USE OF UNDERWATER RESISTIVITY IN THE ASSESSMENT OF GROUNDWATER-SURFACE WATER INTERACTION WITHIN THE BURD

RUN WATERSHED

Paul A. Freyer, Temple University Geology, Philadelphia, PA Jonathan E. Nyquist, Temple University Geology, Philadelphia, PA

Laura E. Toran, Temple University Geology, Philadelphia, PA

Abstract

Characterizing groundwater interaction with streams is essential for understanding contaminant transport. We are investigating the use of multielectrode resistivity to improve the detection of seepage points, and the mapping of gaining and losing stream reaches. Our field area is the Burd Run watershed, Shippensburg, PA. Burd Run is of interest as a geophysical test case because the stream is dilute where it flows from the metasedimentary ridge of South Mountain across a colluvium wedge. Fluid conductivity increases from 35 µS/cm to 440 µS/cm in three abrupt increments as the stream flows across the steeply dipping carbonate units of the Great Valley before discharging into Mill Spring Creek. The increases can be attributed to the introduction of carbonate groundwater, both from municipal discharge and from vertical seeps within the streambed. Continuous dipole-dipole surveys were conducted over a 107 m reach using a 28-electrode cable with a 1-m electrode spacing deployed on the streambed. The survey resolved the conductivity contrast of the streambed sediments over the carbonate bedrock and detected a vertical zone of higher electrical conductivity that correlates with observed streambed seeps.

Introduction

The quantification of groundwater-surface water interactions is vital to understanding contaminant movement and distribution, in addition to improving the management of water supplies (Wroblicky et al., 1998, Conant, 2004, Oxtobee et al., 2002). These interactions along streams and rivers can be quantified using point source monitoring equipment such as mini-piezometers, seepage meters and temperature surveys (Oxtobee et al., 2002). Exchange between groundwater-surface water regimes depends on many complex factors. Because these factors include bedrock topography, temporal climatic variations, sediment types, and hydrologic properties of the materials (Oxtobee et al., 2002, Cey et al., 1998), it can be problematic deciding where to deploy monitoring equipment and how to interpolate between point measurements.

Conant (2004) used a grid of temperature sensors and mini-piezometers over a 60 m stretch of the Pine River in Ontario to quantify hydraulic flux through the streambed sediments overlying a contaminated aquifer. Over 400 temperature sensors were used in the study. Most of the streambed area covered by the temperature sensor grid experienced little or no groundwater contribution. Conant (2004) was able to delineate three discharge areas and showed evidence of fine-scale seepage variations. The region of intense study, however, represented only a fraction of the total stream length. Applying these point source measurement techniques over large areas would be prohibitively expensive and time consuming.

The distribution of monitoring equipment needs to be guided by an understanding of the subsurface geology. Electromagnetic and 2D electrical resistivity techniques have been used on land to identify points of accelerated contaminant transport in karst terrain. Ahmed and Carpenter (2003)

focused on geophysical signals that are likely to result from geologic heterogeneities that yield preferential hydraulic conduits or barriers. Zhou et al. (2000) discussed the difficulties in determining the depth to bedrock, thickness of resistivity horizons and the need for calibration with core data. In our previous work using land-based resistivity surveys to characterize karst (Jenkins and Nyquist, 1999; Mackey et al., 1999; Roth et al., 2000: Roth et al., 2002; Nyquist and Roth, 2005), we found that resistivity tomography surveys in areas with limestone bedrock are sensitive to changes in the thickness of overlying, electrically-conductive clay sediments. Thickness and distribution of clay sediments are also important factors controlling seepage in streams.

The goal of our current research is to investigate whether 2D electrical resistivity tomography surveys can aid in the delineation of groundwater-surface water exchange points, either directly, by detecting the upwelling of conductive groundwater, or indirectly, by mapping geologic heterogeneities that control the streambed’s hydraulic transmissivity.

