Geochemical evolution of groundwater in the Mud Lake area, Eastern Idaho, USA

Surface water has variable water temperature, dissolved oxygen (DO), and specific conductance (SpC) (Table 1) because of seasonal changes in air temperature, the dependence of oxygen solubility on water temperature, and evaporation. pH was 7.8 for Camas Creek, 8.0 for Medicine Lodge Creek, and 7.0 for Mud Lake. Nitrate concentrations were ≤0.25 mg/L as N (Table 3). Camas Creek had a tritium concentration of 26.6 ± 2.1 pCi/L and a δ¹³C value of −7.98 ‰ (Table 4).

Thermal springs in the mountains (Lidy Hot Spring and Warm Spring, sites 57 and 58) were hot (48.4 and 27.8 °C), had low pH (7.2 and 7.1), and had SpC of 694 and 430 µS/cm at 25 °C. Lidy Hot Spring had large concentrations of calcium (87 mg/L), potassium (13.6 mg/L), sulfate (191 mg/L), fluoride (7.0 mg/L), lithium (48 μg/L), and strontium (1,000 μg/L) and a measurable concentration of ammonia (0.11 mg/L as N) (Tables 2, 3). Warm Spring had much smaller concentrations of these chemical species than Lidy Hot Spring, but had higher concentrations of magnesium (19.7 mg/L) and bicarbonate (198 mg/L). Tritium concentrations were 0.4 ± 0.3 and 3.1 ± 2.1 pCi/L, and δ¹³C values were −3.9 and −5.55 ‰. Deep geothermal water from the ESRP (site 100) was hot (57 °C), had a pH of 7.9, was brackish (2,870 µS/cm at 25 °C), was presumed to be anoxic, and had small concentrations of calcium (7.5 mg/L) and magnesium (0.5 mg/L) and large concentrations of sodium (390 mg/L), bicarbonate (900 mg/L), sulfate (99 mg/L), fluoride (13 mg/L), boron (560 μg/L), and iron (1,100 μg/L).

Groundwater from wells (sites 3, 32, and 35) and cold springs (sites 55, 56, and 59) in the mountains had temperatures ranging from 5.8 to 12.8 °C, pH ranging from 6.9 to 7.8, SpC ranging from 258 to 523 µS/cm at 25 °C, were anoxic to slightly undersaturated with oxygen (1.1–94.4 % saturation), and had moderate-to-large carbon dioxide (CO₂) concentrations (logPCO₂ of −2.61 to −1.58) and nitrate concentrations of ≤0.88 mg/L as N. Tritium concentrations ranged from −0.5 ± 1.9 to 33.1 ± 2.5 pCi/L and δ¹³C values ranged from −12.72 to −8.18 ‰.

Groundwater from the ESRP aquifer (32 sites excluding site 100) exhibited a wide range in chemistry (Tables 1, 2, 3, 4). Temperature, pH, and SpC ranged from 9.0 to 17.7 °C, 7.0 to 8.5, and 249 to 929 µS/cm at 25 °C, respectively. DO ranged from anoxic to supersaturated (<2.3–122.5 % saturation) and logPCO₂ ranged from −3.25 to −1.72. Cation and anion concentrations (in mg/L) ranged from 15 to 109 for calcium, 4.6 to 33.0 for magnesium, 9.0 to 85.0 for sodium, 2.0 to 7.2 for potassium, 131 to 457 for bicarbonate, 5.6 to 121 for chloride, 5.3 to 91.3 for sulfate, 0.15 to 1.01 for fluoride, and <0.05 to 13.2 mg/L as N for nitrate. Ammonia concentrations >0.03 mg/L as N were measured in water from a few sites (sites 18, 25, 27, and 29), and large lithium concentrations were measured at sites 25 (71 μg/L) and 29 (47 μg/L). Tritium ranged from −22 ± 13 to 96 ± 13 pCi/L and δ¹³C values ranged from −15 to −11.2 ‰. No trends in the spatial distribution of physical and chemical parameters were apparent, although larger SpC and ion concentrations were generally from groundwater in the southwestern part of the study area (Fig. 5).

