Blank Samples

A variety of blank samples were collected to determine the extent environmental analytical results were affected by bias (contamination of environmental samples) introduced during field equipment cleaning, field sample collection, field sample processing, and laboratory analytical procedures. Blank samples, including field-equipment, source-solution, and trip blanks, were prepared with water that was certified to be free of analytes of interest (blank water) before submission to analytical laboratories for analysis. A brief description of each type of blank sample follows:

A quantified result in any blank sample was considered evidence that contamination could have biased groundwater sample results. Analytical results for groundwater samples were compared to the maximum quantified concentration in each blank sample collected during the same sampling event. EPA guidance (EPA, 1989, p. 5–17) recommends a reported concentration in an environmental sample that is less than five times the concentration in a related blank sample be qualified. In this report, water-quality results are presented as they were received from the laboratory. Constituents that meet EPA guidance criteria have been qualified using a footnote in the water-quality data tables presented in this report. For data analysis purposes, these values were considered nondetections. All qualifications were based on quantified results in field-equipment blank or trip-blank samples. Constituents were not detected in the source-solution blank.

Twenty-four field-equipment blank samples were collected during this study (appendix table 1.4). One-third (8) of these field-equipment blank samples were from well JH–DI1 because it was the only well where samples were collected and analyzed for glycol compounds. The remaining field-equipment blank samples (16) were collected before environmental sample collection from 7 other monitor wells. Sixteen constituents were detected at low but measurable concentrations in field-equipment blank samples: 1 major ion (calcium), 1 uncharged species (silica), 2 nutrients (ammonia and total nitrogen), DOC, 6 trace elements (antimony, chromium, cobalt, copper, manganese, and strontium), and 5 organic compounds (dichloromethane, vinyl chloride, diethylene glycol, ethylene glycol, and propylene glycol; appendix table 1.4). Importantly, benzotriazoles were not detected in any blank samples. A brief description of these detections and resultant effects on environmental concentrations follows.

Concentrations of calcium and silica measured in field-equipment blank samples (four detections; two each) were low; two of the detections were below the LRL but above the long-term MDL. Both constituents were detected in environmental samples, but environmental sample results were not qualified because these constituents generally were detected at concentrations several orders of magnitude larger than the concentrations measured in the field-equipment blank samples.

Concentrations of nutrients (ammonia and total nitrogen) measured in field-equipment blank samples (7 detections) were low (appendix table 1.4); 6 of the 7 detections were below the LRL, but above the long-term MDL. Ammonia and total nitrogen results for environmental samples with detections less than five times the maximum concentration detected in the field-equipment blank sample for that sampling event were qualified using a footnote (appendix table 1.8).

Concentrations of DOC measured in field-equipment blank samples (two detections) were low; one of two detections was at the LRL but above the long-term MDL. DOC results for environmental samples with detections less than five times the maximum concentration detected in the field-equipment blank sample for that sampling event were qualified using a footnote (appendix table 1.8).

Concentrations of trace elements detected in field-equipment blank samples (antimony, chromium, cobalt, copper, manganese, and strontium; 11 detections total) were low; 6 of the 11 detections were below the LRL but above the long-term MDL. When detected in environmental samples, concentrations of antimony, chromium, cobalt, and copper were generally low (appendix table 1.8); consequently, concentrations for many environmental samples were qualified due to being less than five times the maximum detected concentration in the field-equipment blank for each associated sampling event. Environmental sample concentrations of manganese ranged from nondetections (below LRL) to greater than 1,000 mg/L, resulting in only one manganese concentration being qualified (appendix table 1.8). Environmental sample concentrations of strontium were high in relation to field-equipment blank sample concentrations, so no strontium concentrations were qualified.

Two field-equipment and two trip-blank samples were collected and analyzed for VOCs; both types of blank samples were collected before environmental sample collection at well JH–DI1 on August 22, 2012, and at well JH–4 on May 7, 2019. One VOC (dichloromethane) was detected in the field-blank sample and two VOCs (dichloromethane and vinyl chloride) were detected in the trip-blank sample collected at well JH-4 during May 2019 (appendix table 1.4), but not in the associated environmental sample (appendix table 1.8). In addition to VOCs, two field-equipment blank samples were analyzed for GRO and DRO; neither organic constituent was detected in the samples (appendix table 1.4).

Eight field-equipment blank samples were collected and analyzed for glycols during sampling at well JH–DI1 (types of glycol compounds analyzed varied during study; appendix table 1.4). Well JH–DI1 was the only well from which samples were collected and analyzed for glycol compounds during August 2012–20. Three different glycol compounds (total of five glycol detections) were detected in four of the eight field-equipment blank samples (appendix table 1.8). Of the five field-equipment blank detections, only one sample (collected August 13, 2019) had detections of a glycol compound (ethylene glycol) in both the field-equipment blank sample and the associated environmental sample; consequently, the environmental sample concentration was qualified (appendix table 1.3).

