Geochemical characterization of groundwater evolution south of Grand Canyon, Arizona (USA)

The purpose of this report is to document understanding of the geochemical evolution of groundwater discharging from the Redwall-Muav aquifer south of Grand Canyon. Establishing an understanding of the geochemical signature of the groundwater discharging south of Grand Canyon will provide baseline geochemical information, identify anomalous groundwater samples and their sources, and help elucidate what potential geochemical changes may occur due to breccia-pipe uranium mining. The scope of this report includes as assessment of 28 groundwater sites (25 springs and 3 wells) located south of the Grand Canyon spanning from Blue Spring in the eastern to Warm Spring in the western parts of the study area (Fig. 1). Targeted samples were collected for this study from 2016 to 2018 and the data were combined with data from previous studies of South-Rim groundwater chemistry by Monroe et al. 2005, Bills et al. 2007, and USGS 2019.

Major ion and trace element chemistry was used to characterize the groundwater and evaluate changes in groundwater chemistry as compared to historical sample collection. The dissolved ion chemistry was used to characterize water type and as input for statistical analysis that formally determined the similarities and differences between groundwater along the South Rim and helped develop a conceptual model of groundwater geochemical evolution. Groundwater recharge sources, proposed flow paths, and estimated mean ages (Solder et al. 2020; Solder and Beisner 2020; USGS 2019) were used in conjunction with dissolved ion and isotopic data in conceptual model development. Isotopic ratios of uranium and strontium provided additional evidence of flow paths through stratigraphic sections and the extent of water–rock interaction.

The study area is located in the Colorado Plateau physiographic province. The youngest rocks in the study area are Tertiary volcanic rocks primarily present to the south of Grand Canyon which are underlain by Mesozoic and Paleozoic sedimentary rocks. Groundwater samples in this study were from the Redwall-Muav aquifer (Mississippian to Cambrian in age), which is lower in the Paleozoic stratigraphic sequence. A less significant portion of groundwater is present in the perched Coconino aquifer (Permian in age) near the top of the Paleozoic sequence. Water is thought to recharge the system to the south of Grand Canyon near the San Francisco peaks (Crossey et al. 2009; Bills et al. 2007; Solder et al. 2020; Solder and Beisner 2020) and some additional smaller areas of localized recharge closer to the South Rim have been proposed (Knight and Pool 2016; Solder et al. 2020; Solder and Beisner 2020; Fig. 2). The extensive system of fractures, faults, and karst features in the area control general groundwater flow direction and speed, but the magnitude of influence and location of these features are not well constrained (Huntoon 1974; Jones et al. 2017).

The study of groundwater geochemistry south of Grand Canyon has a rich history. Metzger (1961) assessed the potential water supply in Grand Canyon National Park. Huntoon (1974, 1982) described surface water and groundwater flow of the North and South Rims of Grand Canyon and presented analysis of karst flow in the system. Foust and Hoppe (1985) and Goings (1985) established seasonal trends and baseline values for major and trace dissolved constituents in spring waters. Zukosky (1995) contributed stable isotopic (ẟ²H and ẟ¹⁸O) and rare earth element analysis. Fitzgerald (1996) contributed tritium and uranium isotope analysis to the spring data set. Ingraham et al. (2001) explored the possibility of using stable isotopes as a tracer for anthropogenic influence at Indian Garden Spring, and Liebe (2003) focused on uranium and uranium isotopes. Monroe et al. (2005) comprehensively explored the hydrogeochemistry of springs discharging along the South Rim of Grand Canyon, adding radiocarbon, strontium isotopes, and an extensive trace element suite. Crossey et al. (2006) utilized noble gas and radiocarbon analysis to understand contribution from fluids lower in the earth’s crust. Bills et al. (2007) built upon the Monroe et al. (2005) study by resampling springs over time. Tobin et al. (2018) provided a review of all hydrogeologic and geochemical studies conducted on groundwater in Grand Canyon.

