Scientific Investigations Report 2008–5071
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2008–5071
Conceptual Model of Hydrologic and Thermal Conditions
The Lower Aquifer of the Eastbank Aquifer system has been used as a water source for the Eastbank Hatchery and the regional water system since the 1980s. Historical water-level and water-temperature data and water-quality data collected in 2007 were analyzed to evaluate how pumping of the Lower Aquifer has affected its hydrologic and thermal conditions. The source of the historical data is a monitoring network operated by the PUD (fig. 10) that has measured hourly water levels and temperatures in 12 wells and 1 river site from 1990 to the present (2008). The network was designed by Water & Environmental Systems Technology, Inc. (1990) as part of a long-term aquifer test that was completed in 1990. The continuous records from January 1991 through August 2007 were used in this study.
Prior to the construction of Rocky Reach Dam, when the Columbia River was a free-flowing river in the study area, the natural low water level of the river was 610 ft (Stone and Webster Engineering Corporation, 1959). The water level in the Lower Aquifer was likely similar to the river level, so the Lower Aquifer was probably largely unconfined and only partially saturated (fig. 5). The Upper Aquifer is presumed to have been unsaturated entirely, except for a possible seasonal, perched unconfined aquifer above the Clay Confining Unit. Once construction of Rocky Reach Dam was completed and the hydroelectric project became operational in 1961, the natural water level was raised by almost 100 ft to normal pool levels ranging from 703 to 707 ft (Public Utility District No. 1 of Chelan County, 2007b). This raised the water level in the Eastbank Aquifer system such that the Lower Aquifer became confined and the Upper Aquifer became partially saturated to form an unconfined aquifer (fig. 5). Water levels are maintained in the Upper and Lower Aquifers by the subsurface cutoff wall that extends to bedrock. This wall consists of a clay curtain across the Upper Aquifer and a grout curtain across the Lower Aquifer. Ground water that seeps through the subsurface cutoff wall from the Lower and Upper Aquifers is captured by drains and flows through the North and South Weirs (fig. 2). Seepage around and through the subsurface cutoff wall likely is not all captured by the drains (G. Yow, Public Utility District No. 1 of Chelan County, written commun., 2008).
Potentiometric-surface maps of the Lower Aquifer prior to the construction of Rocky Reach Dam, which show the altitude at which water levels would have stood in tightly cased wells, were not available for this study. However, in July 1977, prior to pumping of the Eastbank Aquifer system by the regional water system and the Eastbank Hatchery, and years following the decline of ground-water levels to facilitate dam construction, Robinson and Noble, Inc. (as reported by CH2M Hill, 1977) prepared a potentiometric-surface map of the Lower and Combined Aquifers (fig. 11). This map demonstrates that prior to extensive pumping of the aquifer system, water levels in the Lower and Combined Aquifers were lower than the water level in the river, and the primary ground-water flow direction generally was parallel to the river from northeast to southwest (fig. 11). The sparse water-level data also support an alternative interpretation of a more dominant ground-water flow component from the west-northwest along the western extent of the Lower and Combined Aquifers than shown in figure 11. The interpretation shown in figure 11 is reasonable, however. The potentiometric surface shown in figure 11 can be considered a map of post-dam, pre-development water levels, as it represents static water levels prior to the start of extensive, approximately continuous pumping of the Eastbank Aquifer system since 1983.
The sources of ground-water recharge to the Eastbank Aquifer system are flow from the Columbia River, recharge from precipitation, and recharge from irrigation. The mean annual recharge from precipitation is small, because the mean annual precipitation is only 9.1 in. and potential evapotranspiration is large (the mean annual reference evapotranspiration is 44.5 in. at AgriMet weather station MASW in Manson, Washington, located about 25 mi north-northeast of the study area [Bureau of Reclamation, 2008]). In a ground-water recharge study of the Yakima River basin, about 75 mi south of the study area, Vaccaro and Olsen (2007) estimated mean annual ground-water recharge rates of less than 0.5 in. in parts of the lower basin with similar mean annual precipitation and temperatures as the study area and with similarly short vegetation. A regression equation developed by Bauer and Vaccaro (1990) for estimating recharge to the Columbia Plateau regional aquifer system, which is located just to the east of the study area, was used to estimate a mean annual recharge of 0.65 in. to the Eastbank Aquifer system. However, because the study area has thin to no soils and because the surficial geology consists of very coarse gravels, mean annual recharge may be larger. In the northern part of the U.S. Department of Energy Hanford Site, about 80 mi south-southeast of the study area, Bauer and Vaccaro (1990) simulated mean annual recharge from 23 to 39 percent of precipitation with a median of 35 percent of precipitation in areas with coarse sediments and no vegetation (J. Vaccaro, U.S. Geological Survey, written commun., 2008). The median percentage applied to the study area results in a mean annual recharge of 3.2 in. However, because there is vegetation in the study area, actual recharge from precipitation is likely less than 3.2 in. Irrigation is limited to lawn, shrub, and tree irrigation in small parts of the study area and is considered a negligible source of recharge. Thus, the only significant source of recharge to the Eastbank Aquifer system is the Columbia River.
Current (2008) sources of ground-water discharge from the Eastbank Aquifer system are ground-water pumping from the Lower Aquifer and ground-water seepage around and through the subsurface cutoff wall. Although ground-water discharge from the Lower Aquifer to the Columbia River may have been significant prior to the construction of Rocky Reach Dam, it is presumed to be currently (2008) non-existent.
After the completion of Rocky Reach Dam in 1961 and the complete and partial saturation of the Lower and Upper Aquifers, respectively, the Eastbank Aquifer system was not used as a significant resource until 1983, when the regional water system came on-line. This was followed by use of the aquifer system by the Eastbank Hatchery, which started in 1989. (Lincoln Rock State Park, which uses the Eastbank Aquifer system for limited irrigation, started pumping ground water sometime after an agreement was signed with the PUD in April 1980.) In addition to ground-water pumping, ground water also has been removed from the aquifer system by seepage around and through the subsurface cutoff wall.
A history of ground-water pumping was reconstructed using daily flow-meter records for the RW wells provided by the City of Wenatchee (M. Cockrum, written commun., 2007) and seasonal records for the CT and LR wells provided by the Eastbank Hatchery and PUD, respectively (S. Dilly, Public Utility District No. 1 of Chelan County, written commun., 2007). Mostly semi-annual records of discharge through the North and South Weirs, representing seepage through the subsurface cutoff wall, were provided by the PUD (I. Adams, written commun., 2007).
The annual mean pumpage from the RW well field was computed from 1990 through 2006 based on records provided by the City of Wenatchee (M. Cockrum, written commun., 2007; fig. 12A) for years with at least 11 months of data. From 1990 through 2000, annual mean pumpage was relatively constant, with a mean annual pumpage of 10.7 ft3/s. In 2001, the service area of the regional water system was expanded to include the City of East Wenatchee and pumpage increased about 40 percent to a mean annual pumpage of 15.0 ft3/s from 2002 through 2006, excluding 2003. A breakdown of pumpage into quarterly mean pumpage by the RW well field shows that regional water system pumpage is largest in the summer (July–September) and generally smallest in the winter (January–March; fig. 12B). Quarterly pumpage is only shown for quarters with 3 months of data.
