.18 0.71 0.56 0.73 0.69 41 58 45 59

Note: Average values of the three replicate ISC were presented.

E = −0.34 0.00 0.00 0.21 0.15 1.16 1.04 1.23 1.32 0 20 6 20

were well within the accuracy of the concentration measurements (~3.0%). A peak was always observed at the start of the flushing test (Fig. 3; Table 1). This was due to the flushing out of fluorescein in the submerged zone from the previous spiking test, which is regarded as ‘old water’ (Chandrasena et al., 2013). The presence of the peak is important for validation because it implies a high concentration of pollutants in the outflow, and results herein indicated that the in-situ columns (Fig. 3) effectively represented submerged zone behavior. The slopes for the spiking and flushing tests (after the peak) are shown in Fig. 3 and summarized in Table 1. The spike slopes (average: 0.56/0.68 for 1st and 2nd spiking) and flush slopes (average: − 0.40/− 0.56 for the 1st and the 2nd flushing) do not present a pronounced difference from the field results (Table 1). Meanwhile, similar slopes also indicated there were no accelerations or decelerations of the filtration rate, showing that the sampling methods (by pumping using a calibrated rate) applied in this study were acceptable. With regard to the overall fluorescein reduction, ISC tests demonstrated conservative results as compared to FCT except for the 1st spiking tests, when the ISC overestimated the EMC reduction by 5.3%. These results indicate that ISC could be a conservative tool for validation monitoring, when taking into consideration small errors (e.g. 5.3%).

ISC

E = −0.34 0.00 0.00 0.34 0.28 0.97 0.71 1.0 0.93 22 63 27 63

3.3. Testing the ISC tool for herbicides The pollutographs of atrazine, simazine and prometryn over a series of tests using ISC and FCT are shown in Fig. 4, while the slopes, plateaus, peaks and reductions are summarized in Table 2. In general, the ISC and FCT data were notably different (Fig. 4), with only marginal similarities obtained between the ISC and FCT for atrazine (E = 0.36–0.75 for the

54

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55

Spiking Tests 1 & 2; E = −0.10 for the Spiking Test 3). Specifically, lag time of ~ 0.3 PV for atrazine, ~ 0.5 PV for simazine and ~ 0.7 PV for prometryn were observed in Spiking Test 1, which led to relatively lower peak outflow concentrations (up to 30% lower) in ISC test than in the FCT. Almost all reductions of herbicides in the ISC tests were higher than that in the FCT (Table 2). The larger deviations observed here between the FCT and ISCs for the herbicide test (compared with the very small differences seen between the FCT and ISCs for the fluorescein test) is explained by a number of subtle climatic differences between each of the tests conducted. Both the Fluorescein FCT and the Fluorescein ISC tests were conducted during winter months, hence experienced similar environmental temperatures during the tests (12–16 °C for FCT vs. 12–16 °C for ISCs) and similar high antecedent fortnightly rainfall volumes (28 mm for FCT vs. 40 mm for ISCs). However, the Herbicide FCT and Herbicide ISC tests were conducted in different seasons (summer for FCT vs. winter for ISC), and hence experienced different temperatures during the tests (18–40 °C for FCT vs. 10–20 °C for ISCs) and different antecedent fortnightly rainfall volumes (2.6 mm for FCT vs. 32.7 mm for ISCs). According to Chandrasena et al. (2014) lower antecedent rainfall volumes will cause the plants in the biofilter to uptake the water in the submerged zone. This means that the effluent from the 1st spiking of the Herbicide FCT was not diluted with submerged zone water, causing the concentrations to rise faster and reached higher plateaus than those of the Herbicide ISC tests (which had typical antecedent rainfall totals, and were hence expected to have normal saturated zone volumes). Bailey and White (1964) reported that adsorption of triazines favored lower temperatures, with greater adsorption found at 0 °C compared with 50 °C; hence the differences observed between the FCT and ISC could be caused by different operational temperatures. However, in the Spiking Test 2, when both FCT and ISC had similar antecedent conditions (10 h dry after Spiking Test 1), the results from FCT and ISC were more similar, with average plateau concentrations of atrazine, simazine and prometryn being just slightly lower (2 ~ 10%) than FCT, and the reductions between these two tests were closer (Table 2). In both ISC and FCT, the variability of outflow concentrations in Spiking Test 2 was high (coefficient of variation: 12.9%–27.7%), which was attributed to the challenging wet conditions as previously reported (Zhang et al., 2014). The largest differences were observed in Spiking Test 3; the average EMC reductions of atrazine, simazine and prometryn were 17%, 20% and 41% lower than that in FCT in Spiking Test 3 (Table 2). This could also be explained by the climate differences. High environmental temperature during FCT led to higher evapotranspiration over the 21 days dry period. The change of soil moisture over the dry period was estimated to be ~ 0.35 PV in FCT, which was considerably higher than that in ISC tests (~ 0.22 PV); consequently the greater loss of water in FCT led to the higher outflow concentration in Spiking Test 3. In addition, differences in natural precipitation between Spiking Test 2 and Spiking Test 3 (10.3 mm for FCT and 30.3 mm for ISC) resulted in flushing of more adsorbed triazines during ISC tests, as adsorption of triazines onto soils is mainly through weak physical mechanisms, such as protontransfer and hydrogen bonding (Martin-Neto et al., 1994; Flores et al., 2009). Moreover, more prometryn was adsorbed as a result of the release of adsorption sites after the flushing due to its relative higher adsorption capacity (LogKoc = 2.7), hence prometryn concentration started from a lower level compared with that of atrazine (LogKoc = 2.3) and simazine (LogKoc = 2.1). The differences in redox conditions might also contribute to the difference between ISC and FCT. ISC has a pipe that facilities the pumping for outflow samples and the pipe was not closed during the long dry period before Spiking Test 3. This might cause air-exchange between atmosphere and the bottom of the ISC, and consequently promote the biodegradation rate of the herbicides in ISC tests. These results highlight that both ISC and FCT data should be used only when taking into account the inherent boundary conditions of

