Purpose and Scope

This report describes the occurrence and distribution of inorganic analytes and PFAS in groundwater and surface water used as sources for public-water supply in West Virginia. Samples of raw, untreated water were collected at 279 public-water systems in the State from June 2019 through May 2021. This study was conducted by the U.S. Geological Survey (USGS) in cooperation with the WVDEP and the WVDHHR for the purpose of meeting the requirements of West Virginia Senate Concurrent Resolution 46 of the 2020 regular legislative session (SCR 46; West Virginia Legislature, 2020). This resolution required the WVDEP and the WVDHHR to cooperatively propose and initiate a public source-water supply study to sample PFAS for all community water systems in West Virginia.

Specific objectives of this study included the following: (1) identify the drinking-water supplies in West Virginia that have select PFAS compounds in raw source water; (2) determine if there are geochemical, watershed, industrial-use, land-use, or geohydrologic factors that affect the presence of these compounds; (3) inform State agencies and the public of any need for additional PFAS investigation; and (4) assist State regulatory agencies in protecting public health by providing information on statewide PFAS distribution in source water.

Description of Per- and Polyfluoroalkyl Substances

Per- and polyfluoroalkyl substances are widely observed anthropogenic compounds found in water supplies worldwide and increasingly linked with adverse health effects in humans (Hu and others, 2016; Sunderland and others, 2019). These substances are highly recalcitrant and water soluble, which has enabled extensive environmental permeation and organismal exposure underlined by documented PFAS occurrences in remote polar regions (Giesy and Kannan, 2001; Young and others, 2007; Routti and others, 2015; Lin and others, 2020) and occupationally diverse human populations (Calafat and others, 2007; Vestergren and Cousins, 2009). It is presumed nearly all individuals possess detectable levels of PFAS (Kato and others, 2011; CDC, 2015), which is an artifact of both the omnipresence of PFAS contamination in the environment and widespread use in many industrial applications and consumer products (Kotthoff and others, 2015; Glüge and others, 2020). Unique physio-chemical properties, such as resistance to degradation and amphiphilicity, stem from the chemically inert and stable molecular structure inherent to all PFAS. These attributes often provide unmatched performance in targeted applications, but this intrinsic persistence has also propelled these ‘forever chemicals’ to the forefront of contaminants of emerging concern (CEC; U.S. Environmental Protection Agency, 2017). Furthermore, the EPA CEC list includes only two PFAS variants, PFOS and PFOA, among the thousands populating the EPA PFAS Master List of PFAS substances (https://comptox.epa.gov/dashboard/chemical-lists/PFASMASTER) and currently in use globally (KEMI, 2015). This structural diversity presents a multitude of issues related to the analysis, regulation, and publication of PFAS, further inhibiting comprehension of this rapidly evolving multidisciplinary field. The subsequent introductory sections address these intrinsic challenges by establishing a clear, albeit generalized, understanding of PFAS through a brief review of PFAS nomenclature, sources, and human-exposure pathways.

PFAS Nomenclature and Heterogeneity

The PFAS family consists of thousands of compounds comprising polymers and non-polymers of variable reactivity, solubility, and electric charge (for example, anionic) that span multiple states of matter (gas, liquid, solid; fig. 1A; Cousins and others, 2020). The presence of a perfluoroalkyl moiety (C_nF_2n+1-) is the unifying feature of all PFAS, which are generally defined as aliphatic substances where at least one, but typically more, carbon atoms are perfluorinated (in other words, fully fluorinated; fig. 1B). This yields two possible subclasses for nonpolymer PFAS, perfluoroalkyl and polyfluoroalkyl substances. Along with nonpolymer PFAS, the other primary class is polymer PFAS, encompassing the fluoropolymer, perfluoropolyether, and side-chain fluorinated polymer subclasses (fig. 1A). While nonpolymer PFAS may be used as monomers or as processing aids for polymer PFAS production, polymer PFAS are significantly less prevalent at contaminated sites and are considered to pose a lesser threat to the environment and human health in the near future (Buck and others, 2011; Henry and others, 2018). Therefore, further discussion focuses on nonpolymer PFAS.

Most PFAS observed in environmental and biological samples are nonpolymer PFAS, which comprises perfluoroalkyl and polyfluoroalkyl substances. Both per- and polyfluoroalkyl molecules are composed of a ‘head’ in the form of a functional group, typically a carboxylic or sulfonic acid, and a linear or branched carbon ‘tail’ (fig. 1B). Perfluoroalkyl signifies the alkane tail is fully fluorinated, whereas any PFAS not fully fluorinated but possessing at least one perfluoroalkyl moiety is classified as polyfluorinated. Polyfluoroalkyl substances may degrade to perfluoroalkyl substances, however, as the nonfluorinated C–R bond is relatively weaker than the C–F bond (Buck and others, 2011).