Water samples collected for this study from two wells and 12 spring sites between September 2016 and May 2018, are described in detail in the following. The analytical results were combined with reported data from Red Canyon, JT, Miners, Monument, Lonetree, and Salt Creek Springs (published in Monroe et al. 2005), Fossil Canyon, Matkatamiba, Ruby, Sapphire, Serpentine, and Slate Springs (from Bills et al. 2007), and Warm Springs and Bar 4 Well (USGS 2019; Fig. 1). All samples were collected by the USGS using standard field methods (USGS 2006; 2018), and filtered samples were passed through a 0.45-μm capsule filter. Detailed field and analytical descriptions are published in Beisner et al. (2017).

Field parameters including pH, water temperature, specific conductance, dissolved oxygen, and barometric pressure were measured on site just before water samples were collected. Spring samples were collected as close to the point of issuance from the ground as possible using a peristaltic pump with flexible silicon tubing. Wells were purged of three casing volumes and parameters were within stability criteria prior to collection of water samples (USGS 2006).

Water samples were analyzed for major, trace, and rare-earth elements by the USGS National Research Program Laboratory in Boulder, Colorado (USGSTMCO) using methods from Garbarino and Taylor (1996), (2001), and Garbarino and Taylor (1979). The precision for all methods analyzed by the USGSTMCO was 4% or better depending on the element. Nitrate (NO₃) plus nitrite (NO₂) was analyzed by colorimetry at the USGS National Water Quality Laboratory (Patton and Kryskalla 2011).

Strontium isotope ratios (⁸⁷Sr/⁸⁶Sr) were measured by the USGS National Research Program Laboratory in Menlo Park, California, using methods described in Bullen et al. (1996) with precision of ±0.00002 or better at the 95% confidence level. ALS Environmental in Fort Collins, Colorado, analyzed water samples for uranium isotopes (²³⁴U, ²³⁵U, and ²³⁸U) using alpha-counting methods (ASTM D 3972) as a USGS contract laboratory.

As the majority of trace elements had one or more values below a laboratory reporting level, with several elements containing multiple reporting levels, specific statistical methods (Helsel 2012) for data with values below a censoring threshold (laboratory reporting level) were used to analyze the majority of analytes presented in this study. Detailed description of the statistical tests is given in Beisner et al. (2017). A p-value threshold of 0.05 (95% confidence level) was used to indicate statistical significance for all mentioned statistical tests.

Robust regression on order statistics (ROS) using the cenros function from the NADA software package in R (R Core Team 2018; Lee 2017) was applied to chemical analysis from spring samples for each element to calculate the median, mean, and standard deviation of the data. Groundwater in this study was compared to groundwater north of Grand Canyon (Beisner et al. 2017), using the cendiff function from the NADA package in R (Lee 2017; Helsel and Lee 2006).

Nonmetric multidimensional scaling (NMDS) was performed on the ranks of the uscores (Helsel 2012, 2016) using the metaMDS function from the vegan package in R (Oksanen et al. 2016) using Euclidean distance and two dimensions. Zero dissimilarities were handled by adding a small positive value and no autotransform since there are no community data (Helsel 2012). The following analytes were included in the NMDS: As, B, Ba, Li, Mo, Rb, Se, SO₄, Sr, U, V, and Zn; these were chosen based on interest for uranium mining in the region including groundwater north of Grand Canyon (from Beisner et al. 2017) and others that showed large variation across the study area (B, Ba, Rb, and SO₄). NMDS stress values ≤0.1 are considered fair with good ordination and no real prospect of misleading interpretation, values ≤0.05 indicate good fit, and values ≥ 0.2 are deemed suspect (Clarke et al. 2014).

The analytes used in the NMDS analysis were evaluated for correlation by calculating Kendall’s Tau using the cenken function from the NADA package in R (Lee 2017). A cluster analysis was used to identify similar groups in the spring samples by evaluating minimum differences within groups and maximum differences among groups using the hclust function for the analytes used in the NMDS analysis. The Calinski-Harabasz criterion was used to determine the number of clusters that maximizes the difference between clusters, while minimizing the differences within clusters with the cascadeKM function of the vegan package in R (Oksanen et al. 2016).

PHREEQCI version 3.5.0 (Charlton and Parkhurst 2002; Parkhurst and Appelo 2013) was used to determine saturation indices of calcite, dolomite, and gypsum. The WATEQ4F database was used with the default redox potential (pe) value of 4, since none of the water samples indicate reducing conditions.

