Contamination risk of raw drinking water caused by PFOA sources along a river reach in south-western Finland

Science of the Total Environment 541 (2016) 74–82

Contents lists available at ScienceDirect

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

Contamination risk of raw drinking water caused by PFOA sources along a river reach in south-western Finland Maiju Happonen a,⁎, Harri Koivusalo a, Olli Malve b, Noora Perkola c, Janne Juntunen b, Timo Huttula b a b c

Aalto University, Department of Civil and Environmental Engineering, P.O.B. 11000, 00076 Aalto, Finland Finnish Environment Institute, Freshwater Centre, P.O.B. 140, 00141 Helsinki, Finland Finnish Environment Institute, Laboratory Centre, P.O.B. 140, 00141 Helsinki, Finland

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

• Transport of PFOA was simulated in River Kokemäenjoki in Finland. • River and wastewater PFOA mass flows were determined. • Communal wastewater treatment plants caused only 11% of the total PFOA load. • The concentration of PFOA in raw drinking water remains on a safe level.

a r t i c l e

i n f o

Article history: Received 16 July 2015 Received in revised form 1 September 2015 Accepted 1 September 2015 Available online 21 September 2015 Editor: D. Barcelo Keywords: PFOA Wastewater treatment plant Contamination Raw drinking water River modeling Risk assessment

a b s t r a c t Transport of perfluorooctanoic acid (PFOA) was simulated in the beginning of River Kokemäenjoki in Finland using one-dimensional SOBEK river model. River Kokemäenjoki is used as a raw water source for an artificial groundwater recharge plant, and the raw water intake plant is located near the downstream end of the model application area. Measured surface water and wastewater concentrations were used to determine the PFOA input to the river and to evaluate the simulation results. The maximum computed PFOA concentrations in the river at the location of the raw water intake plant during the simulation period Dec. 1, 2011–Feb. 16, 2014 were 0.92 ng/l and 3.12 ng/l for two alternative modeling scenarios. These concentration values are 2.3% and 7.8%, respectively, of the 40 ng/l guideline threshold value for drinking water. The current annual median and maximum PFOA loads to the river were calculated to be 3.9 kg/year and 10 kg/year respectively. According to the simulation results, the PFOA load would need to rise to a level of 57 kg/year for the 40 ng/l guideline value to be exceeded in river water at the raw water intake plant during a dry season. It is thus unlikely that PFOA concentration in raw water would reach the guideline value without the appearance of new PFOA sources. The communal wastewater treatment plants in the study area caused on average 11% of the total PFOA load. This raises a concern about the origin of the remaining 89% of the PFOA load and the related risk factors. © 2015 Elsevier B.V. All rights reserved.

1. Introduction

Abbreviations: CWWTP, communal wastewater treatment plant; LOD, limit of detection; LOQ, limit of quantification; PFAA, perfluorinated alkyl acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctanesulfonic acid; WIP, water intake plant; WWTP, wastewater treatment plant. ⁎ Corresponding author.

http://dx.doi.org/10.1016/j.scitotenv.2015.09.008 0048-9697/© 2015 Elsevier B.V. All rights reserved.

Perfluorinated alkyl acids (PFAAs) are a group of emerging pollutants that have been used in industry and consumer products from the 1950s (Lau et al., 2007). Because of their fluorine-carbon chain PFAAs are inert, show resistance to high temperatures and repel oil and water, which makes them an ideal coating material for consumables,

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

such as clothes and furniture (Kissa, 2001). Other common uses of PFAAs include aviation hydraulic fluids, firefighting foams, paints and metal plating industry (Renner, 2001). The PFAA compound found the most frequently in European wastewaters is perfluorooctanoic acid (PFOA) (Loos et al., 2013). PFOA is extremely persistent in the environment, since it is not affected by biodegradation (Liou et al., 2010) or photodegradation (Vaalgamaa et al., 2011). It is not significantly removed in wastewater treatment (Schultz et al., 2006) and some studies have even reported higher PFOA concentrations in wastewater effluents than influents, presumably because of biodegradation of its precursors (Becker et al., 2008; Murakami et al., 2009; Schultz et al., 2008). Unlike most other persistent organic pollutants, PFOA is water soluble and found in animals in serum rather than in fat (Post et al., 2012), and its partitioning to sediment is low (Ahrens et al., 2010b, 2011). Because of these properties PFOA can be transported long distances with water. Traces of PFOA have been detected even in Arctic areas far from all emission sources (Butt et al., 2010; Lau et al., 2007). Another long range transport pathway for PFOA is suggested to be the transport of its volatile precursors in the atmosphere. Zareitalabad et al. (2013) have reviewed a large number of PFOA concentrations in surface waters and wastewater around the world. Out of all the reported surface water concentrations for PFOA half were in the range of 0.8–13 ng/l and the median concentration was 3.1 ng/l. In Finland Perkola (2014) reported PFOA concentrations in five rivers to be lower, 0.08–1.51 ng/l. For wastewater Zareitalabad et al. (2013) reported a median PFOA concentration of 27 ng/l, and Loos et al. (2013) found the maximum, mean and median PFOA concentrations of 90 European wastewater treatment plants (WWTP) to be 15,900 ng/l, 255 ng/l and 12.9 ng/l, respectively. These studies reveal the wide occurrence of PFOA in the environment and the potential role of WWTPs as its point sources. PFOA has also been detected in drinking water sources and finished tap water (Post et al., 2012), which raises concerns about the safety of drinking water, as exposure to PFOA has been linked to several adverse health effects in humans (Barry et al., 2013; Frisbee et al., 2010; Lam et al., 2014; Melzer et al., 2010; Steenland et al., 2010). Currently the use of PFOA is not regulated by any international agreements, but it has been proposed for restriction at the European Chemicals Agency (2014). Also in 2006 eight major manufacturers of PFOA committed to a voluntary program to reduce PFOA facility emissions and its content in finished products (USEPA, 2006). As reviewed by Zushi et al. (2012), some guideline values have been set for drinking water concentrations: 40 ng/l (NJDEP, 2007), 300 ng/l including both PFOA and PFOS (German Drinking Water Commission, 2006), 300 ng/l (UK Drinking Water Inspectorate, 2009) and 400 ng/l (USEPA, 2009). There is also an Italian guideline value of 500 ng/l (Regione del Veneto, 2014). PFOA is not regulated and no guideline values have been suggested for it in Finland. PFOA can enter surface water or groundwater from several sources, including industrial air emissions, industrial and domestic wastewater, storm water runoff, land application of biosolids, and release of firefighting foams (Post et al., 2012). Methods that have been used to assess the importance of different PFAA sources include mass balance calculations (Filipovic et al., 2013; Scott et al., 2010; Takazawa et al., 2009), modeling applications (Earnshaw et al., 2014; Paul et al., 2012), and estimation of correlations between PFAA concentrations and for example population or catchment surface area (Müller et al., 2011; Murakami et al., 2008; Pistocchi and Loos, 2009; Takazawa et al., 2009). Different studies place varying emphasis on the importance of the known PFOA sources: communal wastewater treatment plants (CWWTPs) (Becker et al., 2008; Müller et al., 2011; Perkola and Sainio, 2013; Yu et al., 2009), industrial WWTPs (Murakami et al., 2008; Pistocchi and Loos, 2009; Takazawa et al., 2009) and atmospheric deposition (Filipovic et al., 2013; Scott et al., 2010) have all been reported to be major PFOA sources. According to one estimate, 60% of perfluorocarboxylates (including PFOA) are emitted originally from fluoropolymer factories,