Perfluoroalkyl acids (PFAAs) are perhaps the most extensively studied group of PFAS compounds. Investigations of PFAAs in the environment indicate these compounds are both persistent and widespread in water, soils, and tissues. These characteristics coupled with health effects have been the basis for PFAS regulation. Two PFAAs, PFOA and PFOS, are identified as CECs by the EPA (U.S. Environmental Protection Agency, 2017; fig. 1A). The virtually unbiodegradable nature of PFAAs has led to the term ‘terminal PFAS,’ with PFAAs representing the endpoint of PFAS degradation pathways for some precursor molecules. Potential precursors include other perfluoroalkyl substances, such as perfluoroalkane sulfonamides (FASAs), as well as polyfluoroalkyl substances that can degrade via abiotic and biotic pathways to form PFAAs (Liu and Avendaño, 2013; Wang and others, 2015). Perfluoroalkyl carboxylic acids (PFCAs) and sulfonic acids (PFSAs) are the two major PFAA subgroups (fig. 1A). However, these acronyms are used for both the acid and anionic forms of PFCAs and PFSAs (for example, PFOA = perfluorooctanoic acid and perfluorooctanoate). Proper reference of PFAA form is critical to interpreting laboratory data, although in most cases this is not an issue, given the proclivity of PFAAs to exist exclusively as anions under normal environmental conditions.

Deciphering PFAA acronyms is relatively straightforward. The first two letters, ‘PF’, are constant and represent the ‘PerFluorinated’ nature of all PFAAs. The third letter pertains to the carbon number, based on conventional alkane nomenclature, and the fourth letter is related to the functional group (A = carboxylic acid or carboxylate, S = sulfonic acid or sulfonate). Thus, PerFluoroOctaneSulfonic acid is abbreviated PFOS because it is a PerFluorinated PFAA with an eight-carbon tail (in other words, octane) and a sulfonic acid head. PFAAs may also be referred to by chain length, with PFCAs bearing at least 7 perfluorinated carbons and PFSAs containing at least 6 perfluorinated carbons considered long-chain PFAAs (Organization for Economic Cooperation and Development, 2013). These classifications are based on the bioaccumulation potential of PFAAs and not carbon number (fig. 1C). While chain length is not a perfect predictor of environmental behavior of PFAA, or any other PFAS, (Ng and Hungerbühler, 2014), it permits a generalized characterization of how a particular PFAS behaves in the environment, partitions among various media, and accumulates in organisms. Ultrashort- and short-chain PFAS are more mobile due to greater aqueous solubility and volatility relative to long-chain PFAS, which tend to exhibit greater adsorption and bioaccumulation potential. As a result, long-chain PFAS are commonly perceived as more toxic than ultrashort- and short-chain PFAS (fig. 1C). However, these short-chain homologues, now widely used as substitutes for long-chain PFAS, possess comparable persistence in the environment and may pose an even greater challenge to remediation efforts due to lower adsorption potential which reduces the effectiveness of granular activated carbon for removal (Brendel and others, 2018; Ateia and others, 2019). Polyfluorinated fluorotelomer substances, precursor molecules to PFCAs, also follow an intuitive naming convention, whereby a ratio of perfluorinated carbons to non-perfluorinated carbons precedes the acronym. The letters ‘FT’ represent FluoroTelomer and the following letter(s) of the acronym correspond to the functional group (in other words, OH = alcohol; CA = carboxylic acid; SA = sulfonic acid). For example, 6:2 FTSA signifies a FluoroTelomer Sulfonic Acid with 6 perfluorinated carbons and 2 non-perfluorinated carbons. The anionic form, 6:2 fluorotelomersulfonate, is abbreviated 6:2 FTS (Buck and others, 2011). The preceding review on PFAS nomenclature provides an adequate foundation to discern the broader implications of this work. Additional information on the diverse PFAS family and pertinent nomenclature may be found in several comprehensive reviews (for example, Buck and others, 2011; Wang and others, 2017; Kwiatkowski and others, 2020).

PFAS Sources

Amphiphilicity and stability are the two principal attributes leading to the widespread use of PFAS in manufacturing processes since commencement of commercialization in the 1950s (Kissa, 2001). The molecular structure of PFAS (fig. 1B), specifically the polar hydrophilic head (in other words, functional group) and nonpolar hydrophobic and oleophobic tail (in other words, per- or polyfluorinated alkane), enables these amphiphiles to impart excellent repellent and surfactant traits to materials fabricated from, or with, PFAS. Furthermore, the per- and polyfluorinated carbon backbones, comprising the strongest bonds in organic chemistry (C–F; O’Hagan, 2008), grant exceptional resistance to biological, chemical, and thermal degradation (Liu and Avendaño, 2013; Rahman and others, 2014). Usage of PFAS is seemingly unbound, found in nearly all industries (for example, automotive, electronics, construction, agriculture), many consumer products (for example, textiles, cosmetics, food packaging), and notably form an essential component of aqueous film forming foams (AFFF) used in fire-fighting applications (fig. 2; Prevedouros and others, 2006; Glüge and others, 2020 and references therein). Consequently, widespread PFAS use has led to a pervasive environmental footprint.

