Scientific Investigations Report 2007–5216
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5216
The seepage meter allows direct measurement of seepage flux across the sediment–water interface. It consists of a bottomless cylinder formed from an inverted drum or bucket connected to a collection bag by a length of tubing. The device is pushed into the bed of a lake or stream, and a collection bag with a known volume of water is attached. The collection bag is then removed after a period of elapsed time, and the rate of vertical ground-water flux through the area enclosed by the seepage meter is calculated from the increase or decrease in the initial volume of water, the length of time elapsed, and the area of the seepage meter, yielding flux rates in units of length/time. An increase in the initial volume indicates a positive vertical flux rate (ground water to surface water), and a decrease in initial volume indicates a negative vertical flux rate (surface water to ground water).
The seepage meter was initially developed to measure losses from irrigation canals (Israelsen and Reeve, 1944), and in the mid-1970s, the design was improved and the use was expanded to measure ground-water discharge into lakes (Lee, 1977; Lee and Cherry, 1978; John and Lock, 1977; Connor and Belanger, 1981; Erickson, 1981; Woessner and Sullivan, 1984; Isiorho and Matisoff, 1990; Shaw and Prepas, 1990b; Lesack, 1995; Rosenberry, 2000; Sebestyen and Schneider, 2001). Because seepage meters provide a quick and simple method for gathering information on the direction, rate, and variability of seepage flux across the sediment–water interface, their use has been expanded to environments other than lakes. Vertical seepage rates have been measured in wetlands (Choi and Harvey, 2000), estuaries (Lee, 1977; Lock and John, 1978; Zimmerman and others, 1985; Yelverton and Hackney, 1986; Boyle, 1994; Linderfelt and Turner, 2001) and nearshore ocean margins (Cable and others, 1997; Shinn and others, 2002; Taniguchi, 2002; Chanton and others, 2003). Seepage meters have also been used to determine water budgets (Fellows and Brezonik, 1980) or to obtain samples for chemical analysis (Lee, 1977; Downing and Perterka, 1978; Brock and others, 1982; Belanger and Mikutel, 1985; Shaw and others, 1990).
A growing interest in investigating the rates of exchange between streams and ground water has led to the use of seepage meters in stream channels (Lee and Hynes, 1977; Connor and Belanger, 1981; McBride, 1987; Libelo and MacIntyre, 1994; Blanchefield and Ridgeway, 1996; Jackman and others, 1997; Cey and others, 1998; Fryar and others 2000; Dumouchelle, 2001; Landon and others, 2001; Murdoch and Kelly, 2003). However, the data obtained are often highly variable because the original design and application were intended for lake and estuary environments, where issues of current and scour are generally negligible. As discussed in more detail later in this section, significant streamflow rates may create hydraulic pressure on the measure bag, creating erroneous gains or losses in the bag volume over time. Scour may lead to a breach in the hydraulic seal around the seepage meter. Flume and laboratory studies show that much of this variation is due to the effects of flow across the seepage meter collection bag, which alters the hydraulic head within the meter and induces seepage flow (Libelo and MacIntyre, 1994). Their study discovered that the induced seepage flow can be significantly reduced by isolating the seepage meter collection bag from the streamflow.
Concerns over the effects that size, thickness, and initial conditions of the attached collection bag have on measured seepage rates have prompted investigations in field and laboratory settings. Shaw and Prepas (1989) found that an anomalous, short-term influx of water into seepage meters occurred immediately after connecting the collection bag to the seepage meter, but showed that the anomaly could be effectively eliminated by attaching prefilled collection bags with a minimum of 1,000 mL before attaching to the seepage meter. The effects of bag type and meter size on seepage-meter measurements were evaluated in a laboratory setting by Isiorho and Meyer (1999). Their study found that there was no significant difference attributed to bag type, but found that smaller diameter seepage meters had a greater variance. Laboratory studies have also examined bag conductance, which is the ratio of the volumetric flow rate into the bag to the hydraulic head required to fill the bag (Murdoch and Kelly, 2003). Harvey and Lee (2000) found that the conductance of a bag formed from a thin compliant film is expected to be large and relatively constant until it is filled with enough water to cause stretching, at which point conductance will decrease. The conductance of a bag may also vary if the bag deforms in an irregular manner and will decrease if kinks or folds develop in the bag (Kelly, 2001). As a collection bag opens and approaches its manufactured shape, a gradual decrease in conductance of the bag with increasing volume was observed by Schincariol and McNeil (2002). Similar results were observed by Shaw and Prepas (1989) and Blanchfield and Ridgeway (1996).
