Groundwater Quality Defined as Relative Concentrations

In this study, groundwater-quality data are presented as relative concentrations (RCs), which are defined as the ratio of a constituent’s concentration measured in a groundwater sample to the concentration of a constituent’s regulatory or non-regulatory benchmark used to evaluate drinking-water quality (eq. 1). The use of RC is similar to the approaches implemented by other studies to place the concentrations of constituents in groundwater in a toxicological context (U.S. Environmental Protection Agency, 1986; Toccalino and others, 2004; Toccalino and Norman, 2006; Rowe and others, 2007). The RC is defined as follows:

An RC value less than 1.0 indicates that the sample concentration was less than the benchmark concentration, and an RC value greater than 1.0 indicates that the sample concentration was greater than the benchmark concentration. Using RCs creates a single scale for comparing constituents that can be present at a wide range of concentrations. An RC can only be computed for constituents with water-quality benchmarks; therefore, constituents without water-quality benchmarks are not included in the status assessment.

Regulatory and non-regulatory benchmarks apply to treated water that is served to the consumer, not to untreated groundwater. To provide some context for the results, however, concentrations of constituents measured in the untreated groundwater were compared to benchmarks established by the U.S. Environmental Protection Agency (EPA), SWRCB division of drinking water (DDW), and USGS. The benchmarks used for each constituent were selected in the following order of priority:

For constituents with multiple types of benchmarks, this hierarchy might not result in selection of the benchmark with the lowest concentration. If a constituent has both an RL-CA and HAL-US, the lower benchmark of the two was selected. Additional information on the types of benchmarks and listings of the benchmarks for all constituents analyzed is provided by Bennett and others (2019).

Toccalino and others (2004), Toccalino and Norman (2006), and Rowe and others (2007) used the ratio of measured sample concentration to the benchmark concentration, either MCL-US or HBSL and defined this ratio as the benchmark quotient (BQ). The term RC is used in this report rather than BQ because the two values are not the same for constituents that have MCL-CA values that differ from their MCL-US values or for constituents that have neither MCL-US nor HBSL values (thus, no associated BQ).

For ease of discussion, RCs of constituents were classified in low, moderate, and high categories. The RC values greater than 1.0 were defined as “high” for all constituents. The RCs for nine constituents were calculated using SWRCB-DDW RL-CAs (sec-butylbenzene; carbon disulfide; diazinon; 1,4-dioxane; n-nitrosodiethylamine; n-propylbenzene; tert-butyl alcohol; 1,2,4-trimethylbenzene; and vanadium). The RL-CAs were paired with SWRCB-defined notification levels (NL-CA) set at concentrations ranging from 10 to 100 times less than the RL-CA (California State Water Resources Control Board, 2022b). For these nine constituents, the NL-CA was the benchmark concentration used to calculate the RC divide between “moderate” and “low.” For boron, which has an RL-CA and HAL-US, the HAL-US was used when calculating the initial RC. The NL-CA, however, was used to calculate the RC divide between “moderate” and “low,” as was done for the constituents that only had an RL-CA.

For constituents without RL-CAs, the divide between “moderate” and “low” was determined by constituent class. For inorganic constituents (trace elements, nutrients, radioactive constituents, and inorganic constituents having SMCL-US or SMCL-CA benchmarks), RC values greater than 0.5 and less than or equal to 1.0 were defined as “moderate,” and RC values less than or equal to 0.5 were defined as “low.” For organic and special-interest constituents, RC values greater than 0.1 and less than or equal to 1.0 were defined as “moderate,” and RC values less than or equal to 0.1 were defined as “low.” Although more complex classifications could be devised based on the properties and sources of individual constituents, use of a single threshold value for each of the two major groups of constituents (except for those with an RL-CA) provided a consistent objective criterion for distinguishing constituents present at moderate, rather than low, concentrations.

Dataset Used for Status Assessment

Groundwater-quality data used for the SacMetro-DSA status assessment came from sites sampled by the USGS for the GAMA-PBP (fig. 4). Detailed descriptions of the methods used to identify and select sites for sampling are given in Bennett and Fram (2014). Briefly, the SacMetro-DSA study unit was divided into three study areas: the North American, South American, and Cosumnes. The study-area boundaries were equivalent to CDWR-defined groundwater subbasins of the same name (California Department of Water Resources, 2016). Each study area was divided into equal-area grid cells (Scott, 1990): 24 cells in the North American study area, 17 cells in the South American study area, and 15 cells in the Cosumnes study area. The target cell size in all three study areas was approximately 22 square miles (mi²). The objective of defining grid cells was for the USGS to collect water-quality samples from one site in each cell. Candidate sites in each grid cell were identified by compiling lists of available wells in each cell from drillers’ log information obtained from the CDWR. The lists of sites (grid-cell inventories) were then taken into the field, and USGS staff went door-to-door to request permission to sample a well in each grid cell. The USGS was granted permission to sample sites in 49 of 56 grid cells (referred to here as USGS grid sites). The USGS grid sites were named with an alphanumeric GAMA identifier consisting of the prefix “S7-SAC-NA,” “S7-SAC-SA,” or “S7-SAC-C” and a number indicating the order of sample collection. The prefix represents the seventh (S7) domestic-supply assessment and the respective study area: North American (NA), South America (SA), or Cosumnes (C).

Samples collected from all USGS grid sites sampled by USGS were analyzed for 379 constituents (table 1). Water-quality data collected by USGS-GAMA are tabulated in Bennett and others (2019) and are available from the SWRCB’s publicly accessible GAMA Groundwater Information System and the USGS GAMA-PBP Groundwater-Quality Results web map (Jurgens and others, 2018; California State Water Resources Control Board, 2021b).

