Professional Documents
Culture Documents
Fractal/multifractal modeling of geochemical data: A review
PII: S0375-6742(15)00074-6
DOI: doi: 10.1016/j.gexplo.2015.04.010
Reference: GEXPLO 5555
Please cite this article as: Zuo, Renguang, Wang, Jian, Fractal/multifractal model-
ing of geochemical data: A review, Journal of Geochemical Exploration (2015), doi:
10.1016/j.gexplo.2015.04.010
This is a PDF le of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its nal form. Please note that during the production process
errors may be discovered which could aect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
T
State Key Laboratory of Geological Processes and Mineral Resources, China University of
IP
Geosciences, Wuhan 430074, China
R
SC
*Corresponding author. zrguang@cug.edu.cn (R. Zuo)
Abstract
NU
Over the past several decades, a wide range of complex structures or phenomena of interest to
MA
geologists and geochemists has been quantitatively characterized using fractal/multifractal
geochemical data, the focus has been on how to decompose geochemical populations or
P
for this purpose have been proposed on the basis of the scaling characteristics of geochemical
AC
data. These include the concentrationarea (CA) fractal model, concentrationdistance (CD)
fractal model, spectrumarea (SA) fractal model, multifractal singularity analysis, and the
concentrationvolume (CV) fractal model. These fractal models have been widely
study, fractal/multifractal modeling of geochemical data including its theory, the way it works,
its benefits and limitations, its applications, and the relationships between these models are
reviewed. The comparison among of CA, SA, and multifractal singularity analysis based on
simulated data suggested that mapping singularity technique can enhance and identify weak
anomalies caused by buried sources. Future study should focus on how to distinguish the true
T
anomalies associated to mineralization with the false anomalies from a fractal/multifractal
IP
perspective.
R
SC
Keywords: Fractal/multifractal; Concentrationarea fractal model; Spectrumarea multifractal
Singularity
NU
MA
D
P TE
CE
AC
1. Introduction
T
environmental problems becoming increasingly important in recent years, discrimination
IP
between anthropogenic pollution and natural background is assuming increasing relevance
R
SC
(Darnley et al., 1995; Plant et al., 2001; Lima et al., 2003, 2005, 2008; Albanese et al., 2007).
Similarly, how to effectively detect geochemical anomalies from background is one of the
NU
major concerns of geochemical exploration, which continues to be a cornerstone to mineral
MA
exploration at all scales ranging from regional reconnaissance to local exploration (Cohen et
al., 2010; Grunsky, 2010). Anomaly patterns, as the end product of either common geological
D
TE
processes over long periods of time or uncommon processes such as oreforming processes,
P
weathering, human activities and element dispersion from an orebody, are defined simply as
CE
geochemical features different from those that usually occur more frequently. These
AC
differences consist not only of the frequency and spatial distribution of geochemical data,
both of which have been investigated widely in the past several decades (e.g., Ahrens, 1954;
Krige, 1966;Turcotte, 1986, 1997; Reimann and Filzmoser, 2000; Agterberg, 2007; Carranza,
2009), but also the geometrical characteristics and scale invariance of geochemical patterns
(Cheng et al., 1994, 1996, 1997, 1999, 2000; Xu and Cheng, 2001; Li et al., 2003; Lima et al.,
2003; Afzal et al., 2010, 2011, 2012, 2013a; Agterberg, 2012a, 2012b; Zuo et al., 2015). It has
been shown that ore elements, especially trace elements, do not follow a normal or lognormal
distribution, but instead follow a positively skewed distribution with a long Pareto tail toward
high values (e.g., Ahrens, 1957). With respect to the spatial distribution of geochemical data,
T
autocorrelation often exists over a certain spatial range. The geometry of geochemical
IP
anomalies, also an important aspect, indicates geological structures. For example, linear
R
SC
anomalies may be associated with underlying faults, while arcuate anomalies may imply
intrusive bodies (Cheng et al., 1999). Recent studies of geochemical patterns at different
NU
scales have shown that self-similarity or self-affinity are fundamental properties of
MA
geochemical data (e.g., Blviken et al., 1992; Cheng et al., 1994; Zuo et al., 2009a, 2009b;
Afzal et al., 2011). The most effective way to distinguish geochemical anomalies from the
D
TE
above.
CE
The typical and most widely used method for detection of geochemical anomalies is setting
AC
threshold values, which contain the upper and lower limits of background variations (Hawkes
and Webb, 1962). Observations outside of this range are referred to as anomalies, whereas
those within background are not. However, traditional methods, including the one mentioned
above, exploratory data analysis (Tukey, 1977; Behrens, 1997; Reimann, 2005a, 2005b;
Carranza, 2010), and multivariate statistics (Zuo, 2011a, 2011b; Zuo et al., 2009a, 2009b,
2013; Yousefi et al., 2012, 2014), are based on the frequency distribution of geochemical
values and, therefore, neglect spatial variation and other potential characteristics that can
provide valuable information. Considering the fact that exploration geochemical data are
T
distance-weighted (IDW) and different kriging methods, have been put forward (Krige, 1978;
IP
Lam, 1983; Zimmerman et al., 1999). Although these methods acknowledge the spatial
R
SC
dependence of element concentrations, they do not consider that spatial variability is rugged
NU
The main attraction of fractal/multifractal theory lies in its ability to quantify irregular and
MA
complex phenomena or processes that exhibit similarity over a wide range of scales, which is
termed self-similarity (Mandelbrot, 1983; Feder, 1988). Since the concept of fractal was
D
TE
processes and phenomena to characterize the spatial distributions of concentrations and the
CE
relationship between tonnage and grade of deposits (e.g., Mandelbrot, 1983; Turcotte, 1986,
AC
1997, 2002; Lavallee et al., 1993; Cheng et al., 1994, 2000). With respect to applied
geochemical patterns over different scales, as indicated by the increasing number of published
papers (Fig. 1). For example, the number of papers published in the Journal of Geochemical
Exploration (JGE) at the five-year scale has increased nearly exponentially as the total
Environment, Analysis (GEEA). These numbers indicate that fractal/multifractal models have
important achievements have been made over the past thirty years. The first one is the
T
proposal of a series of fractal/multifractal models used for separating geochemical anomalies
IP
from background or for determining baseline concentration in environmental studies. The
R
SC
second is the introduction of the concept of singularity, which enables us to study
mineralization from a new nonlinear perspective and provides an effective tool for mapping
NU
local and weak anomalies, and the third is the ability to quantify the vertical distribution of
MA
geochemical elements using fractal methods.
fractal/multifractal modeling of geochemical data, including its theory, the way it works, its
CE
benefits and limitations, its applications, and the relationships among various models.
