JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 20 April 2005
Published in J Environ Qual 34:825-835 (2005)
DOI: 10.2134/jeq2004.0134
© 2005 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Supplemental Data (63 KB)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Browne, B. A.
Right arrow Articles by Guldan, N. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Browne, B. A.
Right arrow Articles by Guldan, N. M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Browne, B. A.
Right arrow Articles by Guldan, N. M.
Related Collections
Right arrow Surface Water Quality
Right arrow Biogeochemical Processes
Right arrow Watershed and Landscape Processes
Right arrow Watershed-Scale Studies
Right arrow Nitrogen

TECHNICAL REPORTS

Landscape and Watershed Processes

Understanding Long-Term Baseflow Water Quality Trends Using a Synoptic Survey of the Ground Water–Surface Water Interface, Central Wisconsin

Bryant A. Browne* and Nathan M. Guldan

College of Natural Resources, University of Wisconsin-Stevens Point, Stevens Point, WI 54481

* Corresponding author (bbrowne{at}uwsp.edu)

Received for publication March 31, 2004.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 CONCEPTUAL MODEL
 STUDY AREA
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The relationship between stream water quality and landscape activities is difficult to evaluate where the principal source of stream flow is ground water seepage because the average travel time from ground water recharge areas to stream discharge positions can be on the order of decades. We tested the idea that past and future baseflow water quality can be predicted based on a synoptic survey of ground water recharge age-dates (based on chlorofluorocarbon [CFC] measurements) and water quality measurements obtained at the ground water–surface water interface. In this study we (i) characterize the discharge-weighted age distribution and water quality of ground water seepage into the Little Plover River (LPR); (ii) use this information to backcast and forecast baseflow NO3– concentrations; and (iii) evaluate NO3– backcasts against historical baseflow data (1960 to 2000). The discharge-weighted apparent CFC age of ground water seepage into the LPR was 23.7 (±7) yr. Baseflow backcasts matched the four decade rise of baseflow NO3– from 2 to 8 mg L–1. Baseflow forecasts included three scenarios. Scenario A projects the historical rise of NO3– in the LPR basin's ground water recharge through 2050. Scenario B projects a leveling off of NO3– in ground water recharge in the year 2000. Scenario C projects a leveling off in the year 1985. Under Scenario A, LPR baseflow NO3– will increase steadily from 8 to 19 mg L–1 between 2000 and 2050. Under scenarios B and C baseflow NO3– will plateau at 13 mg L–1 in 2030 and at 10 mg L–1 in 2010, respectively. The approach developed in this study can be used to (i) reconstruct historical baseflow water quality patterns in the absence of long-term monitoring data and (ii) project the effects of potential management decision on future water quality.

Abbreviations: CFC, chlorofluorocarbon • LPR, Little Plover River • PIE, pumping-induced ebullition


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
CONCEPTUAL MODEL
 STUDY AREA
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
IN RECENT YEARS, water resources managers have become increasingly aware of the importance of understanding the contribution of ground water to the quality of surface water (Winter et al., 1998; Jones and Mulholland, 2000, and references therein; Puckett et al., 2002). In baseflow-dominated streams, where the principal source of stream flow is continuous ground water seepage rather than episodic runoff, the relationship between land use and management practices in ground water source areas and contemporary stream water quality is difficult to decipher. Ground water movement from recharge areas to stream discharge positions (residence time or lag time) can be on the order of decades (Bohlke and Denver, 1995; Modica et al., 1998; Katz et al., 2001). Because of the delayed stream response to land management activities, the scientific rationale and potential benefits of resource management decisions can be difficult to defend politically and economically.

Few watersheds have a sufficiently long water quality record or the detailed historical land use information needed to explore the lagged relationships between land management practices and baseflow water quality trajectories (e.g., rising nitrate concentrations in agricultural landscapes). Moreover, funds and resources to initiate new long-term stream monitoring–geographical information system (GIS) programs remain difficult to come by because the potential to see noticeable changes is so far in the future. This existing and pending information deficit will continue to hamper progress toward an adequate, practical scientific framework for baseflow water quality management decisions tailored to individual basins. Thus, there is a need to establish an alternative basis for (i) assessing baseflow water quality–land use relationships and (ii) making resource management decisions aimed at improving baseflow water quality.

A number of recent studies have coupled predictions of ground water discharge water quality to historical changes in ground water recharge water quality based on measured or modeled ground water residence times (e.g., Kaufman et al., 2001; Bohlke, 2002; Lindsey et al., 2003). The purpose of this paper is to test the idea that past and future baseflow water quality can be predicted based on a synoptic survey of ground water recharge age-dates and water quality measurements obtained at the ground water–surface water interface. In this study we (i) characterize the discharge-weighted age distribution and water quality of ground water seepage into a baseflow-dominated stream in central Wisconsin, (ii) develop an approach to use this information to backcast and forecast baseflow nitrate (NO3–) concentrations, and (iii) evaluate the quality of baseflow concentration estimates against historical baseflow water quality measurements. We show that this approach can be used to reconstruct historical baseflow water quality patterns in the absence of long-term monitoring data and be used to project the effects of potential management decision on future water quality.