In field settings, slow seepage rates and the relatively small area measured also present problems. Very slow seepage rates may require a meter to be in place for several days. The problem of measurement area is of concern because most researchers and watershed managers are interested in seepage processes on a scale of hundreds to thousands of square meters or more, and most seepage meters typically integrate vertical flux over an area of approximately 0.25 m2 or less (Rosenberry, 2005). Rosenberry addressed these issues in a low-permeability zone by connecting multiple seepage meter cylinders together to a single collection bag to increase the area represented by each measurement, thus integrating spatial heterogeneity over a larger area and reducing the time required to collect a measurable change in volume.
Some investigators (Erickson, 1981; Brock and others, 1982; Woessner and Sullivan, 1984; Shaw and Prepas, 1990a; Blanchfield and Ridgeway, 1996; Harvey and Lee, 2000) using seepage meters have expressed guarded concerns that the performance of the meter itself may cause variability that is unrelated to, and can obscure, the natural processes that they are trying to characterize (Schincariol and McNeil, 2002). Erickson (1981) stated that seepage meters disturb the flow field in which they are installed, resulting in consistently lower measured seepage rates. This disturbance is apparently related to frictional resistance along the internal boundaries of the meter. Frictional resistance is inherent to some extent in all seepage meter designs, and as a result, several laboratory studies have come up with a seepage meter coefficient that is used to convert measured seepage rates to true values. Coefficients in the literature range from 1.1 to 1.7 (Erickson, 1981; Cherkauer and McBride, 1988; Asbury, 1990; Belanger and Montgomery, 1992). Many of the less efficient meter designs require larger coefficients primarily because they use small-diameter tubing to connect the bag to the seepage cylinder (Rosenberry, 2005). The use of large-diameter plumbing greatly reduces loss of efficiency, resulting in a smaller correction coefficient (Fellows and Brezonik, 1980; Rosenberry, 2005).
Murdoch and Kelly (2003) developed a theoretical analysis to evaluate the extent to which bag conductance and velocity head may affect flux measurements by a seepage meter. Their analysis showed that bag conductance, radius of the seepage meter, and hydraulic conductivity of the streambed can be combined to give a dimensionless term that characterizes seepage meter performance. Some seepage meters use electronic flowmeters (Paulsen and others, 2001; Taniguchi and Fukuo, 1993; Rosenberry and Morin, 2004) to eliminate problems encountered with the collection bag. Although these devices show promise, their availability and expense make their use limited. This study focused on the conventional seepage meter that is both inexpensive and easily fabricated.
The purpose of using seepage meters to directly measure seepage rates was to gain an overall understanding of the direction, rate, and variability of seepage rates within the study area. The seepage meters used in this study incorporated suggestions made by several investigators (Libelo and MacIntyre, 1994; Shaw and Prepas, 1989; Kelly, 2001; Murdoch and Kelly, 2003) in an attempt to minimize variability in measured vertical seepage rates. The performance of the collection bags used in this study was measured in a laboratory test tank, and improvements were made to seepage meters and collection bags as a result of laboratory test results.
Flux rates across the sediment–water interface were measured directly using twelve drum-style (Lee, 1977) seepage meters. Two sizes of seepage meters were used: the larger meters were 2,500 cm2 in cross-sectional area and 26.5-cm deep, and the smaller meters were 620 cm2 in cross-sectional area and 13.5-cm deep. The field deployment methods for both sizes of meters was to carefully push each meter into the riverbed sediment, leaving approximately 5 cm of the top of each meter above the streambed bottom. The deployment array of the seepage meters placed in the streambed involved pairing a large and small seepage meter spaced approximately 1.5 m apart. The purpose of pairing the meters was to compare flux rates between seepage meter pairs and observe variability in measured rates for the same size meter at different parts of the array over consecutive measurements. Two deployment arrays were used throughout the study. The first deployment array involved placing the six paired meters perpendicular to flow across two transects that were separated by a distance of approximately 10 m, and the second deployment array placed the paired meters across two transects separated by a distance of approximately 100 m (fig. 4). The seepage-meter pairs will be referred to as A-1, B-2, C-3, and so on.
