Vertical distribution and temporal fluctuation of ground water contaminated with road salt downgradient from a six-lane highway completed in 1987 is being monitored by measuring the electrical conductivity of the aquifer with a borehole-induction logger. Induction logs run in wells 30 to 180 meters downgradient from the highway have indicated that the plume of water contaminated with road salt is being transported in a thin zone of aquifer within 6 meters of the water table. Induction logs also have indicated that conductivity of the contaminated ground water varies annually in response to winter road salting. Water-quality samples collected from clusters of 1.5-meter screened wells concurrently with induction logging confirm the limited vertical extent and the annual fluctuation of road-salt contamination. Induction-log data continue to indicate that the multiple well screens installed intersect the full vertical span of the plume contaminated with road salt.

A method has been developed to estimate the horizontal hydraulic conductivity of the aquifer by measuring traveltime of the road-salt plume between wells along the same ground-water-flow path. Hydraulic conductivities obtained by this method ranged from 46 to 162 meters per day. This range is consistent with results of other tests of hydraulic conductivity within this sand and gravel aquifer.

INTRODUCTION

Borehole induction logs have become an integral part of an investigation of the effectiveness of highway-drainage systems in preventing contamination of ground water by highway deicing salts (road salt). Four test sites were selected for monitoring road-salt contamination of ground water along a 5-kilometer section of State Route 25 in southeastern Massachusetts (figure 1). Each test site represents a different type of highway-drainage-system design to control highway runoff. Borehole induction logs have been used since 1988 to monitor the vertical distribution and temporal fluctuation of ground water cone contaminated by road salt applied to this six-lane highway completed in the summer of 1987.

The purpose of this paper is to describe how the use of borehole induction logs has been, and continues to be, an effective and efficient method for delineation of ground water contaminated with road salt in a sand and gravel aquifer. A method developed to estimate horizontal hydraulic conductivity from traveltimes of the road-salt plume by use of induction-log data also is described. Hydrogeological, geophysical, and water-quality data from only one of the four test sites, Site B, is presented to illustrate use of the induction logs. The hydrogeology and background water quality at Site B are similar to those at the other test sites. All hydraulic conductivities referred to in this paper are horizontal hydraulic conductivities.

DESCRIPTION OF TEST SITE

The test site is on a glacial outwash plain in a rural area of the town of Wareham, Mass. (figures 1 and 2). Split-spoon soil samples collected during well construction show that the test site is generally underlain by a 20- to 30-meter thick layer of fine to coarse sand and some gravel. Fine sand and some silt typically are present below this sand and gravel unit.

Altitude of the water table at the test site ranges from 6 m (meters) upgradient to 4.5 m downgradient from the highway, respectively (figure 2). These altitudes correspond to water-table depths below land surface of 11 m upgradient and 5 m downgradient from the highway. The water table, which is within the fine to coarse sand and gravel unit, forms a relatively planar surface compared to the undulating topography of the land surface. Thickness of the aquifer, defined by the vertical distance between the water table and the fine sand and silt unit, is about 17 m upgradient from the highway and 15 m downgradient from the highway. The water-table gradient fluctuates annually from about 0.004 to 0.006, and ground-water flow is southeasterly, nearly perpendicular to the highway. Based on results from a nearby test site where geologic materials are similar (LeBlanc, 1987), estimated horizontal hydraulic conductivity and effective porosity of the aquifer range from about 60 to 110 m/d (meters per day) and 0.35 to 0.40, respectively. Horizontal hydraulic conductivities of 35 to 48 m/d were calculated from slug-test analyses completed in the upper part of the aquifer, within the road-salt-contaminated zone.

Background concentrations of chloride, sodium, and calcium in ground water, the primary constituents of road salt, range from 5 to 20 mg/L (milligrams per liter), 5 to 10 m/L, and 1 to 5 mg/L, respectively. Background specific conductance of the ground water ranges from 40 to 70 µS/cm (microsiemens per centimeter at 25 degrees Celsius).

