Climate

The topography of New Mexico ranges from high elevation forested mountains to lower elevation deserts. Except in the mountainous regions, the climate is primarily arid to semiarid. Average annual precipitation across New Mexico for 1980–2010 ranged from 6 inches in lower elevation areas of northwestern and southern New Mexico to 52 inches in higher elevation areas of northern New Mexico (New Mexico Office of the State Engineer/Interstate Stream Commission, 2018). Precipitation as snowfall in the winter months is a substantial source of surface-water and aquifer recharge and drives most of the streamflow in the Rio Grande in the northern portion of the State. Summer thunderstorms, known as monsoons, are also a highly variable source of precipitation and drive streamflow in the southern part of the State (Moeser and others, 2021).

Groundwater Hydrogeologic Setting

Groundwater sampling locations (117 groundwater well sites) were selected in populated areas across the State, which were also located in unconfined surficial aquifers (fig. 1, table 1). The majority of these wells supply public water systems. Because the majority of these wells were part of public water systems, it should be noted that the wells were sampled prior to any water treatment. Additionally, to provide a comprehensive assessment of PFAS in sources of water to those public water systems, springs and surface-water diversions within those systems were also sampled for the same constituents as the groundwater samples (table 1). Diversions, as defined by USGS, are locations where water is withdrawn or diverted from a surface-water body. For example, a diversion could be a point where water is diverted from a river into a water system or a point where water is withdrawn from a reservoir (USGS, 2022b). Diversions, while technically representing surface water, were analyzed for groundwater constituents because they were part of public supplies that consisted mostly of groundwater and (or) springs (table 1). However, the analytical results for surface-water diversion samples are not included with groundwater analytical results because they do not represent groundwater due to the diversions being derived from surface-water inputs. Springs are considered representative of groundwater chemistry because each is a location where groundwater emerges at land surface at the spring locations. Major aquifers or aquifer systems that were sampled included the Rio Grande aquifer system (39 groundwater wells and 3 surface-water diversions), the High Plains aquifer (52 groundwater wells), the Colorado Plateaus aquifers (5 groundwater wells), and the Roswell Basin aquifer system (4 groundwater wells; fig. 1). Forty-four sites are located in other surficial aquifers; of these, 24 are springs and 17 are wells.

The Rio Grande aquifer system (fig. 1) is a 70,000-square-mile (mi²) area of interconnected basin-fill deposits in southern Colorado, New Mexico, and western Texas (Robson and Banta, 1995). The Rio Grande headwaters are located in southern Colorado, from which it flows southward through New Mexico from deep canyons north of Santa Fe, through broad basins and valleys, to the Texas State line. The Rio Grande rift is the primary geologic feature of the system, and the aquifer ranges in thickness from about 20,000 feet (ft) near Albuquerque, N. Mex., to 2,000 ft near El Paso, Texas. The Quaternary and Tertiary Santa Fe Group is the principal hydrologic member of the system and consists primarily of gravel, sand, and interbedded clays, and lesser amounts of lava flows, tuffs, and breccias. Recharge is primarily from precipitation in mountainous areas, seepage from the Rio Grande and its tributaries, as well as from irrigation conveyance structures and return flow (Houston and others, 2021). Jurgens and others (2022) characterized the age distribution for groundwater in 21 principal aquifers across the United States, including the Rio Grande aquifer system, and determined that if the principal aquifer contained modern water (1953 or newer), the aquifer could be susceptible to land-surface contamination. Jurgens and others (2022) found that the Rio Grande aquifer system had approximately 15 percent modern water and may be less susceptible to land-surface contamination.

The High Plains aquifer occupies a 174,000-mi² area that extends across parts of several States, including eastern New Mexico (fig. 1). The aquifer primarily consists of unconsolidated gravel, sand, and silt, and the Ogallala Formation is the principal hydrologic unit. The Miocene and early Pliocene Ogallala Formation was deposited by streams and therefore consists of deposits of variable thickness that range from 0 to 500 ft in buried paleochannels (Robson and Banta, 1995). Recharge to the High Plains aquifer is primarily from precipitation infiltration, irrigation water infiltration, or seepage from surface water. Jurgens and others (2022) determined that the High Plains aquifer contained approximately 30 percent modern water.

The Colorado Plateaus aquifers encompass 110,000 mi², including northwestern New Mexico (fig. 1). Four principal aquifers compose the system, with the two in New Mexico being the Mesaverde aquifer and the Dakota-Glen Canyon aquifer system (Robson and Banta, 1995). Study sites 1, 2, and 3 are in the Farmington area along the San Juan and Animas Rivers and vary in hydrogeologic setting. Site 1 (well depth, 145 ft) and site 2 (well depth, 40 ft) are not located in the Mesaverde aquifer according to Stewart (2018) and are in smaller, less productive aquifers. Site 1 is completed in the Pictured Cliffs Sandstone, and site 2 is completed in Naha and Tsegi eolian deposits (Stewart, 2018). The site 3 well is shallow and completed at 23 ft in alluvium rather than in one of the principal aquifers. Wells at sites 7 and 8 are around 2,000 ft deep, and they are likely completed in the Dakota-Glen Canyon aquifer system, which is composed of several sandstone and conglomerate water-yielding units (Robson and Banta, 1995). Recharge to these aquifers has been identified along outcrops forming structural boundaries, as water enters the groundwater system through these outcrops. Recharge also enters from stream-channel loss as streams cross the outcrops (Craigg, 2001). Newer studies have shown that there are some anthropogenic recharge sources in the basin, such as seepage from irrigation and leaking water and sewer lines (Robertson and others, 2016). Jurgens and others (2022) determined that the Colorado Plateaus aquifers contained approximately 20 percent modern water.

The Roswell Basin aquifer system occupies a 12,000-mi² area in southeastern New Mexico (fig. 1) and is composed of an alluvial aquifer and an underlying carbonate-rock aquifer. Groundwater is primarily present in the San Andres Limestone and the overlying Queen and Grayburg Formations (Robson and Banta, 1995). The alluvium overlying the carbonate rocks is generally west of the Pecos River and ranges in thickness from 150 to 300 ft. Recharge is mainly from precipitation in the outcrop areas of the San Andres Limestone and the alluvium. Additional recharge occurs from losing streams, as well as through sinkholes and solution fractures (Land and Newton, 2008).

Approximately one-third of the study sites in table 1 did not fall within the aforementioned primary aquifer systems. Although most of these sites are located in southern New Mexico, two sites are located in north central and northeastern New Mexico (fig. 1). Site 5 is a relatively shallow well with a depth of 68 ft and, based on the surficial geology map of New Mexico (Horton, 2017), is located in Tertiary intrusive rock. Site 6, a groundwater well, is 680 ft deep and is located in Quaternary volcanic rocks. In southern New Mexico, most of the sites (including spring sites 98, 109, 113, and 123–138, and groundwater well sites 110–112, 114–119, 121–122, and 139; table 1) are located in the Yeso Formation (a sandstone; Horton, 2017). Sites 96, 97, and 99 (springs) are located in the Abo Formation (a sandstone; Horton, 2017). Site 84 (groundwater well) is located in the Bursum Formation of the Madera Group (shale), and site 120 (groundwater well) is located in the San Andres Limestone (Horton, 2017).