The annual mean pumpage from the CT well field was computed from 1991 through 2006 using data provided by the Eastbank Hatchery, including seasonal estimates of pumpage and run-time information for the wells (S. Dilly, Public Utility District No. 1 of Chelan County, written commun., 2007; fig. 12A). Data conflicted for part of 1998. Data available initially indicated an annual pumpage of 42.0 ft3/s and data available later indicated an annual pumpage of 43.2 ft3/s. The initial data were used in this study and are shown in figure 12. Over the years, the hatchery used different methods to estimate pumpage from the CT well field. Prior to 1999, pumpage estimates were based on the run time and nominal pumping rate of each CT well, which is 12.5 ft3/s. From 1999 until November 10, 2001, pumpage estimates were based on flow measurements over damboards at the end of hatchery ponds, and from November 10, 2001 through 2005, pumpage was measured by a flow meter. Based on the available data, it appears that by 2006, pumpage was again estimated based on the run time and nominal pumping rate of each CT well. The mean annual pumpage from 1999 through 2001, which was almost entirely based on damboard measurements, was 36 percent less than the mean annual pumpage during the 10-year period including 1994 through 1998 and 2002 through 2006 (26.7 ft3/s versus 41.4 ft3/s, respectively). Because there was no known change in operations of the hatchery between 1999 and 2001 (S. Dilly, Public Utility District No. 1 of Chelan County, oral commun., 2007), the pumpage estimates based on damboard measurements were assumed to be in error. Instead, new pumpage estimates were computed based on the run time and nominal pumping rate of each CT well. This increased the estimate of mean annual pumpage from 1999 through 2001 by 41 percent to 37.6 ft3/s (fig. 12).
The information provided by the Eastbank Hatchery notes that around the time the flowmeter was installed in late 2001, wells of the CT well field received maintenance that increased the peak capacity of the well field by about 5 ft3/s and restored the well field to close to its nominal capacity of 50 ft3/s. The maximum pumpage from the CT well field prior to well maintenance was reported to be 43 ft3/s, but it is not known how long the wells had been pumping at reduced capacity. The maximum pumpage from the CT well field measured by the flow meter was 48.5 ft3/s in 2003 and 47 ft3/s from 2004 through 2005. This information indicates that pumpage estimates for the CT well field based on the run time and nominal pumping rate of individual wells may be too high by as much as about 15 percent prior to 2002 and by as much as about 6 percent in 2006. The percentage of overestimation of pumpage prior to 2002 would depend on how long the CT wells had been pumping at reduced capacity.
A breakdown of pumpage from the CT well field into quarterly mean pumpage (fig. 12C) and quarterly pumpage as a fraction of annual pumpage (fig. 12D) shows that the seasonal pumpage pattern has changed over time. Since 1994, summer (July–September) pumpage generally has increased, although the annual mean pumpage has remained relatively constant and increases expressed as a fraction of annual pumpage (fig. 12D) generally occurred prior to 1999. Winter (January–March) pumpage generally decreased from 1994 until about 2002. Starting in 2001, summer pumpage exceeded winter pumpage, except in 2006. From 1999 through 2006, the overall seasonal pumpage patterns were relatively stable. The seasonal pumpage patterns of the CT well field differ from the RW well field because pumpage from the CT well field is determined by fish-production needs of the Eastbank Hatchery and pumpage from the RW well field is determined by public-supply needs. In 2006, the mean annual pumpage from the CT and RW well fields was about 43 and 16 ft3/s, respectively.
Other well fields in the study area are the SW and LR well fields, and pumpage from these well fields is negligible compared to pumpage from the RW and CT well fields. The original purpose of the SW well field was to lower ground-water levels during construction of Rocky Reach Dam. Since completion of the dam, two wells of the SW well field (SW13 and SW14) have continued to be pumped at a combined rate of about 80 gal/min or about 0.2 ft3/s to lubricate turbines at Rocky Reach Dam and one well, SW11, has been used seasonally to irrigate small parts of the study area. Pumpage from the SW11 well is unknown but it is assumed to be negligible. The LR well field provides water for irrigation of Lincoln Rock State Park. The well field has been in operation since about 1980 and the wells are pumped several hours per day for about 6 months per year. Pumpage data for 2004 through 2006 (S. Dilly, Public Utility District No. 1 of Chelan County, written commun., 2007) show that the mean annual pumpage from the LR well field is about 0.14 ft3/s. It is assumed that mean annual pumpage has remained relatively constant from the LR well field since the start of its operation and from the SW well field since the start of its use for irrigation and turbine lubrication.
Ground-water seepage through the subsurface cutoff wall that was captured by drains has been measured in the North and South Weirs since July 1, 1977 (I. Adams, Public Utility District No. 1 of Chelan County, written commun., 2007; fig. 13). Weir stage is recorded by a strip-chart recorder and converted to discharge using the stage-discharge rating curve for the weir. Prior to 1989, weir discharge was recorded monthly and starting in 1989, weir discharge was mostly recorded semi-annually. Other than a note indicating that the weirs were recalibrated sometime between late 2001 and late 2002, the only other historical calibration information available to this study was from K. deRubertis and others (written commun., 2007) stating that measurements made on February 24, 2006, correlated with the recorded values. At 8:43 a.m. on July 17, 2007, measured stage of the water in the box of the North Weir was 0.36 ft and recorded stage was 0.335 ft. These values are close (within 7 percent of each other) and so both the North and South Weirs were assumed to be accurately recording weir stage. A more detailed analysis of the weir recordings was beyond the scope of this study.
Seepage from the Lower Aquifer is assumed to flow through the North Weir and seepage from the Upper Aquifer is assumed to flow through the South Weir. This assumption is based on the significant decrease in discharge that occurred in the North Weir after October 1987 (fig. 13) that coincided with the installation of the CT well field. No other information is available to support this assumption and it is possible each weir captured a combination of seepage from the Lower and Upper Aquifers.
Mean discharges in the North Weir were 5.4 ft3/s from July 1977 through October 1987 and 3.1 ft3/s from November 1987 through December 1998. This means that the long-term mean seepage from the Lower Aquifer through the subsurface cutoff wall decreased by 2.3 ft3/s or 43 percent. In contrast, discharge in the South Weir, and thus seepage from the Upper Aquifer through the subsurface cutoff wall, was relatively constant during the same time period, showing an increase of 0.7 ft3/s or about 13 percent, from 5.4 to 6.1 ft3/s.