DESCRIPTION OF INDUCTION LOGGER

The induction logger is designed to measure electrical conductivity of the formation surrounding a well. Electrical conductivity of the formation is a function of soil type, porosity, moisture content, and the concentration of dissolved solids (McNeill, 1980). Variations in electrical conductivity, either vertically or spatially within a well network, can be detected by use of induction logs. Variations in electrical conductivity of the pore fluid with time also can be detected by repetitive logging of a single well or a well network.

The borehole probe, 3.6 cm (centimeters) in diameter and 133 cm in length, is insensitive to the conductivity of the borehole fluid and the formation immediately surrounding small diameter wells; the peak response of the instrument occurs approximately 28 cm from the borehole axis (McNeill, 1986). The induction logger can be used in wells constructed of polyvinyl chloride (PVC) casing and screen but not in wells constructed of steel casing and screen. Electrical conductivity of the formation can be measured at five different sensitivity ranges (lowest range, 0 to 300 µS/cm; highest range, 0 to 30000 µS/cm) and recorded at five different vertical increments (0.1 to 1.6 m). For the remainder of this report, the electrical conductivity of the formation measured by the induction logger is referred to as "induction conductivity."

MATERIALS AND METHODS

Monitoring wells, constructed of 5-cm-inside- diameter PVC casing and slotted screen, were installed upgradient and downgradient from the highway for geophysical logging and water-quality sampling. Induction log data were compared with water-quality data to confirm the presence or absence of plumes of ground water contaminated by road salt.

Well Design and Construction

Ten wells, referred to as "long-screened wells" because they penetrate the entire thickness of the sand and gravel aquifer, are along two lines extending from 30 m upgradient to 180 m downgradient from the highway (figure 2). Well screens range from 15 to 30 m in length. The lines of wells are separated laterally by 120 m, each parallel to the direction of ground-water flow.

Two clusters of short-screened wells (1.5-m-long screens) were installed for water-quality sampling. These well clusters are midway between and parallel to the two lines of long-screened wells at approximately 30 m upgradient and 30 m downgradient from the highway (figure 2). The screens are placed to include all of the upper 3 m of the aquifer upgradient from the highway and all of the upper 7.5 m of the aquifer downgradient from the highway.

Additional short-screened wells designed primarily for induction logging were installed at three locations along the same ground-water flow path: one immediately upgradient from each well cluster and one about 90 m downgradient from the highway (figure 2). These wells, referred to hereafter as "induction wells," are screened at a depth of about 30 m below land surface with l.5-m-long screens.

Induction Logs and Water-Quality Samples

From March 1988 through March 1991, 131 induction logs were completed at the 10 long-screened wells; and from November 1990 through March 1992, 114 induction logs were completed at the 3 induction wells. Because of the low specific conductance of the ground water, the induction logger was set at its lowest recording range, 0 to 300 µS/cm, so that changes in induction conductivity with depth could be effectively recorded. All wells were logged with the vertical recording increment set at 0.2 m. Induction conductivity data were recorded on analog chart paper in the field and stored simultaneously in a digital data logger.

Induction-log data are used only for qualitative analysis of induction conductivity. Comparison of induction logs completed at the same wells logged on the same day sometimes show differences in measured induction conductivity with depth. However, the shapes of the induction-log profiles are so similar as to indicate a uniform shift in induction conductivity along the well bore. This uniform shift is likely a result of the inability to zero the induction logger sufficiently for accurate measurements at very low conductivities. To account for this shifting, the investigators adjusted all induction logs to a common reference of 10 µS/cm in the unsaturated zone just above the water table. This decision was based on the following factors:

(1) induction conductivity of the unsaturated zone is similar to that of the uncontaminated aquifer upgradient from the highway, (2) induction conductivities of the unsaturated zone upgradient and downgradient from the highway have consistently shown no change with depth, and (3) induction conductivities from wells downgradient from the highway increase significantly at the water table, immediately below the unsaturated zone. The induction conductivity of the unsaturated zone (10 µS/cm) was based on the estimated induction conductivity of the uncontaminated aquifer by use of Archie's Law (McNeill, 1980). Induction logs were adjusted by adding or subtracting a constant value from each measurement forming the log so that the induction conductivity of the unsaturated zone was 10 µS/cm.