The meters were then left undisturbed for 12 to 24 hours to allow trapped air to escape, as well as to permit riverbed sediments to return to equilibrium after being disturbed during installation. Once deployed, the meters were left in place for the duration of the measuring event. Following the equilibration period, the collection bags were prefilled with 500–1,000 mL of water, placed in housing units to protect the bag from the flow of the river, and connected to the seepage meters. The collection bags and housing units were connected to the seepage meters using a 61-cm length of 0.64-cm-i.d. (inside diameter) vinyl tubing. The housing unit for the large, seepage-meter bag was a perforated plastic storage box that was secured to the top of the seepage meter with a bungee cord. The small seepage meter used the half of a drum as the housing unit for the collection bag and was simply pushed into the sediment alongside the seepage meter (figs. 5 and 6).
Careful attention was given to avoid disturbing the sediments around each of the paired meters, and the attachment and retrieval of collection bags was performed by snorkeling to and from each meter to eliminate any contact with the riverbed bottom. Because of the slow flux rates encountered, the collection bags were retrieved and redeployed approximately every 12–24 hours over a few days. The change in volume was either measured using a graduated cylinder or weighed with a scale, and the elapsed time was recorded. The change in volume (mL/min) over the elapsed time (days) was divided by the cross-sectional area of the seepage meter (cm2) to obtain vertical flux rates in cm/day.
Three types of collection bags were used in the study. The collection bag types used were an 1,800-mL Sun Shower Solar Bag, a 2,000-mL medical urine collection bag, and a 2,000-mL Void-Fill packaging bag. These collection bags will be referred to as shower bag, medical bag, and packaging bag, respectively. Initial field measurements used only the shower and medical collection bags attached to the large and small seepage meters, respectively. However, the results of laboratory test runs prompted exclusive use of the packaging collection bag in seepage rate measurements made in the latter part of the study.
A total of six sampling events were made over a period of 10 months beginning in December 2003 and ending in September 2004. The planned period for each field measurement was initially intended to be one week. However, stream scour around the seepage meters resulted in only two or three repeat measurements taken over 48–72 hours. Because improvements were made to the seepage meter and collection bags over the period of the study, the objective and methods of each field visit are discussed individually and are listed in table 2.
Seepage meters did not work because of bed load movement in the upper 0.5 m of the Merced River. Observations of the streambed during site visits revealed that the stream bottom is continually moving—small dunes would develop and disappear several hours later. The problem was further compounded by the slow seepage rates, which required the seepage meters to be in place for several days and be subjected to the moving streambed, resulting in scour at the base of the meter(s) (fig. 7). Scoured-out seepage meters took in river water and filled the collection bag to maximum capacity, resulting in erroneously high seepage rates. In addition, the housing unit for the small seepage meter was not as effective as the housing unit for the large seepage meter. The housing units would scour out, become dislodged, and expose the collection bag to the flow of the river, thereby inducing volumetric flow into the collection bag. In other cases, the housing unit would become buried under the streambed sediments resulting in complete loss of initial volume.
The original intent of pairing the seepage meters was to compare flux rates between the two sizes. In most cases, the flux direction was opposing, and for pairs in which the flux direction was the same for both sizes, the difference in rates often was too great to provide meaningful results. In addition, variability existed in the measured flux rates between the same size meters over consecutive measurement periods. Low flux rates (<3 cm/day) and a moving streambed resulted in scour or burial of seepage meters. As a result, only two consecutive 24-hour measurement periods could be made for each seepage meter. Figures 8,9, and 10 depict the variability in measurements described above for each meter over the six sampling events. The Appendix lists the results of each of the six sampling events.
The smaller seepage meters (numbered seepage meters in figs. 8,9, and 10) showed greater variability over consecutive measurements than the larger seepage meters, consistent with the findings of Isiorho and Meyer (1999). The results of the January and February 2004 (sampling events 3 and 4) gave inconclusive results for comparisons made between measured rates using shower and medical collection bags and comparison of those rates with measured rates using the packaging bags (Appendix). However, repeat measurements made with the shower bag attached to the large seepage meter gave consistent results. During the January 2004 visit, seepage meter B measured rates of 0.21 and 0.24 cm/day in two repeat measurements, and seepage meter C measured the same seepage rate, 0.15 cm/day, over a 48-hour period. The same pattern resulted for seepage meter A during the February 2004 visit, which measured a flux rate of 0.20 and 0.17 cm/day for consecutive measurements. The July and September 2004 site visits (sampling events 5 and 6) used only the thin-walled packaging bags attached to both meters, and inconsistencies in direction and magnitude of measured rates indicated no apparent pattern in seepage rates.