Surface-Water Hydrologic Setting

Surface-water samples were collected from 18 established USGS streamgaging sites across New Mexico (fig. 2, table 2). The rivers sampled were the Animas River, San Juan River, Canadian River, Gila River, Rio Grande, Rio Chama, Rio Puerco, and Pecos River. Land cover was evaluated near each surface-water site (fig. 2, table 3) using the near-site watershed method as described in Medalie and others (2020) and in the “Methods” section herein.

The Animas and San Juan Rivers flow through the northwestern portion of the State (fig. 2). Both rivers supply water valuable for municipal, agricultural, cultural, and recreational uses. From its headwaters in southwestern Colorado, the Animas River flows southwest until it eventually joins the San Juan River in Farmington, N. Mex. (EPA, 2022c). The Animas River at Farmington, N. Mex. sampling site (herein referred to as “Animas Farmington”; site number 09364500) is approximately 1.25 miles (mi) upstream from the confluence with the San Juan River. The land cover near this sampling site is 72 percent shrubland and 14 percent developed (fig. 2, table 3). The San Juan River enters New Mexico through Navajo Lake, a reservoir used for flood control; water supply for irrigation, domestic, and industrial use; water storage; hydropower; and recreational purposes (EPA, 2022c). San Juan River near Archuleta, N. Mex. (herein referred to as “San Juan Archuleta”; site number 09355500), one of the two sampling sites on the San Juan River, is approximately 6 mi downstream from Navajo Lake (fig. 2). This site is surrounded by 62 percent shrubland and approximately 27 percent forested land. From Navajo Lake, the San Juan River flows through agricultural land in the Colorado Plateau until it reaches Farmington, N. Mex., and is joined by the Animas River at the Animas Farmington site. Below the confluence with the Animas River, the San Juan River flows westward, where it provides irrigation water for agricultural communities, past Fruitland, N. Mex., at San Juan near Fruitland, N. Mex., sampling site (herein referred to as “San Juan Fruitland”; site number 09367540). The land cover near this site is 86 percent shrubland. The San Juan River then continues through northwestern New Mexico, flowing past two coal-fired power plants and through a sandstone canyon where it leaves New Mexico (EPA, 2022c). The sampling locations in the San Juan and Animas rivers were sampled seven to nine times throughout the project (table 2).

The headwaters of the Canadian River flow southeast through northeastern New Mexico and the river continues to flow eastward until it leaves the State (Oklahoma History Center, 2022). In downstream order, the three locations sampled along the Canadian River are: Canadian River near Sanchez, N. Mex. (herein referred to as “Canadian Sanchez”; site number 07221500); Canadian River below Conchas Dam (herein referred to as “Canadian Conchas”; site number 07224500); and Canadian River at Logan, N. Mex. (herein referred to as “Canadian Logan”; site number 07227000) (fig. 2). The Canadian River sites were only sampled one or two times throughout the project (table 2), and all three sites are dominated by shrubland land cover with little to no developed land (table 3).

The Gila River near Gila, N. Mex., sampling site (herein referred to as “Gila”; site number 09430500) is in the Upper Gila watershed in southwestern New Mexico (fig. 2). Much of the 5,532 mi of water courses in the Upper Gila watershed are intermittent streams that occasionally flow in the summer after thunderstorms (Natural Resources Conservation Service, 2022). Gila had a mean annual streamflow of about 156 cubic feet per second for 1927–2011 (Natural Resources Conservation Service, 2022). The Gila River was only sampled twice during the study (table 2), and the land cover near this site consists of approximately 73 percent forested land and 27 percent shrubland/grassland (table 3).

The headwaters of the Rio Grande are in southwestern Colorado, and six sites were sampled along the Rio Grande as it flows south through New Mexico (fig. 2). In downstream order, the sampled sites are Rio Grande below Taos Junction Bridge near Taos, N. Mex. (herein referred to as “Rio Grande Taos”; site number 08276500); Rio Grande above Buckman Diversion near White Rock, N. Mex. (herein referred to as “Rio Grande Buckman”; site number 08313150); Rio Grande at Alameda Bridge at Alameda, N. Mex. (herein referred to as “Rio Grande Alameda”; site number 08329918); Rio Grande at Valle de Oro, N. Mex. (herein referred to as “Rio Grande Valle de Oro”; site number 08330830); Rio Grande Floodway at San Marcial, N. Mex. (herein referred to as “Rio Grande Floodway”; site number 08358400), and Rio Grande at El Paso, Tex. (herein referred to as “Rio Grande El Paso”; site number 08364000). Depending on the location, these sites were sampled between 2 (Rio Grande Taos) and 13 (Rio Grande Valle de Oro) times (table 2), and the sampling frequency was variable and coordinated with other USGS activity in the study area. From northern New Mexico, the Rio Grande flows southward to the most upstream sampling site, Rio Grande Taos. This site is surrounded by about 75 percent shrubland/grassland land cover and 22 percent forested land cover (table 3). The Rio Grande continues from the Upper Rio Grande watershed to the confluence with the Rio Chama, which is the largest tributary to the Rio Grande (Natural Resources Conservation Service, 2022). The Rio Chama below Abiquiu Dam, N. Mex., sampling site (herein referred to as “Rio Chama Abiquiu”; site number 08287000) is surrounded by about 70 percent shrubland/grassland, 26 percent forested land, and 4 percent open water. Downstream from the confluence of the Rio Grande and the Rio Chama near Santa Fe, N. Mex., is the sampling site Rio Grande Buckman. The site is surrounded by 84 percent shrubland/grassland and by about 10 percent forested land. Just downstream from this site, water is pumped from the Rio Grande to a treatment plant to serve the City of Santa Fe and surrounding communities (Buckman Direct Diversion, 2015).

Samples were collected at the northern edge of New Mexico’s largest municipality, Albuquerque, at Rio Grande Alameda, and near the southern boundary of the urban area at Rio Grande Valle de Oro. Within the Middle Rio Grande Basin as well as this reach within Albuquerque, the inner valley of the Rio Grande has a system of riverside drains and irrigation canals and ditches. The irrigation canals and ditches are primarily used during irrigation season (typically mid-March through October), when water from the river is diverted by the Middle Rio Grande Conservancy District for delivery to irrigated fields (Bartolino and Cole, 2002). The riverside drains extend parallel to both sides of the river and were constructed to capture lateral groundwater flow from the river. This helps to stabilize the groundwater table to avoid waterlogging of soils near the river. These riverside drains, which also receive flow from interior drains extending across the valley to intercept seepage from canals and irrigated fields, eventually flow back to the Rio Grande (Bartolino and Cole, 2002).

