Multifractal analysis of stream sediment geochemical data: Implications for hydrothermal nickel prospection in an arid terrain, eastern Iran

    Multifractal analysis of stream sediment geochemical data: Implications for hydrothermal nickel prospection in an arid terrain, eastern Iran Mohammad Parsa, Abbas Maghsoudi, Mahyar Yousefi, Martiya Sadeghi PII: DOI: Reference:

S0375-6742(16)30320-X doi: 10.1016/j.gexplo.2016.11.013 GEXPLO 5853

To appear in:

Journal of Geochemical Exploration

Received date: Revised date: Accepted date:

1 October 2016 26 October 2016 13 November 2016

Please cite this article as: Parsa, Mohammad, Maghsoudi, Abbas, Yousefi, Mahyar, Sadeghi, Martiya, Multifractal analysis of stream sediment geochemical data: Implications for hydrothermal nickel prospection in an arid terrain, eastern Iran, Journal of Geochemical Exploration (2016), doi: 10.1016/j.gexplo.2016.11.013

This is a PDF file 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 final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Multifractal analysis of stream sediment geochemical data: implications for hydrothermal Nickel prospection in an arid terrain, eastern Iran Mohammad Parsa1, Abbas Maghsoudi1, Mahyar Yousefi,2, Martiya Sadeghi3 Faculty of Mining and Metallurgical Engineering, Amirkabir University of Technology, Tehran, Iran

T

1

2

RI P

Faculty of Engineering, Malayer University, Malayer, Iran

3

Mineral Resource Department, Geological Survey of Sweden, Uppsala, Sweden

SC

Abstract

In this study, fractal/multifractal modeling methods have been applied for preliminary

NU

hydrothermal Ni prospection using a set of stream sediment geochemical data of Ahangaran district, which is an arid terrain in eastern Iran. The study area has a complex geological and morphological setting, which is dominated by the presence of aeolian sands. Cretaceous ophiolitic rocks hosting

MA

nickel endowment are influenced by extensive listwanitic alteration. Hydrothermal nickel mineralization occurs as veinlets and disseminated forms within the listwanite units. Among the analyzed elements, Ni, Cr and, Co have been recognized as being associated with listwanitic alteration

ED

and Ni mineralization, through robust factor analysis of compositional data. The regional background patterns of these mineralization-related elements were delineated via the concentration – area (C-A)

PT

fractal modeling method. Different multifractal characteristics of the selected elements were recognized using the multifractal spectrum f(α) curves and different multifractal indices. The

AC CE

enrichment and depletion patterns of the related elements in different parts of the study area were also identified using the local singularity mapping technique. The application of local singularity mapping manifested different geochemical patterns that were not recognized via the C-A fractal modeling method. The superiority of the local singularity mapping technique in delineating geochemical populations over the C-A fractal modeling method, was revealed by quantification of the correlation between geochemical anomalies and geological evidence of the mineralization. Keywords: Nickel mineralization; concentration – area (C-A) fractal model; multifractal spectrum; local singularity mapping; Iran.



Corresponding author: Malayer University, Malayer, Iran. Tel.: +98.911.3385443. Postal Code: 65719-95863. E-mail address: [email protected].

1

ACCEPTED MANUSCRIPT 1.

Introduction Recognition of anomalous geochemical patterns is a crucial issue in regional to local scale mineral

exploration, since mineralization is often accompanied by specific patterns of metal dispersion (Zuo

T

and Wang, 2015). Such patterns differ from normal dispersion patterns of geochemical elements

RI P

across the earth's surface in several aspects. These differences can be categorized in four groups, namely, statistical distribution (e.g., Reimann and Filzmoser, 2000), spatial distribution (e.g., Carranza, 2008), geometric shape (e.g., Cheng et al., 1999), and scale independence characteristics

SC

(e.g., Bölviken et al., 1992).

