You are on page 1of 23

A universal power-law scaling exponent for

fracture apertures in sandstones

J.N. Hooker1,2†, S.E. Laubach1, and R. Marrett2


1
Bureau of Economic Geology, Jackson School of Geosciences, The University of Texas at Austin, Austin, Texas 78713, USA
2
Department of Geological Sciences, Jackson School of Geosciences, The University of Texas at Austin, Austin, Texas 78713, USA

ABSTRACT large-fracture spacing estimations. The as- mal or log-normal size distribution (Table 1).
sumption of the scaling exponent should be In contrast, a well-documented distribution of
A high-resolution data set of kinematic supported by the presence of crack-seal tex- fracture sizes is the power law (Gudmundsson,
aperture (opening displacement) of opening- ture within fractures. 1987; Barton and Zoback, 1992; Clark et al.,
mode fractures, from large (up to 2 m long) 1995; Gross and Engelder, 1995; Loriga, 1999;
quartz-cemented sandstone samples, shows INTRODUCTION Marrett et al., 1999; Ortega and Marrett, 2000;
that microfractures are ubiquitous and that Gillespie et al., 2001; Laubach and Ward, 2006;
most natural-fracture sets are better fit by Although many opening-mode fracture arrays, Ortega et al., 2006; Davy et al., 2010; Guerriero
power-law size distributions than by expo- commonly including cement-barren joints, have et al., 2010; Hooker et al., 2011). The power-law
nential, normal, or log-normal distribu- narrow ranges of opening displacement (or equation returns a positive frequency for all posi-
tions. The data set includes 3822 fractures kinematic aperture) sizes (Odling et al., 1999), tive fracture sizes, in a fixed inverse proportion to
within 68 scanlines from eight formations on many other fracture arrays possess a range of fracture size, and thus implies a scale-free frac-
three continents. Kinematic apertures were apertures spanning multiple orders of magni- ture pattern, i.e., a pattern lacking a characteristic
measured along scanlines using scanning tude. Interpreting the cause of variation in frac- size (Bonnet et al., 2001). Because exponential
electron microscope cathodoluminescence ture size within a fracture population requires size distributions (Nur, 1982; Deschamps et al.,
(SEM-CL) and, for field data, using a hand more information than just the range in size. In 2007) return negative fracture sizes at some fre-
lens. Microtextural evidence from SEM-CL principle, even wide ranges of fracture size can quencies, implying a characteristic fracture spac-
shows that power law–distributed fractures be attributed to statistical aberration around a ing, they are better classified as characteristic as
typically have crack-seal texture and are characteristic size if the fractures follow a nor- opposed to scale-free.
composed of opening increments having a
narrow (characteristic) aperture size range.
TABLE 1. COMMON EQUATIONS DOCUMENTED FOR FRACTURE-SIZE DISTRIBUTIONS
In contrast, rare non-power-law–distributed
Documented fracture-size
fracture populations lack crack-seal texture. or slip-magnitude
Power-law exponents, as measured in one Equation type Cumulative distribution function equation distribution examples
dimension, have values of –0.8 ± 0.1. Most 1⎡ ⎛ X − μ⎞⎤
F= 1+ erf ⎜ ⎟⎥
variation among fracture sets results from Normal 2 ⎢⎣ ⎝ 2σ ⎠⎦
power-law coefficients, which constitute a Gillespie et al., 2001
scale-invariant measure of fracture inten- 1⎡ ⎛ ln X − μ⎞⎤ Philipp, 2008
Log-normal F= ⎢1+ erf ⎜ ⎟⎥
sity. We show how observed scaling patterns 2⎣ ⎝ 2σ ⎠⎦ Larsen et al., 2010
Hooker et al., 2012
can be used to improve estimations of large- Gudmundsson, 1987
fracture spacing in cases where fracture Wong et al., 1989
sampling is limited, as by the width of cores. Marrett and Allmendinger, 1992
Sanderson et al., 1994
The low (<1) value of the power-law scal- Clark et al., 1995
ing exponent reflects but a gentle increase McCaffrey and Johnston, 1996
−b
in fracture frequency with decreasing size, Power law F = aX Marrett, 1997
Bohnenstiehl and Kleinrock, 1999
such that microfracture abundances in core Marrett et al., 1999
are commonly too low for statistically robust Ortega et al., 2006
Guerriero et al., 2010
sampling. On the other hand, the consistency Hooker et al., 2011, 2012
of the scaling exponent among fracture sets Le Garzic et al., 2011
within various tectonic settings is such that Exponential F = ce−dX Nur, 1982
the exponent can be assumed, facilitating Deschamps et al., 2007
−b −dX
F = aX e Kagan, 1997
Gamma law

Current address: Department of Earth Sciences, Sornette and Sornette, 1999
University of Oxford, South Parks Road, Oxford, Note: Normal distribution equation is not documented but is included because it was tested in this study.
Oxfordshire OX1 3AN, UK; john.hooker@earth.ox F—fracture cumulative frequency; X—fracture size; erf—error function; μ—average; σ—standard deviation;
.ac.uk. a, b, c, d—best-fitting variables.

GSA Bulletin; September/October 2014; v. 126; no. 9/10; p. 1340–1362; doi: 10.1130/B30945.1; 18 figures; 4 tables; published online 22 May 2014.

For permission to copy, contact editing@geosociety.org


1340
© 2014 Geological Society of America
A universal power-law scaling exponent for fracture apertures in sandstones

The reasons that fracture sizes might follow a power-law with an exponent, as measured in The advantages of the Narr method are its
power laws versus characteristic size distribu- 1-D, of –0.8 ± 0.1. Second, we use SEM-CL simplicity and its ease of application to any
tions are uncertain, although possible models images of fractures to interpret the key growth sampled subsurface interval. However, errors
have been explored. Power-law size distribu- process that engenders this size distribution: associated with this method will potentially be
tions arise in some numerical routines that localization of individual, characteristic-size great if the reservoir (1) is poorly sampled or
postulate rules for fracture interaction during opening increments. Third, we show how (2) displays significantly heterogeneous fracture
growth (Cowie et al., 1993; Clark et al., 1995; microfracture populations measured from core stratigraphy and the preserved core or well-bore
Cladouhos and Marrett, 1996; Davy et al., 2010; samples can be used to extrapolate macrofrac- volume is insufficient to statistically represent
Hooker et al., 2012). A common theme among ture frequency in the inter-well-bore rock vol- each type of fractured layer.
such models is that large fractures emerge from ume, if the typical size distribution is present. Image logs can be used to identify fractures,
dynamic interaction (e.g., linkage) among Readers familiar with fractals (Mandelbrot, and thus to measure fracture abundance, where
nearby smaller fractures. 1982) can appreciate the potential for using core is not retrieved from the (vertical or slant)
Constraining the size distribution of a fracture power-law–distributed patterns at an observed well (Davatzes and Hickman, 2005; Laongsakul
population can help quantify natural-fracture scale to describe unobserved scales. In geologi- and Dürrast, 2011). However, several difficul-
abundance in the subsurface, where direct sam- cal practice the opportunities to actually make ties plague the interpretation of fractures on
pling is difficult. Fractures that are large enough such extrapolations are rare. Here we demon- image logs: sealed or partially sealed fractures
to be important for fluid storage and permeabil- strate how SEM analysis of core can, with care, are difficult to identify, aperture measurements
ity may nonetheless be difficult to detect seis- be used to constrain the fracture pattern in the are imprecise, and natural fractures are difficult
mically because they dip steeply and may not inter-well-bore space via power-law extrapola- to distinguish from drilling-induced fractures.
juxtapose reflecting horizons, making seismic tion. Thus, we apply the fundamental science Consequently, image logs are of limited utility
fracture-network surveys incomplete (Hestham- of fractals to the reservoir-engineering bugbear for natural-fracture surveys where precise aper-
mer and Fossen, 2003; Marrett et al., 2007; Bur- of subsurface fracture spacing. More generally, ture measurements are needed.
nett and Fomel, 2011; Cartwright, 2011). The we believe that our data-driven approach to the Just as the prevalence of power law–distrib-
subsurface is commonly sampled directly using fractal nature of geologic fractures can form the uted fracture sets has been progressively recog-
only vertical well bores, and in typical cases basis of analogous exercises to quantify a vari- nized, so has the value of accounting for fracture
macrofracture spacing is considerably wider ety of fracture-related attributes—such as strain size as a function of fracture abundance (Ortega
than the well-bore diameter (Ladeira and Price, (Gross and Engelder, 1995), porosity and per- et al., 2006; Guerriero et al., 2011). In order to
1981). Fracture frequency (i.e., number of frac- meability (Marrett, 1996; Laubach, 2003; Guer- establish a relationship between fracture size
tures per unit length, area, or volume of rock) riero et al., 2013), and network evolution (Davy and abundance, not only must fracture surveys
in the inter-well-bore space can potentially et al., 2010)—as well as attributes of other natu- be systematic over a distance, area, or volume
be constrained using microfracture scaling, if ral systems having fractal properties (see data of rock, but the resolution of the method of
microfractures and macrofractures are present sets in Clauset et al., 2009). observation must be considered. If the observ-
in relative abundances that may be statistically able size range is limited, then larger or smaller
extrapolated, as may be the case with power- MEASURES OF FRACTURE fractures than those observed might be present
law size distributions (Marrett et al., 1999; ABUNDANCE but uncounted, so that fracture index is exagger-
Gomez and Laubach, 2006; Ortega et al., 2006; ated where fractures are more readily observed.
Hooker et al., 2009; Guerriero et al., 2010; Iñigo Previous Methods Subsurface cores are typically no more than
et al., 2012). 10 cm wide, and sidewall cores are rarely longer
The new fracture-aperture data set presented The simplest measure of fracture abundance than 5 cm. Therefore, only small layer-parallel
here represents an opportunity for studying is the number of fractures observed within a fracture surveys can be made within the subsur-
size-distribution systematics because of its com- given observation space. For example, the frac- face except in the special case of horizontal core
bined breadth and consistent methods of collec- ture abundance of a layer within a core can be (Hooker et al., 2009). The primary difficulty
tion. The data set is large, comprising fracture quantified by counting the number of fractures associated with using microfractures to predict
populations from a wide range of tectonic set- present at a given depth (Haynes, 1984; Wilson macrofracture abundance is that, in core sam-
tings—including high-strain settings within fold and Paulsen, 1998). But for macrofracture spac- ples, few microfractures are typically observed
and thrust belts, as well as comparatively unde- ing wider than a 10 cm core diameter, cores at that can be reliably associated with larger frac-
formed intramontane basins. The same data-col- many depths will lack any visible fractures, even tures. For example, within Travis Peak Forma-
lection method was consistently deployed: all if fractures are present in the layer. tion, East Texas, USA (Cretaceous) samples,
fracture-size measurements quantify kinematic A method of calculating fracture spacing near the number of fractures (of any size) per thin
aperture (cumulative opening displacement) the well bore using the limited sampling typically section, identified by associated fluid-inclusion
along one-dimensional (1-D) scanlines, using encountered with vertical well bores and steeply planes, is as high as 10–15 but is typically fewer
scanning electron microscope cathodolumines- dipping fractures was devised by Narr (1996). In than five (Laubach, 1989).
cence (SEM-CL) to resolve fracture apertures this method the volume of rock represented by The advent of microfracture detection using
down to the micron scale. All data derive from preserved core or image logs is assumed to sta- SEM-CL greatly increased the precision of
sandstones, thus partly controlling for system- tistically represent the reservoir rock, and aver- microfracture measurement (Laubach et al.,
atic differences in fracture patterns by rock type. age fracture spacing is derived by dividing the 1995; Milliken and Laubach, 2000; Hooker
The purpose of this paper is threefold. First, core-face surface area by the cumulative fracture et al., 2009). However, because of the ambigu-
we use new data and a literature review to docu- height. Extensive subsurface observations and ous provenance of intra-granular microfractures
ment a typical aperture-size distribution among calibration with outcrops or other methods can (Laubach et al., 1995; Laubach, 1997; Hooker
natural opening-mode fractures in sandstones: support this spacing derivation. and Laubach, 2007), the enhanced capability

Geological Society of America Bulletin, September/October 2014 1341


Hooker et al.

of SEM-CL to detect microfractures does technique also controls for scanlines of differ-
not always result in substantially larger data ent length collected at a single scale of obser-
sets of microfractures that are clearly related vation. Lastly, the technique facilitates com-
to macrofractures. In this paper, we use evi- parison of data from different places (Ortega A
dence from SEM-CL imagery to illustrate how et al., 2006).
natural-fracture-size distributions develop, and Scanline
we establish a method of extrapolating large- Scanlines in Core
fracture spacing from observed microfracture
size distributions. In all core samples, microfracture scanlines
were constructed from the longest layer-parallel
Scanlines in Bedded Rock (fracture-perpendicular) scanline possible; core
from deviated wells allows longer multi–thin-
In a structurally simple case, extensional section scanlines than does vertical core, whose
strain will be manifested in a single set of paral- diameter limits scanline length to ~100 mm in
lel, opening-mode fractures. Many such fracture typical cases (Fig. 1). Our combined data set 0 cm 5
arrays exist perpendicular to sedimentary bed- includes scanlines from outcrop, slant core, or
ding in flat-lying or gently dipping beds (Pollard vertical core through dipping beds, in which
and Aydin, 1988). Macrofractures in such cases macrofractures and microfractures could be
can be measured along layer-parallel, fracture- observed. Additionally, we independently B
perpendicular scanlines of observation. Along gathered data from short scanlines (from verti- Scanline
the scanline, the kinematic aperture of each cal cores drilled through horizontal beds) and
fracture is measured where the fracture inter- show how these data can be used to predict
sects the scanline (Fig. 1). In the case of a single macrofracture spacing away from the well bore.
set of parallel opening-mode fractures, trigono- Except as noted, vertical cores are unoriented.
metric corrections are not needed. For multiple Where they occur together, microfractures and
fracture orientations, opening displacement can macrofractures are parallel, and elsewhere we
be corrected by fracture strike, or multiple scan- assume macrofractures in vertical cores strike
lines can be drawn. If multiple fracture sets are parallel to (oriented) macrofractures in slant
present that all dip steeply with respect to bed- cores from the same basins, but this may not
ding, circular scanlines can provide a sample set always be the case.
unbiased by orientation (Mauldon et al., 2001;
Rohrbaugh et al., 2002). Scanlines versus Maps
This study is focused on relatively simple
C
fracture arrays, for which linear scanlines suf- To maximize the number of microfractures
fice. Except as noted, for each macrofracture sampled at a given depth within subsurface core Vertical core
scanline measured in the field or in core, a cor- samples, an entire cross section of core could be
responding microfracture scanline was taken. mapped using SEM-CL. For this study, how- 0 mm 100
For microfracture measurement, a sample was ever, fracture apertures were measured along
taken from the rock along which the macrofrac- 1-D transects, as opposed to 2-D maps, for three Approximate scale
Up
ture scanline was made. From this sample, thin reasons. (1) The additional time and comput-
sections were cut parallel to bedding such that ing power required to assemble a fracture map Fracture Top
a microfracture-perpendicular scanline, parallel using SEM-CL increase with the square of
to the macrofracture scanline, could be digitally scanline length. (2) Fracture-size surveys within
drawn across SEM-CL images. Samples were 2-D maps are subjective, because fracture link-
processed such that no rock was lost between age can make it ambiguous where one fracture
serial thin sections for long, multi–thin-section ends and another begins. With scanlines, the Bottom
analyses (Gomez and Laubach, 2006).
Size distributions are here quantified by Thin section
plotting cumulative frequency (fracture-size surface
rank divided by scanline length) versus frac- Figure 1. Fracture-aperture measurement
ture aperture (e.g., Marrett et al., 1999). Com- along scanlines. (A) Scanline drawn across Horizontal core
bined macroscopic and microscopic scanline fractures in outcrop. Black scanline seg-
measurements allow data from different scales ments represent fractures; white segments
of observation from the same body of rock to represent spacings between fractures. Fracture
be compared on a single graph. Because scan- (B) Scanline drawn across microfractures in ~15mm
lines for microfractures are shorter than those scanning electron microscope cathodolumi- Thin section
~15mm
for large fractures, such a comparison requires nescence (SEM-CL) image. (C) Method of surface
Up Fracture
normalizing the cumulative number of frac- cutting thin sections for core-microfracture
tures (fracture size rank) to the scanline length, analysis. Outcrop thin-section method is
the result being the cumulative frequency. The analogous.