PFAS are ideal environmental contaminants. The same qualities desirable for industrial and commercial use, such as thermal and chemical stability, combined with moderate-to-high water solubility, have simultaneously enabled PFAS to effectively permeate and accumulate across all Earth systems on a global scale (in other words, hydrosphere, geosphere, atmosphere, and biosphere). PFAS infiltration of the hydrosphere is particularly concerning, as it has been found in public and treated water supplies (Hu and others, 2016; Boone and others, 2019), as well as in the oceans (Yamashita and others, 2005). Major sources of PFAS contamination in the environment include wastewater treatment plants (WWTPs), biosolids, landfills, industrial manufacturing sites, military bases, and airports (fig. 2; Busch and others, 2010; Lindstrom and others, 2011; Filipovic and others, 2015; Hu and others, 2016). Millions of people are affected by PFAS contamination and more will be impacted unless technological advancements are made in PFAS removal and replacement chemistry (Hu and others, 2016; Andrews and Naidenko, 2020). Conventional wastewater- and drinking-water treatment methods remain largely ineffective at PFAS and PFAS-precursor removal (Hamid and Li, 2016; Boone and others, 2019), which is exacerbated by the improper disposal of ever-increasing electronic waste, among other PFAS-bearing consumer products, in landfills (Garg and others, 2020). This leads to elevated PFAS in landfill leachates (Lang and others, 2017) that may adversely impact proximal groundwater resources, and further strains WWTPs tasked with treating these leachates alongside industrial and household wastewater containing variable levels of PFAS (Guo and others, 2020). In some cases, treatment may produce effluents with higher PFAA concentrations than influents due to precursor degradation (Schultz and others, 2006), with agricultural application of tainted biosolids a concern (Washington and others, 2010; Ghisi and others, 2019). Unlike other examples, emission and atmospheric dispersion of volatile and non-volatile PFAS from manufacturing facilities, as well as other point sources, represents a potentially prominent PFAS sourcing mechanism capable of impacting both local and global environments (Barber and others, 2007; Young and Mabury, 2010; Galloway and others, 2020). Despite legislation restricting the use of PFAS in certain applications (for example, food packaging), significant sources remain, such as the ongoing use of AFFF at military bases and airports (fig. 2; Anderson and others, 2016; Houtz and others, 2016; Cousins and others, 2019) that will sustain [and likely increase] environmental PFAS concentrations for the foreseeable future and perpetuate the attendant risk to public health.

Description of Study Area

The study area is the State of West Virginia and sample sites are groundwater (wells and springs) and surface-water (reservoirs and rivers) sources used as public supplies (fig. 3). Most of West Virginia is within the Appalachian Plateaus [19,960 square miles (mi²)] and Valley and Ridge (4,220 mi²) Physiographic Provinces, although a small area (20 mi²) at the easternmost tip of West Virginia is within the Blue Ridge Physiographic Province. The Appalachian Plateaus Physiographic Province is an area of flat-lying or gently folded rocks that formed when a peneplain was uplifted during the Appalachian Orogeny and then was dissected by stream erosion during the ensuing 300 million years (Fenneman, 1938). Elevation is highest in the east, where some peaks are higher than 4,000 feet (ft), and lowest in the west near the Ohio River, where the valley is lower than 600 ft (fig. 4). Relief is generally greatest at highest elevations. Most of the Appalachian Plateaus Physiographic Province within West Virginia is drained by the Ohio River (19,631 mi²), and the rest (329 mi²) is drained by the Potomac River.

In West Virginia, the Valley and Ridge Physiographic Province consists of layers of folded and faulted sedimentary rocks that are of Mississippian age or older (Cardwell and others, 1968). Rocks of the Valley and Ridge province were folded in the Appalachian Orogeny, the same event that uplifted the Appalachian Plateaus. Linear ridges that run from southwest to northeast alternate with valleys. Ridges are generally underlain by harder, more erosion-resistant rock than the rock underlying the valleys. Streams in this province drain in a trellised pattern. Generally, lithology in the Valley and Ridge is more complex than in the Appalachian Plateaus. Some of the valleys, most significantly near the eastern edge of the province, are underlain by karst developed in limestone and dolomite rocks (West Virginia Geological and Economic Survey, 2022).

Land use, surface geology, and soils all vary with elevation within the study area (Messinger and Hughes, 2000). Generally, the highest population density and concentration of urban land is in river valleys in the northern and western parts and in the eastern tip of West Virginia. Commercial agriculture is limited in scope, and most commercial farms are near the Ohio and Kanawha Rivers and throughout the Potomac and Greenbrier River Basins (fig. 4). Forest cover is most dense in the mountains in central West Virginia. The largest cities in the study area, all with populations less than 50,000, are Charleston, on the Kanawha River; Huntington, Parkersburg, and Wheeling, on the Ohio River; and Morgantown, on the Monongahela River (U.S. Census Bureau, 2021).