Concurrent with the SacMetro-DSA sampling effort, the USGS NAWQA Program staff sampled 22 monitoring wells in the study unit as part of ongoing NAWQA studies of groundwater quality in different land-use settings (fig. 4). Of the 22 wells sampled, 21 were part of an urban land-use network, and 1 was part of a rice land-use network. The NAWQA monitoring wells were sampled for a suite of analytes similar to those of the SacMetro-DSA sampling effort, including field water-quality constituents, volatile organic compounds (VOCs), pesticide and pesticide degradates, nutrients, major ions and trace elements, dissolved organic carbon, perchlorate, hexavalent chromium, stable isotopic ratios, and tritium. Results from NAWQA monitoring wells, presented by Kingsbury and others (2021), are not included in the status or understanding assessment of the SacMetro-DSA because they are not representative of typical domestic wells.

Selection of Constituents for Discussion

Aquifer-scale proportions were calculated and are presented for the 9 constituents that were present at high or moderate RCs in samples from the 49 USGS grid sites (table 2). Aquifer-scale proportion results also are presented for chloroform and bentazon because they had detection frequencies of greater than 10 percent in samples from the USGS grid sites (table 2).

An additional 36 inorganic constituents, 14 organic constituents, and 12 tracers or isotopes were detected but either have no drinking-water quality benchmarks or were only detected at low RCs (table 3). Aquifer-scale proportions are not presented for constituents only detected at low RCs because the proportion of the groundwater resource having low RCs for those constituents was 100 percent. Of the 379 constituents analyzed in samples collected for the SacMetro-DSA, 306 were not detected in any of the samples (Bennett and others, 2019).

Calculation of Aquifer-Scale Proportions

A grid-based statistical approach (Belitz and others, 2010) was used to calculate the proportions of the domestic-supply aquifer system in the SacMetro-DSA with high, moderate, and low RCs of constituents. For ease of discussion, these proportions are referred to as “high RC,” “moderate RC,” and “low RC” aquifer-scale proportions. Aquifer-scale proportions were calculated for each study area and for the study unit. Calculations of aquifer-scale proportions were made for individual constituents and for classes of constituents. The classes consisted of groups of related individual constituents. Aquifer-scale proportions for constituent classes were calculated by using the maximum RC for any constituent in the class to represent the class. For example, a site having a high RC for arsenic, moderate RC for boron, and low RCs for molybdenum, selenium, and other trace elements would be counted as having a high RC for the class of trace elements with health-based benchmarks.

The USGS grid-site dataset was used for grid-based calculations. Aquifer-scale proportions were calculated for each of the study areas for comparison. Study areas had equal-area cell sizes; therefore, the proportions of RCs among study areas can be compared directly. High-RC aquifer-scale proportion was calculated as the fraction of the USGS grid cells in the study area having high RCs for a constituent (eq. 2). The moderate-RC aquifer-scale proportion was calculated similarly.

High RC aquifer-scale proportions for the study unit were calculated as an area-weighted combination of the aquifer-scale proportions for the three study areas (eq. 3). Moderate-RC aquifer-scale proportions for the study unit were calculated the same way.

Study-unit detection frequencies for organic constituents were also calculated as area-weighted detection frequencies. The detection frequency in each study area was calculated by using equation 2 with $N_{SA}{}^{high}$ replaced by the number of samples with detections, and the detection frequency for the study unit as a whole was calculated by using equation 3 after making the corresponding replacement.

Selection of Constituents for Additional Evaluation

The understanding assessment places groundwater quality within physical and chemical contexts based on the potential explanatory variables. A subset of constituents was selected for additional evaluation in the understanding assessment on the basis of the following two criteria:

These criteria resulted in selection of five inorganic constituents (arsenic, chloride, iron, manganese, and TDS), two organic constituents (bentazon and chloroform), and the special-interest constituent perchlorate for evaluation in the understanding assessment. Additionally, hexavalent chromium, which did not meet the listed selection criteria, is included in the understanding assessment because it is considered a special-interest constituent. California’s regulatory statutes were updated in 2014 and established an MCL-CA for hexavalent chromium in drinking water of 10 micrograms per liter (µg/L; California State Water Resources Control Board, 2020b). The MCL-CA was later invalidated in September 2017 by the Superior Court of Sacramento County because it “failed to properly consider the economic feasibility of complying with the MCL.” However, the SWRCB is likely to establish a new MCL-CA for hexavalent chromium, which could be at the same level as the previously suspended MCL-CA of 10 µg/L (California State Water Resources Control Board, 2020b). Hexavalent chromium has been detected at concentrations close to or greater than the previously established MCL-CA in groundwater used for public and domestic supply throughout the State. Therefore, understanding the processes that can lead to elevated hexavalent chromium concentrations in groundwater was the focus of several recent studies (Izbicki and others, 2015; Manning and others, 2015; Hausladen and others, 2018). In this report, hexavalent chromium concentrations are compared to the USGS-HBSL of 20 µg/L because the previously established MCL-CA was suspended.

Statistical Relations Among Potential Explanatory Variables and Groundwater Quality

Nonparametric statistical tests were used to test the significance of correlations among potential explanatory variables and water-quality constituents. Nonparametric statistics are robust techniques that generally are not affected by outliers and do not require that the data follow any particular distribution (Helsel and others, 2020). All statistical analyses were done using Spotfire S+ for Windows, version 8.1 (TIBCO Software, Inc.). The probability value (p-value) used for hypothesis testing for this report was compared to a level of significance (α) of 5 percent (α=0.05) to evaluate whether the relation was statistically significant (p-value<α).

