High Conductivities Associated with an LNAPL Plume Imaged by
Integrated Geophysical Techniques

William A. Sauck, Estella A. Atekwana and Mike S. Nash
Department of Geology, Western Michigan University, Kalamazoo, MI 49008

 

 

ABSTRACT

Geophysical investigations used to characterize hydrocarbon spill sites have been based principally upon the electrical properties of the hydrocarbons. Controlled spill experiments support the model that high electrical resistivity and low relative permittivity (c,) are characteristic of geologic media contaminated by hydrocarbon spills. However, many geophysical field investigations of light nonaqueous phase liquid (LNAPL) contaminated sites report results that contradict the controlled spill experiments. Some of these studies have documented the attenuation of ground penetrating radar (GPR) signal amplitude coincident with areas of hydrocarbon contamination resulting in a ‘shadowing’, or ‘fuzzy’, or ‘halo’, or ‘muted’ appearance. To evaluate this discrepancy and to assess the sensitivity of other geophysical methods to organic contaminant detection, an integrated geophysical investigation was conducted at a former fire training facility using GPR, Electrical Resistivity, and Self Potential (SP) methods.

The results show a region of attenuated GPR signals resulting in a ‘shadow’ zone over areas with LNAPL contamination. Further, low resistivities and high SP anomalies were also observed over the contaminated zones and found to be coincident with areas of GPR ‘shadow’. These results suggest the presence of more conductive groundwater within plume areas. From this, it is proposed that the ‘shadow’ zones observed in the GPR records are an indication of radio wave attenuation due to elevated groundwater conductivity. This explanation is supported by geochemical studies which indicate the presence of highly conductive waters below some LNAPL plumes due to the release of ions from aquifer solids by reaction with organic acids or carbonic acids derived from the biodegradation of the hydrocarbon compounds. Therefore, we hypothesize that hydrocarbon spills (in this case LNAPLS) in the natural environment cause changes from electrically resistive to conductive behavior with time due to biodegradation. If this is true, it explains why the current physical model, supported by controlled spill experiments, is apparently inadequate to explain results obtained from field investigations of uncontrolled spill sites.

Introduction

The detection of organic contaminants such as light nonaqueous phase liquids (LNAPLS) and dense non-aqueous phase liquids (DNAPLS) in the subsurface using geophysical methods, particularly ground penetrating radar, has been the subject of considerable interest among geophysicists in recent years. Their detection using geophysical methods is, among other factors, dependent on their known electrical properties of low apparent conductivities (s) and low relative permittivity (er ) compared to the higher apparent conductivities and higher relative permittivity of background waters (e.g., Campbell et al., 1996; Daniels et al., 1992; DeRyck et al., 1993; Redman et al., 1994).

NAPL contamination may differentiate into at least four distinct bodies, each of which has unique chemical and geophysical characteristics. There is an immiscible phase, or free product, which is mobile or free to migrate under the influence of gravity. Secondly, the residual phase is that portion which is trapped or left behind after the free product has been removed by recovery wells or has migrated down the hydraulic gradient. Third, volatile NA.PLs may have a well-developed vapor plume in the region above free product and residual product. Finally, small amounts of hydrocarbon enter the aquifer as a dissolved phase. The dissolved phase hydrocarbon plume is important for health reasons, but is not directly of consequence for geophysical methods because of the small quantities (ppm) and lack of charge carriers (organic molecules are electrically neutral).

Extensive field and laboratory studies have been conducted to investigate the geophysical response to hydrocarbon contaminants in earth materials. Olhoeft et al. (1988) mapped the Bemidji spill using GPR with moderate success. King and Olhoeft (1989) suggested that the passage of an organic plume left a trail of altered clay minerals detectable by electrical geophysical methods. Investigators at the University of Waterloo have examined the geophysical response of organic spills in detail with very well controlled laboratory and field experiments. Endres and Redman (1996) began with theoretical modeling of electrical properties of sediments containing immiscible contaminants. Next, the scale was increased to a large sand tank where kerosene was "spilled", and geophysical measurements were made at the surface and in vertical probes (DeRyck et al., 1993, Redman and DeRyck, 1994, and Redman et al., 1994). Redman et al., (I 994), noted a decrease in the amplitude of the "water table" reflection after kerosene injection, but showed with modeling that the effect could be an increase or a decrease, depending upon the GPR frequency applied. In the field, tetraperchloroethylene (PCE), a DNAPL, was injected into a larger enclosure and the downward movement of the PCE was monitored using vertical probes for Time Domain Reflectometry (TDR), Microresistivity, and Thermal Neutron logging (Brewster and Annan, 1994; Schneider and Greenhouse, 1992; Endres and Greenhouse, 1996). These experiments all reported resistive behavior for the organic-contaminated sediments at these early times. Further, Daniels et al. (I 992) in controlled sand test pit experiments reported high amplitude GPR anomalies associated with sands saturated with diesel fuel. Recently, Campbell et al. (1996) in controlled spill experiments at the Oregon Graduate Institute also report an enhanced, brighter (higher amplitude) saturated zone reflection under the spill. One suggested mechanism for the enhancement of GPR reflections is that the LNAPL displaces the water from the transition zone above the capillary fringe, thus making a sharper interface at the top of the water saturated zone. Alternatively, Redman et al. (1994) suggest that the LNAPL pool provides a layer of intermediate relative pemittivity between air saturated and water saturated conditions, thus causing a decrease in the GPR reflection amplitude at the top of the saturated zone.