Principal economic activities include coal mining, forestry, and oil and gas production, which are widespread throughout most of the study area; manufacturing, which is most common in areas near the Monongahela, Ohio, and Kanawha Rivers; and agriculture, which is most important commercially near the Ohio and Kanawha Rivers and in the Greenbrier River Basin (Messinger and Hughes, 2000). Of these, forestry is nearly ubiquitous throughout West Virginia, the exception being urban areas.

Aquifers discussed in this report include four basic hydrogeologic terranes (fig. 4), each with their own intrinsic susceptibility to contamination, as described by Kozar and Paybins (2016). A hydrogeologic terrane is a lithologic type of rock or rocks with distinct similar hydraulic properties and groundwater-flow processes. The aquifers consist of (1) alluvial aquifers bordering the Ohio River; (2) karst aquifers in the Valley and Ridge Physiographic Province and the Greenbrier River Valley in southern West Virginia; (3) abandoned underground coal-mine aquifers in southern West Virginia; and (4) fractured-rock aquifers in the Appalachian Plateaus Physiographic Province and Valley and Ridge Physiographic Province.

Groundwater age has been identified as a leading indicator of PFAS contamination in groundwater in the eastern U.S. with aquifers containing recharge water less than 60-years old having higher rates of PFAS occurrence (McMahon and others, 2022). Age-tracer data from previous studies on West Virginia’s groundwater resources showed that all aquifers of West Virginia contain recharge water less than 60-years old (McCoy and Kozar, 2007; Plummer and others, 2013). This indicates that all of the groundwater aquifers in the State are potentially susceptible to PFAS contamination (Kozar and Paybins, 2016) if a source of contamination exists within the recharge area of the aquifer.

Previous Investigations of Water Quality and PFAS in West Virginia

The quality of surface and groundwater in West Virginia has not been extensively investigated. Surface-water quality is routinely monitored throughout the State, but monitoring is more extensive for regulated contaminants than for emerging contaminants (West Virginia Department of Environmental Protection, 2016). Groundwater quality is typically monitored through samples from individual wells or as part of areal assessments done for planning or response to local problems and has been infrequently reviewed at the State scale (Mathes and others, 1998; Chambers and others, 2012). According to current regulatory benchmarks, groundwater quality is generally acceptable for domestic and other uses, with most problems resulting from geology, fossil-fuel extraction and other industrial activities, agriculture, and improper waste disposal. Groundwater chemistry is strongly affected by topographic setting, complicating areal delineations of water quality based on aquifer composition and compounding the effects of anthropogenic activities (Ferrel, 1987). Elevated levels of radon-222, iron, manganese, and arsenic, relative to drinking-water benchmarks set by the EPA, represent the primary constituents adversely impacting water quality in West Virginia (Ferrel, 1987; Chambers and others, 2012; Law and others, 2017).

No statewide assessments of PFAS in West Virginia public-water supplies have been done. Several PFAS investigations in West Virginia center on a fluoropolymer manufacturing facility situated along the Ohio River in Parkersburg, West Virginia. (Barton and others, 2006; Shin and others, 2011; Galloway and others, 2020) The C8 Health Project, based in Parkersburg, yielded critical data linking PFAS exposure to certain health conditions in humans, while the greater C8 Science Panel has served as a platform to better characterize the behavior of PFAS in the environment (http://www.c8sciencepanel.org/c8health.html). Extensive PFAS contamination of the surrounding alluvial and fractured rock aquifers has granted insight into the fate and transport of PFAS (Barton and others, 2006; Shin and others, 2011; Galloway and others, 2020), providing a natural analog to aid remediation efforts at other contaminated sites. A recent PFAS exposure assessment conducted by the Agency for Toxic Substances and Disease Registry (2022) focused on Martinsburg in Berkeley County, West Virginia. This study assessed the exposure to the local population from contamination of a community water supply by AFFF that migrated through the groundwater from a local military facility. The report concluded that average blood levels of perfluorohexanesulfonate (PFHxS) of study participants were higher than national levels and elevated blood levels of PFHxS may be associated with past drinking-water contamination. While these studies have advanced understanding of PFAS contamination in the environment and at local scales, no insight into how PFAS are distributed in public-water supplies across West Virginia has been provided.

Data Collection and Analytical Methods

Public-water systems sampled for this study were identified by WVDHHR and included systems classified as community water systems and all daycares and schools that operate their own water systems. Community water systems are defined as a public-water system that pipes water for human consumption to at least fifteen service connections used by year-round residents or one that regularly serves at least twenty-five residents (West Virginia Legislature, 2022). Raw source water was sampled for both groundwater and surface-water sites at the first available tap in the public-water system, prior to any treatment. Samples were collected to represent chemical concentrations of constituents at the point-of-entry into the system and assess the baseline chemistry of the system at the time of sampling. Samples were collected at sites as they were available, and no attempt was made to target specific seasons or flow regimes. Data collected for this study do not assess variability in results over time, which may be important at some sites based on local conditions. Various water-treatment techniques were employed by many public-water systems, and the raw-water chemistry described in this report may not represent the water served to the public.