It was unclear whether the variability observed between the two types of seepage meters was a result of spatial variability in the hyporheic flowpaths at the sediment–water interface, a result of scour and burial problems encountered with the seepage meters, or unexplainable factors affecting the performance of the seepage meters and (or) collection bags. Because only two or three repeat measurements could be made at each of the field visits before the meters were scoured out, the seepage rates of pairs giving similar estimates are difficult to accept. More repeat measurements should be conducted before accepting the estimates of flux using this method. In addition, measurements should be made every 4–6 hours for several days to examine the response of seepage rates to diurnal variations in stream temperature and discharge resulting from evapotranspiration.
Evapotranspiration losses attributed to stream evaporation, stream-bank evaporation, and transpiration from stream channel vegetation create diurnal discharge patterns that are characterized by decreasing stream discharge during the day, with minimum discharge generally occurring in the afternoon (Taylor and Nickle, 1936; Troxell, 1936a,b). The decrease in discharge results in increased temperature variations downstream. The traditional assumption that reduced afternoon discharge is due entirely to evapotranspiration losses (Penman, 1963; Wisler and Brater, 1959) does not consider the importance of diurnal variations in stream temperature influencing seepage losses. A study conducted by Constantz and others (1994) demonstrates that for losing reaches with significant diurnal variations in stream temperature, the effect of stream temperature on streambed seepage is a major factor contributing to reduced streamflows. Studies using applied methods should incorporate quantitative analysis of diurnal stream–ground water interaction.
Overall, the direct measurement technique as applied to this study resulted in inconclusive flux results. The moving streambed bottom seemed to be the limiting factor in applying this simple technique and overcame any other attempts to limit variability caused by the measurement technique. The seepage meters seemed to fail in this relatively high energy stream with its mobile bed.
The performance of the three types of collection bags used in this study was tested in a laboratory seepage tank to examine the effects of bag thickness on measured rates of vertical flux. A cylindrical test tank with an inside diameter and height of 152 cm contained a 91-cm thick layer (1.67 m3) of medium sand placed over a 15-cm layer (0.28 m3) well-sorted, rounded gravel, with a medium size of 1.9 cm (fig. 11). Vertical flux was generated within the tank by introducing a constant headwater source to the bottom of the tank, and flow was measured by an in-line flow meter. A data logger recorded both the flow rate delivered to the bottom of the tank and pressure head measured by a pressure transducer located in the constant head water source tank. The data logger was programmed to control the pump to maintain a constant water level. An overflow opening was used to maintain water level within the constant head source tank to within 0.15 cm inside the tank.
Four seepage meters were placed in the test tank: two large and two small seepage meters. The system was allowed to equilibrate for 24 hours prior to the tests. The three types of collection bags used were the same as those used in the field (fig. 12). The wall thickness of each of the three collection bags was measured with vernier calipers at 0.41, 0.26, and 0.04 mm, respectively. The collection bags were prefilled with a known volume of water and attached to the seepage meters using a 10-cm length of 0.64-cm-i.d. vinyl tubing. The seepage meters and connected bags were placed in the test tank for 60 minutes for each test run. Seepage meters of the same cross-sectional area were considered pairs for each test run.
Two separate sets of tests were conducted. The objective of the first set of tests (tests 1 through 8) was to compare the results of measured vertical seepage rates with the known vertical test tank seepage rates. For these tests, a packaging bag and a medical bag were connected to the small seepage meters, and a packaging bag and shower bag were connected to the large seepage meters. At the end of a test, the collection bags were disconnected, weighed, and recorded.
The second set of tests (tests 9–12) was conducted using only the packaging bags. The objective of these tests was to evaluate the ability of the collection bag to fill under various test scenarios. The vertical flux rate in the test tank was set to a known constant positive flux rate for the duration of all test runs conducted. The specifics for each of these test runs are presented in the Laboratory Results section.
