Jorn H Kruhl

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Journal of Structural Geology 46 (2013) 2e21

Contents lists available at SciVerse ScienceDirect

Journal of Structural Geology


journal homepage: www.elsevier.com/locate/jsg

Review

Fractal-geometry techniques in the quantification of complex rock structures:


A special view on scaling regimes, inhomogeneity and anisotropy
Jörn H. Kruhl
Tectonics and Material Fabrics Section, Technical University Munich, 80290 Munich, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Fractal-geometry techniques are widely applied to the quantification of complex rock structures.
Received 13 March 2012 Important properties of such structures are (i) different scaling behaviour on different scales, (ii)
Received in revised form inhomogeneity, and (iii) anisotropy. The current paper presents a special view on the quantification of
29 September 2012
these properties by classical and newly developed fractal-geometry methods, discusses advantages and
Accepted 3 October 2012
Available online 26 October 2012
disadvantages of special methods and outlines the correlations between structure quantifications and
rock properties and structure-forming processes, presented in the literature.
Ó 2012 Published by Elsevier Ltd.
Keywords:
Fractal geometry
Complex rock structures
Scaling regimes
Inhomogeneity
Anisotropy

1. Introduction phase versus a second phase. Therefore and in order to keep the topic
to a manageable extent, the present paper does not deal with the
Matter is structured from micro- to mega-scale and these information stored in, e.g. composition, colours or crystallography of
structures appear to us as patterns of different complexity. The structures, but will only address measurement, quantification, and
human race developed in a structured world where it was and still analysis of binary patterns.
is important to sense similar patterns and to rate them as mean- Structures and their binary patterns are always an expression of
ingful or not. Therefore, the human brain is well designed for the processes, by which they are formed or changed. Therefore, the
pattern recognition and for qualitative comparison of patterns. analysis of structures (pattern analysis) provides valuable infor-
However, pattern quantification appears difficult. mation about the structure-forming processes. Structures may be
The present paper deals with structures in rocks. These structures very simple, with Euclidean geometry like the outlines of crystals.
are defined as geometric and, in certain cases, crystallographic But in most cases they are so complex that measurement by
properties of domains, such as shape, size, spatial distribution, methods of Euclidean geometry is not feasible and qualitative
spatial and crystallographic orientation, combined with variations of descriptions cannot sufficiently record them. During the last
mineralogical or chemical composition. These domains may range decades, in all fields of geoscience measurement methods were
from the micrometre- to kilometre-scale. Typically, they are repre- refined and experiments and modelling became significant parts of
sented by layering, cleavage planes, lineations, folds, fractures, faults, investigations. However, comparison of natural or technically
shear zones, joints, or filled veins, but also by lava flows, sediment produced structures among each other or with the results of
fans, rock fragments, crystals, crystal or particle aggregates, trails of experiments or modelling is effective only if such structures can be
inclusions, and coatings on rock faces. Most information about these described precisely, i.e., if they can be quantified. Only quantifica-
domains, specifically with respect to structure-forming processes, tion builds bridges between nature, technical products, experiment
are stored in their geometric properties and, consequently, a struc- and model.
ture can be represented partly or in many cases even totally by From the wide spectrum of geological structures the present
a binary pattern, i.e., by the purely geometric arrangement of one paper attends mostly to those that result from deformation
processes from the micrometre- to the kilometre-scale. Precise
measurement of these structures needs (i) transformation of
E-mail address: kruhl@tum.de. mainly diffuse and incomplete structures to binary patterns and (ii)

0191-8141/$ e see front matter Ó 2012 Published by Elsevier Ltd.


http://dx.doi.org/10.1016/j.jsg.2012.10.002
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 3

methods that are able to quantify the patterns’ complexity on These simple foliation patterns do not require elaborate pattern
different scales. Recording of complex structures is afflicted with analysis. However, the ‘complex’ arrangement of simple portions of
various problems and restrictions, which are presented and dis- a pattern or the way the material is organized, which is not
cussed. With respect to pattern quantification of geological struc- accessible to the observer at first view, contain rich information
tures, methods of fractal geometry (Mandelbrot, 1977) are best about the material’s properties, its history, and last but not least
choice, as convincingly demonstrated by Kaye (1989, 1993) in his valuable information about structure-forming processes, as will be
stimulating books. However, these methods include restrictions shown further below. Therefore, the characteristics of any struc-
that hamper their successful application and demands diligence tured matter are not represented by single fractions of the structure
and a critical approach to their usage. Nevertheless, fractal geom- but by their organization, i.e., the pattern. Analysing such patterns
etry offers highly effective quantification methods which give always means gaining information about the organization of
consideration to the complexity of natural structures and, in matter. Consequently, pattern analysis can be regarded as part of
addition, have a strong potential for further development. cybernetics (Wiener, 1948).
During the last decades the literature, related to the applica- When is a pattern a pattern? Even if patterns can be related to
tion of fractal geometry in geoscience, increased extremely. It is time, colours (e.g. the drip paintings of Jackson Pollock e Taylor
hardly possible, even solely restricted to structural geology, to et al., 1999), tunes (music e Voss and Clarke, 1975) or other prop-
completely treat of and cite these publications. Therefore, this erties, the term is generally related to the spatial arrangement of
review only presents a limited and necessarily subjective selec- small portions. Encyclopaedias define ‘pattern’ as combination of
tion of papers. single or several portions (figures). Recognizing patterns means to
observe higher interrelationships within a flood of small portions,
2. Complex rock structures and patterns which are not obvious at first glance. With respect to ‘pattern’ it is
clear that the summation of single portions is not the ensemble. The
Rock structures consist of matter with specific physical and relationship of the portions, their arrangement, adds an important
chemical properties. In most cases, the matter itself does not component and only this defines the characteristics of the
provide relevant information. For example, the mineralogical ensemble. In this way the patterns are formed, which are well
composition of a rock, such as 25 vol-% quartz, 15 vol-% K-feld- known to us from Persian carpets, architectural ornaments or the
spar, 40 vol-% plagioclase, 15 vol-% biotite and 5 vol-% white drawings of M.C. Escher. Since small portions of the same type can
mica, is just good for the cognition that it is most probably not be arranged in many different ways, already numerous patterns
a sedimentary rock. Distinction between ‘magmatic’ and ‘meta- arise. However, rock patterns comprise additional properties that
morphic’ would be only possible with additional ‘typical’ are immanent to most natural patterns: some randomness and
minerals, not to talk e in case of ‘metamorphic’ e about a more fuzziness and extension over several orders of magnitude, exactly
precise naming of the rock. Even if the metamorphic rock could what represents complexity (see further below).
be categorized the volume percentage of the minerals would The human brain is outstandingly qualified for storing and
provide nearly no information about the physical properties or recognizing patterns e patterns in terms of, for example, faces,
the history of the rock. Such information is mainly created by the sequences of tunes, or the configurations of pieces in a chess game
arrangement of the mineral grains or e in a more general sense e (Amidzic et al., 2001; Chase and Simon, 1973). However, it is
by the structure of the rock. Distribution and orientation of mica difficult for the brain to quantify such complex patterns. What
flakes may already determine if the rock can be categorized as does ‘complex’ mean and when is a pattern ‘complex’? The term
mica schist deformed by co-axial compression or simple shear or originates from the Latin “complexus” (¼ ‘connected’, ‘compre-
if shearing is localized (Fig. 1). hensive’) and is also defined in this way by encyclopaedias.
Systems are regarded as complex if they are multi-dimensional,
cross-linked, self-dynamic, diffuse, probabilistic and unstable to
a certain extent (Malik, 2004; Reither, 2004). Complexity can be
measured by ‘variety’: the number of all possible distinguishable
states of a system (Ashby, 1956, p. 121ff) e the higher the variety
the larger the complexity. With respect to rock structures, this
definition is reduced to the static components of the rock.
Complex structures are not composed of complicated discrete
parts (complicated in the sense of non-simple, without arrange-
ment and repetition on different scales), but of parts that are
simple and equal or similar to each other and clearly smaller than
the total structure, for example crystals, rock pebbles, straight
segments of fractures, cavities in a surface, etc. In complex struc-
tures the discrete parts are linked to each other, that means with
a certain probability and in the way that parts of the structure
slightly differ from each other or are slightly different on different
scales, i.e., are ‘fuzzy’.
Consequently, a complex rock structure can be defined as
a structure consisting of simple equal or similar parts that are
Fig. 1. Schematic sketches of schist or gneiss foliation structures, each of them
composed of roughly 90 bars. The structures are the result of various deformation clearly smaller than the total structure and repeated in a diffuse
processes and their spatial-temporal interaction. (A) Parallel foliation, possibly way over a certain range of magnitude. A complex structure is often
resulting from pressure solution under strong compression in a quartz-mica schist; (B) formed by a simple but iterative process. That means a process that
Weakly conjugate foliation planes, as they would develop during coaxial flattening in changed a structure at a specific moment acts in the next moment
a psammopelitic rock under low-temperature conditions; (C) Simple-shear deforma-
tion at the same conditions would lead to an ‘S-C’ structure; (D) Localized deformation
on this changed structure. Finally the overall change of the struc-
could lead to micro shear zones oblique to pre-given foliation, indicating a two-fold ture results from the summation of all these minimal changes by
deformation history. the iterative process.
4 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

3. Recording of structures in practice or quality of recording can also change and lead to a bias of the
subsequent pattern quantification.
The quality of the pattern quantification depends on the quality As long as techniques for automated detection of structures
of the pattern. The pattern’s quality in turn depends on three are only partly developed (see further below), in many cases the
things: (i) the completeness of the real structure and its extension manual transformation of a structure to a (binary) pattern is still
over as many orders of magnitude as possible, (ii) the quality of the best choice. However, wrong interpretations by the processor
recording, e.g., the resolution of a photograph or scan or the correct or selective sampling may lead to a considerable bias of the
way of recording the structure by a processor, (iii) the quality of pattern, more strongly if parts of the structure are missing, less if
transforming the structure to a binary pattern. Other aspects of the total structure is recorded imprecisely. Investigations on
a structure, which cannot be transformed to a binary code, are crystal distributions in rocks show: If single crystals are wrongly
beyond the scope of this paper and are not discussed. classified, e.g., plagioclase as K-feldspar, the pattern’s complexity
is more strongly changed than by inaccuracies, e.g. small shifts,
during recording of the crystals’ boundaries (Peternell and Kruhl,
3.1. Completeness of structures
2009). The reason is that a wrong classification of crystals tears
holes in the pattern or adds pieces. Through that the arrange-
Ideally, the observed and recorded structure represents the ‘real’
ment of pattern fractions, which is characteristic of the pattern, is
structure as completely as possible. This is nearly never the case
disturbed. In contrast, this arrangement is only weakly affected
with natural structures. On all scales and because of different
by small shifts of pattern fractions or the boundaries between
reasons wholes in the structure occur: a rock specimen is locally
them.
covered by lichens; a fresh quarry wall is impregnated by Fe
precipitation; in a geological map the boundaries between rock
3.3. Quality of transformation of structures to binary patterns
units are covered by younger sedimentary depositions. The repre-
sentation of structures in the map is often based on various
On the one hand, the transformation of structures to binary
assumptions depending on the processor and on the map scale.
patterns, which already exist as black-and-white sketches, is
That means, typically parts of the structure are imagined or the
a relatively trivial process. Possibly the lines have to be thinned and
structure contains gaps that are ignored. Usually no issue is made
artificial irregularities have to be cleared. On the other hand, already
out of the incompleteness of structures and many publications deal
during this process bias can occur, which affects the results of the
with incomplete structures, e.g., fracture systems on map-scale,
subsequent fractal geometry analyses. This is discussed by Gonzato
without discussion. Incompleteness does not necessarily lead to
et al. (2000). However, the binary coding of structures that exist as
inaccurate results. ‘Holes’ in structures should be looked at more
colour or grey-shade photographs is by no means trivial and it
critically if they have shapes and distributions, which themselves
affects the subsequent pattern quantification. The manual recording
show a specific complexity, leading to systematic changes of the
of different grain-boundary patterns, e.g., quartz, K-feldspar,
structure’s complexity. For example, the size distribution of lichens
plagioclase, biotite, in square-decimetre large areas and down to the
is complex (Morse et al., 1985; Shorrocks et al., 1991) and would
100-mm scale may take up to 50 times as long as automated
accordingly change the remaining area of fresh rock surface. This
recording (Peternell and Kruhl, 2009). Therefore, automation of
problem cannot be removed in principle. Possible impacts on
binary coding is a necessary pre-condition for effective pattern
quantification can be estimated if the scale, on which gaps in the
analyses.
structure occur particularly frequently, is considered.
If grey-scale photographs show strong contrast of brightness the
Vice versa, fractal geometry (Mandelbrot, 1977) offers the
structures to be investigated can be generated by application of
possibility to fill gaps and complete the structure-related pattern. If
a threshold. This procedure is available in commercial image pro-
in smaller complete areas the pattern always shows the same
cessing or drawing programs. The threshold can be selected by
fractal dimension, i.e., if it is homogeneous, the same fractality can
various methods. For example, a computer-aided threshold selec-
be also assumed where gaps occur. Consequently, these gaps can be
tion is performed by stepwise approaching the optimal threshold,
filled with an artificial pattern of the same fractality. In addition, all
based on small changes of pixel numbers (Gerik and Kruhl, 2009). If
natural structures are limited in scale. The upper limit is mostly
several patterns are to be generated from colour or grey-scale
determined by the size of the outcrop, e.g., the size of a quarry wall
photographs, images can be segmented by grey-level slicing or
or the area of basement rocks exposed under a sedimentary cover.
colour threshold, based on commercial image processing programs
The lower limit usually corresponds to the resolution of the
(Andriani and Walsh, 2002; Coster and Chermant, 2001). This may
recording method (Section 4.1).
lead, for example, to differentiation between various minerals or
clasts in fragmented rocks and forms a basis for subsequent quan-
3.2. Quality of recording tification of the pattern. In addition, special ranking filters can be
applied in order to reduce the noise without changing the original
Complex rock structures on various scales can be best recorded shapes of material domains (Heilbronner and Keulen, 2006).
by satellite/aerial, outcrop or specimen photographs, scans of rock Further improvement of the pattern requires revised manual
cuts or thin sections, or on the basis of geological or tectonic maps. processing with additional time-consuming input of information.
With all these methods the quality of recording can be highly More sophisticated automated methods like ‘water-flow’ (Baccar
variable. Specifically with field photographs surface alteration, et al., 1996; Vincent and Soille, 1991), suitable for mineral-phase
variable incidence of light intensity, shadow effects on rough segmentation (Barraud, 2006), are extremely time-consuming as
surfaces, distortions on inclined planes and of margins of photo- well. In the end, it is the processor’s task to find a compromise
graphs can generate a local and overall bias of the recorded struc- between a match, as high as possible, of the original structure and
ture. The compensation of margin distortions and different the binary pattern, dependent on the requirements of the intended
illuminations in merged photographs can cause difficulties (further method, and the necessary amount of time. Comprehensive
discussions in Peternell, 2007; Peternell et al., 2011). In addition, investigations on the accuracy of binary coding of a structure and
manual recording may lead to an oversimplification of structures. If the accuracy of quantifying the binary pattern are missing, except
the recording technique changes on different scales accuracy and/ for few approaches (Peternell and Kruhl, 2009).
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 5

