Purpose and Scope

The purpose of this report is to describe the basic groundwater chemistry of the Coconino aquifer along the Interstate-40 corridor between Flagstaff, Ariz., and the area just east of Holbrook, Arizona. Specifically, major ion, trace metal, and TDS concentrations are presented and analyzed to identify differing groundwater chemistry in association with changes in geology and groundwater movement, and to compare groundwater chemistry to drinking-water standards to determine suitable areas for potential groundwater-resource development.

Previous Investigations

The hydrogeology and chemistry of the Coconino aquifer in the study area have been described in several previous studies. Darton (1910) compiled some of the first geologic data from the area between Kingman, Arizona, and Albuquerque, New Mexico, to explore groundwater prospects for the Atchison, Topeka, and Santa Fe Railway. Gregory (1916) described the geography, climate, surface water, and groundwater of the Navajo Nation and Hopi Reservations; the hydrogeology in this area was later expanded on by Cooley and others (1969). Harrell and Eckel (1939) presented a comprehensive groundwater study of the Holbrook area, including chemical analyses from 118 wells and springs. Bills and Flynn (2002) and Bills and others (2007) summarized the hydrogeology of the Coconino Plateau. Hart and others (2002) compiled existing Coconino aquifer data from the Little Colorado River Basin to produce a generalized groundwater budget. Hoffmann and others (2006) presented geological, hydrological, and chemical data from the Coconino aquifer near Leupp, Arizona, and Jones and Robinson (2021) presented groundwater levels and basic chemistry of the Coconino aquifer in northeastern Arizona.

Evaporites in the study area also have been explored. Bahr (1962) described evaporite karst features on the Holbrook Anticline. Mann (1976) characterized Coconino aquifer water in southern Navajo County and produced an early delineation of the extent of salt beds. Neal and others (1998, 2013) and Rauzi (2000) described evaporite karst in the Holbrook sedimentary basin. Neal and Colpitts (1997) and Neal and Johnson (2002) described specific Holbrook Basin karst expressions (Richard Lake and McCauley Sinks, respectively).

Geology

The Coconino aquifer is named after the primary water-bearing rock unit within the aquifer, the Coconino Sandstone, but the saturated and hydraulically connected parts of the Kaibab Formation, the Toroweap Formation, the Schnebly Hill Formation, and the upper and middle part of the Supai Formation also constitute part of the Coconino aquifer in the study area (fig. 2; Bills and others, 2000; Bills and Flynn, 2002; Hart and others, 2002; Bills and others, 2007).

The Supai Formation ranges in age from Pennsylvanian to Permian and consists of red siltstone and sandstone (Irwin and others, 1971; Blakey, 1990). Divided into three parts, only the upper and middle parts of the Supai Formation are hydraulically connected to the Coconino aquifer; the lower part of the Supai Formation acts as a confining layer for underlying groundwater in the Redwall-Muav aquifer (Bills and others, 2000).

The Hermit Formation overlies the Supai Formation in some areas of northern Arizona. Consisting of red-brown siltstone and sandstone, the Hermit Formation is lithologically similar to the Supai Formation. In much of the study area, the boundary becomes unclear and the Hermit and Supai Formations are indistinguishable. For this reason, the Hermit Formation is usually omitted from stratigraphic columns in this area (Irwin and others, 1971; Blakey, 1990; Bills and others, 2000).

The Permian Schnebly Hill Formation is an important Coconino-aquifer component in the Holbrook Basin, with a thickness as much as 1,700 ft (fig. 2; Blakey, 1990; Bills and others, 2000). The Schnebly Hill Formation is reddish brown to reddish orange and comprises sandstone, mudstone, limestone, and evaporites (Blakey, 1990). East of Holbrook, the Corduroy Member of the Schnebly Hill Formation contains halite and other evaporites of early Permian age as much as 650 ft thick (Blakey, 1990; Conway and Cook, 2013). When present, the Schnebly Hill Formation intertongues with the overlying Coconino Sandstone (Bills and others, 2000; Hoffmann and others, 2006).

The Permian Coconino Sandstone is typically the main water-bearing unit of the aquifer (Hart and others, 2002). The Coconino Sandstone is a tan to white, crossbedded, quartz sandstone of eolian origin (Darton, 1910; Blakey, 1990). In Leupp, geologic logs indicate thicknesses from 300 to 960 ft (Hoffmann and others, 2006). Near Winslow, Coconino Sandstone thickness is about 800 ft (Mann, 1976).