A relative percent difference (RPD) was used to describe the variability between the measured vertical seepage rates and the known test tank vertical seepage rate for each of the 12 tests. The RPD was calculated as the difference between the measured and known seepage rates divided by the average of the two values and is expressed as a percentage. An RPD of less than 10 percent between the known and measured seepage rates was considered acceptable for the tests conducted in the test tank. Table 3 lists the resulting variability for tests 1–8 in which the performance of the three different bag types was compared with the known test tank vertical flux. The median RPD for the shower bag connected to a large seepage meter was 121 percent, and the median RPD for the medical bag connected to small seepage meters was 96.5 percent. The median RPD for the packaging bags connected to the large and small seepage meters was 13.7 and 7.6 percent, respectively. The results of tests 1 through 8 are depicted as box plots in figure 13.
The results from tests 1 through 8 indicate that the thin-walled packaging bags performed better than the thicker-walled collection bags under an applied constant head. The thin-walled bags appear to be more compliant than the thicker-walled bags and filled easily. The compliance of bags to fill under an applied hydraulic head is affected by the size, shape, and the membrane thickness of the bag. The hydraulic head required to cause flow into the collection bags is expected to remain relatively constant until the bags fill with enough water to cause stretching, at which point the volumetric flow into the bag will decrease. Furthermore, if the membrane thickness of the collection bag is too large, the volumetric flow into a bag may reach a point where it rapidly decreases or ceases because the hydraulic head is not enough to overcome the resistance of the thicker-walled bag. Additional hydraulic head would be necessary to overcome the resistance and continue filling the bag. The high RPD for both the medical and shower bagsis likely a result of the latter, as the applied constant head during the test runs remained relatively constant throughout each test run, and the collection bags never filled to maximum capacity in any of the test runs. The volumetric flow rate that the flow meter measured at the beginning and end of each test run was recorded, and an RPD was calculated. The RPD between the flow rates at the beginning and end of each test run ranged from 0 to 7.6 percent for all tests (table 4).
In tests 9 through 12, compliance of the thin-walled packaging bags was tested under four different test scenarios (table 5). The results of test 9 indicate that the empty collection bag would fill if the tubing was filled; however, the RPD was twice the acceptable median RPD. Test runs 10 and 11 compare the results starting with a partially filled collection bag and attaching the hose that is filled (test scenario 2) or empty (test scenario 3). In test scenario 3, it was unclear as to whether the applied constant head filled the tubing before flow into the bag occurred, or whether the initial volume that the bag contained filled the tubing to initiate flow into the collection bag. The median RPD for this test scenario was 43.1 percent. The collection bags under test scenario 2 gave the best results. The results indicate that a small initial volume, void of any air bubbles or kinks in the collection bag or tubing, resulted in an average RPD between measured and known flux rates of 6.4 and 5.4 percent for the large and small seepage meters, respectively.
The results of test run 12, in which empty tubing and an empty collection bag were attached to the seepage meters, indicate that the RPD between the known and measured vertical flux rate was much greater than the acceptable difference. The high RPD (median RPD of 90.3 percent) for this test scenario appears to be related to the energy required to open a bag that is initially empty. Other investigators (Erickson, 1981; Shaw and Prepas, 1990a,b; Belanger and Montgomery, 1992; Landon and others, 2001) have reported problems with seepage meter performance when using bags that were initially empty.
Overall, the laboratory results of the three bag types tested indicate that the thin compliant packaging bags performed better than the medical and shower bag under the applied laboratory conditions. In addition, results of test runs 9–12 indicate that the packaging bags performed best when started with an initial volume and filled tubing prior to attachment to seepage meters (test scenario 2). Results of each of the test scenarios summarized in table 5 are depicted in figure 14.
Laboratory results suggested that use of thin-walled collection bags would provide the best seepage meter results for obtaining streambed fluxes. Following laboratory tests, thin-walled bags were exclusively used as collection bags in the field. All bags were filled with approximately 500 or 1,000 mL of water. The bags were connected to seepage meters with filled tubing, and the water in the bags was void of any air bubbles.
U.S. Department of the Interior | U.S. Geological Survey
URL: https://pubs.usgs.gov/sir/2007/5216
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 20:00:31 EST
