HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (23) Citing Articles via Google Scholar Google Scholar Articles by Puckett, L. J. Articles by Cowdery, T. K. Search for Related Content PubMed PubMed Citation Articles by Puckett, L. J. Articles by Cowdery, T. K. GeoRef GeoRef Citation Agricola Articles by Puckett, L. J. Articles by Cowdery, T. K. Related Collections Ground Water Quality Water Pollution Nitrogen

TECHNICAL REPORTS
Ground Water Quality

Transport and Fate of Nitrate in a Glacial Outwash Aquifer in Relation to Ground Water Age, Land Use Practices, and Redox Processes

^a U.S. Geological Survey, 413 National Center, Reston, VA 20192
^b U.S. Geological Survey, 2280 Woodale Road, Mounds View, MN 55112

* Corresponding author (lpuckett{at}usgs.gov)

Received for publication February 8, 2001.

	ABSTRACT

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
A combination of ground water modeling, chemical and dissolved gas analyses, and chlorofluorocarbon age dating of water was used to determine the relation between changes in agricultural practices, and NO^-₃ concentrations in ground water of a glacial outwash aquifer in west-central Minnesota. The results revealed a redox zonation throughout the saturated zone with oxygen reduction occurring near the water table, NO^-₃ reduction immediately below it, and then a large zone of ferric iron reduction, with a small area of sulfate (SO^2-₄) reduction and methanogenesis (CH₄) near the end of the transect. Analytical and NETPATH modeling results supported the hypothesis that organic carbon served as the electron donor for the redox reactions. Denitrification rates were quite small, 0.005 to 0.047 mmol NO^-₃ yr^-1, and were limited by the small amounts of organic carbon, 0.01 to 1.45%. In spite of the organic carbon limitation, denitrification was virtually complete because residence time is sufficient to allow even slow processes to reach completion. Ground water sample ages showed that maximum residence times were on the order of 50 to 70 yr. Reconstructed NO^-₃ concentrations, estimated from measured NO^-₃ and dissolved N gas showed that NO^-₃ concentrations have been increasing in the aquifer since the 1940s, and have been above the 714 µmol L^-1 maximum contaminant level at most sites since the mid- to late-1960s. This increase in NO^-₃ has been accompanied by a corresponding increase in agricultural use of fertilizer, identified as the major source of NO^-₃ to the aquifer.

Abbreviations: PVC, polyvinyl chloride • DOC, dissolved organic carbon • DO, dissolved oxygen • CFC, chlorofluorocarbon

	INTRODUCTION

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
NITRATE IS RECOGNIZED AS THE most ubiquitous contaminant of shallow aquifers (Postma et al., 1991; Spalding and Parrott, 1994). This ground water contamination provides cause for concern because elevated concentrations of NO^-₃ have been associated with human health risks (Spalding and Parrott, 1994), and when NO^-₃ discharges to surface waters it may contribute to eutrophication. Consequently, there has been growing interest in processes and conditions that lead to natural remediation of NO^-₃ contaminated aquifers. The primary pathway for removal of NO^-₃ from ground water is denitrification, wherein nitrogen serves as the terminal electron acceptor following the general reaction:

[1]

Bacteria that derive energy from the oxidation of a reduced species typically mediate this reaction. Organic C (Gillham and Cherry, 1978; Morris et al., 1988; Starr and Gillham, 1993) is a common electron donor in the heterotrophic denitrification process:

[2]

Alternatively, other reduced species such as ferrous iron in pyrite (FeS₂) and silicate minerals such as pyroxenes and amphiboles, can serve as the electron donor in the autotrophic denitrification process (Kolle et al., 1983; Strebel et al., 1985; van Beek et al., 1988; Postma et al., 1991; Bohlke and Denver, 1995):

[3]

It is significant that in all these reactions, NO^-₃ is converted from a potentially harmful form to a benign form as N₂ gas. Under most conditions, the N₂ gas produced as the result of denitrification will remain in solution until the ground water discharges to a surface-water body and equilibrates with the atmosphere (Heaton and Vogel, 1981; Blicher-Mathiesen et al., 1998). By measuring this excess N₂, it is possible to estimate the amount of NO^-₃ originally present in the ground water (Heaton and Vogel, 1981; Vogel et al., 1981; Wilson et al., 1990; Smith et al., 1991; Bohlke and Denver, 1995; Blicher-Mathiesen et al., 1998; Modica et al., 1998).

Dissolved oxygen (O₂) is thermodynamically favored over NO^-₃ as the terminal electron acceptor in the oxidation of organic C (Stumm and Morgan, 1981; Chapelle et al., 1995). When virtually all the available O₂ is consumed, denitrification becomes the terminal electron accepting process, leading to lower NO^-₃ concentrations as well. Once most of the available NO^-₃ is consumed, bacteria utilize manganese oxides and then iron oxides as terminal electron acceptors. Under very strong reducing conditions, bacteria capable of deriving energy from SO^2-₄ reduction and methanogenesis will be favored. The net result of this evolution of water chemistry is a predictable sequence where O₂ and then NO^-₃ are consumed, manganese and iron come into solution, SO^2-₄ is consumed and sulfide may come into solution, and finally methane may appear in solution (Chapelle et al., 1995). This predictable sequence, under some conditions, may allow determination of the redox state of the aquifer at various points along a flow path, and prediction of the fate of NO^-₃ in the aquifer. Few studies have conducted a complete, simultaneous analysis of the various nitrogen species, particularly N₂, coupled with an analysis of the redox processes involved, leaving many questions unanswered with respect to the true fate of NO^-₃.