Rio Grande Alameda is represented by about 80 percent shrubland/grassland and 9 percent developed land (table 3). Downstream from this sampling location, the Albuquerque Bernalillo County Water Utility Authority operates facilities constructed as part of the San Juan-Chama Drinking Water Project to provide a percentage of Albuquerque’s drinking water. The diverted water has been imported via a tunnel from tributaries of the San Juan River in the Colorado River Basin into reservoirs along the Rio Chama, which flow into the Rio Grande (Albuquerque Bernalillo County Water Utility Authority, 2022). A raw-water pump station diverts water from the Rio Grande to a treatment plant to provide drinking water for Albuquerque residents. After use, unconsumed water is directed to the wastewater treatment plant (WWTP; Albuquerque Bernalillo County Water Utility Authority, 2021). The Albuquerque WWTP is the largest treatment facility in New Mexico and discharges into the Rio Grande. Downstream from the Albuquerque WWTP is a large arroyo (132-mi² drainage area), which flows into the Rio Grande from the east. The arroyo serves as the primary channel for snowmelt and stormwater from areas east of Albuquerque (City of Albuquerque Parks and Recreation Department Open Space Division, 2014). Several major stormwater diversions also contribute large amounts of stormflow to the Rio Grande from major residential and commercial areas.

Rio Grande Valle de Oro is approximately 3 mi downstream from the Albuquerque WWTP. Near Rio Grande Valle de Oro, the surrounding land cover consists of about 49 percent shrubland/grassland and 40 percent developed land (table 3). Farther downstream from this site, the Rio Grande is joined by the Rio Puerco (fig. 2). The Rio Puerco watershed (table 2) of west central New Mexico includes approximately 4,834 mi of water courses that usually flow intermittently in the summer after storms. The Rio Puerco contributes a very small percentage of the Rio Grande’s flow; however, it contributes over half of the total sediment load that enters the Elephant Butte Reservoir (Natural Resources Conservation Service, 2022). The Rio Puerco near Bernardo, N. Mex. sampling site (herein referred to as “Rio Puerco Bernardo”; site number 08353000) is surrounded by 87 percent shrubland/grassland and 13 percent barren land.

Rio Grande Floodway is located about 30 mi upstream from the Elephant Butte Reservoir (fig. 2), and the surrounding land cover is composed almost exclusively of shrubland/grassland (table 3). Rio Grande El Paso is the most downstream sampling location on the Rio Grande and is more than 100 mi south of Elephant Butte Reservoir, across the New Mexico border in El Paso, Tex. Land use near this site is composed of 49 percent shrubland/grassland and almost 40 percent developed land.

The headwaters of the Pecos River are in northern New Mexico, and the river flows southward through the eastern portion of the State until it reaches the Rio Grande south of the New Mexico border (Bureau of Reclamation, 2021). Three sites were sampled along the Pecos River (fig. 2). The most upstream site is Pecos River near Puerto de Luna, N. Mex. (herein referred to as “Pecos Puerto de Luna”; site number 08383500) in the Pecos Headwaters watershed. The Pecos River flows southeast to the Upper Pecos-Long Arroyo watershed where the Pecos River near Artesia, N. Mex. sampling site (herein referred to as “Pecos Artesia”; site number 08396500) is located. The next site downstream is Pecos River at Red Bluff, N. Mex. (herein referred to as “Pecos Red Bluff”; site number 08407500) in the Upper Pecos-Black watershed. All three sites have dominantly shrubland/grassland land cover (table 3).

Groundwater

Water samples were collected from 117 wells and 24 springs by following standard USGS protocols (USGS, variously dated) (table 1). Several sites that had detections of PFAS were resampled, and the sampling frequency at each site is indicated in table 1. Sites were resampled, when possible, if there was a PFAS detection or they were resampled to meet objectives for additional sampling in Curry and Otero Counties. Seasonal variation was not evaluated. Before water samples were collected, field properties including pH, water temperature, specific conductance, and dissolved oxygen were measured in a flow-through cell during well purging at each sampling site. All wells were pumped using dedicated pumps installed within each well by the owner, who reported discharge. Water level, casing dimensions, and pumping rate were used to calculate the purge volume and the time required to purge three casing volumes and allow field parameters to stabilize prior to sample collection. Spring discharge could not be measured because of the spring infrastructure and access. Groundwater samples were collected from a raw-water tap before any treatment by the system owner. Water samples were collected using C-flex tubing connected in line with the groundwater raw-water tap or pumped with a peristaltic pump from the spring orifice and filtered (0.45-micrometer pore size) for major cations, trace elements, alkalinity, nutrients, dissolved organic carbon (DOC), and radiocarbon. The major cations, trace element, and DOC samples were preserved to less than pH 2. Tritium, stable isotopes of water (δ¹⁸O and δ²H), and PFAS were collected as unfiltered samples directly from the raw-water tap or directly from the spring. PFAS samples were collected directly into high density polyethylene sample bottles by USGS staff who wore nitrile gloves over elbow length polyethylene gloves during sampling. Sampling at seventeen sites—sites 21, 22, 28, 29, 31–35, 38–40, 47, 49, 52, 57, and 69—only included PFAS and stable isotopes. Nonwaterproof labels were affixed to the bottles and annotated using a writing implement other than permanent marker, which was prohibited.

Surface Water

Eighteen surface-water locations were sampled across the State (table 2), as well as six surface-water diversions. The rivers were sampled during stable flow, and stormflow events were avoided. Additionally, the frequency of sampling was dependent on the location. Some surface-water sites were sampled while sampling for other projects to minimize travel time and most efficiently use available resources. Although some sites were sampled more frequently than others, seasonality could not be captured or evaluated during this study. Before water samples were collected at a river cross section, field properties including pH, water temperature, specific conductance, and dissolved oxygen were measured from five points along the cross section at the sampling location and the median value was used, with the exception of pH, which was measured in still water from the churn. Surface-water samples were collected by using equal width increment sampling or with dip sampling for average velocities less than 1.5 feet per second, following standard USGS protocols (USGS, variously dated). A polycarbonate sample bottle and polyoxymethylene nozzle were used by USGS staff to collect each PFAS surface-water sample while wearing elbow length polyethylene gloves and nitrile gloves over the polyethylene gloves. Surface-water samples from the cross section for PFAS were then composited into a polycarbonate churn (equipment was cleaned with Liquinox, tap water, 5 percent hydrochloric acid, deionized water, methanol, and PFAS-free blank water), while wearing elbow length polyethylene gloves and nitrile gloves over the polyethylene gloves. A separate sample for wastewater tracers was collected using Teflon equipment (cleaned with Liquinox, tap water, 5 percent hydrochloric acid, deionized water, methanol, and organic-free blank water) at selected surface-water sites after processing of the sample collected using the polycarbonate equipment. Water samples were collected using C-flex tubing and filtered (0.45-micrometer pore size) for major cations, trace elements, and alkalinity. The major cation, trace element, and DOC samples were preserved to less than pH 2. DOC samples were collected directly from the centroid of the surface-water cross section because the churns used for sampling had both been rinsed with methanol. DOC samples were filtered in the laboratory. Wastewater tracers were collected as raw, unfiltered samples from the Teflon churn. PFAS were collected as raw, unfiltered samples from the polycarbonate churn directly into high density polyethylene sample bottles by USGS staff wearing elbow length polyethylene gloves and nitrile gloves over the polyethylene gloves. Nonwaterproof labels were affixed to the bottles and annotated using a writing implement other than permanent marker, as recommended by guidance for minimizing PFAS contamination during sample collection (Interstate Technology Regulatory Council, 2023).