However, GPR response over known hydrocarbon spill sites is ambiguous and appears to be site dependent. Nonetheless, most field characterizations of LNAPL contaminated sites using GPR and other electrical methods such as electrical resistivity and electromagnetic induction report results that contradict the controlled spill experiments. Some of these studies have documented the attenuation of GPR signal amplitude in the vicinity of gasoline concentration in the capillary fringe above the water table and describe a ‘shadow’, or ‘fuzzy’, or ‘halo’, or ‘muted’ appearance coincident with areas of known contamination (e.g., Daniels et al., 1992; Grumman and Daniels, 1995; Maxwell and Schmok, 1995). Further, high apparent conductivities (Monier-Williams, 1995) and low apparent resistivities (Gajdos and Kral, 1995; Sauck and McNeil, 1994) associated with areas of known hydrocarbon contamination have been reported.

While controlled spill experiments have provided valuable insights into the potential for geophysical detection of hydrocarbon contaminants, it is clearly apparent that the ‘insulating layer’ model suggested by Mazac et al. (1990) and supported by controlled spill and tank experiments has been inadequate in explaining the anomalous conductivities associated with and observed over hydrocarbon spill sites. We believe that a better understanding of the evolution of the geophysical properties of these hydrocarbon contaminants in the natural setting is critical to the successful application of geophysical methods to their detection. Given the discrepancies described between field studies and controlled spill experiments, this study was undertaken to evaluate this problem in a field setting, while also evaluating the potential of other geophysical methods for organic contaminant detection in uncontrolled field settings. Specifically, we seek to explain the cause of the attenuation of GPR signals leading to the ,muted’ or ‘fuzzy’ or ‘shadow’ character coincident with areas of hydrocarbon contamination and explore the cause for the discrepancy between the controlled spill experiments and field experiments at uncontrolled spill sites.

Study Site

The study site is the former fire training cell (FT-02) located on the decommissioned Wurtsmith Air Force Base (AFB), in Oscoda, Michigan (fig. 1). Part of the base is currently used as the University of Michigan’s National Center for Integrated Bioremediation Research and Development (NCIBRD). The site was used by the United States Air Force for 24 years as a bi-weekly fire training facility. A typical exercise involved the combustion of several thousand gallons of jet (JP-4) and other hydrocarbon fuels. Some, but not all of the fuel would bum, leaving the rest to percolate into the ground along with the fire retardant chemicals (aqueous film forming foam (AFFF), an organic surfactant mixed with butyl carbitol) used to extinguish the fire. In 1982, a concrete fire-containment basin with an oil-water separator was installed to help reduce the amount of fuel entering into the ground. An unknown quantity of fuel had infiltrated into the subsurface prior to this date, and anecdotal reports indicate that overflows at the separator persisted after this date. Chemical analyses of the groundwater show elevated amounts of benzene, toluene, ethyl benzene, and xylene (BTEX) compounds in the subsurface along with elevated conductivities of the groundwater. As much as 30 cm of free product has been observed above the water table during construction of wells in the plume source area to more than 200 in downgradient.

Table 1. Groundwater conductivities from monitoring wells at the FT-02 site. *Indicates wells located outside the hydrochemically defined plume boundaries.
Well Depth (in) Cond. (mS/m)
FT-2
FT-3
FT-4
FT-5
FT-6
FT-7*
FT-8
FT-9
FT-10*
FT-11*
FT-12
FT-13
FT-14		 7.72
7.32
6.10
6.10
6.10
9.45
9.45
9.75
9.45
9.45
9.45
9.14
9.45		 34.8
44.4
50.7
47.6
22.5
13.6
50.2
45.5
24.3
30.0
43.1
57.4
30.1

Similar thicknesses of residual product due to "smearing" by water level fluctuations have been found above and below the free product. Figure I shows the outline of the hydrochemically defined plume. The lateral boundaries were drawn between detect and non-detect samples from transverse rows of wells. NCIBRD laboratory data from water samples drawn from various monitor wells on and off plume show electrical conductivity enhancement of more than a factor of two relative to background values. Typical conductivity values are shown in Table 1.