Two different statistical tests were used because the set of potential explanatory variables included categorical and continuous variables. Groundwater-age class, oxidation–reduction (redox) class, and study area were treated as categorical variables. Land use, septic-tank density, leaking (or formerly leaking) underground fuel-tank (LUFT) density, aridity index, elevation, well depth, depth to top of screened or open interval, pH, and dissolved oxygen (DO) were treated as continuous variables. Concentrations of water-quality constituents also were treated as continuous variables. Correlations between potential explanatory variables and water-quality constituents were tested for significance. Correlations between continuous variables were evaluated using the Spearman’s rho test (cor.test function in Spotfire S+ version 8.1) to calculate the rank-order coefficient (ρ, rho) and to determine whether the correlation was significant (p-value<α). Values of ρ could range from −1 (complete inverse, or negative, correlation) to 0 (no correlation) to 1 (complete positive correlation). Relations between categorical variables and continuous variables were evaluated using the Kruskal-Wallis and Wilcoxon rank-sum test functions in Spotfire S+ version 8.1 (Helsel and others, 2020). The null hypothesis for these tests is that the distributions of the continuous variables among multiple groups (Kruskal-Wallis test), or between two groups (Wilcoxon rank sum test), are not significantly different from one another.

Trace Elements

Most trace elements analyzed in the SacMetro-DSA study unit were detected at low RCs (fig. 9). The RCs (for one or more constituents) of trace elements with health-based benchmarks were high, moderate, and low in 10 percent, 31 percent, and 59 percent, respectively, of USGS grid sites in the domestic-supply aquifer system (table 6). Arsenic was the only trace element present at a high RC; boron and hexavalent chromium were present at moderate RCs (table 7; fig. 10A).

Understanding Assessment for Arsenic

Arsenic is a semi-metallic trace element that is often naturally present in groundwater. Arsenic concentrations in groundwater are typically controlled by the abundance and dissolution rates of arsenic-bearing minerals in the aquifer and specific geochemical conditions that are conducive to arsenic mobilization (reduced or anoxic) and allow for the desorption of arsenic from mineral surfaces (Welch and others, 2000; Smedley and Kinniburgh, 2002). Pyrite is a common sulfide mineral in aquifers and may contain up to several percent of arsenic. Copper ore smelting, coal combustion, arsenical pesticides, and wood preservatives are other anthropogenic sources of arsenic (Welch and Stollenwerk, 2003). Mining activities for copper, gold, and other metals can increase the rate of dissolution of natural arsenic-bearing minerals such as pyrite (Smedley and Kinniburgh, 2002).

The MCL-US for arsenic was lowered from 50 µg/L to 10 µg/L in 2002; however, chronic exposure to arsenic concentrations between 10 µg/L and 50 µg/L in drinking water has been linked to increased cancer risk and to non-cancerous effects including skin damage and circulatory problems (U.S. Environmental Protection Agency, 2001). An estimated 8 percent of groundwater resources used for drinking water in the United States have high concentrations of arsenic (greater than 10 µg/L; Focazio and others, 2000; Welch and others, 2000), and high concentrations of arsenic in groundwater resources used for drinking water are a worldwide concern (Smedley and Kinniburgh, 2002; Welch and others, 2006).

Arsenic was present at high and moderate RCs in 10 percent and 20 percent, respectively, of USGS grid sites in the domestic-supply aquifer system (table 7). The proportion of the domestic-supply aquifer system having high RCs of arsenic was highest in the North American study area (17 percent), followed by Cosumnes study area (8 percent), and then the South American study area (0 percent; table 7; fig. 10). Previous studies and a review of elevated arsenic in groundwater have identified two primary mechanisms for arsenic mobilization related to these conditions: (1) desorption from, or inhibition of sorption to, aquifer materials concurrent with increasing pH and (2) release of arsenic from dissolution of iron or manganese oxyhydroxides under iron- or manganese-reducing conditions (Smedley and Kinniburgh, 2002; Belitz and others, 2003; Welch and others, 2006). The first mechanism (desorption) requires pH values greater than 7.8, the value above which the primary arsenate species (HAsO₄⁻²) is relatively soluble (Smedley and Kinniburgh, 2002). The accumulation of elevated arsenic concentrations in groundwater is also affected by hydrologic conditions in the aquifer system. Longer contact times between groundwater and aquifer materials increases the reaction time with aquifer materials (Smedley and Kinniburgh, 2002).

In the SacMetro-DSA study unit, high and moderate RCs of arsenic were detected mostly in samples from USGS grid sites in the North American study area, either along or relatively close to the Sacramento River (fig. 11A). High RCs of arsenic were primarily detected in anoxic groundwater (DO less than 1 milligram per liter, reducing conditions) and water classified as pre-modern or mixed, although one anoxic sample was classified as modern (fig. 12; table 4). About half of the samples with moderate concentrations were classified as anoxic, and about half were also classified as modern (fig. 12). Arsenic concentrations had a significant negative correlation to DO (table 8) and significant positive correlations to iron and manganese (table 9). Mobilization of arsenic appears to be primarily related to reducing conditions in the aquifer system, particularly in the North American study area. These observations are consistent with previous studies in the Sacramento Valley (Hull, 1984; Dawson, 2001a). Reducing conditions conducive to arsenic mobilization have been described in fine-grained flood basin deposits in the central part of the valley close to the Sacramento River (Hull, 1984; Dawson, 2001a).

Table 8.    