the tests. This means that if ISC is to be used to validate these systems, or draw comparisons with challenge tests, it is important to ensure that the same boundary conditions are carefully chosen. This is particularly important for stormwater biofilters as they often experience highly variable conditions (Blecken et al., 2009; Hatt et al., 2009). In this study, Fluorescein FCT and ISC tests were conducted under similar testing schemes and also similar climatic conditions, hence good agreements were achieved (Fig. 3). However, although herbicides FCT and ISC tests were performed according to the same dry/wet schemes (Fig. 1 right), the antecedent conditions for Spiking Test 1 of the ISC were different from those of the FCT, as the ISC were tested in wet period (winter) while FCT were performed after a very long dry period (summer). 4. Conclusions A new in-situ column (ISC) tool was developed in this study as an alternative tool for the validation of the treatment of stormwater biofilters. The in-situ tool comprises a thin stainless steel column and a small chamber with holes for outflow sampling at the bottom, which can be inserted into field biofilters in a non-destructive manner. The tool was tested against a series of field challenge tests (FCT). The ISC tool is promising in reproducing the Fluorescein FCT results through a series of continuous tests that covered different operational conditions over a long term, with Nash–Sutcliffe coefficient of 0.83– 0.88 assuming the results from ISC as modeled data against the measured data of the field challenge test. Good agreements regarding the outflow time lags, slopes and peaks and plateaus were observed between the results collected from fluorescein ISC and FCT. The relatively lower EMC and load reductions produced by the ISC demonstrate ISC as a conservative tool for validation monitoring. The use of the in-situ column tool to study removal of herbicides (atrazine, simazine and prometryn) indicate that ISC tests produced relatively lower outflow concentrations compared with the field challenge tests, however, the differences were mainly attributed to the different climatic conditions occurring between the tests. The pumping pipe of the ISC may also cause the differences in the redox conditions between ISC and FCT, and should be closed off during dry weather periods. Same inherent boundary conditions should be considered if ISCs are to be harnessed as an alternative method to FCT for validation of stormwater biofilters. Further testing of the tool is required to establish how the ISC compares with FCT for identical antecedent conditions and a larger number of herbicides and micropollutants that may have very different potential removal mechanisms compared with the studied herbicides. Acknowledgment This study was funded by CRC for Water Sensitive Cities and Chinese Scholarship Council (CSC) (grant no. 2011609012).The authors also wish to acknowledge the support of Richard Williamson and Anthony Brosinsky. We also express our gratitude to Sylvie Barraud from INSA Lyon for her assistance. Louise Krol is also highly appreciated for her proof reading of language. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2015.11.150. References Bailey, G.W., White, J.L., 1964. Soil–pesticide relationships, adsorption and desorption of organic pesticides by soil colloids, with implications concerning pesticide bioactivity. J. Agric. Food Chem. 12 (4), 324–332. Becouze, C., Bertrand-Krajewski, J., Dembélé, A., Cren-Olivé, C., Coquery, M., 2009. Preliminary assessment of fluxes of priority pollutants in stormwater discharges in two urban catchments in Lyon, France. System 2, 3–10.

