A 50-year record of NOx and SO2 sources in precipitation in the Northern Rocky Mountains, USA

Ice-core samples from Upper Fremont Glacier (UFG), Wyoming, were used as proxy records for the chemical composition of atmospheric deposition. Results of analysis of the ice-core samples for stable isotopes of nitrogen (δ15N, ) and sulfur (δ34S, ), as well as and deposition rates from the late-1940s thru the early-1990s, were used to enhance and extend existing National Atmospheric Deposition Program/National Trends Network (NADP/NTN) data in western Wyoming. The most enriched δ34S value in the UFG ice-core samples coincided with snow deposited during the 1980 eruption of Mt. St. Helens, Washington. The remaining δ34S values were similar to the isotopic composition of coal from southern Wyoming. The δ15N values in ice-core samples representing a similar period of snow deposition were negative, ranging from -5.9 to -3.2 ‰ and all fall within the δ15N values expected from vehicle emissions. Ice-core nitrate and sulfate deposition data reflect the sharply increasing U.S. emissions data from 1950 to the mid-1970s.


Introduction
The chemical quality of snowfall deposited in high-elevation areas in the Rocky Mountain region can be affected by energy generation and associated population growth [1,2]. High elevation areas in the Wind River Range (WRR) of Wyoming (figure 1) exceed 4 km above sea level and are adjacent to areas of accelerating energy development [3]. For example, over 3,000 natural gas wells are being installed in the Green River Basin, directly west of the WRR. Full development of the Jonah gas field could result in the production of 1,480 metric tons/yr of NO x and 25.7 metric tons/yr of SO 2 [3].
Thin soils and dilute surface-water systems in highelevation areas have limited capacities to buffer increased acidity associated with the airborne contaminants of NO x and SO 2 . Trends in precipitation chemistry at NADP/NTN sites in the western United States have indicated an increase in total N deposition and a decrease in SO 2− 4 deposition from 1981-1998 [4]. In addition to monitoring trends in N and S deposition, the isotopic composition of snow, firn, and ice has been used to differentiate natural and anthropogenic solute sources. Stable isotope ratios of sulfur in SO 2− 4 , expressed as δ 34 S, were monitored in bulk snowpack samples collected from a network of 52 high-elevation sites in the Rocky Mountains from 1993 to 1999 [5]. The δ 34 S values indicated that snowpack S in high-elevation areas is primarily derived from anthropogenic sources [5]. Changes in δ 34 S values in firn and ice-core samples have been used to reconstruct changes in sulfate sources to central Asia, Greenland, and Antarctica. The variation in δ 34 S values in a firn core from central Asia allowed for the identification of S derived from marine evaporites (+15 ‰) during high dust deposition events and anthropogenic emissions (+5.4 ‰) [6]. Preindustrial δ 34 S signatures in Greenland ice cores were comprised of marine biogenic emissions, continental dust sources, background volcanism, and continental biota [7]. Beginning in 1870 A.D., δ 34 S signatures in the ice-core samples indicated anthropogenic S sources.
In Antarctica, shallow firn cores collected from the South Pole contained δ 34 S values that were used to confirm and contrast the different S isotopic signals between the low-latitude Agunge volcanic eruption in 1963 and background marine biogenic sulfate [8]. Composited icecore samples representing coastal and plateau regions on the Antarctic ice sheet were found to have similar δ 34 S values over the past 1,100 years, indicative of no temporal change in influencing sources of S [9]. Values of δ 34 S in two ice cores from east Antarctica over the complete glacial/interglacial cycle were significantly lighter than previous measurements of δ 34 S from the South Pole [10]. A likely mechanism for the observed isotopic difference was Rayleigh-type fractionation as S species are oxidized and transported toward the East Antarctic Plateau [10]. Measurements of δ 34 S values in an ice core collected from west Antarctica were found to be a mixture of marine and volcanic S sources during the time period from 1935-1976. [11].