The Permian Toroweap Formation is only known to be present in the western part of the study area. Bills and others (2000; p. 26) describe the formation as beds of “carbonate sandstone, red beds, silty sandstone, siltstone, limestone, and thin layers of gypsum.” The formation is often indistinguishable from Coconino Sandstone, but according to Sorauf and Billingsley (1991), a distinction between the Toroweap Formation and the white, quartz sandstone of the Coconino Sandstone can be observed near Flagstaff to the west of the study area; where indistinct, the Toroweap Formation is often considered to be part of the Coconino Sandstone (Bills and others, 2000).

The Permian Kaibab Formation is often expressed as a light-gray limestone from 10 to 650 ft thick. Sinkholes and depressions formed by dissolution are present on the surface, as well as fractures formed by jointing and faulting in the subsurface (Irwin and others, 1971; Bills and others, 2000).

In parts of the study area where the Coconino aquifer is not exposed, the red to reddish-brown Triassic Moenkopi Formation overlies the aquifer (Mann, 1976). Consisting largely of mudstone and siltstone, the Moenkopi Formation acts as a confining layer when not heavily fractured. In some areas, the Moenkopi Formation can supply small amounts of water to wells (Cooley and others, 1969; Bills and others, 2000). North of the study area, the shallower Navajo, Dakota, and Toreva aquifers are present and often used for water supply (Mason, 2021).

Evaporites

Beds of halite underlie about 3,500 mi² in the southeastern part of the study area, with a maximum thickness near the center of an aggregate of 655 ft of salt in 1,500 ft of Schnebly Hill Formation strata (fig. 3). Close to the depositional center of the halite, a zone of potash covers about 600 mi². The potash, consisting of sylvite, carnallite, and polyhalite, is nearly 40 ft thick and overlies the halite (Rauzi, 2000). To the south and southeast, halite transitions into gypsum and anhydrite and extends farther than the halite (Rauzi, 2000, pl. 2).

Dissolution of evaporite beds by the movement of Coconino aquifer groundwater has led to numerous solution-collapse features in the study area. Solution-collapse features in evaporite rocks are developed similarly to those in limestone, but the time scale is shorter. Evaporites such as halite and gypsum can form karst features in a matter of days to years due to their high solubility. Evaporite karst features form near the outer edges of a salt deposit (Johnson, 1997). In the study area, the dissolution front is currently migrating to the northeast, and karst features are forming in real time (Bahr, 1962; Johnson, 1997; Neal and others, 1998).

The Holbrook Anticline is present near the southwestern extent of halite, and the axis can be mapped at the surface for more than 60 miles (mi; fig. 3). The northern flank follows a regional dip of about 2 degrees. On the southern side, the regional dip is interrupted and the average dip is about 15 degrees, although some dips can be steeper. Numerous karst sinks are present on the southern flank (Bahr, 1962). More than 500 sinkholes, joints, compression ridges, and other solution-collapse features have been identified along the Holbrook Anticline and the parallel Dry Lake Syncline to the immediate southwest (Mann, 1976; Conway and Cook, 2013). The Holbrook Anticline is not expressed below the salt layer, which may suggest that dissolution is a factor of its formation (Neal and others, 1998).

Just west of the Holbrook Anticline, near the western limits of evaporites of the Schnebly Hill Formation, McCauley Sinks provide a conspicuous karst surface expression (fig. 4). McCauley Sinks include about 50 sinkholes up to 50 meters (m) deep and 100 m in diameter. They appear in three semi-circular “rings” within a 3-kilometer (km) wide depression (Neal and Johnson, 2002). Along with several other, smaller depressions west of the Holbrook Anticline, these structures are related to the dissolution front of the halite and appear similar to breccia pipes on the Colorado Plateau (Neal and Johnson, 2002). However, where these other breccia pipes originate in the Mississippian Redwall Limestone and (or) the Cambrian Muav Limestone of the Tonto Group, the McCauley Sinks and Richard Lake likely originate due to collapse following salt dissolution in the Schnebly Hill Formation. Similar to other karst features in the area, pressure ridges following the general trend of the Holbrook Anticline are present near both structures (Neal and Johnson, 2002). To the southeast of the study location is an area known as “The Sinks,” which includes more than 250 sinkholes, joint fissures, and other collapse features also related to halite dissolution (Neal and others, 1998).

