Analyses of flowmeter measurements give quantitative estimates of in situ permeability, but these analyses are restricted by certain assumptions that may impact the reliability of practical permeability estimates. We compare permeability profiles interpreted from flowmeter logs obtained during borehole pumping in 13 bedrock boreholes at Mirror Lake, New Hampshire, to fracture density distributions determined from borehole image logs, and to interval transmissivity distributions given by the analysis of straddle-packer isolation and injection tests. Although the distribution of fractures identified in the boreholes is very similar to the distribution of fractures identified on adjacent outcrops, the fracture frequency distributions show almost no correlation with either flowmeter or packer test estimates of permeability. Qualitative and quantitative estimates of interval transmissivity derived from the flowmeter data are correlated with transmissivity values given by the hydraulic test data, but only for intervals with transmissivity greater than about 10^-6.5 m²/s. The main limitation in the flowmeter analysis of fracture permeability is the dynamic range of the heat-pulse flowmeter measurements, which spans at most 2 orders of transmissivity magnitude. The quantitative interpretation of fracture zone transmissivity is also sensitive to the measurement of total borehole capacity used to relate relative inflows from specific zones to flow in liters per minute under normalized aquifer test conditions. The differences between transmissivity values inferred from flowmeter analysis and those given by the hydraulic tests for a few specific intervals are attributed to local fracture connections in the adjacent rocks which cause significant departure of fracture zone hydraulics from the infinite confined aquifer model used in flowmeter profile interpretation.

Several different high-resolution borehole flow measurement techniques have been described in the literature (Hess, 1986; Tsang et al, 1990; Molz and Young, 1993; Paillet et al, 1994). These new borehole logging techniques provide the ability to measure the distribution of permeability along well bores by measuring the vertical distribution of flow during pumping (Paillet et al, 1987; Molz et al, 1989; Hess and Paillet, 1990; Kabala, 1994). Flow logging to infer the distribution of permeability in situ has the potential to provide this information more quickly and with the use of less equipment than conventional hydraulic testing using straddle-packer isolation techniques (Davison, 1984; Hsieh et al, 1985; Pickens et al, 1987). However, open hole flowmeter logging is inherently limited by the inability to control aquifer test boundary conditions because fractures can communicate along the open borehole during aquifer tests. Although fracture zones composed of interconnections between finite-length fractures may not always be properly approximated as an infinite confined aquifer, this approximation is routinely made in aquifer testing. In these aquifer test procedures, straddle-packer isolation and injection tests are routinely used to assign transmissivity and storage coefficient values for fracture zones identified in boreholes. While acknowledging the inherent limitations imposed by the assumptions invoked in packer test analysis procedures for fractured aquifers, we use such data as the accepted standard in the evaluation of flowmeter estimations of fracture zone transmissivity. In this paper we compare the qualitative and quantitative interpretation of the distribution of permeability along boreholes given by the analysis of flowmeter profiles obtained during pumping with discrete interval transmissivity values obtained by conventional hydraulic testing. In this comparison, we ask four questions: 1) how do qualitative permeability profiles obtained from the flowmeter profiles compare with those determined using other borehole logging techniques; 2) how accurately do the qualitative permeability profiles given by the flowmeter represent the relative distribution of permeability along the well bore; 3) can the flowmeter data be used to generate quantitative estimates of fracture zone transmissivity; and 4) what specific sources of error are most likely to affect permeability profiles generated from flowmeter logs. The detailed comparison of flowmeter analyses with the results of conventional hydraulic tests presented in this paper provides information needed to address all four of these important questions about flowmeter applications.

The U. S. Geological Survey has selected the Mirror Lake, New Hampshire, site for the long-term investigation of site characterization techniques needed for the prediction of solute transport in fractured crystalline rocks (Shapiro and Hsieh, 1994). The FSE borehole array at Mirror Lake consists of 13 bedrock boreholes cased through from 15 to 25 m of overburden and open to fractured crystalline bedrock below casing (figure 1). The boreholes vary from 60 to 100 m deep, except for borehole FSE-4, which is 220 m deep. All 13 boreholes were logged with a suite of conventional geophysical logs, and with various downhole imaging devices (Paillet et al, 1987; Paillet and Kepucu, 1989; Paillet, 1993). One borehole, FSE-5, was cored and subsequently reamed to the 15 cm diameter of the other boreholes. However, the reamed borehole was offset from the pilot corehole, producing a non-circular borehole cross-section which made packer seating impossible, and no interval transmissivity data are available for this borehole. Various hydraulic tests have been conducted in the FSE borehole array. The most extensive tests consist of single-hole straddle-packer isolation and injection tests (Shapiro and Hsieh, 1994; Hsieh et al, 1985). These measurements are considered the most consistent and reliable techniques for the characterization of the vertical distribution of permeability along bedrock boreholes (Pickens et al, 1987; Paillet et al, 1987). More than 120 such interval measurements were made in the FSE boreholes; these measurements are used as a reliable and definitive data set with which to verify the effectiveness of permeability profiling with borehole flowmeters.