The hydrochemical facies (Fig. 6) of water was calcium bicarbonate for surface water, most groundwater from the mountains, and 19 of 32 groundwater samples from the ESRP. The hydrochemical facies of other groundwater from the ESRP were various combinations of calcium-, sodium-, and magnesium-bicarbonates and calcium chloride bicarbonate (site 33). Thermal water was calcium sulfate (site 57), calcium bicarbonate (site 58), and sodium bicarbonate (site 100).

Most δ²H and δ¹⁸O values (Fig. 7) of groundwater in the study area (plus values for groundwater from the Centennial Mountains; these values are included in Fig. 7 because much of the groundwater in the study area originates from there) plot near and approximately parallel to the local meteoric water line for winter (Benjamin et al. 2004) indicating that most groundwater is of meteoric origin and from winter precipitation. All groundwater from the ESRP plots along a trend line (determined from linear regression of Mud Lake δ²H and δ¹⁸O values) for evaporation of Mud Lake water, but only sites 19, 22, and 23 (southwest of Mud Lake; Fig. 4) have large δ²H and δ¹⁸O values that indicate evaporated water was a significant source of recharge.

δ¹³C values in groundwater are influenced by the δ¹³C values of recharge water that equilibrates with unsaturated zone CO₂ (with a δ¹³C ≈ −24 to −30 ‰), fractionation of CO₂ as CO₂ gas in the unsaturated zone dissolves in groundwater (δ¹³C in bicarbonate enriched by ≈9 ‰), dissolution of carbonates (carbonate δ¹³C ≈ 0 ‰), and decay of organic matter (organic matter δ¹³C ≈ −24 to −30 ‰) (Clark and Fritz 1997). The δ¹³C values measured from the mountains and the ESRP (Table 4) probably reflect a greater influence from dissolution of carbonates in the mountains and infiltrating surface water, decay of organic matter, or both in the ESRP.

The approximate age of water was estimated from tritium concentrations in water samples, monthly records (from 1953 to 2009) of tritium concentrations in precipitation (Michel 1989 and personal communication, International Atomic Energy Agency 2013), and the decay equation for tritium. Tritium concentrations indicating pre-1952 (old), post-1952 (young), and a mixture of old and young water, for the years 1991 and 2012, are shown in Table 5. Old water was estimated for 2 sites in the mountains and 4 sites on the ESRP; young water was estimated for 3 sites in the mountains, 3 sites on the ESRP, and Camas Creek; and a mixture of young and old water was estimated for 3 sites in the mountains and 4 sites on the ESRP (Table 4). The old water on the ESRP was from sites (sites 5 and 17) in the northeastern part of the ESRP that received recharge from old water in the mountains (water similar in age to site 3, Table 4; Fig. 3) or from sites with unusual chemistry (i.e., anoxic water) located in volcanic vent corridors (sites 25 and 29). Groundwater from the other 7 sites on the ESRP was either young or a mixture of young and old water, indicating that a source of young water provided recharge to most of the shallow groundwater in the ESRP.

Solutes in groundwater are derived from recharge water, anthropogenic inputs, and chemical reactions. Potentially important sources of recharge water to the ESRP aquifer are groundwater from the Beaverhead Mountains, groundwater from the Camas Creek drainage basin (Rattray and Ginsbach 2014), infiltration of surface water (Medicine Lodge and Camas Creeks; lakes, ponds, and wetlands at Mud Lake and the CNWR; and irrigation water), and upwelling of geothermal water from beneath the aquifer (Mann 1986).

Upward movement of geothermal water occurs beneath the ESRP aquifer at the INL (site 100, Mann 1986) and may occur in the aquifer in the Mud Lake area within volcanic vent corridors (Anderson et al. 1999). Four sites (sites 25, 27, 29, and 34) located in volcanic vent corridors had either anoxic water (<2.3 and 7.9 % saturation, Table 1) or high water temperatures (16.2 and 17.7 °C, Table 1), which may indicate that geothermal water was a source of water to these sites. Other indicators of geothermal water were the presence of ammonia, elevated concentrations of fluoride, lithium, and boron (Tables 2, 3), the large concentration ratio of sodium to total cations (in meq/L) for sites 25 and 27, consistent with mixing of geothermal water similar in chemical composition to site 100 with other water from the ESRP (Fig. 6), and hydraulic head measurements at and near site 25 that indicates upward flow of groundwater (U.S. Geological Survey 2014).