3.4. Automated recording of structures impossible. Therefore, structures are still recorded in a descriptive
or half-quantitative way. In contrast to fractal geometry methods,
Automated recording of structures and binary coding are classical techniques of structure quantification (presented in detail
worthwhile not only because they are less time-consuming than by Higgins, 2006) do not measure structures as patterns but treat
manual recording and processing but also because of the higher their fractions mainly by statistical methods: particle-size
accuracy of the resulting pattern and thereby its possible larger frequencies in sedimentology (Pettijohn, 1975); Crystal Size Distri-
size. Consequently, structures can be recorded over more orders of bution (CSD e Marsh, 1988); Shape Preferred Orientation (SPO e
magnitude. The quantifications of these structures have higher Panozzo, 1984; Launeau et al., 1990). Other methods deal with
significance and potential changes of the structures on different patterns but are only designed for simple ones (Launeau and Robin,
scales can be identified more easily. Particularly quantifications of 1996).
anisotropy and inhomogeneity depend on the size of the recorded In contrast, fractal-geometry based (or fractal-geometry related)
structures. Above all, small-sized structures, as common with methods at least partly record the internal connection of patterns.
manual recording, do not allow determination of inhomogeneity However, clear differences exist between different methods. Since
with adequate accuracy (see further below). Benoit B. Mandelbrot’s famous question “How long is the coast of
During the last years, methods of automated structure identifi- Britain?” (Mandelbrot, 1967) and his ground-breaking book on the
cation have been developed. For example, phase boundaries can be ‘fractal geometry of nature’ (Mandelbrot, 1982) a wide auditorium
identified and recorded with the aid of a watershed segmentation became familiar with fractal geometry, also through numerous
procedure (Barraud, 2006) or the constrained automated seeded textbooks that demonstrated the various applications of fractal
region growing (CASRG) algorithm (Choudhury et al., 2006). For high geometry in different fields of geosciences (Barnsley, 1988; Bunde
accuracy however, manual corrections of the images are necessary. and Havlin, 1994; Falconer, 1990, 1997; Feder, 1988; Hasting and
Further possibilities of determining phases and phase boundaries Sugihara, 1993; Kaye, 1989, 1993; Peitgen and Richter, 1986; Peitgen
are offered by the automated recording of crystal orientations, and Saupe, 1988; Peitgen et al., 1992a,b; Stanley and Ostrowsky,
either by computer-aided polarizing microscopy (Fueten, 1997; 1988; Turcotte, 1997; and many others).
Panozzo Heilbronner and Pauli, 1993) or by electron microscopy Right away, fractal geometry methods were recognized in geo-
methods, e.g., electron backscatter diffraction (EBSD e Adams et al., sciences as attractive ways of quantifying complex structures and
1993). However, the latter are elaborate and of only limited use for gaining important information about material properties,
processing large-scale structures. Automated pore-space measure- structure-forming processes and for comparison with results of
ments in sandstone, based on scanning-electron-microscope experiments and technical processes. Or from another viewpoint,
images, have been performed by Krohn and Thompson (1986). fractal geometry is attractive because it “provides a proper
The multi-spectral image processing method (Marschallinger, language and symbolism for studies of ill-defined geometries”
1997) is a promising technique for automated recording of (Avnir et al., 1998). In addition, interpolation may be applied to
mineral grains, which applies neural networks that represent incomplete structures or extrapolation may allow extension of
adaptive systems (Baykan and Yilmaz, 2010; Fueten and Mason, a structure to smaller or larger scales (Mareschal, 1989; Watterson
2007; Goodchild and Fueten, 1998; Thompson et al., 2001). With et al., 1996). The latter, however, only works if the structure (i) is
the aid of such methods, phase boundaries and minerals can be homogeneous and (ii) shows the same scaling behaviour over the
identified and recorded with high probability. In principle, based on entire range of scale. In many cases these conditions are not ful-
further development of these methods, any complex structure in filled. Even if the fractality of many geological structures was
rocks could be recorded automatically. Significant progress in future recognized limitations of the methods were pointed out together
structure recording and processing can be expected in this field of with the fact that by far not all geological structures are fractal
investigation. (Bonnet et al., 2001; Korvin, 1989; Nicol et al., 1996). For example,
Nearly all rock structures occur in 3D, however, because of they may be log-normal or exponential. Log-normal distributions
practical reasons are mostly recorded and analysed in 2D. On the are reported from fracture systems (Foxford et al., 2000; Gerik and
basis of automated recording of structures in 2D, grain fabrics can Kruhl, 2009; McCaffrey et al., 1993; Narr and Suppe, 1991) or from
be reconstructed and visualized in 3D (Marschallinger, 1998, 2001) mineral distribution patterns (Gerik and Kruhl, 2009; Gerik et al.,
and presented as World Wide Web documents using special visu- 2010; Peternell et al., 2011). Strictly seen, such distributions are
alization techniques, partly with interactive user control outside the field of fractal-geometry analysis but nevertheless may
(Marschallinger and Johnson, 2001). Pore-space, fracture systems provide useful information, not least about pattern inhomogeneity
or crystal distributions can be measured in 3D as accurately as and anisotropy.
necessary for most applications, for example, by X-ray or neutron This touches the more general question why natural data should
tomography (Denison et al., 1997; Ketcham and Iturrino, 2005; be fitted to one specific law (if necessary ‘by force’) and not just to
Vontobel et al., 2005; Winkler et al., 2002), computer tomography the one with the relatively best fit. One answer is that ‘fit’ is not
based on 2D sections (Marschallinger, 1998, 2001), or by laser and sought for its own sake but in order to gain information about the
photogrammetric methods (Tonon and Kottenstette, 2006). Most of process (or the processes) behind the natural data. Since natural
these methods are elaborate and time-consuming. data may show natural variation, e.g. a random component with
local or temporal variation, ‘best fit’ would possibly catch such
4. Quantification of structures by fractal-geometry methods meaningless component. Consequently, ‘best fit’ could indicate
laws, i.e. processes, or their variations which are meaningless. A
Measurement of complex structures (or patterns) by Euclidean second answer would be that log-normal and exponential distri-
geometry is difficult or simply not possible. Complex mathematical butions are limited to a certain scale whereas fractal distributions
patterns, in fact, can be quantified based on their design instruction. are not. Therefore, e.g. log-normal distributions satisfyingly
However, when a pattern extends over an infinite number of orders describe scale-limited data sets, such as grain-size distributions,
of magnitude the failure of applying Euclidean geometry methods whereas fractal distributions better represent structures resulting
becomes obvious. Natural structures extend over only a limited from processes, such as fracturing or diffusion, which are seen as
range of scale but their irregularity and their only statistical self- invariant over certain ranges of scale, based on observations as well
similarity make measurements extremely difficult, if not as theoretical considerations (Sammis et al., 1987). These answers,
6 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

however, do not question the fact that the fractality of a pattern Self-similarity of complex patterns generally leads to the power-
may be different on different scales (Section 5). law relationship N(s) w sD, with N ¼ number of objects  s
Methods of fractal geometry are only applicable to the quanti- (Mandelbrot, 1982). The exponent D, the fractal dimension, charac-
fication of complex structures. They fail in quantifying simple terizes the complexity of the pattern. A logarithmic representation
structures or yield results that appear useful but are meaningless. changes the function to log (N) w D log (1/s), i.e., to a straight line
In this respect, computer programs that work as a black-box are with slope D in a double-logarithmic diagram. The advantage of such
particularly dangerous. Further discussion of this topic is given by graphical representation is that variations of measurement points
Harris et al. (1991). In addition, bias is caused by (i) insufficient from the linear arrangement are easily recognizable and can be
sampling, (ii) low number of data points, (iii) incorrect set of the eliminated or, if systematic, used for analysis. Variations from line-
analysis area in relation to the pattern, and (iv) if the resolution of arity often occur at the upper and lower limits of measurement range
the analytical instrument, e.g., the box size in case of the Box- (Carpinteri and Chiaia, 1997; ‘roll off’ e Blenkinsop and Sanderson,
Counting Method or the step length of the Divider Method, 1999; Pickering et al., 1995) (Fig. 3) and define upper and lower
approaches the resolution of the structure. These and other sources limits of fractality. Mainly at the lower limit the structure is often
of error are discussed in detail elsewhere (Gonzato et al., 1998, incompletely recorded because of poor outcrop situation on outcrop
2000). or map scale, or because small parts of the structure are less well
‘visible’ and therefore not fully recorded (Fig. 2). Consequently, they
are under-represented, the measurement points deviate from the
4.1. Natural fractals and their properties
trend (undersampling e ‘truncation’; Ackermann et al., 2001;
Blenkinsop and Sanderson, 1999; Ford and Blenkinsop, 2008a,b;
From a mathematical viewpoint rock structures (or patterns) are
Jackson and Sanderson, 1992; Pickering et al., 1994, 1995; Watterson
not fractal because they do not fulfil fundamental requirements of
et al., 1996) and are omitted for the analysis. Typically, structures in
fractality: They are not self-similar over an infinite range of scale.
thin section, i.e., with ‘100% outcrop’, do not show such an effect
Their self-similarity, like that of all other natural structures, partly
(Kruhl, 1994).
extends over less than one order of magnitude or different scaling
In addition, a complex structure may not be fractal on small
behaviour occurs on different scales, with often diffuse transitions.
scale or shows insufficient resolution, or the actual features are
At the upper limit, each complex rock structure is too coarse, i.e.,
Euclidean. For example, crystal or grain aggregates are frequently
does not contain enough substructures or fractions for statistically
fractal on larger scale, but Euclidean on smaller scale, caused by
safe analyses. At their lower limit, rock structures are either
straight crystal or grain faces (Kruhl and Nega, 1996). The same is
incomplete, due to the limit of resolution, or show Euclidean
true for particle sizes in fragmented rocks. Their frequency distri-
geometry (Fig. 2). In contrast to mathematical fractal patterns,
bution changes from fractal to non-fractal below a certain size that
geological structures always bear a component of randomness,
is governed by cleavage planes or faces of crystals (Sammis and
mainly caused by the inhomogeneity of the material and by the
Biegel, 1989). Fracture patterns often consist of short straight
interaction of the different structure-forming processes. The main
segments, partly because they are actually straight, but also
differences, however, to mathematical structures are: Geological
because they are straightened during transfer from nature to the
structures are often different (i) on different scales (pattern
line pattern, e.g., by the processor or based on the resolution of the
scaling), (ii) in different areas (pattern inhomogeneity), and (iii) in
recording instrument. Pattern fractions with Euclidean outlines can
different directions (pattern anisotropy). These three structure or
assemble to larger fractal patterns (Kaye, 1984, 1989, p. 87). The
pattern properties are geologically significant and often reflect
‘roll-off’ effect is also related to large objects with, due to various
various geological processes.
reasons, low numbers (‘censoring’ e Ackermann et al., 2001;

Fig. 3. Measurement points in a double-logarithmic plot, generated by the application


of the Structured-Walk (Divider) Method to a complex grain-boundary curve. The total
lengths of polygons are plotted versus their stride (step) lengths. The roughly linear
arrangement of data points changes at its upper and lower limits (‘roll off’). Belo
a certain small scale (solid arrow) Euclidean geometry of the curve prevents further
increase of total length (stippled line) and, consequently, leads to a fractal dimension of
1. This change of linear arrangement of data points may be also caused by incomplete
Fig. 2. Complex fracture pattern in a sandstone layer composed of several sets of recording of small parts of the structure (‘truncation’). Above a certain large scale
sub-parallel fractures. The pattern is bounded by a felsic dyke (D) and over- and (stippled arrow) large parts of the structure occur in generally low number
underlying rock layers (bottom and top) without fractures. Locally, the pattern is (‘censoring’). In the case of the Structured-Walk Method the low number of strides leads
incomplete due to steps in the surface and salt precipitation (arrows). Ellipses: regions to a scatter of data points, increasing with increasing stride length and caused by
where fractures ‘fray’, possibly to arrays of microfractures, and become invisible in the various starting points of the structured walk. In the example only the highest and
photograph. Coastal outcrop, Mallacoota region, eastern Victoria/Australia. Coin for lowest values are shown, connect by solid lines. Steps of the data point arrangement
scale. are highlighted by stippled lines. Modified after Kruhl and Nega (1996).
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 7

Jackson and Sanderson, 1992; Pickering et al., 1994). In addition, the (Kruhl and Nega, 1996). This relationship represents a possibility for
weak statistics of large pattern fractions is reflected by the scatter of quantifying and investigating regularity of patterns (Şuţeanu and
measurements, which does not allow any fit of the data points Kruhl, 2002). Cyclic stepping of measurement points frequently
(Fig. 3). appears in publications but is nearly never mentioned or discussed.
The relatively imprecise application of quantification methods is How many orders of magnitude should a complex structure
criticized from a mathematical view (Tricot, 1995). Such criticism is cover for meaningful quantification by fractal geometry? This is
legitimate insofar as in many investigations (i) the boundary a controversial matter (Avnir et al., 1998). Usually one to two orders
conditions of applications are not specified, (ii) the ranges of of magnitude is seen as lowermost limit, probably also because
measurements or analyses are defined visually, i.e., subjectively, (iii) most geological structures do not extend much beyond this range
the range of analysis is too short and the data basis too small for (Fig. 5). In rare cases geological structures may range over up to 6e7
statistically significant conclusions, (iv) the methods are applied to orders of magnitude (Castaing et al., 1996; Jensen et al., 2011; Long
patterns, for which they are not suitable. On the other hand, the and Imber, 2011; Zhao et al., 2009). On the other hand, in some
analysis of natural structures mostly does not intend to find classical examples self-similarity is only present in an interval close
a mathematical model that fits an ‘irregular’ pattern as accurately as to one order of magnitude or even below (Kaye, 1989, 1993).
possible. Application of fractal geometry methods, e.g. to rock However, there is no reason to believe that the physical relevance of
structures, mainly serves to detect regularity or systematics in these self-similarity of a structure increases with increasing range of
patterns and correlate them with pattern-forming processes. In the scale. If self-similarity indicates activity of a specific process a small
foreground is the connection between different pattern character- range of self-similarity only means that this process was active over
istics, such as inhomogeneity, anisotropy and their variation on this small range. Activity over a larger range does not necessarily
different scales, and their comparison to potential pattern-forming indicate greater importance for material behaviour or develop-
processes. Mathematical precision is important but not essential. ment. The confirmation of self-similarity, however, depends on the
Moreover, variations of natural patterns from mathematical fractals resolution of the structure over the range of measurement and on
do not represent ‘defects’ that, if not ‘repairable’, should at least be the accuracy of measurement. Self-similarity of a structure, well
minimized. In fact, such variations are characteristic of natural resolved over a large range of scale, is definitely more meaningful
patterns and they often represent an important source of informa- than of a structure poorly resolved over a short range of scale. The
tion for investigating natural structures and the processes behind latter may be even completely meaningless.
them. Since rock structures result from the interplay between material
The single fractions of a mathematical fractal, e.g. of the triadic properties and material-forming processes, self-similarity over
Koch Curve (Mandelbrot,1982), show specific sizes and are arranged several orders of magnitude would imply constancy of processes
with specific distances and angles to each other. Measuring such and properties over such a range. This is rarely the case with
regularity of a pattern by fractal geometry methods results in cyclic deformation or other structure-forming processes. Many geological
stepping of the measurement points in a logelog plot (Kaye, 1989, p. structures are self-similar only over relatively short ranges of scale
41). Such an effect of pattern regularity becomes specifically but may be differently self-similar on different scales (Section 5).
apparent with mathematical fractals that extend over a limited For example, different deformation processes on different scales are
range of scale. Typically, the few but distinct steps caused by the well known from feldspar deformation experiments (Tullis and
pattern’s regularity increasingly disappear with decreasing regu- Yund, 1985) where single crystals are stronger than crystal aggre-
larity (Fig. 4). Vice versa, such systematic deviations from a power- gates. The reason is that single crystals are dominantly deformed by
law distribution are also generated by the measurement of regular intra-crystalline processes whereas in crystal aggregates defor-
complex natural structures, independently of the applied method mation also acts along crystal boundaries. In rocks, fracture
patterns may change from the centimetre to 1000-kilometre scale,
as a result of different material fabrics and deformation behaviour

Fig. 4. Fractal analysis of the outlines of a ‘regular’ (left) and an ‘irregular’ Koch island
(right), performed by the Structured-Walk Method and presented in a double-loga-
rithmic plot. The ‘irregularity’ is produced by (i) shifts of the boxes parallel to their base
and (ii) by small changes of the box sizes without changing the total area of the island.
Both fractal dimensions D are similar and slightly different from the fractal dimension
D ¼ 1.292, given by the construction procedure. The measurement points related to the
regular Koch island show distinct steps, i.e., neighbouring measurement points with
different stride length but nearly equal polygon length. These steps are not present in Fig. 5. Frequency distribution of intervals of fractality, i.e. of linearity of measurement
the arrangement of measurement points of the irregular Koch island. For reasons of points in a double-logarithmic plot. The data are mostly related to rock structures and
clarity, measurement points of the ‘irregular’ curve are shifted one order of magnitude based on 109 publications with 123 diagrams from 46 geoscience journals in the
to the right. period 1983 to 2012.
8 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

on different scales, leading to an according change of fractality complete decomposition of the pattern to single pieces. These
(Castaing et al., 1996; Nicol et al., 1996). Often, fractality is valid on pieces are thrown in one pot and represented statistically (Fig. 6A),
relatively short ranges of scale, partly even below one order of e.g., as cumulated frequency distributions. This type of presentation
magnitude. Such short intervals are only acceptable in relation with and analysis has long tradition in geosciences and also rock struc-
the application of a strict methodology. Since slopes of linear point tures are studied in this way. It is equivalent to the analysis of data
arrangements and, consequently, changes of slopes can be tracked sets. Such sets can have a structure, e.g., order of magnitude of
more precisely over relatively large intervals, the widest possible numbers, or are chronologically structured. For example, tectoni-
range of analysis is requested (Walsh and Watterson, 1993). cally induced events like earthquakes can show scale-invariant
behaviour with respect to their sizes and chronological appear-
4.2. Fractal-geometry methods ance (Kagan, 2010; King, 1983; Korvin, 1992; Smalley et al., 1987;
Rundle, 1989; Tang et al., 2012; Turcotte, 1986b; and others). Spatial
Since different fractal-geometry methods applied to the same or temporal-spatial variations of seismicity show scaling properties
pattern may lead to different fractal dimensions, usually the (Lunina et al., 2005; Şuţeanu and Ioana, 2007; Xie and Pariseau,
method is indicated together with the dimension. This results in 1993). In most cases, however, chronological data sets from rocks
a complete zoo of fractal dimensions. Different methods are var- cannot be measured since (i) dating the development of deforma-
iably useful for different patterns. Since these methods are inten- tion structures is nearly never possible or (ii) the underlying
sively presented in the literature (Feder, 1988; Kaye, 1989, 1993; processes are too slow to allow practicable observation periods.
Mandelbrot, 1982; and many others), only the most common will Therefore, nearly all data sets are recorded without chronology.
be briefly discussed here in relation to quantification of rock Particle sizes form a common type of data set related to rock
structures. deformation. Their cumulative frequency curves generally follow
Different fractal-geometry methods consider the internal a power law, for example in relation to particles from fragmented
structuring of a pattern in different ways. At one end stands the rocks (Blenkinsop,1991; Blenkinsop and Fernandes, 2000; Imre et al.,