The FSE boreholes at Mirror Lake were logged over the period from 1983 to 1994 to generate a description of the character and distribution of fractures intersecting each borehole. Fracture depth, strike, dip, and a qualitative estimate of fracture size were determined from acoustic borehole televiewer logs (Paillet and Kapucu, 1989). Fracture size is expressed as a relative score ranging from 1 for an isolated, possibly open fracture, to 5 for a major cavern-like opening (table 1). Fracture strike and dip are corrected for magnetic declination and borehole deviation using the methods given by Lau (1983) and Kierstein (1983). The relative amount of flow produced by each fracture is estimated from the flowmeter profiles obtained during pumping. A total of over 300 possibly permeable fractures have been identified and described on the logs for the 13 FSE boreholes (figure 2). The distribution of fractures in the boreholes (figure 2A) appears similar to the distribution of fractures described in outcrop in the Mirror Lake watershed (figure 2B). The surface outcrop data indicates relatively more steeply dipping fractures than the borehole data, but this likely represents the biases built into each method of sampling. Horizontal surface outcrops are more likely to intersect near-vertical fractures than horizontal fractures, while vertical boreholes are more likely to intersect horizontal fractures than vertical fractures. Except for the bias towards over-representation of horizontal fractures, we assume that the borehole data provide an accurate description of the distribution of fractures in situ. This distribution shows a concentration of fractures aligned with the steeply southeast dipping foliation of the bedrock, and a more diffuse distribution of fractures with many different orientations. The distributions of the populations of larger fractures (figure 2C) or water-producing fractures (figure 2D) show that these more limited subsets of fractures have no particular orientation, and are clearly not aligned with foliation. The distribution of these fractures appears to indicate that water-producing fractures are more likely to be horizontal than other fractures, but this may be an artifact of the sampling bias.

The application of flow profile analysis in estimating the transmissivity of fracture zones intersecting an open borehole is based on the concept of quasi-steady flow into a borehole being pumped at a constant rate. In flowmeter logging at Mirror Lake, we pump boreholes at about 4 liters per minute and measure the vertical distribution of flow within the borehole 30 to 60 minutes after the pump is turned on. This period of time allows for drawdown to stabilize and the rate of inflow from one or several producing zones to approach a constant value. The relative inflow rates are determined from the vertical changes in flow or "steps" in the profile, and are given as the percent of total inflow associated with each producing zone. The percent inflow from fracture zones measured during quasi-steady pumping cannot be directly related to fracture zone transmissivity because the rate of inflow depends on both transmissivity and the hydraulic head gradient driving the flow out of the fracture zone. Previous applications of flowmeter profiles in permeability measurements have assumed that vertical head gradients are negligible (Kabala, 1994; Molz et al, 1989). Such an assumption cannot be made in the case of fractured crystalline rocks because the isolation of individual fracture zones allows for significant hydraulic head differences between zones. The effects of hydraulic head differences and transmissivity can be separated by taking the difference between two different pumping conditions. For example, if steady radial flow into a well is assumed, the difference between flows measured under two different hydraulic head gradients is:

Q₁ = 2T(h₀-h₁)ln(R₀/R_w)

Q₂ = 2T(h₀-h₂)ln(R₀/R_w)

Q₂- Q₁ = 2T(h₁-h₂)ln(R₀/R_w) (1)

where Q₁ and Q₂ are the measured flow rates, T is the fracture transmissivity, h₀ is the far field hydraulic head in the fracture at some distance R₀, and h₁ and h₂ are the water levels in the borehole at the time when Q₁ and Q₂ are measured, and R_w is the borehole radius. In these equations, h₀ is generally unknown and different for each fracture, while h₁ and h₂ are known (they are measured in the borehole) and the same for all fracture zones. Under these conditions, the unknown head differences driving the inflow to the borehole are subtracted out of the expression, and the difference between the flow profiles is proportional to fracture zone transmissivity. In flow logging at Mirror Lake, we subtract the flow measured under one condition (ambient hydraulic head) from the flow obtained under another condition (during pumping) to generate the relative permeability profile from the flowmeter data.