AC
2. Fractal/multifractal models
The study of Blviken et al. (1992) was among the first papers to address the importance of
fractal models used for geochemical landscape studies and to predict their profound impact on
geochemical exploration. Cheng et al. (1994) proposed the concentration-area fractal model
geochemical data (Zuo et al., 2012) and is a fundamental technique used frequently for
modeling geochemical anomalies (Carranza, 2009). The spectrum-area fractal model (SA),
introduced by Cheng et al. (1999) as a version of the CA model in the frequency domain,
can separate overlapping populations using more than one cutoff value. Based on
T
Mandelbrots radial-density law, by replacing density with concentration, Li et al. (2003)
IP
proposed the concentrationdistance (CD) fractal model, which is used for discriminating
R
SC
geochemical anomalies from background. Carranza (2009) reviewed the fractal/multifractal
methods used to model geochemical data in his book Geochemical Anomaly and Mineral
NU
Prospectivity Mapping in GIS and demonstrated the advantages of these models. Afzal et al.
MA
(2011) extended the C-A fractal model to 3D and developed the concentrationvolume (CV)
fractal model to identify various zones of mineralization. These models have been
D
TE
acknowledged widely as powerful tools for identifying anomalies as well as for determining
P
geochemical baseline in environmental studies (e.g., Cheng et al., 1994, 1997, 2000, 2010;
CE
Gonalves et al., 2001; Xu and Cheng, 2001; Li et al., 2003; Panahi et al., 2004; Albanese et
AC
al., 2007; Afzal et al., 2010, 2011, 2013b; Deng et al., 2010; Zuo, 2011a, 2011b, 2014;
Sadeghi et al., 2012; Ali et al., 2007; Heidari et al., 2013; Jesus et al., 2013; Wang et al., 2013;
He et al., 2013; Asadi et al., 2014; Sun et al., 2014; Luz et al., 2014). As for the concept of
characterize the degree of uniqueness of geological features, and it can detect potential targets
that are often smoothed by traditional contouring methods (e.g., Cheng, 2007; Xie et al., 2007;
Zuo and Cheng, 2008; Cheng and Agterberg, 2009; Zuo et al., 2009, 2013, 2015; Bai et al.,
2010; Sun et al., 2010; Arias et al., 2012; Xiao et al., 2012, 2014; Zhao et al., 2012, 2013; Liu
T
et al., 2013; Zhao et al., 2014; Chen et al., 2015). Considering the fact that different points
IP
within a small vicinity may have different singularity exponents, Chen et al. (2007) proposed
R
SC
a local singularity iteration algorithm as an improvement of parameter estimation involved in
the local singularity analysis. Cheng (2012) proposed a densityarea power-law model,
NU
attempting to systematically confirm that singularity analysis is effective for recognition of
MA
weak geochemical anomalies. Agterberg (2012a, 2012b) noticed that singularity exponents
are linearly related to logarithmically transformed element concentration values, which can be
D
TE
used to measure the small-scale nugget effect caused by measurement error and microscopic
P
randomness. Zuo et al. (2015) found that the local singularity index calculated by the original
CE
etc., usually need to be estimated. A number of studies have demonstrated that the distribution
surficial materials (Sanderson et al., 1994; Monecke et al., 2001; Zuo et al., 2009b; Wang et
al., 2012; Nazarpour et al., 2014). These properties can provide useful information about
mineralization and can be used to distinguish mineralized zones from those that are not
mineralized.
T
2.1. Numbersize model
IP
The very important fractal model, i.e., the numbersize model (NS), was firstly proposed by
R
SC
Mandelbrot (1983) to characterize the relationship between the size of objects and the number
of objects with size greater than or equal to a given size. Based on this model, several variants
NU
have been developed and successfully applied in earth sciences (Carlson, 1991; Sanderson et
MA
al., 1994; Agterberg, 1996; Turcotte, 2002; Zuo et al., 2009b; Hassanpour and Afzal, 2013).
This model was probably the first one with geochemical relevance, but it fails to characterize
D
TE
geochemical patterns effectively due to its emphasis on only the frequency distribution of
P
variables.
CE
From the multifractal point of view, Cheng et al. (1994) derived the CA
fractal model, which relates the element concentration to the area enclosed by concentration
A v 1 ; A v 2 , (1)
where A denotes the area with concentration values greater than or equal to the contour
value , is the mathematical symbol for proportional to, and 1 and 2 are exponents
associated with minimum and maximum singularity, respectively. Two approaches can be
used to calculate the enclosed area. One is based on the contour map created by interpolation
procedures, and the other is based on superimposing a grid with cells on the study area and
T
calculating the area by means of a box-counting method. Distinct patterns, each
IP
corresponding to a set of similarly shaped contours, can be separated by different straight
R
SC
segments fitted to the values of the contours and enclosed areas on the loglog plot. The
slopes of these straight lines can be taken as an estimation of the exponents of the power-law
NU
relation in Eq. (1). The optimum threshold for separating geochemical anomalies from
MA
background is the concentration value common to both linear relationships on the loglog plot.