Quality-Control Sample Collection

Quality-control samples consisted of replicate and blank samples collected using the same procedures to collect the environmental samples (USGS, variously dated). Field blanks for groundwater sites and certified PFAS-free water were poured directly into the PFAS sample bottles. Field blanks for surface-water sites were collected as described in the National Field Manual (USGS, variously dated). Field blanks were collected at several different sites and processed by different field personnel. Inorganic blanks at surface-water sites were not collected because of the use of a different type of blank water (inorganic-free blank water) that would negate the PFAS-cleaning procedures. Certified inorganic-free blank water was used for major ions, trace elements, nutrients, and DOC. Certified organic-free blank water was used for wastewater tracers, and certified organic and PFAS-free water was used for PFAS. Replicates were collected using the methods outlined in the National Field Manual (USGS, variously dated), concurrently for surface water and sequentially for groundwater. Replicate locations were chosen randomly, as it was unknown whether PFAS would be present at a given site. When possible, however, replicates were added during subsequent sampling events at sites where detections were found. Quality-control sample data are provided in appendix 1, tables 1.2 and 1.4.

Groundwater and Otero County Surface-Water Diversion

Water samples (including surface-water diversion samples) were analyzed for major cations, trace elements, and nutrients by the USGS National Water Quality Laboratory in Denver, Colorado. Analytical methods from the National Water Quality Laboratory included inductively coupled plasma-mass spectrometry to determine arsenic concentration (Garbarino and others, 2006). Inductively coupled plasma atomic emission spectrometry was used to analyze for cations (calcium, iron, magnesium, manganese, potassium, and sodium) (Fishman, 1993). Anions (chloride, fluoride, and sulfate) were analyzed by ion chromatography, and silicon dioxide (SiO₂) was analyzed by discrete analyzer colorimetry (Fishman and Friedman, 1989). Nitrate (NO₃) plus nitrite (NO₂) was analyzed by colorimetry (Patton and Kryskalla, 2011).

Stable isotope ratios of oxygen (δ¹⁸O) and hydrogen (δ²H) of water were measured at the USGS Reston Stable Isotope Laboratory in Reston, Virginia. Standardization is based on international reference materials, Vienna Standard Mean Ocean Water, and Standard Light Antarctic Precipitation. Reston Stable Isotope Laboratory samples were analyzed using mass spectrometry following methods by Révész and Coplen (2008a, b). The two sigma uncertainties are 0.2 parts per thousand (per mil) for oxygen and 2 per mil for hydrogen isotopic ratios reported relative to Vienna Standard Mean Ocean Water.

The University of Miami Tritium Laboratory in Miami, Florida, measured tritium in samples using the electrolytic enrichment and gas-counting method, with a reporting limit of 0.3 picocuries per liter (pCi/L; Östlund, 1987). Carbon-14 (¹⁴C) and δ¹³C (normalized ratio of carbon-13 and carbon-12) were analyzed by the National Ocean Sciences Accelerator Mass Spectrometry at the Woods Hole Oceanographic Institution, Massachusetts (Stuiver and Polach, 1977). ¹⁴C was reported as absolute percent modern carbon and δ¹³C as per mil Vienna Pee Dee Belemnite.

Surface Water

Surface-water samples were analyzed for organic chemicals (wastewater tracer compounds) at the USGS Integrated Water Chemistry Assessment Laboratory in Boulder, Colo. DOC and aqueous inorganic chemicals (trace elements, rare earth elements, and major cations and anions) were measured by the USGS Analytical Trace Element Chemistry Common Services Laboratory in Boulder, Colo. Neutral organic wastewater tracer compounds were extracted by continuous liquid-liquid extraction and measured by gas chromatography-tandem mass spectrometry in multiple monitoring mode following methods by Barber and others (2000), with quantified compound values reported above the reporting level. Surrogate standards were added prior to extraction and workup procedures, and isotopically labeled internal standards were added to the extract immediately prior to analysis. DOC samples were analyzed by platinum catalyzed persulfate/ultraviolet light oxidation with infrared detection, and ultraviolet light absorbance at 254 nanometers was measured in a 1-centimeter quartz cell (Weishaar and others, 2003). Major anions were measured by ion chromatography with conductivity detection using a Dionex Model ICS3000 ion chromatograph with suppressed conductivity detection, an IonPac AS18 analytical column with an IonPac AG18 guard column, and a 28-millimolar (mM) potassium hydroxide mobile phase (Pfaff, 1993). Metals and major cations were measured by inductively coupled plasma-optical emission spectrometry (Garbarino and Taylor, 1979). Trace elements and rare earth elements were measured by inductively coupled plasma-mass spectrometry (Garbarino and Taylor, 1996). Individual samples were analyzed in triplicate and the reporting level was determined for each batch sample run.

Total Per- and Polyfluoroalkyl Substances Concentrations

The total PFAS concentration was calculated for each site listed in this report, and any individual PFAS concentrations of the 28 compounds were reported below the laboratory reporting level, with a “<” were treated as “0” values when calculating the total PFAS concentration. Additionally, when the average PFAS concentration was plotted on figures, if there were no detections of PFAS in the whole sample, the value was considered “0” when calculating the average.

Land Cover Assessment

The surface-water sites selected for this study were located at established USGS streamgaging sites. Therefore, the sites were often not at the downstream end of a watershed, and selecting a way to evaluate surrounding land cover needed to be more specific than evaluating land cover across the entire watershed a given site was located in. The surrounding land cover at surface-water sites (table 3) was determined using the methods outlined in Medalie and others (2020), which showed that glyphosate detections were correlated to near-site watershed land use more strongly than land use within the whole watershed. A 15-kilometer buffer was created around each surface-water site and then clipped to the upstream watershed using ArcGIS Pro (Esri, 2023). The most recently available land cover data (Dewitz and others, 2021) were used to determine distribution within the buffer. The land cover in each near-site watershed was then used to evaluate associations between PFAS detections and land cover. The sites were categorized into a near-site land cover category on the basis of the percentage of developed (includes a range of developed areas mixed with grass cover with some constructed materials to areas that are composed entirely of impervious surfaces), agricultural (includes pasture, hay, and cultivated crops), and undeveloped land cover (includes forests, shrublands, grasslands, open water, wetlands, and barren land) (Dewitz and others, 2021). The categories were defined with the following criteria: (1) developed, near-site watershed contained greater than 30 percent developed land cover; (2) mixed, near-site watershed contained greater than 10 percent mixed land cover (developed plus agricultural ranged from 10 to 29.9 percent); and (3) undeveloped, near-site watershed contained greater than 90 percent undeveloped land cover (table 3). The medians of total PFAS concentration were calculated by land cover category, and total PFAS concentrations for which concentrations of the 28 compounds were not detected were treated as “0” values. Additionally, when calculating the medians, the sites were put into two groups: sites that were sampled more than five times and sites that were sampled less than five times (table 2). Land cover was not evaluated for groundwater locations because of the complex flow paths of groundwater and the difficulty of delineating relevant areas contributing recharge.

Per- and Polyfluoroalkyl Substances Flux Calculations

Instantaneous loading rates, also known as flux, for PFAS in surface water were calculated using an approach of multiplying concentration times streamflow (Meals and others, 2013),

where The instantaneous flux is the instantaneous rate at which the load passes a point in the river, converted into a daily rate. When evaluating total PFAS concentrations in this report, individual quantified PFAS concentrations were added for a total concentration; concentrations below the detection level were treated as a value of zero. This PFAS concentration calculation applies to land cover analysis, average total PFAS concentrations, PFAS proportion graphs where total PFAS concentrations are displayed, flux calculations, and other plots where total PFAS concentrations are shown.

