Published in Vadose Zone Journal 3:183-189 (2004)
© 2004 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
SPECIAL SECTION: UNDERSTANDING SUBSURFACE FLOW AND TRANSPORT PROCESSES AT THE IDAHO NATIONAL ENGINEERING & ENVIRONMENTAL LABORATORY (INEEL) SITE
Well Design to Reduce Barometric Pressure Effects on Water Level Data in Unconfined Aquifers
J. M. Hubbell*,a,
J. B. Sissona,
M. J. Nichollb and
R. G. Taylorc
a Idaho National Engineering and Environmental Laboratory, Geosciences Research Department, P.O. Box 1625, MS 2107, Idaho Falls, ID 83415
b Mining and Geological Engineering, University of Idaho, Moscow ID 83843
c Idaho Department of Environmental Quality, 1410 North Hilton, Boise, ID 83642
* Corresponding author (jmh{at}inel.gov).
Received 19 February 2003.
 |
ABSTRACT
|
|---|
Barometric pressure fluctuations may influence measured water levels in wells where thick vadose zones or low permeability materials overlie unconfined aquifers. Here, we present a well completion method designed to reduce the effects of barometric pressure fluctuations on measured water levels. This well configuration, called the isobaric well, seals the interior of the well from atmospheric pressure, and vents the reference side of the water level pressure transducer to the gas phase pressure above the water table. We tested the isobaric design on a well completed about 180 m below land surface in the Eastern Snake River Plain aquifer at the Idaho National Engineering and Environmental Laboratory. Water level data collected during 14 mo show more than an order of magnitude decrease in diurnal fluctuations when the well was operated in the isobaric mode. Decreasing the noise level allowed clear definition of water level trends that would otherwise have been at least partially obscured by barometric fluctuations. This well configuration allows direct monitoring of water level changes, without the need to rely on postprocessing to mitigate barometric influences.
Abbreviations: bls, below land surface • ESRP, Eastern Snake River Plain • INEEL, Idaho National Engineering and Environmental Laboratory
|
INTRODUCTION
|
|---|
TEMPORAL FLUCTUATIONS in barometric pressure can significantly complicate measurement of water level in unconfined aquifers, particularly where the vadose zone is thick or contains low permeability zones. First, water levels are commonly measured using differential pressure transducers referenced to barometric pressure at the wellhead. Second, unsealed observation wells provide a direct connection to atmospheric pressure changes, while the surrounding aquifer is partially buffered by the intervening vadose zone materials. Thus, water levels in the well may not be at equilibrium with the aquifer, leading to inaccurate measurements. Accurate water level measurements are particularly important at sites with closely spaced wells, small hydraulic gradients, high transmissivity, and/or low storage coefficients, all of which are issues of concern at the Idaho National Engineering and Environmental Laboratory (INEEL).
The effects of barometric fluctuations on water table measurements have been documented by numerous authors (e.g., Jacob, 1940; Weeks, 1979; Rojstaczer, 1988a; Rasmussen and Crawford, 1997; Hare and Morse, 1997). Low air permeability materials in the vadose zone can restrict communication to the atmosphere. As a result, gas pressures in the vadose zone will change more slowly and to a lesser extent than barometric pressure. Conversely, gas pressure in an unsealed well will equilibrate almost instantaneously to changes in barometric pressure. Where gas pressure in the vadose zone exceeds atmospheric pressure, water will move from the aquifer into an unsealed well bore, creating an unnaturally high water level, and vice versa.
Unsealed wells penetrating deep unconfined aquifers frequently exhibit barometric induced fluctuations in water levels on the order of 6 cm d–1, and up to 30 cm over several months (Weeks, 1979; Rasmussen and Crawford, 1997). Data collected at the INEEL (Wylie and Hubbell, 1994) and the Savannah River Site (Bennett et al., 1997) show barometric pressure induced fluctuations at depths to water of 180 m and 15 to 30 m, respectively. Similar water potential fluctuations are observed from advanced tensiometers measuring sediments and basalt in the vadose zone (Sisson and Hubbell, 1999). At sites with closely spaced wells and/or small gradients, barometric induced fluctuations in measured water level may significantly impact estimates for the magnitude and direction of hydraulic gradient. Spane (2002) presented several examples representative of the Hanford Site, where barometric induced fluctuations could lead to estimation errors of up to 180° for groundwater flow direction, and a factor of four for the hydraulic gradient.