75

out of which 23% is distributed to air, 65% to water and 12% to land (Prevedouros et al., 2006). The elevated levels of PFOA in the environment, especially in drinking water sources, and the uncertainty about the sources of PFOA, call for raw water intake risk assessment and modeling the fate and transport of PFOA. For example Harada et al. (2003) reported a case where Tama River was heavily contaminated by PFOS originating from a WWTP. Because the downstream of Tama River was used for intake of raw drinking water, also the drinking water was contaminated and PFOS concentrations in tap water were measured to be as high as 51 ng/l. Concerns about drinking water contamination and damage to the vulnerable esker ecosystem rose in Finland after a regional water company, Turku Region Water Ltd., in 1999 announced its plans to extract raw water from River Kokemäenjoki to be used in artificial groundwater recharge in the Virttaankangas esker (Lyytimäki and Assmuth, 2014). Despite the public concern the artificial groundwater recharge system (Fig. SD 1 in Supplementary data) was built and taken into use in 2011 to provide drinking water to the 285 000 inhabitants of the Turku region. The background information about the artificial groundwater recharge plant and the related public debate has been covered in detail by Lyytimäki and Assmuth (2014). In this study the transport of PFOA and an artificial sweetener acesulfame was modeled using SOBEK river model that was parameterized to describe a 100 km distance of a water course in the Kokemäenjoki River basin. The aim was to assess the transport of the compounds from CWWTPs and the main tributaries to the downstream location where raw water is extracted for the artificial groundwater recharge plant. Since water soluble contaminants such as PFOA are not effectively removed during infiltration and percolation through soil layers, the contaminants in the raw water can remain in the finished drinking water produced at the plant (Davis et al., 2007). The main objectives of the study were to find out whether PFOA concentration in raw water can exceed a safe level and what percentage of the total PFOA load to River Kokemäenjoki is caused by the CWWTPs of the study area. Acesulfame, which was found in surface waters in much higher concentrations than PFOA, was used as a surrogate variable to assess the performance of the water quality model. 2. Materials and methods 2.1. SOBEK river model and the study area SOBEK is a modeling suite developed for integral water solutions by Deltares in The Netherlands. It involves seven modules, which can be combined for different modeling purposes related to water quantity and quality (Deltares, 2014). This study utilized two modules, “D-Flow 1 D Open Water” and “D-Water Quality 1 D”. “D-Flow 1 D Open Water” is a one-dimensional hydraulic model that can be used to model water velocity and level in rivers. It describes water flow as a numerical solution of the complete de Saint Venant equations. “D-Water Quality 1 D” is a one-dimensional water quality model that can simulate transport and mixing of substances using a numerical solution of the advection– diffusion equation. It also includes additional water quality processes, such as sorption and degradation, and supports simultaneous simulation of multiple substances. The flow model was parameterized to describe water flow in a river reach and three connected lakes in the Kokemäenjoki area as a part of project CONPAT (Assmuth et al., 2015; Perkola et al., 2015). The flow model was available from the project for this study and it was tested by Happonen (2015), where the model performance was assessed against measured river and lake water levels and river flows. The flow model simulated the water velocity and water level for a 100 km distance from River Nokianvirta to the Kolsi hydropower plant that is located in River Kokemäenjoki (Fig. 1). The model included 676 river and lake bed cross sections, which were determined based on sonar data and water depth maps. The raw water intake plant (WIP) is located in

76

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

Fig. 1. The modeling area, 8 wastewater treatment plants and 13 surface water sampling locations.

Karhiniemi before the Kolsi hydropower plant in River Kokemäenjoki. The Kolsi hydropower plant is situated approximately 55 km from the mouth of Kokemäenjoki, where the river discharges to the northern part of the Baltic Sea, the Gulf of Bothnia. Water flows into the modeling area through the two upstream boundaries in River Nokianvirta and Siuronkoski (Fig. 1). The distance to the WIP is 63 km from the Nokianvirta boundary and 55 km from the Siuronkoski boundary. There are five hydropower plants in the modeling area: one in Nokianvirta, one in Siuronkoski, and three in River Kokemäenjoki. All of the hydropower plants practice heavy flow regulation on a daily, weekly and yearly scale. The yearly average flows for River Nokianvirta, tributary Siuronkoski and River Kokemäenjoki near the WIP are 134 m3/s, 27 m3/s and 174 m3/s respectively. Hourly or daily water levels and discharges were available from all of the hydropower plants, and they were used as model input and for model evaluation. The upstream boundary condition was set to the measured water level at Nokianvirta. The downstream boundary condition was set to the measured water level at the Kolsi hydropower plant 36 km downstream of the WIP. Time series of discharge were used as model input for all but the last hydropower plant. In this study the wastewater treatment plants, their wastewater flows and contaminant mass loads were added to the model and treated as contaminant sources. There are five CWWTPs in the modeling area, and two more, Viinikanlahti and Rahola, in the city of Tampere, from where water flows into the modeling area through Nokianvirta (Fig. 1). The total wastewater flow of these CWWTPs is about 155 000 m3/d and they serve a population of 307 000. Out of the population 85% is covered by the two largest CWWTPs that are situated in the city of Tampere and discharge into Lake Pyhäjärvi. The load was introduced to the model near the upstream boundary in Nokianvirta, because Lake Pyhäjärvi was not included in the modeling area. This gives an estimate of the maximum effect that the Viinikanlahti and Rahola WWTPs could have on the raw water at the WIP, even if the timing of the load differs from reality. The surface water sampling locations were described as monitoring stations in SOBEK to facilitate the evaluation of modeled concentrations against the measurements.