The first wells of the CT well field were completed in November 1987, followed by CT3 in December 1987 and CT4 in January 1988 (table 3). Even though the Eastbank Hatchery did not become operational until July 1989 and expanded to full operations by about November 1989 (I. Adams, Public Utility District No. 1 of Chelan County, written commun., 2008), it is postulated that the CT well field started pumping as soon as wells were installed and that this explains the decrease in seepage from the Lower Aquifer.
From 2004 through 2006, the mean total discharge through the North and South Weirs was 8.4 ft3/s. This indicates that from 2004 through 2006, the mean annual ground-water pumpage from the Eastbank Aquifer system was about seven times the measured portion of the mean annual ground-water seepage through the subsurface cutoff wall.
Water levels have been measured hourly in 12 wells and 1 river site by the PUD since 1990 (fig. 10) for the purpose of monitoring hydrologic conditions of the Eastbank Aquifer system. Daily means of the hourly water levels for January 1, 1991 through September 30, 2006, are shown in figure 14. In addition to the continuous, hourly measurements, occasional manual water-level measurements have been made to verify the continuous measurements.
The historical hourly water levels have been measured using sealed probes that measure both water levels and water temperature (see Water & Environmental Systems Technology, Inc., 1990, for a detailed description of the instrumentation). Because the probes are not vented to the atmosphere, changes in atmospheric pressure are recorded as apparent changes in water levels, even if the actual water levels did not change in response to changes in atmospheric pressure. Without corrections for atmospheric effects, errors between actual and recorded water levels by non-vented probes may be on the order of inches to feet (Wardwell, 2007). Other sources of error in the recorded water-level data are instrument drift, possible incomplete information about the timing of probe replacements, and noise that resulted from spliced transmission cables that had to be repaired. Probes were replaced in case of obvious instrument failure (D. Davies, Public Utility District No. 1 of Chelan County, oral commun., 2007).
To be able to correct for instrument drift and other possible sources of error, manual water-level verification measurements are needed at regular intervals throughout the year for comparison to recorded water levels. The first water-level verification measurements since the start of the monitoring network in 1990 known to this study were made in July 1998, when the monitoring network was recalibrated. This recalibration resulted in significant shifts in some recorded water levels at that time (fig. 14). (Sudden rises in water levels during the spring of the mid-1990s correspond to seasonal reductions in pumping by the CT wells.) The next known verification measurements were made in February, 2006 (K. deRubertis and others, written commun., 2007) followed by periodic measurements made by this study from July 2007 through January 2008. Differences between water levels measured manually in February 2006 (K. deRubertis and others, written commun., 2007) and July 2007 and those recorded by the monitoring network are summarized in table 4. Other than the outlier of 9.51 ft, the differences in the measurements range from -2.39 to 3.39 ft. Because few verification measurements have been made over the period of record of the monitoring network, there is significant uncertainty in the historical continuous water-level data.
To evaluate the accuracy of the long-term record of the river probe (RIV in fig. 14), river water levels measured by the monitoring network were compared to river water levels recorded by USGS gaging station 12453679 (period of record July 25, 1975 to the present [2008]). The gaging station is located along the west bank of the river at the forebay of Rocky Reach Dam (fig. 2). The gaging station measures hourly water levels but only measurements at midnight are reported by the USGS (fig. 14). Recorded water levels are verified at regular intervals with manual measurements and the water-level record is adjusted accordingly. A comparison of the water-level record of the USGS station and the probe labeled RIV shows that probe RIV has been subject to a steady downward instrument drift of about 3 ft since 1990. Probes in wells of the monitoring network are of the same type and it is likely that they have drifted also, but the magnitude and direction of their drifts over the life of the monitoring network cannot be determined due to a lack of manual water-level measurements.
The pre-development potentiometric surface measured in the Eastbank Aquifer system in 1977 (fig. 11) shows that potentiometric gradients in the study area are small and thus accurate water-level measurements are required to be able to use the historical continuous water-level data for analyses of long-term changes in the aquifer system. The uncertainty of the historical continuous water-level data is too large for this purpose and so these data were not further analyzed in this study.
To make equipment repairs at the Eastbank Hatchery, the PUD scheduled a shutdown of the CT well field from 1 to 3 p.m. on July 18, 2007. Because a complete shutdown of all CT wells is rare, the opportunity was used to schedule a shutdown of all significant pumping wells in the study area and to measure water levels before and after the shutdown. Arrangements were made with the City of Wenatchee and Lincoln Rock State Park to simultaneously shut down the RW and LR well fields. In addition, the PUD shut down wells SW13 and SW14, and only continued to pump about 20 gal/min from well SW11. From about 9 a.m. until about 5 p.m. on July 18, 2007, the recording frequency of the monitoring network was increased from once per hour to once per minute and a team of PUD and USGS personnel made multiple manual water-level measurements from about 10 a.m. until shortly after 3 p.m. River water levels measured at USGS gaging station 12453679 were nearly stable between 10 a.m. and 3 p.m., ranging from 704.33 ft at 10 a.m. to 704.59 ft at 1 p.m. Water levels were not measured in the LR and SW wells because they were not accessible, and water levels were not measured in well CD6 because it had not been located at the time. In addition, before pumping ceased, water levels were not measured in pumping wells except for one measurement in well RW4. The water-level data were used to document the ground-water flow pattern of the Lower and Combined Aquifers before and after the shutdown of pumping.
During the shutdown, the water-level probe in well TH1 was not working properly and the water level in well TH9 did not respond to the shutdown of pumping. The lack of response in well TH9 is considered anomalous and remains unexplained. A slug test was performed in well TH9 to determine if the well was isolated from the aquifer due to clogging of its perforations. However, changes in water levels induced by the slug test dissipated quickly and thus the well perforations were not clogged. A subsequent down-hole camera survey of well TH9 revealed that it contained a 1-inch inner-diameter PVC pipe at depth attached to a heavy object that was presumed to be a pump. An attempt to remove the pipe and pump was unsuccessful (see appendix 2 for additional detail).
Multiple water levels measured manually prior to the shutdown of pumping wells were averaged, plotted, and contoured to create a potentiometric-surface map of the Lower and Combined Aquifers (fig. 15). At the time, wells CT1, CT2, CT3, CT4, RW2, RW4, LR1, LR2-W, SW11, SW13, and SW14 were pumping. The nominal pumping rate of each CT well was 12.5 ft3/s, and wells RW2 and RW4 were pumped at rates of 18.5 and 22.3 ft3/s, respectively. The combined pumping rates of the LR and SW wells were estimated to be 0.3 and 0.2 ft3/s, respectively. Figure 15 shows that with four CT and two RW wells pumping, a cone of depression surrounds the RW field and another cone of depression surrounds the CT well field. The cone of depression of the CT well field widens towards the south because some ground water drains through the subsurface cutoff wall. The cones of depression of the RW and CT well fields intersect and create a ground-water divide along an approximately east-west line going through the general area between wells RW1 and TH8. Based on the potentiometric contours, the horizontal ground-water flow direction is approximately radial towards the pumping wells with source water originating along the aquifer boundary with the Columbia River (fig. 15). A steep potentiometric gradient exists between the river and the western extent of the Lower and Combined Aquifers, indicating that the bottom of the Columbia River is blanketed with materials of lower permeability than the sediments of the Lower and Combined Aquifers. An aquatic habitat study of Lake Entiat confirmed the presence of fine-grained sediments at the bottom of the Columbia River near the study area (Duke Engineering & Services, Inc., 2001). The layer of fine-grained sediments impedes ground-water recharge from the Columbia River but it does not prevent it, as demonstrated by the potentiometric-surface maps (figs. 11 and 15). In addition to the approximately radial horizontal ground-water flow toward the pumping wells, ground water also flows towards the subsurface cutoff wall. The majority of flow to the CT well field may come from the small embayment northeast of and adjacent to Rocky Reach Dam (fig. 15). The aquatic habitat study mapped large cobble and gravel at the bottom of the embayment (Duke Engineering & Services, Inc., 2001), which is riprap that was put at the bottom of the embayment after it was excavated as part of the construction of Rocky Reach Dam.