Statistical Analysis

To help with interpretation of major- and trace-element concentrations, this large dataset was analyzed using Spearman’s rank-order correlation, defined as the assessment of the increasing or decreasing relationship between the rank of each data point to assess monotonic relationships in the data. Next, principal components analysis was used to determine relationships between multiple analytes. Principal components analysis is a method of reducing the number of attributes (variables) of a large dataset while preserving both statistical information (in the form of variability) and meaningful properties of the original dataset (Jolliffe and Cadima, 2016). For the principal component analysis, new uncorrelated variables were created by solving an eigenvalue/eigenvector problem using R (version 4.2.3; R Core Team, 2023) that successively maximizes variance. Eigenvectors determine the directions of the principal component feature space, and eigenvalues determine the magnitude of the eigenvectors. The eigenvectors that correspond to the largest eigenvalues (the principal components) were then plotted as arrows overlain on the principal component analysis graph to show a reconstruction of the variance of the original data to reveal important geochemical analytes that separate the sample data into different areas of the principal component analysis graph.

Nonmetric multidimensional scaling (NMDS) is an alternative to the principal components analysis biplot described above (Helsel and others, 2020), which employs distances measured between ranks of the sample data. The NMDS method reduces the complex data structure (many samples and many elements) to represent the pairwise dissimilarity between objects in a low-dimensional space (Clarke and others, 2014, p. 5–6). The “uscore” function for R (version 4.2.3; R Core Team, 2023) from Helsel (2016) was used to compute Uscores of the data, utilizing default values to calculate the ranks of the scores (Helsel, 2012, 2016). Uscores are defined as the ranks of the sample data with lowest values having the lowest numerical rank. NMDS was performed on the uscores to compare dissimilarities within the dataset using metaMDS from the vegan package in R (Oksanen and others, 2016), utilizing Euclidean distance, where zerodist = “add” and autotransform = “false” (Helsel, 2012). Euclidean distance refers to the length of line segment between two points calculated using the Pythagorean theorem. NMDS stress values are calculated with the “metaMDS” function and reflect how well the ordination summarizes the observed distances among the samples. Values less than or equal to (≤) 0.1 are considered fair with good ordination and no real prospect of misleading interpretation; values ≤0.05 indicate good fit, and values greater than or equal to (≥) 0.2 are deemed suspect (Clarke and others, 2014, p. 5–6).

A cluster analysis was used to identify similar groups of samples by evaluating minimum differences within groups and maximum differences among groups using the “hclust” function with Euclidian distance matrix for the elements used in the NMDS analysis. A cluster analysis is defined as an analysis to assess which data points are more similar to each other, thus belonging to a group, than they are to data points in a separate group. The Calinski criterion (a measure of the variance between clusters) was applied with the “cascadeKM” function of the vegan package in R (Oksanen and others, 2016) to determine the number of clusters that maximizes the difference between clusters while minimizing the differences within clusters. The analytes used in the NMDS analysis were evaluated for correlation by calculating Kendall’s tau (a measure of the relationships between ranked data) using the “cenken” function from the NADA package in R (Lee, 2017).

Interpretation of Age Tracers

Age tracers are isotopes of elements associated with water that provide information on the timing for the recharge water to enter the subsurface then undergo subsequent decay, mixing, and water-rock interactions that provide an indication of the length of time the water has been in the subsurface (also referred to as the “age” of the water). ¹⁴C values, reported by National Ocean Sciences Accelerator Mass Spectrometry as absolute percent modern carbon, were denormalized using equation 5 of Plummer and others (2012) to percent modern carbon (pmc). This study did not have enough groundwater samples along a flow path from recharge to sampling location to perform detailed geochemical modeling, so NetpathXL (a spreadsheet interface to program Netpath) was used to compute corrected groundwater ages using model 11 “Revised F&G solid ex” (Parkhurst and Charlton, 2008; Han and Plummer, 2013). Groundwater age was computed using ¹⁴C values of 0 pmc for carbonate rock and 100 pmc for soil CO₂, assuming δ¹³C values of 0 per mil for carbonate and −11.5 and −21.4 per mil for soil CO₂ (Plummer and others, 2012).

The concentration of tritium in precipitation varies spatially, and in the contiguous United States, the concentration is generally lowest in the southwest region including New Mexico (Michel and others, 2018). Categorical classification groups for groundwater age were determined for the sites sampled for this study by using the measured tritium and the tritium precipitation data from Michel and others (2018) and the methods described in Lindsey and others (2019).

Normalization of Rare Earth Elements

Rare earth elements (REEs) typically follow a pattern of concentration change between elements, but some artificial processes can cause some of the REEs to be greater or less than what they would be in the natural pattern, and these differences are considered to be anomalous when the magnitude of the variation is high. Since the studies of Bau and Dulski (1996) and Bau and others (2006), a positive gadolinium (Gd) anomaly, which is related to the use of Gd-based contrast agents in magnetic resonance imaging (MRI), is now considered worldwide to be a distinctive signature of water inputs from wastewater treatment plants in areas with MRI facilities. For this study, aqueous concentrations of REEs were normalized to the North American shale composite (Gromet and others, 1984; Piper and Bau, 2013). Background concentrations of Gd (Gd_background) from geogenic sources were interpolated from a third order polynomial regression fitted to the REE distribution from lanthanum (La) through ytterbium (Yb) at each site. The background concentrations were compared with the sample concentrations to identify samples with a Gd anomaly, which is ratio of the sample Gd to the background Gd greater than 1.

Per- and Polyfluoroalkyl Substances

The majority of the groundwater samples from 141 groundwater sampling sites in this study did not have any detections of PFAS (fig. 5); that is, concentrations were reported by the laboratory as being below the laboratory detection level. Twenty-seven sites had PFAS detected at one or more sampling events. Fourteen sites only had one PFAS present above the laboratory detection level (table 16), 7 sites had two PFAS detected, and 7 other sites had more than two PFAS detected. Some sites were resampled if PFAS were detected. If a site was resampled, a “.1” was added to the report identification number to indicate the first sampling event and a “.2” for the second sampling event, and so on (table 16). Total PFAS concentrations ranged from 0.91 (site 136) to 80.3 ng/L (site 44, second sampling). The most frequently detected PFAS at groundwater sites were PFBS (11 sites), PFPeA (10 sites), and PFHxA (9 sites). The High Plains aquifer had a higher sample density compared to other aquifers in the study, representing 51 of the total groundwater sites, and had detections at 13 of the sites.

Some sites were sampled more than once over the course of the study. Generally, any detected PFAS and reported concentrations were similar over time at a given site. The signature of the PFAS detected (PFBA, PFPeA, PFHxA, and PFBS) was consistent over time for site 44 (figs. 1 and 5), which was sampled three times between December 2020 and June 2021 (fig. 6A). Variability in the PFAS concentrations over time was within the 12-percent variability from replicate samples in this study. Site 18 had two compounds—PFPeA and PFHxA—detected during each of three sampling events between September 2020 and October 2021 (fig. 6B). Three separate sites in a similar geographic area within Curry County—sites 25, 26, and 64—had multiple PFAS detected, with PFBS and perfluorohexanesulfonic acid (PFHxS) present at all three and perfluorobutanoic acid (PFBA), PFPeA, PFHxA, perfluoropentanesulfonic acid (PFPeS), and PFOS present at one or more of the sites in January and October 2021 (fig. 6C,D).