Statistical Analysis

Nonparametric techniques were used for computing descriptive statistics of water-quality data that were in some instances censored at multiple levels. Statistical analyses were conducted with the R statistical computing environment version 4.1.1 (R Core Team, 2022). Censored values are water-quality results that are reported as less than a specified laboratory value and for which an exact value was not determined. Spearman correlation analysis was used to examine relations between variables. The Kaplan–Meier (KM) and Maximum Likelihood Estimation (MLE) models were used for estimation of summary statistics for censored data following methods described by Helsel (2012). When more than 80 percent of a constituent was censored, the maximum concentration was reported, and the minimum concentration was reported as less than the lowest reporting level. When a sample had 50 to 80 percent censored values, the MLE model was used. The MLE model was used for this range of censoring because of its ability to make accurate estimates and handle multiple censoring levels in datasets with greater than 50 observations. When the data were censored at less than 50 percent, the KM model was used to estimate the summary statistics. The KM estimate was used to account for multiple censoring levels because it does not depend on the assumption of a distributional shape with data that are censored at greater than 50 percent. When no values were censored, nonparametric estimates were unnecessary and summary statistics were computed using standard methods.

Data Review and Validation

Where sufficient data were available, the ionic charge balance error (CBE) was calculated to evaluate the electroneutrality of the water sample, identify transcription errors during field activities, and identify laboratory analytical errors. The CBE is calculated by the following formula:

The CBE has a positive value when the sum of cations exceeds the sum of anions and a negative value when the sum of anions is greater than the sum of cations. Calculated CBEs are rarely zero and values up to 10 percent are typically considered acceptable, but CBEs may exceed this threshold for waters with low-ionic strength or in acidic waters where hydrogen ions are significant (Fritz, 1994). Electroneutrality and the definition of the CBE are further discussed by Freeze and Cherry (1979).

Sufficient data were available to calculate CBEs for 272 samples. Calculated CBEs generally were less than or equal to 10 percent. Exceptions consisted of three samples, NIC03401 collected at a surface-water site on August 8, 2020 (-42 percent); Intake on J BAILEY RESERVOIR 6909475 collected at a surface-water site on October 10, 2020 (-22 percent); and Mal-0406 collected at a groundwater site on April 14, 2021 (-20 percent). Upon review for transcription and analytical errors, low-ionic strength was identified as the probable reason for the high CBEs in both surface-water samples. Dilute samples typically have higher CBEs (Fritz, 1994), and these samples had the lowest values in the dataset for alkalinity, sodium, calcium, and magnesium. The groundwater site, Mal-0406, was also scrutinized for analytical and transcription errors that may lead to high CBEs, but none were found. Considering the low CBEs for the rest of the dataset, it was assumed that the data were of high quality and this sample was included in further analyses.

Data review and validation for five samples analyzed for PFAS by EPA method 537.1 followed guidance described by U.S. Environmental Protection Agency (2018a), which is intended to aid data reviewers in assessing the usability and extent of bias in laboratory results for data analyzed by EPA method 537. All other PFAS samples were analyzed according to the DoD QSM table B-15. Data validation for these samples followed U.S. Department of Defense (2020), which provides guidance for validating PFAS results generated by Liquid Chromatography/Tandem Mass Spectrometry compliant with DoD QSM table B-15 criteria in non-drinking water matrices.

Blank Samples

A combination of equipment blanks, field blanks, and laboratory-method blanks was used to identify and quantify potential sources of contamination. An equipment blank consists of a volume of water of known quality that is processed through the sampling equipment in a laboratory environment. A field blank consists of a volume of water processed through the sampling equipment under the same field conditions in which the samples were processed. A laboratory-method blank consists of a volume of water of known quality that is processed by the analytical laboratory. The qualification process of PFAS samples associated with blanks used for this study are shown in table 1 and followed recommendations by U.S. Environmental Protection Agency (2018a) and U.S. Department of Defense (2020).

Five equipment blanks, 16 field blanks, and laboratory-method blanks batches were evaluated for possible bias caused by PFAS contamination. Blank samples were evaluated only for results detected above the reporting level. A review of estimated results for PFAS detected below the reporting level is discussed in appendix 1. No PFAS was detected in equipment blanks at values above the reporting level, but 6:2 FTS was detected above the reporting level in 2 field blanks. The environmental sample for Min-0177 had a detection of 16 ng/L and the associated field blank was detected above the reporting level at 8.5 ng/L. The environmental sample for Intake on North Branch Potomac River 14364986 had an estimated value of 3.6 ng/L and the associated field blank was detected above the reporting level at 57 ng/L. These two samples had concentrations less than 10 times the concentration of the associated field blank, so the reported concentrations for 6:2 FTS in those samples were qualified with the remark code “<” in accordance with the recommended guidance in U.S. Department of Defense (2020).