Following the shutdown of the pumping wells at 1 p.m., multiple water levels were measured manually in monitoring wells and wells that were previously pumping. The manual measurements continued until shortly after 3 p.m., when the pumping wells that were shut down at 1 p.m. were turned back on. The last set of manual water-level measurements that were made from 3 to 26 minutes before 3 p.m. were extrapolated to 3 p.m. to obtain a set of estimated water levels that represented the potentiometric surface 2 hours after the start of water-level recovery. For wells that are part of the monitoring network, extrapolations of manual water-level measurements were based on the water-level recovery pattern recorded every minute by the monitoring network; for wells that are not part of the monitoring network, extrapolations were made visually based on trends in multiple manual water-level measurements. Water-level adjustments ranged from -0.1 to 0.5 ft with a median adjustment of 0.1 ft. Figure 16 shows that after 2 hours of water-level recovery, the cones of depression surrounding the RW and CT well fields remain, although the potentiometric gradients are less steep. Excluding water-level recoveries of 0 ft in well TH9 and 8.7 ft in well RW4, the water levels recovered from 0.8 ft in well TH5 to 4.7 ft in well CD47, with a median recovery of 4.1 ft. The water levels recorded every minute by the monitoring network indicated that although recovery had slowed down after 2 hours, complete recovery was not achieved. Complete recovery would be expected to return the potentiometric surface to the predevelopment conditions measured on July 19, 1977 (fig. 11). Based on the pattern of the water-level contours, the horizontal ground-water flow directions after 2 hours of recovery were still similar to the flow directions prior to the shutdown of the pumping wells.
Water temperatures have been measured hourly in 12 wells and 1 river site by the PUD since 1990 for the purpose of monitoring thermal conditions of the Eastbank Aquifer system. The river probe (RIV in fig. 10), which also measures the water level of the river, is in a PVC pipe draped along the bottom of the river, about 200 ft offshore. Hourly water temperatures measured in the monitoring-network wells represent temperatures at particular depths and do not provide vertical temperature profiles. Daily median temperatures were computed from hourly temperatures for 2006 (fig. 17), which is an example of a typical 1-year period. The temperature records indicated that different wells have different annual temperature ranges and different time lags between changes in river temperatures and subsequent changes in well temperatures (fig. 17).
Water & Environmental Systems Technology, Inc. (1990) measured vertical temperature profiles in wells CD6, CD47, TH1, TH4, TH5, TH6, TH7, TH8, and TH9 at selected times between July 1989 and April 1990. These profiles indicated that there were significant vertical temperature gradients in the Eastbank Aquifer system that changed seasonally. In July 1987, CH2M Hill (1988) measured vertical temperature profiles in wells TH4, TH5, and TH6 but these data were not available to this study. In February 2006, K. deRubertis and others (written commun., 2007) measured vertical temperature profiles in the same wells as CH2M Hill and also wells CD8, CD10, CD47, LR2-W and TH7. This study measured vertical temperature profiles in wells CD10, TH1, TH4, TH6, TH7, and TH9 between August and September 2007 and started a network of monthly measurements of vertical temperature profiles in 12 wells (CD6, CD8, CD10, CD47, TH1, TH2, TH4, TH5, TH6, TH7, TH8, and TH9) in December 2007 that is now maintained by the PUD.
On August 20, 2007, 32 vertical temperature profiles were measured using CTD (conductivity-temperature-depth) casts in the Columbia River near the study area to determine if the river was thermally stratified in the area of likely ground-water recharge. Profiles were located in an approximately 500-ft-wide band extending from the north shore near the boat ramp (fig. 2) to the shore along the embayment adjacent to Rocky Reach Dam. One profile was measured in the center of the river. Excluding two anomalous temperature profiles located near the outfall from the Eastbank Hatchery, negligible temperature variation was measured. Water temperatures ranged from 19.0 to 19.5°C, with a median of 19.3°C and a standard deviation of 0.1°C. Previous temperature studies also showed a lack of both vertical and lateral stratification of the river near the study area. For example, in a series of approximately monthly vertical temperature-profile measurements from October 1999 through September 2000 located in the center of the river approximately due west of well TH6, Parametrix, Inc. and Rensel Associates Aquatic Science Consultants (2001) measured a maximum temperature range of 0.32°C on July 14, 2000. Parametrix, Inc. and Thomas R. Payne & Associates (2002) recorded nearly constant temperatures in a temperature transect of the river adjacent to the study area in the morning and afternoon of September 2, 2001, except for a slight warming of the surface layer in the afternoon ranging from 0 to 1.3°C. Based on these data, thermal stratification of the Columbia River near the study area was insignificant and water temperatures measured at any nearby river location were representative of the temperature of water that recharged the Eastbank Aquifer system within about ±0.5°C.
The historical hourly water temperatures were measured with the same sealed instruments used to measure water levels. Similar to the lack of manual measurements to verify water levels recorded by the monitoring network, there also was a lack of manual measurements to verify water temperatures recorded by the network. Temperature probes also are subject to failure, instrument drift, possible incomplete information about the timing of probe replacements, and the addition of noise due to transmission-cable repairs. However, water-temperature probes generally are more robust than water-level probes and their measurements are not affected by day to day changes in atmospheric pressure.
To evaluate the accuracy of the river probe (probe RIV), hourly river temperatures measured by the monitoring network were compared with hourly river temperatures measured at the forebay of Rocky Reach Dam (S. Hayes, Public Utility District No. 1 of Chelan County, written commun., 2007) for the time period of concurrent data, March 25, 2003 through August 15, 2007. River temperatures are measured at the forebay for the purpose of monitoring fish habitat and the time series used in this study consisted of merged records of four separate temperature probes (S. Hayes, Public Utility District No. 1 of Chelan County, written commun., 2007). Figure 18 shows that except for some scatter at the low-temperature range that represents several days in January 2007, there is a close match between the temperatures measured at midnight at the forebay and by probe RIV. Statistics based on all hourly data for the concurrent period show that the differences in temperature range from -0.7 to 2.1°C, with a median difference of 0.3°C and a standard deviation of 0.2°C. These results indicate that the temperature measurements of probe RIV since March 2003 have been accurate within a comparable margin of error (generally less than 1°C) as may result from assuming that water temperatures measured at any nearby river location are representative of the temperature of water that recharges the Eastbank Aquifer system. In addition, because probe RIV has not been replaced since its installation in 1990, it is reasonable to assume that temperature measurements by probe RIV have been reliable since 1990.