From the statistical point of view, it is known that ore-forming elements do not always follow

NU

normal and log-normal distributions; instead they often show positively skewed distributions (Reimann and Filzmoser, 2000). Moreover, from the spatial distribution point of view, abrupt changes may exist within anomalous geochemical patterns, while background geochemical patterns more

MA

typically show gradation across the area (cf. Zuo et al., 2015). The geometric shape of anomalous geochemical patterns is typically a function of structural controls of mineralization (i.e., faults and fractures), which could give exploration geochemists the best clues for recognition of anomalous

ED

patterns (e.g., Wang et al., 2012). Additionally, geochemical patterns are scale invariant, and thus, they could be modeled via power-low functions (e.g., Cheng et al., 1994; Cheng, 2007). Therefore, precise

PT

modeling of such anomalous patterns requires simultaneous consideration of all the aforementioned characteristics of geochemical anomalous patterns (Zuo and Wang, 2015). By using any of the traditional statistical methods of anomaly and background separation, such as

AC CE

setting specific threshold values (Hawkes and Webb, 1962), probability plots (Sinclair, 1974, 1976; Stanley and Sinclair, 1989), and exploratory data analysis (Turkey, 1977), several characteristics of anomalous geochemical patterns might be neglected (e.g., Sun et al., 2010; Xiao et al., 2012). In contrast, fractal/multifractal methods consider almost all the above-mentioned characteristics of geochemical anomalies (Zuo and Wang, 2015). Some of the most popular fractal/multifractal models, which have frequently been applied in identifying geochemical anomalies (e.g., Agterberg, 1994, 1996, 2012; Carranza, 2008, 2010, 2011; Cheng, 2006; Cheng and Agterberg, 2009; Cheng et al., 2010; Zuo et al., 2009; Wang et al., 2011; Afzal et al., 2010, 2013a, 2015; Luz et al., 2014; Yousefi and Carranza, 2015a, b; Yousefi and Carranza, 2016; Parsa et al., 2016a, b, c) are the number-size fractal model (N-S: Mandelbrot, 1983), the concentration-area fractal model (C-A: Cheng et al., 1994), the spectrum-area fractal model (S-A: Cheng et al., 1999, 2000), the concentration-volume fractal model (C-V: Afzal et al., 2011), the multifractal spectrum modeling method (Evertsz and Mandelbrot, 1992), and the local singularity mapping (SM: Cheng, 2007) multifractal method. Besides, exploring the plausible links between elements and mineral deposits could be useful for recognition of mineralization-related elemental associations, for which multivariate analyses have been applied (e.g., Reimann et al., 2002; Kumru and Bakac, 2003; Helvoort et al., 2005; El-Makky, 2011; Yousefi et al., 2012, 2014; He et al., 2013; Sadeghi et al., 2013, 2015). In this regard, robust

2

ACCEPTED MANUSCRIPT factor analysis of compositional data (RFA: Filzmoser et al., 2009), as a multivariate technique, could be applied to (a) account for the compositional nature of geochemical data (Aitchison, 1986) and (b) modulate the effects of outliers in recognition of element associations by focusing on the major data structures rather than outliers (Pison et al., 2003).

T

The main objective of this research is to use a set of stream sediment geochemical data for

RI P

prospecting Ni mineralization in the Ahangaran district, eastern Iran, a poorly-explored area. In the study area, due to arid climate condition, widespread presence of aeolian sands and strong winds causing the weathered materials to be transported far away from their original sources (Guillou et al.,

SC

1980), the chemical contents of stream sediment geochemical data are extremely diluted. Thus, interpretation of geochemical signatures is highly challengeable (Sadid, 2011). In such conditions, recognition of geochemical anomalies and generation of reliable target areas for further exploration

NU

are critical issues. To achieve the research objective, we applied (a) RFA of compositional data to recognize elemental associations related to the hydrothermal Ni mineralization and (b) different

MA

fractal/ multifractal methods to squeeze out different levels of geochemical anomalies of mineralization-related elements. Then we have compared the modeling methods and ground-truthed the delineated geochemical anomalies. Several researchers have applied different fractal/multifractal

ED

methods to delineate mono-element or multi-element geochemical anomalies (Arias et al., 2012; Zuo and Wang, 2015; Parsa et al., 2016a, b, c). In this study, we further describe the sequence application of multivariate and multifractal approaches to recognize mineralization-related mono-elemental

PT

geochemical signatures and to extract the weak and concealed anomalous patterns of the mineralization. However, we used a set of extremely diluted stream sediment data in an arid and

2.