Anthropogenic inputs include fertilizer in irrigated areas and road salt and anti-icing liquid (beginning in 2000) on highways (Fig. 2). Fertilizer may be sources of nitrogen, potassium, and chloride; road salt and anti-icing liquid may be sources of sodium, magnesium, and chloride.

Potential chemical reactions include carbonate reactions, dissolution of evaporite minerals, silicate weathering, redox reactions, and cation exchange. Carbonate, evaporite, and silicate minerals are present throughout the study area. Carbonate and evaporite minerals dissolve readily in dilute groundwater and should be important contributors of solutes to groundwater, with potentially large amounts of evaporite mineral dissolution within the Mud Lake subbasin (Fig. 3). Silicate minerals dissolve slowly in groundwater, so the only silicate minerals likely to dissolve in significant quantities during the short residence time of shallow groundwater in the ESRP aquifer (based on average linear flow velocities and the approximate age of groundwater) are plagioclase and volcanic glass (Rattray and Ginsbach 2014). Increasing potassium concentrations with decreasing depth in geothermal water below the ESRP aquifer (Mann 1986) indicate that upward-moving geothermal water may dissolve potassium feldspar in rhyolite. Incongruent dissolution of silicate minerals produces clay minerals; a calcium–sodium aluminosilicate stability diagram (with typical aluminum and silica activities for groundwater in the study area) indicates that the stable clay mineral in contact with groundwater is calcium montmorillonite (Fig. 8). Redox reactions may include oxidation of organic matter in wetlands and reduced species in geothermal water; δ¹⁵N values of 6.9 and 9.5 ‰ for sites 25 and 29 (Knobel et al. 1999), respectively, indicate that denitrification was not an important process in these anoxic waters. Clay is an efficient cation exchange substrate, so cation exchange may be important in areas where surface water infiltrates through sediment. For dilute surface water with much larger concentrations of calcium than magnesium, such as water from Camas Creek (site 78, Table 2), calcium should replace sodium on exchange sites (Drever 1997).

Geochemical modeling attempts to identify the net chemical reactions that account for observed changes in chemistry between initial (one or more) and final (one) water compositions (solutions) along a single flowline or joined (mixture) flowlines. Modeling was performed with PHREEQC using the minimal mode, which means that models were reduced to the minimum number of phases needed to satisfy model constraints (Parkhurst and Appelo 2013). Solutions used in the models were the water compositions in Tables 1, 2, and 3. Water from Camas Creek was used to represent relatively unevaporated Mud Lake water because the water from Mud Lake collected for chemical analyses in Tables 1, 2, and 3 was significantly evaporated. Solution uncertainties were the larger of the CB (absolute value) of a solution (rounded up to the nearest 1 %) or 5 %. Where measurements of aluminum and iron were not available, these constituents were assigned values of <10 and <20 μg/L, respectively. Values reported (or assigned) as “less than” were modeled as one-half the “less than” value with an uncertainty of 100 %. All solutions were assigned a pe of 4 except for the deep geothermal groundwater from site 100. Site 100 was assigned a pe of −5 so that speciation of constituents with PHREEQC would produce reduced species of carbon (methane) and sulfur (hydrogen sulfide).

Elements included in the models were hydrogen, oxygen, carbon, silica, nitrogen, sulfur, chloride, fluoride, calcium, magnesium, sodium, potassium, aluminum, and iron. Phases included carbonates (calcite, dolomite), evaporites (gypsum, halite, sylvite, bischofite), silicates [rhyolitic volcanic glass, plagioclase (An₂₅ and An₆₀), potassium feldspar], fluorite, calcium montmorillonite, goethite, fertilizer (ammonium nitrate, sylvite), road salt (halite), anti-icing liquid (MgCl₂, represented as bischofite), organic matter, and gasses (methane, hydrogen sulfide, DO, CO₂). DO was included as a phase because the water sample from Camas Creek (site 78) had a temperature of 19.3 °C and was undersaturated with DO; however, most water from Camas Creek would recharge at colder temperatures and be saturated with DO. Based on saturation indices, kinetic considerations, and redox conditions, all phases should dissolve in groundwater except for calcium montmorillonite and goethite (precipitate) and calcite (dissolve or precipitate).