different frequencies. Based on this argument, Cheng et al. (1999) proposed the SA fractal
CE
model to separate geochemical anomalies from background using spectral analysis in the
AC
frequency domain combined with the CA model, which can provide an effective tool for
determining an optimum threshold between different patterns on the basis of the scaling
property. This fractal model also can be expressed as a power-law relation between the power
spectrum and area A E with the power spectrum above the given value as follows:
-2d
A E
, (2)
concentration. Different patterns can be recognized by fitting several straight lines, each with
Submitted to JGE SI: Geoinformatics 10
ACCEPTED MANUSCRIPT
a different slope, to the data pairs of area and power spectrum on the loglog plot. An
irregular fractal filter can then be built according to these distinct patterns with background
T
and noise corresponding to low and high power spectrum values being removed. The
IP
geochemical anomalies of interest can be obtained by converting the filtered pattern back to
R
SC
the spatial domain (Cheng, 1999a; Xu and Cheng, 2001; Cheng et al., 2010; Zuo, 2011a,
levels, leading to the spatial distribution of element concentration being clustered at different
D
TE
scales (Li et al., 2003). Mandelbrot proposed the radial-density model as an approach to
P
characterize the clustering of point events (Mandelbrot, 1983). Based on this model, Li et al.
CE
(2003) developed the concentrationdistance (CD) fractal model by replacing the density
AC
with element concentration. This model can directly process original element concentration
data, which can avoid the error caused by any interpolation procedure. Determination of the
optimum threshold for this model is nearly the same as for the CA model.
On the basis of the same idea as the CA model, the CV fractal model was proposed by
Afzal et al. (2011) to quantify the relationship between element concentration and the
accumulative volume with concentration greater than or equal to the given value (Afzal et al.,
2011; Delavar et al., 2012; Sadeghi et al., 2012; Wang et al., 2012; Lin et al., 2013; Soltani et
T
2.6. Local singularity analysis
IP
Singular processes, including earthquakes, mineralization, etc., usually result in anomalous
R
SC
amounts of energy release or mass accumulation, which are generally confined to a relatively
NU
complex processes, and they can often be expressed as multifractal distributions. The
MA
singularity mapping technique is developed on the basis of the local singularity exponent,
multifractal context, the singular geochemical distribution can be described by the following
CE
powerlaw relation:
AC
A =cA 2 , (3)
where A denotes the total amount of metal within an area of size A , c is a constant also
termed the fractal density, and a is the singularity exponent, which can be estimated using the
= log 1 2 log
A1 A2 . (4)
geochemical dispersion patterns, the case with difference value greater than zero indicating
for mapping a local singularity, which has been used widely for mineral prospecting.
T
As showed in the Lebesgue decomposition theorem, any complex measure can be
IP
decomposed into two components: one is an absolutely continuous measure and the other a
R
SC
singularity. Therefore, for the power-law relation in Eq. (3), the value of c should be a
nonsingular component (Cheng, 2005; Chen et al., 2007). In the context of GIS, a point is
NU
different from a pixel because a point has no size. Therefore, the singularity exponent based
MA
on the raster model represents a local singularity in the sense of small vicinity but not a point-
wise singularity, i.e., the estimated value of c still contains the singularity. This kind of
D
TE
knowledge led to the proposal by Chen et al. (2007) of an iterative algorithm to obtain the
P
optimal local singularity exponent, which has been demonstrated to be superior to the non-
CE
iterative algorithm originally proposed by Cheng (2007). Considering that the original
AC
algorithm used to estimate the local singularity exponent could not directly process a data set
containing negative values and that the estimated singularity is influenced by background
values, Zuo et al. (2015) developed a modified algorithm to overcome the aforementioned
shortcomings. The improvement associated with the new algorithm involves the introduction
of a step in which the minimum value within the maximum window for each given location is
previously subtracted. This kind of straight processing solves the existing problem and
enables the singularity of similar anomalous patterns within different background values to be
comparable.
T
2.7. Other parameters related to fractal methods
IP
The Hurst exponent (H) was proposed by Hurst (1951) as a measure of long-term dependence,
R
SC
and it has been applied widely in the geosciences (Turcotte, 1997). The expected value of H
lies between 0 and 1. A larger H means a stronger persistence of the time series. The rescaled
NU
range statistic (R/S) analysis as a measure of how the apparent variability of a series changes
MA
with the length of the time-period being considered is one of the most popular methods to
calculate the Hurst exponent. The rescaled range is calculated through dividing the range of
D
TE
the values within a portion of the time series by the standard deviation of the values for the
P
whole same portion of the time series. The Hurst exponent can be estimated from the slope of
CE
regression line with log (n) versus log(R/S), where n is the length of sub-series.
AC
In addition, the box dimension also can be used for measurement of the complexity of
geochemical patterns, which can be estimated by the box counting method. Further details on
these parameters and models can be found in related literature (e.g., Sanderson et al.,
3. Application
T
The past several decades saw the emergence of fractal/multifractal models and their utility in
IP
the field of applied geochemistry. Particularly, various types of geochemical datasets,
R
SC
including stream and lake sediments, soil, borehole, and lithogeochemistry datasets have been
NU
pollution and natural sources (e.g., Cheng et al., 1994, 2000,2010; Gonalves et al., 2001; Xu
MA
and Cheng, 2001; Li et al., 2003; Panahi et al., 2004; Albanese et al., 2007; Afzal et al., 2010,
2011; Deng et al., 2010; Zuo, 2011a, 2011b, 2014; Sadeghi et al., 2012; Heidari et al., 2013;
D
TE
Jesus et al., 2013; Wang et al., 2013; He et al., 2013; Asadi et al., 2014; Sun et al., 2014; Luz
P
et al., 2014). Many researchers also found that the components decomposed from the original
CE
geochemical data often have good correspondence to the geological processes or phenomena.