Replicate Samples

Replicate samples are two or more samples collected and processed sequentially, so that the samples are considered essentially identical in composition. By analyzing these samples for the same constituents, replicate sampling results can be used to evaluate the cumulative effects of field and laboratory procedures on the reproducibility (precision) of analyte concentration values reported by the laboratory. Although two replicate samples are not adequate to assess precision, they can be used to evaluate whether variability of results for the samples is within the range of expected precision based on the larger available dataset. Potential variability was estimated by comparing replicate groundwater samples. Twenty-two replicate samples were collected from eight monitor wells during 2012–20 and analyzed for many constituents (appendix table 1.5). Variability for each analyte was estimated as the relative percent difference (RPD) between constituent concentrations measured in the primary environmental sample and the replicate environmental sample using equation 1:

RPDs were calculated only for constituents with detections in both the environmental sample and associated replicate environmental sample. The RPD could not be calculated if a concentration was censored in either or both samples. For this study, RPD values greater than 20 percent indicated that analytical results might be affected by high variability. Calculated RPDs are presented in appendix table 1.5. Overall, analytical results for 16 constituents were qualified due to high variability. These results have been identified using a footnote in table 7 and appendix table 1.8. RPDs for environmental samples with associated replicate environmental samples (“replicate pairs”) generally were acceptable. Many of the replicate pairs with high variability were a result of low constituent concentrations in both the environmental and replicate samples. Small differences in low concentrations can result in large RPDs. Of all the replicate pairs collected during this study, only two were identified as having RPDs of concern. One replicate pair (well JH–DI1, August 22, 2012) indicated high variability for iron (RPD=88 percent), and the second replicate pair (well JH–DI1, April 30, 2013) indicated high variability for iron (RPD=163 percent) and manganese (145 percent; appendix table 1.5). These sample results with high RPDs were not rejected because these samples were reanalyzed by the NWQL, and the resulting concentrations were similar. During initial review of the sampling results, there was concern sediment might have broken through the 0.45-micron filter used for sampling; however, major ions analyzed with these constituents in the same samples had RPDs less than 10 percent. The only samples collected during August 2012 and April 2013 and associated with these higher RPDs were from well JH–DI1.

Major-Ion Balance

Major-ion data were quality-assured by calculating a cation-anion balance (“ionic charge balance”). The sum of concentrations of dissolved cations in milliequivalents per liter should equal the sum of concentrations of dissolved anions in milliequivalents per liter (Hem, 1985). The percent difference between the sum of concentrations of cations and anions in milliequivalents per liter was calculated using equation 2:

Ionic charge balances were calculated for groundwater samples with major-ion analyses. The ionic charge balances ranged from −3.08 to 6.61 percent. An ionic charge balance within ±5 percent is preferable (Friedman and Erdmann, 1982; Clesceri and others, 1998). Four ionic charge balances during this study slightly exceeded 5 percent and ranged from about 5.1 to 6.6 percent; these balances were considered acceptable for this study.

Dissolved Oxygen

DO concentrations measured each time a well was sampled were used as the first indication of the redox condition of water in each well (refer to “Determining Redox Conditions” section). Concentrations less than 0.5 mg/L indicated anoxic/reducing conditions, and greater concentrations indicated oxic conditions. Measured DO concentrations during 2011–20 ranged from <0.1 to 9.75 mg/L (median=1.65 mg/L; table 5; fig. 6A). DO concentrations indicative of anoxic/reducing conditions (DO<0.5 mg/L) were measured in 1 or more water samples from 7 of 10 sampled monitor wells, including all 6 monitor wells with benzotriazole detections (JH–1.5R, JH–1.5D, JH–2.5, JH–3, JH–3D, and JH–3.5) and 1 well without benzotriazole detections (well JH–4; fig. 6A; appendix table 1.7). DO concentrations in the 4 wells without detections of 4-MeBT (JH–1, JH–2, JH–4, and JH–DI1) were significantly higher (p-values<0.001, Peto-Peto test) than in the 6 wells with detections (median DO concentrations=6.6 and 0.4 mg/L, respectively; fig. 6A). DO concentrations in 3 of 4 wells without detections of 4-MeBT (JH–1, JH–2, and JH–DI1) were consistently oxic (DO>0.5 mg/L) during 2011–20 (fig. 6A).

Median DO concentrations in the four wells without detections of 4-MeBT ranged from 2.9 to 7.9 mg/L (wells JH–4 and JH–1, respectively; figs. 6A, 7). Median DO concentrations in the 6 wells with detections of 4-MeBT were mostly lower than in the 4 wells without detections and ranged from 0.2 to 3.4 mg/L (wells JH–1.5D and JH–2.5, respectively; figs. 6A, 7). Of these six wells, only well JH–2.5 had a median DO concentration greater than 1 mg/L (figs. 6A, 7). Wells with detections of 4-MeBT also had significantly lower median pH values, significantly higher median specific-conductance values, and significantly higher dissolved iron and manganese concentrations than wells without detections (p-values<0.05; figs. 6B–C, 7; also refer to the next two report subsections that follow, “pH” and “Specific Conductance,” and the “Dissolved Iron and Dissolved Manganese” section).

DO concentrations in the alluvial aquifer varied substantially (anoxic to oxic), both spatially and with time (figs. 6–8). Although anoxic conditions were present in seven wells, these conditions were not persistent during 2011–20 (fig. 8). For example, samples collected during 2011 from well JH–2.5 were all anoxic, whereas samples collected during 2012–20 were all oxic (fig. 8E). Conversely, samples collected during 2011–16 from well JH–4 were oxic, whereas samples collected during 2018 were suboxic or mixed (oxic-anoxic), mixed (anoxic and oxic-anoxic) during 2019, and both oxic and anoxic during 2020 (fig. 8I; no samples were collected from well JH–4 during 2017; also refer to the “Redox Conditions” section).

Although consistently oxic, DO concentrations varied in wells JH–1, JH–2, and JH–DI1 during 2011–20. For example, an upward trend in DO concentrations (p-value<0.001) was determined at well JH–2, and concentrations ranged from a low of 3.3 mg/L in November 2011 to a high of 7.1 mg/L in April and August 2017 (figs. 7, 8D; appendix table 1.9). Similarly, an upward trend in DO concentrations was determined at well JH–DI1, and a probable downward trend was determined at well JH–1 (figs. 7, 8A, J; appendix table 1.9).