Stable isotopes have also been used to gain a better understanding of N sources in atmospheric deposition. Historical records of the isotopic composition of N 2 O in trapped gases (ice cores) from Greenland and interstitial air (snowpack) from the South Pole have been used to differentiate between natural and anthropogenic sources [12]. The δ 15 N value of atmospheric N 2 O has dropped by 1.7 ‰ during the 20 th century, likely due to increasing agricultural activities [12]. In support of these measurements, simulations of N 2 O (g) have indicated a -1.8 ‰ shift in δ 15 N over the last two centuries, primarily due to anthropogenic influences [13]. The δ 15 N of NO − 3 in wet deposition from 33 sites in the northeastern United States was strongly correlated with the location of coal-fired powerplants [14]. Ice-core samples from Greenland indicated decreasing δ 15 N (NO − 3 ) values with increasing NO − 3 concentrations; however, more information is needed to confirm this trend and differentiate source area vs. post-depositional processes during the firn-to-ice transition [15]. A more recent icecore record from Greenland, spanning deposition from 1718 to 2006, revealed a clear trend of decreasing δ 15 N ( NO − 3 ) values from 11 ‰ (pre-industrial) to -1 ‰ (~1996-2006) [16].
Although previous studies have shown that δ 15 N ( NO − 3 ) values of snow and ice samples have excellent potential for providing information on NO − 3 sources, post-depositional changes in the concentration and isotopic composition of NO − 3 needs to be considered [17,18]. Year-round measurement of δ 15 N (NO − 3 ) values in snow pits from Dome C, Antarctica, indicated strong enrichment relative to atmospheric NO − 3 and loss of NO − 3 mass from the snow surface due to UV-photolysis [17]. In contrast to the Antarctica results, a similar study in Greenland [18] indicated minimal influence of photolysis on the isotopic composition of δ 15 N (NO − 3 ) in firn and ice. One possible reason for the different research results between the two studies may be the higher snow accumulation rates at the Greenland study site [18]. Higher rates of snow accumulation were found to mitigate the magnitude of post-depositional processing and loss of NO − 3 in the snowpack [19].  Glaciers present in the high elevation regions of the WRR, Wyoming, present a unique opportunity to couple short-term (1980 to present) NADP/NTN data (sites WY06 and WY98) trends in the chemistry of atmospheric deposition with similar and longer term data preserved in glacial ice from the nearby UFG (figure 1). The UFG is the only glacier within the continental U.S. where ice cores have been documented to contain paleoenvironmental and paleoclimatological records [20][21][22][23][24][25][26][27]. Characteristics present at UFG conducive to preserving paleonvironmental signals include: (1) icecoring site altitudes that exceed 4 km above sea level (ASL) to minimize meltwater modification of the snow and ice chemistry and (2) large ice thicknesses (ranging from 60 to 172 m in the upper half of the glacier) to provide long-term paleoenvironmental records.
Ice cores exceeding 160 m in length were recovered from UFG in 1991 and 1998 [23,24]. The 1991 ice core was estimated to contain 250 years of record as determined from carbon-14 dating of an insect leg recovered from near the bottom of the core [23]. This length of record was subsequently confirmed by a higher-resolution chronology established with continuous electrical conductivity measurements [26]. On the basis of these data, ice at the bottom of UFG in proximity to the 1991 and 1998 drilling sites was likely deposited as snow prior to 1710 AD.
The overall objective of this paper is to demonstrate the effectiveness of using the isotopic composition and concentration of NO − 3 and SO 2− 4 in ice-core samples collected from the UFG to supplement and extend existing NADP/NTN records. Specific objectives are to: (1) determine atmospheric deposition sources since the late 1940s by determining the isotopic values of δ 34 S (SO 2− 4 ) and δ 15 N (NO − 3 ) in ice-core samples; (2) reconstruct changes in SO 2− 4 and NO − 3 wet deposition rates from ice-core samples representative of snow deposited from the early-1950s to early-1990s; and (3) couple the results from objectives 1 and 2 with existing NADP/NTN deposition records and NO X /SO 2 emission records to identify processes controlling long-term trends in the loading of S and N at high-elevation sites.