Results of Spearman's rho (ρ) tests for correlations between selected potential explanatory variables (Bennett, 2022) and selected water-quality constituents (Bennett and others, 2019), Sacramento Metropolitan Domestic-Supply Aquifer study unit, 2017, California Groundwater Ambient Monitoring and Assessment Priority Basin Project (U.S. Geological Survey, 2022).

[ρ values are shown for tests in which the variables were determined to be significantly correlated on the basis of p-values (not shown) being less than the level of significance (α) of 0.05. Abbreviations: LUFTs, leaking (or formerly leaking) underground fuel tanks; ns, Spearman's test indicates no significant correlation between factors; blue text, significant positive correlation; red text, significant negative correlation]

Table 8.    Results of Spearman's rho (ρ) tests for correlations between selected potential explanatory variables (Bennett, 2022) and selected water-quality constituents (Bennett and others, 2019), Sacramento Metropolitan Domestic-Supply Aquifer study unit, 2017, California Groundwater Ambient Monitoring and Assessment Priority Basin Project (U.S. Geological Survey, 2022).

ρ	Percentage land use			Density of septic tanks	Density of LUFTs	Aridity index	Elevation	Well depth	Depth to top of per foration	pH	Dissolved oxygen concen tration
	Agri cultural	Natural	Urban
Nutrients and trace elements with health-based benchmarks
Arsenic	ns	ns	ns	ns	ns	ns	−0.41	ns	ns	ns	−0.52
Hexavalent chromium	ns	ns	ns	0.44	ns	ns	ns	ns	ns	−0.29	0.66
Nitrate	ns	ns	ns	ns	ns	ns	ns	ns	ns	−0.49	0.60
Inorganic constituents with aesthetic-based benchmarks
Chloride	0.33	−0.39	ns	ns	ns	ns	−0.37	−0.31	ns	ns	ns
Iron	ns	ns	ns	−0.37	ns	ns	−0.32	−0.33	ns	0.31	−0.50
Manganese	ns	ns	ns	ns	ns	ns	−0.42	−0.29	ns	0.38	−0.70
Total dissolved solids (TDS)	0.46	−0.37	ns	ns	ns	ns	−0.47	−0.36	ns	ns	−0.42
Organic and special-interest constituents
Chloroform	Ns	ns	0.36	ns	0.36	ns	ns	ns	−0.29	ns	ns
Bentazon	0.38	ns	ns	ns	ns	ns	−0.43	ns	ns	ns	ns
Perchlorate	ns	ns	ns	ns	ns	ns	0.42	ns	ns	−0.62	0.62

Table 9.    

Results of Spearman's rho (ρ) tests for correlations between selected water-quality constituents (Bennett and others, 2019), Sacramento Metropolitan Domestic-Supply Aquifer study unit, 2017, California Groundwater Ambient Monitoring and Assessment Priority Basin Project (U.S. Geological Survey, 2022).

[ρ values are shown for tests in which the variables were determined to be significantly correlated on the basis of p-values (not shown) being less than the critical level (α) of 0.05. Abbreviations: ns, Spearman's test indicates no significant correlation between factors; blue text, significant positive correlation; red text, significant negative correlation]

Table 9.    Results of Spearman's rho (ρ) tests for correlations between selected water-quality constituents (Bennett and others, 2019), Sacramento Metropolitan Domestic-Supply Aquifer study unit, 2017, California Groundwater Ambient Monitoring and Assessment Priority Basin Project (U.S. Geological Survey, 2022).

ρ	Arsenic	Hexavalent chromium	Nitrate	Chloride	Iron	Manganese	Total dissolved solids (TDS)	Chloroform	Bentazon	Perchlorate
Arsenic		ns	−0.30	ns	0.42	0.40	ns	ns	0.32	−0.50
Hexavalent chromium			0.55	ns	−0.46	−0.63	ns	ns	ns	0.55
Nitrate				ns	−0.45	−0.56	ns	ns	ns	0.57
Chloride					ns	ns	0.86	ns	0.33	ns
Iron						0.77	0.32	ns	0.29	−0.52
Manganese							0.30	ns	ns	−0.59
Total dissolved solids (TDS)								ns	0.47	ns
Chloroform									ns	ns
Bentazon										−0.35
Perchlorate

The foothills of the Sierra Nevada east of the SacMetro-DSA study unit has a long history of mining (U.S. Geological Survey, 2021b), and gold has been extracted in some locations from gravel deposits associated with the American River and along the eastern edge of the North American study area (fig. 11A). Mining activities, however, likely are not responsible for elevated concentrations of arsenic because wells with high or moderate concentrations of arsenic were not close to any of the USGS grid sites with a history of gold extraction (fig. 11A).

Understanding Assessment for Hexavalent Chromium

Chromium is a transition metal commonly found in minerals associated with ultramafic igneous rocks (Hem, 1985). Chromium has also been shown to be present at high enough concentrations in granitic rocks to contribute appreciable amounts of chromium to groundwater (Izbicki and others, 2015). Chromite, a mineral highly resistant to weathering, is the most common chromium bearing mineral. However, chromium can substitute into minerals like amphiboles, garnets, micas, pyroxenes, and spinels, which are more easily weathered and therefore could be important sources of chromium to the environment (Izbicki and others, 2015). There are several industrial uses for chromium, such as its inclusion as an alloy in stainless steels, in chrome plating, as a pigment in paints, and as a wood preservative (California State Water Resources Control Board, 2020b). Chromium is also discharged to the atmosphere during the manufacture and combustion of gas and coal (California State Water Resources Control Board, 2020b).