Wright (2013) reported anoxic/reducing conditions for an area southwest of JHA operations that included six wells, all of which also had detections of 4-MeBT (JH–1.5R, JH–1.5D, JH–2.5, JH–3, JH–3D, and JH–3.5; fig. 7). Upward trends in DO concentrations during 2011–20 were determined at all six of these wells (p-values ranging from <0.001–0.021; figs. 7, 8B–C, E–H; appendix table 1.9). By 2020, DO in five of these wells (JH–1.5R, JH–1.5D, JH–2.5, JH–3D, and JH–3.5) had increased to concentrations greater than the 0.5-mg/L threshold for anoxic water, substantially decreasing the areal extent of groundwater with reducing conditions (fig. 8B–C, E–H). Although DO in well JH–3 was measured only once at a concentration greater than 0.5 mg/L (1.0 mg/L; May 21, 2020), an upward trend in DO concentrations during 2011–20 and change in the redox class from exclusively anoxic to also suboxic, mixed (oxic-anoxic), or oxic during 2018–20 indicated groundwater conditions were returning to conditions similar to background conditions measured in the alluvial aquifer upgradient from airport operations towards the end of the study period (fig. 8F; also refer to the “Redox Conditions” section). A downward trend in DO concentrations during 2011–20 was determined at well JH–4 (p-value=<0.001; fig. 8I), which is outside of the area with reducing conditions and detections of 4-MeBT (fig. 7). DO concentrations in 3 of 6 samples collected from well JH–4 during 2018–20 were less than the 0.5-mg/L threshold for anoxic water (fig. 8I).

pH

Values of pH measured during 2011–20 ranged from 7.1 to 8.1 (median=7.5; table 5). All measured pH values were within the EPA SDWR range of 6.5–8.5. Median pH for the 4 wells without detections of 4-MeBT (about 7.6) was significantly higher (p-values<0.001, rank-sum test) than median pH for the 6 wells with detections (about 7.4; fig. 6B). The lower pH in wells with detections of 4-MeBT may be indirectly attributable to 4-MeBT degradation. Biological degradation of organic compounds typically results in increases in microbial activity and alkalinity (Chapelle, 2001). Increases in alkalinity are due to the production of carbon dioxide as a byproduct of the increased microbial activity (respiration). Alkalinity helps maintain groundwater pH by buffering groundwater from acids (Hem, 1985). Consequently, groundwater in contaminated wells where anaerobic biological degradation of organic compounds such as 4-MeBT is occurring would be expected to have higher alkalinity (and correspondingly lower pH) than in uncontaminated wells. Median alkalinity for the 4 wells with detections of 4-MeBT (about 153 mg/L as calcium carbonate [CaCO₃]) was significantly higher (p-value<0.001, rank-sum test) than median alkalinity for the 6 wells without detections (about 122 mg/L as CaCO₃). This indicates increased microbial activity and associated carbon dioxide production may be responsible for the lower pH in wells with 4-MeBT.

An upward trend in pH was determined at 5 of 6 wells with detections of 4-MeBT (JH–1.5R, JH–1.5D, JH–2.5, JH–3, and JH–3D; figs. 7, 9B–C, E–G); no trend was determined at the sixth well (JH–3.5; figs. 7, 9H; appendix table 1.9). For the 4 wells without 4-MeBT, a probable downward trend in pH was determined at 1 well (JH–1), a downward trend was determined at 2 wells (JH–4 and JH–DI1), and no trend was determined at 1 well (JH–2; figs. 7, 9A, D, I–J; appendix table 1.9).

Specific Conductance

Specific-conductance values measured during 2011–20 ranged from 186 to 571 microsiemens per centimeter at 25 degrees Celsius (µS/cm; median=316 µS/cm; table 5; fig. 6). These values are within the range of values for specific conductance (112 to 863 µS/cm) reported by Nolan and Miller (1995) for water from 34 wells completed in Quaternary alluvium, colluvium, and gravel, and pediment, fan, and glacial deposits in Teton County, Wyo. Specific-conductance values generally were lower in wells without detections of 4-MeBT than in wells with 4-MeBT detections (fig. 6C). Median specific conductance for the 6 wells with detections of 4-MeBT (325 µS/cm) was significantly higher (p-values<0.001, rank-sum test) than median specific conductance for the 4 wells without detections (261 µS/cm; fig. 6C).

A downward trend in specific conductance was determined at all 6 wells with detections of 4-MeBT (JH–1.5R, JH–1.5D, JH–2.5, JH–3, JH–3D, and JH–3.5; figs. 7, 10B–C, E–H; appendix table 1.9). For the 4 wells without 4-MeBT, an upward trend in specific conductance was determined at 1 well (JH–4), a probable downward trend was determined at 1 well (JH–2), and no trends were determined at 2 wells (JH–1 and JH–DI1; figs. 7, 10A, D, I–J; appendix table 1.9).

Summary of Detections and Concentrations

Dissolved iron and manganese were detected in at least one sample collected from all 10 monitor wells during 2011–20, including 52 percent (93 of 179) of samples analyzed for iron and 64 percent (114 of 179) of samples analyzed for manganese (table 6; fig. 11). Concentrations of dissolved iron and dissolved manganese ranged from <3.2 (lowest reporting level) to 1,052 micrograms per liter (µg/L; median=5.72 µg/L), and from <0.15 (lowest reporting level) to 1,730 µg/L (median=1 µg/L), respectively (table 6; fig. 11).

