Scientific Investigations Report 2007–5237
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5237
The La Pine study area of the upper Deschutes River basin is a sediment-filled graben feature that lies within the geologically complex transition area between three geologic provinces: the Cascade Range, the Basin and Range Province, and the High Lava Plains. The La Pine study area shares attributes characteristic of each of these geologic provinces. The Cascade Range forms the western margin of the La Pine study area. Newberry volcano forms the eastern study area margin and transition between the extensional Basin and Range Province and the High Lava Plains. North to north-northwest trending and northeast trending en echelon faults define the eastern boundary of the graben forming the La Pine study area.
The Cascade Range is a constructional feature of north-south trending eruptive centers that extends from northern California to southern British Columbia and has been volcanically active for the past 35 million years (Sherrod and Pickthorn, 1989). In central Oregon, these volcanic eruptive centers are stratovolcanoes, such as the North, Middle, and South Sister, and Mount Jefferson, all with elevations greater than 10,000 ft (Lite and Gannett, 2002).
The Basin and Range Province is a region of crustal extension that covers most of the western United States and is characterized by north to northwest trending sub parallel fault-bounded down-dropped grabens forming fault-block ranges with basins typically 10–20 mi wide. In central Oregon, the La Pine graben is defined by north-northwest trending and northeast trending down-dropped faults (Allen, 1966; MacLeod and Sherrod, 1992) forming the sediment-filled 10–15-mi wide subbasin with characteristics similar to those of Basin and Range extensional features along the northwestern transitional boundary in northeastern California and northwestern Nevada.
The north–south trending graben that underlies the La Pine study area is estimated to have down-dropped 1,800 to 2,400 ft based on gravity and aeromagnetic anomalies (Couch and Foote, 1985; MacLeod and Sherrod, 1992) Formation of the basin began during the mid-Pleistocene between 0.6 and 1 my (million years) ago (Couch and Foote, 1985; Gettings and Griscom, 1988; Sherrod and Pickthorn, 1989; MacLeod and Sherrod, 1992; MacLeod and others, 1995) and subsequently has been filled with several hundred feet of sediment. Nearly 1,400 ft of sediment have been penetrated by deep water-supply wells drilled in the study area. Descriptions of sediment lithology in drillers’ reports for water wells indicate that graben formation was concurrent with volcanism and late Pleistocene glacial outwash deposition. Periods of active volcanism and quiescence and fluvial and lacustrine deposition during graben development created a complex sequence of intercalated lava flows, ignimbrites, and alluvial deposits. The depositional history is further complicated by fluvial-lacustrine deposition during the Pleistocene.
Many basin-fill sediments in the La Pine study area are fine-grained lacustrine silt and clay. Within these fine-grained deposits are fine to coarse fluvial sand and gravel channel-fill deposits and discontinuous cinder, pumice- and ash-fall beds. Methane, ammonium, and reduced iron detected in water well samples (Hinkle and others, 2007a) from these fine-grained deposits indicate a predominant quiescent marsh and lake depositional environment with episodic volcanic deposition. This low-energy depositional environment may be related to the onset and development of Newberry volcano about 0.7 my ago. MacLeod and others (1995) indicated that Newberry lavas backed up against Cascade Range lava flows, blocking the channel of the Deschutes River which created a lake and marsh environment over much of the study area. A pumice bed exposed near the top of the lacustrine deposits and about 35 ft (10.8 m) below the fluvial/lacustrine contact at Pringle Falls on the Deschutes River is 0.22 to 0.17 my old (Herrero‑Bervera and others, 1994).
About 0.2 my ago the depositional environment probably underwent an abrupt change from lacustrine to predominantly fluvial with deposition of heterogeneous silt, fine to coarse sand, gravel, and pumaceous sand and gravel. These deposits are likely associated with Pleistocene glaciation of the Cascade Range and Newberry Volcano. These high-energy deposits are capped by a study area wide 3–5 ft-thick pumice- and ash-fall deposit from the Mt. Mazama eruption of 7,627 ± 150 years ago (Zdanowicz and others, 1999). The Deschutes, Little Deschutes, and Fall Rivers have reworked and down-cut through the Mt. Mazama pumice and ash-fall deposit. The high degree of heterogeneity noted in lithologic descriptions in drillers’ reports indicate an active fluvial depositional environment in the central and southern parts of the La Pine study area and deposition more characteristic of a lacustrine environment in the northern part of the study area. Based on interpretation of more than 460 water-well logs, fluvial silt, sand, and gravel deposits were determined to range in thickness from less than 10 ft in the northern study area, to as much as 100 ft in the central and southern parts of the study area (fig. 2). A thin veneer of gravel overlying a paleosol was observed in several monitoring wells in the study area. The presence of the paleosol may represent post-Pleistocene glacial soil development. The thin sand and gravel veneer overlying the paleosol may represent the brief early Holocene glacial event that predates the mid-Holocene Mt. Mazama eruption.
A detailed characterization of subsurface heterogeneity can substantially improve the reliability of models of ground-water contaminant transport (Fogg, 1986; Anderson 1987; Johnson and Dreiss, 1989). In this study, standard hydrogeologic interpretation and analysis techniques, including construction of two-dimensional geologic sections and surface maps, were used in conjunction with transition probability geostatistics to develop a three-dimensional hydrogeologic model that represents the heterogeneity of the complex glaciofluvial system.
The primary source of subsurface geologic data was the nearly 5,400 drillers’ reports available (as of 1999) for domestic and public water supply wells in the study area. Although these reports are plentiful, the quality of lithologic descriptions in the reports is inconsistent. The spatial distribution of domestic wells also is not ideal for developing a detailed, three-dimensional geologic model of the ground-water system; as would be expected, domestic wells are concentrated in areas of residential development and tend to be completed at the shallowest depth that a satisfactory yield can be obtained. A subset of 346 wells was visited as part of this and previous studies (Gannett and others, 2001) to determine accurate locations and collect water-level and other data. The subset of wells visited was selected to provide the best spatial distribution within the study area and good descriptions of geologic materials penetrated by the well. Drillers’ reports for an additional 118 wells were selected to provide information in areas where field-located wells were not available. Although these wells were not visited, the drillers’ reports included superior descriptions of geologic materials and well locations could be accurately estimated using tax lot information.
Descriptions of geologic materials were transcribed to a database from the drillers’ reports using a standardized set of lithologic descriptors developed for this study. A two-letter descriptor was assigned to the primary lithology, and, if needed, secondary lithology reported by the driller for each depth interval. For example, if the driller reported a layer of “gravel with sand” between 40 and 80 ft, the interval was assigned a primary descriptor of GR for gravel and a secondary descriptor of SA for sand. This system retained most of the detail of the original description by the driller and allowed for comparison and analysis of lithology between wells. Thirty-six descriptors were used to describe primary and secondary lithology. Combinations of the primary and secondary descriptors resulted in 157 unique lithologic descriptors for the 464 wells used in the analysis.