When boreholes intersect relatively transmissive fracture zones, drawdown stabilizes shortly after the start of pumping, and differences between measured rates of vertical flow at different depth stations in the borehole can be unambiguously related to inflow or outflow at depths between measurements. In these situations, water produced by the steady pumping is derived almost entirely from the aquifer, and wellbore storage can be neglected. The inflows (outflow being negative inflow) are then differenced for the two pumping conditions (ambient and steady pumping) to give net changes in flow associated with net changes in hydraulic head. The relative net inflow (inflow during pumping minus inflow under ambient conditions) from each "hydraulically active" fracture zone is then expressed as a percent of the total inflow to the borehole. However, relative amounts of inflow expressed in percent need to be normalized using estimates of total borehole capacity to give a qualitative transmissivity profile. We normalize the measured flows by representing borehole capacity as the rate of inflow to the borehole after steady pumping to produce 6 m of drawdown after 1 hour. If a fracture zone is assumed to behave as an infinite confined aquifer, the normalized borehole capacity is given by

Q = Q₀[6/d₀] (2)

where Q is the borehole capacity, Q₀ is the steady pumping rate used for the flow profiling, and d₀ is the drawdown measured in meters 60 minutes after the start of pumping. This normalization is similar to the correction used by Molz et al (1989) in correcting for the increase in drawdown during the flowmeter experiments reported in that reference. The 1-hour test period is taken as typical of the periods used in flow profiling in Mirror Lake boreholes, and the 6 meter drawdown as the maximum drawdown usually allowed without dewatering the shallowest fracture zones. When borehole capacity is less than a few liters per minute, the definition is modified to give the inflow to the well under normalized conditions:

Q* = Q - Q_w (3)

where Q_w represents the rate at which water is being withdrawn from storage in the borehole when 6 meters of drawdown is achieved after 1 hour of pumping.

In some situations, the fractures intersected by a borehole are not transmissive enough to allow flow measurements under quasi-steady conditions. In these situations, the drawdown is found to approach or exceed the 6 meter limit imposed by the requirement that water levels remain above potentially productive fractures. When drawdown quickly exceeds the 6 m limit, the flow profiling is performed during borehole recovery rather than during pumping. The one major drawback with performing a flow profile in this manner is that the net rate of inflow to the borehole is changing during the experiment. As recovery proceeds, the rate of measured flow declines at any given measurement station. Therefore, differences in flow measurements made at different times and different depths can be caused by changes in flow with time, by inflow or outflow between measurement stations, or by a combination of both effects. The effects of recovery (flow changes over time) and the effects of inflow or outflow (flow changes with depth) are separated by normalizing flow measurements to correspond to a constant drawdown of 6 meters:

F(z) = 6F₀(z,t)/d(t) (4)

where F(z) is the normalized flow profile, F₀(z,t) are the flow measurements made at depth z and time t after the start of recovery, and d(t) is the drawdown (in meters) measured at time t. In practice, a few drawdown measurements are made during the experiment, and the appropriate values of d(t) are interpolated between these values.

The flow profiles obtained either by logging during pumping or by logging during recovery are expressed as normalized flow profiles corresponding to inflow under 6 m of drawdown after about 60 minutes of steady pumping. The "steps" in the profile are then interpreted as the difference in inflow rates in liters per minute between ambient and pumping test conditions. Each net inflow is then expressed as the percent, p of the total net inflow (inflow during pumping less the inflow under ambient conditions). The relative transmissivity of fractures intersected by different boreholes can be compared by multiplying the percent inflow at each individual fracture zone by total borehole capacity Q* as defined by equations (2) and (3). The resulting profile of discrete inflows under normalized test conditions at discrete fracture zones gives a qualitative profile of transmissivity along the borehole.