2.2. Sampling and chemical analysis The wastewater and surface water samples were taken during seven sampling campaigns in 2012–2014. The campaigns during this period were scheduled to collect a set of samples that represent the annual water regime (winter low stage, spring flood, summer low stage and autumn flood). During each campaign surface water samples were taken as grab samples from 13 different locations (Fig. 1). Samples from the top water layer were taken with an open bottle, and other samples were taken with a bottle with an opening mechanism at a fixed depth. Wastewater samples were taken as 24 h composite samples from seven CWWTPs and three industrial WWTPs. The industrial WWTPs do not represent PFAA target industries, but one of them had a noticeable outflow of PFOA, so it was included in the model application. Samples were collected in 1 l polypropene bottles that were rinsed with methanol. Sample analysis was performed at the Finnish Environment Institute, which is accredited by the Finnish Accreditation Service (FINAS) as an environmental testing laboratory T003 following the standard SFS-EN ISO/IEC 17025. The samples were transported to laboratory in cool boxes within 48 h, and analyzed within one week from sampling or stored frozen. After addition of internal standards (e.g. d4-acesulfame potassium and 13C4-PFOA), the samples were extracted and cleaned with solid phase extraction. Evaporated and reconstituted extracts were analyzed with liquid chromatography negative electrospray ionization tandem mass spectrometry (LC–ESI–MS/MS, Waters, Milford, MA, USA). The analytical procedure is described in detail in Supplementary data. The limit of quantification (LOQ) and limit of detection (LOD) for PFOA were 0.5 and 0.1 ng/l, respectively, for wastewater, and 0.25 and 0.05 ng/l for surface water. The LOQ and LOD for acesulfame were 200 and 20 ng/l, respectively, for wastewater, and 25 and 2.0 ng/l for surface water. For quality assurance and control, procedural blanks and spiked control samples were prepared in each sample set. A field blank was collected and analyzed to ensure that sampling or containers did not affect the results. All blank concentrations were below LOD. The expanded method uncertainties (U) that were calculated using spiked control samples were 30% for PFOA and 34% for acesulfame (Table SD1 in Supplementary data).

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

2.3. Modeling scenarios and assumptions In order to assess the model performance, the simulated concentrations were compared with the measured surface water concentrations. Then to assess the PFOA contamination risk of raw water, three scenarios were simulated: In scenario 1 the model was run with an input of loads that change in time according to the concentration series interpolated from the measured concentrations. In scenario 2 the model was run using constant maximum loads that were based on the maximum values of the measured concentrations in order to find out the maximum concentration the current PFOA loads can cause. In scenario 3 we estimated the magnitude of a continuous constant PFOA load that would cause the PFOA concentration at the WIP to rise to a harmful level. For a threshold level of PFOA concentration in the river water we chose the guideline value of 40 ng/l (NJDEP, 2007). Finally, using scenario 1, we modeled the distribution of the PFOA load between different sources to find out how much the CWWTPs contribute to the total PFOA load at the WIP. As PFOA practically does not degrade in the environment (Liou et al., 2010; Vaalgamaa et al., 2011) or absorb to sediment (Ahrens et al., 2010b, 2011), it was modeled as a conservative substance unaffected by these processes. Acesulfame was found in much larger concentrations and more consistently above the detection limit than PFOA, so it was used for the evaluation of the water quality model. Because of its abundant use and high stability, acesulfame has been suggested for use as a chemical tracer to indicate domestic wastewater contamination (Buerge et al., 2009; Harwood, 2014; Tran et al., 2014). Like PFOA, acesulfame is resistant against degradation in surface waters and it is hydrophilic (Buerge et al., 2009), so it was also modeled as a conservative substance. In order to differentiate between the different contaminant sources, each source was assigned its own tracer. SOBEK 1DWAQ module offers five conservative and five decayable tracers. For the decayable tracers the decay rate was set so small (10−7 d) that they behaved as conservative tracers, giving us ten conservative tracers in total. For each source, the load originating from that source was input as both the source specific tracer and one additional tracer that was common for all of the sources. This type of tracer modeling allowed us to see at any point of the model both the effect of any individual source, and the effect of all sources combined on the concentration of the river water. PFOA or acesulfame loads originating from the WWTPs were fed into the model as mass per time (μg/s). However concentrations (μg/m3) were used for the water incoming from the model boundaries, because mass loads combined with the short term water flow regulation of the hydropower plants created unrealistic concentrations: for example in Nokianvirta the measured PFOA concentrations ranged 0.43–0.96 ng/l, but during a dry period the modeled concentrations rose up to 28 ng/l when mass loads were used. The high concentrations were created when a constant mass load was discharged during a very low flow rate. In reality the load would have already mixed in the incoming water during its transport in Lake Pyhäjärvi and the preceding waterbodies, and such high concentrations would not occur. The simulation period was Dec. 1, 2011–Feb. 16, 2014 (809 days) in all model runs. The time step for the simulation was set to 10 min and output time step to 24 h. The seven measured mass loads or concentrations for each source were fed for the corresponding sampling dates, and between the sampling dates the values were linearly interpolated. 3. Results and discussion 3.1. Comparison of modeled concentrations with measurements The measured PFOA concentrations varied in communal wastewater from 1.0 to 88 ng/l with a median of 4.6 ng/l and average of 16 ng/l. In surface waters (locations shown in Fig. 1) they varied from b LOD (b0.05 ng/l) to 5.8 ng/l, with a median of 0.64 ng/l and average of