   CONCEPTUAL MODEL
TOP
 ABSTRACT
 INTRODUCTION
 CONCEPTUAL MODEL
STUDY AREA
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
In this study we expand on a conceptual model of ground water entry to streams developed by Modica et al. (1997) (hereafter referred to as the MRP model) and applied by Modica et al. (1998). Based on the MRP model, baseflow water quality on a particular date is essentially a volumetric blend of ground water discharge that originates in recharge areas near the stream (short residence time, recharged recently) and far from the stream (long residence time, recharged long ago). In Fig. 1a we convey this idea using a hypothetical ground water recharge age distribution for ground water discharging along a stream course. (The distribution is presented as a normal distribution for conceptual simplicity only and is not meant to suggest an expectation of normally distributed ages in this or other studies.) The relative frequency values, wi, are weighted by discharge to reflect the proportional contribution of each recharge date i to baseflow. This distribution is superimposed on a pattern of the basin-wide (volumetrically averaged) mean concentration () of a hypothetical chemical constituent in ground water recharge, illustrating that the chemistry of baseflow evolves from a volumetric blending of ground water ages of varying chemical concentrations.



View larger version (40K):
[in this window]
[in a new window]
  		Fig. 1. Conceptual model of baseflow water quality estimation procedure (Eq. [1], text) using ground water seepage along a stream meander. (a) Hypothetical ground water recharge age-date distribution of ground water seepage into a stream course (inset) on three baseflow sampling dates superimposed on , the basin-wide mean annual concentration of a hypothetical chemical in ground water recharge. (b) Estimated baseflow concentrations, Cbf, for each baseflow sampling date by Eq. [1]. (c) Minipiezometer network along a stream meander intercepts ground water seepage of varying recharge age-dates and water quality characteristics. Schematic illustrates design of minipiezometers used in this study.

 
The age distribution in Fig. 1a is shifted along the time axis in 20-yr intervals to show how the water quality blend comprising baseflow would vary on three baseflow sampling dates. For each date, the baseflow concentration of a conservative chemical constituent can be represented as a weighted mean:

[1]
where wi represents the volumetric contribution of ground water recharged on date i and represents the basin-wide mean concentration in ground water recharged on date i. Thus, assuming that the discharge-weighted age distribution of ground water entering the stream can be measured and remains relatively constant in time, Fig. 1a raises the idea that estimates of the concentration of a chemical in past and future baseflow, Cbf, can be generated based on historical or predicted temporal patterns of the mean annual concentration in ground water recharge (Fig. 1b).

According to the MRP model, the age-date distribution of ground water discharging to a stream varies with location along the stream, changing systematically with position along and across the channel. Where the stream channel parallels the direction of ground water flow, the central tendency of the age distribution of ground water seepage into the stream should become older with distance downstream. In addition, assuming a hypothetically straight channel to which ground water flow converges symmetrically from opposite directions, the age of ground water seepage should become older with distance from the banks to the center of the channel at a particular position along the stream course.

Real stream systems are generally sinuous (the thalweg meanders even in straight channels) (Leopold, 1997) and the convergence of ground water flow from opposite sides of a stream discharge zone is as likely to be asymmetric (e.g., stronger from one side of the channel than the other) as balanced. So with respect to cross-stream age patterns, chances of finding the oldest water dead center of the channel and finding highly symmetrical age patterns at any position along an actual stream course are fairly limited. It is perhaps more realistic then to describe a gaining stream as an indefinitely positioned ground water seepage area across which a channel meanders (Fig. 1c). From this viewpoint, a meandering thalweg can be seen as intercepting discharge zones of younger and older water, drifting across indefinitely located maximum age positions, as it advances along the centerline of the stream trajectory.

These considerations suggest that the age distribution and quality of ground water entering a gaining stream should be accessible immediately beneath the stream sediment along the meandering thalweg of a stream channel (Fig. 1c). We hypothesized (i) that trends in basin-wide mean annual concentrations () of NO3– in ground water recharge over several decades could be inferred from regression relationships between NO3– concentrations (estimates of Cr,i) and apparent CFC ages measured in ground water seepage along a thalweg, and (ii) that the trends could be used in conjunction with the apparent CFC age distribution data to backcast and forecast Cbf (Eq. [1]) in a manner analogous to Fig. 1a and 1b.


   STUDY AREA
TOP
 ABSTRACT
 INTRODUCTION
 CONCEPTUAL MODEL
 STUDY AREA
METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The Little Plover River is a 10-km-long, third-order west-flowing tributary of the Wisconsin River. The LPR drains approximately 1300 ha of the Wisconsin central sand plain, where an important unconfined aquifer supports a highly productive agricultural landscape (Fig. 2) . Ground water discharge dominates the LPR's annual hydrograph; Weeks et al. (1965) reported that 90% (22.5 cm) of the total annual runoff (25 cm) is released as baseflow and 10% (2.5 cm) as direct runoff. Thus, LPR stream water quality is largely determined by continuous ground water seepage rather than episodic runoff.



View larger version (41K):
[in this window]
[in a new window]
  		Fig. 2. Study area and minipiezometer network. Arrows: downstream ends of Reaches F–I. Shaded area represents an approximation of the Little Plover River (LPR) ground water basin boundary provided by the Central Wisconsin Groundwater Center via MODFLOW (Harbaugh and McDonald, 1996; physical and hydrologic data used in the model are on file at the CWGC).