Similar temperature data were not available to verify the accuracy of temperature measurements by probes in wells of the monitoring network. Instead, probes were pulled from three wells (TH4, TH6, and TH7) and submerged in a water bath with two thermometers to obtain a representative comparison of temperature readings by the monitoring network (table 5). One thermometer was a Cole-Parmer® reference thermometer calibrated against a National Institute of Standards and Technology (NIST) standard thermometer (INNOCAL test no. 22144) and the other thermometer was a 300-ft-long TLC (temperature-level-conductivity) probe made by Solinst®. The accuracy of the TLC thermometer was confirmed using a 4-point verification in the USGS laboratory in Tacoma, Washington, on July 31, 2007. The temperature probe of a fourth well (CD10) was checked in place by submerging the TLC thermometer into the well to the reported depth of the probe. The probe was not pulled from well CD10 because it was stuck. Side-by-side thermometer and monitoring-network probe comparisons in a water bath are considered more reliable because the method leaves no question that all instruments are measuring the same water and that the measurements are made simultaneously. The results of the temperature comparisons indicate that the monitoring-network probes for wells TH4, TH6, and TH7 measured temperatures within 0.3°C of the reference thermometer (table 5). The monitoring-network probe for well CD10 measured temperature within 0.1°C of the TLC thermometer. Based on verifications of the subset of monitoring-network probes, bias and variability in temperature measurements for all monitoring-network probes were assumed to be less than 0.5°C.
In February 2006, K. deRubertis and others (written commun., 2007) made in-place comparisons of temperatures recorded by all probes of the monitoring network, except the probe in well CD47, at reported probe depths. They found that temperatures matched within 0.5°C, except for probes in wells LR2-W and TH8 for which differences exceeded 1.5°C. Water temperatures recorded once per minute on July 18, 2007, indicated that several temperature probes of the monitoring network had noisy recordings (defined as large variability over short periods of time), including the temperature probes of wells LR2-W and TH8. The large temperature discrepancies reported by K. deRubertis and others (written commun., 2007) for the probes in wells LR2-W and TH8 may therefore have resulted from small differences between the times the verification temperatures were read and the times the probe temperatures were recorded by the monitoring network.
The limited verification data available for the temperature probes of the monitoring network indicate that many of the temperature probes may be making reliable temperature measurements and presumably have done so since they were installed or last replaced. Even if water temperatures recorded by the monitoring network cannot be relied on with great certainty due to limited verification data, the relative pattern of the historical temperature record is reliable and the recorded times of the annual minimum and maximum water temperatures are likely accurate within 1 day.
Ground-water recharge transports heat from the Columbia River to the Eastbank Aquifer system, and each location in the aquifer system has an annual temperature record that mimics the annual temperature record of the river. Generally, with increasing distance from the river, the time lag between a change in river temperature and a subsequent change in well temperature increases and the annual temperature range decreases. The spatial and temporal patterns of ground-water temperatures may change as thermal and hydraulic conditions change in the river and/or aquifer system. Water temperatures measured by the monitoring network were analyzed to determine if and how patterns of ground-water temperatures in the Eastbank Aquifer system have changed.
Hourly water temperatures measured by the monitoring network from January 1, 1991 through August 31, 2007, were simplified to time series of daily median temperatures. The entire record of daily values was analyzed to determine if there were trends in the time lags between changes in river and well temperatures. The part of the record starting in 1999 was analyzed to determine if there were trends in the annual minimum and maximum well temperatures and in the annual temperature ranges of wells with respect to the river. The well-temperature record prior to 1999 was not analyzed for trends in annual extreme temperatures and annual temperature ranges due to uncertainty in the data. An analysis of temperature trends in the Eastbank Aquifer system must consider both horizontal and vertical variability of ground-water temperatures. A limited number of vertical temperature profiles were available to illustrate the three-dimensionality of ground-water temperatures, but too few profiles were available to determine interannual trends in vertical temperature profiles.
Analyses of trends in time lags and annual temperature ranges help determine if the ground-water flow system is in thermal equilibrium. In this study, the ground-water flow system is defined to be in thermal equilibrium at a given location if the time lags between changes in river temperatures and subsequent changes in ground-water temperatures are constant at that location. The equilibrium is a dynamic equilibrium because temperatures vary throughout the year. When the ground-water flow system is in thermal equilibrium, the ratios of annual temperature ranges in the wells to annual temperature ranges in the river also should be constant at a given location. Because transport of heat is primarily by advection (flow of water) within aquifers and by conduction within confining units, and because transport by advection is faster than by conduction (Miller and Delin, 2002), decreasing trends in time lags and increasing trends in ratios are likely explained by increasing transport of heat by advection. This transport increases as ground-water fluxes increase due to increases in pumping and/or increases in hydraulic conductivities as fine sediments in the aquifers are removed or rearranged as a result of pumping and preferential flowpaths form.
Because the only significant source of recharge to the Eastbank Aquifer system is the Columbia River, it is important to know whether there have been trends in river temperature during the periods of analysis. Annual minimum, maximum, and mean river temperature measured by probe RIV (fig. 10) from 1991 through August 2007 is shown in figure 19. Straight-line linear regressions of each time series for the entire time period show that there are no statistically significant trends in the annual minimum, maximum, and mean river temperature at a confidence level of 95 percent. However, straight-line linear regressions of each time series starting in 1999 show that there are statistically significant trends in the annual mean and maximum river temperature at a confidence level of 95 percent, indicating a mean annual increase in the annual mean and maximum river temperature from 1999 through 2006 of 0.07 and 0.17°C, respectively (fig. 19). There are no statistically significant trends in the annual minimum river temperature since 1999 at a confidence level of 95 percent, nor are there statistically significant trends in the annual minimum, maximum, and mean river temperature from 1991 through 1998 at a confidence level of 95 percent.