77

1.0 ng/l. Only three of the measured PFOA concentrations were b LOD and they were set to 0. The levels of PFOA in River Kokemäenjoki were thus quite low compared to levels in most other countries: the median concentration 0.64 ng/l is about a fifth of the worldwide median 3.1 ng/l reported by Zareitalabad et al. (2013). The measured acesulfame concentrations were considerably higher, in communal wastewater 6 820–56 600 ng/l with a median of 36 700 ng/l and average of 34 400 ng/l. In surface waters they varied from 39.0–1740 ng/l, with a median of 279 ng/l and an average of 317 ng/l. The measured and modeled acesulfame and PFOA surface water concentrations at sampling locations 5–13 are presented in Fig. 2. The measured and modeled concentrations were the same at location 5 near the Nokianvirta boundary, where the input concentration was set equal to the measured value. The locations are in order from upstream to downstream, although it is worth to note that not all samples were taken in temporal order from upstream to downstream. For acesulfame (Fig. 2 a-f) the mean absolute error divided by the range of measured concentrations was 12–30% in different locations, or 15–20% without the 5th sampling round (Fig. 2 e). For PFOA (Fig. 2g–l) the error was 18–50% in different locations. The modeled concentrations are alternately higher and lower than the measured concentrations, suggesting that the model error is not systematic. The differences between the measured and simulated concentrations seen in Fig. 2 are not only due to uncertainties in the model: As the surface water samples were taken as grab samples, the concentrations fluctuated greatly in different locations especially in the case of PFOA, while the modeled concentrations were highly dependent on the concentrations of incoming water, as shown in Section 3.3. The fluctuation in the measured concentrations due to incomplete mixing was demonstrated by samples taken at the same time from three different depths in Lake Pyhäjärvi (not located in the modeling area): for example the first sampling campaign results for PFOA were 1.82, 1.10 and 3.85 ng/l taken from the depths of 1 m, 10 m and 40 m, respectively. Meanwhile in a one-dimensional model the water is always completely mixed in the cross section. The modeling area is shallower than Lake Pyhäjärvi, with water depth ranging approximately 1.5–9 m. Stratification is unlikely in the actual river reach of River Kokemäenjoki, which has a width of the order of 100 m. However the lakes preceding River Kokemäenjoki have a width of the order of 1000 m and thus a much lower water velocity, so there stratification is likely. This is why in the future it would be recommendable to take samples from different depths of the waterbody and use the average concentration in onedimensional modeling. It is likely that the uncertainty originating from sampling is greater than the uncertainty in the model: Happonen (2015) studied the transport of a short contaminant pulse and found that there was numerical dispersion in the model. However the numerical dispersion affected notably only a short pulse, and its effect on continuous contaminant loads used in this study was very small. Another possible source of model uncertainty is an error in the water flows, but as the flow was forced to match the measured flows of the hydropower plants through the model, the flow error should be relatively small. Our model performance results are in line with other PFAA modeling studies: Earnshaw et al. (2014) reported about 2.0 or 2.5 times higher modeled PFOA concentrations compared to measured concentrations based on different parameters. For PFOS the maximum ratios were 4.6 and 3.8. The ratios of modeled PFOS concentrations to observed concentrations of Miyake et al. (2014) were 0.79–1.8 for surface seawater, 0.73–1.4 for intermediate seawater and 0.59–1.3 for bottom seawater. In our study the ratios ranged 0.1–2.3 for PFOA, when excluding the measured results that were bLOQ (locations 10–13 in Fig. 2 h and location 8 in Fig. 2 l). The measured PFOS concentrations were considerably higher than the PFOA concentrations in our study: from b LOQ (0.3 ng/l) to 20.7 ng/l (Earnshaw et al., 2014) and 1.5–7.3 ng/l (Miyake et al., 2014). Thus the uncertainty from sample analysis was likely lower in these studies.

78

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

Fig. 2. The measured and modeled concentrations of acesulfame (a–f) and PFOA (g–l) at surface water sampling locations 5–13 during six sampling campaigns: 1) Oct. 2012, 2) Jan. 2013, 3) Apr. 2013, 4) Aug. 2013, 5) Oct. 2013, 6) Jan. 2014. Note the differing concentration scale in figure g.

3.2. Contamination risk of raw water used in artificial recharge of groundwater In order to find out the probable range of PFOA concentrations in raw water under the current PFOA load, the transport of PFOA was simulated using scenario 1 with load input based on the measured concentrations, and scenario 2 with constant maximum loads determined from

the maximum values of the measured concentrations (Fig. 3). The PFOA concentrations at the WIP were 0.36–0.92 ng/l in scenario 1, and 1.17–3.12 ng/l in scenario 2. The modeled maximum concentrations 0.92 ng/l and 3.12 ng/l are 2.3% and 7.8%, respectively, of the NJDEP (2007) guideline value 40 ng/l. It is thus very unlikely that PFOA concentration in raw water at the WIP would reach a harmful level with the current PFOA load. Cornelis et al. (2012) have estimated in Belgium

Fig. 3. The modeled PFOA concentration at the raw water intake plant with measured (scenario 1) or maximum (scenario 2) PFOA loads as model input.