Geochemical models were run with water compositions from sites 6–31 and 33–34 as final solutions. From one to many plausible results (i.e., model results that were consistent with land cover, hydrology, geochemistry, and saturation indices) were produced for each site except for site 17 (the results for site 17 were not consistent with δ²H, δ¹⁸O, and tritium values). Plausible model results from individual sites were similar in overall mass transfers of carbonates, silicates, evaporites, etc., but differed as to which specific phases were involved in mass transfers. A representative model result for each site is shown in Table 6 (small amounts of fluoride dissolved and goethite precipitated in nearly all models but are not shown in Table 6).

The model results indicate that sources of water to the ESRP aquifer were groundwater from the Beaverhead Mountains and the Camas Creek drainage basin; infiltration of surface water from Medicine Lodge Creek, Camas Creek, Mud Lake, and irrigation water; and upward flow of geothermal water. Because Camas Creek was the primary source of surface water to Mud Lake and for most surface water used for irrigation, and because surface water recharge occurred over most of the ESRP aquifer, Camas Creek was a very important source of recharge to the aquifer.

Mixing of groundwater with surface water or other groundwater occurred throughout the ESRP, and mixing of groundwater with surface water, other groundwater, and geothermal water was modeled for four sites (sites 25, 27, 29, and 34) located in volcanic vent corridors. Evaporation was modeled to reproduce the chemistry of groundwater from sites downgradient of the CNWR (site 8) and Mud Lake (sites 19 and 22).

Carbonate reactions, silicate weathering, and dissolution of evaporites and fertilizer explain most of the change in chemistry in the ESRP aquifer (Table 6). Large amounts of gypsum and halite dissolution support the interpretation that evaporite deposits are present within the Mud Lake subbasin and are a significant source of chloride, sodium, sulfate, and calcium in groundwater. Redox reactions were important at the CNWR and Mud Lake (from oxidation of organic matter) and at locations (sites 25, 27, 29, and 34) where upwelling geothermal water mixed with groundwater (from oxidation of reduced carbon and sulfur species in the geothermal water). For example, anoxic groundwater at two sites (sites 25 and 29) was modeled by mixing geothermal water (site 100) with surface water and groundwater upgradient of these sites and oxidation of reduced species in the geothermal water. The groundwater would only be anoxic at these sites if the groundwater was relatively stagnant because a continual inflow of oxidized groundwater would produce oxidized conditions. Cation exchange was locally important in areas where water had significant contact with sediment, for instance, areas where surface water infiltration occurred through wetlands or from irrigation.

Bischofite (could also be carnallite or anti-icing liquid) appears to be a source of chloride in a few locations. At one location (site 33), a large mass transfer of bischofite into solution was required (Table 6) because water from this site had an unusually large chloride/sodium ratio (4.3, Table 2). A similar chloride/sodium ratio, 4.1, was observed at site 22 in 1950 (Olmstead 1962); however, since 1950, the SpC at site 22 (1,000 μS/cm at 25 °C in 1950; Olmstead 1962) decreased 28 % and the chloride/sodium ratio decreased to 1.1 (Table 2). The SpC at site 33 decreased 25 % since measurements were first made in 2002 (U.S. Geological Survey 2014), although the chloride/sodium ratio has remained stable. A possible explanation for the initial large SpC and chloride/sodium ratio at these sites followed by a decreasing SpC and ratio (at site 22) is dissolution of evaporite minerals, including evaporite minerals that precipitated from a magnesium chloride brine. Magnesium chloride brines are present in some present-day salt lakes (Krupp 2005) and may have formed locally in the study area during the multiple climate-derived fluctuations of Lake Terreton.