Quantitative interpretation of transmissivity is based on comparison of the measured inflow during pumping with models of inflow to finite-diameter boreholes intersecting fractures of given transmissivity. A significant complication in flowmeter profiling is that several fracture zone aquifers may interact with each other by means of the open borehole during flow profiling. Therefore, the inflow from individual fractures during pumping cannot necessarily be treated as would the measured inflow from a borehole penetrating a single aquifer during a production test. The boundary conditions will generally be different for each combination of fractures interacting in an open borehole during steady pumping. For example, the inflow from a single fracture under the steady pumping needed to produce 6 meters of drawdown after 1 hour will be different from that when the same drawdown is achieved over the same time period, but with another fracture providing inflow to the borehole.

We propose that the inflow to the borehole in the multiple fracture flow problem can be approximated by the inflow predicted for a single fracture intersecting an open borehole, where the fracture acts as an infinite confined aquifer characterized by transmissivity, T, and storage coefficient, S. The solution for steady pumping from a confined aquifer into a finite-diameter borehole is given by Papadopulos and Cooper (1967). The relationship between drawdown, D, pumping rate, Q, borehole radius, R, time after start of pumping, t, and aquifer parameters T and S is:

D = Q/(4T) * F(u,) (5)

u = R²S/4tT

where is a parameter depending on the ratio of storage coefficient to wellbore storage, and F is a well function analogous to that appearing in the Theis equation. Under the standard borehole test conditions of 6 m of drawdown after 60 minutes of pumping, this equation gives a relationship between pumping rate and transmissivity:

Q = f(T;R,S) (6)

Equation (5) defines a relationship between Q and T for given values of R and S; an example is plotted as the solid line in figure 3. The solid curve cannot be inverted to determine estimates of T for all measured values of Q because the curve becomes independent of T for values of T less than about 10^-6 m²/s. The relation can be modified by using the Papadopulos and Cooper (1967) solution to compute the water flowing into the well after 60 minutes of pumping, Q*, rather than the constant rate of pumping. The modified relation between Q* and T is given as the dashed line in figure 3. In general, computations show that the relationship is effectively independent of S for a 15 centimeter diameter borehole whenever S is smaller than 10^-4. This relationship gives:

Log(T) = 1.136 Log(Q*) - 5.720 (7)

We take this computed relationship between the amount of inflow to a borehole intersecting a single fracture zone under standard test conditions as an approximation for the transmissivity T_i of a fracture or fracture zone producing a flow q_i under standard test conditions. This relationship is used to translate the standardized inflow values q_i determined from the flowmeter logging as estimates of percent inflow, p_i multiplied by borehole capacity, Q*, to estimates of fracture zone transmissivity using the relation:

q_i = p_i Q*

Log(T_i) = 1.136 Log (q_i) - 5.720 (8)

COMPARING FLOWMETER ANALYSIS WITH OTHER GEOPHYSICAL ESTIMATES OF FRACTURE PERMEABILITY. The flowmeter data clearly correlate much better with fracture zone transmissivity than do other estimates of fracture permeability derived from geophysical measurements (figure 4). Simple estimates of fracture density given by summing the number of fractures identified on geophysical logs in each packer isolation interval show no apparent correlation with fracture zone transmissivity (figure 4A). The fracture density estimates can be improved by summing fracture size scores within each packer interval, but the revised estimates of size-weighted fracture density show only a weak correlation with interval transmissivity (figure 4B). However, the normalized inflow during pumping measured with the flowmeter does show a strong correlation with interval transmissivity, although the low-transmissivity limit of the flowmeter measurement appears to be about 10^-6.5 m²/s. These results demonstrate that local fracture density is not closely related to fracture permeability, and that fracture connectivity in large-scale flow paths is probably more important in determining fracture flow than is local fracture aperture (Long and Witherspoon, 1985; Paillet et al, 1987).