In addition to denitrification, other factors may affect spatial and temporal patterns of NO^-₃ along ground water flow paths. For example, inputs of nitrogen in atmospheric deposition, fertilizer, and animal wastes have increased markedly over the past five decades (Puckett, 1995). When ground waters of increasing age are analyzed, the older waters will appear to have less NO^-₃, simply because they received less at the time of recharge. Also, conditions favoring denitrification can be patchy in some environments (Addy et al., 1999; Gold et al., 1998; Jacinthe et al., 1998) making it possible to have discordant NO^-₃ concentrations in an aquifer. All of these conditions make it necessary to develop a complete understanding of both the hydrogeologic and biogeochemical processes involved in the transport and fate of NO^-₃.

Puckett et al. (1999) presented a mass-balance analysis for the Otter Tail aquifer system in west-central Minnesota and concluded that although the 212.4 km² system receives about 2000 Mg of nitrogen each year, only 229 Mg (11%) actually enters the ground water system. Their analysis indicated that about 181 Mg (44%) of the excess nitrogen leaving the root zone was removed through denitrification. Their evidence for denitrification was based on the loss of NO^-₃ from the system in the mass-balance model as well as an analysis of the oxidation-reduction (redox) conditions in ground water sampled as part of the study. However, that analysis focused entirely on landscape and watershed processes, including the upper 2 m of the shallow ground water system; no information was provided on the transport and fate of nitrogen as it moved through the aquifer. Because of the potentially harmful nature of NO^-₃, it is important to understand its transport from the shallow ground water system into deeper portions of aquifers where most drinking water well screens are located. Also, developing a complete understanding of the biogeochemical reactions involved in the transformation of NO^-₃ to other forms may allow us to more accurately predict when and where drinking water supplies may be contaminated.

In this paper, we present the results of a study designed to examine the hydrogeologic and biogeochemical processes controlling the evolution of water chemistry along flow paths of a representative cross-section within the Otter Tail aquifer studied by Puckett et al. (1999). We examine the hypothesis that denitrification is responsible for the loss of NO^-₃ and further that it is driven by oxidation of organic matter. We also consider the hypothesis that changes in concentrations of NO^-₃ are a function of changes in agricultural practices over time and, therefore, the age of water sampled along the transect. In evaluating these hypotheses, we develop an analysis of the transport and fate of the major nitrogen species, the redox reactions involved, and the age-distribution of waters along the transect.

	STUDY AREA

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
This study was conducted in Otter Tail County in west-central Minnesota, USA (Fig. 1) in a 36 km² subunit of the Otter Tail outwash aquifer north of the Otter Tail River near Perham. The Otter Tail outwash aquifer consists mostly of fine to coarse, fairly well-sorted sand, but also contains some gravel and silt. Cowdery (1997) reported that the outwash material consisted of 21% gravel, 45% coarse sand, 27% medium sand, and 7% fine sand and silt. Aquifer thickness is generally about 18 m but, may be >30 m in places (Reeder, 1972). It was deposited by meandering debris-laden streams flowing from the disintegrating margin of the Des Moines glacial lobe sometime between 14000 and 12000 yr ago (Wright, 1972). After the outwash sands were deposited, stagnant ice blocks within the aquifer melted, forming the abundant lakes and wetland depressions common throughout the area (Goldstein, 1985).

View larger version (45K): [in this window] [in a new window]   Fig. 1. Location of the study area north of the Otter Tail River.

Climate of the West Central Minnesota climate division has been described as subhumid continental and annual average temperature and precipitation in the area during the 30-yr period from 1961 to 1990 was 5.6°C and 68 cm, respectively (Stoner et al., 1993). Open water evaporation in the study area exceeds precipitation by about 10 cm yr^-1 (Winter and Woo, 1990); consequently ground water recharge occurs primarily during spring snowmelt when potential evaporation is small. Historically the study area was prairie, surrounded by hardwood and conifer forests on the morainal uplands; however, at the time of the study, approximately 73% of the land was cropland.

Fertilizer nitrogen use for agricultural purposes in Otter Tail County (Fig. 2) increased from about 28 Mg in 1945 to about 13000 Mg in 1994 (Jerald Fletcher, West Virginia Univ., personal communication, 1996). A major step change in the application rate occurred about 1967 and the rate of increase has been essentially linear since about 1970. More than half of the total amount of nitrogen used during the 1945 to 1994 period has been applied since 1982.

View larger version (16K): [in this window] [in a new window]   Fig. 2. (a) Annual fertilizer N applications in Otter Tail county since 1945; (b) cumulative N applications during the same period.

	METHODS

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

View larger version (33K): [in this window] [in a new window]   Fig. 3. Cross-section of the study transect showing land use, piezometer nest locations, screened intervals, stratigraphy, and approximate locations of MODFLOW layers.