Fig. 6. Different types of quantification of a fracture pattern by various fractal-geometry methods. (A) Disassembling of the pattern to single pieces for further statistical treatment. (B)
Measurement of segments or spacing between parts of the pattern along scan-lines, again, for further statistical treatment. (C) Measurement of spacing or widths of planar parts of
the pattern for further statistical treatment. (D) Sequence of intersection points between the pattern and a scan-line, leading to a Cantor set. (E) Box Counting applied to the pattern,
providing information about the internal structure of the pattern, i.e. the arrangement of parts of the pattern relative to each other. (F) Box-Counting applied to small areas that are
shifted over the pattern leads to additional information about the inhomogeneity of the pattern. (G) One-dimensional methods, e.g. Cantor-Dust method (D), applied to various
directions lead to quantification of the anisotropy of the pattern. (H) Combined quantification of inhomogeneity and anisotropy leads to a maximum of information about the
pattern’s variation. The pattern is drawn after a field photograph (Fig. 2). Light grey: areas where the fracture pattern is masked by a salt cover. Ellipses: regions where fractures ‘fray’,
possibly to arrays of microfractures, and become invisible in the photograph.
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 9

2010; Korvin, 1989; Main et al., 2000; Nagahama, 1993; Shepherd, distributions (Cello et al., 2000; Hooker et al., 2011; Renshaw and
1989; Şuţeanu et al., 2000), gouge formation and experimental Park, 1997), as well as length and displacement of deformation
fragmentation (Biegel et al., 1989; Billi, 2005, 2007; Billi and Storti, bands (Fossen and Hesthammer, 1997), length and throw or
2004; Heilbronner and Keulen, 2006; Keulen et al., 2007; Korvin, displacement of faults (Bonnet et al., 2001; Bour and Davy, 1997;
1989; Main et al., 2000; Mair et al., 2002; Marone and Scholz, Davy, 1993; Davy et al., 1992; Dawers et al., 1993; Jackson and
1989; Monzawa and Otsuki, 2003; Morgan and Boettcher, 1999; Sanderson, 1992; Main et al., 1990; Marrett and Allmendinger,
Sammis and Biegel, 1989; Sammis et al., 1986, 1987; Shepherd, 1989; 1991; Nagahama, 1991; Odling et al., 1999; Pickering et al., 1995;
Wilson et al., 2005), pseudotachylites (Hisada, 2004; Shimamoto and Renshaw, 1999; Villemin et al., 1995; Walsh et al., 1991; Watanabe
Nagahama, 1992), or volcanic material (Kueppers et al., 2006; and Takahashi, 1995; Watterson, 1986; Zygouri et al., 2008), and
Perugini et al., 2011; Perugini and Kueppers, 2012). However, there spacing of faults, joints and veins (Brooks and Manning, 1994; Cello
are distinct exceptions (Heilbronner and Keulen, 2006; Keulen et al., et al., 2000; Fagereng, 2011; Gillespie et al., 1993, 2001; Ledésert
2007). Even if the analysis of data sets does not access a larger part of et al., 1993a,b; Manning, 1994; Marrett et al., 1999; Zazoun, 2008).
the information stored in a pattern this type of analysis rightly has However, vein spacing may show log-normal frequency distribution
a long tradition. It is not only fast and straightforward but frequently (McCaffrey and Johnston, 1996) and geometries of veins and faults
is the only practicable possibility of data acquisition. may not be fractal, i.e. self-similar, but self-affine (Blenkinsop, 1989;
Complex structures are characterized by the arrangement of Hull, 1988; Johnston and McCaffrey, 1996; Walsh and Watterson,
smaller fractions of the structure. Consequently, a method is more 1988).
‘valuable’ if it takes this arrangement and the internal geometry of Analysis of spacing and thickness of planar structures along
‘sub-structures’ into consideration e more valuable because less scan-lines leads to alignment of dots or segments, which is usually
information about the geometry of the structure gets lost and more quantified by the Cantor-Dust Method, either (i) the Spacing-Pop-
information about structure-forming processes is preserved. Most ulation Technique (Harris et al., 1991; La Pointe, 1988; Merceron and
information gets lost by the disassembling of a structure or pattern, Velde, 1991) or (ii) the Interval-Counting Technique (Harris et al.,
respectively. Slightly more information, however, is conserved if 1991; Velde et al., 1990, 1991). Whereas the Spacing Population
a structure is segmented along scan-lines of certain orientation Technique performs a statistical analysis of the segment lengths,
(Fig. 6B) and if the segments, e.g., fragment widths parallel to the which does not consider the connection between the segments, the
scan-line, are statistically treated. With this procedure the internal Interval-Counting Technique works like a 1D version of Box Counting:
connection of the structure is not totally lost because the rela- The initial 2D pattern is transferred to a new 1D pattern that at least
tionship between structure and scan-line orientation contains partly preserves the spatial connection of the 2D pattern. However,
information that can be used for the analysis of anisotropy (Section since the Interval-Counting Technique requires that the scan-line
7). The situation is similar if spacing or widths of planar structures profile represents a Cantor Set, application of this technique is
(joints, faults, dykes) or of planar fractions of more complex limited to fractal patterns. In contrast, the Spacing-Population
3D structures are measured (Fig. 6C). This also leads to data sets Technique can also operate with log-normal distributions. This type
with a relationship to directions as additional information. If the of analysis of planar structures allows anisotropy quantification
sequence of segments or intersection points along the scan-line is (Section 7).
preserved (Cantor Set, Fig. 6D) the information content and the Complex linear rock structures predominantly result from
pattern is represented more ‘completely’, at least along the direc- outlines or cross-sections of complex 3D structures. Based on the
tion of analysis. Since many portions of structures are more or less Structured-Walk Method (or Divider-, Ruler- or Yardstick method),
planar and parallel to each other, such as extension veins or folia- established by Richardson (1961) already in ‘pre-fractal’ times,
tions the possible complexity of their width and spacing can be various modified methods were developed, e.g., Equipaced-Polygon
sufficiently characterized based on measurements along cross Method (Kaye, 1978; Kaye and Clark, 1985), Hybrid-Walk Method
profiles. For non-linear or non-planar structures most information (Clark, 1986), Cell-Count Method (Goodchild, 1980), Area-Perimeter
is conserved with those methods that consider the orientation of Method (Lovejoy, 1982; Mandelbrot, 1977), Cumulative Number-Area
pattern fractions relative to each other. The best known method is Method (Wang et al., 2007) or Euclidean-Distance Mapping (EDM e
Box Counting (Fig. 6E). Russ and Russ, 1989) for fragment morphologies (Bérubé and
However, all these methods do not catch the most typical prop- Jébrak, 1999). These methods were mainly applied to outlines of
erties of rock structures or their patterns, respectively: anisotropy sediment particles (Hyslip and Vallejo, 1997; Orford and Whalley,
and inhomogeneity. In order to receive information about these 1983, 1987), mineral grains (Kaye, 1989; Petford et al., 1993; Wang
properties, combinations of methods are required. Inhomogeneity et al., 2007), rock fragments (Lorilleux et al., 2002), profile lines
(of pattern complexity) is best quantified by applying Box Counting across rugged surfaces (sea-floor bathymetric profiles: Barenblatt
to systematically selected sub-patterns (Fig. 6F; see Section 6 for et al., 1984; sea-floor topography: Mareschal, 1989; fracture
details). Anisotropy (of pattern complexity) can be quantified by surfaces: Kristáková and Kupková, 1994) and to sutured grain
applying 1-dimensional quantification methods to different direc- boundaries (Kruhl and Nega, 1996; Kruhl et al., 2004; Majumder
tions (Fig. 6G; see Section 7 for details). In conclusion, most and Mamtani, 2009; Takahashi et al., 1998). All these methods can
information about a pattern is provided by combined anisotropy- be additionally used for anisotropy quantification (Section 7).
inhomogeneity quantification (Fig. 6H). Box Counting (Falconer, 1990; Kaye, 1989; Mandelbrot, 1982) is
In structural geology analysing frequency distributions of thick- excellently suitable for analysing complex and isotropic 2D
ness and spacing of fractures (faults, joints) or veins has long patterns. It is simple and can be easily programmed (Gonzato, 1998;
tradition. In many, but by no means all, cases these analyses of data Peitgen et al., 1992b) and is non-centred, i.e. does not rely on
sets lead to power-law distributions. Widths of veins with different a central symmetry of the pattern (Vignes-Adler et al., 1991). Frac-
fillings and from different geological environments have been ana- tures patterns are classical examples of the successful application of
lysed by André-Mayer and Sausse (2007), Clark et al. (1995), Box Counting on the micro- to mega-scale and to various materials.
Fagereng (2011), Gillespie et al. (1999, 2001), Johnston and Numerous examples appear in the literature (Aviles et al., 1987;
McCaffrey (1996), Loriga (1999); McCaffrey and Johnston (1996); Ayunova et al., 2007; Bornyakov et al., 2006; Castaing et al., 1996;
McCaffrey et al. (1993), Kruhl (1994), and Sanderson et al. (2008). Cello, 1997; Hirata, 1989; Nanjo et al., 1998; Park et al., 2010; Zhao,
Also fault zone and fracture aperture sizes yield power-law 1998; etc.). In addition, many other structures are suitable for Box
10 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

Counting, such as distribution patterns of crystals in magmatic rocks such extrapolation is valid. Certainly, homogeneity and isotropy of
(Armienti and Tarquini, 2002; Peternell and Kruhl, 2009) or the the structure are preconditions. Most methods of measuring frac-
distribution of flank cones of volcanoes (Mazzarini and Armienti, tality in 1D and 2D can be transferred to 3D, most prominently Box
2001). Moreover, 2D distributions of events represent patterns Counting. Automated 3D Box Counting is increasingly used, e.g., for
that can be analysed by Box Counting, for example clustering of quantifying the 3D fractures patterns formed during rock failure
earthquakes (Nanjo and Nagahama, 2000; Smalley et al., 1987). (Yuan and Li, 2009). Furthermore, the 3D situation is important
Furthermore, the method is well suitable for recording pattern for analyses of spatial distributions of earthquake hypocentres and
inhomogeneity (Section 6). Recently, an improved Box-Counting fracture networks (Sahimi et al., 1993), simulations of fracture
method has been presented, which guarantees higher accuracy (Li networks (Hamdi, 2008), and connectivity and permeability of
et al., 2009). fracture networks (Bour and Davy, 1998; Fomin et al., 2003;
The Mass Dimension or Density Dimension Method (Kaye, 1993) is Ledésert et al., 1993a,b, 2010) as well as pore structure (Feranie
mainly used for quantifying centred patterns, e.g., crystal aggre- et al., 2011).
gates or dendritic structures (Roach and Fowler, 1993). Related with However, many rock structures do not fulfil the preconditions of
these methods is the Weight-Repartition Method, which was homogeneity and isotropy. Already on sample scale, grain or
intensely used for the analysis of fracture centres (Bonnet et al., particle distribution patterns can be different in different regions of
2001; Chilès, 1988; Davy et al., 1990; Okubo and Aki, 1987; the study volume and fracture patterns in shear zones are
Sornette et al., 1993; Zazoun, 2008). The Correlation-Dimension frequently anisotropic. Inhomogeneity can be detected by analysing
Method (Grassberger and Procaccia, 1983) is applied to crystal different parts of the study volume or by 3D-related kriging
morphologies (Fowler, 1990; Fowler et al., 1989; Singer and Bilgram, procedures. The analysis of serial sections is helpful insofar as it is
2006). able to unveil inhomogeneity but the 2D fractalities of these
Despite all these methods, new techniques or modifications of sections cannot be readily combined to the 3D fractality of the study
existing techniques are definitely needed as demonstrated by the volume.
analysis of pattern inhomogeneities and anisotropies (Sections 6 In order to quantify anisotropy in 3D, modified 2D methods are
and 7). The necessity already arises from the fact that components necessary, e.g., a Cantor-Dust method that analyses intersection
of complex patterns often hold different weight and, therefore, points of stepwise rotated scan-lines through a 3D pattern, or a 3D
different treatment of these components during analysis may modification of the Mass-Dimension or Density-Dimension methods.
provide additional information about the pattern. For example, the However, the anisotropy quantification of rock structures in 3D is
intersections or triple-points in fracture systems with hierarchical not particularly relevant. Structures formed in stress fields are
structures may have different importance, which can be specified by predominantly planar or linear. Such type of anisotropy can be
a weighted analysis. However, problems can arise from the variably analysed by 2D methods in orthogonal sections. Nevertheless,
good applicability of methods on the quantification of different methods for 3D quantification of structures are considerably
structures and from the fact that different methods can lead to needed, the more because structures are increasingly recorded in
different results (Bérubé and Jébrak, 1999; Walsh and Watterson, 3D by automated procedures that provide detailed, mostly digital
1993). data sets (Section 3.4).
Since the beginning of the application of fractal-geometry
techniques there is an increasing amount of literature on multi- 5. Scaling regimes and structuring
fractal analysis. The concept was introduced in order to fulfil
requirements not met by ordinary fractal (‘monofractal’) analyses As discussed above, scaling properties of rock structures usually
(Hentschel and Procaccia, 1983; Mandelbrot, 1985, 1989; Evertsz are not constant over several orders of magnitude. It is well-known
and Mandelbrot, 1992). Multifractals are interweaved fractals with that self-similarity of natural patterns can vary on various scales
continuous or discrete spectra of fractal dimensions (Stanley and (scaling regimes) and different fractality on different scales is
Meakin, 1988; Cheng and Agterberg, 1996; Cheng, 1997). Multi- standard, not the exception. Scaling is a fundamental feature
fractal analyses can provide more information about measurements of properties in most crystalline as well as non-crystalline
on spatial objects than monofractal analyses insofar as, for example, material. A classical example is the difference between textural
variations of values related to different parts of a structure can be and structural fractality (Horovistiz et al., 2010; Kaye, 1978, 1989 p.
analysed (Mandelbrot, 1989; Cheng and Agterberg, 1996). With 27; Kindratenko et al., 1994; Orford and Whalley, 1983), which
respect to rock structures, multifractal analyses have been describes the difference between the fractality of small entities, e.g.,
successfully applied to fracture systems (Agterberg et al., 1996; mineral grains or dust particles, and the fractality of agglomerates of
Stakhovsky, 2011), strength and failure of matter (Carpinteri and these entities. One process forms the entities, a second the
Chiaia, 1997; Carpinteri and Pugno, 2003; Carpinteri et al., 2010; agglomerates. In general, this means that the synchronous or
Paggi and Carpinteri, 2007), fracture energy (Carpinteri and Chiaia, subsequent activity of different processes leads to patterns of
1995), spatial and temporal distribution of earthquakes (Hirata and different fractality on different scales. A special case is represented
Imoto, 1991; Tang et al., 2012), rough surfaces (Herzfeld and by the transition of fractal patterns to Euclidean geometry at very
Overbeck, 1999), or grade of ore deposits (Agterberg, 1997; Ford small scales (discussed in Section 4.1).
and Blenkinsop, 2009). Since a more detailed discussion of the In logelog representations scaling regimes are most easily
topic is beyond the scope of this paper, the reader is referred to the detected by intervals of linear point arrangement with different
cited literature. slopes. In many cases, however, only small variations from linearity
make it difficult to decide if only one or two or several scaling
4.3. Fractality in 3D regimes exist. The discussion goes back to the nineties (Kranz and
Velde, 1994; Velde et al., 1993). Meanwhile, many examples have
Methods of fractal-geometry do exist nearly exclusively for been published that show different fractality on different scales,
1D or 2D. Generally it is assumed that 3D fractality can be extrap- such as fracture patterns (Castaing et al., 1996; Cello, 1997; Nicol
olated from 2D fractality and that the 3D fractal dimension is et al., 1996; Şuţeanu et al., 2000; Volland and Kruhl, 2004), grain
grained by adding 1 to the 2D fractal dimension (Barton, 1995; and grain-fragment sizes (Heilbronner and Keulen, 2006; Keulen
Mandelbrot, 1982). However, there is discussion about whether et al., 2007), crystal sizes and crystal structures (Armienti and
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 11

Tarquini, 2002; Singer and Bilgram, 2006), and clustering of veins Inhomogeneity with respect to scaling behaviour means that
(Manning, 1994). A large number of additional examples are given sub-structures may show highly diverse fractal dimensions or even
by Kaye (1989). On the other hand, in many publications patterns may not be fractal. However, variation of fractal dimension in
are related to only one scaling regime, which could be by no means different parts of the pattern and one single fractal dimension of
related to two or more scaling regimes. the entire pattern do not exclude each other. Henceforth, the term
Mainly two reasons exist for different fractality of rock struc- inhomogeneity is used as inhomogeneity of the scaling behaviour
tures on different scales. (i) The pre-given rock fabric and the rock of a pattern or variation of the fractal dimension in different parts of
composition are different on different scales and therefore the the pattern.
same process acts differently (Flook, 1979; Kaye, 1978, 1989; In a simple way the presence of inhomogeneity can be detected
Şuţeanu and Kruhl, 2002). For example, fracturing of rocks on the by analysing different parts of the pattern separately and
specimen-scale is affected by the rock’s grain fabric. On a larger comparing the results (Gerik and Kruhl, 2009). In more detail,
scale, layering on the dmem scale and variation in lithology may variation of inhomogeneity can be unveiled if the pattern is sepa-
govern the development of fracture patterns. In addition to the rately analysed in an array of sub-regions that cover the pattern
fractal dimensions of the different scaling regimes, the thresholds totally, e.g. boxes of a rectangular network (Lunina et al., 2005). The
between these regimes, which represent thresholds between most detailed picture of inhomogeneity is obtained if a gliding-
regimes of different pattern-forming processes, may be an addi- window (kriging) procedure with wide overlap of windows is
tional important source of information, as pointed out by Şuţeanu applied (Map Counting: Peternell et al., 2010). This method typically
et al. (2000). (ii) Two processes act synchronously or subse- provides a contour map or a pixel map with colour or grey-scale
quently with different intensity on different scales. For example, shading (Figs. 7 and 8). Quantification of inhomogeneity requires
different fragmentation events occur in succession (Keulen et al., large-area and highly resolved patterns which ideally extends over
2007). Fragmentation in a regional stress field can be followed by several orders of magnitude because the fractal-geometry method
fluid-overpressure-related fragmentation on a smaller scale, which has to be applied to each single window. If the window is too small
comminutes the existing fragments and changes the fragment-size the statistics may be too weak. If the window is large the number of
distribution (Volland and Kruhl, 2004). Subsequent diffusion at windows is possibly too small for a detailed and, thus, meaningful
different temperatures leads to suturing of already sutured grain contour or pixel map. Constant fractality of the pattern over the
boundaries and changes the fractality of the grain boundaries on entire range of scale is no precondition for analysing inhomoge-
different scales (Kruhl and Nega, 1996). In addition, different scaling neity. For example, patterns often show two different fractalities on
regimes may result from non-geologic reasons, for example from a larger and on a smaller scale: structural and textural fractality
human activity in mining (Blenkinsop, 1994). (Section 5). The patterns’ inhomogeneities with respect to these
Nevertheless, structures with self-similarity over several orders two fractalities can be determined and presented independently,
of magnitude do exist. Specifically, fractality of fracture patterns however, under the premise that the threshold between the
may be constant from millimetre- to 10-kilometre-scale (Long and
Imber, 2011) and veins thickness distributions may show similar
fractal dimensions, independently of the size range of the veins and
type and structure of host rocks (Kruhl, 1994). The reasons are not
known and it can only be speculated that certain characteristics of
the stress field play a part. Investigations into this topic are desirable.
Variation of scaling properties of structures over increasingly
many and short ranges of scale will eventually lead to size-
dependency of properties. Consequently, these properties cannot
be regarded as material constants, however, they can be studied by
multifractal analysis (Section 4.2).