The 157 unique lithologic descriptors were grouped into six lithologic units. Five lithologic units comprised unconsolidated sediments: clay-silt, pumice-sand, sand, sand-gravel, and gravel. The sixth lithologic unit represented consolidated and semi-consolidated volcanic rocks such as basalt, basaltic andesite, and tuff. The lithologic unit data for each well were stored in a Geographic Information System (GIS) database that facilitated data interpretation and construction of 34 two-dimensional cross sections through the study area showing the thickness and extent of the primary lithologic units. Locations of the sections are shown in figure 2 and selected sections are shown on plate 1.
Although deep wells are somewhat sparse in the study area, the sections show that as much as 100 ft of fluvial silt, sand, and gravel overlie predominately fine-grained lacustrine sediments throughout the study area (plate 1). A relatively sharp transition exists between the overlying fluvial sediments and the basal lacustrine silt and clay, basalt, and volcaniclastic deposits. The elevation of the top of the lacustrine sediments was mapped using data from the 464 wells (fig. 2). Infrequent sand and gravel lenses as well as pumice and ash beds are present within the basal lacustrine sediments, but because of their discontinuous nature, they are not regionally significant sources of water to wells. The thickness of the fluvial sediments that constitute the primary aquifers in the study area was computed by subtracting the elevation of the top of the basal lacustrine sediments from land surface elevation (fig. 2). The greatest thicknesses of fluvial sediments are in the central, east-central, and south-central parts of the study area where higher elevations occur within the study area. Deposition of Pleistocene alluvial sediments from fans emanating from Newberry volcano contributes to the thickness of coarse sediments on the eastern margin of the study area. The total thickness (unsaturated and saturated) of fluvial sediments exceeds 60 ft over much of this area. The fluvial sediments are thinner in three areas: (1) within the floodplain of the Little Deschutes River where they have been eroded, (2) at the margins of the study area where they overlie shallow tuffs and basaltic and andesitic volcanic rocks, and (3) in the northern part of the study area where lacustrine sediments are closer to the surface (fig. 2).
Interpretation of the geologic data during construction of the two-dimensional sections revealed that there is a greater degree of heterogeneity within the fluvial part of the system than could be delineated using traditional methods of interpretation using drillers’ reports as the primary data source. The drillers’ reports were useful for defining the vertical heterogeneity at each well location; however, because of the distance between wells and the uncertainty in the lithologic descriptions, it was not possible to accurately represent the lateral heterogeneity of the fluvial hydrofacies. Another consideration was the difficulty in constructing a fully three-dimensional hydrogeologic model from the two-dimensional representations in the lithofacies sections.
Important information on the nature and proportion of lithologic units and dimensions and interconnectedness of depositional features was gained through the process of constructing the lithologic unit sections. This information and the location of the lower boundary of the fluvial aquifer system were used as supplemental information in the development of the three-dimensional hydrogeologic model described in the next section.
A transition probability geostatistical approach was used to model the heterogeneity of the fluvial sediments that constitute the shallow aquifer near La Pine. The method was applied using the suite of programs, Transition Probability Geostatistical Software (T-PROGS), documented by Carle (1999). T-PROGS implements a transition probability/Markov approach to geostatistical analysis and simulation of spatial distributions of categorical variables (such as hydrofacies). The approach uses the probabilities of lateral and vertical transitions between hydrofacies to define spatial variability. The transition probabilities, volumetric proportion of facies, and mean lengths of depositional features are the parameters used to fit a three-dimensional Markov chain model to geologic data from drillers’ logs and subjective knowledge based on an understanding of the depositional environment. Once a satisfactory Markov chain model is obtained, it is used in sequential indicator simulation with simulated annealing to produce realizations of the subsurface facies distributions that are conditioned using the observed geologic data. One advantage of the transition probability method is the ability to easily incorporate subjective or “soft” geologic information into the Markov chain model. Details on the theoretical development of the transition probability method, as wells as examples and comparisons with other methods, are given in Carle and Fogg (1996, 1997), Carle and others (1998), and Weissmann and others (1999).
The transition probability method was used to model the distribution of facies within the fluvial sediments only. The lacustrine sediments were not included because the transition probability method is predicated on the assumption that all the sediments within the modeled area were emplaced under similar depositional conditions.
A hydrofacies is defined as one or more lithologic units that have similar hydraulic characteristics. Three hydrofacies were used in the hydrogeologic model: (1) clay-silt, (2) sand, and (3) gravel. The five lithologic units used to subdivide the fluvial sediments in the sections on plate 1 were reduced to three hydrofacies by combining units with similar hydraulic characteristics. The clay-silt hydrofacies represent flood plain deposits with low hydraulic conductivity. The sand hydrofacies include varying amounts of silt and represent levee and proximal overbank deposits with moderate hydraulic conductivity. The gravel hydrofacies includes varying amounts of sand and represents channel deposits with high hydraulic conductivity.
In the rare case where there are abundant, high quality core data, geophysical logs, and detailed lithologic descriptions, the parameters for the Markov model can be estimated directly from bivariate statistics (variograms) for the hydrofacies. More commonly, as in the La Pine study area, most available geologic data are obtained from drillers’ reports. Another limitation of the well data stems from the lack of correspondence between the scale of depositional features (for example, channel deposits) that have mean lengths of 10s or 100s of feet, and the spacing of data points (wells) that may be 100s or 1,000s of feet. The disparity in the scales of the system and the data available to characterize it makes quantifying lateral transition probabilities based solely on well data difficult. However, the transition probability approach provided a means of including the geologic knowledge gained through the process of developing the two‑dimensional lithologic unit sections.
The mean lengths of the high-hydraulic conductivity gravel hydrofacies and the low hydraulic conductivity clay-silt hydrofacies were estimated by manually measuring the length of lithologic units interpreted in the 34 lithologic unit sections (examples shown on plate 1) and computing average lengths for each hydrofacies in the lateral (strike and dip) and vertical dimensions. The depositional strike was aligned with the northeast-southwest strike of the study area (fig. 2). The estimates of lateral mean lengths for the gravel hydrofacies based on the lithologic unit sections were larger than expected for channel deposits typically found in fluvial systems analogous to the Little Deschutes and Deschutes River systems in the La Pine study area. The lateral mean lengths for the gravel hydrofacies were reduced based on measurements of present channel geometry, and meander and oxbow features identified from aerial photography. The estimated mean lengths for the gravel hydrofacies were modified through a trial and error process until the Markov chain model produced a hydrofacies realization that was consistent with geomorphologic observations.