 
The LPR drainage network has an average sinuosity (stream length divided by straight line distance) of 1.6. It consists of headwater agricultural ditches (Reaches A–E) that drain into to a third-order natural meander (Reaches F–I). The upper drainage network, which originates immediately below the terminal (Arnott) moraine, includes five agricultural drainages (Reaches A–E) and a section of natural meander (Reach F). These reaches derive most of their perennial discharge from ground water flow systems recharged in the sandy and stony unsorted till of the glaciated landscape above the terminal moraine (Fig. 2). The lower drainage network is a section of natural meander that includes three reaches (G, H, I) partitioned to identify important breakpoints in the hydraulic connection with the landscape. Reaches G and H are permanent and intermittent losing reaches, respectively, where ground water is recharged by surface water (i.e., by ground water previously discharged in upstream reaches, which potentially reequilibrates with modern atmosphere within the stream channel). Reach I is an intense gaining reach where a major portion of the LPR's ground water discharge is derived. The lower drainage network derives ground water seepage from glacial outwash recharge areas below the terminal moraine but may also capture regional ground water that escapes capture by the upper drainage network.

The amplitude of thalweg meander in the LPR is at least half the width of the stream channel in straight sections of the LPR (i.e., the upper drainage ditches established >40 yr ago) due to recurring stream reworking during bankfull events. Stream reworking has also produced meanders with amplitudes far greater than the width of channel in the natural meander sections of the LPR. Hence, the drift of the LPR thalweg is probably much larger than the theoretical age symmetry of the ground water discharge system throughout most of the LPR. Thus, as the thalweg progresses down the stream corridor, it has the potential to provide an unbiased selection of sampling positions to capture young and old ground water discharge. Consistent with these observations, in the early stages of our work on the LPR, stream cross-section measurements at four locations along the LPR (i) showed the absence of both age symmetry and hydraulic head symmetry beneath and transverse to the stream channel, and (ii) revealed maximum ages and minimum hydraulic heads independent of the thalweg position.

Measurements of stream water quality and streamflow at Hoover Road (Fig. 2), collected sporadically since 1967, have been reported by Albertson (1998) and Mechenich and Kraft (1997). Mechenich and Kraft (1997) reported that baseflow NO3– concentrations increased from 210 µmol L–1 (about 3 mg N L–1) in 1967 to 570 µmol L–1 (about 8 mg N L–1) in 1996. Assuming an average annual baseflow of 0.23 m3 s–1 (8 ft3 s–1) at Hoover Road (Albertson, 1998), these concentration changes suggest that the stream NO3– load delivered by ground water discharge has increased 98 kg N d–1 (from 58 to 156 kg N d–1) over a 30-yr period. Albertson (1998) attributed the dramatic rise of baseflow NO3– to basin-wide ground water contamination associated with a shift from nonirrigated to irrigated agriculture and greater use of fertilizer N.


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 CONCEPTUAL MODEL
 STUDY AREA
 METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
A network of minipiezometers (Fig. 2) at the ground water–surface water interface was used to determine the concentrations of NO3– and dissolved gases (Ar, N2, O2, CFCs) at the tail end of ground water flowpaths as they discharge to the LPR. The network included 163 minipiezometer positions spaced at 60-m intervals along the approximate centerline of the stream meander. Minipiezometers (Fig. 1c) were constructed from polyethylene tubing (0.43-cm i.d., 0.63-cm o.d.). Well screens, which consist of 8 to 10 lines of perforations 2.5 cm long, were created in four to five passes through a sewing machine. A retractable tempered stainless steel insertion rod was used to position the well screens at a depth 60 cm below the stream–sediment interface. The selection of the 60-cm depth was based on several practical considerations including: (i) the minipiezometers would generally stay put at this depth even during stream reworking events, (ii) the minipiezometers could be readily installed simply by pushing the well to the 60-cm depth, and (iii) there was an apparent general absence of sediment stratification or lenses of coarser or finer sediments over the 0- to 60-cm interval based on probing during well insertion and based on a survey of the sediments with a tile probe. Interstitial water samples from the minipiezometers and baseflow grab samples from the stream were collected in five synoptic surveys between 1996 and 2000 (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Synoptic survey data.

 
Specific discharge (m3 m–2 s–1) was calculated using Darcy's law (Lee and Cherry, 1978):

[2]
where Kj and dhj/dzj represent the vertical hydraulic conductivity of streambed sediment and the hydraulic gradient, respectively, at the jth sampling position. Due to (i) the evident looseness of the streambed sediments to a depth of 60 cm, (ii) the apparent absence of lensing or layering to the 60-cm depth inferred by probing with a tile probe, and (iii) good agreement between estimates of Kj obtained by the Hvorslev falling head test and by vertical seepage meter measurements early in the study, we assumed isotropic (Kx = Ky = Kz) conditions within the 0- to 60-cm sediments of the streambed. Based on this assumption, values of Kj were approximated at all locations from measurements of the horizontal hydraulic conductivity by the Hvorslev (1951) falling head test, which can be performed quickly and reproducibly in the field. The hydraulic gradient was determined from the height (dhj) of the static head relative to the stream water surface and the depth of the well screen (dzj) below the streambed. Specific discharge per unit of stream length (m3 m–1 s–1) was obtained by:

[3]
where wavg,j is an estimate of the average width of stream between positions j – 1 and j + 1. Ground water seepage (m3 s–1) was obtained by:

[4]
where lj is the distance between positions j and j + 1.

