USGS Home
Contact USGS
Search USGS

U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5239

Back to Table of Contents

Study Design and Methods

Ground water was sampled to characterize the geochemistry of water discharging to the Deschutes and Little Deschutes Rivers. Where ground water discharges to rivers, that shallow ground water represents the ends of flow paths converging at and discharging into rivers. Ground-water samples were analyzed for O₂, Fe, NO₃^-, NH₄⁺, Cl^-, as well as specific conductance and temperature. Dissolved oxygen was the primary redox indicator species of interest because previous work demonstrated that redox gradients in the study area are sharp and that most NO₃^- is removed by denitrification near the oxic/suboxic boundary (Hinkle and others, 2007). Thus, the presence of suboxic ground water in sediment under a river could be used to indicate the presence of a barrier that would prevent most ground-water NO₃^- from discharging to the river at that point. Note, however, that ground water at a point beneath the riverbed may not flow directly up into the river at that point. A point measurement beneath a river to determine the redox state of ground water entering the river above the measurement point is an approximation. Iron also was used as a redox indicator species. Nitrate and NH₄⁺ can be used as redox indicator species. Alternatively, other redox indicator species (here O₂ and Fe) can serve as predictor variables for the presence of conditions favorable for NO₃^- and NH₄⁺ occurrence. Chloride is useful as a tracer of septic tank effluent (Hinkle and others, 2007).

Temporal and Spatial Considerations for Sampling Network

Exchanges between ground water and surface water are complex as a result of spatial and temporal variations in hydraulic gradients, spatial heterogeneities in aquifer hydraulic properties, and convergence of multiple ground-water flow paths at rivers (rivers commonly serving as regional drains for ground-water discharge). A network design that accounts for these complexities was used in this study. The network consisted of temporary wells installed directly in the riverbeds. Transects of wells were installed at 10 locations: 3 along the Deschutes River and 7 along the Little Deschutes River (fig. 1). At each transect, 2–9 wells (40 wells total among the 10 transects) were installed to allow characterization of redox state at different points. Wells were installed with the center of the screen generally at a depth of 1.0 ft, but several were screened at 2.0 ft, and one at 3.0 ft. Shallow ground-water samples collected where ground water discharged to rivers represent the ends of flow paths converging at and discharging into rivers.

Aquifers may alternately discharge to rivers and receive recharge from rivers at different times, or at the same time but at different places. Given the need to characterize the geochemical nature of ground water discharging to rivers, water samples were collected during periods of low river flow. Previous field measurements of hydraulic gradients between ground water and the Deschutes and Little Deschutes Rivers in the study area indicated that at almost all measurement sites, vertical hydraulic gradients were upward (toward the rivers) during low flow (Morgan and others, 2007). Thus, in the present study, ground-water sampling occurred during periods of low river flow in March and October 2006 (fig. 2). This sampling structure minimized the potential for sampling ground water that originated in the river.

Within a single cross section of a river, ground water often consists of multiple flow paths of varying ages that can vary from days to millennia (Winter and others, 1998). Characterization of ground-water inputs to a river requires attention to this spatial variability. According to one conceptual model that describes the nature of ground-water discharge to a gaining river, the oldest ground water (generally the most geochemically evolved) enters in the center of the river channel. The age of ground water decreases as the point of entry into the river moves from the center to the edges of the river (Modica and others, 1998). “Geochemically evolved” is a relative term that refers to ground water that has changed to a relatively great extent due to chemical reactions with geologic materials. Geochemically evolved ground water may be more reduced, may contain higher concentrations of dissolved solids, and (or) may be older than ground water that has had less interaction with geologic materials.

Browne and Guldan (2005) suggest an alternative conceptual model in which a river meanders across a relatively flat valley floor and ground water discharges to this valley-floor system rather than solely into the river. In this conceptual model, the river intercepts relatively young ground water where the river bends toward the edge of the valley floor, and intercepts relatively old ground water where the river course lies near the center of the valley. The oldest ground water within a given transect might enter the river at the inside edge of a meander bend (edge closest to the center of the valley floor), and the youngest ground water might enter the river at the outside edge (edge closest to the margin of the valley floor) (Brown and Gulden, 2005). Therefore, in any given river transect, younger ground water may tend to enter a river by flow paths near one edge of the river (Browne and Guldan, 2005) or near both edges (Modica and others, 1998). In either conceptual model, though, ground water entering a river near the edges may be particularly young, and river edges may be particularly vulnerable to discharge of young ground water. To account for such spatial variability in ground-water age, and implicitly, geochemical character, the sampling regime used in this study incorporated multiple samples across river transects.