1342 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

locations of fractures are relatively straight- A The largest fractures in cumulative frequency
forward. (3) If fractures are present in parallel data are inherently poorly sampled (Guerriero
sets, then scanlines provide the same aperture- et al., 2010). Because of poor sampling, the
size information that fracture maps do. These large-size end of a fracture size–frequency
reasons notwithstanding, more microfractures curve has high associated uncertainty, which
might be sampled if core cross sections were commonly results in scattering of data in size-
to be mapped in 2-D; 2-D map scaling might frequency space. However, such scattered data
therefore be preferable where data are scarce are not expected to systematically under- or
and could be the subject of future study. over-record fracture frequency for 1-D scan-
line data.
Barren and Cemented Fractures
Statistical Tests for Power-Law Scaling
Fractures observed in this study that lack
cement entirely have been omitted from this B To identify consistent patterns among
analysis. Such barren fractures may have formed TF1 natural-fracture-size distributions, we regress
TF2 Quartz
naturally, but near the Earth’s surface, and so Chert grain the fracture frequency data without removing
have little to do with subsurface fracture for- Intra- fragment any data points. The assembled data sets may
mation, in which diagenetically reactive fluids granular be afflicted by the two types of bias described
commonly precipitate cement (Laubach et al., cement above (i.e., small-size truncation from omis-
2010). Barren fractures might also have formed IF sion of intragranular fractures and poor large-
in the subsurface but in response to fluid pres- fracture sampling), but whether the observed
sure from natural gases that could potentially data are consistent with biased sampling of an
suppress cement precipitation (Lacazette and TF2 ideal type of size distribution is an interpreta-
Engelder, 1992). Lastly, barren fractures may be TF3 tion which must be supported by independent
artificial. Barren fractures in core likely formed argument—bias should not be assumed at the
during retrieval or handling. Barren fractures in outset. Patterns to fracture scaling that are not
thin section may represent damage during saw- attributable to these types of bias require alter-
Figure 2. (A) Scanning electron microscope
ing and polishing. By restricting our attention native explanations, possibly including natu-
cathodoluminescence image of quartzarenite
to cemented fractures, we focus on fractures rally complex scaling.
with microfractures. (B) Interpretation of A.
with apertures that are easily and objectively Aperture-size distributions from each scan-
Intragranular fractures (IF) have unknown
measureable, and we omit possibly artificial or line can be regressed with a variety of equa-
provenance and may have been inherited
surficial fractures. tions (Table 1). The equation type that best
with grains (i.e., may have formed before the
fits each size distribution—among power law,
sandstone was deposited). At least three sets
Intra- and Trans-Granular Microfractures exponential, normal, and log-normal—can
of transgranular fractures (TF) are pres-
be judged by the chi2 value for each equa-
ent, whose relative timing (1—oldest; 3—
Data collection in this study adheres to the tion type:
youngest) can be assessed by crosscutting
fracture-selection criteria used in Hooker et al.
relationships.
(2009) with respect to intra- or trans-granularity chi2 = Σ[(Fobs – Fmod)2/|Fmod|], (1)
of fractures (Fig. 2). By omitting fractures that
do not crosscut grain boundaries and/or inter- where Fobs is observed fracture frequency and Fmod
granular cement, this procedure omits any Curve-Fitting and Sampling Biases is frequency modeled using the regressed equa-
fractures that may have already been present tion. We prefer the chi2 test to the Kolmogorov-
within host sand grains at deposition. Inspec- For the purpose of accurate extraction of Smirnov test suggested by Clauset et al. (2009)
tion using SEM-CL shows that many such frac- statistical information from a fracture set, data because chi2 is equally sensitive to fracture fre-
tures exist in sandstones (Laubach, 1997; Ber- that appear to be biased may be removed before quencies across the observed size range.
net and Bassett, 2005; Bernet et al., 2007), and equations are regressed, in the interest of deriv- Natural-fracture sizes may not perfectly
some of these can be identified by textural or ing an equation that most closely approximates adhere to simple size-distribution equations
color criteria (Laubach, 1997; Seyedolali et al., the underlying distribution (Pickering et al., (Table 1). Indeed, many ideal equations have
1997; Götze et al., 2001; Hooker and Laubach, 1995; Bonnet et al., 2001). A disadvantage to geometrical limits to the range of fracture-size
2007; Augustsson and Reker, 2012). However, such an approach is that it can be difficult to dis- abundances that the equations can approxi-
these criteria cannot unambiguously identify all tinguish real and artificial complexities within mate (Hooker et al., 2011). Moreover, as we
such fractures. All intragranular fractures are fracture intensity curves. show below (see Interpretation of Aperture-
therefore removed for this analysis. It is likely Truncation bias represents a systematic Size Distributions), residuals from such simple
that filtering out all intragranular fractures also under-recording of fractures among the small- equations potentially reveal information about
omits some post-depositional fractures. Because est fracture sizes measured. Truncation bias how fractures grow. Therefore, rather than
such fractures are sub-granular in scale, their may affect our data by omission of sub-grain- hypothesizing that an observed fracture-size
omission is equivalent to a truncation bias, scale microfractures and by our inability to distribution was sampled from an ideal power-
whereby the smallest fractures observed are resolve fractures below the SEM pixel size law or characteristic population, and testing
systematically under-recorded (Baecher and (0.77 microns in width) at the magnification this hypothesis against random subsamples of
Lanney, 1978). used (150×). an ideal, artificial distribution, the approach

Geological Society of America Bulletin, September/October 2014 1343


Hooker et al.

TABLE 2. OVERVIEW OF FRACTURE-SCALING DATA SET


Formation/group Folk (1981) classification Structural setting Major fracture-cement phases Average microfracture strain
Travis Peak Formation Quartzarenite Unstructured* Quartz 1.4 × 10–3
Mesaverde Group Sublitharenite, litharenite Unstructured* Quartz, carbonate 4.7 × 10–3
Formation X Sublitharenite Foreland basin, fold-thrust belt Quartz, carbonate 1.4 × 10–2
Huizachal Group Feldspathic litharenite, litharenite Sub-décollement Quartz, carbonate 1.6 × 10–2
Eriboll Formation Quartzarenite Sub-thrust belt Quartz 2.6 × 10–2
Mesón Group Quartzarenite Fold-thrust belt Quartz 4.8 × 10–2
*Fractures from Travis Peak Formation and Mesaverde Group are not related to obvious folds or faults; see text.

pursued here will be to examine natural-frac- GEOLOGIC SETTINGS C


ture-size distributions of any number of frac-
A
tures in aggregate, so that patterns present Among study areas, strain recorded among
among natural-fracture networks can be iden- microfracture populations (therefore over the
tified and interpreted. A relevant description same fracture-size range) varies over more
of each fracture set includes which common, than an order of magnitude. At the low end,
simple equation best fits the data relative to the strain near 10–3 is present among fracture
other simple equations and judged via the chi2 sets in unstructured settings; at the high end, QC
criterion. However, such equations should not strain near 5 × 10–2 is present among fracture QC
be construed as the underlying or real distribu- sets in fold-and-thrust belts (Table 2). The
tion without accounting for natural complexi- Piceance Basin and East Texas Basin settings
ties that are neither encompassed by the simple (see Appendix) are unstructured in the sense
equation nor even attributable to general, non- that samples were collected on broad (tens to CC
100 μm 500 μm
site-specific growth histories. hundreds of kilometers), gentle arches within
This method could potentially result in a sedimentary basins, distant (>0.5 km in most
power-law size distribution being classified as cases) from mapped faults or folds. These are B D
characteristic, but likely not vice-versa. This is “regional fractures” (Nelson, 1985) in that cur-
because truncation bias tends to result in con- vature associated with these arches is unlikely
cave-downward curves to scaling data. If this to affect fracture development (see Becker QC
effect is combined with a downward roll-off et al., 2010; Fall et al., 2012; and English,
at the large-size end, then data collected from 2012, for discussion of fracture origins). The
a natural-fracture set that strongly follows a remaining fracture-sample locations are within
power law or gamma law could nonetheless be fold-and-thrust belts. In each case we assume
better fit statistically by a characteristic distribu- that the fractures represent tectonic deforma-
tion because of biased sampling. Therefore, the tion, on the basis of their association with
effects of biases should result in overestimates large-scale folds and faults. In any case, non-
of the occurrence of characteristic-sized fracture tectonic fractures arising from local fluid injec- QC+CC
CC
sets in this study. tion (Mandl and Harkness, 1987) may scale
Here we do not regress data using gamma-law differently and are outside the scope of this
or stretched-exponential equations, although paper. Sample locations are listed in Table 2 in 100 μm
10 cm
such equations might closely fit the data. The increasing order of maximum fracture strain.
best interpretation of a gamma law known to A brief description of each geologic setting is Figure 3. Natural fractures in sandstone.
us is a power law with an upper size limit (Sor- provided in the Appendix. (A) Scanning electron microscope cathodo-
nette and Sornette, 1999; Bonnet et al., 2001), luminescence (SEM-CL) image of a thin
akin to the truncated power law (Kolyukhin and FRACTURE DESCRIPTION microfracture sealed with quartz cement
Torabi, 2013). Therefore, a best fit by a gamma (QC). (B) SEM-CL image of a larger micro-
law would require the same interpretation that Fracture populations in this study include fracture lined with quartz cement and sealed
a power law would, with an added implication opening-mode fractures in distributed, parallel by later calcite cement (CC). (C) SEM-CL
that the maximum size has been sampled. We sets. We analyzed only individual sets; where image of a macrofracture with bridging quartz
prefer to independently establish whether the multiple fracture orientations are present, we cement, sealed by calcite cement. (D) Outcrop
maximum size might have been sampled using identified a dominant strike based on fracture photograph of crosscutting fractures, each
field observations. Likewise, the stretched- strain and crosscutting relationships. Microfrac- filled with quartz and calcite cement.
exponential equation (e.g., Clauset et al., 2009; tures systematically strike parallel to the domi-
Gudmundsson and Mohajeri, 2013) is an expo- nant macrofracture strike. Fractures measured at
nential equation with an added parameter to the both scales of observation are typically planar, form only thin veneers that line fracture walls
exponent that stretches the curve. Adding this with sharp, discrete walls (Fig. 3). Macrofrac- or form fibrous or euhedral bridges that extend
parameter can potentially improve the fit of the tures dip steeply with respect to bedding. across fractures (Fig. 3). Calcite linings over-
equation, but we would still interpret such a best All fractures represented in the data set are lap quartz and typically seal fractures entirely,
fit as a characteristic size distribution, as dis- cemented by either quartz or some carbonate although rare euhedral calcite faces indicate par-
cussed in the Introduction. mineral, typically calcite. Quartz cement may tial filling by calcite cement.

1344 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

Although this study is limited to fractures that Euhedral quartz cement log-normal (samples 21 and 22) size distribu-
contain cement, the type of cement played no role tions. No observed fracture set is best fit by a
Anhedral quartz cement
in the scanline selection process. Only upon obser- normal size distribution.
vation in thin section was it confirmed that the
first cement phase deposited within all fractures in Vertical-Core Samples
each sample was quartz. Typically, this phase was
synkinematic (deposited during fracture opening). c-axis Scanlines were constructed from vertical-core
In the Piceance Basin, the Sierra Madre Oriental, thin-section samples from the Piceance Basin,
and Basin X, overlapping, postkinematic carbon- the East Texas Basin, and Basin X. Scanlines
ate cements are common, but in all three cases, Quartz grains from 46 vertical-core samples (from shallowly
later carbonate cements are heterogeneously dis- dipping beds; Table 3) have an average scanline
tributed. Such heterogeneous distributions of late- length of 77.4 mm, limited by the width of the
stage carbonate cement are common in fractured sampled core, parallel to bedding. These scan-
c-axis
tight-gas sandstones (Laubach, 2003), but causes lines intersect 12.6 fractures on average (mini-
of this heterogeneity are obscure. mum zero, maximum 104). Fracture frequency,
Crack-seal texture is common among irrespective of fracture size, averages 0.17
observed fractures and is readily apparent in per mm and ranges from zero to 1.2 per mm.
SEM-CL images. Crack-seal texture consists of Microfracture strain values are similar to those
Non-
bands of cement that run parallel to the fracture observed in horizontal and slant cores from the
quartz
walls, suggesting repeated cracking of the frac- Piceance Basin (Table 3), from which most verti-
grain
ture along fracture wall–parallel surfaces within cal-core samples are derived .
the cement or at the cement-wall interface (Ram- Of the 44 vertical-core scanlines that intersect
say, 1980; Urai et al., 1991). These bands form Fracture fractures, 27 are best fit by power-law aperture
kinematic
as cement is deposited during fracture growth. distributions; 17 are best fit by characteristic size
aperture
During growth, the fracture undergoes recurrent distributions. The average number of fractures
opening, and cement fills in across each ephem- for power-law and characteristic size distribu-
eral gap. These cement bands are broken during tions is similar (13.5 and 12.7, respectively), but
subsequent opening increments, and crosscutting Figure 4. Local preservation of opening in- strain is higher among power law–distributed
relations are evident in SEM-CL images (Becker crements (crack-seal structure) in fracture sets (8.9 × 10–3 versus 1.5 × 10–3).
et al., 2010). As this process repeats, crack-seal cement. Crack-seal texture will form if ce-
texture forms within the growing fracture. ment precipitation can keep up with frac- Power-Law Size Distributions
Whether crack-seal develops within an opening ture opening. Quartz-cement precipitation
fracture therefore depends on the relative rates of can be locally variable, depending on the Power-law size-distribution curves have a
fracture opening and cement deposition. If frac- substrate mineral and crystallographic axis coefficient a and exponent b (Table 1). The a
ture opening outpaces cement deposition, the orientation (Lander et al., 2008). parameter of the power-law equation represents
local sealing step is not achieved, and crack-seal the y-intercept of the line in log-log space [with
texture does not form. Because the rate of cement the y-axis at log(x) = 0]; the b parameter rep-
deposition is much faster on freshly broken sur- resents the line’s slope. It can be qualitatively
faces than on euhedral crystal faces (Lander et al., the pixel resolution for SEM-CL images and appreciated (Fig. 5 and Table 3) that variation in
2008), crack-seal cementation commonly forms close to the width of fractures that are ubiquitous a among individual data sets is greater than vari-
isolated bridge deposits on host-rock grains whose in natural quartz grains (Sprunt and Nur, 1979; ation in b. In fact, a increases systematically with
fast-growth crystallographic axis is oriented at a Seyedolali et al., 1997; Bernet and Bassett, increasing microfracture strain among power
high angle to the fracture (Fig. 4). 2005; Hooker and Laubach, 2007). The fracture law–distributed fracture sets (Fig. 6), whereas b
Typically, pore space is present in the largest sizes observed are therefore likely bounded at remains roughly constant with increasing strain.
fractures observed of each data set. In such large the small end by image resolution and the inde- In general, less variation in b exists among popu-
fractures, synkinematic quartz cement is present terminate genesis of intragranular fractures, as lations containing more fractures (Fig. 7), sug-
in isolated bridges, with porosity or late-stage, discussed above. However, it is not necessar- gesting that some variation of b can be attributed
porosity-occluding carbonate cement in between. ily true that the fracture sets observed contain to statistically inadequate sampling rather than
Fracture size–dependent porosity preservation has narrower, unmeasureable fractures. In the next natural variation.
been described for fractures above a characteristic section (Interpretation of Aperture-Size Distri- The greatest (steepest) value of b from those
size within a given population (Laubach, 1988, butions) we address whether the data set might data sets best fit by power-law equations is
2003; Laubach et al., 2004b; Gale et al., 2010). include the smallest extant fracture sizes. Larger 1.75, observed among microfractures in the
fractures could be present, but at each field site, Mesón Group at Yacoraite Canyon (sample
SCALING DATA or within each slant core, the sets studied each 19). The size distribution of these fractures is
feature a millimeter- to centimeter-scale maxi- poorly characterized by all of the four tested
Outcrop and Slant-Core Samples mum observed fracture aperture. equation types. Yacoraite Canyon size data are
Of 22 outcrop and slant-core fracture-size nearly equally as well fit by an exponential
Fracture sizes (apertures) within the com- data sets, 19 are best fit by power-law equations, equation as by a power law. The quality of fit
bined data set range between 0.0005 and 14 mm judging by the chi2 criterion (Table 3). The other of an exponential equation, relative to that of
(Fig. 5). The small-size end of this range is near three are best fit by exponential (sample 20) or a power law, is better among these fractures

Geological Society of America Bulletin, September/October 2014 1345


Hooker et al.