From November 1990 to March 1992, 174 water-quality samples were collected from the two well clusters concurrently with induction logging. All water samples collected were analyzed for concentrations of dissolved chloride, sodium, and calcium, and for specific conductance.

RESULTS AND DISCUSSION

Delineation of Road-Salt Plume from Induction Logs

Analysis of induction-log data for the long- screened wells logged since 1988 show that induction conductivity varies little with depth upgradient from the highway; however, a distinct zone of high induction conductivity, relative to background conductivity, is present downgradient from the highway in the upper 3 to 6 m of the aquifer. A comparison of upgradient and downgradient induction logs is illustrated in figure 3. Induction conductivity in the upper 2 to 3 m of both logs is affected by the steel casing installed to protect the wells. Induction conductivity in well B2102, upgradient from the highway, is slightly higher below the water table than in the unsaturated zone, but the induction conductivity did not change significantly with depth. In contrast, the induction conductivity increases sharply at the water table in well B2303 downgradient from the highway. The maximum induction conductivity is at about 1 m below the water table; the conductivity declines rapidly downward within the next 2 m. Induction conductivity below this approximately 3-m-thick zone is similar to the upgradient induction conductivity; it is slightly higher than in the unsaturated zone and it changes little with depth. The zone of relatively high induction conductivity at the water table has been detected in wells 30 to 180 m downgradient from the highway.

Examination of induction logs also has shown that the relatively high induction conductivity at the water table downgradient from the highway changes significantly with time. Three induction logs completed at well B2303 in January, February, and March 1990 are shown in figure 4. The spike in induction conductivity immediately below the water table is recorded each month; however, the amplitude of the spike changes considerably with time. The maximum induction conductivity of the spike increases slightly from January to February and significantly from February to March. Induction conductivity does not change significantly through time below this zone. Temporal variation of induction conductivity is further illustrated in figure 5, which shows maximum induction conductivity of the water-table spike from all induction logs completed at well B2303 from March 1988 to March 1991. In general, these data indicate that induction conductivity varies annually and that the maximum occurs in the spring or summer of any given year. Induction conductivities from 1988 are at a maximum in the summer, however, because of the scanty data for this year, it is uncertain when the maximum actually occurred. Induction conductivities in 1989 and 1990 increase rapidly to a maximum in the spring and decrease less rapidly to a minimum in the autumn.