Groundwater temperature ranged from 9 to 26.5 °C with a median temperature of 18.3 °C. pH ranged from 6.4 to 8.8 with a median of 7.4. Specific conductance ranged from 293 to 4,890 uS/cm with a median of 604 uS/cm. Dissolved oxygen ranged from 3 to 11.4 mg/L in the data set, except for Bar 4 Well which had the lowest dissolved oxygen value (0.7 mg/L), with a median dissolved oxygen of 7.2 mg/L.

The majority of sites can be classified as having calcium-magnesium-bicarbonate type water. Except for Blue Spring, which was dominated by sodium type water, the major cations were calcium and magnesium. For the major anions, the majority of sites were dominated by bicarbonate. Some springs were a mixture of bicarbonate-sulfate (Lonetree, Horn, Salt Creek, Slate, Sapphire, Serpentine, Royal Arch, and 140 Mile Springs). Sulfate groundwaters were generally located on the west side of the study area and include Ruby, Fossil Canyon, Matkatamiba, Bar 4 Well, and National Canyon Springs. Blue Spring was dominantly chloride type water (Fig. 4).

Nitrate ranged from <0.002 to 1.4 mg/L as N, with a median value of 0.51 mg/L as N. All sample concentrations of nitrate were below the US Environmental Protection Agency (USEPA) drinking water maximum contaminant level (MCL) of 10 mg/L (USEPA 2019) except for one elevated value at Monument Spring. This outlier was noted in Monroe et al. 2005 as possibly influenced by wastewater effluent from the South Rim of Grand Canyon or by exposure to geologic features not encountered by water discharging at other springs.

Some groundwater sample concentrations in this study exceeded the USEPA drinking water standards for As and U (Table 1; USEPA 2019). Red Canyon, Miners, JT, Havasu, and Warm Springs waters exceeded the USEPA drinking water standard for As of 10 μg/L (Fig. 5a), and Salt Creek Spring water exceeded the USEPA drinking water standard for U of 30 μg/L (Fig. 5b).

Median and mean trace element concentrations for all 27 groundwater sites south of Grand Canyon in this study (25 springs and 2 wells; Supai Well No. 3 not included as it is primarily supplied by water from Havasu Spring) are listed in Table 1. Some elements were not statistically quantified due to the following reasons: the majority of the sample concentrations were below the censoring threshold from blanks in Beisner et al. 2017 (Al, Be, Cd, Cu, Mn, Pb, Sb, and Zn), more than 80% of the sample concentrations were below the highest laboratory reporting level (Co and Fe), or more than 60% of the sample concentrations were below the laboratory reporting level (Ni).

Eleven sites (Blue, JT, Miners, Grapevine, Pipe, Canyon Mine Well, Horn, Salt Creek, Hermit, East Boucher, and National Canyon Springs) had three or more samples over time reported in Monroe et al. 2005, Bills et al. 2007, and USGS 2019. Change of uranium concentration over time at these sites was small, varying between 10 and 30%. There are not enough data to statistically analyze trends over time, but clear monotonic trends are not obvious. Monroe et al. (2005) sampled the majority of the springs presented in this report during different times of the year and found that there was temporal consistency of chemical results for individual springs. Generally, springs discharging south of Grand Canyon presented in this report show less temporal variability than springs north of Grand Canyon on the margins of the Kaibab Plateau that show rapid response from seasonal recharge pulses (Jones et al. 2017; Tobin et al. 2018).

The groundwater north of Grand Canyon is physically isolated from the groundwater to the south as the stratigraphic units are cut through by the Colorado River. Baseline and anomalous concentrations for groundwater north of Grand Canyon, assessed for 36 spring samples from the Toroweap Formation to the Redwall Limestone (Beisner et al. 2017), were compared to 27 groundwater samples collected south of Grand Canyon from this study, to assess regional similarities for select elements (As, B, Li, Se, Sr, Tl, U, V).