A number of authors have suggested numerical methods for correcting water level data by first estimating, then removing the influence of barometric pressure fluctuations (e.g., Clark, 1967; Weeks, 1979; Rojstaczer, 1988a, 1988b; Rojstaczer and Agnew, 1989; Furbish, 1991; Rasmussen and Crawford, 1997; Spane, 2002). To do so, one must have both an accurate record of barometric pressure at the wellhead and adequate models for the relevant transfer functions. However, most existing approaches employ a single constant (barometric efficiency) to describe system response to barometric pressure changes (see summary in Rasmussen and Crawford, 1997). This is a significant assumption, as it has been suggested that the barometric efficiency can vary seasonally and with depth (e.g., Rojstaczer, 1988a).
As an alternative to cleaning up the data in postprocessing, we present a well completion technique designed to reduce the impact of barometric pressure fluctuations on water level measurements in unconfined aquifers. We begin by presenting a conceptual model for an isobaric well (Hubbell and Sisson, 1999) design that isolates the measurement system from atmospheric influences along the well bore, while using the gas pressure above the aquifer as the reference pressure. Then, we describe a field implementation of our conceptual model. Field data collected during a 14-mo trial clearly demonstrates that the isobaric design provides greatly improved water level data and can allow additional insight that would not be possible with a conventional measurement approach. We then conclude by summarizing our results and suggesting possible extensions to our approach.
|
DESIGN APPROACH
|
|---|
Groundwater flow is driven by changes in hydraulic head, which represents potential energy per unit weight of fluid (units of length). Thermal, chemical, and inertial energy are expected to have a negligible influence on most groundwater flows; thus, total hydraulic head (ht) is commonly taken to be the resultant of gravitational potential energy and fluid pressure (P):
 | [1] |
where z is the elevation above datum of the measurement point, and 
w is the unit weight of water (w = 
wg, where w is the density of water and g is gravity). In practice, ht is commonly taken to be the water level elevation in an uncapped well and is monitored by submerging a differential pressure transducer to a known elevation (z) within the well (Fig. 1a)
. For this situation, pressure measured by the transducer (Pobs) will be
 | [2] |
where the first term in the parentheses represents fluid pressure produced by the water column above the transducer location, hp is the height of that column, Pbar represents barometric pressure acting on the free water surface within the well, and Pref is the transducer reference pressure. Assuming that barometric pressure at the well head is equivalent to barometric pressure within the well bore, venting the transducer reference at the well head (Fig. 1a) leads to a true estimate of hp, and hence ht (hp = P/w). However, unless fluids within the well are at equilibrium with the surrounding aquifer, ht measured in the well will not accurately reflect conditions in the aquifer.
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 1. Pressures relevant to measurement of water level in (a) a conventional (nonisobaric) well and in (b) an isobaric well. In both cases, water level is measured by using a pressure transducer to measure water pressure (Pobs) at a given elevation within the well. In the nonisobaric well, water levels in the well respond to changes in atmospheric pressure (Pbar) and the pressure transducer is referenced (Pref) to atmospheric pressure. The isobaric well is sealed from the atmosphere, and the pressure transducer is referenced (Pref) to gas pressure above the water table (Psg).
|
|
In shallow vadose zones containing materials with high air permeability, gas phase pressure above the water table (Psg) may equilibrate rapidly with changes in barometric pressure. In that case, the well configuration shown in Fig. 1a will produce water level measurements that accurately reflect conditions in the surrounding aquifer. However, if the vadose zone is thick, or has zones of low air permeability that restrict pressure transmittal, Psg can differ significantly from Pbar (Weeks, 1979). As a result, water levels in an unsealed well (Fig. 1a) will show an exaggerated response to barometric fluctuations that does not reflect actual changes in ht within the surrounding aquifer. The magnitude of the divergence will depend on both pressure transmittal through the vadose zone, and water transfer between the aquifer and well bore. Because these transfer functions are unknown, simple correction for barometric fluctuations measured at the well head cannot fully correct measured data. We also note that manual water level measurements using a steel tape or e-line, are equivalent to those collected by a conventional data logger–pressure transducer and will also be affected by barometric pressure fluctuations.