The volumetric proportions of the hydrofacies were 10, 50, and 40-percent for gravel, sand, and clay-silt, respectively. The sand hydrofacies was selected as the “background” category because it represented the highest proportion of materials in the system. The transition probabilities (table 1) need only be specified for the nonbackground hydrofacies. The probabilities are expressed as decimal percentages; for example, there is a 30-percent probability of transitioning from the gravel facies to the clay-silt facies in the strike or dip direction, but only a 15-percent probability of transitioning from clay-silt to gravel.
The grid used to create the fluvial hydrofacies model had cell dimensions of 500 ft horizontally along the depositional strike and dip and 5 ft vertically. These cell dimensions were sufficiently small to allow good representation of the gravel and clay-silt facies (within the sand matrix) with mean horizontal and vertical lengths of 1,500–8,100 ft and 8–10 ft, respectively (table 1). The overall dimensions of the modeled region are 138,000 by 50,000 ft along the depositional strike and dip, and 120 ft along the vertical dimension. The T-PROGS program was used with the Markov chain parameters to generate the three-dimensional distribution of the fluvial hydrofacies. The elevation of the base of the fluvial sediments was contoured using drillers’ log data; and cells within the hydrogeologic model below the base of the fluvial sediments were assigned to the lacustrine hydrofacies. The surficial geologic map (fig. 2) was used to assign cells to the basalt hydrofacies and land surface elevation from a DEM was used to “remove” cells from the hydrogeologic model that were above land surface. Sections through the complete three-dimensional hydrogeologic model are shown in figure 3.
Once the appropriate parameters for the Markov Chain model are determined, many equally probable realizations of the hydrostratigraphy of the fluvial hydrofacies can be simulated using the T-PROGS program. Each realization would be consistent with the data from drillers’ logs and interpretations made during construction of the two-dimensional sections used to estimate the mean lengths. The capability to generate many realizations of the hydrostratigraphy represents a potentially powerful tool for evaluating the effects of hydrogeologic uncertainty on simulation models. Using multiple realizations, uncertainty in simulated nitrate concentrations resulting from uncertainty in the hydrogeologic model could be evaluated. While time consuming, this capability could provide important information on the sensitivity of model results to hydrogeologic uncertainty.
This capability would be especially useful for contaminant transport models in highly heterogeneous systems with large contrasts in hydraulic conductivity between hydrofacies where the objective was to simulate individual contaminant plumes. This approach was not used in the current study because (1) although the fluvial sediments in the La Pine study area are somewhat heterogeneous, the hydraulic conductivities of the hydrofacies that compose them fall within a relatively small range; and (2) the purpose of the model in this study was to estimate mean concentrations over relatively large areas as a tool for evaluating watershed-scale wastewater management approaches. Variability at the scale of individual model cells caused by uncertainty in the hydrogeologic model would not impose significant limitations on the use of the model.
Hydraulic conductivity is the characteristic of a porous medium that describes its ability to transmit water, and is expressed in units of length per unit time (for example, feet per day). The hydraulic conductivity of most geologic materials is vertically anisotropic; that is, hydraulic conductivity is greater in the horizontal direction than in the vertical direction. Simulation of three dimensional ground-water flow, as was done for this study, requires estimates of both the horizontal and vertical components of hydraulic conductivity. Horizontal hydraulic conductivity estimates were made for alluvial sediments and volcanic rocks using data from well-yield tests reported in 221 drillers’ reports and slug tests done in 24 monitoring wells constructed for this study.
Transmissivity is another measure of the ability of porous materials to transmit water, and is equal to the product of hydraulic conductivity and the saturated thickness of the material. Transmissivity has units of length squared per time (for example, feet squared per day) and can be estimated using the Theis nonequilibrium equation and solving as reported in Vorhis (1979). Minimum data required from the well-yield test include pumping rate, drawdown, and time pumped. Only field-located wells with complete construction information that included well depth, casing diameter, and open or screened intervals were used in the analysis. The method of Vorhis (1979) yields a solution to the Theis equation in which transmissivity is expressed as a function of storage coefficient. Appropriate transmissivity values were selected by assuming a storage coefficient in the range normally assumed for confined aquifers (0.001) for deep wells and a storage coefficient (specific yield) in the range normally assumed for unconfined aquifers (0.10) for shallow wells. Wells were assigned unconfined storage coefficients if their open intervals did not extend more than 30 ft below land surface in the northern part of the study area or 50 ft in the southern part (south of the confluence of Paulina Creek and the Little Deschutes River) (fig. 1). Horizontal hydraulic conductivity was estimated at each well by dividing the estimated transmissivity by the length of the perforated and (or) uncased intervals of the well.
Horizontal hydraulic conductivities computed using the well-yield test data range from 0.15 to 2,500 ft/d, with a geometric mean of 60 ft/d and median of 21 ft/d. Gannet and others (2001) reported similar results from an analysis of 175 well-yield tests in the La Pine basin, in which the median transmissivity was 901 ft2/d. If a typical effective aquifer thickness of 50 ft is assumed, the transmissivity reported by Gannett and others (2001) is equivalent to a hydraulic conductivity of 18 ft/d. If well-yield data are reported accurately, the principal sources of error using this approach are head losses due to poor well construction and uncertainty in storage coefficient estimates. Sensitivity analysis showed that transmissivity, and therefore hydraulic conductivity, computed for a range of storage coefficients spanning three orders of magnitude (10-1 to 10-4) differed by an average of 50 percent. Uncertainty related to storage coefficient is not considered an important source of error. Overestimation of hydraulic conductivity can occur using this method if a well only partially penetrates a thick, homogeneous aquifer in which substantial vertical flow toward the well can occur. The effects of partial penetration are not likely to be important in data from wells in the La Pine study area because the heterogeneity of the alluvial sediments would limit vertical flow near the well.