AC
For example, Cheng et al. (1994) applied the CA fractal model to lithogeochemical data of
the Mitchell-Sulphurets precious metal district, British Columbia, and found that different
fractal patterns exist inside and outside the potassic, sulfidic, and silicic alteration areas. Afzal
et al. (2011) employed the CV fractal model to separate supergene enrichment and hypogene
zones from oxidation zones and barren host rocks, and the zones interpreted on the basis of
the CV fractal model are consistent with the geological settings. In addition, these fractal
models are often combined with principal component analysis or factor analysis methods to
delineate multi-element association anomalies (e.g., Cheng et al., 1997; Panahi et al., 2004;
Zuo, 2011a, 2011b, 2014; Lin et al., 2013; Asadi et al., 2014; He et al., 2014; Shamseddin et
T
al., 2014; Hosseini et al., 2014). New models also can be established based on these models.
IP
For instance, Wang et al. (2011) proposed a new fractal model to estimate the reserve. In this
R
SC
model, an orebody can be divided spatially into several parts with different degrees of
mineralization via the cutoff values obtained from the CA fractal model. Furthermore, on the
NU
basis of the CA fractal model, a median CA model was derived, and the ore tonnage
MA
thickness and the metal tonnagegrade thickness models then were established (Wang et al.,
2012). These new models can be helpful for understanding orebody spatial distribution. Cao
D
TE
and Cheng (2012) combined the SA fractal model and generalized scale invariance to
P
develop a new method, which can not only separate anomalies from background but can also
CE
Cheng (2007) proposed the concept of the singularity exponent to depict the local structure of
geochemical patterns, and many case studies have been conducted to demonstrate its utility.
Note that some methods, such as principal component analysis, spatially weighted principal
analysis, etc., could be applied to the geochemical data before singularity mapping to
integrate information of ore-forming elements (Xiao et al., 2012; Zhao et al, 2012, 2013; Zuo
et al., 2015). Considering the difference of elemental mobility in areas with overburden, Xiao
T
coefficient analysis to determine a set of suitable indicator elements for further analysis. This
IP
procedure is necessary for identification of weak anomalies from buried mineralization. From
R
SC
the perspective of application, singularity analysis cannot only identify geochemical
anomalies but can also map igneous rock bodies if different element associations are analyzed
NU
(Zhao et al., 2012). For example, stream sediment geochemical data of K2O, Na2O, SiO2, and
MA
Al2O3 can be integrated using principal component analysis and then one component score is
processed to map acid igneous rocks using singularity analysis. In addition, on the basis of
D
TE
Considering the advantage that the singularity exponent can identify heterogeneity, Wang et
CE
al. (2013) proposed a new model, the fault trace-oriented singularity mapping technique, to
AC
Note that a singularity distribution map derived from a singularity mapping technique requires
further analysis for modeling mineral prospectivity, which usually involves setting a series of
thresholds to divide the singularity exponents into binary patterns and then calculating the
spatial correlation between these patterns and known deposits and occurrences. These
correlation indices, termed students t-values, can be estimated by the weights of evidence
method. Further details on this method can be found in related literature (Bonham-Carter,
T
1994; Cheng et al., 1994).
IP
There are also some other applications of singularity mapping techniques that deserve to be
R
SC
mentioned. Local singularity analysis has been proposed in the multifractal context, and it can
NU
conventional methods, one prominent advantage of this approach is its ability to determine all
MA
singularities including positions with extreme values, which often disable statistical methods
due to the small sample size problem. Agterberg (2012) found that estimated singularities are
D
TE
relation, the small-scale nugget effect can be measured. This argument has been demonstrated
CE
by several cases, including simulated and practical geochemical data. In addition, singularity
AC
exponents can be used for interpolation. Conventional methods, such as inverse distance
weighting and Kriging, fail to take into account the local properties of geochemical data. The
incorporating local singularity into traditional models. In this regard, this method takes into
account the local structure and singularity in assigning weights for data interpolation. Both
one- and two-dimensional cases have been used to demonstrate its superiority over
conventional methods (Cheng, 1999b, 2000, 2005, 2008). In fact, ordinary moving average
T
3.3. Characterization the vertical distribution of geochemical element concentration
IP
Characterization of the vertical distribution of geochemical element concentration plays an
R
SC
important role in economic planning in the mining industry. Cheng (2014) proposed a new
non-linear differential equation which assumes the decay rate of concentration is negatively
NU
proportional to the concentration itself, with a functional coefficient dependent on vertical
MA
distance from the underlying surface of mineralized rocks to quantify the decay behaviour of
revealed that the distribution of geochemical elements exhibits fractal properties, which can
P
be characterized by fractal or multifractal models. The routine procedure for this purpose
CE
Hurst exponent, etc. The box dimension often can be used for quantification of the complexity
and non-mineralized zones (Zuo et al., 2009). The Hurst exponent is associated with
continuity of mineralization, and the larger the Hurst exponent, the better the continuity. In
addition, the power-law frequency distribution also plays an important role in characterizing
the vertical distribution of geochemical elements. It can reveal different patterns of elemental
Cheng (2014) briefly reviewed the state-of-the-art the vertical distribution of elements in
T
upper regolith over mineral deposits, and proposed a power-law decay function to model the
IP
regolith decay trends with increasing distance from the underlying altered rocks or saprocks.