Assuming a screened interval that extends across the water table, the aforementioned issues can be ameliorated by implementing an isobaric well design (Fig. 1b). By sealing the well bore against atmospheric pressure, the well is only connected to the atmosphere through the portion of the screen above the water table. Therefore, gas pressure within the well bore will equilibrate to the surrounding media (Psg), and water levels within the well will accurately reflect conditions in the surrounding aquifer. For a sealed well screened across the water table, Eq. [2] becomes:
 | [3] |
Finally, we eliminate barometric fluctuations from Eq. [3] by referencing the pressure transducer to Psg immediately above the water table (Fig. 1b). Since Psg equals Pref then:
 | [4] |
thus removing barometric fluctuations from Pobs.
|
FIELD IMPLEMENTATION
|
|---|
The approach outlined in the previous section was implemented in an observation well penetrating the Eastern Snake River Plain (ESRP) aquifer (Fig. 2)
. The ESRP aquifer is a highly permeable basalt sequence in southeast Idaho that extends about 270 km along the principal flow direction (northeast to southwest), and varies in width from approximately 50 to 100 km. The Eastern Snake River Basalts consist of numerous discrete flows intercalcated with thin fluvial and aeolian sediments. The basalt matrix is tight, as are many of the sedimentary interbeds; thus, flow is primarily restricted to fractures within the basalt and rubble zones separating the flows. The active portion of the aquifer is thought to be about 100 to 200 m thick and underlain by rhyolite and tuffaceous sediments (Pierce and Morgan, 1992). Depths to water in the ESRP aquifer range from a few meters below land surface (bls) in the upland recharge areas to more than 300 m bls in the center of the Snake River Plain. Flow is predominantly horizontal; hydraulic gradients range between 0.5 and 16.4 m km–1, with an average of about 2 m km–1 (Lindholm, 1986). Transmissivity of the ESRP aquifer is highly variable, generally ranging from 102 to 105 m2 d–1, and may locally exceed 107 m2 d–1 (Whitehead, 1992). Estimated storage coefficients vary from 10–5 to 10–1, suggesting that the ESRP aquifer acts as a confined system at some locations, and unconfined at others.
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 2. Location of the field implementation site. USGS-118 is located about 30 m south of the Subsurface Disposal Area at the Radioactive Waste Management Complex (RWMC), a low-level radioactive waste disposal site in the southwest quadrant of the INEEL.
|
|
The observation well used to evaluate our approach, USGS-118, is located just outside the Radioactive Waste Management Complex at the INEEL (Fig. 2). Transmissivity of the ESRP aquifer at this site is about 100 m2 d–1, and the hydraulic gradient is about 0.5 m km–1 (Wylie and Hubbell, 1994). Significant local sources of recharge are from losses in the ephemeral Big Lost River channel and the spreading areas (infiltration basins) used to control local flooding (Fig. 2).
The drill log for USGS-118 indicates sedimentary materials from land surface to 6.1 m bls, and basalt with thin sedimentary interbeds to the bottom of the well at 190 m bls (Fig. 3)
. The 5-cm-diam. observation well was screened across the water table from 179.9 to 185.3 m bls, adjacent to basalt. Above the water table, four vapor extraction ports were added to the outside of the casing for monitoring the basalt (Hubbell et al., 1998). The borehole was packed with coarse sand (2–5 mm diameter) over intervals adjacent to the screen and the vapor ports, with the remainder of the annular space sealed using bentonite and neat cement (Fig. 3). The well cap was designed and built to isolate the interior of the well from atmospheric pressure fluctuations, while allowing electrical leads to pass through (Fig. 4)
.