AC CE

poorly-explored area.

Methods

2.1. Robust factor analysis of compositional data Factor analysis (FA) is a multivariate statistical procedure, which unravels the underlying patterns and manifests the possible links among a set of measured variables, using a correlation or covariance matrix of measured variables (Treiblmaier and Filzmoser, 2010). FA decomposes the correlation or the covariance matrix of variables into two matrices of factor loadings and factor scores. Factor loading is the correlation between the unraveled patterns and measured variables and explains which variables are significantly associated (Tripathi, 1979). FA is a widely used method for recognition of mineralization-related elemental associations (e.g., Yousefi et al., 2012, 2014). However, the direct application of FA on raw geochemical data yields unrealistic and biased factor estimation. This is due to the (a) compositional nature of geochemical data (Aitchison, 1986), without considering to which unrealistic factors would result (Aitchison, 1986; Egozcue et al., 2003; Filzmoser et al., 2009; Buccianti and Grunsky, 2014; Buccianti, 2015; Buccianti et al., 2015) and (b) the typical existence of outliers in geochemical data (Reimann and Filzmoser, 2000), which adversely

3

ACCEPTED MANUSCRIPT affect the estimation of correlation and covariance matrix, and thus, the results of FA carry exploration bias (Filzmoser et al., 2009). The former can be addressed by using any of the existing transformations from the log-ratio family (Aitchison, 1986; Egozcue et al., 2003) and the latter should be modulated by robustification of factor estimation using robust estimators, e.g., the minimum

T

covariance determinant (MCD), instead of using a correlation or covariance matrix (Pison et al.,

RI P

2003).

Thus, robust factor analysis (RFA) of compositional data should be applied for proper recognition of mineralization-related elements. Prior to RFA, a log-ratio transformation should be

SC

applied to open the closed (i.e., compositional) geochemical data (e.g., Parsa et al., 2016b). There are three log-ratio transformations, namely, additive log-ratio transformation (alr: Aitchison, 1986), centered log-ratio transformation (clr: Aitchison, 1986) and isometric log-ratio transformation (ilr:

NU

Egozcue et al., 2003); of which the clr transformation not only yields in somehow symmetric results but also results in an one-to-one explanation of transformed variables (Aitchison, 1986). Thus, the clr

MA

transformation is the most proper transformation for multivariate geochemical data analyses (Filzmoser et al., 2009). However, due to the singularity of transformed variables, application of MCD and the other robust estimators and thus robustification of FA is not possible in the clr space

ED

(Filzmoser et al., 2009). According to Filzmoser et al. (2009), one way out of this is to apply MCD on ilr-transformed data for estimating the covariance matrix. Then, for interpreting the results, the estimated matrix should be back-transformed to the clr-space where interpretation of factors is

AC CE

and Filzmoser, 2010).

PT

possible via compositional biplots (Aitchison and Greenacre, 2002) and loading matrices (Treiblmaier

2.2. Multifractal modeling methods Fractal/multifractal methods have been applied in various fields of geosciences (Agterberg and Cheng, 1999). These methods could unravel precise information on statistical distribution, spatial distribution and geometrical properties of geological, geophysical and geochemical patterns (de Souza and Rostirolla, 2011). Various natural and synthetic patterns as well as anomalous geochemical patterns follow fractal geometry (Mandelbrot, 1983). Mandelbrot (1983) proposed the concept of fractals for describing objects, which have selfsimilarity or self-affinity, meaning that a part of the object is similar to the whole pattern. The fractal geometry is scale-invariant, meaning that the properties of an object within a specific scale are somewhat similar to the properties of the same object within a different scale (Feder, 1988). The objects, which follow the fractal geometry, have a power-law relationship with scale. This power is called fractal dimension (Mandelbrot, 1983). Multifractals describe the statistical distribution of objects with fractal geometry. Multifractals are spatially intertwined fractals with continuous and variable fractal dimensions (Cheng and Agterberg, 1996; Panahi and Cheng, 2004; Arias et al., 2012). A number of fractal/ multifractal modeling methods have been developed for the recognition of anomalous geochemical patterns, of