Redox conditions, specifically low DO concentrations, were a strong indicator of high iron and manganese concentrations in the alluvial aquifer at the JHA during 2011–20. With few exceptions, dissolved iron and manganese concentrations were substantially higher when groundwater conditions near a well were anoxic, suboxic, or mixed, than when conditions were oxic (table 6). In fact, concentrations of dissolved iron and manganese typically were near or below LRLs in samples with oxic waters (DO greater than 0.5 mg/L, unless samples had mixed redox conditions), which were mostly from wells without detections of benzotriazoles (figs. 11–12; table 6). Except for 4 samples collected during 2018–20 from 1 well (JH–4), anoxic, suboxic, or mixed conditions were present at least once only in the 6 wells with detections of benzotriazoles, especially 4-MeBT (refer to the “Dissolved Oxygen” and “Redox Conditions” sections). Dissolved iron and manganese concentrations in samples from the 6 wells with detections of 4-MeBT were significantly higher (p-values<0.001, Peto-Peto test) than from the 4 wells without detections (fig. 11). These results mostly reflect the significantly lower DO concentrations (and associated anoxic conditions) in wells with detections of 4-MeBT than in wells without detections (refer to the previous “Physical Properties” and “Redox Conditions” sections and the next report subsection, “Relation to Geochemical Conditions;” fig. 6A; table 6).

Median dissolved iron concentrations for the six wells with detections of 4-MeBT ranged from 0.48 to 228 µg/L (wells JH–3.5 and JH–3D, respectively; table 6; figs. 7, 11). Median dissolved iron concentrations for three of four wells without detections (JH–1, JH–4, and JH–DI1) were low and ranged from 1.5 to 4.0 µg/L (wells JH–1 and JH–DI1, respectively; table 6; figs. 7, 11). Median dissolved iron concentrations for three wells (JH–1, JH–2.5, and JH–3.5) in table 6 are smaller than the lowest reporting level (<3.2 µg/L) and were imputed (statistically estimated; refer to the “Statistical Analysis” section). Median dissolved manganese concentrations for the 6 wells with detections of 4-MeBT ranged from 0.02 to 641 µg/L (wells JH–2.5 and JH–3, respectively), and median dissolved manganese concentrations for 3 of 4 wells without detections (JH–2, JH–4, and JH–DI1) were very low and ranged from 0.18 to 0.23 µg/L (wells JH–2 and JH–DI1, respectively; table 6; figs. 7, 11). The median dissolved iron concentration for well JH–2 and median dissolved manganese concentration for well JH–1 was not computed because greater than 80 percent of the values were censored (table 6; Helsel, 2012).

Dissolved iron concentrations exceeded the EPA SDWR of 300 µg/L (EPA, 2018) in about 11 percent (19 of 179) of samples collected during 2011–20 (table 6). Concentrations of dissolved iron greater than the EPA SDWR were identified only when groundwater conditions near a well were anoxic (DO<0.5 mg/L) and 4-MeBT was detected in the sample (table 6; fig. 11). Four of six wells with detections of 4-MeBT (JH–1.5D, JH–3, JH–3D, and JH–3.5) had one or more samples with dissolved iron concentrations greater than the EPA SDWR (table 6; fig. 11). For these four wells, the percentage of samples with dissolved iron concentrations greater than the EPA SDWR, from largest to smallest, was about 42 percent for well JH–1.5D (8 of 19 samples), 32 percent for well JH–3D (6 of 19 samples), 21 percent for well JH–3 (4 of 19 samples), and 6 percent for well JH–3.5 (1 of 16 samples; table 6).

Dissolved manganese was detected in more wells at concentrations above its respective EPA SDWR (7 wells) than dissolved iron (4 wells; table 6). Dissolved manganese concentrations exceeded the EPA SDWR of 50 µg/L (EPA, 2018) in 39 percent (70 of 179) of samples collected during 2011–20 (tables 5, 6). Concentrations of dissolved manganese greater than the EPA SDWR were identified when groundwater conditions in the vicinity of a well were anoxic, suboxic, or mixed and when 4-MeBT was detected in the sample except for four samples with anoxic or mixed conditions collected during 2018–20 from well JH–4, which never had a detection of 4-MeBT (table 6). Twenty-five percent (4 of 16) of samples from well JH–4 had dissolved manganese concentrations greater than the EPA SDWR (table 6). All six wells with detections of 4-MeBT had at least three samples with dissolved manganese concentrations greater than the EPA SDWR. For these six wells, the percentage of samples with dissolved manganese concentrations greater than the EPA SDWR, from largest to smallest, was about 90 percent for well JH–1.5D (17 of 19 samples), 79 percent for well JH–3 (15 of 19 samples), 77 percent for well JH–1.5R (13 of 17 samples), 68 percent for well JH–3D (13 of 19 samples), 31 percent for well JH–3.5 (5 of 16 samples), and 17 percent for well JH–2.5 (3 of 18 samples; table 6). Of note, well JH–1.5D had the highest percentage of samples with dissolved iron and manganese concentrations greater than their respective EPA SDWRs (42 and 90 percent, respectively).

Relation to Geochemical Conditions

Iron and manganese are naturally present in most aquifers, mostly because of dissolution and weathering of minerals in soils, sediments, and rocks (Hem, 1985). Geochemical conditions (redox conditions and pH) affect the solubility and mobility of iron and manganese in groundwater, and reduced species are much more soluble than the oxidized species (Hem, 1985; Chapelle, 2001; McMahon and Chapelle, 2008; Ayotte and others, 2011; DeSimone and others, 2014; McMahon and others, 2019).