Based on several borings at the site completed by the-United States Geological Survey (USGS) and from previous investigations at the site (e.g.. Robbins et al., 1995), the shallow subsurface stratigraphy is uniform and consists of wellsorted fine to medium sands coarsening with depth. Underlying the sandy deposits at approximately 19.8 in (65 feet) is a lacustrine silty clay unit ranging in thickness from 6.1 to 30.5 in (20 to 100 feet). The silty clay unit is underlain by a thin glacial till deposit, resting upon the Paleozoic bedrock consisting of the Mississippian Marshall Sandstone and the Coldwater Shale. The depth to ground water ranges from 3.7 in to 5.3 in (1 2 to 17.4 feet) in ground water monitoring wells. Ground water flow is to the southeast, toward a large wetland located 366 in (1200 feet) south to southeast of the site in the flood plain of the Au Sable River (Robbins et al., 1995).
Figure 1. Map of the FT-02 site with locations of geophysical survey lines. The stippled area represents the location of the plume as determined from hydrochemical studies. The heavy lines are the representative geophysical profiles discussed in the text. The donut-shaped structure is the concrete fire containment basin. (Base map adapted from NCIBRD.)

wpeB.jpg (65271 bytes)
 

Field Study

An integrated geophysical investigation was undertaken at the study site using GPR, electrical resistivity (dipole-dipole profiling and Schlumberger vertical electrical sounding), and self potential (SP) methods. A survey grid of 15.24 m x 15.24 m (50 x 50 ft) already established at the site was utilized (fig. 1). The established grid intersects contaminated and uncontaminated areas.

Ground Penetrating Radar

Measurements were made with 100 MHz bistatic antennae recording for a total of 400 nanoseconds. The 100 MHz Transmit-Receiver pair were set at a fixed separation of 1.4 meters (between centers). Acquisition parameters were: 512 samples/scan, 20 scans/second, 3-stage vertical (along scan) IIR low pass filter at 120 cycles/scan, 4-stage vertical IIR low pass filter at 15 cycles/scan, and gains automatically set at the beginning of the line to equalize amplitudes to approximately 80% of full scale. Note that this system uses constant gain setting, and has no AGC or other gain changes along the profile line. Slight horizontal smoothing was done by applying a 3-scan moving average filter. GPR lines were acquired on 30.48 m (100 foot) intervals in the east-west direction with several north-south lines to tie the data together (see fig. I for location of profiles). Two-dimensional (x,z) profile sections were produced for interpretation, and representative profiles along line 305 N and 396 N are shown in figure 2. The data presented are the field data, as no post acquisition filtering was deemed necessary due to the excellent record quality. The GPR profiles show several strong reflectors. Of particular interest is a very strong reflector seen in the records at approximately 80 ns which is equivalent to a depth of 4.35 m (14.3 ft), which is the depth to the water table at the site (resulting in an er Of 9). This strong reflection at the water table is due to the large change in the relative permittivity from moist sand (er ~ 9) to saturated sand (er~ 25). There are several broad antiforms in the record below the water table at 150 ns and greater, which are possibly paleodune surfaces. The most interesting component in the radar record is a region of ‘muted’ (attenuated) reflections or ,shadow’ zone 60-70 m wide extending from just west of 290 m to 366 m on line 305 N and from 260 m to 321 m on line 396 N. This zone of attenuated GPR reflections can be spatially correlated with the area of known hydrocarbon contamination, as determined from soil borings and hydrochemical studies. We observe that the signal strength begins to decrease not above the interpreted water table reflection, but right below it and remains low to the end of the record. No noticeable changes in the character of the GPR signals are observed above the saturated zone across the lines, except perhaps for minor interruptions of the water table reflection along line 305 N. This is consistent with observations on other lines over the entire site. When the positions of the pattern is spatially coincident with the mapped position of the plume from hydrochemical studies.
Figure 2. Two-dimensional display of representative GPR survey lines. The top image is the profile along line 305N and the bottom image is the profile along line 396N. note the area of attenuated GPR signal (shadow zone) over observed contamination. (scan length is 400nsec)

wpeF.jpg (109914 bytes)

Electrical Resistivity

Two types of resistivity survey arrays were utilized in this study, dipole-dipole profiling and vertical electrical soundings (Schlumberger array). The data were acquired using the Iris Syscal R2 resistivity system. Both data sets have been inverted to provide geoelectric sections and I -D models.