Three springs in a similar geographic area of Otero County—sites 124, 125, and 128 (figs. 1D and 5D, table 16)—were sampled in April 2021 and had similar concentrations of PFBA and PFBS (fig. 6E). Two of the springs also had detections of PFHxA, PFPeA, perfluoroheptanoic acid (PFHpA), and PFOA. In addition, spring 124 had detections of perfluorononanoic acid (PFNA) and PFDA (fig. 6E). These three springs were resampled in October 2021 and no PFAS detections were reported by the laboratory for these later samples, despite laboratory detection levels being similar during both time periods. The flow at the sites was greater during the October 2021 sampling, which could reflect contributions of discharge from different groundwater flow paths during this time and (or) dilution of any PFAS that were present at concentrations below laboratory detection levels.

Field Properties

The field properties water temperature, pH, specific conductance, and dissolved oxygen concentration were measured onsite prior to collection of water-quality samples. Groundwater temperature ranged from 5.0 to 32.9 degrees Celsius (°C) and varied on the basis of the location and depth of the groundwater sampled. Field pH values were circumneutral and ranged from 6.7 to 8.8. Specific conductance ranged from 274 to 52,700 microsiemens per centimeter at 25 degrees Celsius (µS/cm at 25 °C). The majority of samples had dissolved-oxygen concentrations greater than 3 milligrams per liter (mg/L), with some sites having concentrations less than 0.5 mg/L, suggesting suboxic or anoxic conditions (sites 7, 8, 12, 15, 16, 140, and 141) (Jurgens and others, 2009).

Water-Quality Standards

Water-chemistry results were compared with EPA maximum contaminant levels (MCL), secondary maximum contaminant levels (SMCL), and health advisories (EPA, 2022a, 2023) (table 17). In 2016, the EPA established a health advisory limit of 70 ng/L for PFOA and PFOS (EPA, 2022a), and after this study was completed, in June 2022, the EPA issued draft health advisory limits for PFOA and PFOS to 0.004 and 0.02 ng/L, respectively. The draft health advisory also set limits of GenX to 10 ng/L and PFBS to 2,000 ng/L (EPA, 2022a). No groundwater sites exceeded the 2016 health advisory limit. The 2022 draft health advisory limits for PFOS and PFOA are below the analytical method detection level (EPA, 2018). A total of seven sites had samples that exceeded an MCL. Two sites (51 and 53) exceeded the MCL for nitrate as nitrogen (10 mg/L), with the maximum concentration being 23.9 mg/L at site 53. The arsenic MCL (10 µg/L) was exceeded at four sites (13, 14, 15, and 17), with the maximum concentration being 41.7 µg/L at site 13. The uranium MCL (30 µg/L) was exceeded at two sites (2 and 53), with the maximum concentration being 214 µg/L at site 2.

Among SMCLs, the total dissolved solids (TDS) SMCL (500 mg/L, defined by the EPA as being for TDS) was exceeded at 59 sites, with the maximum value of dissolved solids (dried at 180 °C) being 58,100 mg/L at site 2. Sulfate had the second most SMCL (250 mg/L) exceedances (41 sites), with a maximum concentration of 33,800 mg/L at site 2. The chloride SMCL of 250 mg/L was exceeded at 24 sites, with a maximum chloride concentration of 5,730 at site 2. No sites exceeded the MCL (4 mg/L) for fluoride; however, 20 sites exceeded the SMCL (2 mg/L), with a maximum fluoride concentration of 2.97 mg/L at site 43. The iron and manganese SMCLs (300 and 50 µg/L, respectively) were exceeded at three and four sites, respectively, with maximum concentrations of iron (1,010 µg/L) and manganese (459 µg/L) at site 12. The SMCL for pH is a range from 6.5 to 8.5, and pH exceeded 8.5 at four sites (7, 8, 14, and 42), with a maximum value of 8.8 at site 7.

Major Ions

Major ions such as calcium, magnesium, sodium, potassium, chloride, sulfate, and bicarbonate dissolve into water as a result of water-rock interactions and also enter into the groundwater through recharge. Major-ion proportions can be represented by a Piper diagram that shows the relative proportions of cations and anions, with sample locations within the plotting regions indicating the dominant ion types (Hem, 1992). Water type can be assessed from the Piper diagram first by the location of the water sample on the lower triangles, where if a sample plots above 60 percent of a given ion on the triangle sides, then it would be predominantly that ion water type, and if it plots lower than 60 percent for any of the ions it is considered a mixture of water types. The left triangle represents positively charged cations and the right triangle represents negatively charged anions. The large central diamond shows the water samples projected up from their respective locations on the triangles to combine the cation and anion water types. Water type varies throughout the State (fig. 7) and is driven primarily by the rock types and associated minerals present along groundwater flow paths. Therefore, water type tends to be relatively consistent within local geographic areas with the same underlying geology, although water type can evolve along a flow path. Figure 7A includes sites located across New Mexico, except in Curry, Roosevelt, or Otero County (fig. 7B, C); sites in figure 7A are discussed next, mostly in order from north to south.

Drinking water in San Juan County is primarily sourced from surface water, so only one public supply well was available to be sampled and two observation wells in the County were also sampled. Sites 1 and 2 are located near a coal deposit and had sodium as the dominant cation; sulfate was the dominant anion for site 2, whereas the anions at site 1 were dominated by a mixture of bicarbonate and sulfate. Site 3 is located near the Animas River and had calcium-bicarbonate-sulfate type water. Groundwater sites 4 and 5 in Taos County had calcium as the dominant cation; site 4 had bicarbonate as the dominant anion, whereas site 5 had a mixture of bicarbonate and sulfate anions. Site 6 in Union County is screened in a young volcanic aquifer and had mixed cation and bicarbonate type water. Two sites in McKinley County (7 and 8) had sodium-bicarbonate-sulfate water, which is likely to represent old water that has undergone cation exchange (Beisner and Jones, 2020).

Groundwater from counties near central New Mexico (Bernalillo, Sandoval, and Santa Fe) generally had calcium and bicarbonate as dominant ions, but with some anomalies. Two sites (9 and 10) in Santa Fe County had calcium-bicarbonate type water. Sites 11 and 12 in Sandoval County had calcium-sodium-bicarbonate type waters. Sites 13, 14, and 15 in Bernalillo County had a mixture of calcium and sodium as the dominant cations and have bicarbonate as the dominant anion, with some chloride influence. Plummer and others (2012) found that groundwater in the Albuquerque area was generally sodium-bicarbonate type west of the Rio Grande and calcium-bicarbonate type east of the Rio Grande, with a narrow north-south trend of mixed-ion water type that may be related to faults parallel to the mountain front in the vicinity of site 13 in this study. South of the aforementioned Counties, two sites (16 and 17) in Socorro County had sodium-chloride type water and may represent interaction with deep basin groundwater or geothermal fluids (Anderholm, 1987). Sites 70–73 in Chaves and Eddy Counties had generally similar calcium-sulfate water types, although site 70 had more influence from sodium and chloride.