6:2 FTS was also found above the reporting level in four method blank batches analyzed on February 26, 2021; March 9, 2021; March 18, 2021; and March 25, 2021, with results of 6.8, 4.3, 13, and 6.1 ng/L, respectively. During that time 15 associated environmental samples also had detections for 6:2 FTS above the reporting level. These 15 samples had concentrations less than 10 times the concentration of the associated laboratory blank, so the reported concentrations for 6:2 FTS in those samples were qualified with the remark code “<” in accordance with the recommended guidance in U.S. Department of Defense (2020).

Perfluorobutanoate (PFBA) was detected above the reporting level in two method blank batches, one analyzed on November 3, 2020, and the other on November 9, 2020, with results of 5.9 and 12 ng/L, respectively. During that time there were three associated environmental samples that also had detections for PFBA above the reporting level: Mal-0106 (10 ng/L), Brk-0047 (6.3 ng/L), and Grb-0189 (14 ng/L). These three samples had concentrations less than 10 times the concentration of the associated laboratory blank, so the reported concentrations for PFBA in those samples were qualified with the remark code “<” in accordance with the recommended guidance in U.S. Department of Defense (2020).

Replicate Samples

Variability for a replicate sample pair was quantified by calculating the relative percent difference (RPD) of the samples, using the following formula:

Generally, concentrations in replicate sample pairs differed by small amounts, typically less than 15 percent RPD for constituents in filtered samples and less than 20 percent RPD in unfiltered samples.

Replicate samples were collected at nine sites for inorganic analytes, as described in McAdoo and others (2022), and RPDs generally showed good agreement between replicate pairs with most analyses within 5 percent. However, cobalt, copper, lead, and orthophosphate had RPDs greater than 15 percent in some replicate pairs. Specifically, cobalt results for Intake on Twelve Pole Creek 3964858 and Mrg-0070 had RPDs of 30 and 23 percent, respectively. Absolute difference for cobalt in these replicate pairs was low, with 0.084 and 0.114 micrograms per liter (µg/L) for Intake on Twelve Pole Creek 3964858, and 0.100 and 0.079 µg/L for Mrg-0070, at a reporting level of 0.03 µg/L. Copper results for Ber-0856 and Ber-0468 had RPDs for replicate pairs of 29 and 88 percent, respectively. Absolute difference for copper in these replicate pairs was low, with 5.2 and 7.0 µg/L for Ber-0856, and 1.2 and 3.1 µg/L for Ber-0468, at a reporting level of 0.4 µg/L. The replicate pair collected at Wet-0111 had an RPD for lead of 18 percent and the replicate pair collected at Mcd-0108 had an RPD of 18 percent for orthophosphate. As with the previously described RPDs greater than 15 percent, these sample pairs had low absolute difference between sample results and were close to the reporting level.

From 12 replicate pairs sampled for PFAS, 20 RPD values were calculated with 2 values having greater than 20 percent RPD. Results for Ber-0468 showed a 24 percent RPD between replicate pairs for PFOA. The environmental sample had a concentration of 11 ng/L and the replicate had a result of 14 ng/L, with reporting levels of 5 and 5.3 ng/L, respectively. Results for Ber-0856 showed a 50 percent RPD between replicate pairs for PFOS. The environmental sample had a concentration of 5.5 ng/L, and the replicate had a result of 3.3 ng/L, which were close to the reporting levels of 5.3 and 5.2 ng/L, respectively. There were 13 instances in which detections were reported between the reporting level and minimum detection level for either the replicate or the environmental sample but not both. An RPD value could not be calculated for several replicate pairs with low-level estimated detections for PFAS. Estimated values between the reporting level and minimum detection level are further discussed in appendix 1.

Water Quality in Relation to Drinking-Water Standards

The sites sampled for this study represent raw-water supplies for every community public-water system in West Virginia, including all daycares and schools regulated by the WVDHHR. Many of these sites have additional treatment after the point sampled, so the results outlined here represent source water that is not necessarily representative of supplied drinking water. Nevertheless, these data were compared to human-health benchmarks established by the U.S. Environmental Protection Agency (2018b) to understand source water for public-water systems relative to drinking-water standards such as maximum contaminant levels (MCLs). Other non-regulatory EPA drinking-water guidelines used to assess this dataset include HA and secondary maximum contaminant levels (SMCLs). Health advisories are non-enforceable and non-regulatory but provide technical information to State agencies and other public health officials on health effects and are listed by the EPA for selected constituents that have no MCL or, in some cases, in addition to the MCL. SMCLs are listed for selected constituents that pose no known health risk but may have adverse aesthetic effects, such as staining or undesirable taste or odor.

The quality of surface water and groundwater collected from sources used for public supply in West Virginia generally remained within regulatory and non-regulatory guidelines established by EPA for measured constituents. However, the data presented in McAdoo and others (2022) showed that some samples had concentrations for certain analytes that exceeded MCLs, HAs, or SMCLs, which may require additional treatment at affected public-water systems. It is important to emphasize that public-water systems follow regulatory requirements established to protect public health by EPA and WVDHHR. Sites that have been identified by this study as exceeding benchmarks may have already installed mitigating treatment.