Background

The Burd Run Watershed is located in Shippensburg, PA 150 miles west of Philadelphia in Cumberland County, as shown in Figure 1. The stream flows northwest from South Mountain down into Great Valley. Metasedimentary, metaigneous, and carbonate complexes underlie Burd Run. The metasedimentary and metaigneous formations are topographically expressed as South Mountain. Fractured, steeply dipping, Cambrian and Ordovician carbonates dominate the Great Valley. A prominent sedimentary feature affecting regional hydrologic flow is the presence of a colluvium wedge between South Mountain and the valley floor. The majority of the carbonate units are mantled by variable thicknesses of clays and fluvial sediments, limiting the bedrock exposures (Lindsay, 2005).

Lindsay (2005) presented fluid conductivity measurements around the region in association with groundwater modeling of local and regional flow systems. The increase of stream water conductivity between the headwaters of Burd Run (30 µS/cm) and the discharge point in Middle Spring Creek (500 µS/cm) implied the influx of groundwater throughout the valley. We collected additional fluid conductivity measurements to constrain the losing and gaining reaches of Burd Run. This allowed us to focus our resistivity tests on stream reaches most likely to have groundwater-surface water exchange. The results presented here cover a stretch of Burd Run that we refer to as Craig’s Reach. Craig’s Reach is a channelized portion of Burd Run bounded by soy fields and landscaped private property. The geology provided a significant contrast in resistivity values. The geologic setting combined with the visual confirmation of seeps along the reach made it an ideal study location.

Field Methods

Preliminary Conductivity Survey

To discern segments of Burd Run that experience the greatest influx of groundwater, we collected fluid conductivity measurements using a handheld conductivity and temperature meter (Figures 2 and 3). All measurement values were recorded in µS/cm and degrees Centigrade. We began with a reconnaissance conductivity survey using a coarse distribution of sampling points across the watershed. For convenience, conductivity measurements were primarily collected where Burd Run intersected a road, driveway, or railroad bed, which resulted in measurements roughly 500 m apart. We also collected water samples at each conductivity measurement point for analysis of major cations and anions using ion chromatography. We identified the reaches of Burd Run with the greatest increases in fluid conductivity values by plotting measurements in ArcGIS. Stream segments that showed fluid conductivity increases greater than 50 µS/cm between measurement points in the reconnaissance survey

Figure 1. (a) General outline of Pennsylvania with a black box highlighting the Shippensburg area between Cumberland and Franklin Counties. (b) Combined portions of the Walnut Bottom Quad and Shippensburg Quad. The numbers on the topographic map represent: (1) Shippensburg University, (2) South Mountain, (3) piped well water discharging into Burd Run, (4) spring water that flows through a ditch into Burd Run, and (5) Craig’s Reach. Note: The colluvium cover extends from the slope of South Mountain (near 3) to the valley floor (near 4). (c) Aerial photograph over a segment of Burd Run highlighting Craig’s Reach.

were then surveyed on a meter scale by walking down those reaches. We combined a visual inspection of the streambed and banks with the results of the refined conductivity survey to locate seeps and springs.

Underwater Resistivity Surveys

A Marine SuperSting® R8/IP (Advance Geosciences Inc.) was used to carry out a 107 m dipole-dipole tomography survey along Craig’s Reach. For this survey we employed a waterproof cable with 28 passive stainless steel electrodes spaced 1 m apart (Figure 3). Current and voltage switching was controlled within the SuperSting®. The resulting profile depth for a dipole-dipole sounding was roughly 6 m. To produce a continuous 107-m profile, we overlapped each successive sounding by 17 m.

EarthImager2D® was used to invert the composite pseudosection, yielding the continuous resistivity profile along Craig’s Reach. We measured water depths above each electrode, so the water layer thickness and conductivity was included in the data inversion process. Bed forms, spring locations, and notable channel features were recorded while each profile was being sounded. Sediment samples were collected to determine the resistivities of the fluvial sediments and the mantle of clay over the carbonate bedrock.

EM31 Conductivity Survey

We collected terrain conductivity data using a Geonics® EM31 (Figure 3) for comparison to the dipole-dipole survey. The EM31 was kept at a height of 1 m above the water surface, with the exception of one deep pool where the instrument was close to the water surface. The penetration depth of the EM31 is roughly 6 m, which is comparable to the maximum depth of investigation for the dipole-dipole configuration used in the resistivity survey.