Lithologic and natural-gamma logs indicate no significant vertical changes in Ethology in the approximately 15-m-thick aquifer downgradient from the highway. Lithologic and natural-gamma logs from well B2303 are included in figure 4 for comparison with the induction logs. The aquifer--the zone between the water table and the layer of fine sand and some silt -is composed of fine to coarse sand and some gravel and is about 14 m thick at this location. The spike in induction conductivity is within the upper 4 m of the aquifer. The natural-gamma log does not indicate any significant changes in Ethology within the aquifer at this location. On the basis of the lithologic and natural gamma log data, the zone of relatively high induction conductivity is a result of vertical and temporal changes in conductivity of the aquifer fluid rather than the Ethology. Because this zone of high induction conductivity is downgradient from the highway and not upgradient from the highway, it must be related to highway runoff. Because the induction conductivity of this zone changes with time on an annual cycle, the changing induction conductivity could be related to the winter application of road salt. The design of the well clusters installed upgradient and downgradient from the highway for water-quality sampling was based on this interpretation. Water-quality data collected from the well clusters have confirmed that the layer of relatively high induction conductivity near the water table downgradient from the highway is a plume of ground water contaminated by road salt. Concentrations of chloride, sodium, and calcium, and specific conductance in water samples collected downgradient from the highway from depths below 3 m from the water table are not significantly different from those measured upgradient from the highway, which represent background levels. However, chloride, sodium, and calcium concentrations and specific conductance in water samples collected from the upper 3 m of the aquifer downgradient from the highway are 1.5 to 12 times greater than the background levels. Water-quality data and induction logs for well clusters sampled in July 1991 are shown in figure 6. Concentrations of chloride, sodium, and calcium, and specific conductance in the two wells spanning the upper 3 m of the aquifer upgradient from the highway represent background levels and are about 10, 6.0, and 1.0 mg/L, and 54 µS/cm, respectively. The induction log at this well site shows no significant change with depth; therefore, it is assumed that very little change in these water-quality characteristics occurs with depth. In the upper 7.5 m of the aquifer downgradient from the highway, however, significant changes in chloride, sodium, and calcium concentrations and specific conductance occur with depth, as indicated by the induction log. The chloride concentration in the upper 1.5 m of the aquifer is 120 mg/L, 12 times the background concentration. Sodium and calcium concentrations in this zone are about 10 and 7 times the background concentrations, respectively. Specific conductance is about 8 times the background level. In the zone 1.5 to 3 m below the water table, concentrations of chloride, sodium, and calcium, and the specific conductance are about 3.5, 3, 2.3, and 2.5 times the background levels, respectively. The concentrations of all three constituents and specific conductance in the zone 3 to 7.5 m below the aquifer are very close, and in some cases they are below the background levels upgradient from the highway. Water-quality data from July 1991 represent maximum levels measured during this annual cycle.

It is apparent that induction logs have successfully delineated the road-salt plume and provided for correct placement of well screens for collection of water samples contaminated by road salt. Induction logs will continue to be used to monitor the vertical distribution of road-salt-contaminated ground water to verify that the sampling well clusters are screened through the full vertical span of the plume.

Traveltime Measurements and Estimation of Hydraulic Conductivity from Induction Logs

A comparison of repetitive induction logs (1988- 93) indicates that most of the dissolved constituents from road salt that are applied during a given winter and reach the water table are transported a minimum of 90 m downgradient from the highway before the next winter season. A method to determine hydraulic conductivity of the road-salt-contaminated zone was developed on the basis of this annual fluctuation of induction conductivity.

The road-salt plume--derived from a non point source consisting of sodium and calcium cations and the conservative, nonreactive chloride anion--is assumed to travel at about the same flow rate as the ground water. On the basis of Darcy's Law, hydraulic conductivity can be calculated whereby average linear velocity is equal to the distance between two monitoring wells along the same ground-water-flow path divided by the traveltime of a tracer between the wells (Davis and others, 1985), as shown in the following equation:

K = (l/t)n/(dh/dl)

where K is hydraulic conductivity (m/d),
l/t is distance between monitoring wells divided by traveltime of tracer (m/d),
n is effective porosity (dimensionless), and
dh/dl is hydraulic gradient (dimensionless).

The traveltime of the road-salt plume is interpreted from plots of the maximum induction conductivities of the plume at two wells and their corresponding dates (figure 7). Two different traveltimes for the road-salt plume between two monitoring wells were used: (1) the time interval between first annual arrivals as the plume is first detected by increased induction conductivities, and (2) the time interval between the temporal centroids, or first moment (Fischer and others, 1979), of the areas defined by the temporal distributions of maximum induction conductivities through their annual cycles. The temporal centroid is an estimate of the time at which half of an annual plume, as detected by induction logs, has passed by a monitoring well.

In 1989, the road-salt plume was first detected at monitoring well B2303 on March 23, Julian Day 82 (JD 82), and 62.5 m further downgradient at monitoring well B2403 on May 8 (JD 128). If the plume is assumed to have arrived between its first detection and the previous induction log, the midpoints being March 11 (JD 70) for B2303 and April 16 (JD 106) for B2403, a traveltime based on times of first arrival is 36 days (figure 7). The traveltime based on the temporal centroids of the plume, June 28 (JD 179) and September 4 (JD 247) for B2303 and B2403, respectively, is 68 days (figure 7).