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

that when drinking water PFOA concentrations are 1–2 ng/l, drinking water forms only around 1% of the total PFOA exposure. This indicates that with the current raw water PFOA concentrations food is likely to be a considerably more important source of PFOA to humans than drinking water. In scenario 3, we determined the magnitude of a continuous PFOA load that would cause the PFOA concentration at the WIP to exceed the guideline value of 40 ng/l (Fig. 4). By trial and error the critical load was found to be approximately 57 kg/year. The size of the critical load depends both on the flow situation and the distance of the emission point from the WIP. At the assumed continuous load level of 57 kg/year, the threshold of 40 ng/l was exceeded during a dry period in October 2013. Fig. 4 presents the effect of the 57 kg/year load on PFOA concentration at the WIP when the load is released from two alternative emission points: from Äetsä 7 km away and Nokianvirta 63 km away from the WIP. For comparison, the total median and maximum PFOA loads of the WWTPs are 0.38 and 1.0 kg/year, and the total median and maximum loads of all sources are 3.9 and 10 kg/year, respectively. The total PFOA load would probably only reach a magnitude of 57 kg/year if new PFOA target industry was founded in the area. However Post et al. (2012) have questioned whether the reported guideline values are strict enough in the light of recent studies on the health effects of PFOA. It must also be noted that even if PFOA alone does not exceed a harmful level in drinking water, there is no guarantee it would not cause harmful effects together with other similar substances such as other PFAAs. This is why there is a need to continue monitoring PFAA levels in drinking water sources and study the collective effects of harmful substances. One suggested approach for the latter is whole effluent assessment of wastewater (Finnish Environment Institute, 2010). 3.3. The role of CWWTPs as PFOA sources to surface waters Fig. 5 presents the distribution of PFOA load between the sources, including the WWTPs and the mass loads through the model boundaries in Nokianvirta and Siuronkoski. The loads were simulated according to modeling scenario 1. In reality the load incoming from Nokianvirta includes the loads of Viinikanlahti and Rahola WWTPs, but here their loads were subtracted from the Nokianvirta load, as the loads from Viinikanlahti and Rahola are presented separately. In other words the red and light blue areas together in Fig. 5 form the load

79

incoming through the boundary at Nokianvirta, and the light blue area represents the part that at most can be attributed to the WWTPs of Tampere. According to the simulation results all WWTPs combined caused on average 11% of the total PFOA flow that reaches the WIP. The WWTPs located in the modeling area (i.e. excluding Viinikanlahti and Rahola) caused on average 5% of the total PFOA load. By far the largest PFOA load came through the Nokianvirta boundary from sources other than the Tampere WWTPs (on average 74%), and the second largest from the Siuronkoski boundary (15%). Out of the WWTPs only Äetsä (1%), Kullaanvuori (3%) and Viinikanlahti and Rahola (6%) had a noticeable effect on the total PFOA load. The effect of the industrial WWTP was negligible. In addition to the model simulations, a simple mass balance calculation was used to get a straightforward estimate of the importance of different contaminant sources: Using the measured wastewater concentrations and wastewater flows, we calculated a median PFOA load for each WWTP and summed them to get a total median load for all the WWTPs in the area, 11.9 μg/s (0.38 kg/year). Then using the median flow of a hydropower plant near the WIP we calculated the concentration that the WWTP PFOA load could cause if it was discharged and immediately mixed into the river water right before the WIP. The median flow of the river was 164 m3/s, so the load of 11.9 μg/s would cause a concentration of 0.073 ng/l. This value is 13% of the median concentration 0.55 ng/l that was measured from the river at the WIP, which is in agreement with the 11% average calculated by the model. Perkola and Sainio (2013) have determined the average yearly PFOA loads of the WWTPs of Turku, Espoo and Helsinki to be 0.35; 0.43 and 0.99 kg/year respectively. These are the three largest WWTPs in Finland, and the WWTP in Helsinki is the largest in the Nordic countries. By the mass balance calculation described above, even these loads together still produced only 62% of the PFOA mass flow arriving at the WIP. Further considering that the Tampere area is by far the largest population center in the catchment and that both of its WWTPs are involved in the calculations, there must be other major sources apart from CWWTPs that discharge PFOA to surface waters in the area. Several researchers have studied PFAA sources by calculating the correlation between PFAA concentrations and population. Müller et al. (2011) studied the correlations of PFAA concentrations with river catchment populations and acesulfame concentrations in Switzerland. They found that both PFOA and PFOS correlated well with population

Fig. 4. The modeled PFOA concentration at the raw water intake plant (WIP), when a continuous PFOA load of 57 kg/year was released from Äetsä 7 km away or Nokianvirta 63 km away from the WIP. 57 kg/year is the magnitude that the PFOA load should rise to for the PFOA guideline value 40 ng/l to be exceeded at the WIP.

80

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

Fig. 5. Modeled distribution of the PFOA load at the raw water intake plant between different sources. The sources include the wastewater treatment plants and the model boundaries in Nokianvirta and Siuronkoski. The light blue and red area form together the load incoming from Nokianvirta.

(R2 = 0.71 and 0.60, respectively) and population marker acesulfame but less with catchment area. From these results the authors deduced that consumer products were the most important source of PFOA and PFOS to surface waters in Switzerland, while atmospheric deposition was a minor source. However Pistocchi and Loos (2009), who studied PFOA and PFOS concentration correlations with population in Europe, found that while PFOS correlated well with population, PFOA concentrations were strongly influenced by industrial point emissions. Their data on PFOA and PFOS concentrations covered approximately 40% of the estimated European drainage area. Also Murakami et al. (2008) found in Japan that while PFOS, PFHpA and PFNA concentrations correlated strongly with population density, for PFOA the correlation was weak, and at some places PFOA was strongly influenced by industrial emissions. It is worth to note that PFOA concentrations in Japan were found to be high: in 11 out of 20 rivers above 40 ng/l. According to this study atmospheric deposition was not an important source of PFOA. In Tokyo, Japan, Takazawa et al. (2009) found that neither PFOA nor PFOS had a linear relationship with population. In addition they compared WWTP and river mass loads: They calculated the PFOS load per person for a CWWTP using wastewater concentration and the total PFOS load using the concentrations of river water. By comparing the load from wastewater per person to the total load from all sources per person they found the total load to be seven times higher than the wastewater load. A few studies have also been conducted on PFOA sources to the Baltic Sea, where also River Kokemäenjoki discharges (in Bothnian Sea). Ahrens et al. (2010a) report PFAA concentrations in the North Sea, Baltic Sea and Norwegian Sea to be highest near industrial and highly populated areas, and atmospheric deposition of PFAA to be negligible close to contaminated areas. However Filipovic et al. (2013) argue that atmospheric deposition is the most important source of PFAAs to the Baltic Sea, while CWWTPs only have a minor effect on the total PFAA loads. This conclusion was reached by calculating PFAA inputs through rivers entering the sea, atmospheric deposition, wastewater discharges (directly to the Baltic Sea and to the Oder River watershed) and inflow from the North Sea. Of the total PFOA input to the Baltic Sea, rivers accounted for 48–59%, atmospheric deposition 34–43% and coastal CWWTPs less than 2%. However, the estimates vary greatly between the basins: river runoff, atmospheric deposition and CWWTP input for PFOA were estimated to be 71%, 29% and 0%, respectively, for The Bothnian Sea, and 12%, 73% and 15%, respectively, for the Gulf of Finland, for example. Mass balance calculations for