Groundwater Movement

Groundwater in the Coconino aquifer generally moves northward, parallel to the regional dip of the strata (Mann, 1976; Hart and others, 2002). Most recharge occurs as snowmelt near the Mogollon Rim to the south; rain events are often flashy and contribute to runoff (Mann, 1976). The Coconino aquifer is unconfined in most of the study area. To the north, the Moenkopi Formation creates confined conditions. The age of groundwater in the Coconino aquifer around Flagstaff in the eastern part of the study area has been estimated as modern to about 7,000 years (Bills and others, 2000).

Well yields from Coconino aquifer wells inventoried by previous studies varied substantially in the study area, from a few gallons per minute to about 2,800 gallons per minute (Mann, 1976; Bills and others, 2000; Hoffmann and others, 2006). Although several factors affect well yields, including pump design and formation lithology, Bills and others (2000) suggested that the greatest effect on Coconino aquifer well efficiency probably is due to proximity to faults and fractures.

Jones and Robinson (2021) discussed wells monitored as part of the USGS C-Aquifer Monitoring Program between Flagstaff and Holbrook, Arizona. They found that measured groundwater levels fluctuate seasonally, and suggested that infiltration from surface water from summer monsoon events and spring snowmelt have the potential to influence wells, as does higher rates of pumping in the summer months. Although some monitored wells have shown little change in groundwater levels (for example, USGS site number 351023111062002, near Leupp), others have shown decreasing water-level trends (for example, USGS site number 345023110111401, south of Holbrook, has decreased about 11 ft from 1969 to 2018; Jones and Robinson, 2021; U.S. Geological Survey, 2023).

Data Compilation

The USGS NWIS database was queried to find existing groundwater sites (wells and springs) that had water-chemistry data associated with them. Those groundwater sites with wells screened-in or springs discharging from the Coconino aquifer and having major ion and (or) TDS data were selected for inclusion in this study.

A total of 130 sites with samples dating from 1933 to 2008 were identified (fig. 1; table 1; U.S. Geological Survey, 2023). These sites were generally in proximity to the Little Colorado River and Interstate-40. Few wells are drilled into the Coconino aquifer in Hopi Tribal Lands or Navajo Nation north of Interstate-40. Some wells may be screened in multiple formations, and it is not always clear which unit(s) the well is producing from. The aquifer coded in NWIS is considered to be the producing unit for this study.

Most wells used (118) were screened in the Coconino Sandstone (fig. 1). Additionally, four of the spring sites discharge from the Coconino Sandstone where it is exposed in canyon walls along Clear and Chevelon Creeks. Three wells and one spring are sourced by the Kaibab Formation in the western part of the study area. Four wells are screened in the Supai Formation. No wells or springs sourced from the Hermit, Toroweap, or Schnebly Hill Formations were present in NWIS in the study area.

Numerous study sites have been sampled multiple times. When computing the median values from all sites for the parameters of pH, specific conductance, and total dissolved solids, the most recent values from each site were used. In three cases, the date when the most recent sample was collected had two samples collected; in those cases, the average value of the two samples was used in the statistical analysis. Additionally, there were 23 samples with estimated results for TDS that were used in the statistical analysis. When computing the water type for sites with multiple samples the most recent sample collected containing all the constituents necessary to compute water type was used.