Figure 2. Plot of fluid conductivity values collect along Burd Run during our preliminary s(a) Increapiped into Burd Run. (b) Increase in conducti

urvey. se in conductivity attributed to the addition of well water from unfinished subdivision

vity resulting from a spring that discharges via a ditch into Burd Run. (c) Contributions made by springs and seeps along Craig’s Reach.

0

50

100

150

200

250

300

350

400

450

500

0 1 2 3 4 5 6 7 8 9

Flui

d C

ondu

ctiv

ity (µ

S/cm

)

Distance (km)

Surface-water Conductivity Variability Along Burd Run

a

b

c

Figure 3. Field photographs: (a) Fluid conductivity and temperature data being collected at Shippensburg University. (b) EM31 survey line at Craig’s Reach in the deep pool seen in the resistivity profile. (c) Dipole-dipole 2D resistivity sounding along Craig’s Reach. The inset inthe bottom left corner shows a magnification of the electrode.

Results and Discussion

Surface Water Conductivity Results The reconnaissance fluid conductivity survey yielded conductivity values that ranged from 35

µS/cm near the headwaters of South Mountain to 440 µS/cm at Shippensburg University before Burd Run discharges into Middle Spring Creek. The primary increase in conductivity along the flank of South Mountain is attributed to a pipe that discharges groundwater from a well in a subdivision development project into Burd Run. A second jump in fluid conductivity occurs where a spring-fed pond discharges into Burd Run via a drainage ditch. This is located southwest of the intersection of Pennsylvania State Route 174 and Airport Road. The third notable increase in fluid conductivity results from a series of springs located in a stretch of Burd Run that begins at Craig’s Reach and extends to Shippensburg University. Consequently, we targeted this area for our investigation. Underwater Resistivity Results

The resistivity profile in Figure 4a shows that the resistivity of the shallow streambed sediments ranges from 100 ohm-m to 400 ohm-m. Based on auger samples, this shallow, relatively resistive layer is a sandy-clay loam with coarse gravel at the stream/sediment interface. Beneath this layer is a significantly less resistive unit, with resistivity values ranging from 20 ohm-m to 60 ohm-m. This corresponds to a clay unit we encountered while augering. Below the clay layer, from positions 11 m to 56 m along the line, there is a more resistive unit at the base of the resistivity section. We were not able to hand auger down to the bedrock, but this resistive feature (Figure 4a) is probably a bedrock pinnacle with a subvertical expression of the dip in the Stonehenge and Shady Grove Formations. The lower-resistivity vertical features within the pinnacle might be attributed to downward translocation of clays along the bedding planes, weathering of micritic limestone, or the vertical flow of conductive fluids. The seeps that were discovered along Craig’s Reach during the investigation correlated to this bedrock high in the resistivity profile.

EM31 Results

Figure 4b shows EM31 data plotted as resistivity versus distance along the line for the same stretch as the underwater resistivity profile. In general, the apparent resistivity values ranged from 60 ohm-m to 90 ohm-m. The data collected from the first 5 m of the line were omitted from the plot. The anomaly removed was due to a rebar-enforced bridge at the start of the Craig’s Reach survey line that exceeded the operating range of the EM31.

The EM31 profile failed to differentiate the location of the seeps within the first 60 m of the line. The signal seems most sensitive to the depth of the clay layer, returning lower resistivity values at the east end of the profile line and higher resistivity values at the west end of the profile line.

Conclusions

Using fluid conductivity values on a coarse interval throughout a watershed identified stretches where groundwater contributions to the surface water were being made on a scale of 50-100 m. The change in fluid conductivity in the surface waters did not help in identifying the losing stretches of Burd Run. Furthermore, meter-scale mapping revealed that the groundwater discharge into the stream was occurring at distinct points controlled by geologic heterogeneity.

The continuous dipole-dipole resistivity survey along Craig’s Reach shows the variation of sediment types with depth beneath Burd Run. Given the large contrast in resistivity between the sediment and rock units, resistivity is highly sensitive to the thickness of the sediments and depth to

carbonate bedrock. Two lower resistivity zones within the carbonate bedrock correspond to the seeps observed in the streambed above. The vertical orientation of these zones is parallel to the dip of the bedding of the Stonehenge and Shady Grove Formations. Underwater 2D electrical resistivity tomography was successful in locating an interaction point along Craig’s Reach. Consequently, this method has the potential for application in other such similar settings, particularly in guiding the placement of seepage point monitoring equipment.