In Doña Ana County, site 140 had calcium-mixed anion water type, and site 141 had a higher proportion of sodium. Site 141 had results more similar to other groundwater sampled farther south in Doña Ana County at sites 144, 145, and 146 (with a higher proportion of sodium). Sites 142 and 143 in Doña Ana County had sodium-chloride type waters.

Groundwater sampling sites provide dense spatial coverage of parts of Curry and Roosevelt Counties in the High Plains aquifer and of Otero County in south-central New Mexico. Water samples were placed into groups on the basis of their geographic location within each county, which for Curry and Roosevelt Counties were divided up into even north and south, then east and west halves and sometimes given a designation of far (cardinal direction) for areas with high sample density. For Otero County, similar geographic areas were designated with additional subareas including central areas within the cardinal direction designations for areas with high sample density. Water types for samples collected in Curry and Roosevelt Counties ranged from calcium-bicarbonate to mixed cation-mixed anion, with one sample categorized as mixed cation-sulfate type (fig. 7B). Analytical results for samples collected in Curry and Roosevelt Counties by Langman and Ellis (2010a, b) suggest groundwater from the Ogallala Formation of the High Plains aquifer typically was a mixed cation-bicarbonate type, with groundwater present in the center of paleochannels having a higher proportion of bicarbonate and upland samples having a higher proportion of sulfate. The Dockum Group underlies the Ogallala Formation and generally had a sodium-chloride type water (Langman and Ellis, 2010a, b); however, groundwater from the Dockum Group was not sampled in this study, and there were no sodium-chloride type waters from samples in Curry and Roosevelt Counties in this study. Groundwater with higher TDS had a higher proportion of sulfate, with site 68 having the highest value (fig. 7B). Site 68 is located at the southern end of Roosevelt County and may represent an upland or other distinct water source compared with the other samples collected in Curry and Roosevelt Counties for this study (fig. 1).

Groundwater samples were collected from sites in the Sacramento Mountains on the east side of Otero County, as well as within the basin-fill sediments west of the mountain block (fig. 1). Generally, samples from the mountain-block area were collected from springs or shallow wells completed in bedrock and had a calcium-bicarbonate water type (fig. 8A). Samples collected from the mountain block had lower TDS (598 mg/L average) compared with wells screened in the basin-fill sediments west of the mountain block, which had generally higher TDS (1,552 mg/L average) and were more of a mixed cation-sulfate to mixed cation-mixed anion water type. There was one brackish water well in the northern area of Otero County (site 74) that had calcium-sulfate type water. The dominant geologic units of the Sacramento Mountains are the Permian San Andres Limestone (primarily limestone with some dolomite) and the Yeso Formation (sandstone, limestone, dolomite, and gypsum units).

Sites 124, 125, and 128 had samples collected in April 2021 (samples ending with “.1”), and October 2021 (samples ending with “.2”) (table 16). For the April 2021 samples, the water types differed among all three sites, with site 124 having a sodium-chloride type water. On the Piper diagram, the April 2021 sample from site 124 plots farthest from the compositions of groundwater from other sites in that area (112–119, 123, 126, 127, 129–132), with the April sample from site 128 plotting closest to those sample compositions (fig. 8). The October 2021 samples for sites 124, 125, and 128 were more similar to the general group of samples in that area, which have calcium-bicarbonate type waters.

Dissolved Organic Carbon

DOC was detected at low concentrations in 95 percent of the groundwater samples. Three sites had DOC above the highest blank concentration threshold of 3.63 mg/L: site 2 (21.9 mg/L), site 124 (April 2021, 8.83 mg/L), and site 99 (5.06 mg/L).

Nutrients

Nitrate, nitrite, ammonia, and orthophosphate were measured in groundwater samples. Ammonia was only detected at nine sites and ranged from 0.02 to 0.92 mg/L. Orthophosphate as phosphorus was detected in most groundwater sites and ranged from 0.004 to 0.066 mg/L at the 98 sites where it was detected.

Nitrate as nitrogen was less than the laboratory detection level at five sites, with detections ranging in concentration from 0.06 to 23.9 mg/L (site 53). Occurrence of nitrate concentrations greater than 5.0 mg/L as nitrogen may be due to the presence of human-related sources of nitrogen on the land surface, transport to the aquifer by natural and human-related recharge mechanisms, and (or) persistence in the aquifer as a result of favorable geochemical conditions for nitrate (Bexfield and others, 2011). Wells where nitrate as nitrogen was detected above 5 mg/L were located in the High Plains aquifer in Curry and Roosevelt Counties and were screened/completed either no more than 400 ft deep (sites 18, 25, 26, 37, 44, and 53) or were of unknown well depth (sites 27, 51, and 68). Two of these sites (51 and 53) exceeded the MCL of 10 mg/L.

Stable Isotopes

Stable isotopes of oxygen and hydrogen of the water molecule were measured for groundwater samples from this study. The ratio of the two stable isotopes gives an indication of recharge elevation, seasonality, and evaporation (Craig, 1961; Rozanski and others, 1993). The statewide sampling represents water from different regions and aquifers; samples within a region and aquifer can be compared with each other to understand local differences in recharge sources.

Stable isotopic ratios for samples collected across New Mexico ranged from −15.2 to −5.81 for δ¹⁸O and from −121 to −40.6 per mil for δ²H (fig. 9). Depleted (more negative) values generally represent higher elevation and (or) winter recharge, whereas enriched (less negative) values generally represent lower elevation and (or) summer recharge (Kendall and others, 1995). The most depleted values are located at sites in the northern part of the State (sites 1–8), as well as at sites near the Rio Grande (sites 11–15) (figs. 1 and 9A). The most enriched values, which also indicate an evaporative signature (Kendall and others, 1995), are located in the south-central (sites 70–147) and eastern (sites 18–69) parts of New Mexico (figs. 1 and 9).

Three groundwater samples from Albuquerque, the largest urban area of New Mexico (sites 13–15, fig. 1), have some of the most depleted isotopic values from the study (fig. 9A). Plummer and others (2012) conducted a detailed geochemical analysis of water in the Middle Rio Grande Basin in the 1990s, which included site 13 and wells within the same well field as sites 14 and 15. Generally, the most depleted stable isotopic values in Albuquerque were located in the center of the basin, attributed to recharge from the Rio Grande, which is sourced largely from precipitation at higher elevations in the mountains of southern Colorado and northern New Mexico; more enriched isotopic values were located closer to the mountain front. Sites 14 and 15 are located in the central area of the basin, and samples from these sites were more depleted than the sample from site 13, which is located at a higher elevation to the east towards the mountain front. Isotopic signatures at sites 14 and 15 were similar to those of samples collected in 1997 from Plummer and others (2012), being within 0.2 and 0.5 per mil for δ¹⁸O and δ²H, respectively. The site 13 value was more depleted compared to the 1997 value, differing by 0.74 and 5.9 per mil for δ¹⁸O and δ²H, respectively, and suggesting a change in the source of water for that well that may warrant further investigation.