4

ACCEPTED MANUSCRIPT which the concentration- area (C-A: Cheng et al., 1994) fractal modeling method, the multifractal spectrum method (Evertsz and Mandelbrot, 1992) and the local singularity mapping (SM: Cheng, 2007) multifractal modeling method could unravel different aspects of geochemical anomalies (e.g., Zuo and Wang, 2015). In the following subsections we describe the aforementioned fractal/

RI P

T

multifractal modeling methods. 2.2.1 C-A fractal modeling

Based on the concept of fractal geometry (Mandelbrot, 1983), Cheng et al. (1994) proposed the

SC

concentration-area (C-A) fractal model to describe the power-law relationship between the concentration (C) and the occupied area by enclosed contours with concentration values greater or

D1

(1)

A( C Ct )  q 2 .C

D2

(2)

MA

A( C Ct )  q1.C

NU

equal to the contour values t (C ≥ C t):

where A (C ≥ C t) and A (C < C t) denote the area enclosed by the contour values of greater than or

ED

equal to the C t and the area enclosed by the contour values smaller than C t, respectively; q1 and q2 are constants and D1 and D2 are fractal dimensions of anomaly and background populations, respectively. In a log-log plot, Eq. [1] and Eq. [2] are linear relationships. It means that: (3)

Log ( A( C Ct ))  Log( q 2 )  D 2 . Log ( C )

(4)

PT

Log ( A( C Ct ))  Log( q1 )  D1. Log ( C )

AC CE

Thus, the slopes of straight lines, which could be fitted to the log-log plots of concentration versus the area, denote the fractal dimensions of different geochemical populations. The intersections of straight lines determine the threshold, which may delineate geochemical populations (Cheng et al., 1994). There are two methods for estimating the A (C ≥ Ct) values, including (a) the measurement of the area enclosed by the contour value Ct on a geochemical contour map and (b) by counting the number of pixels with the average values greater than or equal to Ct and then calculation of the corresponding area (Cheng et al., 1994). 2.2.2 Multifractal spectrum

Several complicated geological processes, such as dispersion pattern of ore-forming elements across an area, are formed as the outcome of non-linear dynamic processes (Panahi and Cheng, 2004). Such processes can cause inhomogeneity over a wide range and thus, the resultant measures of these processes could not be modeled by traditional statistical models (cf. Zuo and Wang, 2015). Fractal geometry could be applied to describe scale-invariant processes. However, due to widespread inhomogeneity, measures of complicated non-linear dynamic processes could not be described via a single scaling exponent (Evertsz and Mandelbrot, 1992). Instead, the scaling exponent or the fractal dimension of such systems varies as a function of

5

ACCEPTED MANUSCRIPT spatial variation (Cheng, 1999). These processes are called multifractals (Feder, 1988), which are formed as the results of non-linear multiplicative cascade processes (Agterberg, 2007). In other words, multifractals are spatially intertwined fractals, with a continuous range or a spectrum of fractal dimensions (Agterberg, 1996).

T

For modeling continuous spectrum of fractal dimensions of geochemical variables, multifractal spectrum analysis by the momentum method (Halsey et al., 1986) can be applied. This analysis is in

RI P

fact the numerical modeling of the partition function χ(q) of the order q, the mass exponent τ(q) of the order q, the singularity index α(q) of the order q and the multifractal spectrum F[α(q)] of the order q.