Ten transects were installed to represent some of the spatial variability of ground water redox state along the axes of the rivers. The geochemical character of ground water flowing into a river may vary along the length of the river in response to a spatially variable geologic and hydrologic framework and to spatially varying qualities of the riparian environment. The riparian zone has been broadly defined as the zone at which terrestrial and aquatic ecosystems overlap (Gregory and others, 1991; Martin and others, 1999). Riparian zone sediment tends to be a zone of enhanced geochemical cycling; therefore, riparian zones can be particularly effective zones of denitrification (Hill, 1996). In riparian zones, NO₃^- loss primarily is due to denitrification; plant and microbial uptake generally is less important than denitrification (Martin and others, 1999). Denitrification in riparian zones is due, in large part, to the abundance of labile (reactive) organic carbon derived from wetland plants and other riparian zone plants, as well as that available in sediment deposited by rivers. This organic carbon serves as an electron donor for redox reactions including denitrification. Therefore, riparian zone processing of ground-water NO₃^- potentially represents a useful explanatory variable for ground water redox conditions along rivers. Published literature on N cycling in riparian environments has tended to focus on individual sites rather than on larger spatial scales (Vidon and Hill, 2004), and elucidation and characterization of N dynamics at larger spatial scales (for example, river scales) currently is in its infancy (Boyer and others, 2006). For an effective river-scale assessment, identification of differences among transect sites is needed to understand the processes governing redox state along rivers.

Site Selection

The sampling framework calls for installation of multiple wells along multiple transects. Site selection within this framework was designed to provide broad coverage within the study area (fig. 1). Site selection also was designed to emphasize the identification of oxic ground water. This emphasis was introduced into the sampling design because it was assumed that most ground water discharging to the Deschutes and Little Deschutes Rivers in the study area would be suboxic. Although identifying suboxic ground water is highly useful for constraining the redox character of ground water discharging to rivers, oxic ground water presents the dominant vulnerability for NO₃^- transport into the rivers; these factors that accommodate retention of O₂ in ground water need elucidation.

The site selection focus on identifying oxic ground water lead to emphasizing coarse-grained sediments (sand, gravel) and avoiding macrophyte beds. Oxic ground water more commonly is detected in coarse-grained sediments than in fine-grained sediments (silt and clay) because of greater concentrations of organic carbon in and slower water movement through fine-grained sediments (Vidon and Hill, 2004; Pinay and others, 2007). Macrophyte beds were avoided because macrophytes can trap fine-grained sediments, and exudates from macrophyte roots could create local conditions of anoxia (Martin and others, 1999).

Transect site names begin with a “D” (Deschutes River) or “LD” (Little Deschutes River), followed by the location in river miles (as indicated on USGS topographic maps). For example, “D194.1” is a site on the Deschutes River at river mile 194.1. Individual wells are assigned a name beginning with the transect site name, followed by a decimal and a number to represent the well. For example, “D194.1.1” is one well in the transect “D194.1”.

Sample Collection and Analytical Methods

Ground-water samples were collected by installing temporary stainless steel wells in riverbed sediment. Screen lengths were 0.25 ft (36 wells) and 0.05 ft (4 wells). Wells were developed and then sampled by pumping with a peristaltic pump.

Field parameters (O₂, specific conductance, and temperature) were measured on-site in flow-through cells. The flow-through cell for O₂ measurements was created from a capped, high-density polyethylene beaker fitted with a fluorocarbon polymer inlet tube to which the pump tubing was attached and with an outlet hole to allow sample water to exit the flow-through cell. The inlet tube allowed sample water to enter the bottom of the flow-through cell, where the tip of the O₂ probe was located. The outlet hole was located at the top of the flow-through cell. The flow-through cell was large enough to accommodate the O₂ probe, but sized sufficiently small as to allow operation with low pumping rates. The flow-through cell allowed O₂ measurements on ground water that was not exposed to atmospheric gasses during measurement. A similar flow-through cell was used for specific conductance and temperature measurements.