Slug tests were done at 24 monitoring wells installed along two transects for this study. One transect was located in the northern part of the study area near South Century Drive and the other was located in the south-central part of the area near Burgess Road (fig. 1). The Burgess Road transect included tests of 17 wells at 6 sites and the South Century Drive transect included tests of 8 wells at 6 sites (table 2). The wells ranged from 8.9 to 68 ft deep and were completed with 2-ft long screens. Falling- and rising-head tests were run with initial-head displacements of 1–2 ft. Test data were analyzed using the method of Bower and Rice (1976), with the assumptions that the effects of storage can be ignored and the change in saturated thickness of the aquifer is negligible. Horizontal hydraulic conductivities computed using the slug test data range from 0.3 to 98 ft/d, with a geometric mean of 25 ft/d and median of 15 ft/d (fig. 4, table 2). The 10th and 90th percentile values of hydraulic conductivity (0.6 and 60 ft/d) span only two orders of magnitude which, even when considering that the tests selectively target coarser materials, suggest that hydraulic conductivity is relatively uniform in the area compared to some alluvial environments. The 10th–90th percentile range in hydraulic conductivity for the well-yield test data was 4 to 94 ft/d. The 10th percentile hydraulic conductivity is greater for this data set because the slug tests are conducted on water-supply wells selectively screened to the most permeable sediments and more thoroughly developed.
Median values from well-yield and slug test data were similar (21 and 15 ft/d, respectively). Geometric mean hydraulic conductivity values were greater for the well-yield test data (60 ft/d) because several basalt wells included large (> 100 ft/d) hydraulic conductivity values that skewed the mean (fig. 4). The agreement between the median values from both data sets indicates that both estimation methods provide good information on the hydraulic characteristics of the basin-fill sediments.
The shallow alluvial ground-water system in the study area is recharged primarily by infiltration of precipitation (rainfall and snowmelt), with lesser amounts resulting from lateral ground-water inflow and effluent from on-site wastewater systems. Gannett and others (2001) estimated ground-water recharge to the upper Deschutes River basin from infiltration of precipitation using a water balance model (Deep Percolation Model) developed by Bauer and Vaccaro (1987). The water balance model is based on empirical relations that quantify processes such as interception and evaporation, snow accumulation and melt, plant transpiration, and runoff. The model computes a complete daily water balance for the soil zone using measured precipitation and temperature and data describing land cover, vegetation, and soil properties. The water balance was computed within rectangular areas (cells), each with an area of 1.3 mi2. A detailed description of application of the water balance model to the upper Deschutes River basin, including model input, is available in Boyd (1996) and a summary of the results for the entire upper basin is given in Gannett and others (2001).
Computed recharge from the water balance model was used to specify the initial spatial distribution of recharge within the boundaries of the simulation model. Based on calibration of simulation models in the La Pine area (see “Nitrate Fate and Transport Simulation Models”), the recharge distribution was reduced slightly; the final calibrated distribution is shown in figure 5. The mean annual recharge distribution ranges from about 2 to 20 in/yr with the greatest recharge occurring in the upland areas; however, most of the model area lies at lower elevations adjacent to the Deschutes and Little Deschutes Rivers where mean annual recharge is about 2 in/yr (fig. 5). The mean annual recharge rate for the simulation model is 3.2 in/yr (58 ft3/s).
Effluent from on-site wastewater systems also contributes to recharge of the shallow ground-water system. The annual rate was estimated for 1999 when there were about 5,200 on-site wastewater systems in the study area. The average daily volume of effluent produced per system depends on the number of occupants and their water-use habits. Deschutes County monitored several on-site systems as part of the La Pine National On-Site Demonstration Project and determined that the average daily effluent volume was 45 gallons per day per person (B. Rich, written commun., Deschutes County, 2003). La Pine households averaged 2.55 persons in 2000 (U.S. Bureau of the Census, 2000), so each household would produce an average of 115 gal/d. In 1999, about 5,200 on-site systems would have contributed 0.6 Mgal/d, (0.9 ft3/s), of recharge to the ground-water system.
Hydraulic head, or simply head, is a measure of the force that drives ground-water movement. Ground water flows from areas of high head to areas of low head. In an unconfined aquifer, such as the shallow alluvial deposits in the La Pine study area, the elevation of the water table represents the head at the upper surface of the aquifer. The rate of change in head with distance is called the hydraulic gradient, which is the slope of the water table. Ground water in deeper aquifers may be confined by lower permeability layers (for example, silt or clay). Ground water in confined aquifers may be under pressure, such that when a well penetrates the aquifer, ground water will rise in the well casing to levels above the top of the aquifer. The hydraulic gradient in the vertical direction is positive if head increases with depth, which indicates ground water is flowing upward; conversely, it is negative if head decreases with depth, and ground water is flowing downward.
Water-level measurements were made in 192 wells in June 2000. Water levels from 170 wells with screened intervals in the upper 100 ft were used to construct a contour map of the water-table surface (fig. 5) used to calibrate the simulation model. Ground water in the shallow part of the system in the La Pine study area generally flows toward the rivers, which generally are gaining in the study area. Water-table contours also indicate that shallow ground water moves toward low-lying Long Prairie south of the La Pine core area (fig. 1).
The water table is typically within 20 ft of land surface in the study area. The shallowest depths to water in the study area are along the lower reach of the Deschutes River and the entire Little Deschutes River. In many areas, the water table is within 5 ft of land surface and seasonally inundates low lying areas adjacent to the streams during high flows in the spring.
Ground-water levels vary with time in response to changes in rates of recharge to and discharge from the ground-water system. To characterize seasonal and shorter variations, bi-monthly water-level measurements were made in 48 wells and digital recorders measured water levels every 2 hours in 9 additional wells. Most wells were monitored between March 2000 and December 2001; however, several wells in the bi-monthly network also were monitored between 1994 and 1998 as part of a previous study of the upper Deschutes basin (Gannett and others, 2001). Two long-term observation wells also are in the study area and are measured by the Oregon Water Resources Department (OWRD). These wells, one of which has been measured quarterly since 1945, provide insights on the effects of withdrawals (pumping), decadal-scale climate variation, and other influences.
Three well hydrographs are shown in figure 6 to illustrate general observations regarding the response of ground-water levels to seasonal and long-term variation in recharge. Basic information for all wells included in the monitoring network is listed in table 3. Additional information for these wells, including the complete history of water-level data, is available from the USGS National Water Information System (NWIS) at http://waterdata.usgs.gov/or/nwis.
Recharge to the ground-water system from infiltration of precipitation and snowmelt occurs primarily in winter and early spring when evaporation and transpiration losses are at a minimum and available moisture is at a maximum. When recharge to the ground-water system exceeds discharge, ground water is added to storage and water levels rise. The response of hydraulic head to the annual recharge pulse in the winter and early spring is attenuated in deeper wells, such as KLAM 136 (fig. 6); however, the annual recharge pulse causes seasonal water-level increases of as much as 5 ft in shallow wells like DESC 53655 (fig. 6). The effects of longer-term variation in climate are evident in shallow and deep wells. Shallow (DESC 53655) and deep (KLAM 136) wells responded to the wet years from 1996 to 1998, although the response in the deep well was delayed and more gradual. The long-term monitoring well DESC 7620 is relatively shallow (100 ft) and shows that several wet and dry periods have occurred in the basin since the mid-1940s (fig. 6). The dry periods have an average length of 11 years. Based on the long-term hydrograph at well DESC 7620, water levels measured at wells in June 2000 and used to construct the water-table map in figure 5 are representative of long-term (1945–2005) mean water levels in the shallow part of the ground-water system.