Hydrologic Modeling
The ground water flow models MODFLOW (McDonald and Harbaugh, 1988; Harbaugh and McDonald, 1996) and MODPATH (Pollock, 1989) were used to develop a three-dimensional steady-state flow model of the study area, delineate flow paths, and estimate ground water travel times. A three-layer model was constructed for a 26.7-km² portion of the study area using approximately 100 by 100 m cell dimensions. The model was based on a detailed study of the physical and hydrologic characteristics of the aquifer reported by Reeder (1972). The bottom of Layer 1 was set at 4 m below the initial water table and configured as unconfined with an average effective recharge rate (Total precipitation - Losses from runoff and evapotranspiration) of 17.7 cm yr^-1 (Puckett et al., 1999). Layer 2 was 4 m thick and the thickness of Layer 3 varied depending on the total thickness of the aquifer. Based on reported hydraulic conductivities of 0.009 m d^-1 for the underlying till, and 27 to 183 m d^-1 for the aquifer materials (Reeder, 1972), the bottom of Layer 3 was designated as a no-flow boundary as were the northern, eastern, and western edges of the model area; porosity of the sand was assumed to be 0.3. Although flow may occur across the boundary at the till-sand contact, because of the great disparity in hydraulic conductivities, this component of flow was not considered to be significant. The boundary with Little Pine Lake on the southeastern corner of the area was modeled with constant head cells. The Otter Tail River on the southern side was modeled using the river option with a conductance of 300 m³ d^-1 m^-1, assuming a vertical hydraulic conductivity of 6 m d^-1, an average width of 50 m, and a thickness of 1 m. Wetlands were modeled as drains with a conductance of 83 m³ d^-1 m^-1, assuming a vertical hydraulic conductivity of 0.5 m d^-1, an average width of 50 m, and a thickness of 0.3 m. Peat hydraulic conductivities used in estimating wetland conductance were based on a range of measurements reported by Siegel and Glaser (1987), Siegel (1988), and Romanowicz et al. (1993). Isotropic conditions of horizontal hydraulic conductivity and a ratio of horizontal to vertical hydraulic conductivity equal to 5 were assumed.

The ground water model was calibrated by adjusting the hydraulic conductivity until horizontal hydraulic gradients and predicted water levels reasonably matched those reported in previous investigations and determined by water-level measurements in this study (Reeder, 1972). The model-applied hydraulic conductivity of 30 m d^-1 was at the lower end of the reported range of 27 to 183 m d^-1 and all factors used in the simulations were within reasonable limits of measured or reported values (Reeder, 1972; Fetter, 1994, p. 321–325). The path-line tracking option of MODPATH was used to compute ground water travel times and delineate flow paths. When initiating particle tracking, particles were placed at the midpoint depth of each well screen (which was the point where we sampled), and the reverse tracking option was used to track the particle back to the position at the water table where it would have originated.

Sample Collection and Analysis
Sediment samples were collected using a continuous piston core barrel with a polycarbonate sleeve. Selected samples from Sites 1 and 3 were sorted into bulk and <1 µm size fractions and analyzed for mineralogy by x-ray diffraction. Samples for these analyses were selected to provide a description of mineralogy at several points across the thickness of the aquifer at these locations. A separate aliquot of each sample was analyzed for total C, using a Carlo Erba elemental analyzer (CE Elantech, Inc., Lakewood, NJ). A duplicate sample was exposed to hydrochloric acid fumes to remove inorganic C after which organic C content was measured, and inorganic C was calculated as the difference between total and organic C (Hedges and Stern, 1984).

Piezometers were sampled up to seven times during the period from April 1994 to August 1997, using a stainless steel positive displacement pump with Teflon tubing. Sampling procedures are detailed in protocols by Koterba et al. (1995), and Menheer and Brigham (1997) report the specific application of these protocols to this study. Water temperature, specific conductance, and dissolved-oxygen (O₂) concentration were continuously monitored before sample collection. When these measures stabilized, and at least three casing volumes of water had been extracted from the piezometer, sample collection began.

Samples were filtered through a 0.45-µm nitrocellulose filter [a silver filter was used for dissolved organic carbon (DOC) samples]. Mercuric chloride preservative was added to nutrient samples during 1994 but afterward, no preservative was added. Instead, samples were shipped on ice the same day to the U.S. Geological Survey National Water Quality Laboratory in Denver, CO, for immediate analyses; cation samples were preserved with nitric acid. Alkalinity was determined in the field by incremental titration of filtered samples. Specific conductivity, pH, temperature, and O₂ were all measured in a sealed flow-through cell attached to the sample discharge line.

Major cations and silica were analyzed by inductively coupled plasma spectroscopy, anions by ion chromatography, nutrients by colorometric methods (Fishman and Friedman, 1989; Fishman, 1993; Wershaw et al., 1987), and DOC by persulfate oxidation and infrared spectrometry (Brenton and Arnett, 1992). Equipment blank samples, field blank samples, replicate samples, and spiked samples (Koterba et al., 1995; Menheer and Brigham, 1997) showed a general lack of sampling contamination.

Sulfide (S^2-) was measured by the methylene blue method (APHA, 1975). Samples were processed immediately using a HACH 2000 portable spectrophotometer (Hach Co., Loveland, CO) with a detection limit of 1.87 µmol L^-1. The geochemical model WATEQ as implemented in NETPATH (Plummer et al., 1994) was used for speciation and calculation of saturation indices. The WATEQ output was then used in the NETPATH program (Plummer et al., 1994) to evaluate various biogeochemical reactions controlling water chemistry along the transect.

Dissolved gases (N₂, Ar, CO₂, CH₄, and O₂) were analyzed in ground water samples by gas chromatography after extraction in headspaces of glass samplers (Busenberg et al., 1998). Results were corrected for solubility in sample water at laboratory temperatures and have an error of ±2 to 4%. Concentrations of N₂ and Ar gases dissolved in water were used to estimate the amount of nitrogen resulting from denitrification (Vogel et al., 1981; Wilson et al., 1990; Bohlke and Denver, 1995; Blicher-Mathiesen et al., 1998). This approach required the assumption that the atmosphere and denitrification were the only sources of N₂ in ground water and the atmosphere was the only source of Ar. However, excess air can also become trapped in recharging ground water due to entrainment of air bubbles, and increase dissolved gas concentrations accordingly. This is more of a problem in arid zones where rapid recharge occurs and in fine-grained sediments, but is usually small in coarse sediments and fractures. Where excess air alone affects dissolved gas concentrations, N₂ and Ar increase linearly in a ratio defined by their relative solubilities in water and concentrations in the atmosphere, making it possible to estimate the amount of excess air, which typically is negligible (<3 mL kg^-1) and often ignored in studies of this nature (Busenberg et al., 1993; Bohlke and Denver, 1995; Plummer and Busenburg, 1999). However, we made corrections to the excess N₂ estimates by subtracting the air-saturated water plus excess air concentrations from the measured concentrations. We estimated the original or reconstructed concentration of NO^-₃ (before denitrification) as the sum of measured NO^-₃ and excess N₂. Denitrification rates were calculated by dividing the excess N₂ gas by sample age.