Following well purging (more than three well bore volumes) and stabilization of field parameters, wells were sampled for nitrite-plus-nitrate (NO₂^- + NO₃^-), NH₄⁺, Fe, and Cl^- (U.S. Geological Survey, 1999). In this report, NO₂^- + NO₃^- is assumed to contain negligible NO₂^-; this assumption is based on previous experience in this basin (Hinkle and others, 2007) and in other aquifers. Thus, NO₂^- + NO₃^- is referred to as NO₃^-. Four wells also were sampled for isotopes of NO₃^- (δ¹⁵N–NO₃^- and δ¹⁸O–NO₃^‑) (isotopic content of N and O in NO₃^-).

Samples were filtered through in-line 0.45-μm nominal-pore-size one-time-use capsule filters. Capsule filters were flushed with 1 L of deionized water and then flushed with sample water to remove deionized water prior to use.

Samples for NO₃^- and NH₄⁺ were preserved by chilling to near-freezing. Fe samples were preserved with ultrapure nitric acid (pH < 2).

Samples for NO₃^-, NH₄⁺, Fe, and Cl^- were analyzed at the USGS National Water Quality Laboratory in Denver, Colorado. Methods are described in Fishman and Friedman (1989) and Fishman (1993).

Samples for δ¹⁵N–NO₃^- and δ¹⁸O–NO₃^- were analyzed at the USGS Reston Stable Isotope Laboratory (http://isotopes.usgs.gov). Analyses were done by bacterial conversion of NO₃^- to N₂O and subsequent measurement on a continuous flow isotope-ratio mass spectrometer (Sigman and others, 2001; Casciotti and others, 2002; Coplen and others, 2004), with values of δ¹⁵N and δ¹⁸O reported relative to N₂ in air (AIR) and Vienna Standard Mean Ocean Water (VSMOW), respectively. Analyses of δ¹⁵N–NO₃^- and δ¹⁸O–NO₃^- were calibrated by analyzing laboratory solutions with known isotopic compositions, and data were normalized against internationally distributed reference materials (Böhlke and Coplen, 1995, Böhlke and others, 2003) for: δ¹⁵N–NO₃^-, IAEA-N3 = +4.7 per mil (‰) and USGS32 = +180 ‰, and for δ¹⁸O–NO₃^-, IAEA-N3 = +25.6 ‰ and USGS34 = –27.9 ‰. Average reproducibility of normalized values for samples analyzed more than once was about ±0.2 ‰ for δ¹⁵N–NO₃^- and ±0.4 ‰ for δ¹⁸O–NO₃^-.

For plotting purposes, NO₃^- and NH₄⁺ concentrations less than the method reporting level (MRL) were plotted at one-half of the MRL. The MRL for NO₃^- was 0.016 mg N/L, and for NH₄⁺, 0.04 mg N/L.

Project quality-control data are presented in table A1 (appendix A). These data demonstrate contamination-free sampling and analysis, tight analytical precision, and negligible analytical bias.

Particle Tracking

Potential ground-water flow paths for advection of NO₃^- from septic tank sources to near-river environments were delineated using a numerical ground-water flow model (Morgan and others, 2007) in conjunction with the MODPATH particle tracking program (Pollock, 1994). For these simulations, four particles were placed at the water table in each water-table cell in the model and tracked to the river location where they were simulated to discharge. Aquifer-scale denitrification was represented in the flow model using the redox-boundary approach in which NO₃^- transported across the redox boundary (from oxic to suboxic conditions) was assumed to be instantaneously denitrified. Pathlines for particles that were transported into the suboxic part of the system were removed from the ensemble of oxic and suboxic pathlines. In these simulations, aquifer-scale denitrification processes were the only chemical reactions represented in the model; additional near-river reactions were not represented.

Although flow-model boundary conditions accounted for and allowed evapotranspiration and discharge to wells, particles in the particle-tracking simulations were not allowed to leave the system by evapotranspiration or discharge to wells and particles did not represent volumes of ground-water discharge. Therefore, greater densities of pathlines should not be interpreted as greater volumes of ground-water discharge to rivers.

Back to Table of Contents