K. Zhang et al. / Science of the Total Environment 544 (2016) 48–55 Blecken, G.-T., Zinger, Y., Deletić, A., Fletcher, T.D., Viklander, M., 2009. Influence of intermittent wetting and drying conditions on heavy metal removal by stormwater biofilters. Water Res. 43 (18), 4590–4598. Braeckevelt, M., Rokadia, H., Imfeld, G., Stelzer, N., Paschke, H., Kuschk, P., Kästner, M., Richnow, H.-H., Weber, S., 2007. Assessment of in situ biodegradation of monochlorobenzene in contaminated groundwater treated in a constructed wetland. Environ. Pollut. 148 (2), 428–437. Bratieres, K., Fletcher, T.D., Deletic, A., Zinger, Y., 2008. Nutrient and sediment removal by stormwater biofilters: a large-scale design optimisation study. Water Res. 42 (14), 3930–3940. Brown, J.N., Peake, B.M., 2006. Sources of heavy metals and polycyclic aromatic hydrocarbons in urban stormwater runoff. Sci. Total Environ. 359 (1–3), 145–155. Chandrasena, G.I., Deletic, A., Ellerton, J., McCarthy, D.T., 2012. Evaluating Escherichia coli removal performance in stormwater biofilters: a laboratory-scale study. Water Sci. Technol. 66 (5), 1132–1138. Chandrasena, G., Deletic, A., McCarthy, D., 2013. Evaluating Escherichia coli removal performance in stormwater biofilters: a preliminary modelling approach. Water Sci. Technol. 67 (11), 2467–2475. Chandrasena, G., Pham, T., Payne, E., Deletic, A., McCarthy, D., 2014. E. coli removal in laboratory scale stormwater biofilters: influence of vegetation and submerged zone. J. Hydrol. 519, 814–822. Chevron Cottin, N., Merlin, G., 2007. Study of pyrene biodegradation capacity in two types of solid media. Sci. Total Environ. 380 (1–3), 116–123. DHV, 2013. Guidelines for Validating Treatment Processes for Pathogen Reduction — Supporting Class A Water Recycling Schemes in Victoria. Department of Health, Victoria. Fletcher, T., Mitchell, V., Deletic, A., Ladson, T., Seven, A., 2007. Is stormwater harvesting beneficial to urban waterway environmental flows? Water Sci. Technol. 55 (4), 265–272. Flores, C., Morgante, V., González, M., Navia, R., Seeger, M., 2009. Adsorption studies of the herbicide simazine in agricultural soils of the Aconcagua valley, central Chile. Chemosphere 74 (11), 1544–1549. Geyer, R., Peacock, A.D., Miltner, A., Richnow, H.H., White, D.C., Sublette, K.L., Kästner, M., 2005. In situ assessment of biodegradation potential using biotraps amended with 13 C-labeled benzene or toluene. Environ. Sci. Technol. 39 (13), 4983–4989. Hatt, B.E., Deletic, A., Fletcher, T.D., 2006. Integrated treatment and recycling of stormwater: a review of Australian practice. J. Environ. Manag. 79 (1), 102–113. Hatt, B.E., Fletcher, T.D., Deletic, A., 2009. Hydrologic and pollutant removal performance of stormwater biofiltration systems at the field scale. J. Hydrol. 365 (3–4), 310–321. Jeng, H.A.C., Englande, A.J., Bakeer, R.M., Bradford, H.B., 2005. Impact of urban stormwater runoff on estuarine environmental quality. Estuar. Coast. Shelf Sci. 63 (4), 513–526. Kalin, L., Govindaraju, R.S., Hantush, M.M., 2003. Effect of geomorphologic resolution on modeling of runoff hydrograph and sedimentograph over small watersheds. J. Hydrol. 276 (1), 89–111. Le Coustumer, S., Fletcher, T.D., Deletic, A., Barraud, S., Lewis, J.F., 2009. Hydraulic performance of biofilter systems for stormwater management: influences of design and operation. J. Hydrol. 376, 16–23.

