| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Idaho National Engineering and Environmental Laboratory, P.O. Box 1625, Mail Stop 2107, Idaho Falls, ID 83415-2107
b 13786 Schoger Rd., Lathrop, CO, 81236
c Idaho State University, Pocatello, ID 83209-8072
d 2540 Grandville Ave., Henderson, NV 89052
* Corresponding author (helmcc{at}inel.gov).
Received 15 April 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: API, American Petroleum Institute • bls, below land surface • CAL, caliper logs • CGC, Century Geophysical Corp. • cps, counts per second • DEN, 
- [log] • INEEL, Idaho National Engineering and Environmental Laboratory • N, neutron [log] • NGR, natural • PNNL, Pacific Northwest National Laboratory • wt%, percentage (w/w)
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Previous logging studies in basalts by Crosby and Anderson (1971), Siems (1973), Siems et al. (1974), and Morin et al. (1993) have improved our understanding of neutron log data in unsaturated conditions, although these authors did not compare detailed physical properties to log response in a quantitative manner. Presently, no successful published calibration exists between neutron log response and moisture for unsaturated continental basalt despite some previous noteworthy attempts. For example, Poeter (1988) discussed quantitative neutron logging in basalt and other rocks, especially for identifying perched water layers in the vadose zone, but did not address all the properties of basalt that can complicate these measurements. Knutson et al. (1994) derived nuclear log correlations for saturated continental basalts and attempted extrapolation of these results for unsaturated basalts with variable success. Quantitative moisture results may exist for unsaturated basalts at the USDOE Pacific Northwest National Laboratory (PNNL), but these measurements were made using custom-built downhole "moisture meter" tools different from the conventional wireline neutron tools used in this study. Regretfully, most of the PNNL moisture results remain unpublished (Carl Koizumi, DOE Grand Junction, 2002 and 2003, personal communication).
Interpreting neutron logs in basalt is complicated by the fact that measured neutron flux is moderated not only by H in water, but also by H in hydrous minerals. These hydrous minerals are the clay fraction of sediments that infiltrate fractures and other voids in basalt, or the clays and zeolites formed by the alteration of continental basalts. Where hydrous minerals are present, the calculated neutron porosity will exceed the actual porosity due to the moderation provided by the addition of chemically bound H (Crosby and Anderson, 1971; Broglia and Ellis, 1990). This problem is essentially the same as the "apparent porosity paradox" observed in oceanic basalts logged by the Deep Sea Drilling Program and its successor, the Ocean Drilling Program (e.g., Broglia and Ellis, 1990). Excess apparent porosity has also been observed in continental basalts (Crosby and Anderson, 1971; Siems, 1973; Blackwell et al., 1982; Helm-Clark et al., 2004) and in some sedimentary rocks (Schlumberger Wireline & Testing, 1989).
Previous research on basalt morphology, alteration, and sediment infiltration suggests that the amount of hydrous minerals in thick basalt sequences may be nontrivial. Broglia and Ellis (1990) noted that hydrous alteration minerals in sea-floor basalts could compose up to 45% (w/w) of the rocks in the borehole environment, with an estimated statistical mode of about 10% (w/w). In their study (Broglia and Ellis, 1990), hydrous minerals typically contributed an error of 0 to 30% to the neutron porosity. A similar situation exists for continental basalts, where in extreme cases, hydrous alteration minerals can completely occupy all of the original total porosity (Morse and McCurry, 1997, 2002; Cheney, 1981). In some cases, the apparent neutron response in altered basalt cannot be distinguished from that of saturated unaltered basalt without supplemental information from other borehole geophysical tools (Helm-Clark et al., 2004). Sediments that infiltrate fractures in basalt can also contribute significantly to the apparent neutron porosity. Using data from East Snake River Plain basalts, Welhan et al. (2002) demonstrated that 15% of the total porosity at the tops and bottoms of flows consists of wide-aperture fractures and voids (aperture width >3 cm). They determined that approximately 50% of these wide-aperture fractures are filled with younger basalts or sediments during burial. The other 50% of the wide-aperture fractures, along with thermal-contraction columnar joints in the interiors of thick basalt flows, are usually preserved during burial (Welhan et al., 2002), although an unknown percentage of these can be completely infiltrated after burial by silt or finer sediments, resulting in apparent neutron porosities sometimes indistinguishable from sedimentary interbeds between flows (Helm-Clark et al., 2004).