View larger version (37K):
[in this window]
[in a new window]
|
Fig. 3. Stratigraphy and well completion for Well USGS-118, which was used to evaluate the isobaric technique. The 5-cm casing was slotted with 1-mm saw cuts over the screened interval, and four vapor ports were added to the outside of the casing. Vapor ports were fabricated from lengths of 9.5-mm-diam. tubing. The bottom of each tube was crimped to prevent backfill from entering, and the next 1.5 m was perforated with 2-mm-diam. holes. Individual vapor ports were attached to the outside of the casing, then routed to the surface. The annular region was backfilled with coarse sand around the screened interval and vapor ports, the remainder of the borehole was sealed with bentonite and neat cement (Hubbell et al., 1998).
|
|
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 4. The airtight PVC well cap. The cap is sealed to the casing when the rubber gasket on the lower exterior of the cap is forced out against the 5-cm well casing by tightening the wing-nut. The one-holed, tapered rubber stopper is forced down into the tapered shaft by tightening the cap nut, thus sealing around the transducer leads. A stock rubber gasket from a mechanical test plug was used in this test.
|
|
A 70-kPa (10 psig) pressure transducer (Druck model PTX-161/D) and a Hermit 1000C datalogger (In-Situ Inc., Laramie, Wyoming) were used for this investigation. The published accuracy at constant temperature for this combination is ±0.05% full-scale range or 0.23 cm. Before fixing the depth of the transducer at 183 m bls, we raised and lowered it within the borehole to verify proper operation. Normally, the pressure sensor is referenced to barometric pressure at the datalogger (gauge pressure); however, we implemented the isobaric design by referencing the transducer to the basalt interval at 166.7 to 168.2 m bls (lowermost vapor port in Well USGS–118). The well was sealed from the atmosphere (Fig. 4), and all other external vents on the datalogger were sealed.
We chose the vapor port as a reference for two reasons. It was simple to access the gas pressure above the water table with the existing vapor port and this design allowed use of an off-the-shelf pressure transducer. As an additional benefit, this particular design allowed us to compare isobaric data to that collected under nonisobaric conditions. The downside of our choice is that the reference point is located about 12 m above the water table, which is not optimal. Noting that there are numerous ways to implement an isobaric well, it would seem natural to extend the sand pack 1 to 2 m above the screened interval and insert a vapor port.
|
RESULTS
|
|---|
Data were collected from Well USGS-118 for 14 mo (March 1997–May 1998) and are reported as height of water above the transducer (Fig. 5)
. The water table remained within the screened interval throughout this test period. For the first month (March–April 1997), data were obtained with the transducer referenced to barometric pressure and the well unsealed (conventional approach). During this first month, water level data showed high frequency variation across a range of about 25 cm (2.25–2.5 m above the transducer), suggesting a noise signal approximately two orders of magnitude larger than the transducer resolution (0.23 cm). The significant fluctuations observed during this period are primarily due to barometric pressure changes at land surface relative to gas phase pressure above the water table and are of sufficient magnitude to mask the target information, which is water level changes resulting from recharge or discharge. Less significant sources for the observed fluctuations would include temperature changes, and the effects of fluids moving into and out of the well bore (delayed response).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 5. Water level above the pressure transducer located at 183 m bls in Well USGS-118 measured for a 14-mo period. Data was collected from 25 Mar. 1997 to 29 May 1998 at 60-min intervals. Flow in the nearby Big Lost River (see Fig. 2) is shown for comparison. Boxes marked A and B are expanded in Fig. 6 to illustrated the difference between isobaric (Box A) and non-isobaric (Box B) data.
|
|
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 6. Comparison of water levels measured in Well USGS-118 for two 7-d periods. Barometric pressure in meters of water is shown for comparison. Data collected under isobaric conditions from 3 June 1997 to 10 June 1997 (Box A in Fig. 5). Linear regression of the data gives a slope of 0.0025 m h–1. Data collected from 26 June 1997 to 3 July 1997 (Box B in Fig. 5) begins and ends with isobaric data, while data in the center of the 7-d period (27 June 1997 to 3 July 1997) was collected under nonisobaric conditions. The isobaric portion of the data shows a slope similar to the earlier data (0.0026 m d–1), while the nonisobaric portion actually shows a declining trend (–0.0143 m d–1) of much larger magnitude.
|
|
On 1 May 1997, the isobaric well configuration was implemented. The well cap (Fig. 4) was sealed, and the transducer reference pressure was switched from barometric to gas phase pressure at the lowest vapor port (Fig. 3). Rapid water level fluctuations observed with the conventional approach were moderated in <1 h following conversion to the isobaric configuration. With two exceptions, data for the remainder of the test show a very small high frequency signal superimposed on a long-term increase in water level that peaked on 24 Oct. 1997 and then declined for the remainder of the test. Within this long-term rise–decline cycle there were several smaller peaks, all of which were of considerably smaller magnitude than the fluctuations observed during nonisobaric operation. The low noise data produced during isobaric operation show that water levels in USGS-118 are apparently correlated to flow in the Big Lost River, and subsequent recharge to the spreading areas located 2.4 km to the west (Fig. 2), with a time lag on the order of 5 mo. Noise levels in nonisobaric data would have obscured this apparent correlation.