As and Sr were significantly different between the north and south groups (p <0.05). As values were higher and Sr values were lower in the group of springs south of Grand Canyon compared with the group of springs north of Grand Canyon. Elevated As concentrations may be related to the volcanic rocks present near the recharge source for water on the South Rim of Grand Canyon (Fig. 2), that are not a present in the presumed recharge area for springs north of Grand Canyon (Beisner et al. 2017). Strontium concentrations generally increase with sulfate concentration and may suggest that water in contact with gypsum along its flow path has elevated strontium. Groundwaters north of Grand Canyon include groundwater discharging from stratigraphic units including and above the Redwall Limestone, which show geochemical evidence of interaction with gypsum units in the Toroweap Formation (Beisner et al. 2017). The other elements (B, Li, Se, Tl, U, and V) were not statistically different between the north and south groups.

Five springs in this study (Pipe, Horn, Indian Gardens, National Canyon, and 140 Mile) discharge from alluvial material rather than directly from bedrock. Changes in trace element chemistry in Horn Creek as water discharges from bedrock and moves through alluvial material have been observed (Beisner and Tillman 2019). Water at the head of the eastern fork of the Horn Creek drainage emerges from bedrock with elevated uranium concentration (257 μg/L), low UAR (1.12), calcium-magnesium-sulfate type water, and low tritium (0.9 pCi/L), then infiltrates into the alluvium and emerges lower in the drainage with lower uranium concentration (23 μg/L), similar UAR (0.95), calcium-magnesium-bicarbonate type water, and elevated tritium (2.6 pCi/L; Beisner and Tillman 2019). Monroe et al. 2005 similarly discussed changes in chemical constituent concentration as groundwater discharging at a higher elevation in Cottonwood and Monument Creek drainages moves through alluvial material, noting increasing concentration of several major ions and decreases of several trace elements and nutrients. Unfortunately, blank and replicate samples were not assessed in the Monroe et al. 2005 study, which limits quantitative assessment of some of the elements from the chemical comparison along a drainage, but similar trends as those observed in Horn Creek by Beisner and Tillman (2019) suggest that additional chemical changes within the drainages are taking place. The mechanisms for the chemical changes within the drainages are not well understood and could vary depending on the local condition and possibly be influenced by mixing with another source of groundwater or geochemical changes as the water moves through unconsolidated material. However, the overall significance is that springs discharging from alluvial material may not have the same chemical signature of the bedrock groundwater, and characterizing secondary processes is critical for understanding Grand Canyon spring discharge chemistry.

Multivariate statistical analysis of groundwater south of Grand Canyon shows a good ordination distinction between sites based on their elemental assemblage (Fig. 8). Analytes associated near uranium in the NMDS were Mo, SO₄, Se, and Sr (Fig. 8). These same analytes were related to U with a significant positive Kendall’s Tau value (Fig. 9). Uranium had a significantly negative Kendall’s Tau relationship with Ba and As (indicating that higher values of uranium are associated with lower values of Ba and As), which are located across the NMDS plot from uranium (Fig. 8).

A cluster analysis on the data used in the NMDS analysis using the Calinski-Harabasz criterion indicates that there are three distinct groups of groundwater sites (Fig. 10, and dashed outlines on Fig. 8). Blue, Havasu, Fern, and Warm Springs are in one group and have higher values of B, Li, Rb, and V and lower values of Mo and Se compared to other sites. Lonetree, Horn, Salt Creek, Slate, Sapphire, Ruby, Serpentine, Royal Arch, Fossil Canyon, 140 Mile, Matkatamiba, Bar 4 Well, and National Canyon Spring form another group and are distinguished by higher values of Mo, Se, SO₄, and Sr and lower values of Ba. Red Canyon, JT, Miners, Grapevine, Pipe, Indian Gardens, Monument, Hermit, East Boucher Springs and Canyon Mine Well form another group and are distinguished by lower values of B, Li, SO₄, and Sr. The groundwater site groupings appear to overlap spatially from east to west with clustering of sites locally (Fig. 10).

Multiple geochemical tracers were used to understand recharge source, groundwater residence time, and water–rock interaction of groundwaters analyzed in this study. A companion paper (Solder et al. 2020) and USGS (2019) present environmental-age tracer analysis for groundwaters presented in this report. Groundwater-age tracer analysis is combined with geochemical information and discussed in the following paragraphs. Detailed geochemical modeling was not performed for this study, as there were not enough samples along a flow path from recharge source to discharge available for the study area.