Most hydrous minerals in basalt may be qualitatively identified by an increase in passive flux measured by a natural logging tool coupled with a decrease in the attenuated neutron flux measured by a neutron logging tool (Crosby and Anderson, 1971; Siems, 1973; Siems et al., 1974). In saturated basalts, quantitative separation of the effects of free water vs. hydrous minerals has been demonstrated using neutron and other logs from the Ocean Drilling Program (Broglia and Ellis, 1990), a method that depends on correctly estimating the amounts of all elements present in the borehole environment and modeling their contribution to neutron moderation.
The essential problem in using neutron tools in unsaturated basalt is identifying and quantifying the relative contributions to neutron moderation by H in free water vs. H in the chemically bound water of hydrous minerals. Following the original suggestion of Crosby and Anderson (1971), our approach to nuclear logs in basalt looks at the actual physical properties measured by these wireline tools (e.g., the moderation of neutron flux) and not the calculated properties more appropriate for logging in sedimentary rocks (e.g., neutron porosity). This study examined whether downhole neutron logs could (i) provide a quantitative estimate of H present in unsaturated basalt, and (ii) whether we could distinguish the relative contributions of H in free water vs. hydrous minerals. To this end, we compared the lithological, petrographic and physical properties of basalt with the response of neutron, - and natural tools in both saturated and unsaturated conditions.
To derive a usable correlation for neutron tool response vs. H content, we included data collected in saturated basalts if we knew both density and porosity and we assumed that the total porosity was 100% saturated with water. With such a correlation in hand, we then examined its applicability in unsaturated basalt and whether we could subsequently separate the effects of free water vs. hydrous minerals on the neutron flux. We focused on one location: the approximately 550-m-deep C1A borehole at the Idaho National Engineering and Environmental Laboratory (INEEL) on the Eastern Snake River Plain. In this well, the vadose zone is a 180-m-thick sequence of basalt flows on top of a 360-m-thick basalt-hosted aquifer.
All unpublished data cited in this study is archived at the Hydrogeological Data Repository at the INEEL and is available for public scrutiny upon request.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
and caliper measurements, with the former using a NaI detector with a radius of investigation approximately 30 cm or less (CGC, 2002; Keys and MacCary, 1971). The 9055 neutron tool used a 1-Ci 241Am-Be neutron source, a single 2.5 x 15 cm He detector and a source-detector spacing of 41 cm. The 9069 - tool used a 125-mCi 137Cs source, NaI detectors, and near and far source-detector spacings of 20 and 36 cm respectively. The depth of investigation of neutron and - tools varies, depending on lithology and borehole conditions, but is always less than the source-detector spacing.
Natural (NGR) and caliper logs (CAL) were collected in the uncased borehole (maximum diameter 15 cm) when the hole was filled to 12 m below land surface (bls) with a polymer gel drilling fluid. Neutron (N) and - (DEN) logs were collected through a NX or HX drill rod inside a 10.16-cm-i.d. steel casing. The N and DEN logs were collected after the drilling fluid had been removed from the borehole and replaced by air and unconfined aquifer water in the vadose and saturated zones, respectively.
Natural response in basalt was assumed to be linearly proportional to 40K content and has been effective for differentiating between low-K basalt flows and K-enriched sediments (Anderson and Bartholomay, 1995). - logs measure the attenuation of induced rays by Compton scattering (i.e., by the collisions of induced rays with orbital electrons). Gamma flux attenuation is therefore a function of electron density, which in turn is proportional to bulk density for the common rock forming minerals (Schlumberger Wireline & Testing, 1989).
Lithological logs for C1A were based on cuttings. Continuous core was collected for C1A and is now archived at the INEEL core library maintained by the USGS Idaho Falls field office. Samples from core (2.5 by 2.3 cm diam.) were collected for porosity, permeability, bulk density, grain density, and petrographic analyses. The samples were collected from the vesicular top, massive interior, and vesicular base of flows. In addition, samples were typically collected from the top and bottom of each flow unit in basalts displaying the characteristics of near-vent facies, that is, shelly pahoehoe and/or sequences of thinly bedded fractured and scoriaceous basalt lacking coherent flow interiors.
Bulk density was derived from measurements of mass and volume both in the field and in the laboratory. Helium porosity was measured in the laboratory using an American Petroleum Institute (API) compliant He gas porosimeter (API, 1952, 1998). Grain density was calculated based on the measured bulk density and He porosity, since the He porosity can be used as total porosity for most materials (Danielson and Sutherland, 1986; API, 1998). Permeability was measured using an API-compliant air permeameter (API, 1952, 1998). The equipment and procedures used for these analyses are described in Knutson et al. (1990).