During 2011–20, the frequency of detection and concentration of dissolved iron and dissolved manganese in individual wells (figs. 11, 13) varied in response to changing geochemical conditions of groundwater near a well, although much more so for redox than for pH, and more commonly in the 6 wells with detections of benzotriazoles than in the 4 wells without detections. Using data collected from all 10 monitor wells during 2011–20, this relation between redox conditions and concentrations of both constituents was reflected by significant negative correlations between concentrations of DO and dissolved iron and manganese (Kendall’s τ=−0.424 and −0.611, respectively; both p-values<0.001; fig. 12). When the sample data were subdivided by wells, DO concentrations were negatively correlated (p-values<0.05) with dissolved iron concentrations in 3 wells (JH–1.5D, JH–3, JH–3D; Kendall’s τ=−0.40 to −0.71) and negatively correlated (p-values<0.05) with dissolved manganese concentrations in 5 wells (JH–1.5R, JH–1.5D, JH–3D, JH–3.5, JH–4; Kendall’s τ=−0.50 to −0.69; appendix table 1.10). A nonsignificant (0.05≤p-value≤0.09) negative correlation between DO and dissolved iron concentrations was noted for 1 well (JH–1.5R; Kendall’s τ=−0.50; appendix table 1.10), and weak, significant negative correlations between DO and dissolved manganese concentrations were noted for 3 wells (JH–2.5, JH–3, and JH–DI1; Kendall’s τ=−0.32 to −0.33; appendix table 1.10).

Using data collected from all wells during 2011–20, pH values were weakly negatively correlated with dissolved iron concentrations (Kendall’s τ=−0.176, p-value<0.001) and negatively correlated with dissolved manganese concentrations (all sample data; Kendall’s τ=−0.359, respectively; p-value<0.001; fig. 14). When the sample data were subdivided by wells, pH values were negatively correlated (p-values<0.05) with dissolved iron concentrations in 3 wells (JH–1.5D, JH–3, JH–3D, Kendall’s τ=−0.51 to −0.58) and negatively correlated (p-values<0.05) with dissolved manganese concentrations in 5 wells (JH–1.5D, JH–3, JH–3D, JH–3.5, JH–4; Kendall’s τ=−0.51 to −0.61; appendix table 1.10).

As expected, given their redox-sensitive nature, dissolved iron and manganese were measured at higher concentrations when anoxic conditions in each well were strongly reducing, which most commonly was when manganese or iron reduction was identified as the predominant redox processes during 2011–20 (table 6). High dissolved iron concentrations greater than the EPA SDWR for iron typically indicated the presence of high dissolved manganese concentrations greater than the EPA SDWR for manganese, but the opposite did not occur as often (table 6). This observation is consistent with typical redox sequence progression after oxygen and nitrate have been reduced, in which manganese is reduced before iron (refer to the “Redox Conditions” section).

Redox conditions progressed beyond iron reduction near some wells during 2011–20, as indicated by the following characteristics of some samples:

High dissolved iron and manganese concentrations were common in samples with these characteristics (table 6). Except for the few samples where the predominant redox processes were identified as sulfate reduction or methanogenesis, the predominant redox processes identified for most samples with high dissolved iron and manganese concentrations were manganese or iron reduction (and infrequently, mixed redox conditions and processes), despite evidence of more reduced conditions in some samples (table 6). This indicates that parts of the aquifer screened by the wells may have contained waters that were very strongly reduced and, thus, “more reduced” than the redox process (manganese or iron reduction) indicated in table 6. Evidence of more than one redox process in water samples (including samples classified as mixed) could be explained by the mixing of waters from different redox zones upgradient from the well, a well screen open to more than one redox zone, or spatial variability in microbial activity/redox processes in the aquifer (Chapelle, 2001).

Wright (2013) observed that all wells with benzotriazole detections were downgradient from airport operations, including the airport terminal where most plane deicing was done before construction of the deicing pad during 2011. Reducing conditions present in an otherwise oxic aquifer indicate an upgradient or natural, in-place source of organic carbon or oxidizable minerals (Chapelle, 2001). No substantial anthropogenic sources of dissolved organic carbon were detected in the alluvial aquifer during 2011–20. DOC concentrations in the alluvial aquifer were low (typically <1 mg/L), despite significantly higher concentrations in wells with benzotriazoles (refer to the “Redox Conditions” section). Other potential sources of organic carbon such as human-made organic compounds (for example, VOCs, GRO, DRO, and glycols) were rarely detected (refer to the “Anthropogenic Compounds” section). Additionally, unconsolidated sand and gravel aquifers such as the Snake River alluvial aquifer tend to have very low amounts of sedimentary organic carbon (app. A of Interstate Technology & Regulatory Council, 2015). Consequently, the absence of substantial sources of organic carbon and the consistent detection of benzotriazoles in wells downgradient from airport operations during 2011–20 strengthens the conclusion of the previous study (Wright, 2013) that microbial oxidation of ADAFs produced anoxic conditions that lead to reductive dissolution of naturally occurring iron and manganese from aquifer sediments. However, microbial oxidation of ADAFs was not likely the cause of suboxic, mixed, and anoxic conditions and resultant high manganese concentrations in well JH–4 during 2018–20 because no benzotriazole compounds (or any other anthropogenic compounds) were detected in the well (table 6). Well JH–4 is a flush-mounted well next to a staging area that has been used to support JHA construction activities. The staging area included a concrete plant during 2015–17. In 2017, a temporary building associated with the concrete plant was placed over well JH–4, preventing access for data collection. These and other construction activities since the well was installed have increased the surrounding grade and land surface, causing the well to be in a low point in this area. Thus, it is possible that surface water may have collected and entered the well, potentially transporting surficial organic matter into the aquifer that subsequently was oxidized and created the reducing conditions (and resultant high manganese concentrations) observed at this well.