2
1
1 2 3
0
–1 Piceance Basin
–2
–3
–4
–5

2
4 5 6 7
1
0
Log cumulative frequency (fractures/mm)

–1 Mexican SMO
–2
–3
–4
–5

2
8 9 10 11
1
0
–1 Basin X
–2
–3
–4
–5

2
1
12 13 14
0
–1 NW Scotland
–2
–3
–4
–5
Argentinian Andes
2
15 16 17 18 19
1
0
–1
–2
–3
–4
–5
–4 –3 –2 –1 0 1 –4 –3 –2 –1 0 1 –4 –3 –2 –1 0 1 –4 –3 –2 –1 0 1 –4 –3 –2 –1 0 1

Log kinematic aperture (mm)


Figure 5. Apertures of power law–distributed fracture data sets, from outcrop and slant-core scanlines. Each graph is labeled with the
sample number (see Table 3). For reference, the plotted line in each graph has a slope –0.8 (b = 0.8) and y-intercept a = 10–3 fractures/mm.
SMO—Sierra Madre Oriental.

than any other power law–distributed data set distribution equation, compared with fractures Observed b values, excluding the dubious
(Fig. 8). Therefore, despite the large number of within smaller data sets. We further investigate sample 19, range from 0.35 to 1.24, with an
fractures present in the fracture set measured the anomalous scaling behavior of the fractures average of 0.77 and a standard deviation of 0.12.
from Yacoraite Canyon (n = 499), it is less clear in sample 19 using a numerical experiment in If sample 19 is included, b averages 0.82 (stan-
that this fracture set follows a power-law size- the next section. dard deviation 0.25). Values of b near 0.8 have

1346 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

(continued )
been reported for fracture sets from previous

1.31
2.50
1.18

1.08

1.78

1.16
1.16
2.04
1.30

1.49
Cv
studies (Table 4) measured using the same 1-D
sampling technique.

2.6E–04
6.9E–03
8.5E–04

9.4E–03

1.1E–02

1.8E–02
2.8E–02
2.0E–03
1.1E–02

4.3E–02
Strain
Characteristic Size Distributions

Macrofractures
Scan-line length
Among outcrop and slant-core samples, the
Vertical core samples

only characteristic fracture-size distributions

(mm)

52151
2045
33054

2972

14235

2308
3091
7198
3594

1121
were measured from samples taken from the
Mesón Group, where it is exposed at Huasamayo
Canyon (Table 3). Outcrops contain poorly
cemented, bedding-perpendicular, opening-

12*

10
26

25

114

62
123
48
42

24
N
mode fractures with regular spacing and a low
range of submillimetric aperture sizes (Hooker
et al., 2013). These macrofractures are not sys-

1.01
2.38

1.78
1.01

3.69

1.82

0.88

1.20
1.63
0.80

1.09
1.27

1.53

1.20
1.08

0.96

0.94
1.65
1.77
1.87
1.85
1.37
1.65
1.53
1.80
1.02
1.87
1.79
1.47
1.98

0.79

1.70
1.51
1.07
1.40
1.89
Cv
tematically parallel to microfractures and may be
unrelated. Average fracture aperture for Huasa-

1.0E–04

3.0E–02

5.4E–04
3.8E–03

9.3E–04

4.7E–03
5.4E–04

3.9E–03
3.4E–03
2.8E–03

1.8E–02
1.5E–03

3.3E–03

9.6E–04
1.5E–04

9.5E–03
9.1E–03
2.4E–02
3.6E–02
5.4E–02
1.3E–02
6.7E–03
1.9E–02
8.8E–03
1.6E–02
3.6E–02
2.3E–02
9.9E–02
1.9E–01

6.0E–03

1.1E–04

3.8E–03
1.9E–02
1.1E–03
4.2E–03
7.6E–03
mayo Canyon microfractures is lower than those Strain
of power law–distributed fracture sets, as is the
Microfractures

fracture strain (Table 3).


Scan-line length

Opening-Increment Width and


(mm)

1017

41

85
120

94

86
2067
554

195
195

94

75
91

83

79
89

61
222
230
122
68
205
183
125
30
48
112
97
61
131

90

84
71
46
69
46
TABLE 3. FRACTURE SCANLINE-DATA STATISTICS

Overall Fracture Aperture

For fractures within sample 8, increment


sizes are well fit by a log-normal size distri-
20

499

17
155

12

18
174
16

14
26
101
131

64

24
4

18

7
3

8
157
160
65
77
153
38
104
134
150
97
416

27
6
6
3
5
bution (Fig. 9), indicating a characteristic size
N

for opening increments. Similar distributions


of opening-increment sizes within crack-seal
8.0E–05

9.0E–05

1.0E–04
2.0E–04

1.1E–03

4.2E–03
8.0E–04
3.0E–04

3.1E–03
8.0E–04
2.0E–03
1.2E–02

4.0E–04

2.0E–06
2.0E–04

6.0E–05

7.4E–03
2.0E–02
2.4E–02
5.0E–03
3.2E–03
4.2E–03
5.5E–03
2.9E–03
5.5E–03
5.0E–03
2.4E–02
1.6E–02

8.5E–03

3.0E–05

1.1E–02

2.1E–03
1.4E–02
4.0E–04
1.0E–02
1.4E–02
textures were documented by Laubach et al.
a

(2004a), Renard et al. (2005), and Hooker et al.


(2012). This relationship appears to be typical 1.75

1.13
1.31
0.74
0.74
0.87
0.52
0.89
0.87
0.60
0.63

1.17
0.73
0.64
0.78
0.83
0.78
0.93
0.85
0.90
0.66
0.82

for observed fractures. However, note that pre-

0.7

0.7
1.9
0.8

1.2

0.4
0.3

1.2

0.3

0.8
0.3
1.1
0.2
0.4
b

served crack-seal increment widths may not


accurately reflect opening-increment sizes. If
the fracture-opening rate accelerates or if the
3.51
0.21
0.07
0.27
0.12

785.52

0.04
9.28
2.82
5.34

0.12
0.05
0.44
0.09
0.02
0.02

0.06
10.34
1.79
0.72
23.84
17.42
21.73
13.75
115.18
26.33
56.14

0.06

0.01
0.04
0.14

0.32
0.09
0.03
0.02
0.04
LN

cement-precipitation rate slows, opening incre-


ments may not be preserved in the crack-seal
record. Therefore, preserved opening-increment
18.64
0.23
0.10
0.43
3.21
217533
949.58
4.E+08

1.22

0.23
0.01
8.E+11
6.E+09
9381.7
33.04
17.42
28.12
54.22
28.28

6.E+10
1.E+10
2.E+07

0.24

0.35

0.09
0.10

0.01

0.36

741.90
0.12
0.03
0.02
0.10
1.15
5.E+09

0.75
widths may overestimate true opening-incre-
N
Chi2 error

ment width. Moreover, for any preserved open-


ing increment, the fracture may have widened
in excess of what is preserved then closed to the
2.85
0.14
0.20
0.41
0.22
23.12
8.43
28.10
45.65
3.80
3.56
65.55
33.42
35.42
37.31
45.04
29.23
294.36
345.66
0.06

0.09
0.01
0.03
0.03
0.04
0.00
0.02
0.10

0.89
0.04
0.01
0.00
0.02
0.06
2.55

0.07
E

size at which cement bridged across.


Most fracture sets within the collective data
set feature some crack-seal texture, but fractures
0.17
0.02
0.01
0.02
0.01
0.65
0.64
0.56
1.80
0.07
0.19
1.14
0.83
3.52
1.08
4.60
1.28
10.04

0.34
0.01
0.04
0.01
0.15
0.00
0.01
0.02
1.75
0.22
22.08

0.06
0.24
0.01
0.02
0.00
0.00
PL

that follow characteristic size distributions are


330.6

mostly devoid of crack-seal (Fig. 10). Char-


acteristic-size microfractures from the Mesón
Formation

Group are thin (<10 microns in aperture), dark


WF

WF
WF
WF
WF
WF
WF
WF
WF
EA
EA
CZ
CZ

TP
TP
TP
TP
TP
TP
FX
FX
FX
FX
LB
LB

M
M
M
M
M
M
M
M
E
E
E

luminescing, and mostly textureless. No charac-


teristic size–distributed fracture sets have frac-
tures more than two to three increments wide,
as preserved within crack-seal texture. Because
Location

MSM
MSM
MSM
MSM

NWS
NWS
NWS
NPB

NPB
NPB
NPB
NPB
NPB
NPB
NPB
NPB
SPB
SPB

ETB
ETB
ETB
ETB
ETB
ETB
AA
AA
AA
AA
AA
AA
AA
AA
BX
BX
BX
BX

of limited resolution of the SEM-CL imaging


technique, these observations do not preclude
the possibility of submicron-scale crack-seal
bands within these fractures. However, such
Sample

hypothetical, submicron-scale, crack-seal bands


c01
c02
c03
c04
c05
c06
c07
c08
c09
c10

c12
c13
c14
c11
10

12
13
14
15
16
17
18
19
20
21
22
11

do not achieve wide variations in fracture size,


1
2
3
4
5
6
7
8
9

Geological Society of America Bulletin, September/October 2014 1347


TABLE 3. FRACTURE SCANLINE-DATA STATISTICS (continued)

1348
Chi2 error Microfractures Macrofractures
Scan-line length Scan-line length
Sample Location Formation PL E N LN b a N (mm) Strain Cv N (mm) Strain Cv
c15 NPB WF 0.00 0.00 0.02 0.02 0.5 2.4E–03 3 61 5.2E–04 1.36
c16 NPB WF 0.02 0.06 0.18 0.08 0.5 6.5E–03 10 92 4.8E–03 1.28
c17 NPB WF 0.24 2.79 102226 18.22 0.8 3.5E–03 37 86 1.0E–02 2.90
c18 NPB WF 0.02 0.04 0.25 0.05 0.3 1.9E–02 10 75 1.7E–02 1.03
c19 NPB WF 0.00 0.01 0.05 0.03 0.6 1.7E–03 5 94 6.7E–04 0.95
c20 NPB WF 0.04 0.29 3.18 0.51 0.5 7.9E–03 14 93 1.3E–02 1.24
c21 NPB WF 0.04 0.14 0.32 0.19 0.4 1.2E–02 12 78 1.2E–02 1.31
c22 NPB WF 0.02 0.02 0.04 0.05 1.2 8.0E–05 7 75 4.5E–04 1.41
c23 NPB WF 3.95 28.63 50.97 25.93 0.8 6.4E–03 104 90 1.0E–01 2.03
c24 NPB WF 0.04 0.01 0.06 0.03 1.5 1.0E–05 7 95 4.2E–04 0.67
c25 NPB WF 0.00 0.00 0.02 0.02 0.3 8.5E–03 3 91 5.0E–03 1.40
c26 NPB WF 0.00 0.00 0.01 0.01 1.5 5.0E–06 2 88 1.1E–04
c27 NPB WF 0.02 0.09 0.31 0.16 0.4 1.0E–02 6 53 1.5E–03 0.74
c28 NPB WF 0.00 0.02 0.07 0.04 0.3 7.5E–03 5 65 2.7E–03 0.94
c29 NPB WF 0.00 0.00 0.02 0.02 0.6 2.3E–03 3 90 7.6E–04 1.27
c30 NPB WF 0 87 0
c31 NPB WF 0.04 0.15 1.24 0.12 0.6 3.1E–03 12 88 3.0E–03 1.46
c32 NPB WF 0.33 0.61 780.89 0.11 0.8 3.7E–03 36 89 6.8E–03 2.70
c33 NPB WF 0.03 0.27 3.12 0.42 0.6 3.6E–03 14 93 3.3E–03 1.08
c34 NPB WF 0.00 0.00 0.01 0.01 0.3 9.5E–03 3 91 4.9E–03 0.60
Vertical core samples

c35 SPB WF 0.06 0.02 0.11 0.06 1.8 1.0E–06 14 108 2.1E–04 1.61
c36 SPB WF 0.01 0.00 0.04 0.02 1.4 2.0E–05 10 98 3.7E–04 1.39
c37 SPB WF 0.00 0.01 0.04 0.03 1.6 3.0E–06 7 93 1.9E–04 0.86
c38 SPB WF 0.04 0.10 0.33 0.17 1.2 7.0E–05 20 98 7.0E–04 1.38
c39 SPB WF 0.01 0.00 0.12 0.08 2.0 2.0E–06 6 22 7.4E–04 0.90
c40 SPB WF 0.47 0.16 0.74 0.16 1.3 8.0E–05 14 38 7.2E–04 1.30
c41 SPB WF 0 20 0
c42 SPB WF 0.05 0.63 21.89 0.24 0.7 4.8E–03 18 39 6.1E–03 1.38
c43 SPB WF 0.29 0.03 0.18 0.06 0.8 2.1E–03 12 43 1.6E–03 1.03
c44 BX FX 0.07 0.01 0.13 0.03 1.1 1.0E–04 12 94 6.0E–04 1.06
Hooker et al.

c45 BX FX 0.03 0.01 0.02 0.02 1.1 2.0E–04 8 82 4.7E–04 1.39


c46 BX FX 0.22 0.06 1.28 0.08 1.6 3.0E–05 19 90 1.3E–03 1.30
Note: Locations: NPB—north Piceance Basin; SPB—south Piceance Basin; MSM—Mexican Sierra Madre Oriental; BX—Basin X; NWS—northwest Scotland; AA—Argentinian Andes; ETB—East Texas Basin.
Formations: WF—Williams Fork Formation; CZ—Cozzette Sandstone, Iles Formation; LB—La Boca Formation; EA—El Alamar Formation; FX—Formation X; E—Eriboll Formation; M—Mesón Group; TP—Travis
Peak Formation. PL—power-law; E—exponential; N—normal; LN—log-normal; b—power-law exponent; a—power-law coefficient; N—number of fractures; Cv—coefficient of variation, standard deviation of
spacings divided by average spacing. Best-fit chi2 error value in bold.
*All 12 macrofractures within sample 1 are included in the microfracture total, which spans the entire cored interval.

b Log10 a
0
1
2

0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
–6
–5
–4
–3
–2
–1
0

Geological Society of America Bulletin, September/October 2014


0.0001

in sample 19.
0.001

distribution equations.

fracture-bridging cements.

SIZE DISTRIBUTIONS
0.01
0.1

Curves among Small Fracture Sizes


Microfracture strain

INTERPRETATION OF APERTURE-

Truncation Bias versus Fracture Localization

be the case. One is the aforementioned charac-


mum opening distance achieved when fractures
may be real and represent a characteristic mini-
Alternatively, the concave-downward curves
sample 6 and among fractures <0.004 mm wide
are present among fractures <0.03 mm wide in
bias. Representative examples of such curves
bution that are not consistent with truncation
ward curves at the small-size end of the distri-
examples, other samples have concave-down-
log space, but the residuals are progressively
above the pixel size form a straight line in log-
power law–distributed data, because the data
scaling curves are consistent with truncation of
sured. In samples 7 and 15, for example, such
present among the smallest aperture sizes mea-
Concave-downward scaling curves are locally
involved repeated breaking and re-sealing of
was incremental, then it does not appear to have
opened incrementally. But if fracture opening
does not preclude the possibility that fractures
ture sets. Furthermore, absence of crack-seal
as is the case for power law–distributed frac-
all data sets best fit by power-law aperture
Figure 6. Power-law coefficient (a) and ex-
1

propagate. Two observations suggest this might


negative near the pixel size. In contrast to these
ponent (b) versus microfracture strain, for
A universal power-law scaling exponent for fracture apertures in sandstones

2 A (O
v
fra erall)
Sample 19 ap cture
ert
1.5 siz ure
e
Fra
op cture
inc e n i n g
1 re
wid ment
b

ths
Quartz grain

0.5
Slant core and outcrop
Vertical core 1000
0 B Power law
0 100 200 300 400 500 600 Exponential

Cumulative number
Normal
Number of fractures Log-normal
100

Figure 7. Power-law exponent (b) versus number of fractures intersected by scanline, for Crack-seal
fracture sets best fit by power-law size distributions. In general, b converges with higher increment widths

number of fractures sampled. Sample 19 is a notable exception; see text for discussion. 10

100 1
(chi2EXP-chi2PL) / chi2PL

0.0001 0.001 0.01 0.1 1


Increment size (mm)

relative to exponential
10 Better power-law fit,
Figure 9. (A) Scanning electron microscope
cathodoluminescence image of fracture
1
from Formation X (sample 8) that grew by
repeated, incremental opening (crack-seal).
(B) Cumulative plot of widths of 11 such
0.1 fractures’ opening increments, within the
same sample. The opening increments are
19 best fit by a log-normal equation.
0.01
1 10 100 1000

Number of fractures teristic size of crack-seal increment widths—


fractures from Formation X, for example,
Figure 8. The y-axis is a relative measure of the quality of the power-law equation fit to the appear to be composed of multiple characteris-
exponential equation fit. It is the difference between the chi2 error from the two fits, divided tic-size opening increments (Fig. 9). The other
by the chi2 error from the power-law fit. This is plotted versus number of fractures that observation is apparently imperfect localization
are best fit by power-law size distributions. As more fractures are sampled, generally the of fractures, which is present locally along frac-
relative fit of the power-law equation becomes progressively better; as in Figure 7, sample tures within Formation X (Fig. 9) and within
19 is here again anomalous. Though best fit by a power-law size-distribution equation, this the El Alamar Formation, sample 6 (Fig. 11A).
data set is nearly as well fit by an exponential equation; other large data sets are easier to SEM-CL images reveal that neighboring open-
distinguish. ing increments may be generally connected and
thought of as parts of a single fracture, but are
locally dispersed, with host rock in between.
TABLE 4. VALUES OF POWER-LAW EXPONENT b REPORTED IN LITERATURE, FOR OPENING-MODE-
FRACTURE APERTURES MEASURED ALONG ONE-DIMENSIONAL SCANLINES
Therefore, a large fracture may be recorded
Author Year b Rock type
within the size distribution as either a single,
Gross and Engelder 1995 0.45 to 0.85 Dolostone
large fracture or a cluster of tiny fractures,
sedimentary rocks

Marrett 1997 0.758, 0.764 Sandstone depending on the location at which the scanline
Exclusively

Marrett et al. 1999 0.769, 1.042 Dolostone, sandstone intersects fractures that locally branch. Note that
Ortega et al. 2006 0.69 to 1.0 Limestone, dolostone
Ortega et al. 2010 0.2 to 2.0 Limestone, dolostone fractures within Formation X (sample 8) do not
Hooker et al. 2011 0.56 to 1.17 Quartzarenite feature a concave-downward curve in Figure 5
Guerriero et al. 2011 0.95 to 1.44 Limestone, dolostone because in that sample the fracture-trace disper-
Hooker et al. 2012 0.5 to 1.0 Limestone, dolostone
Wong et al. 1989 0.843, 0.804 Granite, quartzite
sion is minor, present only where fractures cut
Other rock types

Clark et al. 1995 1.33 Meta-turbidite monocrystalline quartz grains (Fig. 9).
McCaffrey and Johnston 1996 0.58 to 1.13 Metasediments Fractures within sample 19 show the same
Gillespie et al. 1999 0.49 to 1.24 Metasediments, granite, limestone, sandstone
Sasaki 2006 0.89 to 1.35 Volcanics, sedimentary rocks phenomenon—dispersed fracture traces with
André-Mayer and Sausse 2007 1.04 to 1.92 Igneous-hosted porphyry veins host rock in between crack opening incre-
Fagereng 2011 1.24 to 3.01 Mélange ments—to a greater degree (Fig. 11B). In this

Geological Society of America Bulletin, September/October 2014 1349


Hooker et al.