High-frequency noise on all the logs was removed with bandpass filters. The natural log was normalized for hole size and casing configuration. Most log normalization strategies developed for sedimentary rocks are not relevant for DEN and N logs in basalt, so casing configuration and collar effects were normalized using an adaptation of the basalt-specific normalization method developed and described by Knutson et al. (1994). We used only the results from the far density detector for our bulk density correlation analysis, since this detector is the least affected by conditions in the borehole, thereby maximizing the influence of the formation rocks on the DEN log (Keys, 1990; Schlumberger Wireline & Testing, 1989).
We compared physical properties with N and DEN log response to determine if correlations existed between the datasets, and if so, to determine the nature of those correlations. Though the lateral dimensions of the physical property samples and the radii of investigation for the logs were an order of magnitude different, we considered their comparison valid based on an assumption used in most core-to-log integration studies, that is, that the formation sampled by the cores and the logs was homogeneous at scales less than the thickness of the formation (Corbett et al., 1998). This is a valid assumption for continental basalt where flows are laterally homogeneous on the scale of kilometers, and subflow morphologies such as flow interiors or upper vesicular zones are vertically homogeneous throughout. Though rare when compared with the volume of flows, small heterogeneities can occur at the centimeter scale. To minimize the effect of any localized heterogeneities (e.g., pipe vesicles), we applied a 30-cm sliding average to the response of the DEN and N logs before attempting any regression analyses. Physical property data were not used in any analysis from samples closer than 30 cm to the boundary between two basalt morphologies. The length of the sliding average was based on the empirically observed minimum vertical resolution of the borehole geophysical tools used in this study.
We did not use any of the traditional log-derived properties for comparison with physical properties except to demonstrate why such comparisons are inadvisable in basalt. Instead, we based our analyses on the actual phenomenon that each log measured. We also compared the log behavior in saturated vs. unsaturated conditions to distinguish the log effects due to unsaturated conditions from the log effects due to basalt, both of which are seldom logged with wireline tools. In addition, we attempted to identify a relation between the N log response and H content using data from where we were confident that all the neutron moderation was due solely to H in free water (i.e., in saturated nonvesicular basalt).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
flux on the DEN log in counts per second (cps) (indicative of denser material in the borehole wall); and high relative neutron flux on the N log in API-N units (indicative of low H content). Mud- and silt-rich interbeds are characterized by high relative NGR response, high relative DEN response (indicative of less dense sediments), and low relative N flux (indicative of increased H content). Sediments can also infiltrate fracture zones in basalt (Nace et al., 1975), which is the case in Well C1A at approximately 30m bls (Helm-Clark, unpublished data), where NGR response is still <50 API-GR, but DEN is highly variable, suggesting fractures and localized lowered density, and N flux is variable and relatively decreased, due to increased H from the hydrous minerals in the infiltrating clayey silts.
|
|
|
 | [1] |
is porosity, Gsample is the specific gravity of the rock, Gwater is the specific gravity of water, and the mass fraction of H in water is 0.112. Some assumptions must be met to estimate wt% H with this method. First, there can be no sources of H other than the free water in the saturated pore spaces. Second, there are no other neutron moderators present. Unaltered, massive, and saturated basalt with no fractures, vesicles, or infiltrated sediments is the only basalt morphology where we can meet these assumptions, so we are restricted to using data collected in these conditions to develop a valid correlation. Figure 4 shows neutron log response plotted against estimated wt% H in saturated massive basalt. Three outlier data points were excluded from the dataset because the NGR and DEN logs indicated that hydrous minerals were present. There are four possibilities for the nature of the correlation equation. For N tools, the moderated neutron flux should be negatively and logarithmically proportional to the amount of H in the absence of other neutron moderators (Keys and MacCary, 1971; Schlumberger Wireline & Testing, 1989). This means that valid correlations should have the form of logarithmic and/or exponential equations. It is physically impossible, however, to have negative neutron flux or a negative amount of H. Hence, a valid correlation should asymptotically approach both zero H at maximum flux and zero neutron flux at maximum H. This implies that an inverse polynomial equation may be a better choice than exponential and logarithmic equations, both of which can permit negative values for H and neutron flux. A positive linear correlation is also possible if the distance between the neutron source and detector is <Ls, the distance that a fast neutron travels before it reaches the thermal energy range (<0.025 eV) (e.g., Morin et al., 1993). This last possibility is examined in greater detail in the Discussion section below. Figure 4 also shows several regression curves based on these considerations. Some of the curve fits are constrained by boundary conditions of (300 API-N < N < 20000 API-N) and (0.109 wt% < H < 2.76 wt%). The lower bound of 300 API-N is twice the background noise for the N tool used at Well C1A. The upper bound of 20000 API-N is the maximum measurement possible for the N tool (CGC, 2002). The minimum wt% H is twice the lower limit of detection for saturated porosity in basalt for the N tool used at Well C1A (Knutson, 1992, unpublished data).