An east-west dipole-dipole resistivity (DDR) profile was collected along Line 305 N, using an "a" spacing of 10 meters, and n varying from 1 to 4. The apparent resistivities are presented and contoured to produce the apparent resistivity pseudosection of figure 4a. The DDR data were inverted using RES2DINV, which uses a non-linear least squares optimization technique for the inversion subroutine (Loke and Barker, 1996), to produce the inverted or model resistivity section of figure 4b. A finite-difference forward modeling subroutine is then used to calculate the apparent resistivity values of the model, which can then be compared with the field data. The field data and the inversion result presented in figures 4a and 4b show a region of low resistivity (high conductivity) extending from 288 m to 358 m with apparent resistivities less than 300 ohm-m close to the surface and extending to depth. Given the fact that there is very little change in topography along the line, the conductive body at depth most likely represents conductive groundwater and an impacted vadose zone. Further, the area of elevated conductivities is also coincident with the area of muted GPR reflections and the area of known hydrocarbon contamination of groundwater.

Vertical electrical soundings (VES) were taken at VES 01, VES

Figure 3. Map showing location of shadow zoned from the GPR profiles, indicationg the geophysically defined plume.

wpe18.jpg (32054 bytes)

Figure 4. (top) Dipole-dipole resistivity profile along line 305N,   (bottom) inversion results of data shown in (a).

wpe19.jpg (14341 bytes)
02 and VES 04 (see fig. 1 for sounding locations). VES data are plotted and inverted using SCHLINV (Merrick, 1977) as modified by Sauck (1990) which creates a 1-D model from the sounding data. The sounding curves and model results are presented in fig. 5.

Soundings VES 01 and VES 02 are over the area of known hydrocarbon contamination. During the inversion, we used the known average depth to the silty clay unit (19.8 m) and average depth to water table (4 m) to constrain the models shown in figures 5a and 5b. Our best results are obtained using a six layer case. Our model shows a very thin soil layer (<0.1m) with low resistivities of 60 to 210 ohm-m, followed by a high resistivity vadose zone of at least 4500 ohm-m. At the water table depth, a very low resistivity layer (<100 ohm-m) is encountered, occupying the upper several meters of the aquifer unit. This is the horizon sampled by most of the shallow monitoring wells, which show groundwater conductivities of greater than 500 [µS/cm (50mS/m), about double the background for the area. This is also the layer that we believe is imaged in the dipole-dipole resistivity pseudosection, and which is responsible for the attenuation of the GPR signal, causing the amplitude ‘shadow’ below the water table. Below this horizon is a unit with resistivities of 300-330 Ohm-m, which we interpret as clean, fresh-water saturated sands. Both curves bottom at a conductive unit with resistivities of 26-38 ohm-m, interpreted to represent the silty clay unit. This value is consistent with direct measurements of 26 Ohm-m made on outcrops of the silty clay unit two miles south of the FT-02 site, along the Au Sable River.
Figure 5. a) Vertical electrical sounding model for VES 01, b) model for VES 02, and c) model for VES 04. See fig.1 for VES locations.

wpe1C.jpg (29353 bytes)

Figure 6. Self potential (SP) anomaly map of the FT-02 site. Note the SSE trending zone of high SP anomaly values over the contaminated areas.

wpe1F.jpg (23042 bytes)

 The third sounding, VES 04 was outside the contaminated zone. The sounding, was inverted usine, a 5 layer model (fig 5c). The result from this location is different from the soundings obtained over the impacted areas (VES 01 and 02) in two respects. Firstly, the apparent resistivity of the vadose zone is considerably higher, and the inversion required a very high vadose zone resistivity of 20,000 ohm-m for a fit. This result complements the DDR results, which also showed lower vadose zone resistivities over plume areas, compared to offplume areas. Thus, it is apparent that the vadose zone above the contaminated ground water plume has been impacted well beyond the surface impact zone (donut-shaped area in fig. 1). Secondly, the low resistivity layer seen in sounding curves VES 01 and 02 below the saturated zone is absent. However, the overall aquifer interval is fit by resistivities of 290 ohm-m, consistent with uncontaminated aquifer values at the site. Finally, this curve also terminates with a very low resistivity layer of 35 Ohm-m, which we again interpret as the silty clay unit. The electrical resistivity results complement the GPR results and assist in the interpretation of the GPR data by providing important insights into the vertical resistivity distribution causing the attenuation of the GPR signals over the plume.

 Self Potential

Self Potential (SP) measurements were taken using two non-polarizing electrodes (Cu in CuSO4 solution) and a high-input impedance voltmeter. The field system measured the potential difference between a fixed reference electrode and a roving electrode placed at each grid node. The raw SP data were plotted and contoured using the Geosoft® mapping system (fig. 6). The SP data show a linear positive anomaly (+8 to + 25 mV) trending NW-SE compared to background SP values of -5 to 0 mV. This linear trend starts at the concrete fire containment basin (top of the survey area) between 250 m to 360 m and continues to the southeast consistent with the plume track. Values of +8 mV and above are observed over areas coincident with known groundwater hydrocarbon contamination, GPR shadow zones, and elevated conductivities on the DDR pseudo-section. The high SP values are apparently due to the electrochemical potential, which is a combined potential due to chemical concentration gradient and ion diffusion as would be expected for a more conductive groundwater plume interfacing with a cleaner groundwater (liquid junction) (Telford et al., 1990). Thus, this method provides further evidence that a conductive groundwater plume is located below the LNAPL free and residual product contamination zone and coincident with the dissolved hydrocarbon plume further downgradient. There is a secondary linear trend branching off to the east at approximately 220 N with a range of +8 to + 14 mV. The source of this anomaly is a previously unmapped buried utility line.

Discussion and Conclusions 

Organic contaminants like fuels and other hydrocarbons generally have a high electrical resistivity on the order of 101 ohm-m and low relative electrical permittivity, between 2-3. Consequently the contrast in physical properties between hydrocarbon contaminants and the surrounding medium should be readily detectable using electrical geophysical methods. Therefore, it is not surprising that the prevalent physical model among geophysicists is that the presence of hydrocarbon contaminants in the subsurface should produce measurable anomalies by acting as an insulating layer. This physical model has apparently been supported by controlled spill and laboratory experiments that report high resistivities and enhanced GPR reflections of the water table in areas coincident with known hydrocarbon contamination compared to non-contaminated areas (e.g., Campbell et al., 1996; Daniels et al., 1992; Redman et al., 1994).

However, GPR response over known hydrocarbon spill sites is contradictory to the observed controlled spill experiments. Previous investigations of uncontrolled hydrocarbon spill sites have reported a distinctive, ‘muted’/’fuzzy’ GPR response over volatile organic contaminant plumes (e.g., Daniels et al., 1992; Grumman and Daniels, 1995; Maxwell and Schmok, 1995). In their studies, the area of muted GPR response was observed to occur in the vadose zone, above the water table and capillary fringe, suggesting an association with contaminant vapor effects. While the cause of the ‘muted’ or ‘fuzzy’ GPR response coincident with LNAPL contaminant zones is not well understood, Daniels et al.(1992) have suggested that the attenuation of EM waves is caused by high loss tangent scattering resulting from small dispersed concentrations of hydrocarbons in the capillary fringe, or from contaminant vapor effects (Grumman and Daniels, 1995), while Maxwell and Schmok (1995) suggest diffuse reflection resulting from the effective roughness of water, gasoline, and soil mixtures,

In this study, we had excellent conditions such as shallow groundwater, clean, clay-free, homogeneous sands at the FT-02 site favorable for the detection of the organic contaminants using GPR and other geophysical techniques. The integration of different geophysical methods provided us with important insights to suggest possible explanations for the origin of the anomalies observed in our data. We have provided evidence in our GPR records over the FT-02 site to support the presence of attenuated GPR signals in areas coincident with hydrocarbon contamination, but with one major difference from those of previous studies. The GPR records from this site show a marked decrease in the strength of the GPR reflections and the presence of a shadow zone in the region immediately below the water table and not above it as was observed in the studies of Grumman and Daniels (1995) and Maxwell and Schmok (1995). In fact, except for minor interruptions of the water table reflection along line 305 N, no noticeable difference is seen in the water table reflection or vadose zone character across the profiles in the region coincident with known contamination (see fig. 2). This was not an artifact of the GPR recording process, as this system used a constant gain function for the entire survey line. However, the resistivity data present slightly different results. We see evidence of some increased conductivity within the vadose zone, where lower apparent resistivities are observed to be coincident with contaminated areas, compared to higher vadose zone resistivities in off plume areas. The exact reason for the absence of vadose zone contaminant effects within the GPR records is not known. Perhaps these results suggest that the resistivity method is a more sensitive method for detection of the shallow vadose zone conductivity enhancement than the GPR method, while the attenuation of the GPR reflections is an excellent mapping tool for enhanced conductivity below water table.

Although several hypotheses have been advanced to explain the ‘muted’ or ‘shadow’ or ‘fuzzy’ effect observed in GPR records over LNAPL spill sites, we present here an alterntive explanation for the effects observed at the FT-02 site. We suggest that the shadowing effect is simply an indication of radio wave attenuation due to elevated conductivity associated with the biodegradation of the LNAPL plume in the subsurface. As organic acids are produced by biodegradation of the hydrocarbon, the water in the plume area is able to dissolve ions from the surrounding host sediments resulting in waters with significantly higher dissolved solids content and higher conductivities in and below the LNAPL contaminated area relative to background. This explanation is supported by our DDR, VES, and SP results, which all showed a conductive body in areas coincident with the known hydrocarbon contaminant. Further, groundwater specific conductances measured from monitoring wells within the plume area show elevated conductivities (about twice the background) compared to off plume areas, while vadose zone gas sampling indicates the presence of CO2, H2S, H2, and CH4, evidence of active biodegradation (Mark Henry, pers. communication; Robbins et al., 1995).

At a given frequency, it is well known that attenuation of a radar pulse is primarily a function of earth conductivity. As conductivity increases, the depth of penetration of the pulse decreases (Telford et al., 1990). Hence, in areas of enhanced conductivities, there is less penetration resulting in the ‘shadow’ or ‘muted’ or ‘fuzzy’ appearance observed on the GPR records. Although it can be argued that the enhanced conductivities that we observe at the FT-02 site might be due to the fire retardant chemicals used to extinguish the fires during the fire training exercise, the proportion of foaming agent to the water is less than 1 %. Furthermore, our results are not unique to the FT-02 site. Previous investigators such as Monier-Williams (1995), Gajdos and Kral (1995), and Sauck and McNeil (1994) have all reported anomalous (higher) conductivities associated with areas of hydrocarbon contamination. In fact, in his study, Monier-Williams (1995) observed an obvious correlation between the presence of LNAPL and hiah apparent conductivity. He measured EM-31 conductivity values greater than 200 mS/m near wells with LNAPL contamination and corresponding groundwater sample conductivities of 2760 µS/cm (276 mS/m). At wells with no LNAPL contamination, EM-31 conductivity values of less than 50 mS/m and groundwater conductivities of 1300 µS/ cm (130 mS/m) were observed. He proposes that the anomalously high apparent conductivities may be due to interaction between the fuel-oil and the groundwater through possible formation of an emulsion and/or enhanced surface conductance at the fuel oil-water contact, or imbibition.

We submit that the high apparent conductivities observed in Monier-Williams’ study can be explained equally well by our previously stated hypothesis. Further, we have found support for our explanation from geochemical studies conducted at a crude oil pipeline spill near Bemidji, Minnesota (Baedecker et al., 1987; Cozzarelli et al., 1990; Bennett et al., 1993; Baedecker et al., 1993; and Eganhouse et al., 1993) which indicate the presence of highly conductive waters below some hydrocarbon (LNAPL) plumes. The occurrence of conductive groundwater is likely due to the release of ions (bicarbonate, sulfate, nitrate, iron, manganese, silica, and others) from aquifer solids by reaction with organic acids or carbonic acids derived from the biodegradation of the hydrocarbon compounds. A similar effect was reported by Banaszak and Fenalon (1988) for hydrocarbon contamination at the south end of Lake Michigan. Further, Bennett et al. (1988) showed in a laboratory study that large amounts of silica could be dissolved by organic acids, while McMahon and Chapelle (1991) showed that organic-rich silt/clay layers within aquifers supplied organic acids to adjacent parts of the aquifers.

Thus, we hypothesize that hydrocarbon spills (in this case LNAPLS) in the natural environment cause changes in bulk electrical properties with time, from resistive to conductive, due to biodegradation. These changes are largely due to production of secondary inorganic leachate by the increasingly acidic environment within and below the free/residual product zone. The leachate is intimately mixed with the free and residual product, and drains to the aquifer, forming a secondary inorganic plume generally coincident with the dissolved organic plume. Further, the enhanced conductivity observed in the vadose zone above the distal free/residual product plume (not directly impacted by downward hydrocarbon seepage in the spill zone) can be explained by acidification of soil moisture by rising CO2 gas generated by biodegradation of the hydrocarbon product. As in the free/residual product zone discussed earlier, the downward percolating meteoric waters, enriched in carbonic acid, leach ions from the native soil grains, make the vadose zone more conductive, and add to the leachate groundwater plume. If this model is true, then it is not surprising that the current physical model, supported by controlled spill experiments, is inadequate in explaining results obtained from investigations of mature, uncontrolled spill sites. Geophysical studies of controlled LNAPL "spills" have all been made at early times, days to months after the initiation of the experiment, and perhaps may represent only the early-time behavior of organic spills. However, the field characterizations have been late-time studies, usually after several to many years have elapsed following the spill and when active biodegradation has taken place. Also, the controlled experiments have been protected from seasonal temperature and rainfall effects, and are often made in homogeneous, washed and sieved sands, not taking into consideration the intrinsic variability of conditions at geologic sites such as seasonal groundwater dynamics, bacterial degradation, and biochemical reactions. Hence, it is likely that the time elapsed after the spill and before the geophysical survey is ma e is a crucial factor. Early surveys (within the first months or year of the spill) may yield results and interpretations consistent with the controlled spill experiments but entirely contrary to those from much later surveys. Further, there may well be an intermediate time at which there will be little or no anomalous response, midway in the transition from resistive to conductive behavior of the system.

Field studies by both geochemists and geophysicists have rarely been coordinated in the past. Yet it is apparent that the key to unraveling the anomalous geophysical behavior of these contaminants may lie in the understanding of their interactions with their host matrix as well as the biogeochemical processes that subsequently alter them in the subsurface. Further, as most of this work is done from the surface, the inherent ambiguities in interpreting exact depths to changes in physical properties has led to considerable uncertainty in the vertical location of critical boundaries. Coupled with the natural variation from site to site, and the lateral and vertical migration of the hydrocarbon products with time, it is understandable why there are many different geophysical descriptions of these hydrocarbon spill sites. Hence, the success of geophysical investigations of such sites depends upon using the correct model(s) of physical properties contrasts in the subsurface, and then applying the correct methods using the appropriate survey parameters.

Acknowledgments

The authors would like to thank M. Barcelona for giving us access to the FI’-02 site. Logistical support during field work and hydrochemical data were provided by Mark Henry and the Univ. of Michigan NCIBRD staff. We thank them most sincerely for all their help. Appreciation is also extended to all our field helpers, especially K. Sprietzer, B. Pelland and D. Meyers. Eliot Atekwana helped with the preparation of the figures. The many constructive comments of reviewers Anthony Endres and John Greenhouse significantly improved the manuscript and are greatly appreciated. Equipment used in this project was funded by NSF grant number DUE9550874.

References

Baedecker, M.J., Cozzarelli, I.M., and Hopple, J.A., 1987, The composition and fate of hydrocarbons in a shallow glacial-outwash aquifer. United States Geological Survey Open-File Report 87-109.

Baedecker, M.J., Cozzarelli, I.M., Eganhouse, R.P., Siegel, D.1, and Bennett, P.C., 1993, Crude oil in a shallow sand and gravel aquifer - III. Biogeochemical reactions and mass balance modeling in anoxic groundwater; Applied Geochen-dstry, vol. 8, pp. 569-586.

Banaszak, K.J. and J.M. Fenelon, 1988, Water Quality in a Thin Water-table Aquifer Adjacent to Lake Michigan Within a Highly Industrialized Region of Indiana; Proc. of the Symposium on The Great Lakes: Living With North America’s Inland Waters; American Water Resources Association, pp. 247-257.

Bennett, P.C., M.E. Melcer, D.I Siegel, and J.P. Hassett, 1988, The Dissolution of Quartz in Dilute Aqueous Solutions of Organic Acids at 25’C; Geochimica et Cosmochimica Acta, Vol. 52, pp. 1521-1530.

Bennett, P.C., D.E. Siegel, M.J. Baedecker, and M.F. Hult, 1993,

Crude oil in a shallow sand and gravel aquifer - I. Hydrogeology and inorganic geochemistry; Applied Geochemistry, Vol. 8, pp. 529-549.

Brewster, M.L. and Annan, A.P., 1994, Ground-penetrating radar montoring

of a controlled DNAPL release: 200 MHz radar; Geophysics, Vol. 59, No. 8, pp. 1211-1221.

Campbell, D.L., Lucius, J.E., Ellefsen, K.J., and Deszcz-Pan, M., 1996, Monitoring of a controlled LNAPL spill using Ground Penetrating Radar. Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP’ 96), Keystone, CO., pp. 511-517.

Cozzarelli, I.M., R.P. Eganhouse, and M.J. Baedecker, 1990, Transformation of Monoaromatic Hydrocarbons to Organic Acids in Anoxic Groundwater Environment; Environ. Geol. Water Sci., v. 16, No. 2, pp. 135-141.

Daniels, J.J., Roberts, R., and Vendl, M., 1992, Site Studies of Ground Penetrating Radar for Monitoring Petroleum Product Contaminants; Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP’ 92), Oakbrook, ]IL, pp. 597-609.

De Ryck, S.M., Redman, J.D., and Annan, A.P., 1993, Geophysical Monitoring of a Controlled Kerosene Spill; Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP’93), San Diego, CA, pp. 5-20.

Eganhouse, R.P., M.J. Baedecker, I.M. Cozzarelli, G.R. Aiken, K.A.