Field
Ice cores were collected from UFG in 1991 and 1998 using a thermal drill as described in [23]. The ice cores were collected in 1-and 2-m segments. On-site processing of the cores included visual inspection, logging, and density determinations by personnel wearing Tyvek suits and powder-free Latex gloves. The cores were sealed in polyethylene bags, placed in plastic core tubes, and stored in snow vaults until removal from the site to a freezer truck via a 10-minute helicopter flight. The UFG ice cores are currently archived at the National Ice Core Laboratory (NICL) in Lakewood, Colorado.

Laboratory
Ice-core samples were melted according to strict protocols [21] to minimize sample contamination. Ice cores were subsampled using a bandsaw frequently cleaned with ethanol in cold room laboratories at NICL. Multiple core sections from each interval were composited in order to obtain sufficient S and N mass for isotopic analyses. The surface ice from each subsample was scraped away with a stainless steel microtome. Each ice sample was thoroughly rinsed with ultrapure (18.0 megaohm) deionized water and placed in a prerinsed and covered plastic container. Each sample was allowed to melt at room temperature for one hour (or until approximately 15 mL of meltwater had accumulated). After this initial melt period the sample was rinsed in the accumulated meltwater, and the melt was discarded. The remaining sample was allowed to melt in the covered plastic container at room temperature.
For S isotopic analysis, meltwater volumes corresponding to 90-100 micrograms of sulfate (0.5-6 L) were evaporated to dryness within covered canisters flushed by dry nitrogen. The samples were transferred from 4-L Pyrex beakers to smaller Savillex beakers when they had been reduced to about 100 mL. Drydown salts were redissolved in a few mL of deionized water which was pipetted into tin capsules and evaporated a few hundred microliters at a time. One to 2 mg of powdered V 2 O 5 was added and the capsules were pinched closed.
Filled capsules were combusted in an elemental analyzer to produce SO 2 for analysis by isotope ratio monitoring using a Micromass Optima [28]. The elemental analyzer-mass spectrometer system was standardized using reference materials previously calibrated against IAEA-SO-5 and NBS 127 assuming δ 34 S values for them of -34.05 and 21.1 ‰, respectively [29]. To evaluate accuracy, synthetic Ca 2+ -Na + -K + solutions were prepared with reagents of known isotopic composition and then dried and analyzed. Sodium-dominated solutions gave δ 34 S values indistinguishable from the reagents used to prepare them; however, solutions with Na + :Ca 2+ ratios equivalent to the UFG samples were about 1 ‰ low. The cause of this inaccuracy is uncertain. No attempt was made to correct the UFG measurements for a possible similar inaccuracy. The S isotopic compositions are reported in δ-notation relative to Vienna Canon Diablo troilite. Reproducibility for triplicate analysis of sample ISO-1 was ±0.4 ‰ (one standard deviation).
For N isotopic analysis, meltwater volumes corresponding to 1.2 micrograms of nitrate (6-80 mL) were evaporated to <5 mL by the same method used for sulfur. Nitrous oxide (N 2 O) was produced by the denitrifier method using the Pseudomonas chlororaphis bacterium (ATCC 43928) [30,31], and the isotope ratios were measured using a ThermoFinnigan Delta Plus XL equipped with an autosampler and gas bench device. To calibrate the measurements and correct for media blanks, aliquots of the reference materials USGS32 KNO 3 (δ 15 N = 180 ‰), USGS34 KNO 3 (δ 15 N = -1.8 ‰), and USGS35 NaNO 3 (δ 15 N = 2.7 ‰) were analyzed along with the unknowns.
Because the NO − 3 in UFG samples is atmospheric in origin, it is likely to contain excess 17  in UFG samples has Δ 17 O values of 30 ‰ or lower, the error in our δ 15 N results is no more than 1.6 ‰. The N isotopic compositions are reported in δ-notation relative to atmospheric N 2(g) . Reproducibility for triplicate analysis of USGS34 was ±0.27‰ (one standard deviation).