We plan to collect additional ground truth information to ascertain the layer thickness in the resistivity section, including a tile probe survey of bedrock depths and three sediment cores along Craig’s Reach. We will also install a series of temperature loggers to detect inflow or outflow for comparison with the resistivity results. The resistivity survey along Craig’s reach was conducted under base flow conditions. In the spring of 2006, when the stream conductivity falls due to increased rainwater and snowmelt, we will repeat the resistivity line to determine if the increase between the fluid conductivity of the stream and groundwater improves seepage detection.

Figure 4. (a) Continuous dipole-dipole survey over Craig’s Reach. Vertical exaggeration is 3x. (b) EM31 profile of the first 100 m of Craig’s Reach. The gray shaded area represents the section of the reach where seeps are known to be. We interpreted the low resistivity layer starting at a depth of 1.6 m to be clay. More resistive sandy-clay loam lies above the clay layer. The resistive body at the base of the resistivity profile, between 11 m and 56 m, is interpreted to be a limestone pinnacle that contains two vertical low-resistivity zones that correspond to the known seep locations on the surface.

References Ahmed, S., Carpenter, P.J., 2003, Geophysical response of filled sinkholes, soil pipes and associated

bedrock fractures in thinly mantles karst, east-central Illinois, Environmental Geology, 44, 705-716.

Conant, B., 2004, Delineating and quantifying ground water discharge zones using streambed temperatures, Ground Water, 42, 243-257.

Jenkins, S. A., and Nyquist, J. E., 1999, An investigation into the factors causing sinkhole development at a site in Northampton County, Pennsylvania, in Proceedings of the Seventh Multidisciplinary Conference on Sinkholes and the Engineering and Environmental Impacts of Karst, 45-49.

Lindsay, B., 2005, Hydrogeology and simulation of source areas of water to production wells in a colluvium-mantled carbonate-bedrock aquifer near Shippensburg, Cumberland and Franklin Counties, Pennsylvania, USGS Scientific Investigations Report 2005-5195.

Mackey, J.R., Roth, M.J.S., and Nyquist, J.E., 1999, Case study: Site characterization methods in karst, in Geo-Engineering for Underground Facilities, in Geotechnical Special Publication No. 90, G. Fernandez and R. Bauer, eds., ASCE, Reston, VA, 695-705.

Nyquist, J. E. and M. J. S. Roth, 2005, Improved 3D pole-dipole resistivity surveys using radial measurement pairs, Geophy. Res. Lett., 32, no. 21, L21504.

Oxtobee, J.P.A., Novakowski, K., 2002, A field investigation of groundwater/surface water interaction in a fractured bedrock environment, Journal of Hydrology, 269, 169-193.

Roth, M. J. S., Mackey, J.R., Mackey, C., and Nyquist, J.E., 2002, A case study of the reliability of multielectrode earth resistivity testing for geotechnical investigations in karst terrains: Engineering Geology, 65, 225-232.

Roth, M.J.S, Nyquist J.E., and Guzas B., 2000, Locating subsurface voids in karst: a comparison of multi-electrode earth resistivity testing and gravity testing: Proceedings of the Symposium on the Application of Geophysics to Environmental and Engineering Problems (SAGEEP 2000), Washington, D.C., February 20-24, 359-365.

Wroblicky, G.J., Campana, M.E., Valett, H.M., Dahm, C.N., 2002, Seasonal variation in a surface-subsurface exchange and lateral hyporheic area of two stream-aquifer systems, Water Resources Research, 34, 317-328.

Zhou, W., Stephenson, J.B., 2000, Reliability of dipole-dipole electrical resistivity tomography for defining depth to bedrock in covered karst terranes, Environmental Geology, 39, 760-766.

Acknowledgements

We thank the National Science Foundation for funding that allowed us to purchase the Marine

SuperSting® used in this project (Grant EAR-0509853). We also thank the Craig family for permission to work on their property. Invaluable information on the Shippensburg area and updates on the stage conditions of Burd Run were provided by Dr. Christopher Woltemade of Shippensburg University. Bruce Lindsey of the USGS introduced us to the study site and provided extremely useful background data.