Data used to calculate hydraulic conductivity, and the conductivities calculated, are listed in Table 1. The effective porosity of 0.38 is based on the range of porosities (0.35-0.40) measured at a nearby test site (LeBlanc, 1987). Hydraulic gradients are averaged over the traveltime interval. Hydraulic conductivities of 140 and 66 m/d were calculated for the 1989 times of first arrival and temporal centroids of the road-salt plume, respectively. In 1990, traveltime of first arrivals is 68 days and traveltime of the temporal centroids is 103 days (figure 7), resulting in hydraulic conductivities of 74 and 46 m/d, respectively (table 1). As would be expected, estimated hydraulic conductivities calculated for times of first arrival are greater than those from temporal centroids because of the effects of longitudinal hydrodynamic dispersion. However, times of first arrival may indicate approximate maximum velocities of ground- water flow (Davis and others, 1985).

The range of hydraulic conductivities determined from induction-log data (1989 and 1990), 46 to 140 m/d, compare reasonably with the range of hydraulic conductivities obtained from slug-test analyses completed in this road-salt contaminated zone, 35 to 48 m/d. Because of these initial favorable results from existing data collected approximately monthly, the methods of determining hydraulic conductivities from times of first arrival and temporal centroids were pursued by collecting induction logs weekly in 1991 from the induction wells at the B2 cluster and 66.5 m downgradient at B3 (fig. 2). Induction-log data were collected more frequently to improve definition of the annual road-salt plume for traveltime measurements. The traveltime of the road-salt plume from the weekly data was 47 days from the annual first arrivals and 84 days from the temporal centroids (fig. 8), resulting in hydraulic conductivities of 103 m/d and 61 m/d, respectively (table 1). These hydraulic conductivities also compare reasonably with the hydraulic conductivities from the slug-test analyses, and they define a range similar to conductivities determined at a nearby test site of 60 to 110 m/d (LeBlanc, 1987).

Times at which water-quality samples were collected concurrently with induction logs also are indicated in figure 8 for comparison of results from weekly and monthly data. Traveltimes of the annual first arrivals and temporal centroids from the monthly induction logs are 30 days and 94 days, resulting in hydraulic conductivities of 162 m/d and 55 m/d, respectively (table 1). Traveltimes and resulting hydraulic conductivities from the weekly and monthly induction logs differ considerably for the annual first arrivals (103 m/d and 162 m/d), but differ little for the temporal centroids (61 m/d and 55 m/d). This contrast suggests that monthly induction logging may be sufficient for determination of traveltimes of temporal centroids, but, more frequent data are needed for determination of traveltimes of annual first arrivals.

CONCLUSIONS

The borehole induction logger has proven to be an efficient and effective instrument in monitoring road- salt contamination of ground water. Induction logs have been used in the design of wells and placement of screens by defining the vertical location and extent of a road-salt plume downgradient from a highway. The induction logger continues to be used to verify that the well screens span the full vertical extent of the plume. Induction logs can be used to detect or monitor any contaminant plume if its electrical conductivity differs from that of the surrounding uncontaminated ground water. Favorable results were achieved in developing method to determine hydraulic conductivity from induction-log data by monitoring traveltimes of annual road-salt plumes between wells along the same ground water-flow path. The range of hydraulic conductivity determined by use of the induction-log data, 46 to 162 m/d, compares reasonably with the range determined from slug-test analyses in the road-salt-contaminated zone. Although the induction logging method may require frequent monitoring during a relatively long period of time, it does not require injection of a tracer and collection of water samples to capture the tracer. By monitoring the road-salt plume (an existing non point source fluctuating on an annual basis) with induction logs from existing wells, hydraulic conductivity can be estimated under natural flow conditions.