Springs 124, 125, and 128 had PFAS detected in samples collected in April 2021 (table 16), and they plot to the right of the meteoric water line, suggesting an evaporative stable isotopic signature at that time (fig. 10). The springs were resampled in October 2021, when they did not have any PFAS detected and had stable isotopic signatures similar to those of other springs and wells sampled in the area during the April 2021 sampling event (sites 112–128; fig. 10). The samples with PFAS detections may indicate concentration in evaporated water that is localized and not representative of other groundwater in the area. The flow at springs 124, 125, and 128 was higher in October 2021 than in April 2021.

Groundwater Age

Groundwater ages discussed in this section, based on radiocarbon and tritium concentrations, represent an approximation of the length of time since precipitation entered the subsurface and interacted with carbonate minerals in the aquifer matrix.

Radiocarbon (¹⁴C) can be used to estimate the length of time since groundwater moved through the subsurface into the saturated part of an aquifer. The carbon species in the water sample are used to interpret the interaction with young carbon from gases in the soil zone, old carbon present as carbonate minerals in the aquifer matrix, and from interaction with organic carbon within the aquifer matrix such as oil and gas deposits. Some of the carbon species interactions occur in an open system condition while the water is moving through the unsaturated zone in contact with soil gas. Once the water enters the saturated zone, then carbon species interact under closed system conditions, and interactions in both open and closed systems influence the carbon species of the water.

The δ¹³C ratio varies as the water interacts with carbon sources in the soil zone during recharge and continues to do so as the water interacts with older solid carbonate sources in unconsolidated and bedrock materials (Han and Plummer, 2016). The δ¹³C ratio used for the soil and solid carbonate sources can influence the calculation of groundwater age. Solid carbonate δ¹³C values generally vary over a few per mil, whereas soil gas δ¹³C values vary over a larger range depending on the type of plant respiring CO₂ into the soil zone. Knowledge of the plant community or direct measurement of soil gas δ¹³C during groundwater recharge is needed to constrain the groundwater age calculation. Often with older groundwater, soil gas δ¹³C is assumed because the groundwater recharged prior to human measurement of soil gas in the recharge area. Table 18 presents a range of possible groundwater ages based on different soil gas values for waters that plot in a region below ¹⁴C of 50 pmc and below the zero-age area on figure 11 that indicates radiocarbon decay, where minimum is the younger age and maximum is the older age based on the given soil gas value.

The majority of groundwater samples had ¹⁴C values greater than 50 pmc, which precludes the groundwater age from being quantified because of radiocarbon decay (fig. 11; Han and others, 2012). For the samples with ¹⁴C values less than 50 pmc that also plot below the zero-age area line, the groundwater has undergone radiocarbon decay and an age can be estimated using the revised Fontes and Garnier solid exchange equations (Han and Plummer, 2013). Soil gas and solid carbonate ¹⁴C and δ¹³C values are used to calculate the zero-age lines on figure 11. For samples that plot between the zero-age lines, radiocarbon age cannot be quantified and may be explained by geochemical reaction with no radiocarbon decay. Samples that plot above the zero-age area are likely mixtures containing some old and young recharged water. The intersection of Tamers X (δ¹³C) and Y (¹⁴C) represent the starting isotopic composition of the exchange process (Han and Plummer, 2016).

Groundwater age estimates vary on the basis of the δ¹³C of the soil gas used in the revised Fontes and Garnier solid exchange equations, and a range of possible soil gas values (–11.5 to –21.4 per mil; Plummer and others, 2012) was used to represent the possible range of groundwater age, which could be refined for areas if soil gas in recharge areas is measured. Some sites only had a quantifiable groundwater age when using −11.5 per mil for the δ¹³C of soil gas, including sites 44, 70, 71, 72, 73, 83, 90, 86, 88, and 104, which are located in Curry, Chavez, Eddy, and Otero Counties. Additional investigation of the groundwater flow paths to these wells could help inform what factors influence radiocarbon interaction and establish a reasonable δ¹³C soil gas for the water recharging those aquifer systems. Some of the sites (1, 7, 8, 11, 15, 16, 89, 101, 103, 106, 142, 143, 144, 145, and 146) had groundwater age estimates greater than 10,000 years before present. Site 44 had detections of PFAS and only sample 44.1 had a low groundwater age estimate (64 years before present) when using −11.5 per mil for soil gas. Site 106 is located on a golf course and had a detection of PFOA, while also having low tritium below the laboratory reporting level and an old groundwater age (table 18). Irrigation using groundwater with an old radiocarbon age that reinfiltrated into the subsurface is a possible explanation for the presence of PFOA in old water at the site. The other groundwater sites from this study with quantifiable radiocarbon groundwater age had PFAS below the laboratory detection level.

Two groundwater samples from the Colorado Plateaus aquifers in McKinley County (sites 7 and 8) had the most depleted δ¹³C values from the dataset (−12.67 and −13.38 per mil, respectively) and low ¹⁴C (0.96 and 2.66 pmc, respectively) (fig. 11). The estimated groundwater ages for these samples ranged from 25,728 to 39,406 years before present (table 18). Other groundwater studies in the area have similarly low δ¹³C, suggesting interaction with fossil organic matter (Han and others, 2012), potentially related to naturally occurring hydrocarbon deposits (Dam, 1995).

Tritium is a useful indicator of water recharged following the aboveground nuclear testing of the 1950s (Lindsey and others, 2019). Semiannual (February and August) tritium concentrations from 1953 to 2012 for quadrangles covering the majority of New Mexico (latitude 31 through 35 degrees north and longitude −100 through −110 degrees west) from Michel and others (2018) and Jurgens (2018) are shown on figure 12. The premodern threshold for New Mexico is 0.35 pCi/L, which was determined as part of this study by assuming that the average of the 2008–12 precipitation data represents a background value for the area and then decaying that concentration from 1952 to the sample year for this study, which was primarily 2021, with a few samples collected in late 2020. The modern threshold was determined by finding the minimum post bomb-pulse tritium value decayed to 2021 for the aforementioned tritium data from precipitation. Modern thresholds determined for the two latitude-longitude quadrangles containing the most samples from the study were slightly different at 3.19 for latitude 33 to 35, longitude −100 to −105 (which includes sites 18–69, 74−76, and 98) and 3.30 pCi/L for latitude 31 to 33, longitude −105 to −110 (which includes sites 77−97, and 99−146) (fig. 12). Other quadrangles contained less samples and had similar modern thresholds including: 2.11 pCi/L for latitude 31 to 33, longitude −100 to −105 (including sites 70−73), 3.78 pCi/L for latitude 33 to 35, longitude −105 to −110 (including sites 16 and 17), 3.70 pCi/L for latitude 35 to 37, longitude −100 to −105 (including site 6), and 4.58 pCi/L for latitude 35 to 37, longitude −105 to −110 (including sites 1–5 and 7–15).

Tritium from this study ranged from less than the laboratory sample specific critical level to 17.56 pCi/L. The majority of the samples were above the laboratory sample specific critical level and ranged from 0.2 to 17.56 pCi/L. There were 41 samples in the premodern category (including one sample from site 11 that had an elevated censoring level of 0.58 pCi/L), 40 samples in the mixed category, and 48 samples in the modern category (table 18 and appendix 1).