|
|
In general, the permeability and calculated wt% H loosely track one another in both saturated and unsaturated conditions (Fig. 5), although calculated wt% H in the vadose zone is sometimes spurious. For the linear and logarithmic correlations, the calculated wt% H values are occasionally negative. At some depths, low permeability is sometimes associated with relatively high calculated wt% H, or vice versa (e.g., 
25, 50, and 120 m bls), even where there are no indications that hydrous minerals are present either as fracture and vesicle coatings, as alteration minerals, or as infiltrating sediments.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
It is obvious that the traditional interpretive rules for saturated sedimentary rocks should not be applied to basalts, saturated or unsaturated, without significant study and modification. As demonstrated by the results shown on Fig. 2 and 3, the relations between log response and physical properties in basalts are more complex than the simplified interpretive rules commonly used for sedimentary rocks, a result already suggested by Knutson et al. (1994). Similarly, the relations between porosity and permeability are rarely as straightforward in basalt. For example, vesicularity can be as great as approximately 75% in some quickly erupted extremely gaseous basalts (Mangan et al., 1993), but the effective porosity through which any free water will travel could be restricted to fractures which link up vesicle pipes and trains, thus making it difficult to distinguish the relative contributions of vesicularity and fracture porosity to both the total and effective porosity.
It is clear that relating H content to N log response is a useful approach (Fig. 4), especially since such a relation can be applied regardless of any complexities of porosity and permeability in basalt. We can establish a correlation between the estimated wt% of H and the N log response, but only if several assumptions are true. First, the only source of H is free water. Second, no other moderators are present. Third, total porosity and bulk density are known. Fourth, neither porosity nor density are derived using wireline logging tools. Fifth, saturated porosity and total porosity are the same; that is, the total porosity is 100% saturated with free water. Sixth, all porosity is intergranular and interconnected, with no contributions from fractures, voids, and/or vesicles. The only basalt morphology that can successfully approximate these conditions is massive, unaltered, nonvesicular saturated basalt found in the interior of lava flows distal to their vent. Fractures, vesicles, infiltrated sediment, and alteration minerals are all rare in the massive interiors of basalt flows in Well C1A between 180 and 300 m bls, so their contributions to the neutron moderation should be negligible.
Our preliminary statistical analyses show that deriving a neutron flux vs. H content relation is possible in basalt, but it is unclear what sort of correlation equation is most appropriate. The logarithmic correlation used for Fig. 5 (red curve) results in negative H concentrations in the vadose zone. The other logarithmic correlation (see Fig. 4), which was not plotted on Fig. 5, does not produce negative H concentrations. Other than the correlation coefficient, however, it is not initially obvious that the logarithmic correlation used on Fig. 5 would produce physically impossible (negative) H concentrations since its fit residuals are slightly smaller and its 95% confidence intervals are as good as those for the other logarithmic correlation. Other forms of logarithmic correlations, those without a [Ln(N)]2 term, have regression coefficients between the two shown on Fig. 4, but their fit residuals and confidence intervals are much larger and increase with increasing N response. A similar trend exists for almost all of the inverse polynomial correlations, excepting a few which incorporate 1/x2 terms where fit residuals and confidence intervals increase with decreasing N response below about 800 API-N. Obviously an expanded dataset and more analyses are required to determine the best correlation for H vs. N response, one which will not produce negative values for wt% H in unsaturated basalt.