ACKNOWLEDGMENTS

The authors thank the Research and Materials Section and the District No. 5, Wareham Office of the Massachusetts Highway Department, for providing ongoing logistical support, including drilling and well installation, surveying, and recording of road-salt application during this study. The authors also thank the Research and Materials Section and the Department of Transportation, Federal Highway Administration, for their technical comments and suggestions.

REFERENCES CITED

Davis, S.N., Campbell, D.J., Bentley, H.W., and Flynn, T. J., 1985, Ground water tracers: National Water Well Association, 200 p.

Fischer, H.B., List, E.J., Koh, R.C.Y., Imberger, J., and Brooks, N.H., 1979, Mixing in inland and coastal waters: New York, Academic Press, Inc., 483 p.

LeBlanc, D.R., 1987, Fate and transport of contaminants in sewage-contaminated ground water on Cape Cod, Massachusetts in Franks, B. J., ea., U.S. Geological Survey program on toxic waste--ground water-contamination; Proceedings of the third technical meeting, Pensacola, Florida, March 23-27, 1987, U.S. Geological Survey Open-File Report 87 109, p. B3-B7.

McNeill, J.D., Electrical conductivity of soils and rocks: Ontario, Canada, Technical Note TN-5, Geonics Limited, 22 p.

McNeill, J.D. 1986 Geonics EM39 borehole conductivity meter, theory of operation: Ontario, Canada, Technical Note TN-20, Geonics Limited, 17 p.

ABOUT THE AUTHORS

Peter E. Church is a hydrologist with the U.S. Geological Survey, Water Resources Division, Massachusetts-Rhode Island District, in Marlborough, Massachusetts. Mr. Church earned an M.S. degree in Water Resources Management and an M.S. degree in Geography, both from the University of Wisconsin- Madison. Since 1984, he has been leading an investigation to determine the effectiveness of various highway-drainage systems in preventing road- salt contamination of ground water.

Paul J. Friesz is a hydrologist with the U.S. Geological Survey, Water Resources Division, Massachusetts-Rhode Island District, in Marlborough, Massachusetts. Mr. Friesz received his B.S. degree in Civil Engineering from Ohio University and participated in the investigation of the effectiveness of the highway-drainage systems from 1989 through 1992. Currently, Mr. Friesz is investigating the availability of ground water in a drainage basin in western Massachusetts.

Table 1.

FIGURE CAPTIONS

Figure 1.--Location of study area in southeastern Massachusetts and test sites A, B, C, and D along Route 25.

Figure 2.--Locations of wells relative to Route 25, altitude of water table on August 1, 1989, and direction of ground-water flow at site B in Wareham, Massachusetts.

Figure 3.--induction logs obtained in August 1988 from monitoring wells B2102 (upgradient from the highway) and B2303 (downgradient from the highway) along the same ground-water-flow path (plots are shifted vertically to show differences in altitudes of land surface and the water table).

Figure 4.--Induction logs, lithologic log, and natural-gamma log for monitoring well B2303, 30 meters downgradient from the highway.

Figure 5.--Maximum induction conductivities of the water-table spike from March 1988 through March 1991 from all induction logs at well B2303, 30 meters downgradient from the highway.

Figure 6.--induction logs and water-quality data collected on July 25,1991 from well clusters B1 (upgradient from the highway) and B2 (downgradient from the highway) along the same ground-water-flow path (concentrations in milligrams per liter; specific conductance in microsiemens per centimeter at 25 degrees Celsius; plots are shifted vertically to show differences in altitudes of land surface, water table, and well screens).

Figure 7.--Maximum induction conductivities of the road-salt plumes from March 1988 through March 1991 at monitoring wells B2303 and B2403, 30 meters and 92.5 meters downgradient from the highway, respectively, along the same ground-water-flow path.

Figure 8.--Maximum induction conductivities of the road-salt plume from November 1990 through March 1992 from induction wells B2 and B3, 30 meters and 96.5 meters downgradient from the highway, respectively, along the same ground-water-flow path.