SC

This analysis starts with interpolation of a geochemical variable with different scales, i.e., different cell sizes (ε1, ε2,…, εn). At each scale, a measure of µi can be calculated for each cell according to the Eq. [5]:  x . i2

(5)

NU

i

where, x is the average concentration of an element within the examined cell, and εi2 is the area of the

MA

cell (Evertsz and Mandelbrot, 1992). For a set of real numbers as the orders q, -∞ ≤ q ≤ +∞, the partition function χ(q) of the order q, can be defined as (Evertsz and Mandelbrot, 1992): N ( i )

 i 1

i q

(6)

ED

( q ) 

where N(εi) is the total number of cells of size ε. If the distribution of µi is multifractal, the relationship between the partition function, χ(q), and the cell size ε should satisfy the simple power-

PT

low conditions, where the scaling exponent is the mass exponent τ(q) of the order q. It means that (Evertsz and Mandelbrot, 1992): (7)

AC CE

( q )  c   ( q )

In a log-log plot, the Eq. [7] would be of linear relationship and thus the mass exponent τ(q) of the order q, could be estimated as the slopes of fitted straight lines to the log- log plots of χ(q) versus ε: Log (  ( q ))  ( q ).Log (  )  Log( c )

(8)

where, Log (c) is the intercept of the fitted straight line. The singularity index α(q) of the order q can be derived by taking the differential of the mass exponent τ(q) as (Evertsz and Mandelbrot, 1992):

( q ) 

 ( q ) q

(9)

The multifractal spectrum F[α(q)] of the order q, then could be calculated with the Eq. [10]: f[ (q)]  (q).q  (q)

(10)

If the curve of F[α(q)] versus α(q), or the so-called multifractal spectrum curve, is parabolic and satisfies the conditions described by Evertsz and Mandelbrot (1992), the distribution, which is being modeled could be considered as a continuous multifractal. More details about the multifractal spectrum modeling method could be found elsewhere (e.g., Evertsz and Mandelbrot, 1992; Cheng, 1999; Agterberg, 2012). For assessing the multifractal distribution pattern of geochemical contents, the parameters of Δα, ΔαL, ΔαR, R and Δf(α) have been used (e.g., Panahi and Cheng, 2004; Zuo, 2012; Arias et al., 2012),

6

ACCEPTED MANUSCRIPT which are defined by below equations:

 max min L 0 min

(11) (12)

 L   R  L   R

(14)

RI P

R

T

R max 0 (13)

f (  )  f (  )max  f (  )min (15)  '' ( 1)  ( 2)  2 ( 1)  ( 0)

SC

and (16)

where α0 is the singularity index of the order q = 0, αmax and αmin are the maximum and minimum

NU

singularity indices, respectively, ΔαL and ΔαR are the left and the right branches of the multifractal spectrum curve, respectively and they are representative of dispersion pattern of low and high

MA

geochemical contents (Evertsz and Mandelbrot, 1992). R is the asymmetric index and is used for quantification of deviation from the multifractal spectrum curve (Xie and Bao, 2004). The values of R vary in a [-1, 1] range, where the values of R < 0 denote the left deviation of multifractal spectrum

ED

curve and probably are due to the element local enrichment. On the contrary, the values of R > 0 denote the right deviation of multifractal spectrum curve and probably are due to the element local enrichment (Xie and Bao, 2004). The second derivative of the mass exponent function, at the q = 1,

PT

τ''(1), is an index of multifractality, which describes the irregularity of geochemical distribution

AC CE

patterns (Panahi and Cheng, 2004). For a multifractal distribution, τ''(1) is < 0 (Cheng, 1997). 2.2.3 Local singularity mapping

Singular physical processes are defined as non-linear processes, which are accompanied by a huge amount of energy release or material accumulation within a narrow spatial-temporal range. Cheng (2007) extended such definition to geology in order to describe geological processes which involve energy release or material dispersion within narrow spatial-temporal intervals. Due to the occurrence of mineralization within a relatively short period of geological time and consequent enrichment of elements in relatively small ore-bodies, this phenomenon could be considered as an example of singular geological processes (Cheng, 2007). From the multifractal point of view, singularity could be described via power-law relationships. If we describe the amount of material within a surface A, as μ(A), then the concentration of the material, C(A), could be modeled with C(A) = μ(A) / A. It is obvious that the amount of material μ(A) decreases as the area, A, decreases. However, the concentration is the function of the geological and ore deposition processes and as the area decreases the concentration may increase, decrease or remain constant. The measures of μ(A) and C(A) have power-law relationships with the area as (Cheng, 2007):