Ground water samples were collected using a peristaltic pump attached directly to the minipiezometer. Samples were field filtered (0.45-µm nitrocellulose membrane filter), chemically preserved (sulfuric acid for nutrients, nitric acid for metals), and stored at 4°C for analysis of NO3– in the laboratory. Field measurements, including water temperature, pH, specific conductance, dissolved oxygen, and oxidation reduction potential, were obtained using a Model 650 DMS sonde (YSI, Yellow Springs, OH). Total dissolved gas pressure (D'Aoust and Clark, 1980) (PT) was measured in the field using a total dissolved gas pressure sensor, calibrated using an air equilibration procedure. A sealed flow-through cell, connected to the outlet of the peristaltic pump, was used to obtain sonde and PT readings from the minipiezometer samples without exchanging gases with the atmosphere. The combined concentration of NO3– and NO2– was analyzed as N by automated colorimetry. Because NO2– is essentially undetectable (<0.1 mg N L–1) in the LPR ground water the results are reported as NO3– alone.

Dissolved gas samples were collected using two approaches. Before 2000, CFC samples were collected in the field as water samples. Dissolved CFC concentrations (mol L–1) were then determined in the laboratory using a custom purge and trap gas chromatograph equipped with an electron capture detector (ECD). Field and laboratory sampling equipment and procedures are described in Busenburg and Plummer (1992).

Starting in 2000, samples for gas analyses (CFCs, Ar, N2) were collected as gas samples in the field using pumping-induced ebullition (PIE) (Browne, 2004). The PIE gas samples were maintained in gas-tight 10-mL syringes until dry gas mole fractions (X) could be determined by gas chromatography in the laboratory (<48 h). Argon and N2 measurements were performed using high purity helium carrier gas and a pulse discharge detector (PDD) in helium ionization mode (Wentworth et al., 1994); CFC measurements were performed using high purity nitrogen carrier gas and a 63Ni ECD. Chromatographic conditions (columns, temperatures, and carrier gas flow rates) were similar to those reported in Busenburg and Plummer (1992). Field blanks for CFCs were accomplished by installing and PIE sampling minipiezometers at a nearby stream location known to discharge ground water devoid of CFCs (i.e., having an apparent CFC age-date < 1945). Chlorofluorocarbon measurements by the USGS (Busenburg and Plummer, 1992) and PIE procedures in the LPR yielded similar results during this study. Contemporaneous, co-located CFC measurements (USGS and PIE) have been shown to be in close agreement (Browne, 2004).

Dissolved gas concentrations at field water temperature were obtained by Henry's law calculation:

[5]
where C is the concentration of gas in mol L–1; KH is a gas specific Henry's law constant (mol L–1 MPa–1); PT and w are the total dissolved gas pressure (MPa) and water vapor pressure (MPa), respectively; and Fc is a unitless gas-specific coefficient that adjusts for fractionation during the PIE sample collection process [values of Fc are reported in Browne (2004)].

The dissolved concentrations of CFCs were used to develop apparent CFC recharge age-dates using a Henry's law approach (described in more detail in Plummer and Busenburg, 2000). Dry gas mole fractions (in parts per trillion by volume, pptv) of CFCs at the time of recharge were obtained using the following relationship:

[6]
where C (mol L–1) is the dissolved concentration of CFC11, CFC12, or CFC113; P and w are the general atmospheric pressure (MPa) based on elevation and the water vapor pressure (MPa) at the recharge temperature, respectively; and KH is the Henry's law constant (mol L–1 MPa–1) of CFC11, CFC12, or CFC113. The apparent temperature of recharge was determined from the measured concentrations of dissolved Ar and N2 (Busenburg et al., 1993), as described below, after correction of total N2 concentrations for N2 produced by denitrification. Dry air mole fractions of CFC11, CFC12, and CFC113 obtained by Eq. [6] were referenced to chronological records of atmospheric mixing ratios for CFCs to determine a unique "apparent" ground water age-date or recharge date i for each minipiezometer position j. (Chlorofluorocarbon dry air mole fractions greater than possible within the modern troposphere were attributed to local environmental contamination and were excised from the dataset.) Apparent age was assigned as the date of sampling minus the apparent recharge date. Where there was close age agreement between apparent CFC ages (e.g., <5 yr), age estimates were averaged. For wider age discrepancies under low dissolved O2, the apparent CFC12 age was accorded highest priority, followed by CFC113 and CFC11, respectively, based on relative susceptibilities (CFC12 < CFC113 << CFC11) to microbial degradation (Plummer and Busenburg, 2000). For wider age discrepancies under higher O2 concentrations, a younger apparent CFC age (e.g., CFC12 vs. CFC113) was assumed to reflect slight environmental contamination and was excluded from age assignments.