the Oder River watershed suggested that CWWTPs caused 21% and atmospheric deposition the majority of the total PFOA input to the catchment. The reported values of the CWWTP contribution are in agreement with our 11% average simulation result. The results of Scott et al. (2010) and Castiglioni et al. (2015) are also in agreement with our 11% estimate: Scott et al. (2010) report the PFOA load from CWWTPs to Lake Superior in North America to be 6% of the total PFOA input. The major PFAA sources to the lake were found to be riverine discharges and precipitation. Castiglioni et al. (2015) studied six WWTPs in Milan, Italy, out of which three received around 60% industrial wastewater and another three only 1% industrial wastewater. The PFAA load from the former WWTPs was up to 50 times as large as the PFAA load from the latter WWTPs. The WWTPs receiving mainly municipal wastewater covered only 8% or the total PFAA load from the entire basin. Our results indicate that CWWTPs are not a major source of PFOA to surface waters in south-western Finland, which is supported by the literature cited above. Our results complement the understanding about the role of CWWTPs as sources of PFOA. The results are representative of high-latitude area with temperate climate and relatively sparse population. It is not clear where the rest of the PFOA load originates from, but it seems that depending on the location industrial releases can be an important source. The studies are contradictory regarding atmospheric deposition, but it can be an important source especially in the Baltic Sea area. Also the release of aqueous firefighting foams from e.g. fire training facilities is a potential point source of PFOA (Ahrens et al., 2015). We also calculated the PFOA load per capita for each of the CWWTPs in the study area and the CWWTPs of Turku, Espoo and Helsinki based on the values reported by Perkola and Sainio (2013) (Fig. 6). The PFOA load per capita of the smaller CWWTPs varied considerably (1.0–17 μg/d), while for the large plants of Turku, Espoo and Helsinki the load was relatively even (3.4–4.1 μg/d). Our study area total (the total load of all the study area CWWTPs divided by the total population) was 3.3 μg/d, so it corresponded with the values of the large plants. The differences between the smaller plants might be due to different amounts of industrial wastewater lead to the plants or possibly varying wastewater treatment efficiency. The results are in agreement with Castiglioni et al. (2015) who report that the PFOA load of WWTPs in Milan, Italy, receiving mainly municipal wastewaters is 3–6 μg/d per capita. Pistocchi and Loos (2009) have reported the European average PFOA load per capita to be 19.2 μg/d. The value is based on river PFOA

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

81

Fig. 6. PFOA load per capita and population for each communal wastewater treatment plant (CWWTP) and the study area total. The PFOA loads of Viinikanlahti, Rahola, Kullaanvuori, Siuro, Vammala, Mouhijärvi and Äetsä (in the study area) were determined in this study. The PFOA loads for the CWWTPs of Turku, Espoo and Helsinki, determined by Perkola and Sainio (2013), are displayed for comparison.

concentrations and river basin populations, so it represents the total PFOA load per capita, while our values represent only the load from CWWTPs. It thus makes sense for our values to be smaller, as the results of this study indicate that only a small part of the total PFOA load originates from CWWTPs. 4. Conclusions The prediction of the fate of PFAAs, a group of emerging pollutants, is important since they are widely used and persistent in the aquatic environment. As a water soluble substance PFOA can be transported long distances in surface water and groundwater, and it has been linked to several adverse health effects in humans. Thus it is important to be able to predict PFOA concentrations especially in waterways that are used as raw water sources for drinking water. Using a onedimensional advection-dispersion flow and water quality model, we described the dynamic time dependent transport of acesulfame and PFOA in a strongly regulated river stretch. The model was able to reproduce the measured concentrations in a satisfactory manner during different flow conditions. The model also proved to be a useful tool for assessing the importance of different PFOA sources and the overall risk that PFOA causes to the artificial groundwater recharge practiced in the area. The simulated PFOA concentration in river water at the raw water intake plant reached at its maximum 7.8% of the 40 ng/l guideline value for PFOA in drinking water. It is thus unlikely that PFOA concentration would reach a harmful level in drinking water with the current PFOA load to River Kokemäenjoki. The seven CWWTPs in the study area caused on average only 11% of the total PFOA load that reaches the WIP. This result was also supported by simple mass balance calculations. The minor role of CWWTPs indicates that there must be other major PFOA sources in the area. According to literature, the primary sources other than CWWTPs include industrial point sources and atmospheric deposition. The exact role of these and other possible sources in PFOA contamination remains to be solved in future studies. Acknowledgments We thank Maa-ja vesitekniikan tuki ry (grant no. 29913). that provided the primary funding for this study. Additional funding, the flow model and the water quality sampling results were provided by CONPAT (Aquatic contaminants — pathways, health risks and management). CONPAT is a research project of The Finnish Environment