Transition metals like chromium often have more than one thermodynamically stable oxidation state. Chromium-3 is the most common oxidation state in most minerals (Nriagu and Nieboer, 1988). In groundwater, dissolved chromium is either reduced ionic forms of chromium-3 or oxidized anionic forms of hexavalent chromium, like chromate and dichromate (Hem, 1985). The presence and relative abundance of natural chromium-3 and hexavalent chromium on mineral surfaces or dissolved in groundwater are controlled by the following factors: (1) chromium concentrations in primary minerals; (2) mineral weathering rates; (3) the presence of manganese oxides that convert chromium-3 to hexavalent chromium; and (4) the presence of reducing agents that allow for the reduction of hexavalent chromium to chromium-3, which can be mediated through abiotic or microbial pathways (Izbicki and others, 2015). Dissolved total chromium concentrations were compared to hexavalent chromium concentrations in paired samples collected by the GAMA-PBP from more than 900 public-supply wells throughout California; about 90 percent of the dissolved chromium was hexavalent chromium (Izbicki and others, 2015). Similarly, dissolved chromium concentrations were predominantly hexavalent chromium in the SacMetro-DSA samples from public-supply wells (fig. 13).

Hexavalent chromium was detected in 83 percent of wells sampled in the SacMetro-DSA study unit, with detected concentrations ranging from 0.06 to 12.1 µg/L (Bennett and others, 2019). The hexavalent chromium sample was accidentally missed at one well during sample collection, so hexavalent chromium data are available for 48 of the 49 grid wells. Wells with moderate concentrations of hexavalent chromium, greater than one-half of the USGS HBSL of 20 µg/L, were observed in all three study areas (fig. 10A). Comparing hexavalent chromium concentrations by age class showed that samples classified as mixed had higher hexavalent chromium concentrations than samples classified as modern. Comparing hexavalent chromium concentrations by redox classification showed oxic samples had higher concentrations than anoxic samples (table 4; fig. 14). Significant differences were not observed in the concentration ranges of hexavalent chromium by study area (table 4). Concentrations of hexavalent chromium were significantly positively correlated to the density of septic tanks and dissolved-oxygen concentrations but were significantly negatively correlated to pH (table 8). Significant positive correlations to nitrate and perchlorate and significant negative correlations to iron and manganese were also observed for hexavalent chromium concentrations (table 9). In addition to the statistical comparison made with constituents selected for discussion in the status assessment, hexavalent chromium was significantly positively correlated to vanadium (p<0.0001) and selenium (p=0.03). Vanadium and selenium share similar chemistries with chromium and form soluble oxyanions in the plus six oxidation state (Izbicki and others, 2015).

Geogenic hexavalent chromium that is naturally derived from aquifer materials containing chromium-bearing minerals appears to be the source of hexavalent chromium observed in samples collected in the SacMetro-DSA. Hexavalent chromium is positively correlated to oxic conditions, negatively correlated to pH (more hexavalent chromium in more alkaline waters), and positively correlated with nitrate, which in some areas of agricultural land use has been suggested as an indicator of hexavalent chromium mobilization related to irrigation returns (Hausladen and others, 2018). Hexavalent chromium that is naturally derived from aquifer materials containing chromium-bearing minerals has been shown be a significant source of hexavalent chromium in aquifers for large areas of California (Izbicki and others, 2015; Manning and others, 2015; Hausladen and others, 2018).

Conversely, anthropogenic discharges of hexavalent chromium related to industrial processes or oxidation of chromium-3 to hexavalent chromium resulting from remediation efforts to clean up organic solvent pollution are known to occur in some urbanized areas of California (Hausladen and others, 2018). Hexavalent chromium concentrations in monitoring wells (wells typically installed near sites with known industrial contamination) throughout the State exhibited correlations to volatile organic compounds (VOCs), specifically 1,1-dichloroethene (1,1-DCE), 1,1-dichloroethane (1,1-DCA), and 1,4-dioxane, which track with their uses in industrial processes related to metal manufacturing and chrome-plating processes (Hausladen and others, 2018). However, no VOCs were detected in the SacMetro-DSA samples that indicated industrial contamination. For additional analysis, monitoring wells in the SacMetro-DSA study unit for which hexavalent chromium was measured within the last 10 years were retrieved from the GAMA Groundwater Information System (California State Water Resources Control Board, 2021b). Because these monitoring wells are associated with known groundwater contamination sites, they were assumed to indicate the locations of potential industrial sources of hexavalent chromium and the relative proximity of these potential anthropogenic sources were compared to the SacMetro-DSA grid wells (fig. 11B). SacMetro-DSA grid wells with moderate concentrations of hexavalent chromium were not near any of the monitoring wells in which hexavalent chromium was at high or moderate RCs, further indicating that hexavalent chromium observed in the grid wells likely did not originate from an industrial source.

Nutrients

As a class, nutrients were not present at high RCs but were present at moderate RCs in 12 percent of USGS grid sites in the domestic-supply aquifer system of the SacMetro-DSA study unit. Nitrate was present at moderate RCs (table 7). Nitrate in groundwater has anthropogenic and natural sources (Dubrovsky and others, 2010); however, concentrations greater than 2 milligrams per liter (mg/L) as nitrogen (RC of 0.2) generally are considered to indicate presence of nitrate from anthropogenic sources (Mueller and Helsel, 1996).