55

Mandelbaum, R.T., Shati, M.R., Ronen, D., 1997. In situ microcosms in aquifer bioremediation studies. FEMS Microbiol. Rev. 20 (3–4), 489–502. Martin-Neto, L., Vieira, E.M., Sposito, G., 1994. Mechanism of atrazine sorption by humic acid: a spectroscopic study. Environ. Sci. Technol. 28 (11), 1867–1873. Merz, R., Blöschl, G., 2004. Regionalisation of catchment model parameters. J. Hydrol. 287 (1), 95–123. Nash, J., Sutcliffe, J., 1970. River flow forecasting through conceptual models part I—a discussion of principles. J. Hydrol. 10 (3), 282–290. NHMRC-NRMMC, 2011. Australian Drinking Water Guidelines. National Health and Medical Research Council and Natural Resource Management Ministerial Council, Canberra. Nielsen, P.H., Albrechtsen, H.-J., Heron, G., Christensen, T.H., 1995. In situ and laboratory studies on the fate of specific organic compounds in an anaerobic landfill leachate plume, 1. Experimental conditions and fate of phenolic compounds. J. Contam. Hydrol. 20 (1–2), 27–50. Philp, M., McMahon, J., Heyenga, S., Marinoni, O., Jenkins, G., Maheepala, S., Greenway, M., 2008. Review of Stormwater Harvesting Practices. Urban Water Security Research Alliance. Read, J., Wevill, T., Fletcher, T., Deletic, A., 2008. Variation among plant species in pollutant removal from stormwater in biofiltration systems. Water Res. 42 (4–5), 893–902. Sabatini, D.A., 2000. Sorption and Intraparticle Diffusion of Fluorescent Dyes with Consolidated Aquifer Media. Ground Water. 38 (5), 651–656. Smart, P.L., Laidlaw, I.M.S., 1977. An evaluation of some fluorescent dyes for water tracing. Water Resour. Res. 13 (1), 15–33. Stelzer, N., Büning, C., Pfeifer, F., Dohrmann, A.B., Tebbe, C.C., Nijenhuis, I., Kästner, M., Richnow, H.H., 2006. In situ microcosms to evaluate natural attenuation potentials in contaminated aquifers. Org. Geochem. 37 (10), 1394–1410. USEPA, 2005. Membrane Filtration Guideline Manual. Office of Water, Cincinnati. Ying, G.-G., Toze, S., Hanna, J., Yu, X.-Y., Dillon, P.J., Kookana, R.S., 2008. Decay of endocrine-disrupting chemicals in aerobic and anoxic groundwater. Water Res. 42 (4–5), 1133–1141. Zgheib, S., Moilleron, R., Chebbo, G., 2012. Priority pollutants in urban stormwater: part 1 — case of separate storm sewers. Water Res. 46 (20), 6683–6692. Zhang, S., Guo, Y., 2014. Stormwater capture efficiency of bioretention systems. Water Resour. Manag. 28 (1), 149–168. Zhang, K., Randelovic, A., Page, D., McCarthy, D.T., Deletic, A., 2014. The validation of stormwater biofilters for micropollutant removal using in situ challenge tests. Ecol. Eng. 67, 1–10. Zhang, K., Deletic, A., Page, D., McCarthy, D.T., 2015a. Surrogates for herbicide removal in stormwater biofilters. Water Res. 81 (0), 64–71. Zhang, K., Randelovic, A., Aguiar, L.M., Page, D., McCarthy, D.T., Deletic, A., 2015b. Methodologies for pre-validation of biofilters and wetlands for stormwater treatment. PLoS ONE 10 (5), e0125979.