6. Inhomogeneity

Inhomogeneity is a dominant characteristic of rock structures.


Since inhomogeneity originates from various processes and their
interaction it contains information about these processes. There-
fore, tests for inhomogeneity should be an integral part of rock
structure quantification. Inhomogeneity is also a question of scale.
On small scale rock structures and the resulting patterns tend
towards inhomogeneity and, vice versa, gain homogeneity with
increasing size (Kaye, 1993, p. 132). The main reason is that the
single portions of which the pattern is composed, e.g. crystals, pores,
parts of fractures, layers or veins, usually occur in a specific range of
size. Consequently, the results of pattern quantification close to this
size highly depend on where the line, area or volume of analysis is
placed with respect to the portions of the pattern. It also means that
a pattern becomes inevitably inhomogeneous at its lower resolution
limit. On the other hand, a homogeneous pattern can become Fig. 7. Map Counting exemplified by a crystal distribution pattern (small black or grey
inhomogeneous on larger scale. For example, a fracture pattern may rectangles). (A) The method is performed within a box (solid-line box) and the
be homogeneous in a km-scale outcrop of basement rocks and also resulting fractal dimension Db is plotted in the box centre. Shifting of the box over the
pattern parallel to two orthogonal directions (arrows) results in a rectangular grid of Db
homogeneous in the sedimentary cover, but with different fractal- numbers (B). Based on that grid, a contoured map of fractal dimensions can be
ity. Enlarging the analysis area from the basement into the constructed (C) representing the planar distribution of the pattern’s fractality. Modified
sedimentary rocks would lead to an inhomogeneous pattern. after Peternell et al. (2010).
12 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

structure-forming processes. Quantification of such anisotropic


structures represents an essential tool of structure analysis. Since
deformation processes are generally related to a flow plane and
a flow direction the anisotropy of the resulting structures can be
sufficiently analysed in 2D in a section perpendicular to the flow
plane and parallel to the flow direction.
Different methods are available for determining anisotropy of
structures in 2D (e.g. Launeau and Robin, 1996). The anisotropy of
deformation structures is generally interpreted as strain, i.e., the
analysis leads to a strain ellipse or the directional data are fitted to
an elliptical distribution, respectively. However, these methods are
not designed for capturing the anisotropy of complex rock struc-
tures and the resulting patterns. A method that takes into consid-
eration the connectivity of a line pattern in a far better way was
suggested by Şuţeanu et al. (1995, 1997, 2004): the Angle Distribu-
tion of Percolation Length. The pattern is covered by stripes of

Fig. 8. High-resolution Map Counting exemplified by the distribution pattern of mafic


minerals in a syenite. A gliding-window procedure is applied to an approximately
2.5  2.5 m large area, with 10 cm window size, a gliding distance of 5 mm, and a total
of 254,400 measurements, each of them shown by one colour pixel. The colour
variation of the map represents the variation of the box-counting dimension, i.e., the
variation of the pattern’s fractality. Modified after Peternell et al. (2011).

different scaling regimes is approximately the same for all analysed


parts of the pattern.
Inhomogeneity of rock structures has been quantified in relation
to crystal-distribution patterns (Kruhl et al., 2004; Peternell et al.,
2003, 2010, 2011), fracture patterns (Gerik and Kruhl, 2009;
Lunina et al., 2005), or grain sizes (D-of-grain-size mapping e
Heilbronner and Keulen, 2006). The latter implies the analysis of
the 2D distribution of a data set.
Usually, pattern inhomogeneity is determined in 2D. However,
determination of linear inhomogeneity is possible, e.g. by the
Cantor-Dust Method, and is performed in relation to anisotropy
quantification (Section 7). In addition, the variation of the fractal
dimension D is measured along profiles, i.e., D is determined
independently for single profile sections, for example along sea-
floor bathymetric profiles (Barenblatt et al., 1984).
Manual quantification of highly resolved patterns over large
areas is extremely time-consuming and, therefore, automation is
a pre-requisite for wide applicability of the methods (automated
Map Counting: Peternell et al., 2011; Fig. 8). So far only few auto-
mated quantifications of inhomogeneity have been performed,
probably, because natural patterns have to be recorded with high
resolution. This often requires measurements of structures in
numerous neighbouring sub-areas and merging of these areas with
additional corrections (Peternell, 2007; Peternell et al., 2011).
Nevertheless, quantification of complex patterns’ inhomogeneity
represents an important analytical technique, specifically if applied
to large areas. Based on such analyses, variations of rock properties
and structure-forming processes have been detected, specifically in
syntectonic magmatic rocks with diffuse and large-scale structures
that can rarely be measured by classical techniques (Peternell et al., Fig. 9. Application of the modified Cantor-Dust Method to a fracture pattern. (A)
2011). Fragmented rock: grey, quartz filling: white. Parallel scan-lines cover the pattern in
azimuthal orientations of 10 steps. Along the scan-lines the rock fragments are
segmented (bold lines). For each direction the cumulative frequencies of these
7. Anisotropy segments are presented in a logelog diagram (Richardson plot) leading to a fractal
dimension. (B) This fractal dimension is plotted from a centre towards the outside. In
the present case an ellipse can be fitted to the results, however with locally clear
As many geological processes, deformation is a flow process and variation. Given are the ellipse axes a and b, their ratio (AAD: azimuthal anisotropy of
thus directional. Consequently, deformation leads to anisotropic fractal dimension D), the inclination alpha versus a reference line, and the correlation
structures that may provide important information about coefficient R. Modified after Volland and Kruhl (2004).
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 13

certain thickness and extension, with various azimuthal orienta- anisotropy of complex patterns can be quantified and analysed in
tions. Within each of theses stripes the percolation length of the detail. The amount of data can be increased in two ways. (i) The
pattern is determined in different positions and added up. The number of parallel scan-lines is increased and their spacing is
procedure results in a directional total percolation length that decreased, e.g., down to 1 pixel. That means one-pixel thick scan-
represents a specific type of pattern anisotropy. lines cover half of the measurement area. This enhances the
Fractal-geometry methods usually determine the overall statistics and, consequently, the significance of the fractal analysis
complexity of a pattern but have to be modified for anisotropy for each direction. (ii) The number of directions is increased. For
quantification. In the simplest case anisotropy of complexity in 2D example, variation of azimuthal orientations in 1 steps is possible
is determined by the different fractal dimensions of two orthogonal without problems. Of course, for such resolution the pattern’s
directions (Zhao et al., 2009). Such determination is suitable for resolution has to be high enough to avoid the effect of pixel
structures with orthogonal components of geometry. In general,
application of fractal-geometry methods to simple structures, e.g.,
two orthogonal sets of planar fractures, is problematic and the sub-
sets of the simple pattern are better studied independently (Harris
et al., 1991). That also points to the fact that simple, ultimately non-
fractal, patterns are not suitable for fractal analyses.
Complex structures, however, call for fractal-geometry analyses
that are related to numerous and variable directions. During the
last two decades such analyses were performed manually, typically
with various azimuthal orientations of scan-lines, however in 10
steps as a minimum, due to the exceedingly high expenditure of
time (Fig. 9). Mainly the distribution of fractures or fracture spacing
along scan-lines were measured and analysed by the Cantor-Dust
(or Cantor-Set) Method (Babadagli, 2002; Harris et al., 1991;
Ledésert et al., 1993a,b; Pérez-López et al., 2005; Pérez-López and
Paredes, 2006; Velde et al., 1990, 1991; Volland and Kruhl, 2004).
As another possibility, along scan-lines a structure can be dis-
aggregated into complex linear components that are quantified
separately, e.g., by the Directional Structured-Walk Method (Kruhl
et al., 2004; Fig. 10). The result however can be influenced by the
type of disaggregation.
Only automated methods (Gerik and Kruhl, 2009; Gerik et al.,
2010; Pérez-López et al., 2005) produce such large data sets that

Fig. 11. High-resolution quantification of pattern anisotropy by an automated modified


Cantor-Dust method. (A) Within a fracture pattern a circular area of analysis is
determined. (B) Sets of parallel scan-lines that cover the pattern in rotational steps of
1 generate segments of the fracture lines. The cumulative frequency distributions of
these segments lead to fractal dimensions that are plotted from a centre towards the
Fig. 10. Principle of the Directional Structured-Walk Method. (A) A complex line pattern outside. The resulting point distribution can be only roughly fitted by an ellipse. (C)
is covered by sets of parallel scan-lines in azimuthal orientations of (in this case) 10 Segmentation of the rock fragments (rather than fractures) by the same sets of
steps. (B) Along these lines, based on a specific instruction, curves are constructed, scan-lines results in a quite different arrangement of measurement points with a much
which represent the directional ‘permeability’ of the pattern. (C) Each curve is better elliptical fit, even though ellipticities and orientations of the elliptical fits are
quantified by the Structured-Walk Method. Modified after Kruhl et al. (2004). similar. Modified after Gerik and Kruhl (2009).
14 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

discretization. Thereby the directional variation of pattern the pattern’s directional anisotropy can be quantified by 1D fractal-
complexity can be relatively precisely measured, leading to a highly geometry methods, e.g. the Cantor-Dust method. The resulting
resolved picture of pattern anisotropy, which cannot be fitted by directional fractal dimensions span a, probably complex, 3D surface
simple geometric forms, e.g. an ellipse, but represents a much more that represents the 3D anisotropy of the pattern’s scaling behav-
complex geometry that may reflect dominant orientations of iour. (iii) Conceivable, but currently unrealized, are methods, e.g.
pattern fractions (Fig. 11; Gerik and Kruhl, 2009; Gerik et al., 2010). a modified 3D Box-Counting method, which directly measure the
The results of such anisotropy analysis can vary if the ‘background’ pattern’s anisotropy in 3D.
is analysed, rather than the pattern, e.g. the line drawing of a frac- The combination of automated recording methods and an
ture system (Fig. 11B, C). Since the two complementary parts of automated fractal-geometry technique is the indispensable pre-
a ‘complete’ pattern have independent geological meaning, quan- requisite of an effective and fast analysis of rock properties, with
tifying both of them is advisable. Strictly seen, the geometry of the high potential of utilization (Baker et al., 2008). Determination of
data-point distribution does not represent a simple structure inhomogeneity and anisotropy can be combined (Fig. 6H), leading
property, e.g. strain, but represents the anisotropy of structure to a quantification of inhomogeneity of anisotropy (Peternell, 2007;
complexity or, in other words, the directional scaling behaviour of Fig. 12). Such combination requires large and highly resolved
the structure. patterns but provides information that is not gained by single
With respect to anisotropy quantification and analysis, three methods. The strength of these combined analyses is based on the
aspects should be considered. (i) Irregular areas of analysis lead to fact that substructures of larger patterns can be visualized, which
different lengths of scan-lines in different directions and, conse- are not visible in the field. For example, based on a combined
quently, to a different statistical basis for different directions, which inhomogeneity and anisotropy quantification of diffuse crystal
can falsify the overall result of anisotropy quantification. Therefore, distribution patterns in magmatic rocks, magmatic shear structures
the analysis should be performed within a circle with its centre as or lineations can be detected, which provide information about
rotation axis of the scan-lines. (ii) If the outline of the pattern is flow processes in the magma (Peternell et al., 2011).
strongly irregular reduction to a circle highly reduces the area of
analysis and can weaken the statistical basis of analysis. This should
be kept in mind when choosing the area of analysis. (iii) Depending 8. Second-order fractal dimension
on their thickness the scan-lines cover the area close to the rotation
axis more strongly than further away thus increasing the weight of Fractal-geometry methods, applied to 2D or 3D patterns mainly
the pattern in this area. This problem vanishes if the pattern is in combination with inhomogeneity and anisotropy quantification,
homogeneous in the analysed circular area and, above all, if the may lead to 1D or 2D data sets that can be fractal again and ana-
scan-lines totally cover the area of analysis. lysed by 1D or 2D methods. For example, the anisotropy quantifi-
The 3D measurement of structures with the aid of tomographic cation of a crystal distribution pattern in a syntectonic magmatic
methods (neutron or X-ray; Section 3.4) allows 3D quantification of rock results in a point distribution of fractal dimension, which itself
anisotropy. (i) Oriented sections can be cut through a 3D structure is complex and can be quantified by the Structured-Walk or Box-
and the 2D patterns on these sections can be quantified by fractal- Counting method (Fig. 13). Even though the circular and point-
geometry methods, e.g. Box Counting. However, the results cannot symmetric distribution of azimuthal fractal dimensions can be
be easily assembled to a 3D picture of anisotropy. (ii) In the same approximated by an ellipse of certain axial ratio, which gives
way, oriented scan-lines can be placed through a 3D structure and a rough impression of the intensity of anisotropy, the variation is

Fig. 12. High-resolution Mapping of Rock Fabric Anisotropy (MORFA e Peternell et al., 2011). A combination of a modified Cantor-Dust Method and gliding window procedure is
applied to the distribution pattern of mafic minerals in a syenite. A circle with diameter d ¼ 20 cm is shifted over the approximately 1  4 m sized pattern in steps of 5 cm. The
anisotropy intensities are represented by colour-indexed pixels, varying between 1 and roughly 1.7. A value of 1 indicates isotropy of the pattern’s complexity. The black bars
represent the direction of lowest fractal dimension. This direction represents the approximate direction of long grain axes, as shown in the sample photograph. 1804 measurements.
Modified after Peternell et al. (2011).
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 15