( A ) A/ 2

(17)

7

ACCEPTED MANUSCRIPT and C ( A ) A( / 2 )

1

(18)

where ∞ denotes proportionality and α denotes the singularity index. If the spatial distribution of the singularity index within a mineralized district was constant, the spatial distribution of mineralization

T

would have a monofractal distribution. On the contrary, the spatial variation of singularity index

RI P

within a mineralized district denotes that the spatial distribution of mineralization is multifractal (Cheng, 1999).

The singularity index (α) determines how the concentration changes as the area decreases.

SC

According to the Eq. [18], at locations where the concentration remains a constant as the area decreases, it denotes lack of singular processes or multifractality or α = 2. Moreover, at locations where the concentration increases as the area decreases, positive singularity exists or α < 2. Likewise,

NU

at locations where the concentration decreases as the area decreases, negative singularity exists or α > 2 (Cheng, 2007). Therefore, the spatial distribution of the singularity index (α), could give exploration

MA

geochemists the best clues of geochemical enrichment (α < 2), geochemical depletion (α > 2) and no association with mineralization (α = 2) (Cheng, 2007).

Cheng (2007) proposed a moving average window method for determining the singularity index

ED

(α) in any location, which is as follows. Given a location on a map, a set of sliding windows, A(r) (square windows) with variable edge sizes, ri × ri should be used to calculate average concentration

PT

values for each window Mj[A(ri)] (j=1, 2, 3, …, n) [n represents the total number of grid cells in the raster map]. The window size changes with a predefined specific step, d, i.e., ri+1 – ri = d. The average concentration values Mj[A(ri)] show a power-law relationship with the window size ri, or:

AC CE

M j [ A( ri )]  ri (  2 )

(19)

In log-log plots, the average concentration values Mj[A(ri)] show a linear relationship with the window size (ri), or (Cheng, 2007): Log M j [A(ri )] c  (   2 ) log (ri )

(20)

The points should be fitted by the least square method, and the slope of the linear relationship can be considered as (α - 2), where α is the singularity of the jth location (Cheng, 2007).

2.3. Ground-truth analysis of geochemical anomalies Delineated geochemical anomalies should be ground-truth checked with mineralization-related geological information. In this regard, delineated geochemical anomalies as binary predictor maps should be compared to a binary map of mineralization-related geological data or evidence of mineralization (Carranza, 2011). The classification of each cell then should be categorized in true positive (TP), true negative (TN), false positive (FP) and false negative (FN) groups. TPs are delineated geochemical anomaly cells, which are correctly correlated to the mineralization-related geological data. TNs are delineated geochemical background cells, which are not correlated with any mineralization-related geological features. FPs are delineated geochemical

8

ACCEPTED MANUSCRIPT anomaly cells, which do not coincide with geological evidence for mineralization. FNs are delineated geochemical background cells, which coincide with geological evidences of mineralization (cf. Swets, 1988). Two types of errors exist regarding recognition of geochemical anomalies. The type I error (T I) is

T

associated with the value of the method for delineation of geochemical background populations and

RI P

the type II error (T II) is related to the ability of the method to identify geochemical anomalies. T I and T II errors are estimated based on the equations below (Carranza, 2011): FP TP  FP

T II 

FN FN TN

(21)

SC

T I

(22)

NU

The overall accuracy (OA) is a measurement of accuracy of classification methods (Carranza, 2011), and is calculated by the Eq. [23]: TP TN TP TN  FP  FN

(23)

MA

OA 

This quantity has been used for assessing the quality of anomaly and background separation methods (e.g., Afzal et al., 2013b). The lower the two types of errors are, the better the quality of

ED

anomaly and background discrimination would be; and the higher the OA, the more reliable the

3.

PT

delineated anomalous geochemical populations (Carranza, 2011).