Institute, The National Institute for Health and Welfare and VATT Institute for Economic Research, funded by The Academy of Finland. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2015.09.008. References Ahrens, L., Gerwinski, W., Theobald, N., Ebinghaus, R., 2010a. Sources of polyfluoroalkyl compounds in the North Sea, Baltic Sea and Norwegian Sea: evidence from their spatial distribution in surface water. Mar. Pollut. Bull. 60, 255–260. Ahrens, L., Norström, K., Viktor, T., Cousins, A.P., Josefsson, S., 2015. Stockholm Arlanda Airport as a source of per- and polyfluoroalkyl substances to water, sediment and fish. Chemosphere 129, 33–38. Ahrens, L., Taniyasu, S., Yeung, L.W.Y., Yamashita, N., Lam, P.K.S., Ebinghaus, R., 2010b. Distribution of polyfluoroalkyl compounds in water, suspended particulate matter and sediment from Tokyo Bay, Japan. Chemosphere 79, 266–272. Ahrens, L., Yeung, L.W.Y., Taniyasu, S., Lam, P.K.S., Yamashita, N., 2011. Partitioning of perfluorooctanoate (PFOA), perfluorooctane sulfonate (PFOS) and perfluorooctane sulfonamide (PFOSA) between water and sediment. Chemosphere 85, 731–737. Assmuth, T., Simola, A., Pitkänen, T., Lyytimäki, J., Huttula, T., 2015. Integrated frameworks for assessing and managing health risks in the context of managed aquifer recharge with river water. Integr. Environ. Assess. Manag. (Accepted manuscript). Barry, V., Winquist, A., Steenland, K., 2013. Perfluorooctanoic acid (PFOA) exposures and incident cancers among adults living near a chemical plant. Environ. Health Perspect. 121, 1313–1318. Becker, A.M., Gerstmann, S., Frank, H., 2008. Perfluorooctane surfactants in waste waters, the major source of river pollution. Chemosphere 72, 115–121. Buerge, I.J., Buser, H.R., Kahle, M., Müller, M.D., Poiger, T., 2009. Ubiquitous occurrence of the artificial sweetener acesulfame in the aquatic environment: an ideal chemical marker of domestic wastewater in groundwater. Environ. Sci. Technol. 43, 4381–4385. Butt, C.M., Berger, U., Bossi, R., Tomy, G.T., 2010. Levels and trends of poly- and perfluorinated compounds in the Arctic environment. Sci. Total Environ. 408, 2936–2965. Castiglioni, S., Valsecchi, S., Polesello, S., Rusconi, M., Melis, M., Palmiotto, M., et al., 2015. Sources and fate of perfluorinated compounds in the aqueous environment and in drinking water of a highly urbanized and industrialized area in Italy. J. Hazard. Mater. 282, 51–60. Cornelis, C., D'Hollander, W., Roosens, L., Covaci, A., Smolders, R., Van Den Heuvel, R., et al., 2012. First assessment of population exposure to perfluorinated compounds in Flanders, Belgium. Chemosphere 86, 308–314. Davis, K.L., Aucoin, M.D., Larsen, B.S., Kaiser, M.A., Hartten, A.S., 2007. Transport of ammonium perfluorooctanoate in environmental media near a fluoropolymer manufacturing facility. Chemosphere 67, 2011–2019. Deltares, 2014. SOBEK Suite. https://www.deltares.nl/en/software/sobek/#2 (Accessed Aug. 12, 2015). Earnshaw, M.R., Paul, A.G., Loos, R., Tavazzi, S., Paracchini, B., Scheringer, M., et al., 2014. Comparing measured and modelled PFOS concentrations in a UK freshwater catchment and estimating emission rates. Environ. Int. 70, 25–31. European Chemicals Agency, 2014. Annex XV restriction report, proposal for a restriction. http://echa.europa.eu/documents/10162/e9cddee6-3164-473d-b590-8fcf9caa50e7.

82

M. Happonen et al. / Science of the Total Environment 541 (2016) 74–82

Filipovic, M., Berger, U., McLachlan, M.S., 2013. Mass balance of perfluoroalkyl acids in the Baltic Sea. Environ. Sci. Technol. 47, 4088–4095. Finnish Environment Institute, 2010. Whole Effluent Assessment. Proposed Recommendations for the Use of Toxicity Limits. http://www.cohiba-project.net/publications. Frisbee, S.J., Shankar, A., Knox, S.S., Steenland, K., Savitz, D.A., Fletcher, T., et al., 2010. Perfluorooctanoic acid, perfluorooctanesulfonate, and serum lipids in children and adolescents: results from the C8 health project. Arch. Pediatr. Adolesc. Med. 164, 860–869. German Drinking Water Commission, 2006. Provisional Evaluation of PFT in Drinking Water With the Guide Substances Perfluorooctanoic Acid (PFOA) and Perfluorooctane Sulfonate (PFOS) as Examples. http://prevenzione.ulss20.verona.it/ docs/Sian/IgieneNutrizione/Acque/Pfas/pft-in-drinking-water.pdf. Happonen, M., 2015. Assessing PFOA Contamination Risk in Artificial Recharge of Groundwater Using a Dynamic River Model (Master's thesis) Aalto University School of Engineeringhttps://aaltodoc.aalto.fi/handle/123456789/15562?locale-attribute=en. Harada, K., Saito, N., Sasaki, K., Inoue, K., Koizumi, A., 2003. Perfluorooctane sulfonate contamination of drinking water in the Tama River, Japan: estimated effects on resident serum levels. Bull. Environ. Contam. Toxicol. 71, 31–36. Harwood, J.J., 2014. Molecular markers for identifying municipal, domestic and agricultural sources of organic matter in natural waters. Chemosphere 95, 3–8. Kissa, E., 2001. Fluorinated Surfactants and Repellants. 2nd ed. (New York). Lam, J., Koustas, E., Sutton, P., Johnson, P.I., Atchley, D.S., Sen, S., et al., 2014. The navigation guide—evidence-based medicine meets environmental health: Integration of animal and human evidence for PFOA effects on fetal growth. Environ. Health Perspect. 122, 1040–1051. Lau, C., Anitole, K., Hodes, C., Lai, D., Pfahles-Hutchens, A., Seed, J., 2007. Perfluoroalkyl acids: a review of monitoring and toxicological findings. Toxicol. Sci. 99, 366–394. Liou, J.S.C., Szostek, B., DeRito, C.M., Madsen, E.L., 2010. Investigating the biodegradability of perfluorooctanoic acid. Chemosphere 80, 176–183. Loos, R., Carvalho, R., António, D.C., Comero, S., Locoro, G., Tavazzi, S., et al., 2013. EU-wide monitoring survey on emerging polar organic contaminants in wastewater treatment plant effluents. Water Res. 47, 6475–6487. Lyytimäki, J., Assmuth, T., 2014. Down with the flow: public debates shaping the risk framing of artificial groundwater recharge. GeoJournal 80, 113–127. Melzer, D., Rice, N., Depledge, M.H., Henley, W.E., Galloway, T.S., 2010. Association between serum perfluorooctanoic acid (PFOA) and thyroid disease in the U.S. National Health and Nutrition Examination Survey. Environ. Health Perspect. 118, 686–692. Miyake, Y., Kobayashi, T., Kameya, T., Managaki, S., Amagai, T., Masunaga, S., 2014. Comparison study on observed and estimated concentrations of perfluorooctane sulfonate using a fate model in Tokyo Bay of Japan. J. Environ. Sci. Health A Tox. Hazard. Subst. Environ. Eng. 49, 770–776. Müller, C.E., Gerecke, A.C., Alder, A.C., Scheringer, M., Hungerbühler, K., 2011. Identification of perfluoroalkyl acid sources in Swiss surface waters with the help of the artificial sweetener acesulfame. Environ. Pollut. 159, 1419–1426. Murakami, M., Imamura, E., Shinohara, H., Kiri, K., Muramatsu, Y., Harada, A., et al., 2008. Occurrence and sources of perfluorinated surfactants in rivers in Japan. Environ. Sci. Technol. 42, 6566–6572. Murakami, M., Shinohara, H., Takada, H., 2009. Evaluation of wastewater and street runoff as sources of perfluorinated surfactants (PFSs). Chemosphere 74, 487–493. NJDEP, 2007. Guidance for PFOA in Drinking Water at Pennsgrove Water Supply Company. http://www.nj.gov/dep/watersupply/pdf/pfoa_dwguidance.pdf. Paul, A.G., Scheringer, M., Hungerbuhler, K., Loos, R., Jones, K.C., Sweetman, A.J., 2012. Estimating the aquatic emissions and fate of perfluorooctane sulfonate (PFOS) into the river Rhine. J. Environ. Monit. 14, 524–530. Perkola, N., Juntunen, J., Tuominen, S., Nysten, T., Rosendahl, K., Huttula, T., et al., 2015. Kuluttajakemikaalit ja mikrobit Kokemäenjoen vesistössä (Consumer chemicals and microbes in Kokemäenjoki watercourse). Ympäristö & Terveys 3, 50–55.