(Bornyakov et al., 2006; Sornette et al., 1990a). The self-similarity


has been determined from micro- to km-scale (Jensen et al.,
2011; Korvin, 1992; Long and Imber, 2011; Turcotte, 1997; Vignes-
Adler et al., 1991). However, exemplified by the Red-Sea rifting,
Castaing et al. (1996) state that fracture patterns show different
scaling properties from the centimetre scale up to the 1000-km
scale, which are controlled by the thickness of different mechan-
ical layers that constitute the crust at different scales. That is in
agreement with the fact that rocks exhibit various compositions
and structures on different scales, which affect the fragmentation
behaviour.
Numerous studies are related to pattern quantification of faults,
fractures or joints and provide a wide range of fractal dimensions:
Bonnet et al. (2001), Bour et al. (2002), Cello (1997), Cello et al.
(2000), Hatton et al. (1993), Hirata (1989), Hirata et al. (1987),
Kagan (1991), Kulatilake et al. (1997), Mecholsky and Mackin
Fig. 13. Fractality, with fractal box-counting as well as ruler dimension D ¼ 1.281, of
a complex curve generated by the anisotropy quantification of a fracture pattern as (1988), Nagahama and Yoshii (1993), Park et al. (2010), Piggott,
exemplified in Fig. 11B. 1997; Sornette et al. (1990b), Zhao et al. (1990). Roughness of
rock and fracture surfaces has been frequently quantified (Bahat,
1991; Brown, 1987; Dougan et al., 2000; Power and Tullis, 1991;
not random and potentially bears information about the directional Schmittbuhl et al., 1993), but also of surfaces of gravel and sediment
variation of fractality. particles (Drolon et al., 2003; Qin et al., 2012), pores (Wong and Lin,
The inhomogeneity of fractality can be treated in a similar way. 1988), tektites (Rantzsch et al., 2012) and in technical material
The quantification of inhomogeneity of a crystal distribution (Chen et al., 1997; Mandelbrot et al., 1984; Milman et al., 1994).
pattern by Map Counting results in a distribution map of fractal Baker et al. (2008) established an automated method for quanti-
dimensions, which itself can be analysed by Box Counting or other fying surface roughness of rock faces. Fracture roughness is corre-
methods like topography and, again, leads to a sequence of fractal lated with fracture toughness and tensile strength (Carpinteri and
dimensions (Fig. 14). Puzzi, 2005), with mechanical behaviour (Mechtcherine, 2009;
However, the extent to which it is advantageous to condense Mohsen et al., 2003), with energy of technical processes (Brosch
complex data sets or data point arrangements to single numbers is et al., 1992), or is seen as a function of crack propagation speed
a matter of discussion. On the one hand, one number is gained, (Milman et al., 1994; Sharon and Fineberg, 1999) or of crack front
which represents the complexity of anisotropy or inhomogeneity of positions (Horovistiz et al., 2010). The anisotropy of the fractality of
a pattern, on the other hand, information is lost, which can be fracture roughness has been investigated (Bahat, 1991; Baker et al.,
possibly drawn from such complexity. 2008) and related to anisotropy of fracture propagation (Dougan et
al., 2000). Schmittbuhl et al. (1995) show the self-affinity of rough
9. Material properties and pattern-forming processes fracture surfaces and that the self-affine component is similar in
different rocks and they argue for universality in the cracking
Fractal-Geometry-based investigations on material properties process. Fractal dimensions of molten pseudotachylite surfaces, due
have been performed in virtually all fields of geosciences. Specific to frictional heating, decrease with increasing heat-production rate
attention was paid to fractures in rocks, which influence numerous (Hirose and Shimamoto, 2003).
rock properties and, last but not least, can be transferred relatively In general, particle size distributions in fault rocks show varia-
easily from nature to binary patterns. Properties of scaling and of tion of fractal dimensions, interpreted as resulting from different
pattern inhomogeneity and anisotropy of various rock structures fragmentation processes. Low values of w1.97 are related to
have been determined and related to (mainly) physical properties a “pillar of strength” model (Allègre et al., 1982), medium values of
of rocks. The fragmentation of the lithosphere is regarded as a self- w2.58 to the “constrained comminution” model (Sammis et al.,
organized critical phenomenon that leads to dissipative structures, 1987), and high values of w2.84 to the “plane of fragility” model
i.e., to self-similar fracture patterns that evolve by self-organization (Turcotte, 1986a). See Blenkinsop (1991) for review. It can be

Fig. 14. Fractality, with fractal box-counting dimension D ¼ 1.49, of a complex binary pixel pattern generated by the inhomogeneity quantification (Map-Counting Method) of part of
the mineral distribution pattern shown in Fig. 8.
16 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

expected that fragmentation processes vary spatially and tempo- size distribution (Şuţeanu et al., 2000), which potentially bears
rally and that specifically in fault zones fractal dimensions of useful information about the fragmentation process.
particle-size distributions change according to these processes, as Fragment morphologies in hydrothermal breccias may reflect
observed and suggested by Billi and Storti (2004). However, frag- the degree of maturity and the nature of fluiderock interaction
ment size distributions may not always be fractal. Log-normal (Lorilleux et al., 2002). The degree of deformation in mylonites is
distributions have been described as a result of fracturing under reflected by the irregular outlines of quartz grains (Wang et al.,
extensional stress, in contrast to fractal distributions resulting from 2007). Fractal packing and particle shapes influence the mechan-
fragmentation under compressional stress (Korvin, 1989). More- ical properties of building materials (Hecht, 2000) as the grain size
over, the morphology of irregular-shaped particles can be quanti- of a rock may influence the fragmentation process and the spatial
fied and the results replace typical grain-morphology tables for distribution of microcracks. In this way the spatial distribution of
visual comparison (Orford and Whalley, 1983, 1987). acoustic emission hypocentres has been interpreted (Kusunose
The fractality of various rock structures is correlated with et al., 1991).
numerous rock properties or conditions of formation. Scaling In general, quantification of fracture patterns and fractal size
properties of joint, vein, fault or fracture patterns are related to distributions and shapes of particles are increasingly important in
various parameters: to lithostatic pressure (Gillespie et al., 2001), investigations on seismicity (Allègre et al., 1982; Aviles et al., 1987),
confining pressure in experiment (Velde et al., 1993), strain stress fields (Zhao, 1998; Zhao et al., 2009), rock strength and
(Hadizadeh and Johnson, 2003), stress (Pérez-López et al., 2005; deformability (Feng et al., 2009; Zhang and Sanderson, 1994),
Zhao, 1998), intensity of shearing (Volland and Kruhl, 2004), temperature and strain rate of metamorphism (Kruhl and Nega,
connectivity and percolation (Feranie et al., 2011; Fomin et al., 1996; Mamtani, 2010; Takahashi et al., 1998), geothermal energy
2003; Ledésert et al., 1993a, 2010; Nakaya and Nakamura, 2007; (Ledésert et al., 1993b, 2010), hydraulic properties (Barthélémy
Nolte et al., 1989; Park et al., 2010; Roberts et al., 1999), fluid flow et al., 1996; Main et al., 2000; Micarelli et al., 2006; Rawls and
(Sanderson and Zhang, 1999) and to the distribution of metals and Brakensiek, 1995; Rawls et al., 1993; Shepherd, 1989; Wibberley
ore deposits (Carlson, 1991; Blenkinsop, 1994, 1995; Blenkinsop and and Shimamoto, 2003), magmatic systems (Perugini and Poli,
Sanderson, 1999; Ford and Blenkinsop, 2008a,b; 2009; Kreuzer 2000), and on explosive volcanism (Kueppers et al., 2006;
et al., 2007; Weinberg et al., 2004). Fractality of pore-space is Perugini et al., 2011; Perugini and Kueppers, 2012). They are used in
correlated with electric properties of rocks (Da Rocha and Habashy, cellular-automata modelling for distribution of seismicity (Tejedor
1997) as well as with diagenetic processes (Krohn, 1988). In addi- et al., 2010) or, based on spatial distribution of acoustic emission or
tion, the fractality of rock structures, such as fault networks, microseismicity, provide the possibility of predicting rock failure
roughness or particles sizes, is used in numerical studies and (Yuan and Li, 2009) or rock-burst in underground mining (Xie and
modelling (Bistacchi et al., 2011; Bour and Davy, 1997; Chilès, 1988; Pariseau, 1993).
Mair and Abe, 2011; Yfantis et al., 1988).
Fracture patterns may reflect processes of fragmentation, for
example, ballistic or in a slowly increasing stress field (Kaye, 1989, 10. Conclusions
1994). The power-law relationship of fault-size distributions has
been used for sub-seismic (sub-resolution) fault prediction (1) Fractal geometry techniques represent attractive tools for
(Watterson et al., 1996). Mareschal (1989) suggests to extrapolate studying complex rock structures because they provide
seafloor topography into 2D, based on profile measurements, as well a symbolism for quantifying the poorly defined geometries and
as to interpolate to smaller scales. However, even if the power-law patterns of these structures. Such quantification may lead to
relationship of structures forms a basis for upscaling or down- valuable information not only about material properties but
scaling their geometric features, caution is needed since scaling particularly about structure-forming processes, and it facili-
properties of structures may differ from one to the other scale. tates comparison with results of experiments and technical
As shown by Blenkinsop (1991), particle or fragment-size processes. Quantification builds bridges between nature,
distributions generated during fragmentation are generally fractal experiment and model.
and depend on various factors: the specific type of fragmentation (2) Geological structures are often characterized by being different
process and the number of events, the energy input, confining (i) on different scales (pattern scaling), (ii) in different areas
pressure, selective fracturing of larger particles and, in certain (pattern inhomogeneity), and (iii) in different directions
situations, alteration of particles. For example, fractal particle-size (pattern anisotropy). These three structure or pattern properties
distributions have been related to frictional properties and ener- are geologically significant and often reflect various geological
getics of gouge (Biegel et al., 1989; Morgan and Boettcher, 1999; processes. Quantification of inhomogeneity and anisotropy
Mair et al., 2002; Wilson et al., 2005), to energy density for frac- needs modifications of ‘classical’ fractal-geometry methods. The
turing (Nagahama, 1993), to slip along faults (Monzawa and Otsuki, currently available tools are still limited.
2003) and are seen as a result of comminution where neighbouring (3) Additional advantages of fractal-geometry techniques are (i)
particles of the same size do not exist but one of them will fragment interpolation in order to ‘complete’ patterns that are based on
to smaller pieces (An and Sammis, 1994; Imre et al., 2010; Sammis only few data, and (ii) extrapolation from one scale to another.
et al., 1987; Steacy and Sammis, 1991). However, experiments of However, this only works if the patterns are homogeneous and
shock-related fragmentation demonstrated that (i) fragments may show the same fractality on different scales. In many cases
cluster on dominant size intervals and (ii) neighbouring fragments these conditions are not fulfilled or not proven.
show a high probability to be of similar size (Şuţeanu et al., 2000). (4) Results of fractal-geometry methods often show fluctuations.
These authors point to the fact that, in addition to a power-law It might be worth regarding them not as ‘inaccuracies’ or
relationship, many fragment-size distributions are hierarchically ‘errors’ but search for reasons that are linked to the measured
organized, with multi-modal distributions and ‘preferred dimen- pattern. In addition to different scaling behaviour on different
sions’ (Sadovskii et al., 1984; Sadovskii, 1986; cited in Şuţeanu et al., (partly uncomfortably limited) scales, cyclic deviations
2000). Obviously, particle sizes of fragmented material not only from power-low distribution may indicate regularity of the
show power-law distributions but may cluster on specific size pattern, caused by variations or interactions of pattern-forming
intervals. Such clustering points towards a fractal character of the processes.
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 17

(5) Future steps in developing and applying fractal-geometry Barenblatt, G.I., Zhivago, A.V., Neprochnov, Yu.P., Ostrovskyi, A.A., 1984. The fractal
dimension: a quantitative characteristic of ocean bottom relief. Oceanology 24,
methodology, related to analysis of rock structures, should
695e697.
include (i) establishing new methods for quantification of Barnsley, M., 1988. Fractals Everywhere. Academic Press.
anisotropy and inhomogeneity, because of the importance of Barraud, J., 2006. The use of watershed segmentation and GIS software for textural
these rock properties, (ii) increased application to structures in analysis of thin sections. Journal of Volcanology and Geothermal Research 154,
17e33.
3D, in order to keep up with the advancing development of 3D Barthélémy, P., Jacquin, C., Yao, J., Thovert, J.F., Adler, P.M., 1996. Hierarchical
recording techniques, (iii) further automation of methods, structures and hydraulic properties of a fractures network in the Causse of
which would specifically allow quantification of large-scale Larzac. Journal of Hydrology 187, 237e258.
Barton, C.C., 1995. Fractal analysis of scaling and spatial clustering of fractures. In:
patterns in high resolution, as a firm basis for successful anal- Barton, C.C., La Pointe, P.R. (Eds.), Fractals in the Earth Sciences. Plenum Press,
yses of complex rock structures, and (iv) more integral studies New York, pp. 141e178.
on naturally and experimentally generated structures and Baykan, N.A., Yilmaz, N., 2010. Mineral identification using color spaces and
artificial neural networks. Computers & Geosciences 36, 91e97.
simulations, where quantification would provide the necessary Bérubé, D., Jébrak, M., 1999. High precision boundary fractal analysis for shape
links for better understanding natural processes based on characterization. Computers & Geosciences 25, 1059e1071.
experiments and simulations. This would be also one good Biegel, R.L., Sammis, C.G., Dieterich, J., 1989. The frictional properties of a simulated
gouge having a fractal particle distribution. Journal of Structural Geology 11,
answer to the frequently asked question “what’s the use of it?”.
827e846.
Billi, A., 2005. Grain size distribution and thickness of breccia and gouge zones from
thin (<1 m) strike-slip fault cores in limestone. Journal of Structural Geology 27,
Acknowledgements 1823e1837.
Billi, A., 2007. On the extent of size range and power law scaling for particles of
natural carbonate fault cores. Journal of Structural Geology 29, 1512e1521.
Thanks are due to Mark Peternell, Axel Gerik and Sabine Volland Billi, A., Storti, F., 2004. Fractal distribution of particle size in carbonate cataclastic
for years of fruitful cooperation, resulting in many findings pre- rocks from the core of a regional strike-slip fault zone. Tectonophysics 384,
sented in this paper, and to Cristian Şuţeanu and Diego Perugini for 115e128.
Bistacchi, A., Griffith, W.A., Smith, S.A.F., Di Toro, G., Jones, R., Nielsen, S., 2011. Fault
their thoughtful comments on the manuscript. Katharina Döhler roughness at seismogenic depth from LIDAR and photogrammetric analysis.
generated the data sets for Figs. 5 and 13. Particularly, I am grateful Pure and Applied Geophysics 168, 2345e2363.
to Tom Blenkinsop who suggested and stimulated the present work Blenkinsop, T.G., 1989. Thickness-displacement relationships for deformation
zones: discussion. Journal of Structural Geology 11, 1053e1054.
and provided hospitality, privately and at James Cook University,
Blenkinsop, T.G., 1991. Cataclasis and processes of particle size reduction. Pure and
during several stays in Townsville. One of these stays was finan- Applied Geophysics 136, 59e86.
cially supported by an alumni grant of the Alexander von Humboldt Blenkinsop, T.G., 1994. The fractal distribution of gold mineralisation: two examples
from the Zimbabwe Archaean craton. In: Kruhl, J.H. (Ed.), Fractal and Dynamic
Foundation.
Systems in Geoscience. Springer, Berlin/Heidelberg, pp. 247e258.
Blenkinsop, T.G., 1995. Fractal measures for size and spatial distributions of gold
mines: economic applications. In: Blenkinsop, T.G., Tromp, P.L. (Eds.), Sub-
References saharan Economic Geology. Geological Society of Zimbabwe, Special Publication
3, pp. 177e186.
Ackermann, R.V., Schlische, R.W., Withjack, M.O., 2001. The geometric and Blenkinsop, T.G., Fernandes, T.R.C., 2000. Fractal characterization of particle
statistical evolution of normal fault systems: an experimental study of the distributions in chromites from the Great Dyke, Zimbabwe. In: Blenkinsop, T.G.,
effects of mechanical layer thickness on scaling laws. Journal of Structural Kruhl, J.H., Kupková, M. (Eds.), Fractals and Dynamic Systems in Geoscience.
Geology 23, 1803e1819. Pure and Applied Geophysics, vol. 157. Birkhäuser, Basel, pp. 505e521 (Topical
Adams, B.L., Wright, S.I., Kunze, K., 1993. Orientation imaging: the emergence of Issue).
a new microscopy. Metallurgical Transactions 24A, 819e831. Blenkinsop, T.G., Sanderson, D.J., 1999. Are gold deposits in the crust fractals? A
Agterberg, F.P., 1997. Multifractal modelling of the sizes and grades of giant and study of gold mines in the Zimbabwe craton. In: McCaffrey, K.J.W., et al. (Eds.),
supergiant deposits. Global Tectonics and Metallogeny 6, 131e136. Fractures, Fluid Flow and Mineralization. Geological Society London Special
Agterberg, F.P., Cheng, Q., Brown, A., Good, D., 1996. Multifractal modeling of Publications 155, pp. 141e151.
fractures in the Lac Du Bonnet Batholith, Manitoba. Computers & Geosciences Bonnet, E., Bour, O., Odling, N.E., Davy, P., Main, I.G., Cowie, P.A., Berkowitz, B., 2001.
22, 497e507. Scaling of fracture systems in geological media. Reviews of Geophysics 39, 347e383.
Allègre, C.J., Le Mouel, J.L., Provost, A., 1982. Scaling rules in rock fracture and Bornyakov, S.A., Truskov, V.A., Cheremnykh, A.V., 2006. Dissipative structures in
possible implications for earthquake prediction. Nature 297, 47e49. fault zones and their diagnostic criteria (from physical modeling data). Russian
Amidzic, O., Riehle, H.J., Fehr, T., Wienbruch, C., Elbert, T., 2001. Pattern of focal Geology and Geophysics 49, 138e143.
g-bursts in chess players. Nature 412, 603. Bour, O., Davy, P., 1997. Connectivity of random fault networks following a power
An, L.-J., Sammis, C., 1994. Particle size distribution of cataclastic fault law fault length distribution. Water Resources Research 33, 1567e1583.
materials from southern California: a 3-D study. Pure and Applied Geophysics Bour, O., Davy, P., 1998. On the connectivity of three-dimensional fault networks.
143, 203e227. Water Resources Research 34, 2611e2622.
André-Mayer, A.-S., Sausse, J., 2007. Thickness and spatial distribution of veins in Bour, O., Davy, P., Darcel, C., Odling, N., 2002. A statistical scaling model for fracture
a porphyry copper deposit, Rosia Poieni, Romania. Journal of Structural Geology network geometry, with validation on a multiscale mapping of a joint network
29, 1695e1708. (Hornelen Basin, Norway). Journal of Geophysical Research 107 (B6), 2113.
Andriani, G.F., Walsh, N., 2002. Physical properties and textural parameters of Brooks, B.A., Manning, C.E., 1994. Fractal clustering of metamorphic veins: comment
calcarenitic rocks: qualitative and quantitative evaluations. Engineering and reply. Geology 22, 1147e1149.
Geology 67, 5e15. Brosch, F.J., Pölsler, P., Riedmüller, G.,1992. The use of fractal dimension in engineering
Armienti, P., Tarquini, S., 2002. Power law olivine crystal size distributions in geology. Bulletin of Engineering Geology and the Environment 45, 83e88.
lithospheric mantle xenoliths. Lithos 65, 273e285. Brown, S.R., 1987. A note on the description of surface roughness using fractal
Ashby, R.W., 1956. An Introduction to Cybernetics. Chapman & Hall, London. dimension. Geophysical Research Letters 14, 1095e1098.
Aviles, S.R., Scholz, C.H., Boatwright, J., 1987. Fractal analysis applied to Bunde, A., Havlin, S., 1994. Fractals in Science. Springer, Berlin/Heidelberg/New
characteristic segments of the San Andreas fault. Journal of Geophysical York.
Research 92, 331e334. Carlson, C.A., 1991. Spatial distribution of ore deposits. Geology 19, 111e114.
Avnir, D., Biham, O., Lidar, D., Malcai, O., 1998. Is the geometry of nature fractal? Carpinteri, A., Chiaia, B., 1995. Multifractal nature of concrete fracture surfaces and
Science 279 (5347), 39e40. size effects on nominal fracture energy. Materials and Structures 28, 435e443.
Ayunova, O.D., Kalush, Yu.A., Loginov, V.M., 2007. Relationship of the seismic Carpinteri, A., Chiaia, B., 1997. Multifractal scaling laws in the breaking behaviour of
activity of the Tuvinian and adjacent Mongolian areas with the fractal disordered materials. Chaos, Solitons & Fractals 8, 135e150.
dimensionality of a fault system. Russian Geology and Geophysics 48, 593e597. Carpinteri, A., Pugno, N., 2003. A multifractal comminution approach for drilling
Babadagli, T., 2002. Scanline method to determine the fractal nature of 2-D fracture scaling laws. Powder Technology 131, 93e98.
networks. Mathematical Geology 34, 647e670. Carpinteri, A., Puzzi, S., 2005. The crack surface anomalous scaling and its connec-
Baccar, M., Gee, L.A., Gonzalez, R.C., Abidi, M.A., 1996. Segmentation of range tion with the size-scale effects. International Journal of Fracture 133, 43e60.
images via data fusion and morphological watersheds. Pattern Recognition 29, Carpinteri, A., Spagnoli, A., Vantadori, S., 2010. A multifractal analysis of fatigue
1673e1687. crack growth and its application to concrete. Engineering Fracture Mechanics
Bahat, D., 1991. Tectonofractography. Springer, Berlin/Heidelberg/New York. 77, 974e984.
Baker, B.R., Gessner, K., Holden, E.J., Squelch, A.P., 2008. Automatic detection of Castaing, C., Halawani, M.A., Gervais, F., Chilès, J.P., Genter, A., Bourgine, B.,
anisotropic features on rock surfaces. Geosphere 4, 418e428. Ouillon, G., Brosse, J.M., Martin, P., Genna, A., Janjou, D., 1996. Scaling
18 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