Inorganic Constituents with Aesthetic-Based Benchmarks

The RCs for constituents with aesthetic-based benchmarks (SMCLs) were high, moderate, and low in 16, 2, and 82 percent, respectively, of USGS grid sites in the domestic-supply aquifer system (table 6). Constituents with SMCLs can affect the aesthetic properties of water, such as taste, color, and odor, or can create technical problems, such as scaling and staining. Chloride, iron, manganese, and TDS were the constituents with high RCs in the grid-site network (table 7). Chloride and TDS are indicators of salinity, whereas iron and manganese are trace elements. The major ion chloride and TDS have recommended and upper SMCL-CA values. In this report, data were compared to the upper SMCL-CA values.

Understanding Assessment for Iron and Manganese

Iron and manganese have similar chemistries and mineral associations. Weathering and dissolution of igneous rocks and mafic minerals that are relatively high in iron and manganese is a natural source to groundwater (Hem, 1985). In addition, iron-bearing silicate, sulfide, or oxide minerals are in most rocks and sediments. Once released from their primary minerals, iron and manganese often reprecipitate as oxyhydroxide and oxide coatings on mineral and sediment grains (Hem, 1985). The solubility of iron and manganese is strongly dependent on oxidation–reduction conditions; the more reduced species are more soluble (McMahon and Chapelle, 2008).

Comparison benchmarks used in this study for iron and manganese were the SMCLs of 300 and 50 µg/L, respectively. The SMCL-CA is established as the concentrations at which iron and manganese become nuisances and can affect the aesthetic properties of water but have minimal adverse effects on human health. For USGS grid sites in the SacMetro-DSA domestic-supply aquifer system, high concentrations of iron and manganese were detected in 4 and 14 percent of samples, respectively, and moderate concentrations of iron and manganese were detected in 2 and 4 percent of samples, respectively (table 7). Iron and manganese were significantly higher in samples from USGS grid sites classified as anoxic than from USGS grid sites classified as oxic (table 4; fig. 15). Iron and manganese were significantly negatively correlated to dissolved oxygen, elevation, and well depth and significantly positively correlated to pH (table 8). Iron was also significantly negatively correlated to density of septic tanks (table 8). Iron and manganese were significantly positively correlated to each other, arsenic, and TDS and significantly negatively correlated to hexavalent chromium, nitrate, and perchlorate (table 9). Additionally, iron was positively correlated to bentazon (table 9).

Reducing conditions that are conducive to the mobilization of iron and manganese in alluvial aquifers typically form when DO is consumed by the oxidation of organic matter within the sediments (Appelo and Postma, 2005). In the SacMetro-DSA study unit, groundwater was primarily oxic; however, anoxic, suboxic, or mixed conditions were found in 22 percent of the USGS grid sites sampled, most often along the western boundary of the study unit in wells close to the Sacramento River (fig. 16). This is consistent with a general pattern of decreasing DO concentrations as groundwater flows along the regional flowpath from east to west (Landon and others, 2010). The lowest concentrations of DO typically are near the trough of the basin, notably in the fine-grained sediments of the flood basin deposits near the Sacramento River, as has been described in other studies of the Sacramento Valley (Hull, 1984; Dawson, 2001a).

Chloride and Total Dissolved Solids (Salinity Indicators)

The weathering and dissolution of minerals in soils, sediments, and rocks is a primary source of TDS to groundwater. Concentrations of TDS in groundwater can also be affected by mixing with hydrothermal solutions or other saline waters, interactions with sediments of marine or lacustrine origins, or evaporative concentration of shallow groundwater (Hem, 1985). Groundwater recharged from water used for irrigation and wastewater discharge is a potential anthropogenic source of TDS (Hem, 1985). Total dissolved solids were present at high and moderate RCs in 2 percent of the SacMetro-DSA (table 7). The South American study area had one site with a high RC of TDS, and the North American study area had one site with a moderate RC (table 7). Chloride is a component of TDS; the one site with a high RC of TDS in the South American study area also had high a RC of chloride.

The TDS concentrations were higher in groundwater classified as anoxic than as oxic with respect to redox conditions (table 4). The TDS concentrations were significantly positively correlated to agricultural land use and to concentrations of chloride, iron, manganese, and bentazon (tables 8, 9). Concentrations of TDS were significantly negatively correlated to natural land use, elevation, well depth, and DO concentrations (table 8). The observed correlations are generally consistent with associations or processes known to affect concentrations of TDS. Generally, TDS was higher in agricultural areas that have higher recharge rates because of irrigation at the land surface, which can transport additional solutes to the aquifer. Recharge water from precipitation is low in TDS, and precipitation increases as elevation increases. Higher elevations also correlate to higher proportions of natural land use in the eastern parts of the study unit. The USGS grid sites with high and moderate concentration of TDS were on the western edge of the study unit close to the Sacramento River. Concentrations of TDS in these wells were noticeably higher than the rest of the USGS grid sites in the SacMetro-DSA study unit (fig. 17). The median TDS concentration for USGS grid sites sampled in the SacMetro-DSA was 239 mg/L (Bennett and others, 2019), which is less than the upper SMCL-CA of 500 mg/L. Although beyond the scope of this report, concentrations of TDS in the Sacramento Valley aquifer system (in particular, in the low-lying floodbasin areas) have been discussed as potentially being related to several factors, such as evaporative concentration of agricultural recharge, aquifer residence time, sediment lithology, or contributions from deeper more saline parts of the aquifer (Hull, 1984; Dawson, 2001b).