As in many other CFC tracer studies, apparent age assignments in this study assume piston flow, as if the water flowed without substantial mixing via hydrodynamic dispersion from the point of recharge to the point of discharge (Plummer and Busenburg, 2000). We also assumed that mixing of varying age waters was minimal during the sample collection process. This assumption was based on previous measurements of subsurface age variation in several stream cross-sections in the LPR. These observations showed that horizontal and vertical age variation was fairly uniform relative to the scale of influence (drawdown) at the narrow (2.5 cm) minipiezometer well screen during sample pumping. Furthermore, because of the confounding effect of spotty CFC12 contamination in the LPR and the susceptibility of CFC11 to degradation, the use of CFC tracer ratios (e.g., Lindsey et al., 2003) to diagnose or apply other mixing models (e.g., exponential mixing or binary mixing; Cooke and Bohlke, 2000) to interpret the age variation of ground water mixtures was not globally feasible. Lindsey et al. (2003) found little difference in age interpretations for piston, exponential, and binary mixing for a large number spring samples collected in streams of the Chesapeake Bay watershed.

The dissolved concentrations of N2 and Ar gas in ground water were used to estimate the excess N2 produced by denitrification (excess N2) (Vogel et al., 1981; Martin et al., 1995) and the apparent recharge temperature. The atmosphere and denitrification were assumed to be the only sources of N2 in ground water. The amount of excess N2 in each sample was estimated from the difference between the N2 concentration in the sample and the N2 concentration in air-saturated water with the same Ar concentration as the sample (Bohlke and Denver, 1995). The potential contribution of excess air (supersaturation by dissolution of entrapped air bubbles during recharge; Heaton and Vogel, 1981; Holocher et al., 2003) to dissolved Ar and N2 was checked by iteratively solving Henry's law expressions for the excess air amount providing the same apparent recharge temperature for both the Ar and corrected N2 concentrations. These calculations revealed excess air to be negligible in our samples; thus, apparent recharge temperatures were based on the measured dissolved Ar concentration (and corresponding corrected N2 concentration), assuming equilibration with general atmosphere (i.e., air-saturated water).

Values of the initial total concentration of NO3– in ground water (Cr,i) for each apparent CFC recharge date i were reconstructed from the sum of measured NO3– and excess N2 concentrations in each ground water seepage sample for the 27 June 2000 synoptic survey:

[7]

Denitrification reaction progress in 27 June 2000 ground water was gauged by the following ratio:

[8]

Because these calculations suggested that excess N2 was generally a small component of total NO3– (see Results and Discussion), values of Cr,i for all other synoptic survey sampling dates were approximated by measured NO3– concentrations.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 CONCEPTUAL MODEL
 STUDY AREA
 METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
We assessed the reliability of the minipiezometer network by comparing (i) the summation of ground water seepage ({Sigma}Qj) measured along the stream course to actual streamflow measurements and (ii) the discharge-weighted mean concentration of NO3– in ground water seepage ({Sigma}QjCj/{Sigma}Qj) to the measured concentration of NO3– in baseflow along the stream course. Figure 3 illustrates the results for the 27 June 2000 synoptic survey. The values of Qj were summed in sequence (j = 1 to j = 163) from the upstream end to the downstream end of each reach and the reaches were summed in sequence from Reaches A to I. The running total produced a fairly accurate replica of the actual stream flow along the entire stream course (Fig. 3a). A running estimate of the discharge-weighted mean NO3– concentration was tallied in the same sequence. The results show that the discharge-weighted mean NO3– concentration in ground water seepage converged rapidly with actual concentrations measured along the stream course (Fig. 3b). Similar agreement between the ground water seepage and stream data was observed across all synoptic survey dates (Table 1). This result demonstrates that the minipiezometer network provided a quantitative and representative sample of ground water seepage contributing to stream baseflow. The cumulative accuracy provided by the specific discharge measurements probably reflects (i) generally isotropic conditions of the 0- to 60-cm sediments in the LPR; (ii) the low spatial variation in hydraulic conductivity along the stream corridor (25th, 50th, and 75th percentile Kj values were 0.0043, 0.0092, and 0.0124 cm s–1, respectively, over n = 145 sites); and (iii) the large number of sample sites used in this study.



View larger version (35K):
[in this window]
[in a new window]
  		Fig. 3. Synoptic survey data for 27 June 2000. (a) Ground water seepage (Qj, Eq. [4], solid triangle) along the stream course and comparison of cumulative seepage ({Sigma}Qj, solid line) to measured streamflow (open circles). Letters along x axis indicate downstream ends of Reaches A–I (Fig. 2). (b) NO3– concentrations in ground water seepage (solid triangle) and comparison of discharge-weighted mean concentration ({Sigma}QjCj/{Sigma}Qj, solid line) with measured baseflow concentration (open circles).

 
The relationship between NO3– concentrations and the apparent CFC age-date of ground water discharge to the LPR is shown in Fig. 4 for all synoptic sampling events. Based on denitrification reaction progress (Eq. [8]) data obtained for the 27 June 2000 synoptic survey (discharge-weighted mean denitrification reaction progress = 20 ± 14%), NO3– is by far the largest component of total NO3– in ground water seepage to the LPR. Thus, in the absence of excess N2 measurements for all sampling dates, the Fig. 4 data provide an indication of the landscape variation of total NO3– (Cr,i) in ground water recharged over the last four decades.