Positive correlations between dissolved iron and manganese concentrations indicate a common source of both trace elements in the alluvial aquifer sediments underlying the airport. Using all data collected during 2011–20, a significant positive correlation was found between dissolved iron and manganese concentrations (Kendall’s τ=0.522, p-value<0.001). When the sample data were subdivided by wells, dissolved iron concentrations were positively correlated (p-values<0.05) with dissolved manganese concentrations in four wells: JH–1.5R (Kendall’s τ=0.42), JH–1.5D (Kendall’s τ=0.92), JH–3 (Kendall’s τ=0.82), and JH–3D (Kendall’s τ=0.94; appendix table 1.10). These four wells had varying anoxic, suboxic, or mixed redox conditions during most of 2011–20 and the only exceedances of the EPA SDWR for iron and most of the exceedances of the EPA SDWR for manganese of all sites sampled (83 percent; table 6). Oxidized forms of iron and manganese commonly co-occur together in aquifer sediments, most commonly as oxyhydroxide coatings on sediment grains, but also as minerals composing aquifer sediments (Chapelle, 2001). Thus, the positive correlations between dissolved iron and manganese concentrations indicate aquifer sediment mineralogy, in combination with reducing conditions induced by microbial degradation of ADAFs, may have largely controlled concentrations of both trace elements in parts of the alluvial aquifer at the JHA during 2011–20. However, as described previously, there was no evidence indicating degradation of ADAFs was the cause of suboxic, mixed, and anoxic conditions and resultant high manganese concentrations in the aquifer in the vicinity of well JH–4 during 2018–20.

Trends in Dissolved Iron and Dissolved Manganese Concentrations

Trends in dissolved iron and manganese concentrations during 2011–20 were evaluated with the nonparametric seasonal Kendall test (refer to the “Statistical Analysis” section). The trend test was applied to dissolved iron and manganese concentrations at nine wells. Because of severe censoring (greater than 80 percent of samples censored), trend tests for dissolved iron and manganese were not done for wells JH–2 and JH–1, respectively. Trends in dissolved iron and manganese concentrations for 2011–20, including results of seasonal Kendall tests, are shown in a series of graphs for individual wells in figure 13 and spatially for all wells in figure 7. Results of seasonal Kendall tests for dissolved iron and manganese (including p-values, Kendall’s τ values, and Theil-Sen and Akritas-Theil-Sen slopes) are tabulated in appendix table 1.9.

Downward and probable downward trends in dissolved iron and manganese concentrations during 2011–20 were determined at six of the monitor wells at the JHA (figs. 7, 13; appendix table 1.9). The results of the seasonal Kendall tests for the 6 wells with benzotriazole detections indicated downward trends in iron concentrations at 3 wells (JH–1.5D, JH–3, and JH–3D), probable downward trends at 2 wells (JH–2.5 and JH–3.5), and no trend at 1 well (JH–1.5R; figs. 7, 13B–C, E–H; appendix table 1.9). No trends in dissolved iron concentrations were determined at the four wells without benzotriazole detections (JH–1, JH–2, JH–4 and JH–DI1; figs. 7, 13A, D, I–J; appendix table 1.9). The results of the seasonal Kendall tests for the 6 wells with benzotriazole detections indicated downward trends in dissolved manganese concentrations at 5 wells (JH–1.5D, JH–2.5, JH–3, JH–3D, and JH–3.5) and a probable downward trend at 1 well (JH–1.5R; fig. 13B–C, E–H; appendix table 1.9). No trends in dissolved manganese concentrations were determined at 3 of 4 wells without benzotriazole detections (JH–1, JH–2, and JH–DI1), but an upward trend was determined at 1 well (JH–4; figs. 7, 13A, D, I–J; appendix table 1.9). The upward trend in dissolved manganese concentrations in well JH–4 corresponded with downward trends in DO and pH, and an upward trend in specific conductance (refer to the “Physical Properties” section; fig. 7; appendix table 1.9).

The downward and probable downward trends in dissolved iron and manganese concentrations at individual wells with benzotriazole detections mostly corresponded with upward and probable upward trends in DO concentrations and pH values (refer to the “Physical Properties” section), downward and probable downward trends in specific-conductance values (refer to the “Physical Properties” section), and 4-MeBT concentrations (and nondetections of 5-MeBT by 2018; refer to the “Anthropogenic Compounds” section) for the same wells (fig. 7). Collectively, these trends indicated groundwater-quality conditions near all wells with benzotriazole detections were returning to water-quality conditions similar to background water-quality conditions measured in the alluvial aquifer upgradient from airport operations, potentially in response to the changes in the ADAF formulation used by the JHA, many JHA infrastructure improvements made during 2011–20, and degradation of existing ADAFs in subsurface soils and groundwater, or some combination of these possibilities.

Out of the 6 wells with detections of benzotriazoles, all of which had 1 or more detections of dissolved iron or manganese at concentrations greater than respective EPA SDWRs during 2011–20, only 1 well (JH–1.5R) had detections of either constituent at concentrations greater than these standards by 2020 (fig. 13; table 6). In 2020, dissolved manganese concentrations in well JH–1.5R (83.2 µg/L in May and 51.2 µg/L in July) were higher than the EPA SDWR of 50 µg/L but were substantially lower than the highest concentration (1,639 µg/L, November 2011) measured in that well during 2011–20 (table 6). Of the remaining 5 wells with benzotriazole detections during 2011–20, dissolved iron and manganese typically were detected at low concentrations or were no longer detected above reporting levels by 2020 (table 6); all but 1 of these wells (well JH–3.5) had downward trends in 4-MeBT concentrations (refer to the “Anthropogenic Compounds” section).