We also applied the methods presented by Blicher-Mathiesen et al. (1998) to evaluate the potential for degassing of ground water. This method involves comparing the total partial pressures of dissolved gases in each sample to the hydrostatic pressure of the sampled piezometer. Where partial pressures exceed hydrostatic pressure the potential exists for degassing.

Samples for analyses of the chlorofluorocarbons CFCl₃ (CFC-11), CF₂Cl₂ (CFC-12), and C₂F₃Cl₃ (CFC-113) were collected using a stainless steel sampling apparatus under an ultra-pure N₂ atmosphere and were flame sealed in 62 mL borosilicate glass ampoules, preventing atmospheric contamination (Busenberg and Plummer, 1992). Five ampoules were collected at each site, three of which were used for analyses by electron-capture gas chromatography with a detection limit of 0.3 pg kg^-1 of water for CFC-11 and CFC-12 and 1.0 pg kg^-1 for CFC-113 (Busenberg and Plummer, 1992). The CFC's are known to exhibit varying degrees of susceptibility to biodegradation and sorption on organic matter, with CFC12 being least affected and CFC11 and CFC113 being the most affected (Cook et al., 1995; Lovely and Woodward, 1992; Plummer and Busenberg, 1999). Consequently, CFC12 was used in most cases for age determinations. Sample ages were assigned based on comparison of CFC equilibrium partial pressures, corrected for recharge temperature, with a chronology of atmospheric partial pressures (Busenberg and Plummer, 1992). Recharge temperatures were estimated in two ways: (i) using Ar concentrations and assuming equilibrium with the atmosphere, and (ii) where Ar was not measured based on the average annual temperature (Busenberg and Plummer, 1992). These methods result in a potential uncertainty of about 2°C, which translates to an age-dating uncertainty of about 1 yr before 1970, 1 to 3 yr for 1970 to 1990, and 3 yr for waters that recharged after 1990. The increasing uncertainty in recent years results from the flattening of the CFC input function due to reductions in CFC usage. Because they typically require some degree of subjective interpretation, CFC age dates are referred to as model or apparent ages and are considered a minimum age (Plummer and Busenberg, 1999).

	RESULTS

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Hydrogeology
Bulk fraction mineralogy (Table 1) was dominated at both sites by five minerals: quartz, plagioclase (probably albite), potassium feldspar, dolomite, and calcite. Quartz was the most abundant except at Site OF3, near the contact with the underlying till where plagioclase dominated. Also, potassium feldspar was more common at Site OF3 and only appeared in the sample from 12.2 m deep at Site OF11. Dolomite (up to 19%) was usually more abundant than calcite (up to 15%) although, values in the 1 to 3% range for each were more common. Examination of sediments during drilling revealed that most of the sand was tinted a yellow-red color indicating the presence of iron oxide coatings. These results are also consistent with those reported by Berndt (1987) and Tuccillo et al. (1999) for outwash deposits nearby in the vicinity of Bemidji, MN.

Total C was in consistently greater abundance at Site OF1, ranging from 1.11 to 3.17%, than at Site OF3, where it ranged from 0.06 to 0.89%. Interestingly the peak abundance at Site OF1 coincided with what had been identified during drilling as a thin black organic layer at 4.6 m in depth. Inorganic C abundance was similar in all samples except for the shallow sample at Site OF3 where it was 0.04% and the black organic layer at Site OF1 where it was 1.71%. Organic C abundance was variable, ranging from 0.01 to 1.45% (median = 0.15) with the largest abundance in the black organic layer at Site OF1 and the smallest in the shallow sample from Site OF3.

Results of the MODFLOW/MODPATH models indicated that ground water generally flowed north to south, toward the Otter Tail River or Little Pine Lake. Predicted and observed water levels (Table 2) agreed within about 0.6 m with a root mean square error (RMS) of 0.44. Simulated flow lines and particle tracking results from the MODPATH simulation (Fig. 4) indicated that travel times increased with increasing depth and distance traveled along the transect. Ground water flowing along the deepest and longest flow paths along the transect, from the northern edge of the aquifer to the Otter Tail River, can be expected to take >70 yr to reach its discharge point in the area of the Otter Tail River (Fig. 4). The net result is a well-stratified age distribution within the aquifer (Fig. 5) .

View larger version (31K): [in this window] [in a new window]   Fig. 4. Ground water flow lines and travel times along the study transect based on the MODFLOW/MODPATH simulation.

View larger version (31K): [in this window] [in a new window]   Fig. 5. Apparent ground water recharge dates as determined from chlorofluorocarbon (CFC) concentrations measured in ground water samples and generalized age distribution along the transect based on the MODFLOW/MODPATH simulation.

View this table: [in this window] [in a new window]   Table 3. Median concentrations of S2-, dissolved gases, NO-3, excess N2 (Xs N2), reconstructed NO-3 (NO-3), and CFC-based apparent ages of samples collected during 1994, 1995, and 1997. (NO-3 concentrations may differ from Table 4 because only samples having paired N2 gas data were included in the calculations.)

View larger version (15K): [in this window] [in a new window]   Fig. 6. Argon and N2 gas concentrations for samples collected during 1994, 1995, and 1997 and their theoretical values. Values to the right of the air-saturated water equilibrium line represent excess nitrogen concentrations. Horizontal tick marks on the equilibrium line indicate Ar and N2 equilibrium concentrations from 0 to 20°C in 2° increments.