Twice during the data collection period (27 June–7 July 1997 and 1–20 Sept. 1997) the transducer was referenced to barometric pressure as a test of the isobaric operation; in both cases, the well remained sealed. To compare measurements made in isobaric mode with those made under nonisobaric conditions, we focus on data from two 7-d-long segments marked A and B on Fig. 5. Nonisobaric data on the rising limb of the water level curve was collected for the time segment marked B (26 June–3 July 1997). For comparison, we arbitrarily chose segment A to cover a 7-d period (3–10 June 1997) of isobaric data on the rising limb; both data sets are shown in Fig. 6
. Fluctuations about a best-fit linear trend through each data set show a range of ± 0.11 m in the nonisobaric mode (Fig. 6b), and ± 0.0012 m in the isobaric mode, which is close to the instrument's precision, and an improvement of nearly two orders of magnitude.
Interval B (Fig. 6) has isobaric measurements at the start and finish of the data and conventional (nonisobaric) water level measurements in the interval between. The time interval for barometric effects to nearly disappear from measured data appears to be <1 h (Fig. 6). This response time can be expected to vary from site to site, depending on the construction of the well, the transducer–datalogger–well configuration, and the geologic properties of the surrounding formation. Linear regression of the isobaric data at the start and finish of Segment B (dashed line) shows a slope (0.0026 m d–1) nearly identical to that for segment A (0.0025 m d–1), while the nonisobaric portion of Segment B shows a negative slope of much greater magnitude (–0.0143 m d–1). We also note that confidence in the linear fit is much poorer for the nonisobaric portion of Segment B (R2 = 0.31) than for the isobaric data in Segment A (R2 = 0.95). These results suggest that conventional measuring techniques (nonisobaric) could not be used to correctly determine the water level trend for a 7-d period at this location.
|
CONCLUSION
|
|---|
A well completion method for reducing the effects of barometric pressure fluctuations on measured water levels was introduced and evaluated. Sealing the well from atmospheric pressure and referencing the water level pressure transducer to the air pressure immediately above the water table reduced the effects of barometric pressure fluctuations by more than an order of magnitude, significantly improving our ability to detect water level trends at the study site. This technique can be applied to many existing wells, and new wells can be constructed with little or no additional cost. Gas pressure above the water table (Psg) can be accessed by screening the well across the water table, or by installing a gas sampling port at the appropriate depth. It is important to note that this approach allows direct measurement of water levels at sites with deep vadose zones without having to correct data for barometric effects using unknown transfer functions.
As presented here, the isobaric well configuration is designed to provide data on changes in water level at a given location. Using this data to calculate hydraulic gradients and flow direction may require slight modification. Because gas pressure above the water table (Psg) contributes to total hydraulic head (ht), it must be factored into calculation of hydraulic gradients, unless of course Psg is the same at all measurement locations. The isobaric technique can also be used to obtain better estimates of aquifer hydraulic properties and hydraulic boundary definition from aquifer tests. Because the response from pumping will not be masked by barometric induced water level fluctuations, the isobaric well completion may allow a decrease in the pumping rate, or the placement of observation wells further from the pumping well.
|
ACKNOWLEDGMENTS
|
|---|
Work was supported by the Laboratory Directed Research and Development (LDRD) program and EM-50 of the U.S. Department of Energy, Assistant Secretary of Environmental Management, under DOE Idaho Operations Office Contract DE-AC07-99ID13727 at the INEEL. M.J. Nicholl would like to acknowledge support from the U.S. Department of Energy through the Basic Energy Sciences Geoscience Research Program under contract number DE-FG03-01ER15122 and the Environmental Management Science Program under contract DE-FG07-02ER63499. The authors thank Indrek Porro, Tom Stoops, and Annette Schafer who provided insightful comments on this document. Mention of trademark propriety products are for the benefit of the readers and do not constitute an endorsement for the products by the Department of Energy to the exclusion of other products that may also be suitable.
|
REFERENCES
|
|---|
- Bennett, C.B., J.K. Reed, and T.C. Rasmussen. 1997. Identifying and correcting hydraulic gradient calculation errors at the Savannah River Site. p. 64. In Abstract Book for the Ninth National Technology Information Exchange Workshop. 26–28 Aug. 1997. INEEL, Idaho Falls, ID.