A the cement precipitation rate had been slower,


A relative to the fracture opening rate, fractures in
Fracture
these samples would have been less thoroughly
Non-quartz sealed with cement during growth and would
grain have formed as discrete, larger fractures. Instead,
the cementation rate was sufficiently fast to
cause later opening increments to imperfectly
Quartz grain
localize along earlier ones, the same way in
which the opening increments are locally diffuse
in Figures 9 and 11. With slower cement pre-
cipitation, the characteristic-size opening incre-
100 μm
200 μm ments would combine to make larger fractures,
B and fewer fractures would therefore intersect the
B
scanline. As observed, diffuse fracturing is mani-
fest in scanline data as a greater number of small
surface pits (characteristic-sized, <0.01-mm-wide) fractures.
Fracture trace Lacking a direct test of this hypothesis, the
effect of diffuse fracturing on size distributions
can nevertheless be examined via a geometric
experiment, eliminating the smallest spacings
from the scanline (Fig. 12A). We rearranged the
scanline data in order to create a hypothetical
arrangement in which the fractures separated by
100 μm 200 μm
distances smaller than a threshold value formed
Figure 10. (A) Cathodoluminescence (CL) Figure 11. Scanning electron microscope together as a single fracture. A threshold spac-
and (B) secondary electron (SE) images of cathodoluminescence images of branching ing value of 0.15 mm was applied, and any
microfracture from sample 20. Most of the microfractures. (A) Sample 6, El Alamar fracture increments separated by less than this
fracture is entirely occluded by textureless Formation. Dashed lines indicate fracture threshold value were treated as a single frac-
quartz cement (luminescing medium gray). walls away from quartz grain, in which ture. The hypothesis is thereby made that frac-
Where the fracture cuts a non-quartz grain, the fracture branches into numerous seg- tures separated by less than this threshold value
visible in CL by irregular and dark lumines- ments. (B) Sample 19, Mesón Group. Crack- would have formed together, as a larger fracture
cence, and in SE by pitted surface, the frac- opening increments show little tendency to with no separation in between, if the cement
ture is porous (bright CL and pitted trace in coalesce into larger fractures with crack- precipitation rate had been slower. Figure 12B
SE). (A) modified from Hooker et al. (2013). seal texture. shows the original microfracture size distribu-
tions from samples 6 and 19, as well as those
resulting from the summation of nearby fracture
sample, most fractures are individual crack increments with local intervening host-rock increments. The latter, hypothetical size distri-
opening increments separated by host rock, and slivers where cement precipitation was fast. butions are better fit by power-law equations;
individual fractures therefore lack crack-seal This local branching of the fracture trace is a as well, the power-law equations best fit to the
texture sensu stricto. consequence of inconsistency in the locations of resulting size distributions have b values closer
If power-law size distributions indeed arise re-cracking surfaces within the cement bridges. to the average for power-law data sets (0.8).
from localization of micron-scale-width open- If a single fracture can be locally discrete where
ing increments into larger fractures, perhaps cement precipitated slowly and locally diffuse Curves among Large Fracture Sizes
deviations from the power law would be intro- where cement precipitated quickly, then an
duced if this localization mechanism were entire fracture, or most of a fracture, might be Fracture-size data also typically deviate from
imperfect. The microfractures imaged in Fig- diffuse if cement precipitation were fast, relative best-fit power-law equations at the large-size
ures 9 and 11 illustrate grain-scale cement- to fracture opening, throughout. Such appears end of the data distribution. This deviation has
precipitation effects on fracture opening. In to have been the case in sample 19 (Fig. 11B). at least five plausible explanations:
Figures 9 and 11A, where fractures cut quartz Microfractures within samples 6 and 19 super- 1. The largest fracture sizes sampled in a
grains on which quartz cement precipitation was ficially resemble fractures described by Caputo scanline are necessarily statistically inade-
rapid enough to bridge the opening gaps as the and Hancock (1998), which these authors inter- quate. As explained by Guerriero et al. (2010),
fracture grew, the bridges had to re-crack with preted to branch because the host-rock material progressively smaller, cumulative sample sizes
every widening increment. In regions of the is weaker than the fracture fill. increase the statistical error of the cumulative
same fractures where cement precipitation was frequency calculation toward the right side of
slower, because the fractures remained open, Experiment to Investigate the Effects of the scaling plot. Greater deviation from any
no cohesive material was there to break as the Localization on the Size Distribution underlying ideal size distribution can therefore
fracture widened. Consequently, what appears Consider the hypothesis that concave-down- be expected to increase among the largest frac-
to be a single fracture with unambiguous walls ward curves among the microfracture size dis- ture sizes encountered in a single data set. This
where cement precipitation was slow appears tributions in samples 6 and 19 are not related to deviation may occur upward or downward from
elsewhere as a cluster of individual opening bias, but instead represent diffuse fracturing. If the distribution regressed to smaller fractures.

1350 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

100 μm
Original data Combine fractures with spacing < threshold

{
15 μm threshold

Result

B 1000 Observed
distribution b = 1.75
Accumulated
Cumulative number

b = 0.98 microfractures

100

10

b = 0.67
b = 0.80
Sample 6 Sample 19
1
0.0001 0.001 0.01 0.1 1 10 0.0001 0.001 0.01 0.1 1 10

Kinematic aperture (mm)


Figure 12. Fracture size distribution numerical experiment. (A) To test what the size distribution would be had separate fracture
increments formed together, fractures separated by less than a threshold value are combined to a single fracture. (B) Original
microfracture data have concave-downward curves to their size distributions. Each resulting size distribution is better fit by a
power-law equation with the exponent b closer to the typical value observed in sandstones (0.8).

2. Rocks undergo finite strain. The large-size fracture size. Because fractures may be sam- in the field and analogous to the upper mag-
end of the distribution may systematically curve pled near their upper size limit, the downward nitude limit of earthquakes (e.g., Sornette and
downward from a power-law equation because curve to the size distribution may be real. An Sornette, 1999).
the rocks have undergone only limited strain. upper size limit is consistent with the paucity 3. Opening-mode fractures are commonly
Finite strain necessitates some upper limit to of centimeter-scale or wider fractures exposed height restricted by their host strata. Assuming

Geological Society of America Bulletin, September/October 2014 1351


Hooker et al.

fracture propagation stops, and aperture goes APPLICATION (1985) and Lorenz et al. (1991), of parallel frac-
to zero, at a layer boundary, linear-elastic frac- tures forming over a broad area in response to
ture mechanics predicts that fracture aperture Open-Fracture Spacing from a single loading event. If fractures formed in
should scale with fracture height, height being Microfracture Scaling response to local folding, then such scaling esti-
the shortest length dimension (Olson, 2003). mates may be applicable only to local structural
Thus a power law among small fracture aper- Open-fracture spacing in the subsurface can domains having equivalent fracture strain, as we
tures might be interrupted among fractures be estimated by solving a power-law fracture- discuss later in this section.
long enough to reach bedding boundaries frequency equation using a fracture size equal Below we subdivide our macrofracture data
(Bonnet et al., 2001). In our data set the frac- to the emergent threshold—the fracture aperture in order to test whether typical data sets derived
ture arrays most strongly bound to sedimentary above which some porosity is preserved (Lau- from core-width scanlines (<100 mm in length)
layering are also among the lowest-strain rocks bach, 2003). The reciprocal of the calculated in tight-gas sandstone reservoirs can be used to
(Piceance and East Texas Basins), so it is dif- frequency is the average spacing of fractures. predict macrofracture frequency. Microfracture-
ficult to distinguish the relative importance of Such a frequency model can be derived by derived scaling extrapolations were tested by
layer thickness, finite strain, and poor sampling regressing a scaling equation to observed size- subsampling scanlines made from horizontal
on the large-size end of these aperture distribu- frequency data (see Curve-Fitting and Sampling and slant cores within the Piceance Basin (sam-
tions. Macroscopic fractures in the beds from Biases, above). The examples presented here ples 1, 2, and 3). Because typical vertical cores
which samples 20–22 were taken are also suggest that a power-law scaling equation more are no wider than 100 mm, the scanline data
strongly layer bound and too thin to reliably commonly best fits natural-fracture data than from these microfracture samples were divided
measure in the field, but it is unclear whether will exponential, normal, or log-normal scaling into 100-mm-long segments, each of which was
these macro- and microscopic fractures are equations. Notable exceptions may be made if best fit with power-law equations. First, these
genetically related (Hooker et al., 2013). the fractures most likely have a characteristic tasks were done using Microsoft Excel’s auto-
4. Progressive fracture strain is commonly size distribution, which may be the case, for mated regression routine, which varies a and b.
accommodated by faults. Faults may form with example, if crack-seal texture is absent. As an alternative method, the frequency equa-
progressive strain, either by fracture propa- Such possibilities notwithstanding, natural- tion was solved by holding b fixed at 0.8 and
gation to depths of higher differential stress fracture sets that grew in diagenetically reac- solving for a at each data point, averaging a val-
(Gudmundsson, 1987) or by a relative rotation tive environments tend to follow power-law ues to establish a frequency equation. For zero
between the stress field and fracture orientation aperture-size distributions, with b near 0.8, as microfractures intersected, zero macrofractures
that produces shear stress on fracture surfaces measured in 1-D. The prevalence of this b value are expected (infinite spacing).
(Gudmundsson, 1992; Myers and Aydin, 2004). has two important ramifications for estimating For comparison, we include an analogous
It is highly likely that a scaling law for opening- open-fracture spacing in the subsurface. subsampling of two outcrop samples: one from
mode apertures would be modified by the emer- First, although power laws in general imply the Mesón Group (sample 18) and one from the
gence of faults, because fault displacements are the presence of systematic microfracture popu- Huizachal Group (sample 6). These samples
not perpendicular to the fault plane. As stated in lations whereas characteristic size distributions were selected for their long microfracture scan-
the Geologic Settings section above, most of our do not, the abundance of microfractures depends lines (Table 3). As well, these fracture popula-
samples derive from fold-and-thrust belts; these on b. The shallow b value (<1) present among tions record higher strain, and allow us to assess
settings commonly include faults that strike par- natural fractures means that for every macro- the accuracy and precision of microfracture-
allel to opening-mode fractures. In this study we fracture there aren’t myriad microfractures based fracture-frequency predictions.
exclusively measured opening-mode fractures, present, especially considering the ambiguity of
so a transition to faults might simply show up in origin of fractures smaller than the grain scale Macrofracture Spacing from Core
our data as a downward roll-off to the size dis- (Hooker and Laubach, 2007). Imagine a hypo- and Outcrop Observation
tribution. Gudmundsson and Mohajeri (2013) thetical case in which the average spacing of
investigated the scaling transition between visible fractures (>0.1 mm aperture) is equal The north Piceance Basin sample (sample 1)
opening- and shear-mode fracture lengths in to a core width (100 mm). If such a population derives from slant core, drilled ~45° from verti-
Iceland, interpreting a steeper scaling exponent were to follow a power-law size distribution of cal. Fracture data were collected from a sand-
for longer fractures with shear offset. b = 0.8, then a, in fractures per mm, would be stone bed within the Williams Fork Formation,
5. Lastly, for fracture populations that fol- 0.0016. Because of this shallow slope, ~40 frac- at least 2 m thick, at a measured depth (MD)
low power-law aperture-size distributions with tures more than 1 micron wide would be present of 3138–3140 m (true vertical depth [TVD]
< 1, the frequency of large fractures must nec- within a core width, a total substantially below 2649 m). Near-vertical fractures intersect the
essarily fall below the best-fit size-distribution the 200 called for by Bonnet et al. (2001) to core obliquely (Fig. 13), and abundant calcite
equations because such equations eventually independently establish a size-distribution equa- cement is present that overlaps quartz cement.
predict a greater frequency of large fractures tion. In contrast, if the scaling exponent were The microfracture scanline for this sample
than can fit within the rock observed (Hooker 2.5, there would be 1000 microfractures pres- comprises 72 bedding-parallel thin sections,
et al., 2011). In each observed scanline, ent, and statistically robust populations could be arranged in a stair-step pattern so as to sample
because ample space is available for more large sampled within a single core width. the maximum layer-parallel length possible
fractures to fit into, such geometric limitations On the other hand, the low variability of (~2 m). The scanline therefore moves progres-
do not apply to curves to the large-size ends the exponent carries an implicit benefit: if b is sively lower within the bed (Fig. 14).
of the distributions. However, observed power roughly constant, then we can assume b = 0.8 South Piceance Basin samples (samples 2 and
laws must end at some upper size limit, regard- and vary only a to regress the data. Such an 3) derive from cores drilled parallel to layering
less of whether the upper limit is sampled in approach presupposes the existence of a regional through the Cozzette Sandstone member of the
this study. fracture set, in the sense described by Hancock Iles Formation, which underlies the Williams