The analysis of well-temperature records was based on the assumption that the water temperature measured at a given depth in a monitoring well was representative of the water temperature at the same depth in the ground-water flow system outside the well. This assumption is justified because convective flow, and thus temperature-controlled density stratification of the water column, is unlikely to occur in the monitoring wells analyzed in this study due to their small diameters (3 to 8 in.; Diment, 1967; Gillespie, 1995). Well CT3, which is a hatchery well with a temperature probe located about 85 ft above the top of the open interval, has a casing diameter of 26 in. for most of its depth. Due to its large diameter, density stratification may occur in well CT3 when it is not pumping. When the well is pumping, the temperature probe measures water temperatures that are likely affected both by the ambient temperature of the ground-water flow system outside the well and the temperature of pumped water moving through a pipe inside the well casing. Well CT3 was pumping the majority of the time, except during 1991 and 1992. Annual temperature extremes measured in well CT3 and that were used in the trend analyses were all measured when well CT3 was pumping, except for the annual maximum temperature in 1991 and the annual minimum temperatures in 1992, 1998, 2005, and 2006. Due to the poorly known variables that affect the temperature record of well CT3, there is greater uncertainty in the interpretation of the temperature records of well CT3 than the records of the monitoring wells.
Figure 20 shows selected historical vertical temperature profiles in three wells, CD6, CD47, and TH8, measured in 1989–90 by Water & Environmental Systems Technology, Inc. (1990), in 2006 by K. deRubertis and others (written commun., 2007), and in 2007–08 by this study. Profiles were selected to illustrate the maximum temperature range that may occur in each hydrogeologic unit in each well. Although wells CD6 and TH8 had multiple temperature-profile measurements throughout 1989–90, none of the wells had a sufficient number of measurements to be sure that the annual temperature extremes had been measured.
The Upper Aquifer at well CD6 (fig. 20A), which is located near the shoreline and subsurface cutoff wall (fig. 15), has a temperature range that is almost identical to the temperature range of the river (fig. 19). In addition, the annual extreme temperatures in the river and Upper Aquifer at well CD6 occur at about the same time of year; usually the annual minimum and maximum river temperatures occur in February and August or September, respectively. Temperatures in the Upper Aquifer at the location of well CD6 track the river temperatures closely, because sediments of the Upper Aquifer near the well were highly disturbed during construction of Rocky Reach Dam and probably consist primarily of high-permeability backfill. The annual temperature range in the Lower Aquifer at well CD6 is smaller than that in the Upper Aquifer and the dates of annual high and low temperatures are different between the aquifers, with the Clay Confining Unit showing transitional patterns.
Wells CD47 and TH8 (figs. 20B and 20C) show complex vertical temperature profiles that are different from each other and well CD6, indicating that each location in the Eastbank Aquifer system has a set of unique, time-varying vertical temperature profiles. The temperature profile at each location at a given time is a function of the temperature and water level of the river; the proximity to the river, pumping wells, and subsurface cutoff wall; the rate and schedule of pumping; the horizontal and vertical hydraulic conductivities and thicknesses of the hydrogeologic units; and the thermal properties of the sediments of and the bedrock beneath the Eastbank Aquifer system. Even though the vertical temperature profiles of wells CD6, CD47, and TH8 differ, temperature changes across the confining units at each of these wells indicate that the flow systems of the Upper and Lower Aquifers are not tightly connected although there is a stronger connection at well TH8.
A comparison of the vertical temperature profiles of the Lower Aquifer of wells CD47 and TH8 shows that generally, the vertical temperature gradients are slightly larger in well TH8 than in well CD47 indicating slightly more flow in the upper part of the Lower Aquifer near well TH8. Well TH8 has no perforated interval, so water inside its casing has likely equilibrated to the temperature of the surrounding aquifer by conduction. One possible explanation for the increased flow in the upper part of the Lower Aquifer is that it has a relatively larger horizontal hydraulic conductivity, although the driller’s log for well TH8 did not indicate a significant change in lithology from the upper to the lower parts of the Lower Aquifer. An alternative explanation for the increased flow in the upper part of the Lower Aquifer near well TH8 is that the well is very near the CT well field, and in particular near well CT4 (fig. 15). Well CT4 is open to the Lower Aquifer from an altitude of 549 to 575 ft and the open intervals for all wells of the CT well field range from 540 to 576 ft.
The smaller vertical temperature gradients in the lower part (below the altitude of the CT well field open intervals) of the Lower Aquifer of well TH8 during much of the year represent colder and thus denser water that probably is pumped by the CT well field at a lower rate than water in the upper part (within the altitude range of the CT well field open intervals) of the Lower Aquifer. The source of some of this colder water may be colder and denser water that settled locally in the bedrock depression north and west of well TH8 (fig. 7) and is captured by well CT4. The smaller vertical temperature gradients in the Lower Aquifer of well CD47 may result from mixing of water at and near the open interval because well CD47 is perforated below an altitude of about 560 ft. The mixing would result in more uniform water temperatures in the well that represent the average ambient ground-water temperatures near the perforated interval. Alternatively, mixing in well CD47 is minimal and the vertical temperature profile measured in the well reflects the vertical temperature profile of the ambient water temperatures. Vertical temperature profiles not shown for other monitoring wells with perforated intervals (for example, wells TH1, TH4, and TH7) include significant vertical temperature gradients (up to about 0.1°C/ft) adjacent to open intervals and so it is assumed that generally, vertical temperature profiles measured in monitoring wells with perforated intervals are representative of vertical temperature profiles of ambient temperatures.
The vertical temperature profiles in figure 20 also indicate that the depth at which a temperature probe is located substantially influences the ground-water temperature measurements. For example, a probe located in the upper part of the Lower Aquifer of well TH8 (fig. 20C) would measure a different temperature record than a probe located in the lower part of the aquifer, where the probe in well TH8 has been located since the start of the monitoring network. Trend analyses of the temperature record of a probe that has remained at a constant depth through time can provide useful insights into the dynamics of the aquifer system.
Temporal trends in time lags between changes in river and ground-water temperatures; annual ranges in ground-water temperatures with respect to annual ranges in river temperatures; and annual extreme ground-water temperatures were analyzed spatially.
Time series of daily median water temperatures were used to estimate the time lag between a change in river temperature and the subsequent change in ground-water temperature at a given well. Particular focus was on the time lag between annual minimum and maximum temperatures. The time lags were estimated as the difference between the date of an annual extreme temperature in the river and the date of the subsequent annual extreme temperature in individual wells. Time lags did not exceed 1 year. The resulting annual time series of time lags of annual minimum and maximum temperatures are shown in figure 21. Time lags of wells in which the subsequent annual extreme occurred in the following year are plotted according to the year of the annual extreme in the river. Data are missing for years when well-temperature records had gaps or when the record was unclear when the annual minimum or maximum occurred.
Trends in the time series of time lags from 1991 through 2007 were evaluated using straight-line linear regressions with each of the time series shown in figure 21. The results of the trend analyses are summarized in table 6, and trends of decreasing time lags that are statistically significant at a confidence level of 95 percent are shown in figure 21. A decreasing trend in time lags is indicated (table 6) if the trend is observed in the time lag for either the annual maximum or the annual minimum temperature. The results indicate that there were no decreasing trends in time lags in the Upper Aquifer and Clay Confining Unit. However, with the exception of wells TH1 and TH4, there were decreasing trends in time lags in the Lower and Combined Aquifers. The lack of decreasing trends in Lower Aquifer wells TH1 and TH4 indicates that the thermal conditions of the Lower Aquifer near wells TH1 and TH4 have been in equilibrium for the pumping and aquifer conditions that have existed since 1991. Any decreases in time lags near wells TH1 and TH4 presumably would have occurred following activation of the CT and RW well fields, respectively, and prior to 1991.