PFAS Occurrence in West Virginia’s Source Water

During the study period, analytical methods for detection of PFAS were being developed and at least 23 PFAS were analyzed at all 279 sites with 264 sites having 28 PFAS analyzed (table 4). Additional PFAS were added as they became available for analysis and 15 sites sampled at the beginning of the project did not have 11-chloroeicosafluoro-3-oxaundecane-1-sulfonate (11Cl-PF3OUdS) or 9-chlorohexadecafluoro-3-oxanonane-1-sulfonate (9Cl-PF3ONS), 10 sites did not have 8:2 fluorotelomersulfonate (8:2 FTS), eight sites did not have 4,8-dioxa-3H-perfluorononanoate (ADONA), and five sites did not have perfluoro-2-propoxypropanoate (HFPO-DA). Summary statistics in table 4 represent values that were detected over the reporting level. Values estimated between the reporting level and minimum detection level were censored at the sample result level and are further discussed in appendix 1.

To date, EPA has not established MCLs for any PFAS under the Safe Drinking Water Act as it applies to public drinking-water systems. The only health-based guidelines that exist for PFAS in drinking water is the lifetime HA for combined concentrations PFOA and PFOS of 70 ng/L (U.S. Environmental Protection Agency, 2016a, b80). Concentrations at five sites exceeded 70 ng/L: Woo-0215 (1,540 ng/L), Woo-0216 (156 ng/L), Woo-0121 (104 ng/L), Mal-0410 (98 ng/L), and Ber-0468 (71 ng/L) (fig. 5). Overall, 37 sites had detections above the reporting level for PFOA, PFOS, or both, with specific sites and results shown in appendix 2, table 2.1, and table 2.4. Figure 5 shows that 18 of the 37 sites with detections for PFOA or PFOS were located in counties that border Ohio on the western side of the State in the Ohio River Valley and 13 sites with detections for PFOA or PFOS were located on the eastern side of the State in the eastern panhandle. Sites with detections for PFOA and PFOS were highest in the western counties, with 4 samples exceeding the HA of 70 ng/L and 3 sites with concentrations between 20 and 40 ng/L. West Virginia’s eastern panhandle had one site that exceeded the HA and one site with concentrations between 40 and 70 ng/L. Five surface-water sites and one groundwater site with detections for PFOA or PFOS were in counties in the interior of the State. All of these interior sites within the Appalachian Plateaus Physiographic Province (fig. 4) had combined PFOA and PFOS concentrations below 20 ng/L.

Out of 279 sites sampled for this study, 212 sites (76 percent) had no PFAS detected above the reporting level. Detections for PFAS consisted of 67 sites (24 percent) with at least 1 PFAS, 27 sites (10 percent) with at least 2 PFAS, and 16 sites (6 percent) with at least 3 PFAS detected above the reporting level (appendix 2; fig. 6A). Forty-seven of the sites with detections for PFAS were sites with groundwater sources and 20 were from surface-water sources. Only 3 surface-water sites had more than one PFAS detected above the reporting level and no surface-water sites had more than 5 PFAS detected. Total PFAS had a significant positive Spearman correlation (rho = 0.71, p < 0.001), with the number of PFAS detected per site (fig. 6A) indicating that higher concentrations of PFAS may be expected at sites with a higher number of PFAS analytes detected. The highest number of PFAS analytes and the highest total PFAS concentrations were detected at groundwater sites that had a mixture of 9 different PFAS analytes (fig. 6A). The total PFAS detected for an individual site was determined by calculating the sum of all PFAS detected at a site above the reporting level.

There were 7,796 individual results reported for PFAS (McAdoo and others, 2022) with 144 results over the reporting level. Out of the 28 PFAS analytes shown in table 4, only 12 were detected at least once above the reporting level. Detections were too infrequent for estimating median values with nonparametric statistics and detections were generally low with a small subset of high concentrations at specific sites (appendix 2). Ten of the 12 PFAS detected were PFCA and PFSA compounds in the 4 to 9 carbon range with PFOA (8 carbons) having the highest number of detections (29, fig. 6B) and also the highest concentration (1,540 ng/L, table 4). Other PFCA compounds that were detected included PFBA (4 carbons), perfluoropentanoate (PFPeA, 5 carbons), perfluorohexanoate (PFHxA, 6 carbons), perfluoroheptanoate (PFHpA, 7 carbons), and perfluorononanoate (PFNA, 9 carbons). PFSA compounds that were detected included perfluorobutanesulfonate (PFBS, 4 carbons), perfluoropentanesulfonate (PFPeS, 5 carbons), PFHxS (6 carbons), and PFOS (8 carbons). Two of the 12 analytes detected were precursors and included 6:2 FTS, which had the second highest number of detections, and HFPO-DA, which was detected at only three sites (table 4; appendix 2).