R
SC
Cheng's work indicated that the element concentration is a function of the depth of mineral
deposits. For instance, Cheng (2012) showed a geological and geochemical profile in Gejiu
NU
Tin district, China (c.f., Fig.2 in Cheng, 2012). It can be observed that in the east of Gejiu,
MA
most of the known deposits with outcropping or shallower ore bodies (Fig. 2G) have large Sn
concentrations; however, in the west, the ore bodies occurred about 1 km below the surface of
D
TE
the Earth, correspond to small Sn concentrations (Fig. 2F). Following the preceding studies,
P
the CA, SA and singularity index were compared based on simulated data, which were
CE
model, we only considered the situation described by Cheng (2012, 2014). In practice, there
could be more complex situations which should be further considered. In this paper, the
anomaly pattern was generated using a method similar to the gravity forward model (Telford
et al., 1990), which exhibited a power-law relation between magnitude of the gravity field and
the departure from the geological body. The anomalies were defined as the projection of
patterns around sources with different burial depths. The model and detailed parameters used
for simulating anomalies can are shown in Table 1 and Fig. 2. The simulated geochemical
pattern (Fig.2c) was created by superimposing the anomalies (Fig. 2b) on the regional
background (Fig. 2a). Figure 2d illustrates the profiles that intersect those patterns in the
T
middle and two anomalous sources with different depth. Obviously, the deeper the source, the
IP
weaker the pattern on the surface. Thus, the anomalous pattern corresponding to the second
R
SC
source becomes almost indistinguishable due to the masking effect of the background. Even
NU
<Insert Fig. 2 here>
MA
Three fractal models CA, SA, and local singularity analysis will soon be employed on the
simulated geochemical data to detect anomalies with the aid of a Matlab program (Wang and
D
TE
Zuo, 2015). Based on the simulated geochemical pattern, the CA plot, as illustrated in Fig.
P
3a, was obtained to reveal the relationship between the threshold value and the number of
CE
cells with values greater than or equal to it. Three straight lines can be fit by means of the
AC
least square method. Two threshold values can be then obtained and used further to divide the
geochemical pattern into three components (Fig. 3b). The left part and the circle within the
middle region can be regarded as anomalies. The middle part, with the exception of the circle,
can be regarded as moderate anomalies, and the rest can be considered as background.
Comparing these results with the anomalies map illustrated in Fig. 2b, only the higher
anomaly can be detected, which is associated with a large part of background being classified
as anomalies. In this regard, the effectiveness of the CA fractal model depends on the
T
<Insert Fig. 3 here>
IP
Converting the simulated geochemical data to the frequency domain using fast Fourier
R
SC
transformation (FFT) and applying the CA model again, the SA plot can then be obtained
(Fig. 4). Two cutoff values can be determined by the same procedure as the CA fractal
NU
model. The filter can be built using the higher cutoff value and further employed on the FFT
MA
map. Converting the filtered map back into the spatial domain, two patterns, i.e., background
Unlike the results obtained by CA and SA can extract anomalies that lie within the varied
AC
background. The decomposed background is nearly the same as the originally simulated
pattern, with the exception of those locations significantly influenced by the two anomalous
sources. As for the anomaly map, both of the sources can be reflected to some degree.
However, there are also some symmetrically distributed anomalous areas that have no direct
correspondence to potential sources. These areas may be the numerical artifact of the S-A
model itself. In this regard, this method has limited capacity to correctly map anomalies
The singularity mapping technique was also used to detect simulated anomalies. Several
parameters need to be set prior to the calculation of singularity exponents. A series of square
T
windows was selected with half window lengths ranging from 4m to 20m at a 4-m interval.
IP
The singularity map (Fig. 6) could then be created. Obviously, two circular regions with
R
SC
relatively low singularity exponents are delineated. The locations and shapes of the regions
are highly consistent with the anomalous sources. In addition, there are scarcely any
NU
interferences, as indicated by the fact that the singularity exponents of the regional field area
MA
approach 2. However, note that the singularity of the left delineated area is relatively stronger
than that of the right one due to their different burial depths. Therefore, the singularity
D
TE
mapping technique is a powerful tool for enhancing weak anomalies and locating anomalous
P
data. These methods also have been proved effectively in practice by many case studies.
However, for analyzing geochemical data itself, there are still some unsolved problems. No
matter what methods are adopted, the anomalies detected are usually of unequal importance
for further exploration. Thus, methods for selecting promising anomalies mathematically are
crucial and require additional research. There are few studies on whether fractal/multifractal
mineralization with false anomalies which may be caused by noise, data processing, or other
T
geological processes. In addition, detection of geochemical anomalies in areas with more or
IP
less overburden remains the principal challenge. Mechanisms of element dispersion and
R
SC
distribution in the surficial environment should be investigated further with the help of new
NU
This study has mainly presented an overview of fractal/multifractal models used to model
MA
geochemical data. These models, including CA, SA, the singularity mapping technique, etc.,
have been used widely to detect anomalies in geochemical exploration and determine the
D
TE
made a comparison of three fractal/mutifractal models, i.e., CA, SA, and singularity
CE
mapping, indicating that singularity mapping technique can enhance and identify weak
AC
Acknowledgments
We thank John Carranza, Peyman Afzal and an anonymous reviewer's comments and
suggestions, which improve this study.This research benefited from the joint financial support
from the National Natural Science Foundation of China (No. 41372007), and the Program for
References
Afzal, P., Fadakar, A.Y., Khakzad, A., Moarefvand, P., Rashidnejad, O.N., 2011. Delineation
T
of mineralization zones in porphyry Cu deposits by fractal concentrationvolume
IP
modeling. Journal of Geochemical Exploration 18, 220232.
R
SC
Afzal, P., Fadakar, A.Y., Moarefvand, P., Rashidnejad, O.N., Asadi, H.H., 2012. Application
NU
enrichment, leached and barren zones in Kahang Cu porphyry deposit, Central Iran.
MA
Journal of Geochemical Exploration 112, 131138.
Afzal, P., Harati, H., Fadakar, A.Y., Yasrebi, A.B., 2013a. Application of spectrumarea
D
TE
Afzal, P., Ahari, H. D., Omran, N. R., Aliyari, F., 2013b. Delineation of gold mineralized
AC
zones using concentrationvolume fractal model in Qolqoleh gold deposit, NW Iran. Ore
Afzal, P., Khakzad, A., Moarefvand, P., Rashidnejad, O.N., Esfandiari, B., Fadakar, A.Y.,
Agterberg, F.P., 1996. Multifractal modeling of the sizes and grades of giant and supergiant
T
Geomechanics Abstracts 33, A365.