Volatile Organic Compounds, Gasoline-Range Organics, and Diesel-Range Organics

As part of the previous study (Wright, 2013), groundwater samples from 10 monitor wells collected during June and July 2011 and April 2012 were analyzed for VOCs, GRO, and DRO. Three VOCs (benzene, ethylbenzene, and total xylene) and GRO were detected infrequently at low, estimated concentrations in at least one sampled well. DRO was not detected in any of the wells. Benzene was detected at a low estimated concentration of 0.09 µg/L in the sample from well JH–1.5D collected during April 2012; however, this value was less than five times the concentration measured in the associated trip blank, so the detection was qualified and considered questionable (Wright, 2013). Ethylbenzene was detected at an estimated concentration of 0.11 µg/L in the June 2011 sample collected from well JH–2.5, and total xylene was detected at estimated concentrations of 0.26 and 0.20 in the June and July 2011 samples collected from well JH–2.5, respectively. GRO was detected at an estimated concentration of 11 µg/L in the July 2011 sample collected from well JH–1.5D.

Groundwater samples from two monitor wells (JH–DI1 and JH–4) collected during August 2012–August 2020 were analyzed for VOCs, GRO, and DRO (table 1). Samples from well JH–DI1 were initially analyzed for these and other constituents to determine the “baseline” groundwater-quality conditions downgradient from and before use of the airport deicing pad, and subsequently to monitor for their presence after deicing activities began at the facility. Unlike well JH–DI1, these organic compounds were added to the analyte list for well JH–4 to determine if their presence was responsible for the increasingly reduced conditions observed in this well during the study period (refer to the “Redox Conditions” section). DO concentrations in well JH–4 significantly decreased, whereas dissolved manganese concentrations significantly increased during 2011–20 (refer to the “Physical Properties” and “Dissolved Iron and Dissolved Manganese” sections), despite no detections of benzotriazoles in samples from the well (table 7).

The groundwater sample collected from well JH–DI1 on August 22, 2012, was analyzed for VOCs, GRO, and DRO. GRO was detected at a low, estimated concentration (16 µg/L) in the environmental sample, but not in the corresponding replicate sample (appendix tables 1.3, 1.5). Because GRO was detected in the environmental sample, but not in the corresponding replicate sample, an additional (confirmatory) sample from well JH–DI1 was subsequently collected on April 30, 2013. GRO and BTEX compounds were not detected in the sample, indicating they were likely not present in the aquifer at this location (appendix table 1.3).

Samples for VOCs, GRO, and DRO were collected twice from well JH–4 during 2019. GRO was detected at low, estimated concentrations in the environmental sample (30 µg/L) and corresponding replicate sample (19 µg/L) collected on August 13, 2019 (appendix tables 1.3, 1.5). Because the environmental sample and replicate sample values were estimated concentrations, and the RPD (about 45 percent; appendix table 1.5) between the samples was large (greater than 20 percent; refer to the “Quality Assurance and Quality-Control Sample Collection” section), there is considerable uncertainty associated with both concentration values reported by the analyzing laboratory.

Glycols

Although glycols were not detected during the two previous studies (Wright, 2010, 2013), samples collected from well JH–DI1 were analyzed for glycol compounds during August 2012–20 to document water-quality conditions near the well and to facilitate comparison with future water samples collected to monitor for potential groundwater-quality changes due to deicing activities at the upgradient deicing pad installed during 2011 (fig. 5). Glycol compounds were detected in five samples collected from well JH–DI1 during August 2012–August 2020 (appendix table 1.3), but evaluation of the field and laboratory quality-control sample data associated with the environmental samples indicated these five detections may be due to contamination introduced during sampling or laboratory analysis, or due to other laboratory-related issues. However, the available information is not sufficient to conclusively determine the presence or absence of glycols in the alluvial aquifer near well JH–DI1.

Two glycol compounds (diethylene glycol and triethylene glycol) that are not components of deicers were detected at the JHA. Diethylene glycol was detected at a concentration of 9.8 µg/L in 1 of 11 environmental samples from well JH–DI1 (August 1, 2017) analyzed for the constituent, but it was not detected in the corresponding replicate sample (appendix tables 1.3, 1.5). Triethylene glycol was detected in 1 of 11 environmental samples from well JH–DI1 analyzed for the constituent at an estimated concentration of 2.7 µg/L (July 13, 2013; appendix table 1.3). The compound also was detected in an associated laboratory blank sample (appendix table 1.3), indicating the detection likely was attributable to laboratory contamination.

Ethylene glycol, a component in deicers, was detected in 3 of 17 samples from well JH–DI1 that were analyzed for the constituent at estimated concentrations of 0.53 µg/L (April 22, 2014), 2.9 µg/L (August 13, 2019), and 2.7 µg/L (May 19, 2020; appendix table 1.3). Two of three ethylene glycol detections (estimated concentrations: 0.53 µg/L, April 22, 2014, and 2.7 µg/L, May 19, 2020) were in the environmental samples but not in the corresponding replicate samples (appendix tables 1.3, 1.5). The remaining ethylene glycol detection (estimated concentration: 2.9 µg/L, August 13, 2019) was at a concentration less than five times the maximum concentration measured in an associated laboratory blank sample (appendix table 1.3; also refer to the “Quality Assurance and Quality-Control Sample Collection” section).

Benzotriazole Compounds

Samples for analysis of benzotriazoles (1H-BT, 4-MeBT, and 5-MeBT) were collected each time a monitor well was sampled during 2011–20. One set of benzotriazole analyses for well JH–DI1 (sample from July 31, 2013) is missing because the sample bottle broke during transit to the analyzing laboratory.