Trends in the time series of time lags also were evaluated from 1991 through 1998 and from 1999 through 2006. The selected time periods were arbitrary, except that the period 1999-2006 coincides with the period of analysis of annual well-temperature ranges described in section “Annual Temperature Ranges.” From 1991 through 1998, there were statistically significant decreasing trends at a confidence level of 95 percent in either the annual minimum or maximum temperatures in wells CD8 and CD10 in the Clay Confining Unit and all wells in the Lower and Combined Aquifers except wells TH1 and TH4 (table 6; fig. 21). (The temperature probes in wells CD8 and CD10 are adjacent to sand lenses near the base of the Clay Confining Unit that may be connected to the Lower Aquifer.) From 1999 through 2006, there were no statistically significant trends in time lags, except for well LR2-W (table 6; fig. 21). These results indicate that during 1999-2006, the Lower and Combined Aquifers were in thermal equilibrium except for the Combined Aquifer near well LR2-W. The thermal equilibrium was reached prior to 1991 in the Lower Aquifer near wells TH1 and TH4, and during 1991-98 in the remainder of the Lower and Combined Aquifers except for the Combined Aquifer near well LR2-W.
The spatial distribution of decreasing trends in time lags of wells in the Lower and Combined Aquifers during 1991–98 and the time lags of the annual maximum temperatures in 1991, 1999, and 2006 are shown in figure 22. The distribution of time lags shows that as early as 1991, the time lag in well TH1 was significantly shorter than in wells TH7, TH8, and TH9, which is likely the result of a large ground-water flux from the Columbia River west-to-southwest of well TH1 to the CT well field. The short time lag in well TH1 as early as 1991 is consistent with the lack of decreasing trends in time lags at the well during 1991–2007 and 1991–98 because the Lower Aquifer near the well had already reached equilibrium by 1991. The same is not true for the Lower Aquifer near wells TH7, TH8, and TH9, which are located approximately along the ground-water divide between the CT and RW well fields (fig. 15). In 1991, 1999, and 2006 these wells showed a pattern of increasing time lags of the annual maximum temperatures with distance from the river (fig. 22). However, during 1991-98, the time lags of the annual maximum temperatures decreased for all three wells (fig. 21J–L), ranging from a mean annual decrease of 7.0 days in well TH9 to 10.0 days in well TH8.
The time series of daily median water temperatures were also used to determine if there were trends in the annual temperature ranges of ground water in wells compared to annual temperature ranges in the river. Time series of ratios of annual temperature ranges in the wells to annual temperature ranges in the river were computed and the results are shown in figure 23. Ratios for wells in which the corresponding annual temperature range spanned more than one calendar year are plotted according to the year of the annual extreme in the river. Data are missing for years when well-temperature records had gaps or when the record was unclear what the magnitude was of the annual minimum or maximum temperature.
Figure 23 indicates that the annual temperature-range ratios for well TH6 have been increasing since 1991. Straight-line linear regressions were performed on each of the time series from 1999 through 2006, to determine quantitatively if there were trends in the time series of annual temperature-range ratios. Values prior to 1999 were not included in the analysis because the monitoring network was recalibrated in July 1998, which may have affected the magnitude but not the timing of annual temperature extremes. The effects of the recalibration are indicated in figure 23 as offsets in 1998 in several of the time series of annual temperature-range ratios. Trends of increasing annual temperature-range ratios that are statistically significant at a confidence level of 95 percent are summarized in table 6. The results indicate that in the Lower and Combined Aquifers, the trends in the annual temperature-range ratios support the 1999-2006 trends in the time lags for 4 of 7 wells (wells LR2-W, TH1, TH7, and TH8). For the remaining 3 wells in the Lower and Combined Aquifers (wells TH4, TH6, and TH9), trends in the annual temperature-range ratios indicate that the Lower and Combined Aquifers near those wells have not reached thermal equilibrium. However, it is assumed that the Lower and Combined Aquifers near wells TH4, TH6, and TH9 are in thermal equilibrium, because trends in time lags are more reliable than trends in annual temperature-range ratios. Trends in time lags are more reliable because time lags are estimated from the relative patterns (and thus timing) of temperature records that are likely accurate within 1 day, and annual temperature-range ratios are estimated from 4 temperature measurements with unknown error. The cumulative errors in annual temperature-range ratios make the metric less reliable and may help explain discrepancies between results of the trend analyses for time lags and annual temperature-range ratios for some of the wells.
The time series of daily median water temperatures were also used to determine if there were trends in the annual minimum and maximum ground-water temperatures from 1999 through 2006. No statistically significant trends were found in the annual minimum temperatures at a confidence level of 95 percent, but statistically significant trends of increasing annual maximum temperatures were found in all but 3 wells of the monitoring network (fig. 24). Annual maximum temperatures of wells in which the corresponding annual maximum temperature in the river occurred the previous calendar year are plotted according to the year of the annual maximum in the river. Data are missing for years when well-temperature records had gaps or when the record was unclear what the magnitude was of the annual maximum temperature. Mean annual increases in annual maximum temperatures from 1999 through 2006 ranged from 0.12°C in well CD47 to 0.26°C in well TH1 (table 6) and averaged 0.19°C for all wells with statistically significant increases in annual maximum temperatures. The lack of an increasing trend in annual maximum temperatures in Lower Aquifer well TH8 may be due to heat attenuation with distance from the Columbia River that decreases the variability of annual maximum temperatures in this part of the aquifer that is minimally affected by pumping (fig. 20C). Alternatively, the lack of a trend could be due to a source of colder water that settled locally in the bedrock depression north and west of well TH8 (fig. 7) and is captured by pumping well CT4.
Increases in annual maximum ground-water temperatures cannot be larger than increases in annual maximum river temperatures. However, the mean annual increases in annual maximum ground-water temperatures in the Lower and Combined Aquifers from 1999 through 2006 ranged from 0.18°C in well TH7 to 0.26°C in well TH1, whereas the increase was 0.17°C in the river; the discrepancy indicates that the trend analysis results in errors in the estimated mean annual increase in annual maximum temperatures of ±0.09°C. Although there is uncertainty in the magnitude of the mean annual increases in annual maximum river and ground-water temperatures, figure 24 and the trend analyses indicate that the river and most annual maximum well temperatures have generally been increasing from 1999 through 2006. Because there was no trend in the annual minimum well temperatures during the same period, the mean annual well temperatures will also have increased in most wells from 1999 through 2006 although less than the annual maximum well temperatures.