IP
Agterberg, F.P., 2007. Mixtures of multiplicative cascade models in geochemistry. Nonlinear
R
SC
Processes in Geophysics 14, 201209.
122, 113122.
NU
MA
Agterberg, F.P., 2012b. Sampling and analysis of chemical element concentration distribution
205212.
Albanese, S., De Vivo B., Lima A., Cicchella, D., 2007. Geochemical background and
Ali, K., Cheng, Q., Chen, Z., 2007. Multifractal power spectrum and singularity analysis for
T
Arias, M., Gumie, P., MartnIzard, A., 2012. Multifractal analysis of geochemical anomalies:
IP
A tool for assessing prospectivity at the SE border of the Ossa Morena Zone, Variscan
R
SC
Massif (Spain). Journal of Geochemical Exploration 122, 101112.
Asadi, H.H., Kianpouryan, S., Lu, Y., McCuaig, T.C., 2014. Exploratory data analysis and C
NU
A fractal model applied in mapping multielement soil anomalies for drilling: A case
MA
study from the Sari Gunay epithermal gold deposit, NW Iran. Journal of Geochemical
Bai, J., Porwal, A., Hart, C., Ford, A., Yu, L., 2010. Mapping geochemical singularity using
P
Behrens, J.T., 1997. Principles and Procedures of Exploratory Data Analysis. Psychological
Methods 2, 131160.
Blviken, B., Stokke, P.R., Feder, J., Jssang, T., 1992. The fractal nature of geochemical
GIS. Computer Methods in the Geosciences (first ed.) Pergamon, New York. 398 pp.
Cao, L., Cheng, Q., 2012. Quantification of anisotropic scale invariance of geochemical
T
Journal of Geochemical Exploration 122, 4754.
IP
Carlson, C.A., 1991. Spatial distribution of ore deposits. Geology 19, 111114.
R
SC
Carranza, E.J.M., 2009. Geochemical Anomaly and Mineral Prospectivity Mapping in GIS.
Elsevier,Amsterdam.
NU
MA
Carranza, E.J.M., 2010. Mapping of anomalies in continuous and discrete fields of stream
10, 171187.
P
Chen, G., Cheng, Q., Zuo, R., Liu, T., Xi, Y., 2015. Identifying gravity anomalies caused by
CE
Chen, Z., Cheng, Q., Chen, J., Xie, S., 2007. A novel iterative approach for mapping local
Cheng, Q., 1999a. Spatial and scaling modeling for geochemical anomaly separation. Journal
T
Cheng, Q., 2000. Interpolation by means of multifractal, kriging and moving average
IP
techniques. In GeoCanada 2000, Proc. GAC/MAC Meeting, Calgary, AB, Canada [CD
R
SC
ROM]. 29 May2 June 2000. Geol. Assoc. Can., St. Johns, NF, Canada.
Cheng, Q., 2005. A new model for incorporating spatial association and singularity in
NU
interpolation of exploratory data. In: Geostatistics Banff 2004, Quantitative Geology and
MA
Geostatistics, Springer Press, edited by: Leuangthong O. and Deutsch C.V., Netherlands
14, 10171025.
D
TE
Cheng, Q., 2007. Mapping singularities with stream sediment geochemical data for prediction
P
Cheng, Q., 2008. Modeling local scaling properties for multiscale mapping. Vadose Zone
Journal 7, 525532.
Cheng, Q., 2012. Singularity theory and methods for mapping geochemical anomalies caused
by buried sources and for predicting undiscovered mineral deposits in covered areas.
Cheng, Q., 2014. Vertical distribution of elements in regolith over mineral deposits
T
Geochemistry: Exploration, Environment, Analysis 14, 277289.
IP
Cheng, Q., Agterberg, F.P., 2009. Singularity analysis of oremineral and toxic trace elements
R
SC
in stream sediments. Computers & Geosciences 35, 234244.
Cheng, Q., Agterberg, F.P., Ballantyne, S.B., 1994. The separation of geochemical anomalies
NU
from background by fractal methods. Journal of Geochemical Exploration 51, 109130.
MA
Cheng, Q., Agterberg, F.P., BonhamCarter, G.F., 1996. A spatial analysis method for
Cheng, Q., BonhamCarter, G.F., Hall, G.E.M., Bajc, A., 1997. Statistical study of trace
P
elements in the soluble organic and amorphous FeMn phases of surficial sediments,
CE
59, 2746.
Cheng, Q., Xia, Q., Li, W., Zhang, S., Chen, Z., Zuo, R., Wang, W., 2010. Density/area
powerlaw models for separating multiscale anomalies of ore and toxic elements in
30193025.
Cheng, Q., Xu, Y., Grunsky, E., 1999. Integrated spatial and spectral analysis for geochemical
the Fifth Annual Conference of the International Association for Mathematical Geology,
T
Cheng, Q., Xu, Y., Grunsky, E., 2000. Integrated Spatial and Spectrum Method for
IP
Geochemical Anomaly Separation. Natural Resources Research 9, 4352.
R
SC
Cohen, D.R., Kelley, D.L., Anand, R., Coker, W.B., 2010. Major advances in exploration
NU
Darnley, A.G., Bjorklund, B., Gustavsson, N., Koval, P.V., Plant, J., Steenfelt, A.,
MA
Tauchid, T.M., Xie, X.J., 1995. A Global Geochemical Database for environmental and
Delavar, S.T., Afzal, P., Borg, G., Rasa, I., Lotfi, M., Omran, N.R., 2012. Delineation of
CE
Deng, J., Wang, Q., Yang, L., Wang, Y., Gong, Q., Liu, H., 2010. Delineation and
explanation of geochemical anomalies using fractal models in the Heqing area, Yunnan
T
Hawkes, H.E.,Webb, J.S., 1962. Geochemistry in Mineral Exploration. Harper and Row, New
IP
York.