Understanding Assessment for Bentazon

Bentazon was the most detected pesticide in the SacMetro-DSA, with a detection frequency of 24 percent (fig. 18). Bentazon was only detected at low RCs (fig. 19B). Bentazon is an herbicide used to control broadleaf weeds and sedges that has been used on a variety of crop types including corn, dry beans, peanuts, peas, rice, soybeans, and sorghum (U.S. Environmental Protection Agency, 1985). In its commercial form (sodium bentazon), bentazon is moderately soluble in water and has a low soil adsorption coefficient that causes bentazon to be leached from soils and carried to the groundwater system (California Office of Environmental Health Hazard Assessment, 1999). Bentazon can degrade in the environment over time; however, degradation can be slowed or stopped under anoxic conditions (California Office of Environmental Health Hazard Assessment, 1999). The use of bentazon on rice in California was suspended in 1989 because bentazon was detected in groundwater wells in the vicinity of rice-growing areas (California Department of Food and Agriculture, 1990). At that time, most of the bentazon use in California was on rice fields, and approximately 60 percent of rice fields were treated annually (California Department of Food and Agriculture, 1990). Although banned on rice, the use of bentazon on corn, sorghum, peas, and dry beans was still permitted in California (with certain limitations) because there had not been any detections of bentazon in groundwater associated with use on those crops (California Department of Food and Agriculture, 1990).

Bentazon was detected in the SacMetro-DSA only in wells classified as modern or mixed with respect to groundwater age, but concentrations were not significantly different between the modern and mixed categories (table 4). Bentazon concentration was significantly positively correlated to agricultural land use and concentrations of arsenic, chloride, iron, and TDS; bentazon concentration was significantly negatively correlated to elevation and concentrations of perchlorate (tables 8, 9). A previous study showed high detection frequencies of bentazon in areas of rice agriculture (Dawson, 2001b). Comparing the locations of wells with detections of bentazon to crop types where bentazon is (or was in the case of rice) approved for use in California showed that most of the detections were near areas of rice agriculture in the North American study area (fig. 20). The four detections of bentazon in wells in the South American and Cosumnes study areas are south of the rice growing regions of the Sacramento Valley. However, a previous study examining groundwater quality beneath recently urbanized areas of Sacramento found bentazon in shallow groundwater wells near the Sacramento River and attributed the detections to the movement of agriculturally affected Sacramento River water to the shallow aquifer system (Shelton, 2005).

In the late 1990s, nearly a decade after being banned for use on rice fields, bentazon was still being detected in the Sacramento River and its tributaries (Domagalski, 2000). Only one of the four detections outside of rice growing areas was close to the Sacramento River. As of 2014, agricultural land use compiled for the California Department of Water Resources by Land IQ, LLC, showed that the predominant crop type in the vicinity of the three detections in the southernmost wells was corn, sorghum, Sudangrass, or beans (fig. 20; Land IQ, 2017). Except for Sudangrass, bentazon use is still approved for use on these crops. Querying the California Pesticide Information Portal (CalPIP) shows bentazon applications in Sacramento County between 1991 and 2017 (California Department of Pesticide Regulation, 2020).

From 1991 to 1995, the use of bentazon in the county was limited to landscape maintenance, ornamental turf, and rights of way; however, the California Department of Pesticide Regulation noted some exceptions to reporting requirements for home-and-garden use and for most industrial and institutional uses (California Department of Pesticide Regulation, 2020). In 1995, 1996, 1998, 2014, and 2017, there were registered applications of bentazon on beans or peas within the county; however, there were no reported uses on corn or sorghum (California Department of Pesticide Regulation, 2020). The Land IQ dataset did not identify peas as a unique crop type. About half of the reported bentazon use in Sacramento County between 1991 and 2017 was accompanied with location information accurate to the township, range, and section (1-square mile), which could then be compared with locations of wells with detections of bentazon and with areal distributions of crop types for which bentazon use was approved (fig. 20). There were no reported uses on corn or sorghum, however, and the identifiable sections of land with known applications were not near the wells with detections. Therefore, we cannot clearly determine whether approved applications of bentazon to crops contributed to detections of bentazon in wells at the southern end of the study unit.

Understanding Assessment for Chloroform

The trihalomethane (THM) chloroform was the most detected VOC in the SacMetro-DSA study unit. Chloroform was detected in 12 percent of USGS grid sites and in all three study areas (figs. 18, 19A). Nationally, chloroform is one of the most frequently detected VOCs in groundwater used for public or domestic supply (Schaap and Zogorski, 2006; Ivahnenko and Zogorski, 2006; Zogorski and others, 2006). Although frequently detected, concentrations of chloroform in drinking-water supplies are usually far below the MCL-US of 80 µg/L. The MCL-US applies to the sum concentration of the following THMs: bromodichloromethane, chloroform, dibromochloromethane, and tribromomethane. In the SacMetro-DSA, only bromodichloromethane (one detection) and chloroform (six detections) were observed in the USGS grid sites. One site had a moderate RC of chloroform and a low RC detection of bromodichloromethane. The combined concentration of the two THMs remained moderate.

Chloroform is a disinfection by-product resulting from the treatment of water supplies or wastewater with chlorine to prevent waterborne diseases (Ivahnenko and Zogorski, 2006; Carter and others, 201231). The distribution of chlorine-treated water for municipal use, or post-treatment discharge back to the environment (in the case of regulated wastewater), provides opportunities for chloroform to be carried into the groundwater system. For example, application of chlorinated public-supply water to lawns, parks, or golf courses as irrigation can discharge chloroform back to the environment. Furthermore, distribution pipes or sewer systems carrying treated water can leak and release chloroform to the groundwater system (Ivahnenko and Zogorski, 2006; Carter and others, 201231). Outside of urban areas, domestic-well disinfection to prevent or treat bacterial issues typically is done by pouring chlorine solutions directly into the well, thus providing a direct pathway for THMs to enter the aquifer system (Wright, 1997; Centers for Disease Control and Prevention and U.S. Department of Housing and Urban Development, 2006). Septic systems in rural areas can also be a source of chloroform to the aquifer system (Ivahnenko and Zogorski, 2006).