Spatial Distribution of PFAS in West Virginia’s Source Water

There were 173 samples collected from groundwater sources (wells and springs) and 106 samples collected from surface-water sources (rivers and reservoirs). Groundwater sites were concentrated in the Ohio River alluvium in the west, the karst terrain in the east, and abandoned underground coal-mine aquifers in the south. Some groundwater sites were located in karst aquifers in the southeastern part of the State and in the Valley and Ridge Physiographic Province (figs. 3 and 4). All of the sites with combined PFOA and PFOS concentrations greater than 40 ng/L were at groundwater sites; only one surface-water site had a concentration greater than 20 ng/L. This suggests that higher concentrations are more common in groundwater, especially in areas near a source or sources of PFAS.

The Ohio River Valley had groundwater and surface-water sites with the highest total PFAS concentrations and several sites with multiple PFAS detected. The Ohio River alluvium had 34 groundwater sites, of which 18 had at least 1 PFAS detected (fig. 7) and 4 sites were determined to exceed the HA for combined PFOA and PFOS (fig. 5). Four of the 6 surface-water sites that withdraw water from the Ohio River had detections of at least 1 PFAS, and 2 of these sites had the highest number of detections (fig. 7) and highest total PFAS (fig. 8) for surface water in the dataset. Groundwater sites with the highest number of detections and highest total PFAS concentrations were located in Wood County, but sites with multiple PFAS detections and relatively high total-PFAS concentrations were found throughout the entire Ohio River Valley of West Virginia (figs. 7 and 8).

There were 66 groundwater sites in West Virginia’s eastern panhandle, 23 of which had detections for at least 1 PFAS. The primary source of water for these sites is generally from karst aquifers, located in Jefferson and Berkeley Counties. Two of these sites had greater than 6 PFAS detected. These two counties also had the highest total PFAS concentrations in the karst aquifers of West Virginia (fig. 8) with one site (Ber-0468) exceeding the HA for combined PFOA and PFOS (fig. 5). There were also 15 surface-water sites in the east, of which 8 sites had detections for 1 PFAS; total PFAS concentrations were less than 20 ng/L for all surface-water sites in this region (fig. 8).

Groundwater and surface water sites located in the Appalachian Plateaus Physiographic Province and karst aquifer in the southeast (fig. 4) had relatively few detections for PFAS. In this area, there were 73 groundwater sites, of which 6 sites had only one PFAS detected. Five of these sites had total PFAS concentrations less than 20 ng/L, and 1 site in southern West Virginia had a detection for 6:2 FTS of 57 ng/L (fig. 8). This area also had 84 surface-water sites, of which PFAS was detected at only 8 sites. Seven of the 8 sites in this region had only one PFAS detected, with 1 site with 4 PFAS detected and a total PFAS concentration of 21.4 ng/L.

Limitations

Drinking-water-quality criteria were used as benchmarks against which to compare samples, but this study did not attempt to evaluate finished drinking-water quality. Many public- water systems employ different treatment techniques for specific contaminants and the raw-water results described here may not be directly applicable to human exposure. The data analyzed for this report represent a one-time sampling event and do not capture temporal variability in concentrations. Concentrations of PFAS and other contaminants may be highly variable at surface-water sites, depending on flow conditions. Some groundwater sites may be directly influenced by surface water or experience changes in concentrations at different times of the year. The data are not uniformly distributed across the study area, with more groundwater sites in aquifers that produce sufficient yield for public supply (Ohio River alluvium, karst terrane, abandoned underground coal-mine aquifers). Surface-water sites were more evenly distributed throughout the State but may be directly impacted by point discharges, more variable than groundwater sites, and should not be used to assume PFAS concentrations in adjacent aquifers lacking groundwater data. Therefore, a large portion of fractured-rock aquifers in the Appalachian Plateaus Physiographic Province and Valley and Ridge Physiographic Province that lack groundwater samples cannot be assessed for the occurrence of PFAS, and PFAS exposure to private homeowners who rely on these aquifers as a primary drinking-water source is currently unknown.

Future Investigations

Data collected from groundwater and surface-water sites for this study represent PFAS concentrations across the State from one moment in time. These data are useful for understanding the statewide distribution and occurrence of PFAS in the source water for public supplies but questions that could be considered for future studies may include the following: (1) what is the PFAS concentration in treated finished water at sites that had detections for PFAS; (2) what is the distribution of PFAS in domestic wells in areas of contamination; (3) what is the PFAS distribution in domestic wells where there is a lack of groundwater data; (4) what are the major sources and exposure pathways of PFAS in West Virginia; and (5) what are influences on transformation and change in PFAS concentrations over time in surface water and groundwater. Possible approaches to answering these questions could include: (1) sampling finished water at public-water systems with detections of PFOA or PFOS over the reporting level; (2) sampling of domestic wells where there is a lack of data for groundwater resources and in areas of contamination; (3) targeted sampling of suspected sources contributing PFAS to public-water supplies and understanding how those PFAS sources affect drinking water, fish tissue, and other pathways of human exposure; and (4) long-term monitoring for PFAS in groundwater and surface water to understand PFAS fate and transport in areas of known contamination.