Scientific Investigations Report 2007–5041
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5041
A number of water budgets for all or part of the SVRP aquifer have been developed in previous studies: these are summarized in Kahle and others (2005). A ground-water budget using primarily new information compiled as part of this study is presented in table 2 and figure 11. Most components of this ground-water budget represent average conditions, 1990–2005, corresponding to the simulation period of the ground-water flow model of Hsieh and others (2007). Exceptions are estimates of ground-water and surface-water interaction, which are based on actual measurements made during one week; inflow from Coeur d’Alene Lake and Lake Pend Oreille and outflow to Long Lake are based on previously published estimates.
The estimated mean annual inflow and outflow to and from the aquifer is 1,471 and 1,468 ft3/s, respectively. The 3 ft3/s imbalance between estimated inflows and outflows of this ground-water budget (table 2, fig. 11) represents less than 1 percent of the total, and may be due to measurement error, uncertainty in the estimation of water-budget components, or the release of ground water from storage in the aquifer.
This brief summary of individual water-budget components draws upon a number of studies that are referenced in the text. The reader is encouraged to examine these sources directly for a complete understanding of the strengths and limitations of the respective components.
Recharge or inflow to the SVRP aquifer occurs from six main sources: the Spokane River, lakes, precipitation over the aquifer, tributaries, infiltration from landscape irrigation and septic systems, and subsurface inflow. Total estimated mean annual inflow to the aquifer is 1,471 ft3/s.
Discharge measurements to determine seepage gains and losses on the Spokane River and some tributaries were made three times during this study: September 13–16, 2004; August 27–September 1, 2005; and April 24–28, 2006. Because a continuous streamflow record is lacking for most of these sites that would allow the calculation of a long-term annual mean, the current report uses the values measured during August 27–September 1, 2005 because they are the most comprehensive measurements and were made under favorable weather and flow conditions. During this period, discharge was measured at 31 sites in and near the study area: 11 on the Spokane River, 9 on the Little Spokane River, and 11 on various tributaries. Measured streamflow at these sites was compared to upstream and downstream measurements to define distinct gaining or losing reaches within the study area. Consequently, five contiguous gaining or losing reaches were defined: four on the Spokane River and one on the Little Spokane River below the “At Dartford” gaging station. Recharge to the aquifer from streamflow loss in the two losing reaches is shown in table 2. The cumulative recharge from both reaches is 718 ft3/s (table 2, fig. 11) representing 49 percent of the total mean annual inflow of 1,471 ft3/s, making this the largest source of recharge to the aquifer.
Nine lakes around the margin of the SVRP aquifer contribute recharge to ground water (pl. 1). These lakes are either perched or hydraulically connected to the aquifer. The magnitude of this recharge is important yet difficult to quantify—because inflow to the aquifer from lakes cannot be measured directly, several approaches have been used to estimate the volume of this inflow. The first approach requires the development of a water balance for the lake itself in which the residual is assumed to be recharge to the aquifer. (The primary component of such an analysis typically is surface-water inflow to the lake from its contributing basin, thus it is often referred to as the basin-yield method.) This approach works best for smaller lakes in which surface-water inflow and evaporation can be determined with more certainty as opposed to large lakes in which it is difficult to measure or estimate the increased number of inflows and outflows. For the SVRP aquifer study L. Murray (University of Idaho, written commun., March 3, 2006) used this approach for all lakes other than Coeur d’Alene and Pend Oreille. Another method of estimating inflow from lakes into the aquifer is to apply Darcy’s law; however this requires water-table gradient, cross-sectional area of the lake/aquifer interface, and hydraulic conductivity. The last two variables are poorly constrained for the aquifer in the area of Coeur d’Alene Lake and Lake Pend Oreille resulting in significant uncertainty in such an analysis. Finally, flow from the lakes into the aquifer can be estimated as a residual of a ground-water budget in which more certain estimates of other water-budget components are used to constrain and estimate less-certain components. All or some of these methods may be combined: a ground-water flow model can combine Darcian analysis with ground-water budget residual analysis.