Ground water discharges to streams, springs, and wells, and by evapotranspiration. Discharge to streams and springs is the primary pathway for ground water to leave the shallow aquifers in the study area. Discharge to streams occurs as diffuse seepage through the streambed (below the water line), as discrete springs or spring complexes that form tributaries (for example, Fall River), and as seepage along the stream bank (above the water line). Evapotranspiration, which includes bare soil evaporation as well as transpiration by plants, is another mechanism for discharge where the water table is shallow. Pumping by wells also accounts for part of the overall discharge from the shallow aquifers in the study area.
The direction and rate of flow between ground water and streams is directly related to the direction and magnitude of the hydraulic gradient between the stream and the ground-water system. For ground water to discharge to a stream, the ground-water level must be greater than the stage of the stream. The rate of discharge to the stream is proportional to the difference between the ground-water level and the stream stage.
Ground water discharge to streams and springs was estimated from gain-loss surveys of the Deschutes and Little Deschutes Rivers between October 1995 and October 2000. Gain-loss surveys, sometimes termed “seepage studies”, are used to measure increases or decreases in streamflow between upstream and downstream measurement sites that can be attributed to exchange of water between the stream channel and the underlying ground-water system. The gain-loss surveys were conducted by measuring streamflow at intervals of 1 to 10 mi on the Deschutes River between Wickiup Dam (river mile [RM] 226.7) and Benham Falls (RM 181.4) and on the Little Deschutes River between Wagon Trail Ranch (RM 43.0) and immediately upstream of the mouth (RM 0.5). Gaging stations are located below Wickiup Dam and at Benham Falls on the Deschutes River and at RM 26.7 on the Little Deschutes River (fig. 7). Tributary inflow to the Deschutes River was measured from Fall, Spring, and Little Deschutes Rivers, and tributary inflow to the Little Deschutes River was measured from Paulina Creek and Long Creek. The gain or loss to or from each reach was computed by summing the upstream and tributary inflow to the reach and deducting the downstream outflow and diversions. Any positive residual was assumed to be discharge from ground water; negative residuals were assumed to represent stream leakage (recharge) to ground water. Streamflow measurements and estimated gains and losses for surveys in October 1995, February 1996, March 2000, and October 2000, are listed in table 4.
The gain-loss surveys of the Deschutes River show net gains of 143 and 166 ft3/s in 1996 and 2000. Most gains occur downstream of Harper’s Bridge, RM 191.7, where large springs discharge near river level from basalt aquifers. Spring discharge to this reach was about 170–180 ft3/s (15–20 percent of flow) in the two gain-loss surveys. Between Wickiup Dam (RM 226.7) and Harper’s Bridge the 1996 and 2000 surveys show a possible net loss of 10–40 ft3/s. Between Harper’s Bridge and General Patch Bridge, surveys showed net losses of 54 and 24 ft3/s in 1996 and 2000, respectively. Although the seepage results show apparent losses in this reach they are similar in magnitude to the estimated error of measurement (32–38 ft3/s). Other data (contours showing horizontal hydraulic head gradients toward the river and upward vertical gradients measured below the river bed) indicate that ground water discharges to the river in this reach and seepage data lack the accuracy to measure the gain in flow. Upstream of General Patch Bridge, the Deschutes River receives ground-water discharge of as much as 25 ft3/s, with most entering the river downstream of the La Pine State Recreation Area. The 10-mi reach upstream of Pringle Falls showed little or no ground-water interaction in 1996 and 2000.
Overall, little net gain or loss was measured in flow of the Deschutes River upstream of Harper’s Bridge, relative to the total discharge of the river. This is probably due to the low permeability of the ash and tephra deposits that constitute much of the streambed and lack of major springs discharging directly to the channel in this reach. However, exchange between the river and the ground-water system may be occurring at a smaller scale that cannot be resolved by gain-loss surveys in which measurements are several miles apart. The multitude of meander loops likely are hydraulically connected by old channel deposits that may provide ground-water flow pathways where stream water can leave the channel and re-enter downstream.
The Little Deschutes River downstream of RM 43 showed net gains ranging from 6.7 to 19.8 ft3/s in the gain-loss surveys conducted in October 1995, March 2000, and October 2000. All three surveys showed that the reach between Bridge Drive (RM 15.9) and South Century Drive (RM 5.5) received ground-water discharge; the gains in the reach ranged from 5.9 to 25.2 ft3/s (0.6 to 2.4 ft3/s/mi), although the measurement error was nearly equal to the apparent 25.2 ft3/s gain in March 2000. Gains ranging from 2.9 to 28.5 ft3/s (0.2 to 1.8 ft3/s/ mi) also were measured in the 16.3 mi reach upstream of the gaging station (at RM 26.7). The reach from South Century Drive (RM 5.5) to just upstream of the mouth (RM 0.5) had little or no gain or loss in the October surveys (1995 and 2000); however, the March 2000 survey indicated a 35 ft3/s loss in this reach.
Long Creek is an intermittent tributary to the Little Deschutes that drains the area between Highways 97 and 31 to the south of La Pine. This topographically low area probably is an abandoned channel of the Little Deschutes River. Aerial photography showing lush vegetation throughout summer indicates that this area has a shallow water table and likely is a ground-water discharge area. In a typical year, Long Creek flows until the water-table declines below the streambed in June–July and then remains dry until the water table rises in response to winter-spring recharge.
The October surveys probably were representative of average conditions (neutral or slight gains) on the lowermost reach of the Little Deschutes River, whereas the March 2000 survey probably was affected by high flows (about 300 ft3/s) that occurred during the seepage measurements. The area between the Deschutes River and Little Deschutes immediately upstream of their confluence is underlain by 10–15 ft of pumice, sand, and fine gravel that overlie a dense, low-permeability clay. The top of the clay is exposed in the banks of the Deschutes and Little Deschutes Rivers a few feet above the water surface at low flow. In late spring, ground-water seepage is visible on the bank at the contact between the sand and gravel and the clay. At high stream stage, as during the March 2000 survey, surface water would move from the channel into the exposed sand and gravel while the hydraulic gradient between the stream and ground water is temporarily reversed (stream stage is greater than the ground-water level).