R
SC
Hassanpour, Sh., Afzal, P., 2013. Application of concentration-number (C-N) multifractal
NU
Iran. Arabian Journal of Geosciences 6, 957970.
MA
He, F., Wang, Z., Fang, C., Wang, L., Geng, X., 2014. Identification and assessment of Sn
He, J., Yao, S., Zhang, Z., You, G., 2013. Complexity and Productivity Differentiation
CE
Heidari, S.M., Ghaderi, M., Afzal, P., 2013. Delineating mineralized phases based on
Hosseini S.A., Afzal P., Sadeghi B., Sharmad T., Shahrokhi S.V.., Farhadinejad T., 2014.
T
Hurst, H.E., 1951. Long-term storage capacity of reservoirs. Transactions of the American
IP
Society of Civil Engineer 116, 770808.
R
SC
Jesus, A.P., Mateus, A., Gonalves, M.A., Munh, J., 2013. Multifractal modelling and
spatial Cusoil anomaly analysis along the southern border of the Iberian Terrane in
NU
Portugal. Journal of Geochemical Exploration 126127, 2344.
MA
Krige, D.G., 1966. A study of gold and uranium distribution patterns in the Klerksdorp
Lam, N.S., 1983. Spatial interpolation methods: A review. American Cartography 10, 129
AC
149.
Lavallee, D., Lovejoy, S., Schertzer, D., Ladoy, P., 1993. Nonlinear Variability and
Li, C., Ma, T., Shi, J., 2003. Application of a fractal method relating concentrations and
Lima, A., Albanese, S., Cicchella, D., 2005. Geochemical baselines for the radioelements K,
T
and gammaray surveys. Journal of Geochemical Exploration 20, 611625.
IP
Lima, A., De, Vivo B, Cicchella, D., Cortini, M., Albanese, S., 2003. Multifractal IDW
R
SC
interpolation and fractal filtering method in environmental studies: an application on
regional stream sediments of (Italy), Campania region. Applied Geochemistry 18, 1853
1865.
NU
MA
Lima, A., Plant, J.A., Vivo B.D., Tarvainen, T., Albanese, S., D. Cicchella, 2008.
Interpolation methods for geochemical maps: a comparative study using arsenic data
D
TE
48.
CE
Lin, X., Zhang, B., Wang, X., 2013. Application of factor analysis and concentrationvolume
AC
Liu, Y., Cheng, Q., Xia, Q., Wang, X., 2013. Application of singularity analysis for mineral
6172.
Luz, F., Mateus, A., Matos, J. X., Gonalves, M. A., 2014. Cu- and Zn-Soil Anomalies in the
T
Multifractal and Geostatistical Analyses. Natural Resources Research 23, 195-215.
IP
Mandelbrot, B.B., 1983. The Fractal Geometry of Nature. Freeman, San Francisco, 468pp.
R
SC
Monecke, T., Gemmell, J.B., Monecke, J., 2001. Fractal distributions of veins in drill core
from the Hellyer VHMS deposit, Australia: constraints on the origin and evolution of the
NU
mineralizing system. MineraliumDeposita 36, 406415.
MA
Nazarpour, A., Sadeghi, B., Sadeghi, M., 2014. Application of fractal models to
Panahi, A., Cheng, Q., BonhamCarter, G.F., 2004. Modelling lake sediment geochemical
CE
Plant, J., Smith, D., Smith, B., Williams, L., 2001. Environmental geochemistry at the global
T
Reimann, C., Filzmoser, P., 2000. Normal and lognormal data distribution in geochemistry:
IP
death of a myth. Consequences for the statistical treatment of geochemical and
R
SC
environmental data. Environmental Geology 39, 10011014.
Sadeghi, B., Moarefvand, P., Afzal, P., Yasrebi, A.B., Saein, L.D., 2012. Application of
NU
fractal models to outline mineralized zones in the Zaghia iron ore deposit, Central Iran.
MA
Journal of Geochemcial Exploration 122, 919.
Sanderson, D.J., Roberts, S., Gumiel, P., 1994. A fractal relationship between vein thickness
D
TE
and gold grade in drill core from La Codosera, Spain. Economic Geology 89, 168173.
P
Shamseddin, M., M., Afzal, P., Gholinejad, M., Yasrebi, A.B., Sadeghi, B., 2014. Delineation
CE
stream sediments data in Sarajeh 1:100,000 sheet, Central Iran. Arabian Journal of
Geosciences 7, 53335343.
Soltani, F., Afzal, P., Asghari, O., 2014. Delineation of alteration zones based on Sequential
Sun, T., Liu, L., 2014. Delineating the complexity of CuMo mineralization in a porphyry
intrusion by computational and fractal modeling: A case study of the Chehugou deposit
in the Chifeng district, Inner Mongolia, China. Journal of Geochemical Exploration 144,
128143.
T
Sun, X., Gong, Q.,Wang, Q., Yang, L., Wang, C., Wang, Z., 2010. Application of local
IP
singularity model to delineate geochemical anomalies in Xiong'ershan gold and
R
SC
molybdenum ore district, Western Henan province, China. Journal of Geochemical
NU
Telford, W.M., Geldart, L.P., Sheriff, R.E., 1990. Applied Geophysics. Cambridge University
MA
Press, 784pp.
Turcotte, D.L., 1986. A fractal approach to the relationship between ore grade and tonnage.
P
Turcotte, D.L., 1997. Fractals and Chaos in Geology and Geophysics. Cambridge University
AC
Wang, G., Carranza, E.J.M., Zuo, R., Hao, Y., Du, Y., Pang, Z., Sun, Y., Qu, J., 2012.
Wang, G., Pang, Z., Boisvert, J. B., Hao, Y., Cao, Y., Qu, J., 2013. Quantitative assessment of
T
porphyry Cu deposit (China). Journal of Geochemical Exploration 134, 8598.