L. Murray (University of Idaho, written commun., March 3, 2006) investigated a variety of techniques including water-level analysis and gradients, basin yield using StreamStats (U.S. Geological Survey, 2006), and water quality. For all but the two largest lakes, Coeur d’Alene and Pend Oreille, these basin-yield values are used in this report and are reported in table 2 and figure 11.
Flow from Coeur d’Alene Lake and Lake Pend Oreille remains very uncertain. Previously published estimates of annual ground-water inflow to the SVRP aquifer from Coeur d’Alene Lake range from 35 to 300 ft3/s; for Lake Pend Oreille they range from 20 ft3/s to greater than 1,000 ft3/s, though most estimates range from 20 to 61 ft3/s (Kahle and others, 2005). As described in the section,“Ground-Water Budget Errors and Uncertainty”, cumulatively the two lakes probably contribute less than 200 ft3/s of inflow to the aquifer annually, which is in the lower range of previously published estimates. For this report, mean annual inflow from Coeur d’Alene Lake is taken as 37 ft3/s: this is the value from Sagstad’s (1977) Darcian analysis of a cross section across the approximate area which Coeur d’Alene Lake (and not the Spokane River) is in contact with the aquifer, which is similar to the 35 ft3/s used in Buchanan’s (2000) ground-water flow model. For Lake Pend Oreille, an estimate of 50 ft3/s is used—this value was estimated in several previous studies using different methods (Kahle and others, 2005). Thus, the total contribution from lakes into the SVRP aquifer is shown in table 2 and figure 11 as 287 ft3/s, of which Coeur d’Alene Lake and Lake Pend Oreille comprise 87 ft3/s. This 287 ft3/s represents 20 percent of the total mean annual inflow of 1,471 ft3/s.
It seems counterintuitive that two of the largest lakes in the western United States contribute so little water to the SVRP aquifer (in addition, Lake Pend Oreille is one of the deepest lakes in the world). Frink (1964) noted that the sediments forming the natural dam of Lake Pend Oreille are glacial till and moraine and are thus of lower permeability—he confirmed this by citing lower specific capacities of wells on Farragut Naval Base as compared to the rest of the valley. Sagstad (1977) analyzed well logs and specific capacity data for four wells in the Coeur d’Alene area and concluded that the aquifer in this area generally was less permeable than in the Post Falls area. Therefore, there seems sufficient basis to support the premise that these two lakes contribute relatively little ground-water inflow to the SVRP aquifer.
Areal recharge of the SVRP aquifer derives from two component sources—permeable and impermeable surfaces. The former is direct infiltration of precipitation through the soil zone, while the latter is precipitation runoff from impervious cover in urban areas that recharges to the aquifer. For the purposes of this ground-water budget, the two are combined into a total term for areal recharge.
Bartolino (2007) used data from six active weather stations in and near the study area to calculate direct areal recharge from precipitation using four different techniques. Bartolino (2007) concluded that the dual-coefficient FAO Penman-Monteith dual-crop evapotranspiration and deep percolation calculations (Allen and others, 1998) with 1990–2005 daily values best represented changes in soil moisture and thus temporal changes in recharge. Using GIS techniques, a triangular network was constructed of the six weather stations using summed daily recharge values (P.A. Hsieh, U.S. Geological Survey, written commun., December 6, 2006). Simple linear interpolation was used to establish recharge values within the triangle thus establishing recharge for the entire SVRP aquifer.
Using aerial photography and GIS coverages of precipitation and dry well locations, B.A. Contor (Idaho State University, written commun., April 20, 2006, and July 19, 2006) mapped changes in impervious cover between 1990 and 2005 and estimated recharge from impermeable surfaces to the SVRP aquifer. Precipitation on impervious cover that drained directly to the Spokane River or other water body was not included. The estimated mean annual areal recharge (inflow) for 1990–2005 to the aquifer from permeable and impermeable surfaces is 233 ft3/s representing 16 percent of the total inflow of 1,471 ft3/s.