The surveys on the Little Deschutes River in October 1995 and October 2000 are surprisingly dissimilar considering that the discharge and measurement error were relatively low for each survey. The apparent net gain in October 2000 was nearly three times that measured in October 1995. The difference between the results of the two surveys likely is explained by the water-table elevation at the time of each survey. In 1995, the basin had undergone several years of less than-normal precipitation and ground-water levels had fallen by nearly 20 ft over the previous decade in long-term observation well DESC 7620 (fig. 6). The lowering of the water table reduced the hydraulic gradient between the shallow ground water and the stream and, therefore, the ground-water discharge to the stream also was reduced. Between 1995 and 2000, ground-water levels recovered almost 15 ft, and based on the gain-loss surveys, ground-water discharge to the Little Deschutes River increased from 7 to nearly 20 ft3/s.
In October–November 2000, a survey was made of the vertical hydraulic head gradient between the shallow ground-water system and the Deschutes and Little Deschutes Rivers. The purpose of the survey was to identify the locations of gaining and losing reaches of the Little Deschutes and Deschutes Rivers at a more detailed scale than possible using the gain-loss surveys in which discharge measurements were made at locations separated by as much as 10 mi. Hydraulic gradient alone cannot be used to determine the volumetric rate of flow between a stream and the aquifer. Stream-channel profiles and bed lithology data (appendix A) also were collected during the survey and later used to estimate the stream bed conductance parameters required by the simulation model. The simulation model was used to estimate the rate of ground-water flow into or out of the stream; the simulated rates were compared with estimates from the seepage surveys during model calibration. Head gradient measurements were made at 12 sites on the Deschutes River between RM 192.5 and RM 218.4 and 20 sites on the Little Deschutes River between RM 2.0 and RM 42.6 (appendix A). The average distance between head gradient measurements was 2 river miles. Measurements were made using a portable hydraulic potentiometer that was driven 1–3 ft into the streambed. The relative head difference between the stream and ground water was measured to the nearest 0.01 ft using a manometer. The design of the potentiometer and details on the procedure for measuring head gradients is given in Winter and others (1988). The dimensionless vertical hydraulic head gradient was computed at each location by dividing the head difference (ground-water level minus stream stage) by the depth of the center of the potentiometer screen below the stream bed.
Ground-water discharge was indicated at 18 of 20 measurement sites on the Little Deschutes River and at all 12 sites on the Deschutes River (appendix A). Vertical hydraulic gradients ranged from -0.003 to 0.075 on the Little Deschutes River (negative gradients indicate downward flow from the river) and 0.010 to 0.060 on the Deschutes River. All gradients were upward (indicating ground-water discharge to the river) except for two sites on the Little Deschutes River (RM 28.1 and 29.5; fig. A1, table A1) where neutral or slight downward gradients were measured.
Ground-water discharge to the atmosphere can occur by evaporation from bare soil and transpiration from the leaves of phreatophytes (plants whose roots draw water from the saturated zone). The rate of evaporation from bare soil diminishes rapidly with increasing depth to the water table and is negligible if the water table is more than 10 ft below land surface (Brutsaert, 1982, p. 236). Transpiration rates are dependent on the type and density of phreatophytes, climatic conditions, quality of water, and depth to water (Robinson, 1958, p. 16). These processes usually are considered together and referred to as evapotranspiration, or ET.
In June 2000, the water table was within 10 ft of land surface over an area of about 22 mi2. The potential ET (PET) rate is the maximum possible rate for the local climate conditions if water is nonlimiting. As part of a regional ground-water budget, Gannett and others (2001) estimated that the PET for the saturated zone in the La Pine study area was 22 in/yr, which is equivalent to about 1.6 ft3/s/mi2. This rate would occur only if the water table was at or near land surface with full coverage of phreatophytes. Under these conditions, the maximum rate of ground-water discharge from the study area by ET would be 35 ft3/s. Because the water table is not at land surface and the coverage of phreatophytes is considerably less than 100 percent, a more reasonable estimate of total ET from the study area is in the range of 10 to 20 ft3/s.
Eleven public-water supply systems (Oregon Department of Human Services, 2005) served about 1,650 people in the study area in 2000. The largest water purveyor, the La Pine Water District, serves 800 residents in the central core area of La Pine. A private water system serves about 350 residents of neighborhoods north of South Century Drive, between the Deschutes and Little Deschutes Rivers. Several other small water systems serve about 500 residents of mobile home and recreational vehicle parks scattered throughout the study area. About 12,350 residents, about 90 percent of the population, obtained their water from individual, privately-owned wells in 2000. Local water use data for self-supplied rural residents in the study area was not available. Mean per capita withdrawals for self-supplied rural domestic use in Oregon is estimated to be 110 gal/d (Hutton and others, 2004); however, given the limited amount of irrigated landscaping in the La Pine area, this value probably is not applicable. Deschutes County monitored wastewater effluent discharge at homes in the area as part of the La Pine Demonstration Project and determined the average per capita rate was 45 gal/d (Oregon Department of Environmental Quality, 2004b). To estimate domestic well withdrawals, it was assumed that outside water use was equal to 25 percent of total use, making the total per capita withdrawal rate equal to 60 gal/d. Assuming this rate for each of the 14,000 residents, about 0.84 Mgal/d (1.3 ft3/s) were withdrawn in 2000. Infiltration of effluent from on-site wastewater systems returns is about 45 gal/d per capita (0.63 Mgal/d total) to the ground-water system as recharge at the water table, resulting in net withdrawals of 0.21 Mgal/d (0.32 ft3/s).
As part of this study, Hinkle and others (2007a) applied geochemical and isotopic tools at various scales to provide a framework for understanding aquifer-scale nitrate source, transport, and fate. (Other wastewater contaminants, such as viruses and pharmaceuticals, were evaluated by Hinkle and others [2005].) The conceptual model resulting from this work was used to develop the transport simulation model.
Nitrogen concentration data, tracer-based apparent ground-water ages, ratios of N/Cl-, N isotope data, and ground-water flow directions indicate that on-site wastewater effluent is the only significant anthropogenic nitrogen source to shallow ground water in the area. High concentrations of ammonium were measured in samples from deep ground water. Nitrogen isotopes, N/Cl- and N/C ratios, tritium-age data, and hydraulic-head gradients support a natural, sedimentary organic matter source for the high concentrations of ammonium in the deep aquifer, as opposed to an origin from on-site wastewater derived nitrogen.