1352 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

A long microfracture sample was taken, intersect- contain more microfractures within 100-mm-
Sample 3 ing 416 microfractures (Fig. 15). long scanlines, and the predicted macrofracture
MD 2177.0 m The Huizachal Group fractures (sample 6) are frequency is closer to the observed frequency
a set of east-west–striking fractures exposed on (Fig. 16). Error bars are substantially narrower
a canyon floor (Fig. 3) containing at least three for high-fracture-frequency samples, but for
other fracture sets (Laubach and Ward, 2006). all samples error is decreased by assuming b =
These fractures contain abundant calcite cement 0.8. Our least-accurate result was derived from
Fracture overlapping quartz and little or no porosity. the slant core (sample 1), from which only 12
face From the Huizachal Group, 114 macrofractures macrofractures were observed in ~2 m of layer-
were measured over a 14.2-m-long exposure, parallel rock (Fig. 14). Thus, the poor predic-
2 cm
from which a 222-mm-long microfracture sam- tion from this sample may result from the short
ple was taken, intersecting 101 microfractures macrofracture scanline.
(Fig. 15). We performed the same analysis on the
B
The average macrofracture spacing for each 46 vertical-core samples from three basins
of the five samples used in this exercise was (Table 3). Microfracture spacing data are highly
calculated simply as the number of macroscopic variable within basins (Fig. 17A). Future work
fractures divided by scanline length. To estimate could identify correlations between these results
error for the observed frequency of macrofrac- and other hypothetical controls on fracture fre-
tures, we assumed that macrofractures were quency, such as layer thickness, rock mechanical
randomly spaced and that macrofracture inter- properties, and proximity to local structures, as
section by the scanline is therefore a Poissonian well as correlations with other methods of frac-
Sample 1
process (Guerriero et al., 2010). Therefore the ture prediction, such as seismic anisotropy.
2 cm
MD 3137.6 m
expected fracture spacing S is: The wide variation in a among core-width
scanlines implies an inverse correlation between
S = L/N, (2) fracture strain and predicted fracture spacing
Figure 13. Macrofractures in slant cores re-
(Fig. 17B). A consistent relationship between
covered from the Piceance Basin. (A) Frac-
where L is scanline length and N is number of fracture strain and macrofracture spacing could
ture has a thin veneer of quartz cement.
fractures intersected. The upper and lower 95% potentially simplify the estimation process. For
(B) Fracture covered in calcite cement;
confidence intervals are: ideal power-law size distributions having the
microscopy revealed quartz-cement lining.
same largest fracture size, strain is expected
Modified from Hooker and Laubach (2010).
S025,975 = L/N025,975, (3) to vary with a1/b (Marrett, 1996). For b = 0.8,
1/b = 1.25. We observe that overall microfrac-
where N025 and N975 are the mean values of the ture strain varies as a1.7. Strain proportional to
cumulative Poisson distributions for which a1.25 roughly describes a lower limit to the data.
Fork Formation. Macrofractures within these the probability of intersecting N or more fractures The observed result is consistent with high strain
samples contain no calcite cement and only thin is 0.025 and 0.975, respectively. As we discuss measurements from varying largest-fracture size
veneers of quartz cement (mostly <1 mm thick) later in this section, if fractures are systematically among the sampled populations. Large frac-
lining the fracture walls (Fig. 13). The micro- clustered, this method of error quantification may tures by definition provide more fracture strain
fracture scanlines for sample 2 comprise 16 total underestimate true spacing uncertainty. than small fractures, but the spacing prediction
thin sections, taken from 2755 m MD (8 thin and a calculation are equally sensitive to each
sections) and 2762 m (8 thin sections). These Macrofracture-Spacing Prediction size-frequency data point. Therefore, chance
scanline data were reported in Hooker et al. using Microfractures intersection of large fractures will skew strain
(2009). All sample 2 core was recovered from calculations higher than frequency (or spacing)
2420 m TVD. The microfracture scanline for To estimate macrofracture spacing using calculations. To control for this effect, strain can
sample 3 comprises 29 thin sections, sampled power laws extrapolated from microfracture be calculated from only microfractures smaller
from 2135 m MD (16 thin sections) and 2167 m size-frequency data, a fracture aperture must be than 0.01 mm. Strain at this size range varies
(13 thin sections). TVD for all scanlines from selected for which average spacing can be cal- as a 0.7 (Fig. 17C), possibly reflecting varying
sample 3 is 2045 m. culated (Ortega et al., 2006). We used 0.5 mm, degrees of truncation from varying grain sizes.
Because all three Piceance Basin cores are which is approximately the smallest macro- As a consequence, strain is a poor predictor of
typically broken along natural fractures, frac- fracture size that can be reliably detected (i.e., macrofracture spacing using this technique.
ture apertures are minimum estimates made by smaller macrofracture sizes may be truncated).
measuring the thickness of cement deposits that Microfracture frequency, measured from Effects of Clustering
line the fracture walls. 100-mm-long scanlines, varies systematically
The Mesón Group fractures (sample 18) rep- with macrofracture spacing (Fig. 16). In all three Many natural fracture sets are not regularly
resent a single set of parallel fractures. All frac- Piceance Basin core samples analyzed, the dis- spaced but, rather, are either randomly arranged
tures are filled mostly with quartz cement, with parity between observed and predicted macro- or systematically clustered. Spacings between
local porosity among quartz-cement bridges fracture spacing is smaller using the constant-b randomly arranged fractures should have a
within the widest fractures. From the Mesón assumption, versus varying both a and b (Fig. coefficient of variation, Cv, near 1:
Group, 24 macrofractures were measured over 16). Relative to the Piceance Basin samples,
a 1.1-m-long exposure, from which a 131-mm- the higher-fracture-frequency outcrop samples Cv = σ/μ, (4)

Geological Society of America Bulletin, September/October 2014 1353


Hooker et al.

(Not preserved) MD = 3,138 m TVD = 2,649 m

Microfracture kinematic aperture (mm)


10
TS Sandstone
TS

TS
1
TS
Sample 1 core
TS
drilled at ~45° to bedding
TS
0.1
TS

TS
True
bedding
TS
0.01 thickness
TS > 2m

TS

TS
0.001
TS

TS
Two apparent macrofracture
clusters intersected TS

TS

Sandstone TS
Shale
0 1000 2000
Horizontal distance (mm)
10 10

Sample 2 macrofractures Sample 3 macrofractures

1 1
Kinematic aperture (mm)

0.1 0.1

0.01 0.01
0 5000 10000 15000 20000 25000 30000 35000 0 10000 20000 30000 40000 50000 60000

10 10
Sample 2 Sample 3
microfractures 6. microfractures
1 1
1. 2. 3. 4. 5. 1. 2. 3. 4. 5. 6. 7. 8. 9.

0.1 0.1

0.01 0.01

0.001 0.001
0 100 200 300 400 500 600 0 200 400 600 800 1000

Distance along scanline (mm)


Figure 14. Slant-core macrofracture (filled symbols) and microfracture (open symbols) scanline data. Sample 1 was
taken from core drilled at ~45° to bedding, so thin-section surfaces (TS) were cut in a stair-stepping fashion. Samples 2
and 3 were taken from layer-parallel cores. Each microfracture scanline was divided into 100-mm-long subsets (num-
bered arrows); the data within subsets were then used to predict macrofracture spacing.

1354 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

10
El Alamar Fm. macrofractures Sample 6

0.1
0 2000 4000 6000 8000 10000 12000 14000 16000
10
El Alamar Fm. microfractures Sample 6
1
1. 2.
0.1
3.

0.01

0.001
0 50 100 150 200 250
Kinematic aperture (mm)

10
Mesón Gp. macrofractures Sample 18

0.1
0 200 400 600 800 1000 1200

10
Mesón Gp. microfractures Sample 18
1
1.
2.
0.1

0.01

0.001

0.0001
0 20 40 60 80 100 120 140

Distance along scanline (mm)


Figure 15. Microfracture (open symbols) and macrofracture (filled symbols) apertures measured along scanlines, samples 6 and 18. Micro-
fracture data are subdivided into 100-mm-long subsets (numbered arrows). Subsets were used to predict macrofracture spacing.

Geological Society of America Bulletin, September/October 2014 1355


Hooker et al.

Sample
number
A
Mesón Gp. 18
El Alamar Fm. 6

Mesaverde Gp.
(Williams Fork Fm.)
1

Mesaverde Gp.
(Iles Fm.)
2

Mesaverde Gp.
(Iles Fm.)
3

10 100 1000 10000 100000

Mesón Gp. 18
B
El Alamar Fm. 6

Mesaverde Gp.
(Williams Fork Fm.)
1

Mesaverde Gp.
(Iles Fm.)
2

Mesaverde Gp.
(Iles Fm.)
3

10 100 1000 10000 100000

Average fracture spacing, mm


Figure 16. Predictions of macrofracture spacing based on power-law extrapolations of microfracture size distributions.
(A) Best-fit method, varying power-law coefficient (a) and exponent (b). (B) Best-fit method holding b constant at 0.8 and vary-
ing a. Vertical bars are average macrofracture spacing (solid) with 95% confidence intervals (dashed). Filled symbols are pre-
dictions made from all microfractures measured; open symbols are the same data split into 100-mm-long scanline segments.

1356 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

1 conditions, including host rocks having extreme


A East Texas Basin
subcritical-crack-index values (Olson, 1993;
Northern Piceance Basin
Olson et al., 2009). Such fractures may open
Southern Piceance Basin
from increased effective tension near the tips
0.1 Basin X
of other growing fractures, resulting in clusters
y = 61.398x-1.677 that formed by feedback inherent to the growing
fracture population. Internally imposed clus-
0.01 ters present a challenge to subsurface fracture-
pattern characterization, which is outside the
scope of this paper.

0.001
Comparison with Narr’s (1996) Method

The method described by Narr (1996) to esti-


0.0001
mate subsurface fracture spacing (see Previous
10 100 1000 10000 100000 Methods, above) was applied, in that study, to
Figure 17. (A) Macrofracture
Predicted fracture spacing, 0.5 mm-wide fractures fractures within cores of the U.S. Department
(aperture ≥ 0.5 mm) spacing
of Energy’s Multiwell Experiment (MWX) site
predictions made from micro- 1
(samples c39–c42; Table 3), which are vertical
fractures measured from scan- B
Microfracture strain

cores drilled at the same site as the SHCT-1 core


lines across vertical cores. y = 155.63x1.677
(sample 2). Thus, the fracture spacings mea-
(B) Measured microfracture 0.1
sured from the SHCT-1 core were used as a test
strain versus best-fit power-law
of Narr’s method. Unlike the scaling method
coefficient (a) value, holding
0.01 presented in this study, the Narr method is not
the exponent (b) constant at 0.8.
specific to a single stratigraphic interval, but,
(C) Same data as in B but with
rather, core-fracture geometries are collected
strain calculated only from frac- 0.001
from as many intervals as are represented within
tures <0.01 mm wide.
the core. Narr’s method was first applied to
0.0001
Theoretical slope = 1.25 sandstone layers within the MWX core and not
1 to intervening fractureless mudstones.
C Open-fracture spacing estimates using the
microfracture scaling method, with associated
0.1
y = 0.0573x0.720 error bars, are similar to those made by Narr
(1996). Estimates overlap with those made
using the Narr (1996) method (Fig. 18).
0.01
Spacing estimates were also calculated for
this study based on microfracture frequencies
0.001 observed from four depths within the Williams
Fork interval of the MWX well. MWX micro-
fracture scanlines span single thin sections only.
0.0001 From one scanline (depth 1503 m), because no
0.0001 0.001 0.01
fractures are present, infinite spacing is pre-
a dicted using the scaling method. Estimates from
the other three scanlines indicate closer frac-
ture spacing within the Williams Fork interval
where μ is the average fracture spacing and 2012), proximity to folds or faults (Anders and compared to the paludal zone and the Cozzette
σ is the standard deviation (Gillespie et al., Wiltschko, 1994; Guerriero et al., 2013; Iñigo Sandstone. This result is consistent with higher
1999). Fracture populations presented here are et al., 2012), preferential fracturing of hetero- fracture frequency in the Williams Fork Forma-
generally more clustered than random arrange- geneously distributed cements (Eichhubl et al., tion, as measured in sample 1 (Fig. 18; Table 3),
ments, by this measure (Table 3). If a fracture 2004, 2009), or local fluid-pressure anomalies although our samples of the two stratigraphic
population is more clustered than random, then (Jolly and Lonergan, 2002; Philipp, 2008), then intervals derive from different parts of the basin
observed fracture frequency in the inter-well- an adequate description of far-field fracture (see Appendix).
bore space may show more variation than the frequency must take into account the external
error bars in macrofracture spacing (Figs. 16, control(s). Perhaps frequency systematically IMPLICATIONS FOR FRACTURE
17) imply. Note that clustering of fractures can varies within different spatial domains of the GROWTH AND THE EVOLUTION
be either externally or internally controlled. fracture population. OF FRACTURE SETS
If fracture clusters are externally controlled, Internally controlled fracture clustering may
as from changes in layer thickness (Schöpfer arise from the dynamic nature of fracture-set Consistency of the natural b value is a phe-
et al., 2011), layer mechanical properties (Lau- evolution. Fracture clusters are predicted by nomenon that could provide evidence of an
bach et al., 2009; Zahm et al., 2010; Ellis et al., linear-elastic fracture mechanics under some underlying process. Higher variation in b is

Geological Society of America Bulletin, September/October 2014 1357


Hooker et al.

MWX 1554 increase could hypothetically be achieved by


MWX 1503: no microfractures (infinite spacing) the addition of new fractures having a variety
MWX 1414 of sizes, although crack-seal texture suggests
MWX 1383 that even large fractures began as microscopic
fractures. Alternatively, the constant slope and
increasing coefficient could arise from addi-
tion of microfractures and gradual widening of
SHCT (Cozzette) existing fractures. Either case suggests that new
microfractures develop throughout the accumu-
SHCT (Total) lation of fracture strain.
SHCT (Cozzette) Narr, 1996
SHCT (Paludal zone)
SUMMARY

10 100 1000 10000 100000 Natural fractures in sandstones from a range


Fracture spacing, mm of geographic and geologic settings follow
power-law size distributions having a consis-
Figure 18. Comparison of this study’s spacing estimates to those tent exponent. Although the reasons for this
of Narr (1996). Spacing estimates are for aperture size 0.5 mm or pattern are unclear, the effects of synkinematic
greater, corresponding to Narr’s (1996) general estimated fracture cement precipitation may play an important role
aperture. Circles are Narr’s (1996) spacing estimates based on visi- because fractures with contrasting size-distri-
ble fractures in the MWX-1 vertical core. Horizontal dashed lines bution statistics also feature contrasting cement
are Narr’s (1996) calculations of probabilities of intersecting visible textures. Namely, power law–distributed frac-
fractures within the vertical core, given the SHCT-1 core macro- tures typically feature crack-seal texture, sig-
fracture spacing. Vertical lines and triangles are SHCT-1 spacing nifying synkinematic cement precipitation; in
estimates based on microfractures (same as Fig. 16). Squares are contrast, fracture sets that have a characteristic
microfracture spacing estimates from the MWX-1 vertical core size do not have significant evidence of synkine-
(Fig. 17). matic cement.
The observed power-law exponent that
characterizes so many natural fracture sets is
present among scanlines with lower numbers commonly have characteristic size distributions. shallow (<1). This shallow exponent has two
of fractures (Fig. 7); some variation in b can Early in the evolution of a fracture set, when important consequences for the application of
therefore be attributed to poor sampling. Further only a small amount of fracture strain has accu- fracture-size scaling to subsurface fracture-
variation in b might be systematically related mulated, most fractures probably result from spacing estimations. First, because the typical b
to some geological conditions. For example, only a single opening increment; very few frac- value is <1, microfractures are not excessively
greater b values have been reported in previous tures are more than five increments wide. There- abundant in natural fracture sets. However, and
studies of metamorphic and hydrothermal sys- fore, the size distribution of the overall fracture second, because b varies little, assuming a fixed
tems (Table 4). set resembles that of the individual opening b of 0.8 can substantially increase the precision
The relatively good fit of power-law versus increments. As fracture strain increases, indi- of subsurface fracture-spacing estimates. Such
characteristic size-distribution equations sug- vidual fractures can become much larger than a method represents a complement or alterna-
gests that as a general tendency, but not as a the characteristic opening-increment size, and tive to existing methods and may be preferable
strict rule, natural fractures that formed in the the power-law distribution can emerge. This when only small amounts of core are retrieved
subsurface follow power-law size distributions. process likely requires some positive feedback or otherwise available.
Previous simulations of fracture growth (Cowie loop between fracture size and fracture-growth
et al., 1993; Cladouhos and Marrett, 1996; Davy rate; otherwise, fracture widening would be APPENDIX
et al., 2010; Hooker et al., 2012) show that equally shared among fractures, and the size East Texas Basin—Cretaceous
power-law fracture-size distributions may arise distribution would remain characteristic. Travis Peak Formation
by interaction among growing fractures, par- The observed range in b among natural power
ticularly when such interaction provides posi- law–distributed fractures is narrow, despite the Fractures are widespread in vertical core from the
Lower Cretaceous Travis Peak Formation, a fluvial-
tive feedback mechanisms that promote further range of tectonic settings and fracture strains deltaic accumulation (Dutton et al., 1991) on the
growth of already-large fractures or inhibit the sampled. This constant b value suggests that western flank of the Sabine Arch, East Texas, USA
growth of small fractures. In contrast, popu- as fracture strain accumulates, the power law (Laubach and Jackson, 1990). Fracture abundance
lations of characteristic-size fractures might maintains its slope as its coefficient increases. reported as cumulative fracture height per core or
signify external controls on fracture size, such This hypothesis is supported by the positive cor- rock-type interval shows that channel sandstones are
highly fractured relative to floodplain and lacustrine
as sedimentary layer thickness (Odling et al., relation between a and fracture strain and by the sediments; layer thickness correlates poorly with frac-
1999; Gillespie et al., 2001). poor correlation between b and strain (Fig. 6). ture abundance in the core (Laubach, 1988, 1989).
Power law–distributed fracture sets appear For fracture strain to increase, fractures must Fluid inclusions in synkinematic quartz cement have
to represent coalescence of characteristic-size become larger or more numerous, or both. A homogenization temperatures (a proxy for precipita-
tion temperature) between 134 and 151 °C (Becker
opening increments, systematically localized power-law size distribution of consistent slope et al., 2010), corresponding to fracture-propagation
into fractures of larger sizes (Fig. 9), partly and increasing coefficient requires that fractures initiation near maximum burial depth and persisting
explaining why lower-strain data sets more of all sizes present increase in number. This for ~48 m.y. (Becker et al., 2010).