Runoff from highlands adjacent to the SVRP aquifer contributes recharge to the aquifer. Because the sediments of the valley floor are highly permeable, few distinct surface-drainage channels have developed other than the Spokane and Little Spokane Rivers, and it may be assumed that all streamflow from highlands and tributary basins infiltrates to the aquifer once drainage debouches onto the valley floor. As described in appendix A (at back of report), a GIS-based technique was used to estimate basin yield, which was then assumed to equal recharge to the aquifer.
Hortness (appendix A) calculated basin yield for 72 basins. Estimated total mean annual recharge to the aquifer from these 72 basins is 112 ft3/s or 8 percent of the total inflow of 1,471 ft3/s. Details of the calculation techniques, a table of the estimated recharge from each basin, and a map of the basins are in appendix A.
Infiltration of water applied at the land surface for irrigation and deep percolation of water from septic systems contribute about 54 and 23 ft3/s of recharge to the SVRP aquifer, respectively. These values were suggested by B.A. Contor (Idaho State University, written commun., March 28, 2006) who estimated that 60 percent of landscape-irrigation water and 5 percent of in-home domestic water use was consumptive, based on values found in the literature. The cumulative mean annual recharge from both sources is 77 ft3/s (table 2, fig. 11) representing 5 percent of the total mean annual inflow of 1,471 ft3/s.
Subsurface inflow into the SVRP aquifer through the Hoodoo and Blanchard Valleys is probably a minor component of recharge, however, because such underflow cannot be measured directly there is a high degree of uncertainty associated with any estimate. Previous work by Walker (1964) and Parliman and others (1980) note the presence of ground-water divides in both valleys that would limit the amount of subsurface inflow. Previous estimates for inflow from the two valleys are discussed in Kahle and others (2005) but because these estimates were often aggregates of recharge from various northern valleys of the Rathdrum Prairie, it is difficult to establish previous estimates for a specific valley. For the Hoodoo Valley, previous estimates range from 0 ft3/s (Buchanan, 2000) to 90 ft3/s (Drost and Seitz, 1978); for the Spirit Valley, previous estimates range from 3 ft3/s (Buchanan, 2000) to 89 ft3/s (Thomas, 1963). Using methodology discussed previously in section, “Lake Recharge,” L. Murray (University of Idaho, written commun., March 3, 2006) estimated that Blanchard Lake, near the south end of Spirit Valley, recharged 44 ft3/s to ground water.
Ground-water levels were measured in only a few wells in this general area in September 2004 (Campbell, 2005). Four wells were measured in Hoodoo Valley: 260 and 261 in the northern end and 262 and 263 to the southwest (these well numbers are those used in Campbell [2005]). Whereas measured water-table altitudes in wells 260, 261, and 263 differ by less than 1.5 ft, the measured water-table altitude in well 262 is approximately 10 ft higher than in these three wells. Further uncertainty is introduced because the land-surface altitude accuracy of well 263 is ±5 ft, as opposed to ±0.1 ft for the other wells, and well 260 was pumped prior to the water-level measurement. However, if these water-table altitudes are accepted, and if well 262 is excluded because it is located on the margin of the Hoodoo Valley and thus probably affected by conditions in the Spirit Valley, the water-table gradient in the Hoodoo Valley is nearly flat. This, in conjunction with the ground-water divide noted in previous work, suggests that ground-water inflow to the SVRP aquifer from Hoodoo Valley is insignificant. Three wells were measured in September 2004 in the Spirit Valley: well 267 on the western end and wells 262 and 264 to the east. The altitude of the water-table declines about 130 ft in the nearly 3 mi between wells 267 and 264. Such a steep gradient could be the result of low-permeability sediments within the aquifer or may indicate a decreased saturated thickness through this area. The maps of wells with fine-grained layers within the Rathdrum Prairie (fig. 8) and of the approximate thickness of the aquifer (fig. 10) in this report could loosely support either interpretation.