Most residential development in the La Pine area has occurred since 1960, with accelerated growth during 1990–2005. As a result, loading of nitrate from on-site wastewater systems is a relatively recent phenomenon that, in combination with low ground-water recharge rates, has resulted in high concentrations of nitrate near the water table. Low recharge rates and flow velocities have, for now, generally restricted nitrate occurrence to discrete plumes within 20–30 ft of the water table. Concentrations of nitrate typically are low in deeper, older ground water due to the nature and timing of nitrate loading and transport, and to loss by denitrification. Denitrification is the process of reducing nitrate into gaseous nitrogen. The process is performed by bacteria when dissolved oxygen (which is a more favorable electron acceptor) is depleted, and bacteria turn to nitrate in order to oxidize organic carbon or other electron donors.
Denitrification was identified as an important geochemical process in the study area by Hinkle and others (2007a), who reported various supporting evidence, including nitrate/chloride relations, presence of excess N2 enriched in 15N, presence of N2O, and nitrate losses within a framework of progressively reduced ground water, reflecting classical geochemical evolution. Ground water in the study area evolves from oxic to increasingly reduced conditions with increasing depth below the water table. Suboxic conditions are achieved in 15–30 years, and the boundary between oxic and suboxic ground water is sharp. Nitrate is denitrified near the oxic-suboxic boundary.
The denitrification process was simulated by specifying the location of the oxic-suboxic boundary within the ground-water system and applying the assumption that nitrate is rapidly converted to nitrogen gas as it is transported across the boundary. The location of the boundary was mapped within the study-area using data from 256 wells where dissolved oxygen concentrations were available (Hinkle and others, 2007a). Oxic ground water was defined as water with dissolved oxygen concentrations greater than to 0.5 mg/L. The boundary between oxic and suboxic water was constrained at each sampling site using information on the depth to the water table and well construction. The elevation of the oxic/ suboxic boundary cannot be exactly determined at wells; however, the elevation can be constrained if depth to the water table and the top of the screened interval of the well are known. The presence of suboxic water indicates that the boundary lies above the top of the screened interval and the presence of oxic water indicates that the boundary lies below the top of the screened interval. In the latter case, the well may yield a mixture of oxic and suboxic water if the boundary lies within the screened interval, or the water may be entirely oxic if the boundary lies below the screened interval. In areas where data were sparse, an understanding of general patterns of occurrence of oxic and suboxic water was used to infer the location of the boundary. The general absence of oxic water below a depth of 50 ft below the water table helped constrain the depth over large parts of the study area. Similarly, the general absence of suboxic water in ground water within 10 ft of the water table constrained the minimum thickness of the oxic part of the aquifer. However, one limitation of the data was the absence of wells in near-stream areas to constrain the boundary location. The thickness of the oxic part of the aquifer was contoured (fig. 8) and used to specify the location of the oxic-suboxic boundary in the transport model. The thickness of the oxic part of the aquifer decreases as ground water moves toward discharge areas along the Deschutes River, Little Deschutes River, and Long and Paulina Creeks. This is consistent with the aging of ground water as it moves along flow paths terminating at surface-water features in the area. Details of the implementation of the boundary are presented in the description of the simulation model.
The annual rate of nitrogen loading to the shallow aquifer system from on-site wastewater systems was estimated for 1960–2005. The simulation model was run for 1960–99 using the estimated annual loading rates as input and simulated nitrate concentrations were compared with measured concentrations. Previous investigations (Century West Engineering, 1982; Oregon Department of Environmental Quality, 1994) have concluded that the only significant anthropogenic source of nitrogen to the ground-water system in the La Pine study area is effluent from on-site wastewater systems. As previously described, Hinkle and others (2007a) also cite several lines of geochemical evidence supporting the conclusion that on-site wastewater systems are the dominant anthropogenic nitrogen source. This conclusion also is consistent with the low level of agricultural activity, limited number of livestock, and infrequent use of turf (and therefore fertilizer) in residential landscaping in the study area (Century West Engineering, 1982; Oregon Department of Environmental Quality, 1994; Hinkle and others, 2007a).
Annual nitrogen loading to ground water from on-site wastewater-system effluent was estimated for each tax lot in the study area annually from 1960 to 1999. The population of the study area prior to 1960 was sufficiently small that nitrogen loading was considered negligible. Nitrogen loading estimates were compiled by tax lot, but later aggregated across larger areas for input to the simulation model. Sources considered for nitrogen loading estimates included residential, schools, motels, restaurants, and recreational vehicle and trailer parks. Tax lot data (location, property class, improvement class, year built) were provided by Deschutes County (Tim Berg, Deschutes County Department of Community Development, written commun., 2001) and Klamath County (Jim McClellan, Klamath County Department of Management Information Services, written commun., 2001). Nitrogen loading was assumed to begin when lots were developed and end when lots were connected to a centralized sewer system (see areas served by centralized sewers in figure 1). Dates of sewer installation were provided by ODEQ (Dick Nichols, ODEQ, written commun., 2001) and La Pine Special Sewer District (Andy Newton, La Pine Special Sewer District, oral commun., 2001). Nitrogen loads were calculated for each of the approximately 5,200 lots that were developed and used on-site wastewater systems in 1999.
Nitrogen loading for homes was assumed to equal the product of the average number of persons per household (2.55 persons/household in the study area [U.S. Bureau of the Census, 1960, 1970, 1980, 1990, 2000]); average volume of effluent discharged to the on-site system by each person (45 gal/person/d or 170 L/person/d [U.S. Environmental Protection Agency, 1980; Oregon Department of Environmental Quality, 2004b]), and average nitrogen concentration in on-site wastewater system effluent (61 mg N/L in the study area [B. Rich, ODEQ, written commun., 2001]). Nitrogen discharged from on-site wastewater systems also was assumed to be completely oxidized to nitrate in the unsaturated zone beneath the drain field. Using these assumptions, the estimated potential average nitrogen loading per home is 21 lb N/yr (9.7 kg N/yr).