View larger version (27K): [in this window] [in a new window]   Fig. 7. Distribution of excess N2 along the transect.

Chlorofluorocarbon Apparent-Age Dates
Chlorofluorocarbon apparent ages ranged from 5 yr in the shallow piezometers to >50 yr at OF1D and are consistent with the modeling results, given the greater uncertainty of the post-1970 apparent ages. There was a pattern of increasing age with increasing depth along the transect (Fig. 5), and a linear relationship (r² = 0.759) between age and depth. The CFC apparent ages and those simulated using the MODPATH model were generally in good agreement (r² = 0.835), although the MODPATH ages tended to be younger (Table 2).

Chlorofluorocarbon apparent ages reflect the year in which ground water was last in contact with unsaturated zone air (Cook and Solomon, 1997) and previous work at this site has shown that it can take from 0.5 to 1 yr for water to percolate through the vadose zone to the water table, and about 2.5 to 4.8 yr for water to then reach the sampling zone of the shallow piezometers (Stoner et al., 1997; Puckett et al., 1999). Therefore, a lag of 3 to 6 yr between the age of shallow samples and the year in which the precipitation fell is expected. Dissolved gas results indicated that at this site excess air was small and therefore the effects on CFC apparent-age dates were negligible (Plummer and Busenberg, 1999). The CFC data for piezometers OF4S and OF3M were much older than could be explained based on model projections. The recharge areas for these sites, however, are in an old outwash channel containing wetlands, factors we feel resulted in sorption of the CFCs, leading to the older apparent ages.

Major Ion Chemistry
Concentrations of NO^-₃, which were as high as 1785 µmol L^-1 with a median of 107 µmol L^-1, behaved like O₂, decreasing with both increasing age and depth (Table 3, Fig. 8) . This decreasing pattern suggested that once O₂ was depleted, NO^-₃ was utilized as the terminal electron acceptor in the oxidation of organic C. Samples more than 30 yr old were essentially depleted of NO^-₃. Concentrations of NH⁺₄ reached 50 µmol L^-1 with a median of 2.1 µmol L^-1. There was a general trend of increasing NH⁺₄ concentration with increasing age and depth. Although the exact source of NH⁺₄ was not determined, in other studies of sediments in similar redox environments, NH⁺₄ production has been traced to mineralization of organic N and not to dissimilatory NO^-₃ reduction (McMahon et al., 1999). Other forms of N were negligible.

View larger version (26K): [in this window] [in a new window]   Fig. 8. Concentrations of NO-3 along the transect.

Iron concentrations displayed a large range, from below detection to as high as 135 µmol L^-1. Such relatively large concentrations of dissolved iron suggests that the predominant terminal electron accepting process had changed from reduction of O₂, NO^-₃, and manganese oxides to iron oxide reduction. Furthermore, these concentrations are consistent with results from similar sediments nearby in Bemidji, MN, that contain sufficient coatings of iron oxides on quartz and feldspar grains to generate Fe²⁺ concentrations of 35000 µmol L^-1 if reduced simultaneously (Tuccillo et al., 1999). Manganese concentrations reached as much as 11 µmol L^-1 and typically peaked when most NO^-₃ had been consumed and Fe²⁺ was increasing. This pattern agrees with the reported redox sequence where manganese oxide reduction occurs at p levels after denitrification and before iron oxide reduction (Stumm and Morgan, 1981).

The dominant dissolved species were Ca²⁺, Mg²⁺, and HCO^-₃, which is consistent with what is known of the outwash mineralogy and agricultural applications. Farmers in the area typically apply 6.7 to 8.9 Mg ha^-1 yr^-1 of limestone containing both calcite and dolomite (Denzel Cooper, Univ. of Minnesota Extension Service, personal communication, 1996). Infiltrating ground waters would therefore be expected to develop a carbonate signature as a result of dissolution of these carbonates. The ratio of Ca²⁺ to Mg²⁺ was approximately 2:1, which suggests dissolution of calcite and dolomite in roughly equal amounts. The HCO^-₃ concentrations generally increased with increasing depth in the aquifer.

Chloride (Cl^-) and sulfate (SO^2-₄) concentrations ranged as high as 2003 and 1103 µmol L^-1, respectively. Chloride commonly occurs in fertilizer as potassium chloride (KCl) and is a common contaminant associated with agricultural areas. With respect to S, there is little available in the outwash sediments; therefore, it is commonly supplied as an agricultural supplement either as gypsum (CaSO₄·2H₂O) or in more recent years as ammonium sulfate [(NH₄)₂SO₄] at rates of 28 to 336 kg S ha^-1 (Denzel Cooper, Univ. of Minnesota Extension Service, personal communication, 1996). Typically, the larger amount is applied once in a 6-yr rotation as a starter for alfalfa (Medicago sativa L.) and the smaller amount is applied as a starter for corn (Zea mays L.). At the estimated ground water recharge rate of 17.7 cm yr^-1 (Cowdery, 1997; Puckett et al., 1999), these application rates would result in a spatially average concentration of 495 to 5934 µmol SO^2-₄ L^-1 and therefore can easily account for the observed concentrations. The distributions of both Cl^- and SO^2-₄ (Fig. 9a, b) reflect the most intensive areas of agriculture along the transect, and the lowest concentrations are generally found in the deeper, down-gradient portions of the transect. These lower concentrations also are associated with the oldest ground waters, which is to be expected since less fertilizer was used at that time and some SO^2-₄ reduction occurred as well.

View larger version (53K): [in this window] [in a new window]   Fig. 9. Distibution of (a) Cl- and (b) SO2-4 along the transect.