After filtration with 0.45 μm disposable polysulfone filters, a modified ion chromatography procedure for low ionic strength samples was used for the analysis of NO − 3 and SO 2− 4 [38]. A Dionex DX-120 was configured with a 250 μL loop to increase instrument sensitivity. Ten percent of each sample batch consisted of quality control standards. Seven-point calibration curves covering the range of solutes being measured were run daily. Each batch contained sets of blanks (18.2 megaohm deionized water) and a series of USGS standard water reference samples (SRWS) at intervals of approximately every 10 ice-core samples. SRWS and blanks were monitored during sample runs. If the SRWS were not within 10% of expected values or the blanks were above detection limits, the sample run results were rejected and repeated.
All chemical data from the ice-core analyses can be found in additional file 1.

Results and Discussion
Ice-core chronology Tritium, chloride-36, and carbon-14 age-dating methods [23,39] were combined with the timing of major volcanic eruption signals that were evident in the core [26] to establish an ice-core chronology. The polynomial fit for the age-depth profile in the ice core [26] was determined to be Age (in years) = 0.00739(D) 2 where D is depth below glacier surface, in m. This age-depth profile is in excellent agreement with known time horizons in the ice core that include (1)   Plot of reported volcanic events and isotopic age dates used to generate a polynomial fit for an age-depth profile of the Upper Fremont Glacier ice core collected in 1991 [26], Wind River Range, Wyoming. A slightly modified age-depth profile was developed for the 1998 ice core (DH-98-4) to account for the additional snow deposited between 1991 and 1998. concentrations [27] and elevated electrical conductivity, SO 2− 4 , and Cl - [26]. Ice samples collected from a 1980 deposition horizon on Knifepoint Glacier, approximately 4 km southwest of UFG, contained ash material that had morphology and chemical composition similar to Mt. St. Helens ash [20].
Except for the elevated δ 34 S value in the layer corresponding to atmospheric deposition from Mt. St. Helens, the remaining δ 34 S values in the ice-core samples ranged from 5.1 to 6.8 ‰ ( figure 3). The δ 34 S values in the UFG ice core were substantially enriched relative to biogenic SO 2− 4 sources (-2.4 ‰) [41] and massive sulfide deposits in Arizona (0 ‰) [42]. The δ 34 S values of massive sulfide deposits were assumed to be representative of smelter emissions associated with ore processing in Arizona and New Mexico [5]. Marine aerosols, which have a δ 34 S value of +20 ‰ [5], could not be a large source of SO 2− 4 to UFG due to the much lighter δ 34 S isotopic values found in the ice.
The WRR and UFG are in close proximity to large expanses of arid and semi-arid lands composed of soils, sedimentary rocks, and dry lakebeds containing gypsum and anhydrite [5]. Therefore, dry deposition of dust on the surface of UFG during the summer and dust solute sources associated with snow deposition could comprise a large natural SO 2− 4 source and corresponding isotopic signal. Marine and non-marine evaporites have δ 34 S values that range from +9 to +33 ‰ [43] and are much heavier relative to δ 34 S values found in UFG ice-core samples Coal from southern Wyoming, commonly used in electricity-producing power plants in close proximity to UFG, has a mean δ 34 S composition of 5.1 ‰ for organic S [45], similar in isotopic composition to the δ 34 S values in ice-core samples from UFG ( figure 3). Assuming limited isotopic fractionation during coal combustion, the δ 34 S isotopic composition in the ice-core samples indicates that coal-fired powerplant emissions could be a dominant source of S deposition to high-elevation areas of the WRR. Based on U.S. emissions data compiled for 1998, electric utilities contributed the majority of SO 2 emissions, representing 68% of total national SO 2 emissions [46]. Coal combustion comprised over 90% of electric utility emissions on a national scale. from UFG and analyzed for δ 15 N in NO − 3 ( figure 4). The δ 15 N values in the ice were all negative and had similar isotopic composition, ranging from -5.9 to -3.2 ‰. Nitrate in snow and corresponding glacier ice can be derived from a variety of natural (lightning, biogenic soil processes, and wildfires) and anthropogenic sources (generation of electricity, industrial processes, and vehicle exhaust) [14]. The mean δ 15 N value in the UFG icecore samples (-4.0 ‰, n = 7) was equal to the mean δ 15 N value of six archived precipitation samples collected during 2000 at NADP/NTN site WY98 (C. Kendall, U.S. Geological Survey, written commun., 2008). The similarity in mean δ 15 N values between the ice-core and NADP samples suggests similar NO − 3 source(s).