relationships in intraplate fracture systems related to Red Sea rifting. Fueten, F., Mason, J., 2007. An artificial neural net assisted approach to editing edges
Tectonophysics 261, 291e314. in petrographic images collected with the rotating polarizer stage. Computers &
Cello, G., 1997. Fractal analysis of a Quaternary fault array in the central Apennines, Geosciences 33, 1176e1188.
Italy. Journal of Structural Geology 19, 945e953. Gerik, A., Kruhl, J.H., 2009. Towards automated pattern quantification: time-
Cello, G., Gambini, R., Mazzoli, S., Read, A., Tondi, E., Zucconi, V., 2000. Fault zone efficient assessment of anisotropy of 2D pattern with AMOCADO. Computers
characteristics and scaling properties of the Val d’Agri Fault System (Southern & Geosciences 35 (6), 1087e1097.
Apennines, Italy). Journal of Geodynamics 29, 293e307. Gerik, A., Kruhl, J.H., Caggianelli, A., 2010. Quantification of flow patterns in sheared
Chase, W.G., Simon, H.A., 1973. Perception in chess. Cognitive Psychology 4, 55e81. tonalite crystal-melt mush: application of fractal-geometry methods. Journal
Chen, Z., Mecholsky, J.J., Joseph Jr., T., Beatty, C.J., 1997. The fractal geometry of Si3N4 Geological Society of India 75, 210e224.
wear and fracture surfaces. Journal of Materials Science 32, 6317e6323. Gillespie, P.A., Howard, C.B., Walsh, J.J., Watterson, J., 1993. Measurement and
Cheng, Q., 1997. Discrete multifractals. Mathematical Geology 29, 245e266. characterization of spatial distributions of fractures. Tectonophysics 226,
Cheng, Q., Agterberg, F.P., 1996. Multifractal modeling and spatial statistics. 131e141.
Mathematical Geology 28, 1e16. Gillespie, P.A., Johnston, J.D., Loriga, M.A., McCaffrey, K.J.W., Walsh, J.J., Watterson, J.,
Chilès, J.P., 1988. Fractal and geostatistical methods for modeling of a fracture 1999. Influence of layering on vein systematics in line samples. In:
network. Mathematical Geology 20, 631e654. McCaffrey, K.J.W., Lonergan, L., Wilkinson, J.J. (Eds.), Fractures, Fluid Flow and
Choudhury, R.K., Meere, P.A., Mulchrone, K.F., 2006. Automated grain boundary Mineralization. Geological Society London Special Publications 155, pp. 35e56.
detection by CASRG. Journal of Structural Geology 28, 363e375. Gillespie, P.A., Walsh, J.J., Watterson, J., Bonson, C.G., Manzocchi, T., 2001. Scaling
Clark, N.N., 1986. Three techniques for implementing digital fractal analysis of relationships of joint and vein arrays from The Burren, Co. Clare, Ireland. Journal
particle shape. Powder Technology 46, 45e52. of Structural Geology 23, 183e201.
Clark, M.B., Brantley, S.L., Fisher, D.M., 1995. Power-low vein thickness distributions Gonzato, G., 1998. A practical implementation of the box counting algorithm.
and positive feedback in vein growth. Geology 23, 975e978. Computers & Geosciences 24, 95e100.
Coster, M., Chermant, J.-L., 2001. Image analysis and mathematical morphology for Gonzato, G., Mulargia, F., Marzocchi, W., 1998. Practical application of fractal anal-
civil engineering materials. Cement and Concrete Composites 23, 133e151. ysis: problems and solutions. Geophysical Journal International 132, 275e282.
Da Rocha, B.R.P., Habashy, T.M., 1997. Fractal geometry, porosity and complex Gonzato, G., Mulargia, F., Ciccotti, M., 2000. Measuring the fractal dimensions of
resistivity: from rough pore interfaces to hand specimens. In: Geological Society ideal and actual objects: implications for application in geology and geophysics.
London Special Publications 122, pp. 277e286. Geophysical Journal International 142, 108e116.
Davy, P., 1993. On the frequency-length distribution of the San Andreas Fault Goodchild, M.F., 1980. Fractal and the accuracy of geographical measures. Journal of
system. Journal of Geophysical Research 98, 12.141e12.151. Mathematical Geology 12, 85e98.
Davy, P., Sornette, A., Sornette, D., 1990. Some consequences of a proposed fractal Goodchild, J.S., Fueten, F., 1998. Edge detection in petrographic images using the
nature of continental faulting. Nature 348, 56e58. rotating polarizer stage. Computers & Geosciences 24, 745e751.
Davy, P., Sornette, A., Sornette, D., 1992. Experimental discovery of scaling laws Grassberger, P., Procaccia, I., 1983. Characterization of strange attractor. Physical
relating fractal dimensions and the length distribution exponent of fault Review Letters 50, 346e349.
systems. Geophysical Research Letters 19, 361e363. Hadizadeh, J., Johnson, W., 2003. Estimating local strain due to comminution in
Dawers, N.H., Anders, M.H., Scholz, C.H., 1993. Growth of normal faults: experimental cataclastic textures. Journal of Structural Geology 25, 1973e1979.
displacement-length scaling. Geology 21, 1107e1110. Hamdi, E., 2008. A fractal description of simulated 3d discontinuity networks. Rock
Denison, C., Carlson, W.D., Ketcham, R.A., 1997. Three-dimensional quantitative Mechanics and Rock Engineering 41, 587e599.
textural analysis of metamorphic rocks using high-resolution computed X-ray Harris, C., Franssen, R., Loosveld, R., 1991. Fractal analysis of fractures in rocks: the
tomography: part I: methods and techniques. Journal of Metamorphic Geology Cantor’s Dust method e comment. Tectonophysics 198, 107e111.
15, 29e44. Hasting, H.M., Sugihara, H., 1993. Fractals: A User’s Guide for the Natural Sciences.
Dougan, L.T., Addison, P.S., McKenzie, W.M.C., 2000. Fractal analysis of fracture: Oxford Science Publications, Oxford, 235 pp.
a comparison of dimension estimates. Mechanics Research Communications 27, Hatton, C.G., Main, I.G., Meredith, P.G., 1993. A comparison of seismic and structural
383e392. measurements of fractal dimension during tensile subcritical crack growth.
Drolon, H., Hoyez, B., Druaux, F., Faure, A., 2003. Multiscale roughness analysis of Journal of Structural Geology 15, 1485e1494.
particles. Application to the classification of detrital sediments. Mathematical Hecht, C.A., 2000. Appolonian packing and fractal shape of grains improving
Geology 35, 805e817. geomechanical properties in engineering geology. In: Blenkinsop, T.G.,
Evertsz, C.J.G., Mandelbrot, B.B., 1992. Multifractal measures. In: Peitgen, H.-O., Kruhl, J.H., Kupkovà, M. (Eds.), Fractals and Dynamic Systems in Geoscience.
Jürgens, H., Saupe, D. (Eds.), Chaos and Fractals. Springer, Berlin/Heidelberg/ Pure and Applied Geophysics, vol. 157. Birkhäuser, Basel, pp. 539e557 (Topical
New York, pp. 922e953. Volume).
Fagereng, Å., 2011. Fractal vein distributions within a fault-fracture mesh in an Heilbronner, R., Keulen, N., 2006. Grain size and grain shape analysis of fault rocks.
exhumed accretionary mélange, Chrystalls Beach Complex, New Zealand. Tectonophysics 427, 199e216.
Journal of Structural Geology 33, 918e927. Hentschel, H.G.E., Procaccia, I., 1983. The infinite number of generalized dimensions
Falconer, K., 1990. Fractal Geometry. Wiley, Chichester, 288 pp. of fractals and strange attractors. Physica D 8, 435e444.
Falconer, K., 1997. Techniques in Fractal Geometry. Wiley, Chichester, 256 pp. Herzfeld, U.C., Overbeck, C., 1999. Analysis and simulation of scale-dependent
Feder, J., 1988. Fractals. Plenum Press, New York, 283 pp. fractal surfaces with application to searfloor morphology. Computers & Geo-
Feng, Z.C., Zhao, Y.S., Zhao, D., 2009. Investigating the scale effects in strength of sciences 25, 979e1007.
fractured rock mass. Chaos, Solitons & Fractals 41, 2377e2386. Higgins, M.D., 2006. Quantitative Textural Measurements in Igneous and Meta-
Feranie, S., Fauzi, U., Bijaksana, S., 2011. 3D fractal dimension and flow properties in morphic Petrology. Cambridge University Press, Cambridge, 276 pp.
the pore structure of geological rocks. Fractals 19, 291e297. Hirata, T., 1989. Fractal dimension of fault systems in Japan: fractal structure in rock
Flook, A.G., 1979. The Characterization of Textural and Structural Profiles by the fracture geometry at various scales. Pure and Applied Geophysics 131, 157e169.
Automated Measurement of Their Fractal Dimensions. In: 2nd European Hirata, T., Imoto, M., 1991. Multifractal analysis of spatial distribution of
Symposium on Particle Characterization, pp. 591e599. microearthquakes in the Kanto region. Geophysical Journal International 107,
Fomin, S., Hashida, T., Shimizu, A., Matsuki, K., Sakaguchi, K., 2003. Fractal concept 155e162.
in numerical simulation of hydraulic fracturing of the hot dry rock geothermal Hirata, T., Satoh, T., Ito, K., 1987. Fractal structure of spatial distribution of
reservoir. Hydrological Processes 17, 2975e2989. microfracturing in rock. Royal Astronomical Society Geophysical Journal 90,
Ford, A., Blenkinsop, T.G., 2008a. Combining fractal analysis of mineral deposit 369e374.
clustering with weights of evidence to evaluate patterns of mineralization: Hirose, T., Shimamoto, T., 2003. Fractal dimension of molten surfaces as a possible
application to copper deposits of the Mount Isa Inlier, NW Queensland, parameter to infer the slip-weakening distance of faults from natural pseudo-
Australia. Ore Geology Reviews 33, 435e450. tachylytes. Journal of Structural Geology 25, 1569e1574.
Ford, A., Blenkinsop, T.G., 2008b. Evaluating geological complexity and complexity Hisada, E., 2004. Clast-size analysis of impact-generated pseudotachylite from
gradients as controls on copper mineralisation, Mt. Isa Inlier. Australian Journal Vredefort Dome, South Africa. Journal of Structural Geology 26, 1419e1424.
of Earth Sciences 55, 13e23. Hooker, J.N., Laubach, S.E., Gomez, L., Marrett, R., Eichhubl, P., Diaz-Tushman, K.,
Ford, A., Blenkinsop, T.G., 2009. An expanded de Wijs model for multifractal Pinzon, E., 2011. Fracture size, frequency, and strain in the Cambrian Eriboll
analysis of mineral production data. Mineralium Deposita 44, 233e240. Formation sandstones, NW Scotland. Scottish Journal of Geology 47, 45e56.
Fossen, H., Hesthammer, J., 1997. Geometric analysis and scaling relations of defor- Horovistiz, A.L., de Campos, K.A., Shibata, S., Prado, C.C.S., de Oliveira Hein, L.R.,
mation bands in porous sandstone. Journal of Structural Geology 17, 1479e1493. 2010. Fractal characterization of brittle fracture in ceramics under mode I stress
Fowler, A.D., 1990. Self-organized mineral texture of igneous rocks: the fractal loading. Material Science and Engineering A 527, 4847e4850.
approach. Earth-Science Reviews 29, 47e55. Hull, J., 1988. Thickness-displacement relationships for deformation zones. Journal
Fowler, A.D., Stanley, H.E., Daccord, G., 1989. Disequilibrium silicate mineral of Structural Geology 10, 431e435.
textures: fractal and non-fractal features. Nature 341, 134e136. Hyslip, J.P., Vallejo, J.P., 1997. Fractal analysis of the roughness and size distribution
Foxford, K.A., Nicholson, R., Polya, D.A., Hebblethwaite, R.P.B., 2000. Extensional of granular materials. Engineering Geology 48, 231e244.
failure and hydraulic valving at Minas da Panasqueira, Portugal: evidence from Imre, B., Laue, J., Springman, S.M., 2010. Fractal fragmentation of rocks within
vein spatial distributions, displacements and geometries. Journal of Structural sturzstroms: insight derived from physical experiments within the ETH
Geology 22, 1065e1086. geotechnical drum centrifuge. Granular Matter 12, 267e285.
Fueten, F., 1997. A computer controlled rotating polarizer stage for the petrographic Jackson, P., Sanderson, D.J., 1992. Scaling of fault displacements from the Badajoz-
microscope. Computers & Geosciences 23, 203e208. Córdoba shear zone, SW Spain. Tectonophysics 210, 179e190.
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 19