Based on ground-water levels in the Hoodoo Valley, the ground-water budget in this report assumes that no subsurface inflow enters the SVRP aquifer through the Hoodoo Valley. L. Murray’s (University of Idaho, written commun., March 3, 2006) estimated value of 44 ft3/s for mean annual recharge from Blanchard Lake is used as the value for subsurface inflow entering the SVRP aquifer through the Spirit Valley. Because Parliman and others’ (1980) small-scale map shows a ground-water divide near Blanchard Lake, this value could represent a maximum and the actual value could be considerably less. This estimate of 44 ft3/s represents 3 percent of the total mean annual inflow to the aquifer.
Discharge or outflow from the SVRP aquifer occurs from five main sources: the Spokane River, the Little Spokane River, pumpage, subsurface discharge to Long Lake, and infiltration of ground water to sewers. Total estimated mean annual outflow from the aquifer is 1,468 ft3/s.
Discharge measurements to determine seepage gains and losses on the Spokane River and some tributaries are described in section, “Spokane River.” Discharge from the aquifer was measured as streamflow gain in two reaches of the Spokane River and is shown in table 2. The measured flow of 1.5 ft3/s from Hangman Creek and 56 ft3/s of discharge from the Spokane Waste Water Treatment Plant were subtracted from the measured streamflow gain in the reach, Spokane River at Spokane to below Nine Mile Dam, to compute actual streamflow gain through this reach of the Spokane River. The cumulative discharge measured as streamflow gain from both reaches is 861 ft3/s (table 2, fig. 11) representing 59 percent of the total mean annual outflow of 1,468 ft3/s—the largest source of discharge from the SVRP aquifer.
Estimated outflow from the aquifer into the Little Spokane River includes measured flow in tributaries downstream of this gaging station because they were primarily ground-water fed during the measurement period (table 2). The measured streamflow gain or aquifer discharge is 232 ft3/s (table 2, fig. 11), which represents 16 percent of the total mean annual outflow.
In developed areas, water-use and pumpage data are necessary in order to assemble a water budget or ground-water flow model. M.A. Maupin (U.S. Geological Survey, written commun., January 4, 2006) and B.A. Contor (Idaho State University, written commun., September 11, 2006) compiled data from public-supply, domestic, irrigation, and industrial wells and wastewater-treatment plants to estimate the amount of water pumped from the SVRP aquifer. Table 2 shows mean ground-water pumpage for 1990-2005 in four categories: public-supply, self-supplied industrial, irrigation (outside purveyor-service areas), and domestic (outside purveyor-service areas). The estimated total discharge from the aquifer for these five categories is 318 ft3/s or 22 percent of the total mean annual outflow of 1,468 ft3/s. Taken individually, percentages of the total outflow for each pumpage category are: 14 percent (public-supply), 2 percent (self-supplied industrial), 3 percent (irrigation), and 2 percent (domestic).
Previously published estimates of underflow from the aquifer range from 0 (CH2M Hill, 1998; Golder Associates, Inc., 2004) to 105 ft3/s (Bolke and Vaccaro, 1981), and were computed as residuals to balance ground-water budgets or by calibration of ground-water flow models. Because such underflow cannot be measured directly, there is large uncertainty associated with this water-budget component. The hydrogeologic framework described in this report and anecdotal evidence suggest some outflow to Long Lake, accordingly, the lowest non-zero estimate is used: 55 ft3/s (Drost and Seitz, 1978). This estimated value falls near the middle of the range of previously published estimates and represents 4 percent of the total mean annual outflow.
The City of Spokane reports that approximately 2.3 ft3/s of ground water infiltrates into sewer lines, is treated at the wastewater-treatment plant and discharged to the Spokane River (L. Brewer, City of Spokane, written commun.. February 8, 2006). This seepage is counted in the ground-water budget as a withdrawal and comprises less than 1 percent of the total mean annual outflow of 1,468 ft3/s.
Under natural conditions, over the long term, recharge to an aquifer is approximately balanced by discharge from the aquifer—inflows approximate outflows and there is negligible change in the amount of ground water stored in the system. However, for developed aquifers, short-term climatic variations and subsequent land-use changes, and (or) changes in ground-water use, may tip this balance and water may be taken into or released from storage in the aquifer. The source of water for withdrawals (or pumpage) is either increased recharge, decreased discharge, removal of water from storage, or some combination of the three.