View larger version (34K):
[in this window]
[in a new window]
  		Fig. 4. Nitrate concentrations and apparent chlorofluorocarbon (CFC) ground water recharge dates of ground water seepage measurements across all synoptic survey dates (Table 1). Hatched and solid lines: historical record of fertilizer N sales in the United States and Wisconsin (USDA, 1997), respectively.

 
For any particular age interval in Fig. 4, the vertical concentration variation reflects spatial variation in land use–land practices. The broadening vertical concentration variation of NO3– in ground water with age-dates between the 1960s to the late 1980s is driven by rising high-end concentrations that appear to track the historical expansion and/or intensification of agriculture (see, for example, historical fertilizer sales curves in Fig. 4; USDA, 1997). Reaching an apparent maximum in ground water recharged in the early to mid-1980s, high-end NO3– concentrations potentially stabilized or dropped in ground water recharged through the 1990s. In contrast, low-end concentrations remained fairly constant in ground water recharge in the 1950s through the 1990s, probably representing nonagricultural recharge areas of the landscape.

We developed our predictions of Cbf (NO3–) using a subset of the Fig. 4 synoptic survey data. We restricted the focus of our analysis to the 27 June 2000 dataset (Fig. 5) because it includes (i) coverage of both the upper and lower drainage networks (Fig. 2, Table 1) and (ii) measurements of excess N2 (see the supplemental data section [available free of charge] in the online version of this manuscript, available at http://jeq.scijournals.org/) that allowed us to explicitly consider the influence of denitrification on baseflow NO3– predictions. Moreover, recognizing that the expense in time and money of accumulating the amount of data in Fig. 4 would be unrealistic in routine watershed assessment, we wished to test whether accurate baseflow water quality estimates by Eq. [1] could be generated for the LPR with the results of only one synoptic survey at the ground water–surface water interface.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 5. Nitrate concentrations and apparent chlorofluorocarbon (CFC) recharge age-dates from the 27 June 2000 synoptic survey of ground water seepage. Best fit lines obtain by weighted (wi, Eq. [9]) linear regression were used to approximate the historical progression of the basin-wide mean annual concentration of a hypothetical chemical in ground water recharge (Eq. [1]).

 
The best fit lines in Fig. 5 approximate the basin-wide mean concentrations (Eq. [1]) of total NO3–, NO3–, and excess N2 in ground water recharged over the four decades captured by the synoptic survey dataset. We used weighted (wi, Eq. [9] below) linear regression, as the simplest and most direct approach, to avoid injecting potential assumptions or bias into the estimation procedure, recognizing that future applications will likely have sparser datasets than Fig. 5, in which it would be unrealistic to perceive nonlinear patterns (e.g., Fig. 4), and would therefore tend to gravitate to the simple linear regression approach. It is noteworthy in this regard that the linear regression estimates obtained from the best fit lines in Fig. 5 appeared to be fairly robust in that more sophisticated fitting procedures (moving average, LOWESS smooth, exponential growth, and sigmoidal models), which allow greater subjectivity in parameter selection, ultimately produced similar Cbf estimates.

Figure 6 illustrates the discharge-weighted age-date distribution used in the calculation of Cbf values. These results represent the age-dates of ground water seepage into perennial gaining sections of the LPR relative to year 2000 based on the dissolved CFC measurements. The weighted relative frequency values for each age-date i necessary in Eq. [1] were calculated for each minipiezometer location j using:

[9]
where values of q'j were based on dhj (Eq. [2]) measurements for 27 June 2000. We assigned wi = 0 to temporary and permanent losing stream segments (Fig. 2) to avoid the confounding influence of ground water recharged by surface water. The weighted mean (±SD) apparent age-date of ground water seepage contributing to baseflow was 1976.3 ± 7 yr, indicating that baseflow was comprised on average of ground water recharged 23.7 yr before year 2000.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 6. Discharge-weighted apparent chlorofluorocarbon (CFC) age-date distribution for ground water seepage samples collected during the 27 June 2000 synoptic survey. Relative frequencies are weighted by discharge per unit length of stream (see Eq. [9] in text).

 
In Fig. 7 we illustrate the computation mechanics involved in combining the and wi data to produce a Cbf estimate for year 2000 baseflow by Eq. [1]. The actual weighted age distribution obtained from the minipiezometer network (Fig. 6 and vertical point diagram Fig. 7a) was used to sample (total NO3–) values from the total NO3– regression line (Fig. 5a) as shown in Fig. 7a. Values of (total NO3–) were then multiplied by their corresponding discharge-weighted frequencies to obtain a set of weighted concentrations (Fig. 7b). The weighted concentrations were then summed (Fig. 7c) to obtain an estimate of Cbf (total NO3–) for year 2000 baseflow. This weighted mean procedure was repeated using the excess N2 regression line in Fig. 5b to obtain Cbf (excess N2). Finally, we calculated Cbf (NO3–) by subtraction:

[10]



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 7. General procedure to estimate baseflow concentration (Cbf) from the discharge-weighted age distribution of ground water seepage. (a–c) The basin-wide mean annual concentration in ground water recharge (total NO3–) from regression line in Fig. 5a. (a) Selection of using the discharge-weighted age-date distribution (vertical point diagram) of the 27 June 2000 dataset (Fig. 6). (b) Weighting of values by the discharge-weighted relative frequency. (c) Summation of weighted concentrations to obtain Cbf.