Jensen, E., Cembrano, J., Faulkner, D., Veloso, E., Arancibia, G., 2011. Development of Li, J., Du, Q., Sun, C., 2009. An improved box-counting method for image fractal
a self-similar strike-slip duplex system in the Atacama Fault system, Chile. dimension estimation. Pattern Recognition 42, 2460e2469.
Journal of Structural Geology 33, 1611e1626. Long, J.J., Imber, J., 2011. Geological controls on fault relay zone scaling. Journal of
Johnston, J.D., McCaffrey, K.J.W., 1996. Fractal geometries of vein systems and the Structural Geology 33, 1790e1800.
variation of scaling relationships with mechanism. Journal of Structural Loriga, M.A., 1999. Scaling systematics of vein size: an example from the Guanajuato
Geology 18, 349e358. mining district (Central Mexico). In: McCaffrey, K.J.W., Lonergan, L.,
Kagan, Y.Y., 1991. Fractal dimension of brittle fracture. Journal of Nonlinear Science Wilkinson, J.J. (Eds.), Fractures, Fluid Flow and Mineralization. Geological
1, 1e16. Society of London Special Publications 155, pp. 57e67.
Kagan, Y.Y., 2010. Earthquake size distribution: power-law with exponent b h ½ ? Lorilleux, G., Jébrak, M., Cuney, M., Baudemont, D., 2002. Polyphase hydrothermal
Tectonophysics 490, 103e114. breccias associated with unconformity-related uranium mineralization
Kaye, B.H., 1978. Specification of the ruggedness and/or texture of a fine particle (Canada): from fractal analysis to structural significance. Journal of Structural
profile by its fractal dimension. Powder Technology 21, 207e213. Geology 24, 323e338.
Kaye, B.H., 1984. Fractal description of fineparticle systems. In: Beddow, K. (Ed.), Lovejoy, S., 1982. Area-perimeter relation for rain and cloud areas. Science 216,
Particle Characterization in Technology, vol. 1. CRC Press, Boca Raton. Chapt. 5. 185e187.
Kaye, B.H., 1989. A Random Walk through Fractal Dimensions. VCH Publishers, Lunina, O.V., Mart, Y., Gladkov, A.S., 2005. Fracturing patterns, stress fields
Weinheim. and earthquakes in the Southern Dead Sea rift. Journal of Geodynamics 40,
Kaye, B.H., 1993. Chaos and Complexity. VCH Publishers, Weinheim. 216e234.
Kaye, B.H., 1994. Fractal Geometry and the Mining Industry, a Review. In: Kruhl, J.H. Main, I.G., Peacock, S., Meredith, P.G., 1990. Scattering attenuation and fractal
(Ed.), Fractals and Dynamic Systems in Geoscience. Springer, Berlin/Heidelberg/ geometry of fracture systems. Pure and Applied Geophysics 133, 283e304.
New York, pp. 233e245. Main, I.G., Kwon, O., Ngwenya, B.T., Elphick, S.C., 2000. Fault sealing during
Kaye, B.H., Clark, G.G., 1985. Fractal Dimension of Extraterrestrial Fineparticles. deformation band growth in sandstone. Geology 28, 1131e1134.
Dept. Physics Laurentian University, Sudbury, Ontario, Canada. 16 p. Mair, K., Abe, S., 2011. Breaking up: comminution mechanisms in sheared simulated
Ketcham, R.A., Iturrino, G.J., 2005. Nondestructive high-resolution visualization and fault gouge. Pure and Applied Geophysics 168, 2277e2288.
measurement of anisotropic effective porosity in complex lithologies using high- Mair, K., Frye, K.M., Marone, C., 2002. Influence of grain characteristics on the
resolution X-ray computed tomography. Journal of Hydrogeology 302, 92e106. friction of granular shear zones. Journal of Geophysical Research 107 (B10),
Keulen, N., Heilbronner, R., Stünitz, H., Boullier, A.-M., Ito, H., 2007. Grain size ECV1eECV4.
distributions of fault rocks: a comparison between experimentally and natu- Majumder, S., Mamtani, M.A., 2009. Fractal analysis of quartz grain boundary
rally deformed granitoids. Journal of Structural Geology 29, 1282e1300. sutures in a granite (Malanjkhand, Central India) e implications to infer
Kindratenko, V.V., Van Espen, P.J.M., Treiger, B.A., Van Grieken, R.E., 1994. Fractal regional tectonics. Journal Geological Society of India 73, 309e319.
dimensional classification of aerosol particles by computer-controlled scanning Malik, F., 2004. Komplexität, Management und patterns of control. In: Mutius, B.
electron microscopy. Environmental Science & Technology 28, 2197e2202. von (Ed.), Die andere Intelligenz. Klett-Cotta, Stuttgart, pp. 174e187.
King, G., 1983. The accommodation of large strains in the upper lithosphere of the Mamtani, M.A., 2010. Strain-rate estimation using fractal analysis of quartz
earth and other solids by self-similar fault systems: the geometrical origin of b- grains in naturally deformed rocks. Journal Geological Society of India 75,
value. Pure and Applied Geophysics 121, 761e815. 202e209.
Korvin, G., 1989. Fractured but not fractal: fragmentation of the Gulf of Suez Mandelbrot, B.B., 1967. How long is the coast of Britain? Statistical self-similarity
basement. Pure and Applied Geophysics 131, 289e305. and fractional dimension. Science 156, 636e638.
Korvin, G., 1992. Fractal Models in the Earth Sciences. Elsevier, p. 396. Mandelbrot, B.B., 1977. Fractals e Form, Chance and Dimension. Freeman, San
Kranz, R.L., Velde, B., 1994. Fractal and length analysis of fractures during brittle to Francisco.
ductile changes. Comment and reply. Journal of Geophysical Research 99 (B8), Mandelbrot, B.B., 1982. The Fractal Geometry of Nature. Freeman & Co., New York,
15645e15650. 468 pp.
Kreuzer, O.P., Blenkinsop, T.G., Morrison, R.J., Peters, S.G., 2007. Ore controls in the Mandelbrot, B.B., 1985. Self-affine fractals and fractal dimension. Physica Scripta 32,
Charters Towers goldfield, NE Australia: constraints from geological, geophys- 257e260.
ical and numerical analyses. Ore Geology Reviews 32, 37e80. Mandelbrot, B.B., 1989. Multifractal measures, especially for the geophysicist. Pure
Kristáková, Z., Kupková, M., 1994. Correlation of fractal surface description and Applied Geophysics 131, 6e42.
parameters with fracture toughness. In: Kruhl, J.H. (Ed.), Fractals and Dynamic Mandelbrot, B.B., Passoja, D.E., Paullay, A.J., 1984. Fractal character of fracture
Systems in Geoscience. Springer, Berlin/Heidelberg/New York, pp. 77e94. surfaces of metals. Nature 308, 721e722.
Krohn, C.E., 1988. Sandstone fractal and Euclidean pore volume distribution. Journal Manning, C.E., 1994. Fractal clustering of metamorphic veins. Geology 22, 335e338.
of Geophysical Research 93 (B4), 3286e3296. Mareschal, J.-C., 1989. Fractal reconstruction of sea-floor topography. Pure and
Krohn, E.E., Thompson, A.H.,1986. Fractal sandstone pores: automated measurements Applied Geophysics 131, 197e210.
using scanning-electron-microscope images. Physical Review B 33, 6366e6374. Marone, C., Scholz, C.H., 1989. Particle-size distribution and microstructures within
Kruhl, J.H., 1994. The formation of extensional veins: an application of the Cantor- simulated fault gouge. Journal of Structural Geology 11, 799e814.
dust model. In: Kruhl, J.H. (Ed.), Fractals and Dynamics Systems in Geoscience. Marrett, R., Allmendinger, R.W., 1991. Estimates of strain due to brittle faulting:
Springer, Berlin/Heidelberg/New York, pp. 95e104. sampling of fault populations. Journal of Structural Geology 13, 735e738.
Kruhl, J.H., Nega, M., 1996. The fractal shape of sutured quartz grain boundaries: Marrett, R., Ortega, O.J., Kelsey, C.M., 1999. Extent of power-law scaling for natural
application as a geothermometer. Geologische Rundschau, 38e43. fractures in rock. Geology 27, 799e802.
Kruhl, J.H., Andries, F., Peternell, M., Volland, S., 2004. Fractal geometry analyses of Marschallinger, R., 1997. Automatic mineral classification in the macroscopic scale.
rock fabric anisotropies and inhomogeneities. In: Kolymbas, D. (Ed.), Fractals in Computers & Geosciences 23, 119e126.
Geotechnical Engineering. Advances in Geotechnical Engineering and Tunnel- Marschallinger, R., 1998. A method for three-dimensional reconstruction of macro-
ling, vol. 9. Logos, Berlin, pp. 115e135. scopic features in geological materials. Computers & Geosciences 24, 875e883.
Kueppers, U., Perugini, D., Dingwell, D.B., 2006. “Explosive energy” during volcanic Marschallinger, R., 2001. Three-dimensional reconstruction and visualization of
eruptions from fractal analysis of pyroclasts. Earth and Planetary Science Letters geological materials with IDL e examples and source code. Computers &
248, 800e807. Geosciences 27, 419e426.
Kulatilake, P.H.S.W., Fiedler, R., Panda, B.B., 1997. Box fractal dimension as a meas- Marschallinger, R., Johnson, S.E., 2001. Presenting 3-D models of geological mate-
ure of statistical homogeneity of jointed rock masses. Engineering Geology 48, rials on the World Wide Web. Computers & Geosciences 27, 467e476.
217e229. Marsh, B.D., 1988. Crystal size distribution (CSD) in rocks and the kinetics and
Kusunose, K., Lei, X., Nishizawa, O., Satoh, T., 1991. Effect of grain size on fractal dynamics of crystallization I. Theory. Contributions to Mineralogy and Petrology
structure of acoustic emission hypocenter distribution in granitic rocks. Physics 99, 277e291.
of Earth and Planetary Interior 67, 194e199. Mazzarini, F., Armienti, P., 2001. Flank cones at Mount Etna volcano: do they have
La Pointe, P.R., 1988. A method to characterize fracture density and connectivity a power-law distribution? Bulletin of Volcanology 62, 420e430.
through fractal geometry. International Journal of Rock Mechanics and Mining McCaffrey, K.J.W., Johnston, J.D., 1996. Fractal analysis of a mineralised vein
Science & Geomechanics Abstracts 25, 421e429. deposit: Curraghinalt gold deposit, County Tyrone. Mineralium Deposita 31,
Launeau, P., Robin, P.Y.F., 1996. Fabric analysis using the intercept method. 52e58.
Tectonophysics 267, 91e119. McCaffrey, K.J.W., Johnston, J.D., Feely, M., 1993. Use of fractal statistics in the
Launeau, P., Bouchez, J.-L., Benn, K., 1990. Shape preferred orientation of analysis of Mo-Cu mineralization at Mace Head, County Galway. Irish Journal of
object population: automatic analysis of digitized images. Tectonophysics 180, Earth Sciences 12, 139e148.
201e211. Mecholsky, J., Mackin, T.J., 1988. Fractal analysis of fracture in Ocala chert. Journal of
Ledésert, B., Dubois, J., Velde, B., Meunier, A., Genter, A., Badri, A., 1993a. Geomet- Materials Science Letters 7, 1145e1147.
rical and fractal analysis of a three-dimensional vein network in a fractured Mechtcherine, V., 2009. Fracture mechanical behavior of concrete and the condition
granite. Journal of Volcanology and Geothermal Research 56, 267e280. of its fracture surface. Cement and Concrete Research 39, 620e628.
Ledésert, B., Dubois, J., Genter, A., Meunier, A., 1993b. Fractal analysis of fractures Merceron, T., Velde, B., 1991. Application of Cantor’s Method for fractal analysis of
applied to Soultz-sous-Forêts hot dry rock geothermal program. Journal of fractures in the Toyoha Mine, Hokkaido, Japan. Journal of Geophysical Research
Volcanology and Geothermal Research 57, 1e7. 96 (B10), 16641e16650.
Ledésert, B., Hebert, R., Genter, A., Bartier, D., Clauer, N., Grall, C., 2010. Fractures, Micarelli, L., Benedicto, A., Wibberley, C.A.J., 2006. Structural evolution and
hydrothermal alterations and permeability in the Soultz Enhanced Geothermal permeability of normal fault zones in highly porous carbonate rocks. Journal of
System. Comptes Rendus Geoscience 342, 607e615. Structural Geology 28, 1214e1227.
20 J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21

Milman, V.Y., Stelmashenko, N.A., Blumenfeld, R., 1994. Fracture surfaces: Peternell, M., 2007. Fabric Patterns in Magmatic Rocks: Method Development,
a critical review of fractal studies and a novel morphological analysis of Automation and Geological Significance. PhD thesis., Technical University
scanning tunneling microscopy measurements. Progress in Material Science Munich, Faculty of Civil and Geodetic Engineering, p. 124.
38, 425e474. Peternell, M., Kruhl, J.H., 2009. Automation of pattern recognition and
Mohsen, A.I., Mahmoud, A.I., Islam, Md.S., Chudnovsky, A., 2003. Fractal dimen- fractal-geometry-based pattern quantification, exemplified by mineral-phase
sion e a measure of fracture roughness and toughness of concrete. Engineering distribution patterns in igneous rocks. Computers & Geosciences 35, 1415e1426.
Fracture Mechanics 70, 125e137. Peternell, M., Andries, F., Kruhl, J.H., 2003. Magmatic flow-pattern anisotropies e
Monzawa, N., Otsuki, K., 2003. Comminution and fluidization of granular fault analyzed on the basis of a new ‘map-counting’ fractal geometry method.
materials: implications for fault slip behavior. Tectonophysics 367, 127e143. Journal of the Czech Geological Society 48, 104.
Morgan, J.K., Boettcher, M.S., 1999. Numerical simulations of granular shear zones Peternell, M., Bitencourt, M.F., Kruhl, J.H., Stäb, C., 2010. Macro and microstructures
using the distinct element method. 1. Shear zone kinematic and the micro- as indicators of the development of syntectonic granitoids and host rocks in the
mechanics of localization. Journal of Geophysical Research 104, 2703e2719. Camboriú region, Santa Catarina, Brazil. Journal of South American Earth
Morse, D.R., Lawton, J.H., Dodson, M.M., Williamson, M.H., 1985. Fractal dimension of Sciences 29, 738e750.
vegetation and the distribution of arthropod body lengths. Nature 314, 731e733. Peternell, M., Bitencourt, M.F., Kruhl, J.H., 2011. Combined quantification of
Nagahama, H., 1991. Fracturing in the solid earth. The Science Reports of the Tohoku anisotropy and inhomogeneity of magmatic rock fabrics e an outcrop scale
University. Second Series, Geology 61, 103e126. analysis recorded in high resolution. Journal of Structural Geology 33, 609e623.
Nagahama, H., 1993. Fractal fragment size distribution for brittle rocks. Interna- Petford, N., Bryon, D., Atherton, M.P., Hunter, R.H., 1993. Fractal analysis in granitoid
tional Journal of Rock Mechanics and Mining Science & Geomechanics Abstracts petrology: a means of quantifying irregular grain morphologies. European
30, 469e471. Journal of Mineralogy 5, 593e598.
Nagahama, H., Yoshii, K., 1993. Fractal dimension and fracture of brittle rocks. Pettijohn, E.J., 1975. Sedimentary Rocks, third ed. Harper & Row, New York, 628 pp.
International Journal of Rock Mechanics and Mining Science & Geomechanics Pickering, G., Bull, J.M., Sanderson, D.J., Harrison, P., 1994. Fractal fault displace-
Abstracts 30, 173e175. ments: a case study from the Moray Firth, Scotland. In: Kruhl, J.H. (Ed.), Fractal
Nakaya, S., Nakamura, K., 2007. Percolation conditions in fractured hard rocks: and Dynamic Systems in Geoscience. Springer, Berlin/Heidelberg/New York,
a numerical approach using the three-dimensional binary fractal fracture pp. 105e119.
network (3D-BFFN) model. Journal of Geophysical Research e Solid Earth 112, Pickering, G., Bull, J.M., Sanderson, D.J., 1995. Sampling power-law distributions.
B12203. Tectonophysics 248, 1e20.
Nanjo, K., Nagahama, H., 2000. Spatial distribution of aftershocks and the fractal Piggott, A.R., 1997. Fractal relations for the diameter and trace length of disc-shaped
structure of active fault systems. Pure and Applied Geophysics 157, 575e588. fractures. Journal of Geophysical Research 102 (B8), 18,121e18,125.
Nanjo, K., Nagahama, H., Satomura, M., 1998. Rates of aftershock decay and the Power, W.L., Tullis, T.E., 1991. Euclidian and fractal models for the description of rock
fractal structure of active fault systems. Tectonophysics 287, 173e186. surface roughness. Journal of Geophysical Research 96 (B1), 415e424.
Narr, W., Suppe, J., 1991. Joint spacing in sedimentary rocks. Journal of Structural Qin, J., Zhong, D., Ng, S.L., Wang, G., 2012. Scaling behavior of gravel surfaces.
Geology 13, 1037e1048. Mathematical Geosciences 44, 583e594.
Nicol, A., Walsh, J.J., Watterson, J., Gillespie, P.A., 1996. Fault size distributions e are Rantzsch, U., Franz, A., Kloess, G., 2012. Fractal dimension and maximum roughness
they really power-law? Journal of Structural Geology 18, 191e199. applied as sculpture descriptor for tektites. Mathematical Geosciences 44, 711e720.
Nolte, D.D., Pyrak-Nolte, L.J., Cook, N.G.W., 1989. The fractal geometry of flow paths Rawls, W.J., Brakensiek, D.L., 1995. Utilizing fractal principles for predicting soil
in natural fractures in rock and the approach to percolation. Pure and Applied hydraulic properties. Journal of Soil and Water Conservation 50, 463e465.
Geophysics 131, 111e138. Rawls, W.J., Brakensiek, D.L., Logsdon, S.D., 1993. Predicting saturated hydraulic
Odling, N.E., Gillespie, P., Bourgine, B., Castaing, C., Chilés, J.P., Christensen, N.P., conductivity utilizing fractal principles. Soil Science Society of America Journal
Fillion, E., Genter, A., Olsen, C., Thrane, L., Trice, R., Aarseth, E., Walsh, J.J., 57, 1193e1197.
Watterson, J., 1999. Variations in fracture system geometry and their implica- Reither, F., 2004. Komplexität als Herausforderung und Chance e Befunde aus
tions for fluid flow in fractured hydrocarbon reservoirs. Petroleum Geoscience Forschung, Praxis und Simulation. In: Mutius, B. von (Ed.), Die andere Intelli-
5, 373e384. genz. Klett-Cotta, Stuttgart, pp. 160e173.
Okubo, P.G., Aki, K., 1987. Fractal geometry in the San Andreas fault system. Journal Renshaw, C.E., 1999. Connectivity of joints networks with power-law length
of Geophysical Research 92, 345e355. distributions. Water Resources Research 35, 2661e2670.
Orford, J.D., Whalley, W.B., 1983. The use of the fractal dimension to quantify the Renshaw, C.E., Park, J.C., 1997. Effect of mechanical interactions on the scaling of
morphology of irregular-shaped particles. Sedimentology 30, 655e668. fracture length and aperture. Nature 386, 482e484.
Orford, J.D., Whalley, W.B., 1987. The quantitative description of highly irregular Richardson, R.L., 1961. The problem of contiguity: an appendix to statistics of deadly
sedimentary particles: the use of fractal dimension. In: Marshall, J.R. (Ed.), quarrels. General Systems Yearbook 6, 139e187.
Clastic Particles. Van Nostrand Reinhold Company, New York, pp. 267e280. Roach, D.E., Fowler, A.D., 1993. Dimensionality analysis of patterns: fractal
Paggi, M., Carpinteri, A., 2007. Fractal and multifractal approaches for the analysis of measurements. Computers & Geosciences 19, 849e869.
crack-size dependent scaling laws in fatigue. Chaos, Solitons & Fractals 40, Roberts, S., Sanderson, D.J., Gumiel, P., 1999. Fractal analysis and percolation
1136e1145. properties of veins. In: McCaffrey, K.J.W., Lonergan, L., Wilkinson, J.J. (Eds.),
Panozzo, R., 1984. Two-dimensional strain from the orientation of lines in a plane. Fractures, Fluid Flow and Mineralization. Geological Society London Special
Journal of Structural Geology 6, 215e221. Publication 155, pp. 7e16.
Panozzo Heilbronner, R.P., Pauli, C., 1993. Integrated spatial and orientation analysis Rundle, J.B., 1989. Derivation of the complete Gutenberg-Richter magnitude-
of quartz c-axes by computer-aided microscopy. Journal of Structural Geology frequency relation using the principle of scale invariance. Journal of Geophys-
15, 369e382. ical Research 94, 12337e12342.
Park, S.-I., Kim, Y.-S., Ryoo, C.-R., Sanderson, D.J., 2010. Fractal analysis of the Russ, J.C., Russ, J.C., 1989. Uses of the Euclidean distance map for the measurement
evolution of a fracture network in a granite outcrop, SE Korea. Geoscience of features in images. Journal of Computer Assisted Microscopy 1, 343.
Journal 14, 99e234. Sadovskii, M.A., 1986. Some results in seismology obtained on the basis of the new
Peitgen, H.-O., Richter, H.P. (Eds.), 1986. The Beauty of Fractals. Springer, Berlin/ media model. Bulgarian Geophysical Journal 3, 3e7.
Heidelberg/New York. Sadovskii, M.A., Golubeva, T.V., Pisarenko, V.F., Shnirman, M.G., 1984. Characteristic
Peitgen, H.-O., Saupe, D. (Eds.), 1988. The Science of Fractal Images. Springer, Berlin/ dimensions of rock and hierarchical properties of seismicity. Izvestiya, Academy
Heidelberg/New York, 312 pp. of Sciences, USSR, Physics of the Solid Earth 20, 87e96 (English translation).
Peitgen, H.-O., Jürgens, H., Saupe, D., 1992a. Fractals for the Classroom, Part 1. Sahimi, M., Robertson, M.C., Sammis, C.G., 1993. Fractal distribution of earthquake
Springer, Berlin/Heidelberg/New York. hypocenters and its relation to fault patterns and percolation. Physical Review
Peitgen, H.-O., Jürgens, H., Saupe, D., 1992b. Chaos and Fractals: New Frontiers of Letters 70, 2186e2189.
Science. Springer, Berlin/Heidelberg/New York. Sammis, C.G., Biegel, R.L., 1989. Fractals, fault-gouge, and friction. Pure and Applied
Pérez-López, R., Paredes, C., 2006. On measuring the fractal anisotropy of 2-D Geophysics 131, 255e271.
geometrical sets: application to the spatial distribution of fractures. Geoderma Sammis, C.G., Osborne, R.H., Anderson, J.L., Banerdt, M., White, P., 1986. Self-similar
134, 402e414. cataclasis in the formation of fault gouge. Pure and Applied Geophysics 124,
Pérez-López, R., Paredes, C., Muñoz-Martín, A., 2005. Relationship between the 53e78.
fractal dimension anisotropy of the spatial faults distribution and the paleo- Sammis, C.G., King, G., Biegel, R., 1987. The kinematics of gouge deformation. Pure
stress fields on a Variscan granitic massif (Central Spain): the F-parameter. and Applied Geophysics 125, 777e812.
Journal of Structural Geology 27, 663e677. Sanderson, D.J., Zhang, X., 1999. Critical stress localization of flow associated with
Perugini, D., Kueppers, U., 2012. Fractal analysis of experimentally generated deformation of well-fractured rock masses, with implications mineral deposits.
pyroclasts: a tool for volcanic hazard assessment. Acta Geophysica 60, In: McCaffrey, K.J.W., et al. (Eds.), Fractures, Fluid Flow and Mineralization.
682e698. Geological Society London Special Publications 155, pp. 69e81.
Perugini, D., Poli, G., 2000. Chaotic dynamics and fractals in magmatic interaction Sanderson, D.J., Roberts, S., Gumiel, P., Greenfield, C., 2008. Quantitative analysis
processes: a different approach to the interpretation of mafic microgranular of tin and tungsten bearing sheeted vein systems. Economic Geology 103,
enclaves. Earth and Planetary Science Letters 175, 93e103. 1043e1056.
Perugini, D., Speziali, A., Caricchi, L., Kueppers, U., 2011. Application of fractal Schmittbuhl, J., Gentier, S., Roux, S., 1993. Field measurement of the roughness of
fragmentation theory to natural pyroclastic deposits: insights into volcanic fault surfaces. Geophysical Research Letters 20, 639e641.
explosivity of the Valentano scoria cone (Italy). Journal of Volcanology and Schmittbuhl, J., Schmitt, F., Scholz, C., 1995. Scaling invariance of crack surfaces.
Geothermal Research 2002, 200e210. Journal of Geophysical Research 100, 5953e5973.
J.H. Kruhl / Journal of Structural Geology 46 (2013) 2e21 21