In the absence of artificial recharge, recharge can be increased over natural conditions by such mechanisms as increased infiltration of wastewater, infiltration of runoff from impervious surfaces such as roads or home sites (driveways, roofs, and so on), or infiltration from lakes or streams. Natural discharge is reduced by pumping or otherwise intercepting water that formerly discharged at springs and gaining stream reaches. A decrease in ground-water storage results in water-level declines. Thus, water levels decline if the rate of ground-water recharge is less than ground-water discharge from the aquifer.
If no water was used consumptively, all well pumpage eventually would be returned to the aquifer through direct infiltration resulting in no net water-level change after some sufficient period of time. However, because some pumpage is lost to consumptive use and a large percentage is treated and discharged to the Spokane River, any deficit must come from increased recharge, decreased discharge, or removal of water from storage. In the SVRP aquifer, no significant water-level declines have been observed in the study area; therefore major changes in storage probably have not occurred.
As with most ground-water budgets, there is some degree of uncertainty in the budget presented in this report because many of the components cannot be measured directly. Uncertainty may be associated with the estimation of such budget components because of incomplete data and (or) simplifying assumptions. However, even with components that can be measured directly, such as streamflow losses and gains, some uncertainty is introduced by measurement standard error and temporal variation. As with the measurement of any physical property, there is intrinsic uncertainty associated with the measurement of streamflow and computed discharge. The standard error of a discharge measurement can range from 2 percent under ideal conditions to 20 percent under poor conditions; most measurements fall in the range of 3 to 6 percent (Sauer and Myer, 1992). In addition, values shown in table 2 for streamflow gains and losses were measured over a week during a single season and single year; thus they do not represent a long-term mean.
Many of the components in previous budgets with large uncertainty have been addressed in the current SVRP study using technology and techniques unavailable to previous workers who also were limited by the generally smaller scale of their studies. Regardless, some components of the ground-water budget still cannot be quantified with a high degree of confidence: notably flow into the aquifer from Coeur d’Alene Lake and Lake Pend Oreille, subsurface inflow from the Spirit Valley, and subsurface outflow to Long Lake. Despite this, refinements to other aspects of the ground-water budget described here considerably narrow the range of possible values for these components.
A rough estimate can be made of the probable range of values for the most uncertain ground-water-budget components by making several simplifying assumptions. First, it is assumed that all water-budget component estimates are correct except Coeur d’Alene Lake, Lake Pend Oreille, subsurface inflow from the Spirit Valley, and subsurface outflow to Long Lake. Second, that the estimate of subsurface outflow to Long Lake is reasonably constrained by previously published values: 0 to 102 ft3/s. Third, that the total inflows and outflows will continue to balance within 3 ft3/s as shown in table 2. Using these assumptions, the only uncertain outflow component is subsurface outflow to Long Lake, thus total ground-water outflow is limited to a range of 1,413 to 1,515 ft3/s. Thus to maintain the balance of the water budget, total ground-water inflow must range between 1,416 to 1,518 ft3/s. Consequently the sum of inflows from Coeur d’Alene Lake, Lake Pend Oreille, and the Spirit Valley must vary between 76 and 178 ft3/s; the ground-water budget in this report uses 131 ft3/s representing 9 percent of the total mean annual inflows (table 2, fig. 11). Although this analysis cannot separate individual values for these three inflow components, it does indicate that cumulatively they probably represent between 5 and 12 percent of the total mean annual inflows.
The 3 ft3/s imbalance between estimated inflows and outflows of this ground-water budget (table 2, fig. 11) represents less than 1 percent of the total, and may be due to measurement error, uncertainty in the estimation of water-budget components, or the release of ground water from storage in the aquifer.
U.S.
Department of the Interior | U.S. Geological
Survey
Persistent URL: https://pubs.water.usgs.gov/sir20075041
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:43:04 EST