- Clark, W.E. 1967. Computing the barometric efficiency of a well. J. Hydraul. Div. Am. Soc. of Civil Engineers. 93 (HY4):93–98.
- Furbish, D.J. 1991. The response of water level in a well to a time series of atmospheric loading under confined conditions. Water Resour. Res. 27:557–568.
- Hare, P.W., and R.E. Morse. 1997. Water-level fluctuations due to barometric pressure changes in an isolated portion of an unconfined aquifer. Ground Water 35:667–671.
- Hubbell, J.M., T.R. Wood, B. Higgs, A.H. Wylie, and D.L. McElroy. 1998. Design, installation and uses of combination ground water and gas sampling wells. Ground Water Monit. Rem. 19:1–7.
- Hubbell, J.M., and J.B. Sisson. 1999. Isobaric groundwater well. U.S. Patent 5 969 242. Date issued: 19 October.
- Jacob, C.E. 1940. On the flow of water in an elastic artesian aquifer. Trans. Am. Geophys. Union 21:574–586.
- Lindholm, G.F. 1986. Snake River Plain Regional Aquifer systems study. p. 88–106. In Sun (ed.) Regional aquifer-system analysis program of the U.S. Geological Survey—Summary of projects, 1974–84. U.S. Geol. Survey Circular 1002. U.S. Geological Survey, Reston. VA.
- Pierce, K.L., and L.A. Morgan. 1992. The track of the Yellowstone Hot Spot: Volcanism, faulting and uplift. p. 1–54. In P.K. Link et al. (ed.) Regional geology of eastern Idaho and western Wyoming. Geological Society of America Memoir 179. GSA, Boulder, CO.
- Rasmussen, T.C., and L.A. Crawford. 1997. Identifying and removing barometric pressure effects in confined and unconfined aquifers. Ground Water 35:502–511.[ISI]
- Rojstaczer, S. 1988a. Determination of fluid flow properties from the response of water levels in wells to atmospheric loading. Water Resour. Res. 24:1927–1938.
- Rojstaczer, S. 1988b. Intermediate period response of water levels in wells to crustal strain: Sensitivity and noise level. J. Geophys. Res. 93(B11):13619–13634.
- Rojstaczer, S., and D.C. Agnew. 1989. The influence of formation material properties on the response of water levels in wells to earth tides and atmospheric loading. J. Geophys. Res. 94(B9):12403–12411.
- Sisson, J.B., and J.M. Hubbell. 1999. Water potential to depths of 30 meters in fractured basalt and sedimentary interbeds. p. 855–865. In M.Th. van Genuchten et al. (ed.) Proceedings of the International Workshop on Characterization and Measurement of Hydraulic Properties of Unsaturated Porous Media. U.S. Salinity Laboratory, Riverside. CA.
- Spane, F.A. 2002. Considering barometric pressure in groundwater flow investigations. Water Resour. Res. 38:14–18.
- Weeks, E.P. 1979. Barometric fluctuations in wells tapping deep unconfined aquifers. Water Resour. Res. 15:1167–1176.
- Whitehead, R.L. 1992. Geohydrologic framework for the Snake River Plain regional aquifer system, Idaho and eastern Oregon. U.S. Geologic Survey Professional paper 1408-B. U.S. Geological Survey, Reston. VA.
- Wylie, A.H., and J.M. Hubbell. 1994. Aquifer testing of well M1S, M3S, M4D, M6S M7S and M10S at the Radioactive Waste Management Complex. Engineering Design File ER-WAG-7. EG&G Idaho Inc., Idaho Falls, ID.
TOP
ABSTRACT
INTRODUCTION