 
Note that Cbf (NO3–) calculations based directly on (NO3–) in Fig. 5c produced essentially the same results as Eq. [10].

To generate values of Cbf (total NO3–) for Eq. [10] over the period 1960 to 2050, we extended the 27 June 2000 total NO3– regression relationship (Fig. 5a) into the past and future under three ground water recharge quality scenarios (Fig. 8a) . Under the least conservative scenario (Scenario A), we assumed that the (total NO3–) would increase indefinitely into the future; in Scenarios B and C we assumed it would stabilize at year 2000 and year 1985, respectively. Extended into the past we set all scenarios at a minimum concentration of 2.6 mg L–1, which allows for a residual amount of NO3– in the stream after adjustment (Eq. [10]) for excess N2. Each scenario was subsampled at 10-yr intervals as in Fig. 7 by shifting the recharge age distribution along the time axis in 10-yr increments. This procedure produced a set of Cbf (total NO3–) values in 10-yr intervals (e.g., curve A', upper shading, Fig. 8b).



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 8. Estimation of baseflow NO3– concentrations for 1960 to 2050. (a) Lines A, B, and C, projected basin-wide mean annual concentration in ground water recharge (total NO3–) scenarios, extended into the past (1920) and future (2050) sufficiently to generate baseflow concentration (Cbf) predictions between 1960 and 2050. Scenarios are described in the text. See Fig. 7a for explanation of vertical point diagram. (b) Lines A', B', C' are corresponding Cbf estimates. Shaded area in A' represents Cbf (excess N2). Cross-hair represents year 2000. (c) Cbf (±SD) values (solid circles) from the lower left quadrant (1960 to 2000) in Fig. 8b are superimposed on historical baseflow NO3– measurements (small open circles). Solid line: best fit historical data (slope = 0.152 mg L–1 yr–1). Dashed line: best fit Cbf backcasts (slope = 0.159 mg L–1 yr–1).

 
To generate corresponding decadal values of Cbf (NO3–) by Eq. [10], we held Cbf (excess N2) constant at the value for year 2000 baseflow (1.6 mg L–1). We took this approach because the denitrification capacity in LPR ground water is probably limited by its low dissolved organic carbon (DOC) (electron donor) concentrations and high O2 concentrations (Tiedje, 1988), conditions that are not likely to change substantially over time. The discharge-weighted mean concentrations of DOC and O2 in ground water seepage to the LPR were 91 and 252 µmol L–1, respectively, in the 27 June 2000 data. Moreover, denitrification reaction progress (Eq. [8]) was negatively correlated with the concentrations of NO3– (mg L–1) and O2 (µmol L–1) [denitrification reaction progress (%) = 54.1 (±4.8) – 1.84 (±0.26) x NO3– – 0.074 (±0.016) x O2; r2 = 0.50, p < 0.0001].

Decadal estimates of Cbf (NO3–) are depicted by curves A', B', and C' in Fig. 8b. The horizontal distance between Scenario A and the upper shaded area of A' captures the influence of the 23.7-yr lag time between ground water recharge and discharge on the quality of baseflow. The vertical thickness of shading in curve A' represents the magnitude of the denitrification correction (Eq. [10]) in curves A', B', and C' using the Cbf (excess N2) value for year 2000 baseflow.

For all scenarios (A', B', C'), Cbf (NO3–) in Fig. 8b rose approximately 2.13 mg L–1 decade–1 from the late 1960s to 8.6 mg L–1 in the year 2000. Beyond the year 2000, Scenario A' suggests that NO3– concentrations will continue to rise at the same rate, approaching 19 mg L–1 in year 2050. The unending increase implied by Scenario A' ignores a real possibility that ground water NO3– loads have begun to approach a steady state under somewhat stable land management practices. Support for this possibility is shown by the upper range of NO3– concentrations in Fig. 4, which appear to flatten or drop slightly after an early 1980s maxima. (See also the fertilizer use curves in Fig. 4, which suggests that agricultural N use rates began a leveling off through the 1980s and 1990s.) Thus, the annual NO3– recharge concentration might be more appropriately represented by the "inclined step" patterns of Scenarios B' and C'. Assuming that annual ground water NO3– concentrations in recharge have stabilized at year 2000 levels, Scenario B' suggests that baseflow NO3– concentration will rise to a plateau of 13 mg L–1 in the year 2030. Under the more optimistic Scenario C, baseflow NO3– concentration will rise to a plateau of 10 mg L–1 in the year 2010.

We checked the quality of the Cbf (NO3–) predictions by comparing them with actual baseflow data. In Fig. 8c we superimpose the Cbf (NO3–) backcasts (from lower left quadrant in Fig. 8b) on baseflow NO3– measurements reported in Albertson (1998) and Mechenich and Kraft (1997) for the Hoover Road site. The Hoover Road stream data represent grab samples collected sporadically between 1967 and 2000 during baseflow conditions (operationally defined as stream discharge less than 0.28 m3 s–1) in a long-term monitoring program. These historical data reveal a linear increase in the baseflow NO3– concentration since the late 1960s, which is closely matched by the Cbf (NO3–). These results show that historical baseflow trends can be accurately reconstructed from one synoptic survey of water quality and age-date measurements at the ground water–surface water interface.