QUALITATIVE EVALUATION OF PERMEABILITY PROFILES GIVEN BY FLOWMETER ANALYSIS. In general, the relative permeability profiles given by the flowmeter experiments for the FSE boreholes agree with the packer transmissivity data (table 2). Two examples of the televiewer, flowmeter, and hydraulic test data for specific boreholes demonstrate the close qualitative agreement between the flowmeter and hydraulic tests data, and the superiority of both over the fracture density data given by the borehole image logs. The data for borehole FSE-6 (figure 5) illustrate the results for most of the boreholes, where the relative magnitude of interval transmissivity from the hydraulic tests closely matches the relative fracture permeability given by the measured percent of total inflow contributed by each fracture. The largest proportion of inflow is associated with the most transmissive fracture zone, and the flowmeter data indicates inflow associated with all intervals having transmissivity values greater than 10^-6.5 m²/s. In contrast, the data for borehole FSE-8 (figure 6) illustrate a significant disagreement between the qualitative transmissivity profile from the flowmeter analysis and the quantitative transmissivity data from the hydraulic tests among all of the more than 120 intervals tested. In this example, the flowmeter analysis correctly identifies the lower two intervals in borehole FSE-8 as having transmissivity values greater than 10^-5 m²/s, but clearly indicates that the upper of the two intervals has only a small fraction of the transmissivity of the lower interval. In this example, the flowmeter analysis identified the two most transmissive intervals in the borehole, while apparently giving an underestimate of the relative transmissivity of one of the two intervals with respect to the other. We also note the correction of permeability profiles to account for ambient flow is no more than one or two percent of total inflow in the one borehole, FSE-6, where the ambient flow was the greatest among all of the FSE boreholes. This would not always be the case; Paillet et al (1994) present a specific example where these corrections would not be negligible.

QUANTITATIVE EVALUATION OF PERMEABILITY PROFILES GIVEN BY FLOWMETER ANALYSIS. Although the technique used to calibrate the flowmeter permeability profiles in terms of fracture zone transmissivity ignores the communication between fracture zones along the open borehole during flow measurement, the calibrated flowmeter transmissivity values agree relatively well with the interval transmissivity values from the hydraulic tests (table 2). A linear regression between flowmeter estimates of transmissivity and the hydraulic test estimates of transmissivity shows a significant correlation (r squared equals 0.85), and the least-squares linear fit to the data gives a straight line very close to the diagonal (figure 7). The scatter about the diagonal indicates no tendency to bias the flowmeter estimates of transmissivity towards over- or under-estimates. The only obvious limitation of the flowmeter analysis with respect to the hydraulic test data is the inability of the flowmeter analysis to identify production from intervals characterized by intermediate transmissivity values less than about 10^-6.5 m²/s.

SOURCES OF ERROR IN FLOWMETER PROFILE ANALYSIS. The comparison between the flowmeter analysis and the hydraulic test data indicate that one important limitation of permeability profiling with flowmeters is the dynamic range of the measurement. The heat-pulse flowmeter provides useful measurements over a range from about 0.05 to 10.0 liters per minute (Hess and Paillet, 1990). This range represents about 3.5 orders of magnitude. However, flowmeter analysis involves distinguishing changes of flow with depth from measurement noise and transient changes in the overall rate of flow in the borehole. This inherent background noise appears to remove more than one order of magnitude from the measurement range of the flowmeter. This result is indicated by a statistical description of the transmissivity distribution (figure 8). The distribution shows that all of the most transmissive intervals are identified by the flowmeter analysis, along with about half of the next transmissive class. The effective dynamic range of the flowmeter measurement is between 1.5 and 2.0 orders of magnitude. Therefore, the flowmeter technique is useful in identifying the most transmissive fracture intersecting a borehole, but will not detect fractures having transmissivity values more than about 2 orders of magnitude less than that of the most transmissive fractures intersecting the borehole.

Another source of error in the quantitative analysis of the flowmeter data is the estimate of total borehole production capacity. In estimating transmissivity from the flowmeter data, the relative percent of flow from a given fracture is multiplied by total borehole capacity in equation (8) to give a normalized fracture zone production in liters per minute. The calibration of these values in transmissivity is as sensitive to the capacity measurement as it is to the relative apportioning of inflow from each zone. For example, the transmissivity values estimated from the flowmeter profile for borehole FSE-13 could be made to agree with those given by the hydraulic test data for all intervals within the borehole if the total capacity of the borehole used in the flowmeter analysis were increased by a factor of about 2.5.