Using the categorical age classification for tritium of <1.3 pCi/L for premodern water, >12.8 pCi/L for modern water, and values between for a mixture of premodern and modern water from Beisner et al. (2017), no samples contained primarily modern water (water recharged after 1952) and the majority of the sites had a mixture of modern and premodern water and some sites were predominantly premodern water (Blue, Havasu, Supai Well No. 3, Fern, Boucher East, Canyon Mine Well, Hermit, Pipe, Indian Gardens Spring, and Bar 4 Well; Bills et al. 2007, Monroe et al. 2005, Solder et al. 2020). Tritium values were not available for Fossil Canyon, Ruby, Sapphire, Slate and Warm Springs (Bills et al. 2007). Spring sites with sulfate or sulfate/ bicarbonate water had tritium in the range of a mixture of modern and premodern water and may indicate a component of water that potentially moved through the subsurface gypsum deposits along a shorter flow path. Bar 4 Well did have sulfate water type but no measurable tritium (Bills et al. 2007; USGS 2019).

Solder et al. (2020) calculated mean age for the springs sampled between 2016 and 2018 for this study as well as for groundwater data published in Monroe et al. 2005 and Bills et al. 2007. Sites classified as premodern water had mean ages ranging from approximately 1,050 to 19,200 years. Of the premodern samples, Pipe Spring had the lowest mean age of 1,050 years and the next lowest age was Indian Gardens Spring with 2,750 years. Tritium content was 3.5 pCi/L in Pipe Spring samples collected in 2000 (Monroe et al. 2005), which would indicate a mixture of modern and premodern water and may explain the low mean age compared with other springs with low tritium values. Waters with tritium concentrations indicating a mixture of modern and premodern water had lower mean age ranging from 90 to 2,900 years. Monument Spring had the highest mean age for the mixed waters of 2,900 years (Monument Spring tritium value was 1.3 pCi/L, which is right at the threshold for premodern water) and the next highest age was Royal Arch Spring with a mean age of 2,400 years.

Dedolomitization may be occurring in some groundwater south of Grand Canyon, and may affect the radiocarbon age calculation (Plummer et al. 1990; Fitzgerald 1996). The calculated saturation index of gypsum was negative for all samples in this study, but values show a logarithmic increase to near saturation with increasing sulfate concentration (Fig. 11). Samples range from Grapevine Spring with 6 mg/L sulfate and SI of gypsum of –3.03 to National Canyon Spring with 991 mg/L sulfate and SI of gypsum of –0.42 (Fig. 11). Five sites (National Canyon, Matkatamiba, Ruby, Fossil Springs, and Bar 4 Well) had high sulfate concentration (>600 mg/L), saturation indices (SI) of gypsum less than 0.7, calcite SI plus or minus 0.25, and dolomite SI plus or minus 0.5, which may indicate dedolomitization

Four springs (Blue, Havasu, Fern, and Warm Spring) had unique chemistry and may represent a contribution of groundwater from deeper in the earth’s crust (Crossey et al. 2006) or long groundwater residence times. These sites are located at discrete locations in the study area near major structural geologic features that may allow for focused groundwater discharge and may also provide a pathway for interaction with deep crustal materials. Characteristics of these waters are premodern tritium values, radiocarbon ages on the order of 4,000–16,000 years (Solder et al. 2020), elevated As, B, Li, Rb, and V concentration, calcium-magnesium-bicarbonate type water (except for Blue Spring with sodium-chloride type water), and intermediate UAR (between 2 and 3)—see Table S1 of the electronic supplementary material (ESM). Strontium isotopic ratio was intermediate for these springs (Table S1 of the ESM), suggesting groundwater chemistry is largely controlled by interaction with Paleozoic carbonate rocks rather than Precambrian basement rocks with more radiogenic strontium isotopic ratios (Crossey et al. 2006).