However, it is important to recognize potential limitations of Cbf backcasts and forecasts by Eq. [1] associated with projecting basin-wide values from a synoptic survey dataset. The regression line in Fig. 5a approximates the basin-wide annual mean concentration () of total NO3– in ground water recharge and provides the backbone for extrapolation of the synoptic survey dataset into the past and future via Scenarios A, B, and C in Fig. 8. Hence, in using a synoptic survey dataset, a potential for systematic error in estimates will exist if the sample numbers are too few or the spatial distribution of samples is too biased to accurately capture ground water ages and NO3– concentrations and their volumetric contribution to baseflow. Thus, it is important to have some measure of the volumetric representativeness of ground water discharge captured by a particular sampling design. In our case, the reconstruction of stream NO3– concentrations at Hoover Road for several dates from synoptic survey seepage measurements (Table 1) leant confidence that the thalweg network of minipiezometers within the LPR affords a fairly quantitative representation of ground water ages and nitrate concentrations entering the LPR during typical baseflow conditions.

Furthermore, for any particular age interval (e.g., 10 yr) in Fig. 5a, the vertical concentration variation of Cr,i values actually reflects spatial variation in land use–land practices of a particular and finite recharge zone. Hence, the central tendency, , approximated by the regression line, technically reflects the spatial mean of that particular and finite recharge zone, rather than the entire landscape. From this standpoint, for instance, the 1960s data range in Fig. 5a contains information about the mean and variation of NO3– in the LPR contributing area most remote from the stream; while the 1990s data range reflects the contributing area nearest the stream. Thus, in projecting from Fig. 5a as a basin-wide mean we had to assume each contributing area (i.e., particular and finite recharge zone) to be a representative element of the entire LPR landscape (e.g., Fig. 9a) through time.



View larger version (64K):
[in this window]
[in a new window]
  		Fig. 9. Hypothetical age intervals (ground water lag times to stream) of recharge zones in two basins: (a) similar and (b) contrasting land use and land management practices across recharge zones.

 
Problems arise in Cbf estimates by Eq. [1] when the validity of this assumption is strongly violated; that is, when large contributing areas (age intervals) have distinctly different land use–land management attributes or histories than the rest of the landscape (Fig. 9b). However, under this case, valuable information can still potentially be extracted from the discharge-weighted age distribution, but a more data-intensive version of Eq. [1] becomes necessary. If, for example, the apparent decline in Cr,i values during 1990s (Fig. 4 and 5) actually reflected a lower landscape NO3– loading in the contributing area near the stream (e.g., forest-dominated area, Fig. 9b) compared with more remote recharge zones (e.g., agriculturally dominated areas, Fig. 9b), Eq. [1] would need to be expanded to include projected mean annual concentrations for each contributing area:

[1a]

The successful series expansion of the numerator of Eq. [1a] is predicated on generating ground water quality predictions for specific recharge areas, a task well suited to GIS-based ground water quality modeling. Further research should be done in a system like the LPR to establish (i) whether models of specific recharge areas can be allied with the discharge-weighted age distribution data from the ground water–surface water interface to generate reliable Cbf backcasts and forecasts, and (ii) under what land use patterns the non-expanded and expanded versions of Eq. [1] are best applied.

Further work should also be done to determine (i) the minimal sampling density along a meander to reliably characterize the discharge-weighted age and water quality distributions, (ii) the stability and resilience of the distribution in time and space (e.g., in response to short-term perturbations such as large rainfall events or snowmelt or long-term trends such as aquifer drawdown due to irrigation or climate change), and, finally, (iii) whether the concepts and approaches presented here have broader applications in other systems.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 CONCEPTUAL MODEL
 STUDY AREA
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
This study presents a conceptual and practical approach for assessing historical and future water quality trends in baseflow-dominated streams based on age-date and water quality measurements at the ground water–surface water interface. Using a longitudinal network of minipiezometers installed along a meandering thalweg, we obtained representative samples of the age distribution and water quality of ground water seepage into a baseflow-dominated stream from which estimates of baseflow water quality were generated. We show that this approach can be used to reconstruct historical baseflow water quality patterns in the absence of long-term monitoring data and be used to project the effects of potential management decisions on future water quality. This approach provides a tool for understanding the implications of land use and land management practices on baseflow-dominated streams.


   ACKNOWLEDGMENTS
 
The authors would like to thank David Saad, George Kraft, Kevin Masarik, Robert Gleason, John Walker, Larry Puckett, and three anonymous reviewers for their technical and editorial suggestions.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 CONCEPTUAL MODEL
 STUDY AREA
 METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2005 34: ix. [Full Text]  



This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
B. A. Browne, G. J. Kraft, J. M. Bowling, W. M. DeVita, and D. J. Mechenich
Collateral Geochemical Impacts of Agricultural Nitrogen Enrichment from 1963 to 1985: A Southern Wisconsin Ground Water Depth Profile
J. Environ. Qual., June 23, 2008; 37(4): 1456 - 1467.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Supplemental Data (63 KB)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow