1995 MGLS Symposium

Santa Fe, New Mexico

ABSTRACT

The Passaic Formation consists of gradational sequences of mudstone, siltstone, and sandstone, and is a principal aquifer in central New Jersey. A hydrologic investigation designed to evaluate anisotropic flow characteristics within this aquifer was initially undertaken in 1966 at a study site north of Trenton (Vecchioli et al., 1969). The existing wells associated with this study are now being utilized in a research project concerned with solute transport in fractured sedimentary rocks. These wells also provide an opportunity to return to this site with more sophisticated tools and techniques in order to expand upon the original study. A comprehensive suite of geophysical logs was obtained in ten closely spaced wells roughly 46 m in depth with a nominal diameter of about 15 cm. Lithologically responsive logs such as electrical resistivity and natural gamma activity provided vertically continuous profiles of depositional sequences that were used for stratigraphic correlation across the site. Inspection of individual fractures intersecting the wellbores by means of caliper and acoustic televiewer logs supplied the basic data from which a statistical analysis of the total fracture population was performed. Fluid-property logs (temperature, electrical conductivity, vertical velocity) delineated the transmissive intervals in each well and allowed the permeable fractures to be distinguished from the general population. About 18 percent of all fractures displayed detectable transmissivities.

Results show that a seemingly complex, heterogeneous network of fractures can be reduced to a simplified series of fracture subsets, with two distinct fracture types emerging from the analysis. Bedding-plane partings that strike N84°W and dip 20° to the north are the most numerous. These features exhibit widely varying hydraulic transmissivi-ties that average roughly 5 m_/day and that generally diminish in magnitude and frequency with depth. High-angle fractures strike subparallel to the bedding planes at N79°E but dip steeply (71°) to the south, thereby forming fracture planes orthogonal to the bedding planes. Transmissivities associated with these fractures are, on average, about half those of the bedding-plane partings. The intersections of these two fracture types form linear conduits that retain the common east-west strike direction and have average transmissivity values that are intermediate between those of the individual fractures. The geophysical logging results portray a distinct hydrogeologic structure within this aquifer that is controlled by fracture orientation and type. Since all permeable features strike approximately east-west, local transmissivity should be strongly anisotropic. However, this directional dependence is minimal near the surface because flow is controlled primarily by the gently dipping bedding-plane partings. Horizontal anisotropy becomes more significant with increasing depth as the high-angle fractures become more dominant hydrologically.

Characterizing the hydrogeologic properties of fractured bedrock aquifers is a complex and challenging task. Three-dimensional fluid transport through a heterogeneous network of fractures is cumbersome to describe in detail in the field, often requiring multi-disciplinary and multi-scale approaches consisting of drilling and sampling, surface geophysics, borehole geophysics, geologic mapping, and pumping and tracer tests. An overview of this type of general testing strategy is presented by Paillet (1993a).

Numerical models have attempted to conceptualize the heterogeneous transport properties of rocks by incorporating generalized field and laboratory observations into their simulations of preferential flow paths. Various approaches have been applied with some degree of success, such as percolation models (e.g., Charlaix et al., 1987; Ewing and Gupta, 1993), network models (e.g., Rouleau and Gale, 1987; David et al., 1990), and statistical averaging models (e.g., Gueguen and Dienes, 1989; Durner, 1994). However, there is no substitute for field experiments to supply the basic assumptions upon which these models are based.

The following study describes the results of geophysical-logging operations in ten closely spaced wells penetrating the Passaic Formation of central New Jersey. These field data provide substantial information regarding the local hydrostratigraphy of this sedimentary-rock aquifer, including the influence of specific types of fractures on the three-dimensional distribution of transmissivity. The results demonstrate the diagnostic value of these particular field methods and will be used as conceptual guidelines in an ongoing study of the solute transport characteristics of this aquifer. This work is being partially funded by the New Jersey Department of Environmental Protection, Division of Science and Research.

The study site is located in Hopewell Township, New Jersey, approximately 16 km north of Trenton, on a 247-ha nature reserve owned by the Stony Brook-Millstone Watershed Association (Figure 1). Hopewell Township is situated in the Newark Basin, an elongate (210 by 55 km), northeast-southwest trending fault trough filled with late Triassic and early Jurassic fluvial and lacustrine sediments (Olsen, 1980; Houghton, 1990). The sedimentary rocks of the Newark Basin are similar to deposits in about 30 inland rift basins along the east coast from South Carolina to Nova Scotia and include important regional aquifers such as the Stockton and Passaic Formations.

The study site is underlain by the late-Triassic Passaic Formation, consisting of red arkosic mudstones, siltstones, and fine-grained sandstones. Three cores of stratigraphically similar Passaic-Formation rock, including one collected about 16 km west of the site, had matrix porosities of 3-5 percent and hydraulic conductivities ranging from undetectable to 3 x 10^-11 cm/sec (Core Laboratories, written communication, 1991). The Hopewell Fault is a major regional structure and lies approximately 2 km northwest of the study area. Associated with this fault are secondary features, including a northwest-plunging syncline centered at a small pond on the east side of the site (Figure 1). On the western limb of the broad syncline, bedding planes strike approximately east-west and dip gently to the north.

Thirteen observation wells were drilled by the air-rotary method at the site in 1966 (prior to the construction of the pond) for a study of regional anisotropic flow in the Passaic Formation (Vecchioli et al., 1969). The wells are constructed with about 6 m of steel surface casing, have a nominal diameter of about 15 cm, and are about 46 m deep, except for Well 6 which was drilled to 61 m in order to penetrate the same bedding planes intersected by Well 1. All of the wells were found to be in good condition upon re-opening in 1994, except for Well 5 which had collapsed at a depth of 21 m shortly after its construction.

GEOPHYSICAL LOG ANALYSIS

These existing wells provide convenient access to an important regional resource and present an opportunity to return to the site with more sophisticated tools and techniques in order to expand upon the original study. Consequently, wells were re-entered in the summer and fall of 1994 and a variety of geophysical logging tools were employed to examine in greater detail the hydrologic characteristics of this fractured sedimentary-rock aquifer. All of the wells shown in Figure 1 were logged except for Wells 5, 7, and 12, which were inaccessible with our field equipment. The suite of geophysical logs consisted of caliper, temperature, fluid conductivity, formation electrical resistivity, natural gamma activity, acoustic borehole televiewer (BHTV), and heat-pulse flowmeter under ambient and pumping conditions. The heat-pulse flowmeter is a custom logging instrument developed by the U.S. Geological Survey (Hess, 1982; 1986). This tool is a high-resolution flow sensor capable of detecting vertical fluid movement in a well as slow as 1 cm/min; it has been successfully used to determine flow conditions in conjunction with numerous ground-water investigations (Hess and Paillet, 1990). Detailed information regarding the other tools and their principles of operation may be found in Keys (1990).

Composites of selected logs from four wells are presented in Figure 2. Wells 1, 3, and 10 were chosen for display because they span almost the entire length of the site and are located along the general directon of strike, while Well 4 (in pond) is perpendicular to strike and its lowermost section intersects bedding planes observed immediately below surface casing in Well 1. Although the logs from only these four wells are presented here for the sake of brevity, they are typical of logs from all ten wells. The caliper log provides an approximate measure of well diameter and helps identify zones of weak and fractured rock. The fluid temperature and conductivity logs yield information regarding water quality and are useful in locating zones of fluid exchange between the borehole and the formation. In numerous cases, evidence of fluid movement is amplified in the temperature gradient log (slope of the temperature log). Shifts in these fluid properties with depth, such as is clearly observed in Well 10 at 17 m (Figure 2c) and in Well 4 at 19 m (Figure 2d), indicate an exchange of fluid between the well and the surrounding aquifer and, consequently, the proximity of transmissive zones.

The plots of BHTV fracture score presented in Figure 2 represent a qualitative measure of fracture frequency and aperture size as determined from inspection of the acoustic televiewer logs. The BHTV provides a magnetically oriented image of the borehole wall, and its design and theory of operation are described by Zemanek et al. (1970). An example of a televiewer log developed from the amplitude of reflected acoustic pulses is shown in Figure 3. This image was obtained from Well 2 across the depth interval 10.9 to 13.5 m; it is a planar, "unwrapped" representation of a cylindrical surface. Also shown is the simple geometric exercise used to compute the strike and dip of each individual fracture intersecting the wellbore. The BHTV fracture scores presented in Figure 2 were determined by assigning a value of 1 to thin, discrete fractures and progressively increasing the score to 5 where fractures were wide and extensive, resulting in a substantially damaged wellbore. These scores were averaged over 2-m intervals and are plotted versus depth as bar graphs.

Heat-pulse flowmeter measurements were initially obtained in all wells under ambient conditions and none indicated any fluid movement above the detection limit of the tool. Subsequently, the flowmeter was left in the static well and pumping was begun either from the surface through a hose or from a submersible pump lowered below the water table but above the submerged logging tool. Flow measurements were recorded at various depths as the tool was raised and lowered in the well; vertical flow, pumping rate, and drawdown were concurrently monitored as a function of time. Particular zones where water entered the well during pumping were identified and the attendant percentages of the total volumetric pumping rate are plotted versus depth in Figure 2. These log composites delineate the highly fractured intervals and the highly transmissive intervals, which are not necessarily coincident.

The two log types that are the most sensitive and responsive to lithology are the natural gamma and the formation resistivity logs. High gamma activity, in counts per second (cps), is typically associated with the presence of fine-grained units that tend to accumulate radioisotopes through adsorption and ion-exchange processes. In arkosic sedimentary rocks such as those encountered in this Mesozoic rift basin, high gamma counts may also correspond locally to potassium feldspar in sandstone units. Conversely, sodium-calcium feldspar has no radioactivity and produces no gamma response. Electrical resistivity logs respond to lithologic variations as well as to changes in pore-water quality and porosity. Low-porosity, coarse-grained sandstones exhibit high resistivities, whereas fine-grained deposits are more efficient electrical conductors. Both types of logs serve as good lithologic indicators and can be effective for stratigraphic correlation among wells.

Wells 10, 2, 1, and 3 form a west-east transect across most of the study site and their resistivity logs are plotted as a function of elevation in Figure 4a. High-resistivity spikes, indicative of low-porosity sandstones, can be correlated across the site to identify a slight easterly component of dip of about 4°. Similarly, a transect of natural gamma logs is presented in Figure 4b and trends in these latter logs are generally the inverse of those observed in their resistivity counterparts. However, small-scale variability is more evident in the gamma profiles because these reflect not only shifts associated with sandstone-siltstone sequences, as do the resistivity logs, but also the additional effects of sporadic potassium-feldspar input. Again, a gentle 4° dip component to the east is identified.

Detailed inspection of BHTV logs obtained from all ten wells was performed to determine individual fracture strike and dip in the manner demonstrated in Figure 3b. Orientations were corrected for local magnetic declination. A non-horizontal fracture appears as a sinusoid in the planar image and the fractures shown intersecting Well 2 (Figure 3a), for example, are dipping to the south (low point). Typical rosette and lower-hemisphere stereographic plots generated from this exercise are presented in Figure 5 for three wells. The equal-area stereographic projections are a representation of pole densities converted to shaded contours using spherical Gaussian statistics (Kamb, 1959). Contour intervals are sequenced in increments of the standard deviation s.

Statistical analyses of fracture orientations computed by means of a Bingham axial distribution (Mardia, 1972; Fisher et al., 1987) were performed and magnitudes of the eigenvalues l and eigenvectors (strike and dip) are listed for each well in Table 1. The value of l is normalized to 1.0 and is considered to be a measure of the relative concentration of poles associated with a statistically significant fracture set. The eigenvector represents the orientation of a representative fracture plane within that set. In each of the three examples shown in Figure 5, two primary fracture subsets emerge.

Combining fracture data from the ten wells (total of 280 fractures) yields the statistical diagrams presented in Figure 6. Slight hole-to-hole variations in fracture populations are melded into these final cumulative plots which clearly depict the two primary subsets. As is listed in Table 1, the predominant subset (l = .532) strikes N84°W (or 276°) and dips 20° to the north. Its slight easterly component of dip is manifested in the orientation of stratigraphic contacts shown in Figure 4. This fracture population represents bedding-plane partings associated with mudstone, siltstone, and sandstone sequence. The second major fracture subset (l = .411) strikes N79°E (or 79°) and dips 71° to the south. This group of high-angle fractures cuts the Passaic Formation roughly orthogonal to the bedding planes. These two fracture populations were identified by Vecchioli et al. (1969), and the statistical analyses formulated from the BHTV images provides a quantitative confirmation.

It should be noted that a third, albeit very minor, fracture set is identified from the Bingham-distribution analysis (see Table 1). It comprises less than 6 percent of the total fracture population and strikes approximately perpendicular to the two primary fracture sets which are roughly parallel. However, because these fractures are nearly vertical, their presence must be considered in terms of the inherent undersampling of steeply dipping fractures that occurs from a vertical borehole.

The probability of intersecting a near-horizontal fracture with a vertical borehole is very high, whereas the probability of intersecting a near-vertical fracture with a vertical borehole is virtually zero. To account for this bias, a statistical correction based on dip angle may be applied to fracture populations. This adjustment predicts the frequency of fractures of a particular orientation that may be intersected by a well that is drilled normal to its plane. This correction was introduced by Terzaghi (1965) and has been applied in numerous fractures studies (e.g., Barton and Zoback, 1992). The number of observed fractures having a given dip angle q is multiplied by 1/cosq to arrive at the number of fractures probably present.

This probability correction was performed with the cumulative fracture population collected from the ten wells and results are shown in Figure 7. A histogram of dip angle versus frequency derived from the original data set is presented in Figure 7a and clearly depicts the bimodal distribution defined by the fracture statistics. When these data are corrected for probability of intersection by a vertical well, the bimodal pattern remains but the number of steeply dipping fractures increases dramatically (Figure 7b). Based on this simple exercise, the third minor fracture set (l = .057) composed of near vertical fractures appears to be more significant than the magnitude of its eigenvalue implies.

Most of the permeable fractures can be distinguished from the general fracture population by means of the flowmeter logs obtained while pumping. Typically, vertical fluid flow was measured in the wells at approximately 3-m intervals. If fluid exchange was detected within that section, the BHTV log was examined to locate the source fracture. In cases where the 3-m depth interval encompassed several fractures, complementary temperature and fluid conductivity logs were inspected for sudden shifts in fluid properties in order to isolate and identify the permeable fracture within the group (see examples of diagnostic temperature and conductivity shifts in Figure 2).

A total of 51 permeable fractures were reliably identified, accounting for about 18 percent of the total fracture population. In eight cases, fluid exchange could not be attributed to a unique fracture because of a highly damaged wellbore and a lack of corroborating responses from the fluid-property logs. In these instances, no specific fracture was assigned to the transmissive zone. The orientations of the remaining 43 permeable fractures are represented statistically in the diagrams of Figure 8 and are listed in Table 1. These results demonstrate that the statistical distribution of this small set of permeable fractures is almost identical to that of the overall fracture population.

The analysis of permeable fractures was extended beyond simple detection and identification to the quantification of the hydraulic properties of individual fractures. This was accomplished by performing the flowmeter-pumping technique at the study site (Morin et al., 1988; Molz et al., 1989), with slight modifications in methodology proposed by Kabala (1994) to consider storage effects. Transient drawdown in each well during pumping was recorded and volumetric flow across each permeable interval was concurrently measured at several times prior to the attainment of quasi-steady state conditions. These data were subsequently used to compute values of transmissivity and storativity for each individual permeable fracture. According to the double-flowmeter analysis reported by Kabala (1994),

T_i = K_ib_i = [Q_i/4p(s_i2-s_i1)] ln(t_i2/t_i1) [1]

S_i = S_sib_i = [0.562Q_i/pr_w_(s_i2-s_i1)]t_i1(t_i1/t_i2)^si_^(si_^-si_⁾ ln(t_i2/t_i1) [2]

where	Ti	=	transmissivity of layer i (L_/T),
	Ki	=	hydraulic conductivity of layer i (L/T),
	bi	=	thickness of aquifer layer i (L),
	si1	=	drawdown for layer i at time 1 (L),
	si2	=	drawdown for layer i at time 2 (L),
	Qi	=	(Qi1 + Qi2)/2
		=	volumetric flow averaged over times 1 and 2 (L_/T),
	ti1	=	time 1 associated with measurements at layer i (T),
	ti2	=	time 2 associated with measurements at layer i (T),
	Si	=	storativity of aquifer layer i (dimensionless),
	Ssi	=	specific storage of layer i (L-_), and
	rw	=	radius of well (L).

This analysis is based on the Theis (1935) solution and is valid when dimensionless time (r_w_S_si/4K_it) < 0.05.

Although fracture aperture can be measured from inspection of the BHTV acoustic images, it is uncertain how useful or accurate these estimates really are. Drilling effects can artificially enlarge fractures at the borehole wall. Also, permeable fractures are not simply formed by parallel plates, are not always isolated features, and may vary in thickness as they propagate away from the well and intersect other fractures. Consequently, it is difficult to assign a representative aperture width to a fracture, or an equivalent thickness to an aquifer layer i (b_i in the equations above). Thus, hydraulic properties are presented in terms of T_i and S_i and are not converted to values of K_i and S_si.

This flowmeter-pumping method provides a quantitative estimate of the small-scale vertical distribution of horizontal transmissivity across a range of approximately two orders of magnitude. This constraint is primarily imposed by the resolution limitation of the heat-pulse flowmeter, which is capable of measuring vertical fluid velocities across a similar range. A sample calculation is presented for a permeable zone in Well 1 located at a depth of 30.8 m. According to Figure 2a, approximately 12 percent of the total pumping rate of 51 L/min is entering the well at this depth. From the attendant drawdown data,

	si1	=	0.63 m at time	ti1	=	3.0 minutes
	si2	=	1.00 m at time	ti2	=	29.0 minutes
	Qi1	=	6.1 L/min	Qi2	=	6.1 L/min

Substituting into [1] and [2],

T_i = 3.0x10^-_ m_/min S_i = 7.5x10^-_

Average transmissivity and storativity values for the permeable fractures identified in this study are listed in Table 2. Of the 43 permeable fractures whose orientations were discernible (Figure 8), four could not be assigned transmissivity values because of equipment malfunctions in the field that produced inaccurate drawdown data or flowmeter measurements. An additional eight pairs of bedding-plane partings and high-angle fractures (16 total fractures) were categorized as forming eight fracture intersections. Thus, 31 out of the original total of 51 permeable fractures could be characterized in terms of both orientation and hydraulic properties.

Values of T range from about 10^-4 to 10^-_ m_/min in this study and are typical of fracture transmissivities reported by others (e.g., Hsieh et al., 1985; Lapcevic et al., 1993). Transmissivities associated with each fracture type are presented as a function of depth in Figure 9. Bedding-plane partings exhibit the widest range of transmissivities (two orders of magnitude) that diminish in magnitude and in frequency with depth. The most transmissive fractures identified at this site are bedding-plane partings observed near the surface, but no permeable partings were detected below about 35 m. High-angle fractures have a slightly narrower range of transmissivity values with no clear depth correlation, whereas fracture intersections have a very narrow range of T (< one order of magnitude) that is independent of depth.

Values of S are larger than expected for the case of a thin, confined aquifer and this overestimation may be the product of several factors. The local presence of surficial fractures intersecting the water table allows the specific yield of these fractures to contribute to the total storativity. In addition, the proximity of other open boreholes in hydraulic communication with the permeable fractures imposes significant borehole storage effects which increase local storativity (Paillet, 1993b). A third factor stems from the effects of borehole rugosity, as illustrated in the caliper logs (Figure 2). The formulation of Kabala (1994) assumes a uniform well diameter and, consequently, a constant wellbore storage with depth. Wellbore enlargements produce excess volumetric storage and artificially increase estimates of storativity in this type of analysis.

This field study demonstrates the scientific and practical value of applying geophysical logging to a ground-water investigation in a fractured sedimentary-rock environment. Descriptive logs such as electrical resistivity and natural gamma provided continuous vertical profiles of lithologic variations and first-order stratigraphic correlation across the site. Caliper and acoustic televiewer logs were essential to the detailed characterization of fractures, thereby supplying the basic data required to conduct a statistical analysis of the total fracture population within this aquifer. Finally, fluid-property logs delineated the transmissive intervals in each well and allowed the permeable fractures to be distinguished from the general population.

This comprehensive collection of logs supplies the field data necessary to develop a conceptual understanding of the hydrologic structure of the Passaic Formation at this site. Within this context, a statistical analysis of fracture orientations reduces a seemingly complex, heterogeneous network of fractures to an organized series of fracture subsets. Fracture statistics listed in Table 1 indicate that bedding-plane partings comprise the most abundant fracture subset and the most abundant permeable fracture subset; their relative concentration was identical for both categories (l = .532). However, this apparent predominance of gently dipping features is likely skewed by a systematic undersampling of near-vertical fractures by a vertical well (Figure 7).

Computed values of transmissivity are typical of fracture transmissivities reported elsewhere. Sedimentary rocks with bulk porosities of roughly 5 percent typically have matrix transmissivities on the order of 10^-8 m_/min or less (Nelson, 1994). Correspondingly, the flow of water through this aquifer is controlled and dictated by specific fractures that have transmissivities four or more orders of magnitude greater than that of the unfractured rock matrix. The transmissivities of the bedding-plane partings exhibit the widest range of values and tend to diminish in magnitude and in frequency as a function of depth. This behavior is probably a mechanical response to the accumulation of overburden stress acting roughly perpendicular to the fracture planes and systematically closing them. The other fracture types show no such dependence on depth. Computed values of fracture storativity are unrealistically large for the case of a thin, confined aquifer and this overestimation is attributed to 1) local, surficial fractures intersecting the water table, 2) the proximity of other open boreholes that are in hydraulic communication, and 3) artificial borehole rugosity effects.

Based on statistical analyses of fracture orientations, local aquifer transmissivity might appear to be highly anisotropic due to the shared east-west strike of the bedding-plane partings and the high-angle fractures, accented by the conductive linear fractures produced by their intersections. However, this directional dependence is not evident near the surface because flow is controlled by transmissive, subhorizontal bedding-plane partings that permit north-south fluid movement. Average transmissivity decreases with depth, yet horizontal anisotropy becomes more significant as the high-angle fractures become more dominant hydrologically. Preliminary hydraulic testing between wells at this site indicates that transmissivities are greater along bedding planes than across them, inferring that these steeply dipping fractures are not necessarily continuous vertically. Houghton (1990) reports that near-vertical structural fractures are seen to traverse entire coarse-grained units, but often terminate at shaly beds.

Finally, it should be recognized that many other fractures of widely varying orientations were identified in this analysis and that transmissivities smaller than about 10^-4 m_/min could not be determined because of resolution limitations of field equipment. Although many of these fractures may be considered hydraulically insignificant because of poor interconnectivity (e.g., Rouleau and Gale, 1987), aquifer transmissivities less than 10^-4 m_/min are not trivial and may represent significant volumetric exchange when regarded over larger spatial and longer time scales.

Acknowledgments

We are grateful to the Stony Brook-Millstone Watershed Association for allowing us to conduct these field studies on its land preserve. We thank R. Almendinger for his generous distribution of the software package STEREONET.

REFERENCES

Barton, C.A., and M.D. Zoback, Self-similar distribution and properties of macroscopoic fractures at depth in crystalline rock in the Cajon Pass Scientific Drill Hole, J. Geophys. Res., 97, 5181-5200, 1992.

Charlaix, E., E. Guyon, and S. Roux, Permeability of a random array of fractures of widely varying apertures, Transp. Porous Media, 2, 31- 43, 1987.

David, C., Y. Gueguen, and G. Pampoukis, Effective medium theory and network theory applied to the transport properties of rocks, J. Geophys. Res., 95, 6993-7005, 1990.

Durner, W., Hydraulic conductivity estimation for soils with heterogeneous pore structure, Water Resour. Res., 30, 211-223, 1994.

Ewing, R.P., and S.C. Gupta, Modeling percolation properties of random media using a domain network, Water Resour. Res., 29, 3169-3178, 1993.

Fisher, E.H., G. Lewis, and R. Embleton, Statistical Analysis of Spherical Data, Cambridge Univ. Press, London, 332 p., 1987.

Gueguen, Y., and J. Dienes, Transport properties of rocks from statistics and percolation, Math. Geol., 21, 1-13, 1989.

Hess, A.E., A heat-pulse flowmeter for measuring low velocities in boreholes, U.S. Geol. Surv. Open-File Rept. 82-699, 40 p., 1982.

Hess, A.E., Identifying hydraulically conductive fractures with a slow- velocity borehole flowmeter, Can. Geotech. J., 23, 69-78, 1986.

Hess, A.E., and F.L. Paillet, Applications of the thermal-pulse flowmeter in the hydraulic characterization of fractured rocks, Am. Soc. Test. Mater. Special Technical Publication 1101, 99-112, 1990.

Houghton, H.F., Hydrogeology of the early Mesozoic rocks of the Newark Basin, N.J., in Proceedings: Aspects of Groundwater in New Jersey, Kroll, R.L. and Brown, J.O., eds., Geological Assc. of New Jersey 7^th Annual Meeting, E1-E36, 1990.

Hsieh, P.A., S.P. Neuman, G.K. Stiles, and E.S. Simpson, Field determination of the three-dimensional hydraulic conductivity tensor of anisotropic media 2. Methodology and application to fractured rocks, Water Resour. Res., 21, 1667-1676, 1985.

Kabala, Z.J., Measuring distributions of hydraulic conductivity and specific storativity by the double flowmeter test, Water Resour. Res., 30, 685-690, 1994.

Kamb, W.B., Ice petrofabric observations from Blue Glacier, Washington, in relation to theory and experiment, J. Geophys. Res., 64, 1891- 1910, 1959.

Keys, W.S., Borehole geophysics applied to ground-water investigations, U.S. Geol. Surv. Tech. Water Resour. Investigations, E2, 149 p., 1990.

Lapcevic, P.A., K.S. Novakowski, and F.L. Paillet, Analysis of flow in an observation well intersecting a single fracture, J. Hydrology, 151, 229-239, 1993.

Mardia, K.V., Statistics of Directional Data, Academic Press, London, 357 p., 1972.

Molz, F.J., R.H. Morin, A.E. Hess, J.G. Melville, and O. Guven, The impeller meter for measuring aquifer permeability variations: evaluation and comparison with other tests, Water Resour. Res., 25, 1677-1683, 1989.

Morin, R.H., A.E. Hess, and F.L. Paillet, Determining the distribution of hydraulic conductivity in a fractured limestone aquifer by simultaneous injection and geophysical logging, Ground Water, 26, 587-595, 1988.

Nelson, P.H., Permeability-porosity relationships in sedimentary rocks, Log Analyst, May-June, 38-62, 1994.

Olsen, P.E., The latest Triassic and early Jurassic formations of the Newark Basin (eastern North America, Newark Supergroup): Stratigraphy, structure, and correlation, New Jersey Academy of Science Bulletin, 25, 25-51, 1980.

Paillet, F.L., Application of borehole geophysics in the characterization of flow in fractured rocks, U.S. Geol. Surv. Water Resour. Investigations Rept. 93-4214, 36 p., 1993a.

Paillet, F.L., Using borehole geophysics and cross-borehole flow testing to define hydraulic connections between fracture zones in bedrock aquifers, J. Appl. Geophys., 30, 261-279, 1993b.

Rouleau, A., and J.E. Gale, Stochastic discrete fracture simulation of groundwater flow into an underground excavation in granite, Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., 24, 99-112, 1987.

Terzaghi, R., Sources of error in joint surveys, Geotechnique, 15, 287- 304, 1965.

Theis, C.V., The relation between the lowering of the piezometric surface and the rate and duration of discharge of a well using ground-water storage, Eos Trans. Am. Geophys. Union, 16, 519-524, 1935.

Vecchioli, J., L.D. Carswell, and H.F. Kasabach, Occurrence and movement of ground water in the Brunswick Shale at a site near Trenton, New Jersey, U.S. Geol. Surv. Prof. Paper 650-B, B154-B157, 1969.

Zemanek, J., E.E. Glenn, L.J. Norton, and R.L. Caldwell, Formation evaluation by inspection with the borehole televiewer, Geophysics, 35, 254-269, 1970.

Table 1. Fracture statistics computed from Bingham axial-distribution analysis.

Well	number of	eigenvalue	eigenvector
	fractures	l	strike	dip
1	24	.501	38°	22°
		.421	261°	74°
		.078	167°	76°
2	39	.560	79°	55°
		.376	257°	34°
		.064	348°	87°
3	33	.697	276°	17°
		.288	77°	74°
		.015	167°	85°
4	13	.747	279°	26°
		.244	93°	65°
		.009	184°	87°
6	45	.508	79°	50°
		.441	267°	40°
		.050	172°	86°
8	26	.603	264°	14°
		.392	79°	77°
		.005	169°	88°
9	27	.709	267°	30°
		.282	77°	61°
		.009	169°	85°
10	41	.569	253°	10°
		.309	76°	81°
		.122	345°	88°
11	14	.535	277°	32°
		.437	91°	58°
		.026	182°	87°
13	18	.592	69°	38°
		.356	247°	52°
		.052	337°	88°
all	280	.532	276°	20°
		.411	79°	71°
		.057	170°	85°
permeable	43	.532	276°	16°
only		.407	79°	75°
		.061	170°	85°

Table 2. Average hydraulic properties of transmissive fractures.

fracture	number of	Tave	Save
type	fractures	(m_/min)
bedding-plane
partings	11	3.5x10-_	3.7x10-_
high-angle	12	1.8x10-_	2.9x10-_
fractures
intersections	8	2.9x10-_	2.2x10-_

Figure 1. Locations of study site and test wells.

Figure 2. Composites of selected logs from Wells 1, 3, 10, and 4.

Figure 3. (a) Magnetically oriented, acoustic-amplitude image of borehole wall generated from BHTV in Well 2. (b) Fracture strike and dip are determined from depth scale and magnetic orientation.

Figure 4. Stratigraphic correlation of (a) resistivity logs and (b) natural gamma logs along west-east transect of study site.

Figure 5. Rosette diagrams and equal-area stereographic plots of fracture orientations for Wells 3, 6, and 10. N is number of fractures, C.I. is contour interval, and s is standard deviation.

Figure 6. Rosette and stereographic plots of fracture orientations for all wells combined (N = 280 fractures).

Figure 7. Histograms of fracture dip versus frequency for entire fracture population (a) before and (b) after vertical probability correction.

Figure 8. Rosette and stereographic plots of fracture orientations for permeable fractures only (N = 43 fractures).

Figure 9. Transmissivities of the three fracture types as a function of depth.