Only two of the three benzotriazole compounds measured (4-MeBT and 5-MeBT) were detected in wells sampled during 2011–20 (table 7). Both benzotriazoles were detected in 7 of 10 monitor wells sampled during 2011–20 (table 7). Benzotriazoles were detected twice from well JH–2 at low concentrations at or slightly greater than the LOD (4-MeBT, July 9, 2016, and 5-MeBT, July 29, 2014). However, both well JH–2 benzotriazole detections are considered suspect for the following reasons: (1) 4-MeBT and 5-MeBT were detected only once each, unlike all other wells with detections where both compounds were detected more frequently; (2) water from well JH–2 was consistently oxic with nondetectable to low iron and manganese concentrations during 2011–20, unlike all other wells with detections, which had anoxic waters and high iron and manganese concentrations for at least some period; (3) 4-MeBT was not detected in the sample with 5-MeBT, unlike all other samples when 5-MeBT was detected; and (4) the 4-MeBT concentration is at the LOD (table 7). For these reasons, both detections were treated as nondetections in this report for summary and analysis purposes because the available information was insufficient to conclusively determine if the detections represented rare occurrence in the alluvial aquifer near well JH–2 or were due to contamination introduced during sampling or laboratory analysis. Therefore, all summary and analysis of data in this report was done assuming benzotriazoles were detected in only six wells. The six remaining wells with benzotriazole detections included wells JH–1.5R (and its predecessor well JH–1.5), JH–1.5D, JH–2.5, JH–3, JH–3D, and JH–3.5 (table 7).

Concentrations of 4-MeBT ranged from less than the LOD (0.35 µg/L) to 13 µg/L (well JH–1.5R, November 2, 2011), and many detections were an order of magnitude greater than the LOD (table 7; fig. 15). Median 4-MeBT concentrations ranged from 0.55 to 3 µg/L (wells JH–3.5 and JH–1.5R, respectively; figs. 7, 15). Concentrations of 5-MeBT ranged from less than the LOD (range=<0.25 to <1.4 µg/L, but mostly <0.25 µg/L) to 2.6 µg/L (well JH–1.5R, April 24, 2014, and well JH–2.5, April 23, 2014; table 7). The benzotriazole 4-MeBT was detected more frequently and consistently at higher concentrations than 5-MeBT (table 7). In fact, many detections of 5-MeBT were reported as estimated concentrations by the analyzing laboratory, and 5-MeBT was not detected in any well after 2017 (table 7). This pattern of more frequent 4-MeBT detections and 4-MeBT concentrations higher than 5-MeBT concentrations is consistent with other benzotriazole research, which determined 5-MeBT is more biodegradable than 4-MeBT (Cornell, 2002; Weiss and Reemtsma, 2005).

Based on samples from monitor well clusters 1.5 and 3, benzotriazole concentrations generally decreased with depth during 2011–20. At well cluster 1.5, 4-MeBT concentrations in the deeper well (JH–1.5D) were lower than in the shallower well (JH–1.5 [2011] and JH–1.5R [2012–20]) for 15 of 19 (about 79 percent) sampling events (table 7). Concentrations of 4-MeBT in the deeper well at well cluster 3 (JH–3D) were lower than in the shallower well (JH–3) for all 19 sampling events (table 7).

Benzotriazoles were detected only in wells that experienced anoxic/reducing (including suboxic and mixed) conditions for at least some period during 2011–20 (table 6; also refer to the “Redox Conditions” section). Using data collected from all six wells with benzotriazole detections during 2011–20, 4-MeBT concentrations were negatively correlated with DO concentrations (Kendall’s τ= −0.467, p-value<0.001; fig. 16A). When the sample data were subdivided by wells, 4-MeBT concentrations in 4 of 6 wells (JH–1.5R, JH–1.5D, JH–2.5, and JH–3D) were negatively correlated (p-values<0.05) with DO concentrations (Kendall’s τ= −0.36 to −0.66; appendix table 1.10).

Concentrations of 4-MeBT were evaluated in relation to pH values. Using data collected from all six wells with benzotriazole detections during 2011–20, 4-MeBT concentrations were negatively correlated with pH values (Kendall’s τ= −0.334, p-value<0.001; fig. 16B). When the sample data were subdivided by wells, 4-MeBT concentrations in 4 of 6 wells (JH–1.5D, JH–2.5, JH–3, and JH–3D) were negatively correlated (p-values<0.05) with pH values (Kendall’s τ= −0.35 to −0.42; appendix table 1.10).

Trends in 4-MeBT concentrations for the six wells with benzotriazole detections during 2011–20 were evaluated with the seasonal Kendall test (refer to the “Statistical Analysis” section). Results of seasonal Kendall tests (including p-values, Kendall’s τ values, and Theil-Sen and Akritas-Theil-Sen slopes) are tabulated in appendix table 1.9. Trends in 4-MeBT concentrations for 2011–20, including results of seasonal Kendall tests, are shown in a series of graphs for individual wells in figure 17 and spatially for all wells in figure 7. The results of the seasonal Kendall tests indicated downward trends in 4-MeBT concentrations at 4 wells (JH–1.5D, JH–2.5, JH–3, and JH–3D), a probable downward trend at 1 well (JH–1.5R), and no trend at 1 well (JH–3.5; figs. 7, 17; appendix table 1.9). No benzotriazoles were detected in well JH–3.5 in 2020. The downward and probable downward trends in 4-MeBT concentrations at wells JH–1.5D, JH–2.5, JH–3, and JH–3D (and nondetections of 5-MeBT after 2017; fig. 17; table 7) corresponded with numerous other trends indicating water quality was returning to conditions similar to background water-quality conditions in these wells during 2011–20, including upward trends for DO and pH, and downward trends for specific conductance, dissolved iron, and dissolved manganese (refer to the “Physical Properties” and “Dissolved Iron and Dissolved Manganese” sections; appendix table 1.9).