Variation of δ 15 N in ice
Ice-core δ 15 N values were compared with δ 15 N values from potential NO X sources from previously published data (figure 4). Although δ 15 N values from coal deposits in the western U.S. were not available, ice-core δ 15 N values were substantially lighter than δ 15 N values of NO X from coal-fired powerplants in South Africa [47], simulated lightning [48], or kerogen in North American coal deposits [49]. Ice-core δ 15 N values were significantly enriched relative to soil-released biogenic NO values [50]. The δ 15 N of the ice-core samples all fall within the δ 15 N values expected from vehicle emissions [47,50]. Based on national NO X emission data compiled by the U.S. Environmental Protection Agency [46], onand off-road vehicle emissions comprised 53% of NO X emissions during 1998 as opposed to electrical generation (25%) and industrial combustion (12%). Data compiled for the western United States indicate that on-and off-road mobile sources comprise 67% of the anthropogenic NO x emissions, with stationary sources comprising only 22% of the anthropogenic NO x emissions [51]. Based on the comparisons of the tightly constrained UFG ice-core δ 15 N values with existing end-member data, it appears likely that vehicular NO X emissions may have a large impact on NO − 3 sources in snowfall deposited in the WRR of Wyoming.
The ice-core results conflict with a recent study in the more densely populated eastern United States where it was found that NO x emissions from stationary sources were strongly correlated with spatial variation of δ 15 N (NO − 3 ) in wet deposition [14]. One possible reason for the observed discrepancy with the ice-core results could be the lower number of stationary NO x sources in the western United States to support the lower population density. Additional δ 15 N values of NO X emissions from regional biogenic sources, as well as natural gas production facilities and coal-fired power plants in the western U.S. are needed to better assess NO − 3 sources.

deposition
In addition to the S and N isotope data, the ice-core samples were used to provide an archive of annual SO 2− 4 and NO − 3 deposition amounts that could be used to qualitatively extend the NADP/NTN records to the early-1950s. Dissolved (0.45 μm) SO 2− 4 and NO − 3 concentrations from the 1998 ice-core samples (site DH-98-4) were used in combination with the average ice-core density value of 0.90 g/cm 3 (n = 12) to calculate deposition masses in kilograms/hectare (kg/ha). The age of each ice-core sample from site DH-98-4 was determined from the modified age-depth profile developed for the 1991 ice core to account for the additional snow deposited between 1991 and 1998. Once an age was assigned to each ice-core  sample, the SO 2− 4 and NO − 3 masses from the same year were summed to represent an annual deposition mass that could be directly compared to the NADP/NTN deposition data from sites WY06 and WY98 [52]. Annual SO 2− 4 and NO − 3 masses in kg/ha were calculated for snow deposited on UFG from 1951 through 1993 ( figure 5).
Comparison of the NO − 3 wet deposition data from the NADP/NTN sites to UFG data suggests that NO − 3 deposition in the WRR may have been less prior to 1980 (figure 5). During the time period of overlapping NO − 3 data there is some general agreement between UFG and NADP/NTN annual deposition masses (i.e. 1985, 1988, 1992, 1993, and 1994). Other years of overlapping data indicate substantially less annual NO − 3 deposition from the ice-core record relative to NADP/ NTN sites WY06 and WY98 (i.e. 1982, 1983, 1984, 1990, and 1991.