Our focus in the model was on the reactions responsible for O₂ and NO^-₃ losses, and increases in C, Ca²⁺, and Mg²⁺, including deducing the electron donors in the redox reactions. In addition to the calcite (CaCO₃), dolomite [CaMg(CO₃)₂], and organic matter, that occur naturally in the soil, agricultural inputs at the site include calcite, dolomite, potassium (as KCl), nitrogen (as NO^-₃ and NH⁺₄), and sulfur [as CaSO₄·2H₂O or (NH₄)₂SO4]. Accordingly, C, Ca, Mg, K, Na, N, S, Cl and redox state were among the constraints in the model and the phases included nitrogen gas (N₂), organic matter (CH₂O), calcite, dolomite, gypsum (CaSO₄·2H₂O), and potassium chloride (KCl). We also modeled ion exchange of both K and Na with Ca and Mg. The phases were chosen on the basis of mineralogy and the saturation indices calculated in WATEQ.

Four NETPATH models (Table 5) were found that fit the constraints, phases, saturation indices, and mineralogy. In all the models organic matter was oxidized and N₂ left solution, small amounts of KCl and gypsum were dissolved, and Na exchanged for Ca and Mg. Calcite and/or dolomite dissolved in all of the models, and in those models where one phase or the other was not selected, K exchange for Ca or Mg balanced the reaction. Because we know that both calcite and dolomite are applied in the area, Models 1 and 2 (Table 5) seem most realistic.

	DISCUSSION

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Redox Processes
Analytical results (Tables 3 and 4) show that the upper 3 to 4 m (<10 yr old) of the saturated zone were well below the equilibrium concentration for O₂ (373 µmol L^-1). However, this zone was not anoxic, indicating that it was a zone where O₂ reduction was occurring (Fig. 10) . Below the zone of O₂ reduction was another 3 to 4 m (10–20 yr old) thick zone essentially depleted of O₂, where NO^-₃ reduction was occurring with essentially complete denitrification at its base. Most of the remainder of the outwash sediments comprised a zone of iron oxide reduction ranging from 20 to 50 yr in age, with a small zone of SO^2-₄ reduction and CH₄ production in the sediments containing ground water >50 yr in age (Fig. 10). The loss of both O₂ and NO^-₃ in the shallow portions of the aquifer (Tables 3 and 4), accompanied by the production of large amounts of HCO^-₃ (Table 4) and excess N₂ (Fig. 7), indicate that oxidation of organic C with NO^-₃ as the electron acceptor was occurring and this is supported by the NETPATH model results. At equilibrium, dissolved O₂ would be 373 µmol L^-1 and therefore would place a demand of 1492 electron equivalents (Postma et al., 1991). The introduction of 463 µmol NO^-₃ L^-1 (the median concentration of NO^-₃ in the shallow aquifer) places an additional demand of 2315 electron equivalents. The combined electron demand is about 2.5 times that of O₂ alone, increasing the amount of C oxidized. As organic C continues to be oxidized in the shallow sediments we can expect to see the depths to which NO^-₃ persists increase; however, the rate of downward movement can be expected to take decades. Similar conclusions were reached by Robertson et al. (1996) for an aquifer in southern Ontario where reduced sulfur compounds served as the electron donor for autotrophic denitrification.

View larger version (40K): [in this window] [in a new window]   Fig. 10. The generalized redox state of the aquifer along the transect.

The redox zonation has implications on the potential transport and fate of NO^-₃ entering the aquifer. Specifically, NO^-₃ is thermodynamically and biologically unstable below a depth of about 3 to 4 m below the water table. Because the denitrification process is dependent on organic C as an electron donor, the future ability of this aquifer to continue removing NO^-₃ is dependent on the availability of reactive organic C. It is generally accepted that leaching from soil horizons is the primary source of dissolved organic C to shallow flow systems (Chapelle, 1993). Korom (1992) has stated that given the stoichiometry of Eq. [2], on a molar basis there must be 25% more dissolved organic C than NO^-₃ for denitrification to proceed to completion. However, this assumption ignores another widely accepted fact that buried organic matter may also provide the organic C required for denitrification (Chapelle, 1993; Appelo and Postma, 1996). Also, because this buried organic matter often is refractory, and may limit microbial processes, its reactivity controls the extent of denitrification (Chapelle, 1993; Appelo and Postma, 1996).

Sediment organic C content was variable but ranged from 0.01 to 1.45% with a median of 0.15%. Applying an analysis similar to that of Trudell et al. (1986), and assuming a porosity of 0.3 and a particle density of 2.65 g cm^-3, results in an estimate of 52 to 7471 mmol organic C with a median of 773 mmol C, in each 3.3-L volume of sediment. Based on the stoichiometry of denitrification in Eq. [2], this would translate to the ability to reduce 42 to 5977 mmol NO^-₃ in each liter of water in contact with each 3.3-L volume of sediment. Given that the maximum concentration of NO^-₃ measured along the transect was 1785 µmol L^-1, it is obvious that even the relatively small amounts of organic C present in the aquifer are adequate to explain the redox conditions. Grischek et al. (1998) reached similar conclusions in their study of a sand and gravel aquifer in Germany that had received organic C from infiltrating Elbe River waters.

At the smallest organic C concentration (0.01%) and the median NO^-₃ concentration at the water table (436 µmol L^-1), all of the O₂ (373 µmol L^-1 assuming equilibrium with the atmosphere) and NO^-₃ originally present should be consumed in the upper 2 m of the saturated zone. That this is not occurring suggests that either the organic C is not readily accessible for O₂ reduction and denitrification, or that other biological, thermodynamic, and/or kinetic constraints exist with the result that most of the aquifer appears to be C limited, resulting in slow denitrification rates.