1358 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

Piceance Basin—Cretaceous Mesaverde Group scopic scanlines (Table 3) to be sampled perpendicular but does not generally affect fracture orientation. The
to fractures and within single fractured layers. Bed- sandstones’ mature mineralogy, crossbedding, and
The Piceance Basin (northwest Colorado, USA) ding planes within the folds of the core are commonly local Skolithos burrows indicate deposition in a high–
comprises sediments consisting of marginal-marine smooth, with faint striae, particularly in horizons rich wave-energy, marginal-marine setting (Turner, 1970).
and nonmarine siliciclastics deposited in or near the in organic matter. Fractures in core A show a tendency Conformably overlying the Mesón Group in the study
Cretaceous seaway that covered much of present-day to develop perpendicular to bedding, but exceptions area is the Ordovician Santa Victoria Group. Absence
North America (Quigley, 1965; Johnson and Keighin, are abundant. Fractures commonly show multiple ori- of a major unconformity between the Cambrian and
1981). The sequence features minor transgressions entations in the same interval; crosscutting and abut- Ordovician sequences suggests that these rocks were
but is broadly regressive, from marine to paludal up- ting relationships indicate simultaneous propagation not strongly affected by Ordovician orogenesis (Mon
section. Samples studied for fracture analysis derive of fractures with different orientations. Downsection and Hongn, 1991; Mon and Salfity, 1995). Regional
from the Mesaverde Group, mostly from the Williams of the folds, macrofractures are less abundant. Core B rifting occurred during the Cretaceous, along with
Fork and the Iles Formations. The sequence’s burial was drilled through horizontal beds, away from folds. deposition of the Pirgua Formation (Kley et al., 1997,
history, constrained by Nuccio and Condon (1996) No macrofractures were observed in core B. 2005; Deeken et al., 2006). Furthermore, Cretaceous
and Nuccio and Roberts (2003) using vitrinite reflec- deformation (horizontal extension) does not appear
tance and heat-flow modeling, indicates that maximum Scottish Highlands—Cambrian to have left an outcrop-scale structural imprint on the
burial was attained near 35 Ma and uplift began near Eriboll Formation study area. Bedding in the study area strikes NNE and
10 Ma. At maximum burial the Cozzette Sandstone, dips moderately (25°–30°) to the west, consistent with
within the Iles Formation, reached a temperature of The Cambrian Eriboll Formation of the Ardvreck regional Andean folding, which was likely most active
~170 °C. Analysis of fluid-inclusion assemblages Group crops out west of and within the Moine thrust from the Eocene to the Miocene in the Eastern Cor-
within synkinematic fracture cements indicates that belt (MTB) in northwest Scotland. The Eriboll Forma- dillera of Argentina, but continues today (Mon, 1987;
fractures were opening throughout this extended tion includes the 75- to 125-m-thick Basal Quartzite Marrett et al., 1994; Marrett and Strecker, 2000; Hor-
period of maximum burial (Fall et al., 2012). Samples Member and the overlying 75- to 100-m-thick Pipe ton, 2005; Carrera and Muñoz, 2008; DeCelles et al.,
were collected from the north Piceance Basin and the Rock Member. The current study focuses on fractures 2011; Siks and Horton, 2011).
south Piceance Basin (Table 3), the dividing line be- from the latter. This sedimentary sequence records a
tween which we place at 39.7°N latitude. Fractures in broad marine transgression (Park et al., 2002). Eriboll ACKNOWLEDGMENTS
all three slant cores studied strike WNW to E-W. sandstones are indurated, quartz-cemented quartz
arenites. This study was funded partly by grant DE-FG02-
Mexican Sierra Madre Oriental— Cambro-Ordovician rocks dip 10°–15°SE beneath 03ER15430 from Chemical Sciences, Geosciences
Triassic–Jurassic Huizachal Group the adjacent thrust belt and are unconformable on and Biosciences Division, Office of Basic Energy Sci-
Archaean Lewisian gneisses and Proterozoic (Torri- ences, Office of Science, U.S. Department of Energy;
Samples studied from northeast Mexico derive donian) sandstone. The ESE-dipping MTB marks Si- the GDL Foundation (fieldwork); the Fracture Re-
from the Triassic–Jurassic Huizachal Group, a fluvial lurian and Early Devonian WNW-directed shortening, search and Application Consortium; and the Geology
succession deposited at the margin of Pangaea (Mixon juxtaposing Neoproterozoic sedimentary rocks to the Foundation of the Jackson School of Geosciences,
et al., 1959; Michalzik, 1991). Barboza-Gudiño et al. east over the Cambro-Ordovician sequence to the west The University of Texas at Austin (UT). Data in part
(2010) established the El Alamar Formation at the (Strachan et al., 2002, 2010; Cawood et al., 2004). The of Figure 17 and Table 3 were collected as part of the
base of the Huizachal Group, distinct from the overly- Eriboll Formation is cut by the lowest thrust faults UT ExxonMobil Unconventionals project. We thank
ing Jurassic redbed succession. These outcrops were of the MTB, which are associated primarily with J. Gale, P. Eichhubl, S. Fomal, B. Horton, R. Ketcham,
previously interpreted to be part of the La Boca For- brittle deformation and repeated imbricate sequences J. Olson, O. Ortega, and J. Sharp for helpful discus-
mation (Laubach and Ward, 2006). The Huizachal (Holdsworth et al., 2007). The Eriboll Formation is sion. We thank Lana Dieterich for editing an early
Group is overlain by Jurassic evaporites and Cre- also cut by later faults whose motion accommodated version of the manuscript. This paper benefitted from
taceous limestones, dolostones, and shales, which extension, tilting, and wrenching in the Paleozoic reviews by H. Fossen, A. Gudmundsson, E. Beutel,
compose the passive-margin paleodepositional envi- and Mesozoic (Peach et al., 1907; Laubach and Mar- C. Koeberl, and an anonymous reviewer.
ronments of the Gulf of Mexico (Goldhammer and shak, 1987; Roberts and Holdsworth, 1999; Wilson
Johnson, 2001; Ocampo-Diaz et al., 2008). The sand- et al., 2010; Ellis et al., 2012). Poorly cemented and REFERENCES CITED
stones studied, where exposed near Galeana, Nuevo therefore likely post-MTB faults are locally present Aceñolaza, F.G., Buatois, L.A., Mángano, M.G., Esteban,
Leon, consist of feldspathic litharenites; grain size is in outcrops that we studied; maximum displacements S.B., Tortello, M.F., and Aceñolaza, G.F., 1999, Cám-
medium in the lower unit (El Alamar Formation) and are centimeter scale. Samples and measurements were brico y Ordovícico del noroeste Argentino: Instituto de
coarse in the upper unit (La Boca Formation). taken at least 10 m away from such faults. Several sets Geología y Recursos Minerales Anales, v. 29: Geología
The overlying evaporites (Minas Viejas Formation) of opening-mode fractures are present, having consis- Argentina, p. 169–187.
represent a detachment layer, separating kilometer- tent crosscutting relationships and reflecting distinct Anders, M.H., and Wiltschko, D.V., 1994, Microfracturing,
scale, Laramide-age folds of the Sierra Madre Orien- fracturing episodes (Laubach and Diaz-Tushman, paleostress and the growth of faults: Journal of Struc-
tal above from complexly deformed, but more gently 2009). This study includes data from fractures strik- tural Geology, v. 16, p. 795–815, doi:10.1016/0191
-8141(94)90146-5.
folded, strata below (Marrett and Aranda-García, 2001; ing 000N (sample 12), 110N (sample 13), and 030N
André-Mayer, A.-S., and Sausse, J., 2007, Thickness and
Zhou et al., 2006; Latta and Anastasio, 2007). (sample 14); therefore, fracture sets included here spatial distribution of veins in a porphyry copper de-
Four fracture sets (A through D of Laubach and likely formed at different times. posit, Rosia Poieni, Romania: Journal of Structural
Ward, 2006) cut the El Alamar Formation sandstone at Geology, v. 29, p. 1695–1708, doi:10.1016/j.jsg.2007
the study area. This study includes data from fractures Central Andes—Cambrian Mesón Group .06.010.
striking NW-SE (set B) and E-W (set C), which are Augustsson, C., and Reker, A., 2012, Cathodoluminescence
the two most abundant fracture sets. Crosscutting re- Fractures were measured in the Mesón Group at spectra of quartz as provenance indicators revisited:
lationships indicate that set B predates set C. Fractures Huasamayo Canyon, ~5 km southeast of Tilcara, Journal of Sedimentary Research, v. 82, p. 559–570,
doi:10.2110/jsr.2012.51.
in the overlying La Boca Sandstone strike E-W and Argentina, and at Perchel Canyon, 10 km to the north.
Baecher, G.B., and Lanney, N.A., 1978, Trace length biases
NE-SW; the relative timing is unclear. All fractures The lowest rocks exposed there are of the Puncoviscana in joint surveys, in Proceedings, 19th U.S. Symposium
dip steeply, near vertical. Formation (latest Precambrian), a mostly turbiditic se- on Rock Mechanics: Reno, University of Nevada,
quence that underwent low-grade metamorphism and p. 56–65.
Basin X—Formation X exhumation before deposition of overlying sediments Barboza-Gudiño, J.R., Zavala-Monsiváis, A., Venegas-
(Turner, 1960; Aceñolaza et al., 1999). Overlying the Rodríguez, G., and Barajas-Nigoche, L.D., 2010, Late
Fractures were sampled from core derived from Precambrian rocks are hundreds of meters of pink and Triassic stratigraphy and facies from northeastern
folded and unfolded beds at a confidential site, more red, well-sorted, medium- to fine-grained sandstones Mexico: Tectonic setting and provenance: Geosphere,
v. 6, p. 621–640, doi:10.1130/GES00545.1.
than 1000 km from the nearest sample site within the of the Cambrian Mesón Group (Turner, 1970; Kumpa
Barton, C.A., and Zoback, M.D., 1992, Self-similar distri-
data set. The samples are dominantly sublitharenites. and Sanchez, 1988; Such et al., 2007). The sandstones bution of macroscopic fractures at depth in crystalline
Two vertical cores were studied. Core A was drilled are quartzarenites and are thoroughly indurated with rock in the Cajon Pass scientific drill hole: Journal of
through a sequence of tight folds that overlies nearly quartz cement. Beds are typically several centimeters Geophysical Research, v. 97, p. 5181–5200, doi:10
horizontal beds. Folded layers allowed long micro- to a few decimeters thick; crossbedding is common .1029/91JB01674.

Geological Society of America Bulletin, September/October 2014 1359


Hooker et al.