A third potential source of error in the qualitative and quantitative estimate of fracture zone transmissivity with the flowmeter analysis is the presence of the open borehole during the experiment. The flowmeter data are calibrated under the assumption that the production from each fracture zone can be treated as a single-aquifer test. In fact, multiple fracture zones are communicating through the open borehole during the flowmeter profiling, while the straddle-packer hydraulic tests isolate each interval during the testing. This open hole communication effect appears to explain the disagreement between the two quantitative estimates of transmissivity in the lowermost two zones in borehole FSE-8. Other tests have indicated that these two zones are in communication in the surrounding rock mass (Paillet, 1993). During open hole testing, the more transmissive lower fracture produces most inflow. When the smaller-looking fracture just above that zone is isolated during straddle-packer tests, all flow is forced into this fracture, but the far-field flow is probably influenced by a connection with the much more conductive fracture just below. A combination of these effects may account for the differences between the two sets of transmissivity data for the most transmissive intervals such as those near 44 and 55 m in depth in borehole FSE-4, which appears as the most obvious outliers in regression shown in figure 7.

Comparison of qualitative and quantitative estimates of the distribution of fracture permeability along 13 bedrock boreholes derived from the analysis of flowmeter profiles obtained during pumping agree with the estimates of interval transmissivity given by conventional straddle-packer hydraulic tests for more than 120 separate intervals. The flowmeter estimates of permeability are clearly superior to permeability profiles based on the geophysical log data, such as size-weighted fracture density profiles generated from borehole image logs. The main limitation of the flowmeter analysis is derived from the dynamic range of the measurement. Although the range of flow detection is nominally about 3.5 orders of magnitude for the heat-pulse flowmeter, background noise and erroneous data normalization effectively limit the transmissivity characterization to about 2 orders of magnitude. Because of this limitation, the flow profile analysis failed to detect production from fractures with transmissivities less than about 10^-6.5 m²/s under test conditions used in the Mirror Lake boreholes. Other differences between the flowmeter-derived transmissivity estimates and those given by the hydraulic test data are probably related to the communication between fracture zones along the open borehole during hydraulic testing, and to connections between fractures in the rock mass surrounding the borehole.

The authors acknowledge the cooperation of the U. S. Forest Service and the Institute for Ecosystems Study in gaining access to the Mirror Lake study site, and for various support activities during the performance of field work. Allen Shapiro and Paul Hsieh provided the straddle-packer isolation and injection test values for interval transmissivity used to verify flowmeter measurements in the FSE boreholes. Chris Barton provided the surface outcrop fracture statistics used for comparison with data from borehole image logs.

Davison, C.C., 1984, Monitoring hydrogeological conditions in fractured rock at the site of Canada's Underground Research Laboratory: Groundwater Monitoring Review, v. 3, no. 4, p. 79-111.

Hess, A.E., 1986, Identifying hydraulically-conductive fractures with a low-velocity borehole flowmeter: Canadian Geotechnical Journal, v. 23, no. 1, p. 69-78.

Hess, A.E., and Paillet, F.L., 1990, Applications of the thermal-pulse flowmeter in the hydraulic characterization of fractured rocks: American Society for Testing Materials, ASTM STP 101, p. 99-112.

Hsieh, P.A., Neuman, S.P., Stiles, G.K., and Simpson, E.S., 1985, Field determination of the three-dimensional hydraulic conductivity tensor of anisotropic media--2. Methodology and application to fractured rocks: Water Resources Research, v. 21, no. 11, p. 1667-1676.

Kabala, Z.J., 1994, Measuring distributions of hydraulic conductivity and storativity by the flowmeter double test: Water Resources Research, v. 30, no. 3, p. 685-690.

Kierstein, R.A., 1983, True location and orientation of fractures logged with the acoustic televiewer: U. S. Geological Survey Water Resources Investigation Report 83- 4275, 71 p.

Lau, J.S.O., 1983, The determination of true orientation of fractures in rock cores: Canadian Geotechnical Journal, v. 20, no. 2, p. 209-221.

Long. J.C.S., and Witherspoon, P.A., 1985, The relationship of degree of interconnection to permeability in fracture networks: Journal of Geophysical Research, v. 90, no. B4, p. 3087-3098.

Molz, F.J., Morin, R.H., Hess, A.E., Melville, J.C., and Guven, Otkay, 1989, The impeller meter for measuring aquifer permeability variations - evaluation and comparison with other tests: Water Resources Research, v. 25, no. 7, p. 1677-1683.

Molz, F.J., and Young, S.C., 1993, Development and application of borehole flowmeters for environmental assessment: The Log Analyst, v. 34, no. 1, p. 13-23.

Paillet, F.L., 1993, Using borehole geophysics and cross- borehole flow testing to define hydraulic connections between fracture zones in bedrock aquifers: Journal of Applied Geophysics, v. 30, no. 3, p. 261-279.