IP
Wang, J., Zuo, R., 2015. A MATLAB-based program for processing geochemical data using
R
SC
fractal/multifractal modeling. Earth Science Informatics. DOI 10.1007/s12145-015-
0215-5.
NU
Wang, Q., Deng, J., Liu, H., Wang, Y., Sun, X., Wan, L., 2011. Fractal models for estimating
MA
local reserves with different mineralization qualities and spatial variations. Journal of
Wang, Q., Deng, J., Zhao, J., Li, N., Wan, L., 2012. The fractal relationship between orebody
P
Wang, W., Zhao, J., Cheng, Q., 2013. Fault traceoriented singularity mapping technique to
AC
Wang, W., Zhao, J., Cheng, Q., Liu, J., 2012. Tectonicgeochemical exploration modeling for
Xiao, F., Chen, J., Agterberg, F.P., Wang, C., 2014. Element behavior analysis and its
implications for geochemical anomaly identification: A case study for porphyry CuMo
T
deposits in Eastern Tianshan, China. Journal of Geochemical Exploration 145, 111.
IP
Xiao, F., Chen, J., Zhang, Z., Wang, C., Wu, G., Agterberg, F. P., 2012. Singularity mapping
R
SC
and spatially weighted principal component analysis to identify geochemical anomalies
NU
Journal of Geochemical Exploration 122, 90100.
MA
Xie, S., Cheng, Q., Chen, G., Chen, Z., Bao, Z., 2007. Application of local singularity in
Xu, Y., Cheng, Q., 2001. A fractal filtering technique for processing regional geochemical
P
147156.
AC
analysis and logistic function to create a fuzzy stream sediment geochemical evidence
4558.
T
Zhao, J., Wang, W., Cheng, Q., 2013. Investigation of spatially nonstationary influences of
IP
tectonomagmatic processes on Fe mineralization in eastern Tianshan, China with
R
SC
geographically weighted regression. Journal of Geochemical Exploration 134, 3850.
Zhao, J., Wang, W., Dong, L., Yang, W., Cheng, Q., 2012. Application of geochemical
NU
anomaly identification methods in mapping of intermediate and felsic igneous rocks in
MA
eastern Tianshan, China. Journal of Geochemical Exploration 122, 8189.
Zhao, J., Zuo, R., Chen, S., Kreuzer, O. P., 2014. Application of the tectono-geochemistry
D
TE
10.1016/j.oregeorev.2014.09.023.
AC
Zimmerman, D., Pavlik, C., Ruggles, A., Armstrong, A.P., 1999. An experimental
Zuo, R., 2011a. Decomposing of mixed pattern of arsenic using fractal model in Gangdese
Zuo, R., 2011b. Identifying geochemical anomalies associated with Cu and PbZn skarn
T
in the Gangdese Belt, Tibet (China). Journal of Geochemical Exploration 111, 1322.
IP
Zuo, R., 2012. Exploring the effects of cell size in geochemical mapping. Journal of
R
SC
Geochemical Exploration 112, 357367.
Zuo, R., Carranza, E.J.M., Cheng, Q., 2012. Fractal/multifractal modelling of geochemical
NU
exploration data. Journal of Geochemical Exploration 122, 13.
MA
Zuo, R., Cheng, Q., 2008. Mapping singularities a technique to identify potential Cu
mineral deposits using sediment geochemical data, an example for Tibet, west China.
D
TE
Zuo, R., Cheng, Q., Agterberg, F.P., Xia, Q., 2009a. Application of singularity mapping
CE
technique to identify local anomalies using stream sediment geochemical data, a case
AC
study from Gangdese, Tibet, western China. Journal of Geochemical Exploration 101,
225235.
Zuo, R., Cheng, Q., Xia, Q., 2009b. Application of fractal models to characterization of
Zuo, R., Wang, J., Chen, G., Yang, M., 2015. Identification of weak anomalies: A multifractal
Zuo, R., Xia, Q., Zhang, D., 2013. A comparison study of the CA and SA models with
T
Geochemistry 33, 165172.
IP
Zuo, R., 2014. Identification of geochemical anomalies associated with mineralization in the
R
SC
Fanshan district, Fujian, China. Journal of Geochemical Exploration 139, 170176.
NU
MA
D
P TE
CE
AC
Figure Captions
T
geochemical data published in Journal of Geochemical Exploration (JGE), Applied
IP
Geochemistry (AG) and Geochemistry: Exploration, Environment, Analysis (GEEA)
R
SC
during 1991-2014.
Fig.2. Maps showing simulated geochemical patterns: (a) a regional field created by linearly
NU
decreased function; (b) anomalies generated by two sources with a different buried
MA
depth; (c) simulated geochemical pattern by superimposing the regional background
Fig. 3. (A) Loglog plot of concentration versus area with values greater than or equal to the
P
given concentration; (b) anomalies map identified by the C-A fractal model.
CE
Fig. 5. Maps showing background (a) and anomalous (b) map obtained by the S-A model.
T
R IP
SC
NU
Fig. 1
MA
D
PTE
CE
AC
T
R IP
SC
NU
MA
Fig. 2
D
PTE
CE
AC
T
R IP
SC
Fig. 3
NU
MA
D
PTE
CE
AC
T
R IP
SC
NU
MA
Fig. 4
D
PTE
CE
AC
T
R IP
SC
Fig. 5
NU
MA
D
PTE
CE
AC
T
R IP
SC
NU
Fig. 6
MA
D
PTE
CE
AC
Table 1
No. 1 P
Ch 4000 80 (400,402)
x
T
3
2
y h
2 2 2
IP
Note: The simulated values range from 1 to 14. P -simulated value; h -buried depth; C -
R
constant; ( x, y) -location of center of anomaly source.
SC
NU
MA
D
P TE
CE
AC
Highlights
T
IP
(2) The C-A, S-A and singularity index are compared based on simulated data.
R
(3)Singularity mapping technique can well detect weak geochemical anomalies.
SC
NU
MA
D
P TE
CE
AC