Denitrification Rates
Denitrification rates are typically measured in controlled incubations under laboratory conditions; however, these rates are typically greater than those determined by in situ studies, or even tracer studies and all appear to overestimate NO^-₃ losses under natural conditions. Smith et al. (1996) attempted to correct for these overestimates by injecting Br^- as a tracer along with NO^-₃ and then monitoring along the flow path for denitrification products. The rates calculated from their experiments were less than slurry and soil core incubations done with the same material but were of the same order of magnitude. Other in situ studies (Trudell et al., 1986; Gillham et al., 1990; Tesoriero et al., 2000) typically have produced rates an order of magnitude greater than those of Smith et al. (1996). Another method is to measure the production of excess N₂ or N₂O gas generated from denitrification and estimate a denitrification rate based on sample age (Vogel et al., 1981; Smith et al., 1996; Tesoriero et al., 2000). Rates based on the excess N₂ method may be an order of magnitude lower than those based on tracer tests.

Denitrification rates based on excess N₂ gas and sample age measured in this study, ranged from 0.005 to 0.047 mmol NO^-₃ L^-1 yr^-1. These rates are orders of magnitude lower than what has typically been measured in laboratory and in situ incubations (Table 6) but are comparable to those from other field-based studies in sand and gravel aquifers (Bohlke and Denver, 1995; Smith et al., 1996; Tesoriero et al., 2000). The slowest denitrification rates (Table 6) were from Vogel et al. (1981), who reported a loss of 1.5 mmol NO^-₃ over a period of about 14000 yr for a C limited system in the Kalahari Desert. Rates reported by Tesoriero et al. (2000) overlap those reported here and were dependent on both C and pyrite weathering providing Fe²⁺ as the electron donor. The somewhat faster rates reported by Smith et al. (1996) were measured in an aquifer contaminated with treated sewage, but denitrification was still C limited. These studies indicate that rates measured under field conditions are generally much lower than either laboratory or isolated in situ incubations and we theorize that they reflect the natural variability of conditions favorable for denitrification. Because there are no additions or modifications to the natural system (other than well installation and sampling) these rates may come closest to actual denitrification rates. Rates derived in this manner therefore are likely better measures of natural denitrification rates and would provide more accurate predictions of NO^-₃ transport and fate.

View larger version (27K): [in this window] [in a new window]   Fig. 11. Distribution of reconstructed (original) concentrations of NO-3 calculated as the sum of excess N2 and NO-3 measured along the transect.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The results of this study demonstrate the importance of developing a complete understanding of the hydrogeologic and biogeochemical processes involved when evaluating the fate of ground water contaminants. Age-dating techniques combined with sampling along flow paths, allow us to develop a chronology of the change in contaminant concentrations over time and to relate those changes to historical changes in land-use and management practices. Because residence times on the order of 50 yr or more are not unusual, there may be adequate time in many aquifers for even slow biogeochemical process to modify original contaminant concentrations. These modifications require us to monitor not only the original contaminant concentrations but the breakdown products as well to understand the processes involved. The final result, however, is a more comprehensive appreciation of the biogeochemical processes involved than could be gained from a one- or two-dimensional sampling approach. More studies are required to document the spatial extent of the role of denitrification in removing NO^-₃ from contaminated aquifers; it is likely, however, that many other aquifers in settings similar to the Otter Tail aquifer contain sufficient organic C to support denitrification and natural bioremediation.

	ACKNOWLEDGMENTS

 
This study was conducted as part of the U.S. Geological Survey National Water-Quality Assessment Program (NAWQA) and benefited from the assistance of Jeffrey Stoner, Lan Tornes, Mark Brigham, and David Lorenz. We particularly appreciated the technical reviews of Timothy Spruill, James LaBaugh, Jeffrey Stoner, and two anonymous reviewers. The use of trade names in this paper is for identification purposes only and does not constitute endorsement by the U.S. Geological Survey.

	REFERENCES

TOP ABSTRACT INTRODUCTION STUDY AREA METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 