Paillet, F.L., Crowder, R.E., and Hess, A.E., 1994, High resolution flow logging - a unique combination of borehole geophysics and hydraulics Part II - borehole applications with the heat-pulse flowmeter: Symposium on the Applications of Geophysics to Environmental and Engineering Problems, Boston, March, 1994, Proceedings, p. 381-404.

Paillet, F.L., Hsieh, P.A., and Cheng, C.H., 1987, Experimental verification of acoustic waveform and vertical seismic profile measurements of permeability: Society of Professional Well Log Analysts 28th Annual Logging Symposium, London, June, 1987, Transactions, p. PP1-PP21.

Paillet, F.L., and Kepucu, Kemal, 1989, Fracture characterization and fracture permeability estimates from geophysical logs in the Mirror Lake Watershed, New Hampshire: U. S. Geological Survey Water Resources Investigation Report 89-4058, 49 p.

Papadopulos, I.S., and Cooper, H.H., Jr., 1967, Drawdown in a well of large diameter: Water Resources Research, v. 3, no. 1, p. 241-244.

Pickens, J.F., Grisak, G.E., Avis, J.D., Belanger, D.W., and Thury, M., 1987, Analysis and interpretation of borehole hydraulic tests in deep boreholes - principles, model development, and applications: Water Resources Research, v. 23, no. 7, p. 1341-1375.

Tsang, C.H., Hufschmied, Peter, and Hale, F.V., 1990, Determination of fracture inflow parameters with a borehole fluid conductivity method: Water Resources Research, v. 26, no. 2, p. 561-578.

TABLE 1: Examples of fracture listings for Mirror Lake bedrock boreholes, FSE-6 and FSE-8.

TABLE 2: Comparison of fracture zone transmissivity estimated from analysis of flow profile obtained during pumping with those given by straddle-packer isolation and injection tests.

1. The FSE borehole array at the Mirror Lake, New Hampshire, research site.

2. Equal-area projections of the poles to fractures showing the distribution of strike and dip for: A) All fractures identified in the FSE boreholes; B) All fractures identified in surface outcrops within the Mirror Lake and adjacent Hubbard Brook watersheds; C) Fractures in the FSE boreholes assigned a fracture size score of 2 or higher; and D) Fractures in the FSE boreholes associated with inflow inferred from flowmeter logging during pumping.

3. Relation between normalized inflow and transmissivity for a 15-centimeter diameter borehole and an assumed value of S = 10^-4; normalized conditions are 6 meters of drawdown after one hour of steady pumping. The solid line, computed using the solution given by Papadopulos and Cooper (1967), gives the relation between pumping rate and transmissivity and the dashed line gives the relation between inflow rate and transmissivity.

4. Comparison of interval transmissivity given by straddle-packer isolation and injection tests with: A) Fracture density given as the number of fractures in the isolated interval; B) Size-weighted fracture density given as the summed fracture size scores in the isolated interval; and C) The normalized inflow during pumping inferred from flowmeter logs.

5. Comparison of flowmeter log interpretation and straddle-packer isolation and injection interpretations of fracture zone transmissivity for borehole FSE-6: A) Borehole televiewer log; B) Interval-averaged fracture permeability scores from televiewer data; C) Borehole flowmeter profile under ambient conditions; D) Borehole flowmeter profile obtained while pumping at about 4 liters per minute; E) Distribution of fracture zone inflow during pumping based on flowmeter log; and F) Interval-averaged fracture zone transmissivity given by straddle-packer tests.

6. Comparison of flowmeter log interpretation and straddle-packer isolation and injection interpretations of fracture zone transmissivity for borehole FSE-8: A) Borehole televiewer log; B) Interval-averaged fracture permeability scores from televiewer data; C) Borehole flowmeter profile under ambient conditions; D) Borehole flowmeter profile obtained while pumping at about 4 liters per minute; E) Distribution of fracture zone inflow during pumping based on flowmeter log; and F) Interval-averaged fracture zone transmissivity given by straddle-packer tests.

7. Comparison of fracture zone transmissivity values estimated from flowmeter data with those from straddle-packer and injection tests.

8. Frequency distribution of transmissivity values from hydraulic tests given as a function of transmissivity magnitude, showing the number of permeable zones in each transmissivity range that were identified using the flowmeter log interpretation.