Chloroform concentrations were not significantly correlated to age class, redox class, or study area (table 4). Chloroform concentrations were significantly positively correlated to percent of urban land use and LUFT density and were significantly negatively correlated to the depth to top of perforation (table 8). Chloroform concentrations were not significantly correlated to other selected water-quality constituents (table 9). The correlation of chloroform to LUFT density likely is coincidental with the fact the LUFTs are more common in urban areas because LUFTs are not a known source of chloroform to the environment.

Understanding Assessment for Perchlorate

Perchlorate is a highly soluble inorganic constituent for which an MCL-CA of 6 µg/L was established in 2007 (California State Water Resources Control Board, 2020c). Perchlorate was present at moderate RCs in 8 percent of the SacMetro-DSA study unit (table 7). Perchlorate was also detected at low RCs in 58 percent of the SacMetro-DSA study unit across all three study areas (table 6; figs. 18, 19C).

Perchlorate concentrations are commonly affected by the same factors affecting nitrate concentrations, given that they have similar oxidation–reduction potentials (Coates and Achenbach, 2004). Concentrations of perchlorate were significantly higher in oxic than in anoxic conditions (table 4). Comparing study areas, the only significant difference observed was that perchlorate concentrations were higher in the Cosumnes study area than in the South American study area (table 4). Perchlorate concentrations were significantly negatively correlated to pH and significantly positively correlated to elevation and dissolved-oxygen concentration (table 8). Finally, perchlorate was significantly positively correlated to hexavalent chromium and nitrate and significantly negatively correlated to arsenic, iron, manganese, and bentazon (table 9).

Perchlorate has natural and anthropogenic sources. An investigation of perchlorate occurrence and distribution throughout California and the southwestern United States, primarily from public-supply wells, showed that concentrations of perchlorate above certain thresholds can indicate anthropogenic activity, and concentrations below these thresholds are likely the result of natural sources (Fram and Belitz, 2011). Methods developed in Fram and Belitz (2011) were used to compare actual and predicted probabilities of perchlorate detections in this study.

Only samples that were oxic or suboxic and had a perchlorate measurement were included in the analysis because perchlorate is reduced at about the same redox conditions as nitrate. This screening process resulted in 40 samples: 10 Cosumnes study-area samples, 19 North American study-area samples, and 11 South American study-area samples.

Median aridity-index values of 0.384, 0.434, and 0.386, and detection frequencies of 0.1 µg/L, 0.5 µg/L, and 1 µg/L, were calculated for the Cosumnes, North American, and South American study-area samples, respectively. For example, of the 19 samples in the North American study area, 15 had detections of perchlorate at concentrations greater than 0.1 µg/L, which equates to a 79-percent detection frequency (fig. 21). In the Cosumnes and North American study areas, detection frequencies of perchlorate at concentrations greater than 0.1 µg/L and greater than 0.5 µg/L were higher than would be expected under natural conditions (fig. 21), indicating anthropogenic contributions of perchlorate to the groundwater system are likely in these areas. All other combinations of study area and threshold had observed detections that were within the 90-percent Jeffreys confidence intervals (Brown and others, 2001) for the predicted detection frequencies and were therefore consistent with expectations for natural conditions (Fram and Belitz, 2011).

Perchlorate is used in rocket fuels, explosives, road flares, automobile air-bag systems, and other products that can release perchlorate to the environment (California State Water Resources Control Board, 2020c). The EPA-designated Aerojet General Corporation Superfund site in Rancho Cordova, Calif., is known to have contaminated a 27-square mile area with chemicals associated with rocket propellants that include perchlorate (Sacramento County Water Agency, 2004; Sacramento Groundwater Authority, 2008; U.S. Environmental Protection Agency, 2021). The Aerojet General Corporation Superfund site is in the north central part of the South American study area (fig. 22). Data from the GAMA Groundwater Information System for all three study areas highlight the locations of monitoring wells where perchlorate was measured, which include monitoring wells downgradient from the Aerojet General Corporation Superfund site (fig. 22; California State Water Resources Control Board, 2021b). The USGS grid sites sampled in the South American study area were not close to the monitoring wells, and concentrations of perchlorate in the closest grid wells were either non-detections or low RCs (less than 0.6 µg/L; fig. 22). As discussed previously, the detection frequencies of perchlorate at concentrations greater than 0.1, 0.5, or 1 µg/L in the South American study area were consistent with or less than the predicted detection frequencies of perchlorate from natural sources.

Another potential anthropogenic source of perchlorate is the transport of natural perchlorate present in the unsaturated zone to the groundwater aquifer by irrigation recharge (Fram and Belitz, 2011). Perchlorate is also a natural minor component of salts associated with a fertilizer imported into the United States from the Atacama Desert in Chile (Böhlke and others, 2009). This Chilean fertilizer has been linked to elevated perchlorate concentrations in groundwater (Böhlke and others, 2009; Fram and Belitz, 2011). The import and use of Chilean nitrate in the United States as fertilizer began in the early part of the 20th century but has declined in recent years (Dasgupta and others, 2006).

The variety of sources of perchlorate to the environment confounds efforts to connect elevated concentrations of perchlorate observed in the SacMetro-DSA study unit to specific pathways of perchlorate contamination that could be mitigated. Perchlorate isotope analyses could potentially help identify sources of perchlorate, but such analyses are beyond the scope of this report.