Sharon, E., Fineberg, J., 1999. Confirming the continuum theory of dynamic brittle Velde, B., Moore, D., Badri, A., Ledesert, B., 1993. Fractal and length analysis of
fracture for fast cracks. Nature 397, 332e335. fractures during brittle to ductile changes. Journal of Geophysical Research 98
Shepherd, R.G., 1989. Correlations of permeability and grain-size. Ground Water 27, (B7), 11935e11940.
633e638. Vignes-Adler, M., Le Page, A., Adler, P.M., 1991. Fractal analysis of fracturing in
Shimamoto, T., Nagahama, H., 1992. An argument against the crush origin of two African regions from satellite imagery to ground scale. Tectonophysics 196,
pseudotachylytes based on the analysis of clast-size distribution. Journal of 69e86.
Structural Geology 14, 999e1006. Villemin, T., Angelier, J., Sunwoo, C., 1995. Fractal distribution of fault length and
Shorrocks, B., Marsters, J., Ward, I., Evennett, P.J., 1991. The fractal dimension of offsets: implications of brittle deformation evaluation e Lorraine Coal Basin. In:
lichens and the distribution of arthropod body length. Functional Ecology 5, Barton, C.C., La Pointe, P.R. (Eds.), Fractals in the Earth Sciences. Plenum Press,
457e460. New York, pp. 205e226.
Singer, H.M., Bilgram, J.H., 2006. Integral scaling behaviour of different morphol- Vincent, L., Soille, P., 1991. Watersheds in digital space: an efficient algorithm based
ogies of 3D xenon crystals. Physica D 219, 101e110. on immersion simulations. IEEE Transactions on Pattern Analysis and Machine
Smalley, R.F., Chatellain, L.L., Turcotte, D.L., Prevot, R.A., 1987. Fractal approach to the Intelligence 13, 583e598.
clustering of earthquakes: applications to the seismicity of the New-Hebrides. Volland, S., Kruhl, J.H., 2004. Anisotropy quantification: the application of fractal
Bulletin of the Seismological Society of America 77, 1368e1381. geometry methods on tectonic fracture patterns of a Hercynian fault zone in
Sornette, D., Davy, P., Sornette, A., 1990a. Structuration of the lithosphere in plate NW-Sardinia. Journal of Structural Geology 26, 1489e1500.
tectonics as a self-organized critical phenomenon. Journal of Geophysical Vontobel, P., Lehmann, E., Carlson, W.D., 2005. Comparison of X-ray and neutron
Research 95, 17353e17361. tomography investigations of geological materials. IEEE Transactions on Nuclear
Sornette, A., Davy, P., Sornette, D., 1990b. Growth of fractal fault patterns. Physical Science 52, 338e341.
Review Letters 65, 2266e2269. Voss, R.R., Clarke, J., 1975. 1/f noise in music and speech. Nature 258, 317e318.
Sornette, A., Davy, P., Sornette, D., 1993. Fault growth in brittle-ductile experiments Walsh, J.J., Watterson, J., 1988. Analysis of the relationship between displacements
and the mechanics of continental collisions. Journal of Geophysical Research 98, and dimensions of faults. Journal of Structural Geology 10, 239e247.
111e139. Walsh, J.J., Watterson, J., 1993. Fractal analysis of fracture patterns using the
Stakhovsky, I.R., 2011. A multifractal model of crack coalescence in rocks. Physics of standard box-counting technique: valid and invalid methodologies. Journal of
the Solid Earth 47, 371e378. Structural Geology 15, 1509e1512.
Stanley, H.E., Meakin, P., 1988. Multifractal phenomena in physics and chemistry. Walsh, J.J., Watterson, J., Yielding, G., 1991. The importance of small-scale faulting in
Nature 335 (6189), 405e409. regional extension. Nature 351, 391e393.
Stanley, H.E., Ostrowsky, N. (Eds.), 1988. Random Fluctuations and Pattern Growth: Wang, Z., Cheng, Q., Cao, L., Xia, Q., Chen, Z., 2007. Fractal modelling of the
Experiments and Models. Kluwer, Dordrecht. NATO ASI Series E 157. microstructure property of quartz mylonite during deformation process.
Steacy, S.J., Sammis, C.G., 1991. An automaton for fractal patterns of fragmentation. Mathematical Geology 39, 53e68.
Nature 353, 250e252. Watanabe, K., Takahashi, H., 1995. Fractal geometry characterization of
Şuţeanu, C., 2004. Characterization of anisotropyescale relations for complex geothermal reservoir fracture networks. Journal of Geophysical Research 100
irregular structures. In: Kolymbas, D. (Ed.), Fractals in Geotechnical Engineering. (B1), 521e528.
Advances in Geotechnical Engineering and Tunnelling, vol. 9. Logos, Berlin, Watterson, J., 1986. Fault dimensions, displacements and growth. Pure and Applied
pp. 103e114. Geophysics 124, 365e373.
Şuţeanu, C., Ioana, C., 2007. Pattern identification in the dynamic fingerprint of Watterson, J., Walsh, J.J., Gillespie, P.A., Easton, S., 1996. Scaling systematic of
seismically active zones. Quaternary International 171-172, 45e51. fault sizes on a large-scale range fault map. Journal of Structural Geology 18,
Şuţeanu, C., Kruhl, J.H., 2002. Investigation of heterogeneous scaling intervals 199e214.
exemplified by sutured quartz grain boundaries. Fractals 10 (4), 435e449. Weinberg, R.F., Hodkiewicz, P.F., Groves, D.I., 2004. What controls gold distribution
Şuţeanu, C., Munteanu, F., Zugrặvescu, D., 1995. Hierarchies, scaling and anisotropy in Archean terranes? Geology 32, 545e548.
in dehydration cracking. Revue Roumaine de Géophysique 39, 21e39. Wibberley, C.A.J., Shimamoto, T., 2003. Internal structure and permeability of major
Şuţeanu, C., Munteanu, F., Zugrặvescu, D., 1997. Scaling regimes and anisotropy: strike-slip fault zones: the Median Tectonic Line in Mie Prefecture, Southwest
towards an effective approach to complex geologic structures. Revue Roumaine Japan. Journal of Structural Geology 25, 59e78.
de Géophysique 41, 25e43. Wiener, N., 1948. Cybernetics or Control and Communication in the Animal and the
Şuţeanu, C., Zugrặvescu, D., Munteanu, F., 2000. Fractal approach of structuring by Machine. The Massachusetts Institute of Technology.
fragmentation. In: Blenkinsop, T.G., Kruhl, J.H., Kupková, M. (Eds.), Fractals and Wilson, B., Dewers, T., Reches, Z., Brune, J., 2005. Particle size and energetics of
Dynamic Systems in Geoscience. Pure and Applied Geophysics, vol. 157. Bir- gouge from earthquake rupture zones. Nature 434, 749e752.
khäuser, Basel, pp. 539e557 (Topical Volume). Winkler, B., Knorr, K., Kahle, A., Vontobel, P., Lehman, E., Hennion, B., Bayon, G.,
Takahashi, M., Nagahama, T., Masuda, T., Fujimura, A., 1998. Fractal analysis of 2002. Neutron imaging and neutron tomography as non-destructive tools to
experimentally, dynamically recrystallized quartz grains and its possible study bulk-rock samples. European Journal of Mineralogy 14, 349e454.
application as a strain rate meter. Journal of Structural Geology 20, 269e275. Wong, P.-Z., Lin, J.-S., 1988. Studying fractal geometry on submicron length scales by
Tang, Y.-J., Chang, Y.-F., Liou, T.-S., Chen, C.-C., Wu, Y.-M., 2012. Evolution of the small-angle scattering. Mathematical Geology 20, 655e665.
temporal multifractal scaling properties of the Chiayi earthquake (ML ¼ 6.4), Xie, H., Pariseau, W.G., 1993. Fractal character and mechanisms of rock bursts.
Taiwan. Tectonophysics 546e547, 1e9. International Journal of Rock Mechanics and Mining Science & Geomechanics
Taylor, R.P., Micolich, A.P., Jonas, D., 1999. Fractal analysis of Pollock’s drip paintings. Abstracts 30, 343e350.
Nature 399, 422. Yfantis, E.A., Flatman, G.T., Englund, E.J., 1988. Simulation of geological surfaces
Tejedor, A., Gómez, J.B., Pacheco, A.F., 2010. Hierarchical model for distributed using fractals. Mathematical Geology 20, 667e672.
seismicity. Physical Review E 82, 016118. Yuan, R.F., Li, Y.H., 2009. Fractal analysis on the spatial distribution of acoustic
Thompson, S., Fueten, F., Bockus, D., 2001. Mineral identification using artificial emission in the failure process of rock specimens. International Journal of
neural networks and the rotating polarizer stage. Computers & Geosciences 27, Minerals, Metallurgy and Materials 16, 19e24.
1081e1089. Zazoun, R.S., 2008. The Fadnoun area, Tassili-n-Azdjer, Algeria: fracture network
Tonon, F., Kottenstette, J.T. (Eds.), 2006. Laser and Photogrammetric Methods for geometry analysis. Journal of African Earth Sciences 50, 273e285.
Rock Face Characterization. American Rock Mechanics Association, Alexandria, Zhang, X., Sanderson, D., 1994. Fractal structure and deformation of fractured rock
Virginia. masses. In: Kruhl, J.H. (Ed.), Fractals and Dynamic Systems in Geoscience.
Tricot, C., 1995. Curves and Fractal Dimension. Springer, New York, 323 pp. Springer, Berlin/Heidelberg/New York, pp. 37e52.
Tullis, J., Yund, R.A., 1985. Dynamic recrystallization of feldspar: a mechanism for Zhao, Y.H., 1998. Crack pattern evolution and a fractal damage constitutive mod-
ductile shear zone formation. Geology 13, 238e241. el for rock. International Journal of Rock Mechanics and Mining Sciences 35,
Turcotte, D.L., 1986a. Fractals and fragmentation. Journal of Geophysical Research 349e366.
91, 1921e1926. Zhao, Z.Y., Wang, Y., Liu, X.H., 1990. Fractal analysis applied to cataclastic rocks.
Turcotte, D.L., 1986b. A fractal model for crustal deformation. Tectonophysics 132, Tectonophysics 178, 373e377.
261e269. Zhao, Y.S., Feng, Z.C., Liang, W.G., Yang, D., Hu, Y.Q., Kang, T.H., 2009.
Turcotte, D.L., 1997. Fractals and Chaos in Geology and Geophysics, second ed. Investigation of fractal distribution law for the trace number of random
Cambridge University Press, Cambridge. and grouped fractures in a geological mass. Engineering Geology 109,
Velde, B., Dubois, J., Touchard, G., Badri, A., 1990. Fractal analysis of fractures in 224e229.
rocks: the Cantor’s Dust method. Tectonophysics 179, 345e352. Zygouri, V., Verroios, S., Kokkalas, S., Xypolias, P., Koukouvelas, I.K., 2008. Scaling
Velde, B., Dubois, J., Moore, D., Touchard, G., 1991. Fractal patterns of fractures in properties within the Gulf of Corinth, Greece; comparison between offshore
granites. Earth and Planetary Science Letters 104, 25e35. and onshore active faults. Tectonophysics 453, 193e210.

You might also like