It is unclear whether the relative contributions of H in water vs. H in hydrous minerals can be separated for vadose zone basalts, since this cannot be determined with certainty until the exact nature of the H vs. N log correlation is known. This is indicated in part by the handful of instances where permeability decreases but calculated wt% H increases and vice versa in unsaturated basalt. Examples of this behavior are at 25, 50, 71, 90, and 120 m bls (Fig. 5). It is certainly possible to have permeable, fractured basalts that are dry and therefore host little appreciable H, especially in the case where there is an abundance of wide aperture fractures, but it is far less likely that elevated H concentrations are associated with massive, impermeable, unfractured, unaltered, and nonamygdaloidal basalts. Infiltration and/or growth of hydrous minerals are major agents in destroying basalt permeability (Morse and McCurry, 1997, 2002; Cheney, 1981), so hydrous minerals are a likely source of increasing H when permeability decreases. There is no indication, however, in the C1A cores, thin sections, or on the DEN and NGR logs that hydrous minerals are present at approximately 50 and 71 m. It is possible, therefore, that the N log for C1A suffers from the measurement inversion effect at these two depths, an effect documented by Morin et al. (1993) in unsaturated basalt, where the "slowing down" length of the N tool exceeds the source-detector spacing.
The slowing-down length, Ls, is the average distance that fast neutrons must travel before they are thermalized, that is, before reaching the thermal energy range of <0.025 eV (Schlumberger Wireline & Testing, 1989). The detectors of thermal neutron logging tools are designed to measure these moderated lower-energy neutrons. Ls is dependent on the concentration of neutron moderators plus the scattering and capture cross sections of those moderators. If H is present, its contribution to Ls will dominate all others due to its favorable scattering cross section and the fact that the mass of a proton is nearly equal to the mass of a neutron, causing large momentum losses when neutrons collide with H protons. For example, Ls in water is an order of magnitude shorter than Ls in unsaturated calcite, despite that fact that calcite is the denser material (Broglia and Ellis, 1990; Schlumberger Wireline & Testing, 1989).
As long as the neutron source and detector are at a distance greater than the maximum Ls possible, then the measured neutron flux will be negatively and logarithmically proportional to the amount of H present in the borehole environs (Keys and MacCary, 1971). If the source-detector distance is <Ls, however, then the tool response will be positively and linearly proportional to H content (Keys and MacCary, 1971). Wireline neutron logging tools are configured using the former source-detector geometry, whereas moisture meters are configured using the latter. When H concentration is very low (i.e., when Ls is dominated by elements other than H), then Ls can exceed the source-detector distance of a N log. In these cases, a N log will behave like a moisture meter and the measured neutron response will be inverted compared with the expected instrument response, as was the experience of Morin et al. (1993) in unsaturated basalt. Such an inversion would greatly complicate any attempts at separating the effects of free water vs. hydrous minerals in unsaturated basalts.
While logging in unsaturated basalts, Morin et al. (1993) observed excess-Ls neutron inversion effects on N logs. On the basis of the behavior of our N log data and on the similar experience of Morin et al. (1993), we suggest that neutron log inversion is the cause of the unexpected and anomalous N log behavior observed at about 50 and 71m bls. This may be a common problem when logging in impermeable, unfractured, and unsaturated basalts with off-the-shelf N logging tools. To eliminate this problem, N logging in unsaturated basalts may require tools with larger-than-usual source-detector spacing coupled with sources to produce larger neutron flux, although such a scheme would be at the expense of vertical resolution. Another solution would be a tool with multiple detectors spaced at different intervals on the sonde so that the inversion effect would be identified on the shorter source-detector pairs in comparison with the expected behavior measured by the longer source-detector pairs. It is this sort of multiple detector neutron tool that Morin et al. (1993) used in their study to investigate the measurement inversion effect in unsaturated basalt. The advantage of the multidetector arrangement is that any measurement inversions could be clearly identified as such simply by comparing the response of the nearer vs. farther detectors, thus eliminating any doubts as to whether a high H measurement was due to actual abundant H or due to a long-Ls measurement inversion effect.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The results of this study are based on a subset of the physical property data available for saturated basalts in C1A and other nearby wells. By incorporating all the saturated basalt data, we hope to determine the exact nature of the N vs. H content relation. Once the form of this relation is known, we plan to incorporate fracture frequency, fracture permeability, vesicularity, vesicle size and shape, alteration, point count phenocryst vs. matrix vs. void percentages, and equilibrium saturation information into this ongoing research.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. M. Holt and M. J. Nicholl Uncertainty in Vadose Zone Flow and Transport Prediction Vadose Zone J., May 1, 2004; 3(2): 480 - 484. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