Quality Assurance

Water-chemistry samples were analyzed using methods described in Fishman and Friedman (1989), Fishman (1993), and Fishman and others (1994). Major ion data included dissolved calcium, magnesium, sodium, potassium, chloride, sulfate, and bicarbonate (computed from alkalinity or acid-neutralizing capacity). When potassium was not measured, sodium was used by itself in the ion balance and analyses; potassium concentrations are considered minor. In order to validate the dissolved-ion data, the ion balance of samples was checked by converting the concentrations of cations and anions in the sample from milligrams per liter to milliequivalents per liter for comparison. Theoretically, if all ions have been correctly determined, the total milliequivalents per liter of cations should equal the total milliequivalents per liter of anions in a sample (Hem, 1985). Most samples had differences of less than 5 percent between cation and anion concentrations expressed as milliequivalents per liter. Four samples had ion balances with differences from 6 to 12 percent; these samples are included in this study because the percent differences are low, and other unmeasured ions and trace metals may potentially contribute to the ion balances (Hem, 1985). Potassium was not measured in all water samples which undoubtably affected the ion balance of samples where it was missing. However, because potassium is usually a minor constituent of natural waters the omission was considered acceptable. The TDS of groundwater was analyzed using the sum of constituents method (Fishman and Friedman, 1989). Specific conductance, or the ability of a solution to conduct an electric current, is a function of the concentration and charge of the ions (Hem, 1985; Fishman and Friedman, 1989). Specific conductance and TDS from samples used in this study showed a strong relationship as should be expected with an R² value of 0.98.

Graphical and Statistical Analysis

The geochemical compositions of water-chemistry samples were graphically depicted with stiff and trilinear diagrams (similar to Piper [1944]). Analyses were performed using R statistical software (v.4.2.2; R Core Team, 2022). Water-chemistry data were downloaded from NWIS using the dataRetrieval package (De Cicco and others, 2022), and stiff diagrams and piper diagrams were created using the smwrGraphs package (Lorenz and Diekoff, 2017). Prior to plotting, concentration data, in milligrams per liter, were transformed to milliequivalents per liter. TDS were plotted in ArcMap (v. 10.8.1, Esri, Redlands, California) and interpolated using the “spline with barriers” method.

In addition to the graphical methods described above, principal component analysis (PCA) on the major ion data was performed in Primer 7 (Clarke and others, 2014; v7.0.17, PRIMER-E Ltd., Plymouth, United Kingdom) to investigate associations in the data. PCA was conducted on transformed and normalized major ion data to better understand how the selected factors explained the observed variation among sites (Clarke and others, 2014). Briefly, PCA captures as much of the variability in the original multi-dimensional space as possible within the two axes of the plane. Output from the PCA includes eigenvalues (variances of each principal component axis), eigenvectors (coefficients for the linear combination of input factors that defines the plane), and principal component scores (coordinates of the samples on the PC axes computed using eigenvector coefficients). A percent variance explained (from eigenvalues) is computed to quantify the extent to which the two principal component axes of the plane provide an accurate representation of the true association between the factors in the original multi-dimensional space.

Potential for Use as Potable Water

High concentrations of TDS in much of the study area affect the quality of Coconino aquifer water for potential potable use. In the southeastern part of the study area, and north of the Little Colorado River, about 73 percent of Coconino aquifer samples contain TDS greater than the SMCL of 500 milligrams per liter (mg/L) up to concentrations greater than 10,000 mg/L. Although this falls into the TDS range that can be potentially remediated, desalination of groundwater for potable use can be costly and energy intensive (Stanton and others, 2017).

Trace metals have not been widely measured, but most concentrations are less than the MCLs for drinking water. Fluoride exceeded the MCL in one sample (table 3). Although both arsenic and uranium samples were less than the MCLs (10 µg/L As and 30 µg/L U), these elements have been a concern for the Navajo Nation and Hopi Tribe. Jones and others (2020) found that both arsenic and uranium exceeded the EPA MCL in western Navajo Nation in unregulated water sources, including around Leupp, Arizona.

Water containing elevated TDS can be used for livestock watering and (or) irrigation in some cases. Irrigation water with specific conductance values ranging from 750 to 1,500 µS/cm may have detrimental effects on sensitive crops, whereas higher specific conductance values may affect many crops (Zaman and others, 2018). Sodium hazard, which describes how sodium affects the soil, and ion toxicity are other potential hurdles.

Less is known about Coconino aquifer water north of the study area on the Hopi Reservation and Navajo Nation. Wells in these areas penetrate the shallower Navajo, Dakota, and Toreva aquifers (Mason, 2021). The Hopi Tribe did drill a single municipal well into the Coconino aquifer at the Village of Moenkopi north of the study area. Water from that well required reverse osmosis treatment demonstrating that salinity concentrations are elevated in that area (Jon Mason, oral commun., 2023). However, to the northeast of the study area, near Arizona’s border with New Mexico, Coconino aquifer water contains less dissolved solids (less than 500 mg/L; U.S. Geological Survey, 2023).