Becker, S.P., Eichhubl, P., Laubach, S.E., Reed, R.M., Cretaceous extension to middle Miocene shortening: southwest Iceland: Tectonophysics, v. 139, p. 295–308,
Lander, R.H., and Bodnar, R.J., 2010, A 48 m.y. his- Tectonics, v. 25, TC6003, doi:10.1029/2005TC001894. doi:10.1016/0040-1951(87)90103-X.
tory of fracture opening, temperature, and fluid pres- Deschamps, A., Tivey, M., Embley, R.W., and Chadwick, Gudmundsson, A., 1992, Formation and growth of normal
sure: Cretaceous Travis Peak Formation, East Texas W.W., 2007, Quantitative study of the deformation faults at the divergent plate boundary in Iceland: Terra
basin: Geological Society of America Bulletin, v. 122, at Southern Explorer Ridge using high-resolution Nova, v. 4, p. 464–471, doi:10.1111/j.1365-3121.1992
p. 1081–1093, doi:10.1130/B30067.1. bathymetric data: Earth and Planetary Science Letters, .tb00582.x.
Bernet, M., and Bassett, K., 2005, Provenance analysis v. 259, p. 1–17, doi:10.1016/j.epsl.2007.04.007. Gudmundsson, A., and Mohajeri, N., 2013, Relations be-
by single-quartz-grain SEM-CL/optical microscopy: Dutton, S.P., Laubach, S.E., Tye, R.S., Baumgardner, R.W., tween the scaling exponents, entropies, and energies of
Journal of Sedimentary Research, v. 75, p. 492–500, Jr., and Herrington, K.L., 1991, Geologic characteriza- fracture networks: Bulletin de la Société Géologique
doi:10.2110/jsr.2005.038. tion of low-permeability gas reservoirs, Travis Peak de France, v. 184, p. 377–387.
Bernet, M., Kapoutsos, D., and Bassett, K., 2007, Diagene- Formation, East Texas: The University of Texas at Aus- Guerriero, V., Iannace, A., Mazzoli, S., Parente, M., Vitale,
sis and provenance of Silurian quartz arenites in south- tin Bureau of Economic Geology Report of Investiga- S., and Giorgioni, M., 2010, Quantifying uncertainties
eastern New York State: Sedimentary Geology, v. 201, tions 204, 89 p. in multi-scale studies of fractured reservoir analogues:
p. 43–55, doi:10.1016/j.sedgeo.2007.04.006. Eichhubl, P., Taylor, W.L., Pollard, D.D., and Aydin, A., Implemented statistical analysis of scan line data from
Bohnenstiehl, D.R., and Kleinrock, M.C., 1999, Faulting 2004, Paleo-fluid flow and deformation in the Aztec carbonate rocks: Journal of Structural Geology, v. 32,
and fault scaling on the median valley floor of the trans- Sandstone at the Valley of Fire, Nevada—Evidence for p. 1271–1278, doi:10.1016/j.jsg.2009.04.016.
Atlantic geotraverse (TAG) segment, ~26°N on the the coupling of hydrogeologic, diagenetic, and tectonic Guerriero, V., Vitale, S., Ciarcia, S., and Mazzoli, S., 2011,
Mid-Atlantic Ridge: Journal of Geophysical Research, processes: Geological Society of America Bulletin, Improved statistical multi-scale analysis of fractured
v. 104, p. 29,351–29,364, doi:10.1029/1999JB900256. v. 116, p. 1120–1136, doi:10.1130/B25446.1. reservoir analogues: Tectonophysics, v. 504, p. 14–24,
Bonnet, E., Bour, O., Odling, N.E., Davy, P., Main, I., Eichhubl, P., Davatzes, N.C., and Becker, S.P., 2009, Struc- doi:10.1016/j.tecto.2011.01.003.
Cowie, P., and Berkowitz, B., 2001, Scaling of fracture tural and diagenetic control of fluid migration and ce- Guerriero, V., Mazzoli, S., Iannace, A., Vitale, S., Carravetta,
systems in geological media: Reviews of Geophysics, mentation along the Moab fault, Utah: AAPG Bulletin, A., and Strauss, C., 2013, A permeability model for
v. 39, p. 347–383, doi:10.1029/1999RG000074. v. 93, p. 653–681, doi:10.1306/02180908080. naturally fractured carbonate reservoirs: Marine and
Burnett, W., and Fomel, S., 2011, Azimuthally anisotropic Ellis, M.A., Laubach, S.E., Eichhubl, P., Olson, J.E., and Petroleum Geology, v. 40, p. 115–134, doi:10.1016/j
3D velocity continuation: International Journal of Geo- Hargrove, P., 2012, Fracture development and diagen- .marpetgeo.2012.11.002.
physics, doi:10.1155/2011/484653. esis of Torridon Group Applecross Formation, near An Hancock, P.L., 1985, Brittle microtectonics: Principles and
Caputo, R., and Hancock, P.L., 1998, Crack-jump mecha- Teallach, NW Scotland: Millennia of brittle deforma- practice: Journal of Structural Geology, v. 7, p. 437–
nism and its implications for stress cyclicity during ex- tion resilience?: Journal of the Geological Society, 457, doi:10.1016/0191-8141(85)90048-3.
tension fracturing: Geodynamics, v. 27, p. 45–60, doi: v. 169, p. 297–310, doi:10.1144/0016-76492011-086. Haynes, F.M., 1984, Vein densities in drill core, Sierrita
10.1016/S0264-3707(97)00029-X. English, J.M., 2012, Thermo-mechanical origin of regional porphyry copper deposit, Pima County, Arizona:
Carrera, N., and Muñoz, J.A., 2008, Thrusting evolution in fracture systems: AAPG Bulletin, v. 96, p. 1597–1625, Economic Geology and the Bulletin of the Society of
the southern Cordillera Oriental (northern Argentine doi:10.1306/01021211018. Economic Geologists, v. 79, p. 755–758, doi:10.2113
Andes): Constraints from growth strata: Tectono- Fagereng, A., 2011, Fractal vein distributions within a fault- /gsecongeo.79.4.755.
physics, v. 459, p. 107–122, doi:10.1016/j.tecto.2007 fracture mesh in an exhumed accretionary mélange, Hesthammer, J., and Fossen, H., 2003, From seismic data
.11.068. Chrystalls Beach Complex, New Zealand: Journal of to core data: An integrated approach to enhance reser-
Cartwright, J., 2011, Diagenetically induced shear failure Structural Geology, v. 33, p. 918–927, doi:10.1016/j voir characterization, in Ameen, M., ed., Fracture and
of fine-grained sediments and the development of .jsg.2011.02.009. In-Situ Stress Characterization of Hydrocarbon Reser-
polygonal fault systems: Marine and Petroleum Geol- Fall, A., Eichhubl, P., Cumella, S.P., Bodnar, R.J., Laubach, voirs: Geological Society [London] Special Publica-
ogy, v. 28, p. 1593–1610, doi:10.1016/j.marpetgeo S.E., and Becker, S.P., 2012, Testing the basin-centered tion 209, p. 39–54.
.2011.06.004. gas accumulation model using fluid inclusion observa- Holdsworth, R.E., Alsop, G.I., and Strachan, R.A., 2007,
Cawood, P.A., Nemchin, A.A., Strachan, R.A., Kinny, P.D., tions: Southern Piceance Basin, Colorado: AAPG Bul- Tectonic stratigraphy and structural continuity of the
and Loewy, S., 2004, Laurentian provenance and an letin, v. 96, p. 2297–2318, doi:10.1306/05171211149. northernmost Moine Thrust Zone and Moine Nappe,
intracratonic tectonic setting for the Moine Super- Folk, R.L., 1981, Petrology of Sedimentary Rocks (2nd ed.): Scottish Caledonides, in Ries, A.C., Butler, R.W.H.,
group, Scotland, constrained by detrital zircons from Austin, Hemphill Publishing Company, 184 p. and Graham, R.D., eds., Deformation of the Conti-
the Loch Eli and Glen Urquhart successions: Journal Gale, J.F.W., Lander, R.H., Reed, R.M., and Laubach, S.E., nental Crust: The Legacy of Mike Coward: Geo-
of the Geological Society, v. 161, p. 861–874, doi:10 2010, Modeling fracture porosity evolution in dolo- logical Society [London] Special Publication 272,
.1144/16-764903-117. stone: Journal of Structural Geology, v. 32, p. 1201– p. 123–144.
Cladouhos, T.T., and Marrett, R., 1996, Are fault growth and 1211, doi:10.1016/j.jsg.2009.04.018. Hooker, J.N., and Laubach, S.E., 2007, The geologic his-
linkage models consistent with power law distributions Gillespie, P.A., Johnston, J.D., Loriga, M.A., McCaffrey, tory of quartz grains, as revealed by color SEM-CL:
of fault lengths?: Journal of Structural Geology, v. 18, K.J.W., Walsh, J.J., and Watterson, J., 1999, Influence Gulf Coast Association of Geological Societies Trans-
p. 281–293, doi:10.1016/S0191-8141(96)80050-2. of layering on vein systematics in line samples, in Mc- actions, v. 57, p. 375–386.
Clark, M.B., Brantley, S.L., and Fisher, D.M., 1995, Power- Caffrey, K.J.W., Longeran, L., and Wilkinson, J.J., eds., Hooker, J.N., and Laubach, S.E., 2010, Using empirical
law vein-thickness distributions and positive feedback in Fractures, Fluid Flow and Mineralization: Geological tends in fracture size-frequency data to constrain sub-
vein growth: Geology, v. 23, p. 975–978, doi:10.1130 Society [London] Special Publication 155, p. 35–56. surface fracture abundance: American Rock Mechan-
/0091-7613(1995)023<0975:PLVTDA>2.3.CO;2. Gillespie, P.A., Walsh, J.J., Watterson, J., Bonson, C.G., and ics Association 44th US Rock Mechanics Symposium
Clauset, A., Shalizi, R., and Newman, M.E.J., 2009, Power- Manzocchi, T., 2001, Scaling relationships of joint and 5th US-Canada Rock Mechanics Symposium, Salt
law distributions in empirical data: SIAM (Society for and vein arrays from The Burren, Co. Clare, Ireland: Lake City, UT, June 27–30, 2010, ARMA 10-325,
Industrial and Applied Mathematics) Review, v. 51, Journal of Structural Geology, v. 23, p. 183–201, doi: 11 p.
p. 661–703, doi:10.1137/070710111. 10.1016/S0191-8141(00)00090-0. Hooker, J.N., Gale, J.F.W., Gomez, L.A., Laubach, S.E.,
Cowie, P.A., Vanneste, C., and Sornette, D., 1993, Statisti- Goldhammer, R.K., and Johnson, C.A., 2001, Middle Juras- Marrett, R., and Reed, R.M., 2009, Aperture-size scal-
cal physics model for the spatiotemporal evolution sic–Upper Cretaceous paleogeographic evolution and ing variations in a low-strain opening-mode fracture
of faults: Journal of Geophysical Research, v. 98, sequence-stratigraphic framework of the northwest set, Cozzette Sandstone, Colorado: Journal of Struc-
p. 21,809–21,821, doi:10.1029/93JB02223. Gulf of Mexico rim, in Bartolini, C., Buffler, R.T., and tural Geology, v. 31, p. 707–718, doi:10.1016/j.jsg
Davatzes, N.C., and Hickman, S., 2005, Comparison of Cantú-Chapa, A., eds., The Western Gulf of Mexico .2009.04.001.
acoustic and electrical image logs from the Coso Geo- Basin: Tectonics, Sedimentary Basins, and Petroleum Hooker, J.N., Laubach, S.E., Gomez, L.A., Marrett, R.,
thermal Field, CA, in Proceedings, Thirtieth Workshop Systems: AAPG Memoir 75, p. 45–81. Eichhubl, P., Diaz-Tushman, K., and Pinzon, E., 2011,
on Geothermal Reservoir Engineering, Stanford Uni- Gomez, L.A., and Laubach, S.E., 2006, Rapid digital quan- Fracture size, frequency, and strain in the Cambrian
versity: Stanford, California, SGP-TR-176, 11 p. tification of microfracture populations: Journal of Eriboll Formation sandstones, NW Scotland: Scottish
Davy, P., Le Groc, R., Darcel, C., Bour, O., de Dreuzy, J.R., Structural Geology, v. 28, p. 408–420, doi:10.1016/j Journal of Geology, v. 47, p. 45–56, doi:10.1144/0036
and Munier, R., 2010, A likely universal model of .jsg.2005.12.006. -9276/01-420.
fracture scaling and its consequence for crustal hydro- Götze, J., Plötze, M., and Habermann, D., 2001, Origin, Hooker, J.N., Gomez, L.A., Laubach, S.E., Gale, J.F.W.,
mechanics: Journal of Geophysical Research, v. 115, spectral characteristics, and practical applications of the and Marrett, R., 2012, Effects of diagenesis (cement
B10411, doi:10.1029/2009JB007043. cathodoluminescence of quartz: Mineralogy and Petrol- precipitation) during fracture opening on fracture
DeCelles, P.G., Carrapa, B., Horton, B.K., and Gehrels, ogy, v. 71, p. 225–250, doi:10.1007/s007100170040. aperture-size scaling in carbonate rocks, in Garland, J.,
G.E., 2011, Cenozoic foreland basin systems in the Gross, M.R., and Engelder, T., 1995, Strain accommodated Neilson, J.E., Laubach, S.E., and Whidden, K.J., eds.,
central Andes of northwestern Argentina: Implications by brittle failure in adjacent units of the Monterey For- Advances in Carbonate Exploration and Reservoir
for Andean geodynamics and modes of deformation: mation, U.S.A.: Scale effects and evidence for uniform Analysis: Geological Society [London] Special Publi-
Tectonics, v. 30, TC6013, doi:10.1029/2011TC002948. displacement boundary conditions: Journal of Struc- cation 370, p. 187–206.
Deeken, A., Sobel, E.R., Coutand, I., Haschke, M., Riller, U., tural Geology, v. 17, p. 1303–1318, doi:10.1016/0191 Hooker, J.N., Laubach, S.E., and Marrett, R., 2013, Frac-
and Strecker, M.R., 2006, Development of the south- -8141(95)00011-2. ture-aperture size—Frequency, spatial distribution,
ern Eastern Cordillera, NW Argentina, constrained by Gudmundsson, A., 1987, Geometry, formation, and develop- and growth processes in strata-bounded and non-
apatite fission track thermochronology: From early ment of tectonic fractures on the Reykjanes Peninsula, strata-bounded fractures, Cambrian Mesón Group,

1360 Geological Society of America Bulletin, September/October 2014


A universal power-law scaling exponent for fracture apertures in sandstones

NW Argentina: Journal of Structural Geology, v. 54, U.S.A.: Journal of Structural Geology, v. 11, p. 603– eds., Fractured Reservoirs: Characterization and Mod-
p. 54–71, doi:10.1016/j.jsg.2013.06.011. 611, doi:10.1016/0191-8141(89)90091-6. eling Guidebook: Denver, Rocky Mountain Associa-
Horton, B.K., 2005, Revised deformation history of the Laubach, S.E., 1997, A method to detect natural fracture tion of Geologists, p. 217–226.
central Andes: Inferences from Cenozoic foredeep and strike in sandstones: AAPG Bulletin, v. 81, p. 604–623. Marrett, R., and Allmendinger, R.W., 1992, Amount of ex-
intermontane basins of the Eastern Cordillera, Bolivia: Laubach, S.E., 2003, Practical approaches to identify- tension on “small” faults: An example from the Viking
Tectonics, v. 24, TC3011, doi:10.1029/2003TC001619. ing sealed and open fractures: AAPG Bulletin, v. 87, graben: Geology, v. 20, p. 47–50, doi:10.1130/0091
Iñigo, J.F., Laubach, S.E., and Hooker, J.N., 2012, Fracture p. 561–579, doi:10.1306/11060201106. -7613(1992)020<0047:AOEOSF>2.3.CO;2.
abundance and patterns in the Subandean fold and Laubach, S.E., and Diaz-Tushman, K., 2009, Laurentian Marrett, R., and Aranda-García, M., 2001, Regional struc-
thrust belt, Devonian Huamampampa Formation petro- paleostress trajectories and ephemeral fracture permea- ture of the Sierra Madre Oriental fold-thrust belt,
leum reservoirs and outcrops, Argentina and Bolivia: bility, Cambrian Eriboll Formation sandstones west Mexico, in Marrett, R., ed., Genesis and Controls of
Marine and Petroleum Geology, v. 35, p. 201–218, doi: of the Moine thrust zone, northwest Scotland: Journal Reservoir-Scale Carbonate Deformation, Monterrey
10.1016/j.marpetgeo.2012.01.010. of the Geological Society, v. 166, p. 349–362, doi:10 Salient, Mexico: The University of Texas at Austin Bu-
Johnson, R.C., and Keighin, C.W., 1981, Cretaceous and .1144/0016-76492008-061. reau of Economic Geology Guidebook 28, p. 31–55.
Tertiary history and resources of the Piceance Creek Laubach, S.E., and Jackson, M.L.W., 1990, Origin of arches Marrett, R., and Strecker, M.R., 2000, Response of intra-
Basin, western Colorado, in Epis, J.C., and Callen- in the northwestern Gulf of Mexico basin: Geology, continental deformation in the central Andes to late
der, J.F., eds., Western Slope Colorado: New Mexico v. 18, p. 595–598, doi:10.1130/0091-7613(1990)018 Cenozoic reorganization of South American Plate
Geological Society 32nd Field Conference Guidebook, <0595:OOAITN>2.3.CO;2. motions: Tectonics, v. 19, p. 452–467, doi:10.1029
p. 199–210. Laubach, S.E., and Marshak, S., 1987, Fault patterns gen- /1999TC001102.
Jolly, R.J.H., and Lonergan, L., 2002, Mechanisms and con- erated during extensional deformation of crystalline Marrett, R., Allmendinger, E.W., Alonso, R.N., and Drake,
trols on the formation of sand intrusions: Journal of the basement, NW Scotland, in Coward, M.P., Dewey, J.F., R.E., 1994, Late Cenozoic tectonic evolution of the
Geological Society, v. 159, p. 605–617, doi:10.1144 and Hancock, P.L., eds., Continental Extensional Tec- Puna Plateau and adjacent foreland, northwestern
/0016-764902-025. tonics: Geological Society [London] Special Publica- Argentine Andes: Journal of South American Earth
Kagan, Y.Y., 1997, Seismic moment-frequency relation for tion 28, p. 495–499. Sciences, v. 7, p. 179–207, doi:10.1016/0895-9811
shallow earthquakes: Regional comparison: Journal of Laubach, S.E., and Ward, M.E., 2006, Diagenesis in porosity (94)90007-8.
Geophysical Research, v. 102, p. 2835–2852, doi:10 evolution of opening-mode fractures, Middle Triassic Marrett, R., Ortega, O.J., and Kelsey, C.M., 1999, Extent
.1029/96JB03386. to Lower Jurassic La Boca Formation, NE Mexico: of power-law scaling for natural fractures in rock:
Kley, J., Müller, J., Tawacholi, S., Jacobshagen, V., and Tectonophysics, v. 419, p. 75–97, doi:10.1016/j.tecto Geology, v. 27, p. 799–802, doi:10.1130/0091-7613
Manutsoglu, E., 1997, Pre-Andean and Andean-age .2006.03.020. (1999)027<0799:EOPLSF>2.3.CO;2.
deformation in the Eastern Cordillera of southern Bo- Laubach, S.E., Hentz, T.F., Johns, M.K., Baek, H., and Clift, Marrett, R., Laubach, S.E., and Olson, J.E., 2007, Aniso-
livia: Journal of South American Earth Sciences, v. 10, S.J., 1995, Using diagenesis information to augment tropy and beyond: Geologic perspectives on geophysi-
p. 1–19, doi:10.1016/S0895-9811(97)00001-1. fracture analysis: Austin, The University of Texas at cal prospecting for natural fractures: The Leading
Kley, J., Rossello, E.A., Monaldi, C.R., and Habighorst, B., Austin Bureau of Economic Geology, report for the Gas Edge, v. 26, p. 1106–1111, doi:10.1190/1.2780778.
2005, Seismic and field evidence for selective inver- Research Institute, contract no. 5082-211-0708, 189 p. Mauldon, M., Dunne, W.M., and Rohrbaugh, M.B., Jr.,
sion of Cretaceous normal faults, Salta rift, northwest Laubach, S.E., Lander, R.H., Bonnell, L.M., Olson, J.E., and 2001, Circular scanlines and circular windows: New
Argentina: Tectonophysics, v. 399, p. 155–172, doi:10 Reed, R.M., 2004a, Opening histories of fractures in tools for characterizing the geometry of fracture traces:
.1016/j.tecto.2004.12.020. sandstones, in Cosgrove, J.W., and Engelder, T., eds., Journal of Structural Geology, v. 23, p. 247–258, doi:
Kolyukhin, D., and Torabi, A., 2013, Power-law testing for The Initiation, Propagation, and Arrest of Joints and 10.1016/S0191-8141(00)00094-8.
fault attributes distributions: Pure and Applied Geo- Other Fractures: Geological Society [London] Special McCaffrey, K.J.W., and Johnston, J.D., 1996, Fractal analy-
physics, v. 170, p. 2173–2183, doi:10.1007/s00024 Publication 231, p. 1–9. sis of a mineralised vein deposit: Curraghinalt gold
-013-0644-3. Laubach, S.E., Reed, R.M., Olson, J.E., Lander, R.H., and deposit, County Tyrone: Mineralium Deposita, v. 31,
Kumpa, M., and Sanchez, M.C., 1988, Geology and sedi- Bonnell, L.M., 2004b, Coevolution of crack-seal tex- p. 52–58, doi:10.1007/BF00225395.
mentology of the Cambrian Grupo Mesón (NW Ar- ture and fracture porosity in sedimentary rocks: Cath- Michalzik, D., 1991, Facies sequence of Triassic-Jurassic
gentina), in Bahlburg, H., Breitkreuz, C., and Giese, P., odoluminescence observations of regional fractures: red beds in the Sierra Madre Oriental (NE Mexico) and
eds., The South Central Andes: Contributions to Struc- Journal of Structural Geology, v. 26, p. 967–982, doi: its relation to the early opening of the Gulf of Mexico:
ture and Evolution of an Active Continental Margin: 10.1016/j.jsg.2003.08.019. Sedimentary Geology, v. 71, p. 243–259, doi:10.1016
Lecture Notes in Earth Sciences 17: Berlin, Springer- Laubach, S.E., Olson, J.E., and Gross, M.R., 2009, Mechan- /0037-0738(91)90105-M.
Verlag, 261 p. ical and fracture stratigraphy: AAPG Bulletin, v. 93, Milliken, K.L., and Laubach, S.E., 2000, Brittle deforma-
Lacazette, A., and Engelder, T., 1992, Fluid-driven cyclic p. 1413–1426, doi:10.1306/07270909094. tion in sandstone diagenesis as revealed by scanned
propagation of a joint in the Ithaca Sandstone, Appa- Laubach, S.E., Eichhubl, P., Hilgers, C., and Lander, R.H., cathodoluminescence imaging with application to
lachian Basin, New York, in Evans, B., and Wong, T.-F., 2010, Structural diagenesis: Journal of Structural characterization of fractured reservoirs, in Pagel, M.,
eds., Fault Mechanics and Transport Properties of Geology, v. 32, p. 1866–1872, doi:10.1016/j.jsg.2010 ed., Cathodoluminescence and Related Techniques
Rocks: San Diego, Academic Press, p. 297–323. .10.001. in Geosciences and Materials: New York, Springer-
Ladeira, F.L., and Price, N.J., 1981, Relationship between Le Garzic, E., de l’Hamaide, T., Diraison, M., Géraud, Y., Verlag, p. 225–243.
fracture spacing and bed thickness: Journal of Struc- Sausse, J., de Urreiztieta, M., Hauville, B., and Cham- Mixon, R.B., Murray, G.E., and Diaz-Gonzales, T.E., 1959,
tural Geology, v. 3, p. 179–183, doi:10.1016/0191 panhet, J.-M., 2011, Scaling and geometric proper- Age and correlation of Huizachal Group (Mesozoic),
-8141(81)90013-4. ties of extensional fracture systems in the Proterozoic State of Tamaulipas, Mexico: AAPG Bulletin, v. 43,
Lander, R.H., Larese, R.E., and Bonnell, L.M., 2008, Toward basement of Yemen: Tectonic interpretation and fluid p. 757–771.
more accurate quartz cement models—The importance flow implications: Journal of Structural Geology, v. 33, Mon, R., 1987, Structural geology of two geothermal areas
of euhedral vs. non-euhedral growth rates: AAPG Bul- p. 519–536, doi:10.1016/j.jsg.2011.01.012. in the Andes: Copahue and Tuzgle (Argentina): Bul-
letin, v. 92, p. 1537–1563, doi:10.1306/07160808037. Lorenz, J.C., Teufel, L.W., and Warpinski, N.R., 1991, letin of Engineering Geology and the Environment,
Laongsakul, P., and Dürrast, H., 2011, Characterization of Regional fractures I: A mechanism for the formation v. 35, p. 79–85.
reservoir fractures using conventional geophysical log- of regional fractures at depth in flat-lying reservoirs: Mon, R., and Hongn, F., 1991, The structure of the Precam-
ging: Songklanakarin Journal of Science and Technol- AAPG Bulletin, v. 75, p. 1714–1737. brian and lower Paleozoic basement of the Central
ogy, v. 33, p. 237–246. Loriga, M.A., 1999, Scaling systematics of vein size: An Andes between 22° and 32°S lat.: Geologische Rund-
Larsen, B., Grunnaleite, I., and Gudmundsson, A., 2010, example from the Guanajuato mining district (central schau, v. 80, p. 745–758.
How fracture systems affect permeability development Mexico), in McCaffrey, K.J.W., Longeran, L., and Mon, R., and Salfity, J.A., 1995, Tectonic evolution of the
in shallow-water carbonate rocks: An example from the Wilkinson, J.J., eds., Fractures, Fluid Flow and Miner- Andes of northern Argentina, in Tankard, A.J., Suarez
Gargano Peninsula, Italy: Journal of Structural Geol- alization: Geological Society [London] Special Publi- Soruco, R., and Welsink, H.J., eds., Petroleum Basins
ogy, v. 32, p. 1212–1230, doi:10.1016/j.jsg.2009.05 cation 155, p. 57–67. of South America: AAPG Memoir 62, p. 269–283.
.009. Mandelbrot, B.B., 1982, The Fractal Geometry of Nature: Myers, R., and Aydin, A., 2004, The evolution of faults
Latta, D.K., and Anastasio, D.J., 2007, Multiple scales of San Francisco, W.H. Freeman, 460 p. formed by shearing across joint zones in sandstone:
mechanical stratification and decollément fold kine- Mandl, G., and Harkness, R.M., 1987, Hydrocarbon mi- Journal of Structural Geology, v. 26, p. 947–966, doi:
matics, Sierra Madre Oriental foreland, northeast gration by hydraulic fracturing, in Jones, M.E., and 10.1016/j.jsg.2003.07.008.
Mexico: Journal of Structural Geology, v. 29, p. 1241– Preston, R.M.F., eds., Deformation of Sediments and Narr, W., 1996, Estimating average fracture spacing in sub-
1255, doi:10.1016/j.jsg.2007.03.012. Sedimentary Rocks: Geological Society [London] Spe- surface rock: AAPG Bulletin, v. 80, p. 1565–1586.
Laubach, S.E., 1988, Subsurface fractures and their relation- cial Publication 29, p. 39–53. Nelson, R.A., 1985, Geologic Analysis of Naturally Frac-
ship to stress history in East Texas Basin sandstone: Marrett, R., 1996, Aggregate properties of fracture popula- tured Reservoirs: Houston, Gulf Publishing, 320 p.
Tectonophysics, v. 156, p. 37–49, doi:10.1016/0040 tions: Journal of Structural Geology, v. 18, p. 169–178, Nuccio, V.F., and Condon, S.M., 1996, Burial and thermal
-1951(88)90281-8. doi:10.1016/S0191-8141(96)80042-3. history in the Paradox basin, Utah and Colorado,
Laubach, S.E., 1989, Paleostress directions from the pre- Marrett, R., 1997, Permeability, porosity, and shear-wave and petroleum potential of the Middle Pennsylvanian
ferred orientation of closed microfractures (fluid- anisotropy from scaling of open fracture populations, Paradox Formation: U.S. Geological Survey Bulletin
inclusion planes) in sandstone, East Texas basin, in Hoak, T.E., Klawitter, A.L., and Blomquist, P.K., 2000-O, 41 p.