Thom, and T.F. Dorsey, 1993, Crude oil in a shallow sand and gravel aquifer - II. Organic geochemistry; Applied Geochemistry, Vol. 8, pp. 551-567.

Endres, A.L. and Redman, J.D., 1996, Modeling the electrical properties of porous rocks and soils containing immiscible contaminants; Jour. of Env. and Eng. Geophysics, Vol. 0, No. 2, pp. 105-112.

Endres, A.L. and Greenhouse, J.P., 1996, Detection and monitoring of chlorinated solvent contamination by thermal neutron logging; Ground Water, Vol. 34, No. 2, pp. 283-292.

Gajdos, V. and Krdl, V., 1995, Influence of Hydrocarbon Pollution to Soil Conductivity; Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP ’95), Orlando, FL, pp.785-789.

Grumman, D.L. and Daniels, J.J., 1995, Experiments on the Detection of Organic Contaminants in the Vadose Zone; Journal of Environmental & Engineering Geophysics. Vol. 0. No. 1. pp. 31-38.

King, T.V.V. and G.R. Olhoeft, 1989, Mapping organic contamination by detection of clay-organic processes; Proceedings AGWSE/NWWA/API Conf. on Petroleum Hydrocarbons and Organic Chemicals in Ground Water - Prevention, Detection, and Restoration, pp. 627-640.

Loke, M.H., and Barker, R.D., 1996, Rapid least-squares inversion of apparent resistivity pseudo section by a quasi-Newton method, Geophysical Prospecting, 44, p.131-152.

Maxwell, M., and Schmok, J., 1995, Detection and Mapping of an LNAPL Plume Using GPR: A Case Study; Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP’95), Orlando, FL, pp. 15-23.

Mazdc, O., Benes, L., Landa, I., and Maskova, A., 1990, Determination of the Extent of Oil Contamination in Groundwater by Geoelectrical Methods; in Geotechnical and Environmental Geophysics (S.H. Ward, ed.), Vol. H, pp. 107-112.

McMahon, P.B. and Chapelle, F.H., 1991, Microbial production of organic acids in aquitard sediments and its role in aquifer geochemistry; Nature, vol. 349, pp. 233-235.

Merrick, N.P., 1977, A computer program for the inversion of Schlumberger sounding curves in the apparent resistivity domain, Hydrogeological Report # 197715, Water Resources Commission, N.S.W., Australia, 36 pp.

Monier-Williams, M, 1995, Properties of Light Non-aqueous Phase Liquids and Detection Using Commonly Applied Shallow Sensing Geophysical Techniques; Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP ’95), Orlando, FL, pp. I 13.

Olhoeft, G.R., Lucius, J.E. and Bochicchio, R.M., 1988, Scales of length and connectivity from ground penetrating radar; in U.S. Geological Survey Toxic Substances Hydrology Program Proceeding of the Technical Meeting, Phoenix, AZ (eds. G.E. Mallard and S.E. Ragone), U.S. Geol. Surv. Water Resources Investigations Report 88-4220.

Redman, J.D. and De Ryck, S.M., 1994, Monitoring Non-aqueous Phase Liquids in the Subsurface with Multilevel Time Domain Reflectometry Probes; from Symposium and Workshop on Time Domain Reflectometry in Environmental, Infrastructure, and Mining Applications; U.S. Bureau of Mines, Special publication SP 19-94, pp. 207-214.

Redman, J.D., DeRyck, S.M., and Annan, A.P., 1994, Detection of LNAPL pools with GPRl Theoretical modeling and surveys of a controlled spill; Proceedings of the Fifth International Conference on Ground penetrating Radar (GPR’94), Kitchener, Ontario, pp. 1283-1294.

Robbins, G.A. McAninch, B.E., Gavas , F.M., and Ellis, P.M., 1995, An evaluation of soil-gas surveying for H2S for locating subsurface hydrocarbon contamination; Ground Water monitoring and Remediation, pp. 124-132.

Sauck, W.A., 1990, Modification of the SCHLINV program, Internal Report, Institute for Water Sciences, Western Michigan University, Kalamazoo, MI.

Sauck, W.A., and J. McNeil, 1994, some Problems Associated with GPR Detection of hydrocarbon Plumes (abs.); Fifth International Conf. on Ground Penetrating Radar (GPR’94); Kitchener, Ontario; Abstracts for poster Papaer, p. 34.

Scheider, G.W., and Greenhouse, J.P., 1992, Geophysical detection of perchloroethylene in a sandy aquifer using resistivity and nuvlear logging techiques. Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems (SAGEEP’92), Oakbrook, IL, pp. 619-628.

Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990, Applied Geophysics, Cambridge University Press, Cambridge, U.K., 770pp.