The potential for nitrogen loss in the unsaturated zone was evaluated by sampling packed-bed (sand) filter on-site systems in the upper Deschutes Basin. Inflow to and outflow from the sand filters was sampled at 15 sites in two networks. The first network consisted of 10 sand filters, installed 0.9 to 9.1 years prior to sampling. The second network consisted of 5 new sand filters. The first network was sampled once (October 2001) and the second network was sampled approximately bi-monthly for 3 years after installation (October 2000 – November 2003). Samples were analyzed by ODEQ as part of the La Pine NDP for nitrate, ammonium, total Kjeldahl nitrogen, and chloride. Nitrogen concentrations were adjusted for dilution and evaporation using chloride concentrations. The adjusted nitrogen concentrations from the first sand-filter network indicated median and mean nitrogen losses of 31 percent and 29 percent, respectively (ODEQ, unpub. data, 2004). Data from the second network showed that the sand filters underwent a period of maturation after installation. Using data from samples collected more than one year after installation, adjusted nitrogen concentrations from the second sand-filter network indicated median and mean nitrogen losses of 11 percent (Oregon Department of Environmental Quality, 2004b). The artificial unsaturated zone created by a sand-filter on-site system would be expected to function much like the natural unsaturated zone beneath standard on-site systems in the study area and similar rates of nitrogen loss would be expected beneath standard and sand-filter on-site wastewater systems. Based on the range of nitrogen loss indicated in the sand-filter sampling study (approximately 10 to 30 percent), nitrogen losses in the unsaturated zone were assumed to average 25 percent for conventional on-site systems (including standard, pressure-distribution, and sand-filter systems) for computing loading rates. Nitrogen loss of 25 percent in the unsaturated zone would reduce the assumed nitrogen concentration of effluent as it enters the saturated part of the ground-water system (water table) from 61 mg N/L to 46 mg N/L. For the same effluent volume, the resulting annual nitrogen loading to the ground-water system would be reduced from 21 lb N/yr (9.7 kg N/yr) to 16 lb N/yr (7.3 kg N/yr).
Nitrogen loading estimates were further adjusted to account for seasonal residency. For many years a significant number of homes in the study area were second homes or vacation rentals occupied only part of the year. In 1980, about 46 percent of the population in the study area was seasonal (Century West Engineering, 1982). Seasonal residents made up an estimated 20 percent of study area population in 2000 (Keven Bryan, La Pine Postmaster, oral commun., 2000; Carla Crume, La Pine Chamber of Commerce, oral commun., 2000). To account for seasonal residents in the estimates of nitrate loading it was assumed that (1) a seasonal resident did not occupy their home for 6 months of the year, (2) seasonal residents were 46 percent of the population from 1960 to 1980, and (3) the percentage of seasonal residents decreased from 46 to 20 percent between 1980 and 1999. Adjusted for seasonal residency, residential nitrate loading estimates ranged from 12 to 14 lb/yr (5.6 to 6.5 kg/yr) per home between 1960 and 1999.
About 9,300 residential lots in the study area depend (or will depend) on on-site systems for wastewater disposal. Nearly 8,200 lots are in Deschutes County and the remaining 1,100 lots are in Klamath County. About 1,800 additional undeveloped lots in Deschutes County may not be suitable for on-site wastewater systems due to extremely shallow water-table conditions (Tim Berg, Deschutes County Community Development Department, written commun., November 2005). Another 485 lots in Deschutes County will be served by a central sewer system scheduled to be completed in 2008 and are not included in these totals (B. Rich, Deschutes County Community Development Department, written commun., November 2005). As of 1999, homes with on-site systems had been built on 4,800 and 390 lots in Deschutes County and Klamath County, respectively. Assuming an average nitrogen loading rate of 14 lb/yr (6.5 kg/yr) per home, total residential loading in 1999 was 74,000 lb/yr (33,750 kg/yr). According to Deschutes County records, about 1,000 new homes were built between 2000 and 2005; this represents about 26 percent of the 2000 inventory of 3,900 lots in the Deschutes County part of the study area. Building data were not available for the Klamath County part of the study area for 2000–2005, however if the same growth rate is assumed, an additional 180 new homes would have been built during this period. The estimated number of homes in the study area in 2005 was 6,370 with nitrogen loading of 91,000 lb/yr (41,400 kg/yr). The remaining inventory of lots suitable for on-site systems in 2005 was 2,930, which could potentially add 42,000 lb/yr (19,000 kg/ yr) of nitrogen loading from on-site systems. These residential loading estimates are based on the assumption that 20 percent of residents were seasonal in 1999 and 2005. If all residents were full-time, the average nitrogen loading per home would be 16 lb/yr (7.3 kg/yr) and the total loading from on-site systems after all lots (9,300) were developed would be about 150,000 lb/yr (67,900 kg/yr), or about 164 percent of the 2005 estimate.
As with homes, nitrogen loading for schools, motels, restaurants, and recreational vehicle and trailer parks was assumed to equal the product of the number of people using the facilities, average volume of effluent generated per person, and the average nitrogen concentration in on-site system effluent (also assumed to be 46 mg N/L for these facilities). In most instances, estimates of the number of people served by on-site systems in study area schools, motels, restaurants (data in terms of restaurant seats), and recreational vehicle and trailer parks were obtained from employees at the facilities and from the Deschutes County Department of Environmental Health. Data could not be obtained directly from some motels and recreational vehicle and trailer park sites; in these instances, data obtained from similar sites were applied. Average volumes of effluent generated in schools were reported to be 3.2 gal/d (12 L/d) per elementary school student and 5.3 gal/d (20 L/d) per middle school student (John Rexford, Operations Manager, Bend-La Pine Public School District, written commun., 2001); volumes for high school students were assumed to be the same as for middle school students. A volume of 14 gal/d (53 L/d) per restaurant seat (obtained by adjusting the value reported by Frimpter and others [1990], to estimated study area restaurant use) was used in estimates for restaurants. A volume of 34 gal/d (130 L/d) per person was used for motels (Tchobanoglous and Burton, 1991), and 45 gal/d (170 L/d) per person (same as for homes) for recreational vehicle and trailer parks. In 1999, nitrogen loading from nonresidential sources totaled about 6,600 lb/yr (3,000 kg/yr).
Annual and cumulative nitrogen loading to the ground-water system from 1960 through 2005 are shown in figure 9. Annual rates for 1960–99 were estimated using the method described above. The 2005 loading rate was based on updated information on the number of homes built as of 2005; the 2000–2004 loading rates were estimated by assuming that the number of homes (and loading) increased linearly during the period and nonresidential loading remained constant at the 1999 rate of 6,600 lb/yr (3,000 kg/yr). Loading increased at a moderate rate as the population grew from 1960 to the mid-1970s. In the late 1970s, a surge in building (and nitrogen loading) occurred prior to enactment of more restrictive rules for on-site systems. Construction of a sewer system for the central business district of La Pine in 1989 caused only a minor reduction in overall nitrogen loading that was followed by rapid population growth and concurrent loading into the 2000s. In 2005, the cumulative mass of nitrogen added to the ground-water system since 1960 was more than 2 million pounds (900,000 kg). As the number of homes contributing nitrogen to ground water increases, the cumulative mass of nitrogen added to ground water rapidly increases. One-half of the total mass of nitrogen added during the 45-year period (1960–2005) was added during the last 12 years (1993–2005) (fig. 9).
U.S. Department of the Interior | U.S. Geological Survey
URL: https://pubs.usgs.gov/sir/2007/5237
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:54:18 EST