Groundwaters with a dominant sulfate type water (Slate, Sapphire, Ruby, Fossil Canyon, Matkatamiba, Bar 4 Well, and National Canyon Spring) are located on the western side of the study area (orange group on multivariate analysis on Fig. 10). Gypsum deposits grade thicker from east to west in the study area (Billingsley et al. 2008; P. Huntoon, University of Wyoming, personal communication, 2019) and may explain the predominance of sulfate type water of samples in the west part of the study area (Fig. 4). Three of these sites have strontium isotopic data (National Canyon, Matkatamiba Springs, and Bar 4 Well), which indicate high strontium concentration and low strontium isotopic ratio that may indicate interaction with gypsum (CaSO₄·2H₂O) deposits in the Kaibab and or Toroweap formations (Fig. 7). Ruby and Fossil Canyon Springs have elevated strontium concentrations (1,900 and 2,300 μg/L, respectively; Table S1 of the ESM), and may also have low strontium isotopic ratio which warrants follow up sampling. Matkatamiba and National Canyon Springs have tritium indicating a mixture of modern and premodern water (Table S1 of the ESM; Solder et al. 2020). National Canyon had a young mean age of 87 years and Matkatamiba had an intermediate age of 1,200 years. Bar 4 Well had no measurable tritium and an old radiocarbon age of 19,100 years, although the age may be less if dedolomitization is occurring along the flow path before the Bar 4 Well (Solder et al. 2020).

Other sites with a mixture of bicarbonate and sulfate type waters (Lonetree, Horn, Salt Creek, Serpentine, Royal Arch, and 140 Mile Springs; Fig. 4) are also part of the orange multivariate group (Fig. 10) and have tritium values indicating a mixture of modern and premodern water, low UAR (no data for Lonetree and Serpentine), and high strontium isotopic ratios (except Royal Arch Spring; Figs. 6 and 7). Lonetree, Horn, Salt Creek, and 140 Mile Spring water samples had low radiocarbon ages ranging from 175 to 1,000 years. Serpentine and Royal Arch Springs waters had older radiocarbon ages ranging from 2,000 to 2,400 years (Solder et al. 2020). Horn and 140 Mile Springs waters had the lowest UAR values from the study (Fig. 6) and had low radiocarbon ages of 360 and 175 years respectively (Solder et al. 2020).

The other springs in the study area (Red Canyon, JT, Miners, Grapevine, Pipe, Canyon Mine Well, Indian Gardens, Monument, Hermit, and East Boucher) had calcium-magnesium-bicarbonate type waters (green group from multivariate analysis in Fig. 10). The spring waters from this group on the east side of the study area (Red Canyon, JT, Miners and Grapevine Springs) have tritium indicating a mixture of modern and premodern waters and radiocarbon ages ranging from 345 to 1,500 years (Solder et al. 2020), high strontium isotopic ratio, high As and low U concentration (Table S1 of the ESM). Other lines of evidence, including groundwater flow modeling (Knight and Pool 2016) and isotopes of hydrogen and oxygen in water (Solder and Beisner 2020), suggest water present at the aforementioned springs has a short flow path and may have a more local source of recharge compared to the regional system. The groundwater from sites located to the west of the aforementioned springs generally have tritium indicating primarily premodern water and radiocarbon ages ranging from 1,050 to 12,000 years (except for Pipe and Monument spring as discussed at the beginning of this section; Solder et al. 2020).

Generally, the majority of water in this study area recharges the subsurface near the San Francisco Peaks and migrates down through permeable structural features to the Redwall-Muav aquifer (Fig. 2), consistent with previous conceptual models of groundwater flow (e.g., Hart et al. 2002; Monroe et al. 2005, Bills et al. 2007). Multiple independent geochemical constituents, as reported in this study and two companion papers (Solder et al. 2020; Solder and Beisner 2020), show additional sources of local recharge to the subsurface located north of the San Francisco Peaks along the Coconino Plateau. Locally, groundwater migrates from upper stratigraphic units containing gypsum to the Redwall-Muav aquifer through permeable structural features. In some parts of the study area, large-scale structural features such as faults, facilitate transport of older groundwater from the deep subsurface. The exchange between the shallow meteoric and deep subsurface groundwater flow systems along large structural features is apparent in the tracer chemistry (Solder et al. 2020; Solder and Beisner 2020), trace element chemistry (Fig. 11), and abundant dissolved carbon dioxide as reported by Crossey et al. (2006) and observed as over saturation and deposition of carbonate mineral species. Conversely, small-scale groundwater interaction with mineralized breccia pipes in the region is not well understood. The observed chemical anomalies that occur around known mineralized breccia pipes are the subject of further research.