Perkola, N., 2014. Fate of artificial sweeteners and perfluoroalkyl acids in aquatic environment. Reports from the Department of Environmental Sciences. Faculty of Biological and Environmental Sciences. University of Helsinki, p. 23. https://helda.helsinki.fi/ handle/10138/136494?locale-attribute=en. Perkola, N., Sainio, P., 2013. Survey of perfluorinated alkyl acids in Finnish effluents, storm water, landfill leachate and sludge. Environ. Sci. Pollut. Res. 20, 7979–7987. Pistocchi, A., Loos, R., 2009. A Map of European Emissions and Concentrations of PFOS and PFOA. Environ. Sci. Technol. 43, 9237–9244. Post, G.B., Cohn, P.D., Cooper, K.R., 2012. Perfluorooctanoic acid (PFOA), an emerging drinking water contaminant: a critical review of recent literature. Environ. Res. 116, 93–117. Prevedouros, K., Cousins, I.T., Buck, R.C., Korzeniowski, S.H., 2006. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 40, 32–44. Regione del Veneto, 2014. Deliberazione della Giunta Regionale. http://bur.regione. veneto.it/BurvServices/Pubblica/DettaglioDgr.aspx?id=273825 (Accessed Aug. 11, 2015). Renner, R., 2001. Growing concern over perfluorinated chemicals. Environ. Sci. Technol. 35, 154A–160A. Schultz, M.M., Higgins, C.P., Huset, C.A., Luthy, R.G., Barofsky, D.F., Field, J.A., 2006. Fluorochemical mass flows in a municipal wastewater treatment facility. Environ. Sci. Technol. 40, 7350–7357. Schultz, M., Barofsky, D., Field, J., 2008. Quantitative determination of fluorinated alkyl substances by large-volume-injection LC/MS/MS—characterization of municipal wastewaters. Environ. Sci. Technol. 40, 289–295. Scott, B.F., De Silva, A.O., Spencer, C., Lopez, E., Backus, S.M., Muir, D.C.G., 2010. Perfluoroalkyl acids in Lake Superior water: trends and sources. J. Great Lakes Res. 36, 277–284. Steenland, K., Tinker, S., Shankar, A., Ducatman, A., 2010. Association of perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) with uric acid among adults with elevated community exposure to PFOA. Environ. Health Perspect. 118, 229–233. Takazawa, Y., Nishino, T., Sasaki, Y., Yamashita, H., Suzuki, N., Tanabe, K., et al., 2009. Occurrence and distribution of perfluorooctane sulfonate and perfluorooctanoic acid in the rivers of Tokyo. Water Air Soil Pollut. 202, 57–67. Tran, N.H., Hu, J., Li, J., Ong, S.L., 2014. Suitability of artificial sweeteners as indicators of raw wastewater contamination in surface water and groundwater. Water Res. 48, 443–456. UK Drinking Water Inspectorate, 2009. Guidance on the Water Supply (Water Quality) Regulations 2000 Specific to PFOS (Perfluorooctane Sulphonate) and PFOA (Perfluorooctanoic Acid) Concentrations in Drinking Water. http://dwi.defra.gov.uk/ stakeholders/information-letters/2009/10_2009annex.pdf. USEPA, 2006. PFOA Stewardship Program Description. http://www.epa.gov/oppt/pfoa/ pubs/stewardship/pfoastewardshipbasics.html (Accessed Aug. 11, 2015). USEPA, 2009. The toxicity of perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS). http://www.epa.gov/opptintr/pfoa/pubs/Final%20PFOA%20PFOS%20RfD% 20memo%2010-28-09.pdf. Vaalgamaa, S., Vähätalo, A.V., Perkola, N., Huhtala, S., 2011. Photochemical reactivity of perfluorooctanoic acid (PFOA) in conditions representing surface water. Sci. Total Environ. 409, 3043–3048. Yu, J., Hu, J., Tanaka, S., Fujii, S., 2009. Perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) in sewage treatment plants. Water Res. 43, 2399–2408. Zareitalabad, P., Siemens, J., Hamer, M., Amelung, W., 2013. Perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) in surface waters, sediments, soils and wastewater — a review on concentrations and distribution coefficients. Chemosphere 91, 725–732. Zushi, Y., Hogarh, J.N., Masunaga, S., 2012. Progress and perspective of perfluorinated compound risk assessment and management in various countries and institutes. Clean Techn. Environ. Policy 14, 9–20.