Water-quality samples were collected from nine ground-water and one surface-water location in the study area on August 20–22, 2007, to measure the concentrations of selected water-quality constituents, including calcium, magnesium, sodium, potassium, bicarbonate, nitrate, chloride, sulfate, fluoride, and silica. The concentrations of these constituents are present in many natural waters and will vary largely due to the extent of interactions between the water and surrounding rock material (Hem, 1985). The objective of the sampling program was to evaluate the spatial variations in the concentration of these water-quality constituents and verify ground-water flowpaths between areas of ground-water recharge and discharge to and from the Lower Aquifer of the Eastbank Aquifer system.
The source of recharge to the Lower Aquifer is the Columbia River. Concentrations of water-quality constituents in Columbia River water typically are low compared to concentrations in ground water of the Columbia Plateau region (Bortleson and Cox, 1986; Turney, 1986). As ground water moves away from recharge areas through relatively unweathered aquifer material such as the sediments that make up the Lower Aquifer, it accumulates solutes from the dissolution of rock and mineral fragments (Drever, 1988). Conversely, particulate matter such as bacteria that is present in river water will be reduced by filtration from passage through the aquifer material. Thus, ground water at locations along a flowpath downgradient from a recharge area can be expected to have increasing concentrations of dissolved constituents and decreasing numbers of live bacterial cells.
Ground-water samples were collected from one well that supplies the regional water system (RW3), one well that is used to irrigate Lincoln Rock State Park (LR2-E), two wells that supply the Eastbank Hatchery (CT3 and CT4); three wells that are used for dam operations or irrigation (SW11, SW13, and SW14), and one monitoring well (TH4; fig. 25A). A sample also was collected from the North Weir, which is assumed to represent ground-water seepage from the Lower Aquifer through the grout curtain of the subsurface cutoff wall. In addition, one sample of Columbia River water was collected near the site where the PUD collects continuous river water-level and water-temperature data, labeled RIV in figure 25A.
Concentrations of many of the analyzed constituents showed spatial patterns in the Lower Aquifer. Samples were collected nearly concurrently from wells SW13 and SW14 to establish a baseline of variability, which includes variability due to sampling, sample analysis, and localized spatial variability within the aquifer. Wells SW13 and SW14 are believed to have similar construction and are located within about 200 ft of each other. For many constituents (table 7), the laboratory analysis of samples from wells SW13 and SW14 were nearly identical. Larger variations of more than 2 percent were reported for dissolved oxygen, nitrate, and the bacterial enumeration; the relative percent difference computed for nitrate and dissolved oxygen was 13 and 16 percent, respectively, and about 60 percent for bacterial enumeration. Smaller variations with a relative percent difference of less than 2 percent were reported for concentrations of alkalinity, calcium, magnesium, bicarbonate, chloride, potassium, sodium, silica, fluoride, and sulfate. This level of variation commonly is reported for analysis of duplicate environmental samples. As a result, differences in constituent concentrations greater than those observed between the samples for wells SW13 and SW14 were considered indicative of spatial variations in water quality of the Lower Aquifer. Concentrations of nitrate and dissolved oxygen were not used to assess spatial variation.
Specific conductance is a measure of the ability of water to conduct an electric current and thus provides a general measure of the amount of dissolved matter in water. Specific conductance of the river sample was 127 µS/cm. A survey of 32 vertical profiles distributed throughout the Columbia River near the study area conducted on the same day the river was sampled showed that the specific conductance in the river varied by less than 3 µS/cm. Specific conductance in ground water from wells near the shoreline and from wells near the center of the Lower Aquifer was in the range of 135 to 148 µS/cm and 163 to 167 µS/cm, respectively. This indicates a pattern of lower concentrations of dissolved constituents near the river and larger concentrations near the center of the Lower Aquifer (fig. 25A).
A similar pattern also was observed for individual dissolved constituents. Concentrations of all dissolved constituents, except sulfate, were smallest in the river sample and largest in ground-water samples from nearer the center of the Lower Aquifer. The largest concentrations of dissolved constituents were generally measured in wells CT4 or RW3 with generally slightly smaller concentrations in LR2-E and the North Weir. Maps of the spatial distributions of specific conductance, potassium, silica, alkalinity, and chloride are shown in figures 25A through 25E. Silica (fig. 25C) and potassium (fig. 25B) show the most pronounced spatial variation, whereas spatial variation is more difficult to discern for sodium (fig. 25F). With few exceptions, this pattern was consistent among different constituents, including alkalinity (fig. 25D) and chloride (fig. 25E). For several constituents, such as sodium (fig. 25F), concentrations in ground water were larger than in surface water but did not show a consistent spatial pattern in relation to possible directions of ground-water flow. A generally northeast-southwest gradient of increasing concentrations of dissolved constituents in the direction of predevelopment ground-water flowpaths (fig. 11) was not observed. Instead, data generally indicate ground-water flowpaths from the western shoreline to the pumping centers in the CT and RW well fields.
Bacterial concentrations are largest in the river and lowest in wells closer to the center of the Lower Aquifer (fig. 25G). In the sample from the well in Lincoln Rock State Park (LR2-E), no viable bacterial cells were observed above the detection limit of 3 cells per milliliter. The occurrence of fewer live bacterial cells in ground-water samples obtained at locations more distant from the surface-water source is consistent with the filtering effect resulting from movement of ground water through an aquifer matrix. As shown in figure 25G, live bacterial concentrations increased along the central axis from wells LR2-E to CT3, which is inconsistent with a ground-water flowpath from the northeast to the southwest. Generally, the live bacterial concentrations indicate ground-water flowpaths from the western shoreline to near the center of the Lower Aquifer. In well CT3, however, live bacterial concentrations are larger than those in wells near the shoreline, which may be related to the large pumping rate of the well, its proximity to the river, and preferential flowpaths from the river to the well. This interpretation is supported by dissolved constituent concentrations in well CT3, which were consistently smaller than those found in wells to the north and were similar to the more dilute concentrations in nearby wells adjacent to the river.
The water-quality data indicate that the ground-water flowpaths that end in the CT well field predominantly originate along the shoreline west and southwest from the pumping wells. If the northeast-to-southwest ground-water flowpath present during predevelopment conditions were predominant, then well CT3, which is the most southerly well that pumps at a high rate, should have had the largest concentrations of dissolved constituents and the smallest concentration of live bacterial cells. However, the water-quality results show nearly the opposite and a very short flowpath is indicated for well CT3. Conversely, the lack of substantial live bacterial concentrations and the occurrence of relatively larger dissolved constituent concentrations in well LR2-E indicate that the flowpath from the river to the well is longer and/or less recharge from the river passes through that area. Ground-water seepage through the grout curtain of the subsurface cutoff wall appears to be a combination of water with both long and short flowpaths and is consistent with a collector drain integrating discharge from the Lower Aquifer along a 2,000-foot-long interface.
U.S. Department of the Interior | U.S. Geological Survey
URL: http://pubs.usgs.gov/sir/2008/5071
Page Contact Information: Publications Team
Page Last Modified: Thursday, 10-Jan-2013 18:46:28 EST