A number of reasons may explain the time periods when NO − 3 deposition rates recorded in the UFG ice core were different than the measured deposition rates at the NADP/NTN sites. At 4,000 m ASL, the UFG icecoring site is subject to wind erosion of previously deposited snow and the corresponding NO − 3 deposition mass. An on-site snow depth sensor on UFG was used in combination with snow-density measurements to document significant snow removal by wind events after deposition [25]. Data collected on UFG during 1999-2000 indicated that the snow removal events were less during the spring, likely due to warmer air temperatures, resulting in higher-density snowfall. At an elevation of 4 km ASL, UFG is subject to free tropospheric and sometimes stratospheric air, increasing the likelihood of receiving atmospheric deposition from hemispheric sources relative to the lower elevation NADP/NTN sites. The NO x produced from the oxidation of nitrous oxide in the stratosphere should contain a δ 15 N of~19 ‰ [35], which is significantly enriched relative to the δ 15 N (NO − 3 ) values observed in the UFG ice-core samples ( figure 4).
Post-depositional elution of solutes from the snowpack deposited on UFG could also contribute to lower NO − 3 masses relative to similar time periods at the NADP/ NTN sites. However, chemical and isotopic analyses of ice from both cores indicate the low-resolution preservation of historical records of environmental change, including above-ground nuclear weapons testing, air temperature, volcanic eruptions, and the chemical quality of atmospheric deposition such as sub-μg/L concentrations of atmospheric mercury [20][21][22][23][24][25][26][27].
The deposition data from UFG (1951 through 1993) indicates an increasing trend in SO 2− 4 deposition until approximately 1980 (figure 5). Beginning in 1980, both the NADP/NTN (1980 to 2006) sites and the UFG data indicate slightly decreasing to stable SO 2− 4 deposition. During the time period of overlapping SO 2− 4 deposition data from both UFG and NADP/NTN sites, there is only one year where similar deposition masses occur (1986). All remaining years of overlapping data indicate substantially less annual SO 2− 4 deposition from the icecore record relative to the NADP/NTN sites. Similar processes controlling NO − 3 deposition also are likely affecting the annual SO 2− 4 deposition rates on UFG relative to the NADP/NTN sites.
The annual NO − 3 and SO 2− 4 deposition data from UFG was compared with total U.S. emissions of NO X and SO 2 from 1950 to 1998 (figure 5) as compiled by the U.S. Environmental Protection Agency. Although this comparison does not address the overall higher NO X and SO 2 emissions in the eastern U.S. relative to Wyoming deposition rates, it does provide a historical perspective on the initiation of the NADP/NTN program relative to total emission amounts. The effect of sharply increasing NO X and SO 2 emissions from 1950 to the mid-1970s occurred before the beginning of the NADP/NTN program; however, the UFG ice-core  Given the observed increase in SO 2− 4 and NO − 3 deposition in the UFG ice-core samples since the mid-1940s, it is unclear why no distinctive temporal trend in the isotopic composition is observed during the same time period. This may suggest that most of the SO 2− 4 and NO − 3 deposition to the WRR since the mid-1940s has been dominated by anthropogenic inputs from a common source and any increases in deposition amounts contain similar isotopic signatures. Additional reasons for the lack of an isotopic trend could include: (1) limited period of record starting after the existence of anthropogenic inputs; (2) large sample composites for isotope samples covering multiple years of deposition that mask finer scale trends evident in the SO 2− 4 and NO − 3 deposition data; (3) varying deposition amounts from local, regional, and global sources; and (4) isotopic fractionation from postdepositional processes in the firn and ice. Additional studies are needed to assess the impact of postdepostional processes on the preservation of δ 34 S (SO 2− 4 ) and δ 15 N (NO − 3 ) values at the UFG coring site. Recent work at the Summit site, Greenland, found a potentially minimal influence of photolysis on the isotopic composition of NO − 3 in firn and ice, in part due to high accumulation rates [18].
Isotopic analyses from other sections of the UFG ice core (including pre-industrial) would provide additional insight into the existence of isotopic trends. Unfortunately, the low concentration of NO − 3 and SO 2− 4 combined with an insufficient amount of ice-core samples representing atmospheric deposition prior to~1945 does not allow for sufficient sample mass to support S and N isotopic analyses. Despite this short-coming, results from these icecore data provide insight to the long-term wet deposition of NO − 3 and SO 2− 4 in pristine areas of the western United States that were previously not available.