Geological Society of America Bulletin, September/October 2014 1361


Hooker et al.

Nuccio, V.F., and Roberts, L.N.R., 2003, Thermal matu- Pollard, D.D., and Aydin, A., 1988, Progress in understand- in Trewin, N.H., ed., The Geology of Scotland: Lon-
rity and oil and gas generation history of petroleum ing jointing over the past century: Geological Society don, The Geological Society, p. 81–148.
systems in the Uinta-Piceance Province, Utah and of America Bulletin, v. 100, p. 1181–1204, doi:10.1130 Strachan, R.A., Holdsworth, R.E., Krabbendam, M., and
Colorado, in USGS Uinta-Piceance Assessment Team, /0016-7606(1988)100<1181:PIUJOT>2.3.CO;2. Alsop, G.I., 2010, The Moine Supergroup of NW
Petroleum Systems and Geologic Assessment of Oil Quigley, M.D., 1965, Geologic history of Piceance Creek– Scotland: Insights into the analysis of polyorogenic
and Gas in the Uinta-Piceance Province, Utah and Eagle basins: Bulletin of the American Association of supracrustal sequences, in Law, R., Butler, R.W.H.,
Colorado: U.S. Geological Survey Digital Data Series Petroleum Geologists, v. 49, p. 1974–1996. Strachan, R.A., and Krabbendam, M., eds., Continen-
DDS-69-B, Chapter 4, 35 p. Ramsay, J.G., 1980, The crack-seal mechanism of rock tal Tectonics and Mountain Building: The Legacy of
Nur, A., 1982, The origin of tensile fracture lineaments: deformation: Nature, v. 284, p. 135–139, doi:10.1038 Peach and Horne: Geological Society [London] Spe-
Journal of Structural Geology, v. 4, p. 31–40, doi:10 /284135a0. cial Publication 335, p. 233–254.
.1016/0191-8141(82)90004-9. Renard, F., Andréani, M., Boullier, A.-M., and Labaume, Such, P., Buatois, L.A., and Mángano, M.G., 2007, Stratig-
Ocampo-Diaz, Y.Z.E., Jenchen, U., and Guerrero-Suastegui, P., 2005, Crack-seal patterns: Records of uncorrelated raphy, depositional environments and ichnology of the
M., 2008, Facies and depositional systems of the stress release variations in crustal rocks, in Gapais, Lower Paleozoic in the Azul Pampa area—Jujuy Prov-
Galeana Sandstone Member (Taraises Formation, D., Brun, J.P., and Cobbold, P.R., eds., Deformation ince: Revista de la Asociación Geológica Argentina,
Lower Cretaceous, northeastern Mexico): Revista Mechanisms, Rheology and Tectonics: From Minerals v. 62, p. 331–344.
Mexicana de Ciencias Geológicas, v. 25, p. 438–464. to the Lithosphere: Geological Society [London] Spe- Turner, J.C.M., 1960, Estratigrafía de la Sierra se Santa Vic-
Odling, N.E., Gillespie, P., Bourgine, B., Castaing, C., cial Publication 243, p. 67–79. toria y adyacencias: Academia Nacional de Ciencias
Chiles, J.-P., Christensen, N.P., Fillion, E., Genter, A., Roberts, A.M., and Holdsworth, R.E., 1999, Linking on- Boletín, v. 41, p. 163–196.
Olsen, C., Thrane, L., Trice, R., Aarseth, E., Walsh, shore and offshore structures: Mesozoic extension in Turner, J.C.M., 1970, The Andes of northwestern Argentina:
J.J., and Watterson, J., 1999, Variations in fracture sys- the Scottish Highlands: Journal of the Geological So- Geologische Rundschau, v. 59, p. 1028–1063, doi:10
tem geometry and their implications for fluid flow in ciety, v. 156, p. 1061–1064, doi:10.1144/gsjgs.156.6 .1007/BF02042283.
fractured hydrocarbon reservoirs: Petroleum Geosci- .1061. Urai, J.L., Williams, P.F., and van Roermund, H.L.M., 1991,
ence, v. 5, p. 373–384, doi:10.1144/petgeo.5.4.373. Rohrbaugh, M.B., Jr., Dunne, W.M., and Mauldon, M., Kinematics of crystal growth in syntectonic fibrous
Olson, J.E., 1993, Joint pattern development: Effects of sub- 2002, Estimating fracture trace intensity, density, and veins: Journal of Structural Geology, v. 13, p. 823–836,
critical crack growth and mechanical crack interaction: mean length using circular scan lines and windows: doi:10.1016/0191-8141(91)90007-6.
Journal of Geophysical Research, v. 98, p. 12,251– AAPG Bulletin, v. 86, p. 2089–2104. Wilson, R.W., Holdsworth, R.E., Wild, L.E., McCaffrey,
12,265, doi:10.1029/93JB00779. Sanderson, D.J., Roberts, S., and Gumiel, P., 1994, A frac- K.J.W., England, R.W., Imber, J., and Strachan, R.A.,
Olson, J.E., 2003, Sublinear scaling of fracture aperture versus tal relationship between vein thickness and gold grade 2010, Basement-influenced rifting and basin develop-
length: An exception or the rule?: Journal of Geophysi- in drill core from La Codosera, Spain: Economic ment: A reappraisal of post-Caledonian faulting pat-
cal Research, v. 108, 2413, doi:10.1029/2001JB000419. Geology and the Bulletin of the Society of Economic terns from the North Coast Transfer Zone, Scotland,
Olson, J.E., Laubach, S.E., and Lander, R.H., 2009, Natural Geologists, v. 89, p. 168–173, doi:10.2113/gsecongeo in Law, R., Butler, R.W.H., Strachan, R.A., and Krab-
fracture characterization in tight gas sandstones: In- .89.1.168. bendam, M., eds., Continental Tectonics and Mountain
tegrating mechanics and diagenesis: AAPG Bulletin, Sasaki, M., 2006, Statistical features of vein systems in the Building: The Legacy of Peach and Horne: Geological
v. 93, p. 1535–1549, doi:10.1306/08110909100. Hishikari epithermal gold deposit, Japan: Resource Society [London] Special Publication 335, p. 795–826.
Ortega, O.J., and Marrett, R., 2000, Prediction of macro- Geology, v. 56, p. 27–36, doi:10.1111/j.1751-3928 Wilson, T.J., and Paulsen, T.S., 1998, CRP-1 fracture arrays:
fracture properties using microfracture information, .2006.tb00265.x. Constraints on the Neogene–Quaternary stress regime
Mesaverde Group sandstones, San Juan basin, New Schöpfer, M.P.J., Arslan, A., Walsh, J.J., and Childs, C., along the Transantarctic Mountains Front, Antarctica:
Mexico: Journal of Structural Geology, v. 22, p. 571– 2011, Reconciliation of contrasting theories for frac- Terra Antarctica, v. 5, p. 327–335.
588, doi:10.1016/S0191-8141(99)00186-8. ture spacing in layered rocks: Journal of Structural Wong, T.-F., Friedrich, J.T., and Gwanmesia, G.D., 1989,
Ortega, O.J., Marrett, R., and Laubach, S.E., 2006, A scale- Geology, v. 33, p. 551–565, doi:10.1016/j.jsg.2011.01 Crack aperture statistics and pore space fractal geom-
independent approach to fracture intensity and average .008. etry of Westerly Granite and Rutland Quartzite:
spacing measurement: AAPG Bulletin, v. 90, p. 193– Seyedolali, A., Krinsley, D.H., Boggs, S., Jr., O’Hara, P.F., Implications for an elastic contact model of rock com-
208, doi:10.1306/08250505059. Dypvik, H., and Goles, G.G., 1997, Provenance inter- pressibility: Journal of Geophysical Research, v. 94,
Ortega, O.J., Gale, J.F.W., and Marrett, R., 2010, Quanti- pretation of quartz by scanning electron microscope– p. 10,267–10,278, doi:10.1029/JB094iB08p10267.
fying diagenetic and stratigraphic controls on fracture cathodoluminescence fabric analysis: Geology, v. 25, Zahm, C.K., Zahm, L.C., and Bellian, J.A., 2010, Integrated
intensity in platform carbonates: An example from the p. 787–790, doi:10.1130/0091-7613(1997)025<0787: fracture prediction using sequence stratigraphy within
Sierra Madre Oriental, northeast Mexico: Journal of PIOQBS>2.3.CO;2. a carbonate fault damage zone, Texas, USA: Journal of
Structural Geology, v. 32, p. 1943–1959. Siks, B.C., and Horton, B.K., 2011, Growth and fragmen- Structural Geology, v. 32, p. 1363–1374, doi:10.1016/j
Park, R.G., Stewart, A.D., and Wright, D.T., 2002, The tation of the Andean foreland basin during eastward .jsg.2009.05.012.
Hebridean terrane, in Trewin, N.H., ed., The Geology advance of fold-thrust deformation, Puna Plateau and Zhou, Y., Murphy, M.A., and Hamade, A., 2006, Structural
of Scotland: London, The Geological Society, p. 45–81. Eastern Cordillera, northern Argentina: Tectonics, development of the Peregrina-Huizachal anticlinorium,
Peach, B.N., Horne, J., Gunn, W., Clough, C.T., Hinxman, v. 30, TC6017, doi:10.1029/2011TC002944. Mexico: Journal of Structural Geology, v. 28, p. 494–
L.W., and Teall, J.J., 1907, The Geological Structure Sornette, D., and Sornette, A., 1999, General theory of the 507, doi:10.1016/j.jsg.2005.11.005.
of the North-West Highlands of Scotland: Geological modified Gutenberg-Richter Law for large seismic mo-
Survey of Great Britain Memoir, 668 p. ment: Bulletin of the Seismological Society of Amer- SCIENCE EDITOR: CHRISTIAN KOEBERL
Philipp, S.L., 2008, Geometry and formation of gypsum ica, v. 89, p. 1121–1130. ASSOCIATE EDITOR: ERIN K. BEUTEL
veins in mudstones at Watchet, Somerset, SW England: Sprunt, E.S., and Nur, A., 1979, Microcracking and healing
Geological Magazine, v. 145, p. 831–844, doi:10.1017 in granites: New evidence from cathodoluminescence: MANUSCRIPT RECEIVED 11 JUNE 2013
/S0016756808005451. Science, v. 205, p. 495–497, doi:10.1126/science.205 REVISED MANUSCRIPT RECEIVED 14 FEBRUARY 2014
Pickering, G., Bull, J.M., and Sanderson, D.J., 1995, Sam- .4405.495. MANUSCRIPT ACCEPTED 30 APRIL 2014
pling power law distributions: Tectonophysics, v. 248, Strachan, R.A., Smith, M., Harris, A.L., and Fettes, D.J.,
p. 1–20, doi:10.1016/0040-1951(95)00030-Q. 2002, The northern Highland and Grampian terranes, Printed in the USA

1362 Geological Society of America Bulletin, September/October 2014

You might also like