• Addy, K.L., A.J. Gold, P.M. Groffman, and P.A. Jacinthe. 1999. Ground water nitrate removal in subsoil of forested and mowed riparian buffer zones. J. Environ. Qual. 28:962–970.[Abstract/Free Full Text]
• American Public Health Association. 1975. Standard methods for the examination of water and wastewater. 14th ed. APHA, Washington, DC.
• Appelo, C.A.J., and D. Postma. 1996. Geochemistry, groundwater and pollution. A.A. Balkema, Rotterdam.
• Berndt, M.P. 1987. Metal partitioning in a sand and gravel aquifer contaminated by crude petroleum, Bemidji, Minnesota. M.S. thesis. Syracuse Univ., Syracuse, NY.
• Blicher-Mathiesen, G., G.W. McCarty, and L.P. Nielsen. 1998. Denitrification and degassing in ground water estimated from dinitrogen and argon. J. Hydrol. 208:16–24.
• Bohlke, J.K., and J.M. Denver. 1995. Combined use of ground water dating, chemical, and isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural watersheds, Atlantic coastal plain, Maryland. Water Resour. Res. 31:2319–2339.
• Brenton, R.W., and T.L. Arnett. 1992. Determination of dissolved organic carbon by UV-promoted persulfate oxidation and infrared spectrometry. Open-File Rep. U.S. Geol. Surv. 92-480.
• Busenberg E., and L.N. Plummer. 1992. Use of chlorofluorocarbons (CCl₃F and CCl₂F₂) as hydrologic tracers and age-dating tools: The alluvium and terrace system of central Oklahoma. Water Resour. Res. 28:2257–2283.
• Busenberg, E., E.P. Weeks, L.N. Plummer, and R.C. Bartholomay. 1993. Age dating ground water by use of chlorofluorocarbons (CCl₃F and CCl₂F₂), and distribution of chlorofluorocarbons in the unsaturated zone, Snake River Plain Aquifer, Idaho National Engineering Laboratory, Idaho. Water Resour. Invest. Rep. U.S. Geol. Surv. 93-4054.
• Busenberg E., L.N. Plummer, R.C. Bartholomay, and J.E. Wayland. 1998. Chlorofluorocarbons, sulfur hexafluoride, and dissolved permanent gases in ground water from selected sites in and near the Idaho National Engineering and Environmental Laboratory, Idaho, 1994–97. Open File Rep. U.S. Geol. Surv. 98-274.
• Chapelle, F.H. 1993. Ground-water microbiology and geochemistry. John Wiley & Sons, New York.
• Chapelle, F.H., P.B. McMahon, N.M. Dubrovsky, R.F. Fujii, E.T. Oaksford, and D.A. Vroblevsky. 1995. Deducing the distribution of terminal electron-accepting processes in hydrologically diverse ground water systems. Water Resour. Res. 31:359–371.
• Cook, P.G., and D.K. Solomon. 1997. The transport of atmospheric trace gases to the water table: Implications for ground water dating with chlorofluorocatbons and Krypton-85. Water Resour. Res. 31:263–434.
• Cook, P.G., D.K. Solomon, L.N. Plummer, E. Busenberg, and S.L. Schiff. 1995. Chlorofluorocarbons as tracers of ground water transport processes in a shallow, silty and aquifer. Water Resour. Res. 31:425–434.
• Cowdery, T.K. 1997. Shallow ground-water quality beneath cropland in the Red River of the North Basin, Minnesota and North Dakota, 1993–95. Water Resour. Invest. Rep. U.S. Geol. Surv. 97-4001.
• Fetter, C.W. 1994. Applied hydrology. 3rd ed. Prentice-Hall, Upper Saddle River, NJ.
• Fishman, M.J. 1993. Methods of analysis by the U.S. Geological Survey National Water Quality Laboratory: Determination of inorganic and organic constituents in water and fluvial sediments. Open File Rep. U.S. Geol. Surv. 93-125.
• Fishman, M.J., and L.C. Friedman. 1989. Methods for determination of inorganic substances in water and fluvial sediments. U.S. Geological Survey techniques of water-resources investigations. Book 5. U.S. Gov. Print. Office, Washington, DC.
• Gillham, R.W., and J.A. Cherry. 1978. Field evidence of denitrification in shallow ground water flow systems. Water Pollut. Res. Can. 13:53–71.
• Gillham, R.W., R.C. Starr, and D.J. Miller. 1990. A device for in situ determination of geochemical transport parameters: II. Biochemical reactions. Ground Water 28:858–862.
• Gold, A.J., P.A. Jacinthe, P.M. Groffman, W.R. Wright, and R.H. Puffer. 1998. Patchiness in ground water nitrate removal in a riparian forest. J. Environ. Qual. 27:146–155.[Abstract/Free Full Text]
• Goldstein, B.S. 1985. Stratigraphy, sedimentology, and Late-Quaternary history of the Wadena Drumlin Region, central Minnesota. Ph.D. thesis. Dep. of Geology and Geophysics, Univ. of Minnesota, Minneapolis, MN.
• Grischek, T., K.M. Hiscock, T. Metschies, P.F. Dennis, and W. Nestler. 1998. Factors affecting denitrification during infiltration of river water into a sand and gravel aquifer in Saxony, Germany. Water Res. 32:450–460.
• Harbaugh A.W., and M.G. McDonald. 1996. User's documentation for MODFLOW-96, an update to the U.S. Geological Survey modular finite-difference ground-water flow model. Open File Rep. U.S. Geol. Surv. 96-485.
• Heaton, T.H.E., and J.C. Vogel. 1981. "Excess air" in ground water. J. Hydrol. 50:201–216.
• Hedges, J.L., and J.H. Stern. 1984. Carbon and nitrogen determinations of carbonate-containing solids. Limnol. Oceanogr. 29:657–663.
• Jacinthe, P.A., P.M. Groffman, A.J. Gold, and A. Mosier. 1998. Patchiness in microbial nitrogen transformations in ground water in a riparian forest. J. Environ. Qual. 27:156–164.[Abstract/Free Full Text]
• Johnston, C.T., P.G. Cook, S.K. Frape, L.N. Plummer, E. Busenberg, and R.J. Blackport. 1998. Ground water age and nitrate distribution within a glacial aquifer beneath a thick unsaturated zone. Ground Water 36:171–180.
• Kolle, W., P. Werner, O. Strebel, and J. Bottcher. 1983. Denitrifikation in einem reduzierenden grundwasserleiter. Vom Wasser. 61:125–147.
• Korom, S.F. 1992. Natural denitrification in the saturated zone: A review. Water Resour. Res. 28:1657–1668.
• Koterba, M.T., F.D. Wilde, and W.W. Lapham. 1995. Ground-water data-collection protocols and procedures for the National Water-Quality Assessment Program: Collection and documentation of water-quality samples and related data. Open File Rep. U.S. Geol. Surv. 95-399.
• Lovely, D.R., and J.C. Woodward. 1992. Consumption of freons CFC-11 and CFC-12 by anaerobic sediments and soils. Environ. Sci. Technol. 26:925–929.
• McDonald, M.G., and A.W. Harbaugh. 1988. A modular three-dimensional finite-difference ground-water model. U.S. Geological Survey techniques of water-resources investigations. Book 6. U.S. Gov. Print. Office, Washington, DC.