Hydrothermal Ni mineralization in the study area and data used

AC CE

The outcrops of ophiolitic rocks in Iran are distributed over the Khoy, Rasht, Kermanshah, Nayriz, Nayin, Sabzevar, Iranshahr and Birjand areas (Fig. 1). The Ahangaran district is located within the ophiolitic and colored mélange belt of Birjand in eastern Iran (Fig. 1). The widespread ophiolitic outcrops in the study area are Cretaceous peridotites and serpentinized harzburgites (Guillou et al., 1980; Ali et al., 2015), which have been affected by strong listwanitic alteration (Stocklin, 1975). Listwanites are carbonate-rich alteration products of ultramafic rocks (e.g., Halls and Zhao, 1995). Hydrothermal Ni mineralization in eastern Iran occurs within the ultrabasic ophiolitic rocks, which are commonly associated with listwanitic alteration (Ghorbani, 2013). The oldest rocks exposed in the Ahangaran district are Cretaceous ophiolitic rocks, Cretaceous volcanic rocks including basaltic andesites and porphyritic andesites, and Cretaceous sedimentary rocks consisting of sandstones and marls (Guillou et al., 1980: Fig. 2). Eocene limestones and sandstones are exposed in the vicinity of Cretaceous units (Fig. 2). The Cretaceous ophiolitic rocks are overlain by Oligo-Miocene quartz microdiorites (Guillou et al., 1980). The structural features of the study area have a general NW-SE trend, which is in alignment with the major and deep NW-SE trending faults in the area (Sadid, 2011). Magma and hydrothermal fluids might have been channeled through these fault systems and have formed Oligo-Miocene quartz diorites in the study area (Sadid,

9

ACCEPTED MANUSCRIPT 2011). These intrusions are spatially and genetically associated with hydrothermal Ni mineralization and listwanitic alteration in the study area (Sadid, 2011; Ali et al., 2015). The hydrothermal Ni mineralization in the study area occurs as veins and veinlets within listwanitic-altered rocks (Sadid, 2011). The major mineral assemblage associated with Ni

T

mineralization in the study area consists of nickelite, pyrite, chalcopyrite, bornite and malachite

RI P

(Sadid, 2011). The NW-SE trending faults structurally control the hydrothermal Ni mineralization in the area (Ali et al., 2015).

For Ni prospecting in the study area, stream sediment geochemical exploration has been carried

SC

out by the geological survey of Iran and a total of 188 stream sediment samples have been collected. The samples were sieved by 180 μm screen and the fractions <180 μm were selected for chemical analysis. The sieved fractions were digested in HNO3 + HCl and then analyzed for multi-elements

NU

(As, Co, Cr, Cu, Ni, Pb and Zn) by inductively coupled plasma optical emission spectrometry (ICPOES). The detection limits were: 0.2 ppm for Cu, 0.2 ppm for Pb, 0.2 ppm for Zn, 0.2 ppm for Co, 2

MA

ppm for Ni, 0.5 ppm for As, 0.1 ppm for Sb, and 2 ppm for Cr. The method of Thompson and Howarth (1976) was applied for assessing the analytical precision using duplicated samples. The precision was better than 10% for the analyzed elements.

ED

The climate in the study area is arid steppe. The typical characteristic of the study area is the presence of aeolian sands and strong winds that cause the weathered materials to be transported far away from their original sources (Guillou et al., 1980). Therefore, strong dilution effect of aeolian

PT

sands results in a complicated dispersion pattern of elements in the stream sediment geochemical data. Thus, the interpretation of such data is extremely challengeable, but it could be facilitated by

4.

AC CE

multifractal modeling approaches (Zuo and Wang, 2015).

Results

4.1. Recognition of mineralization-related geochemical elements For the recognition of mineralization-related geochemical signatures in the study area, robust factor analysis (RFA) of compositional data was applied on the seven analyzed elements (i.e., As, Co, Cr, Cu, Ni, Pb and Zn), all of which are somehow associated with the hydrothermal Ni mineralization (Sadid, 2011). Initially, ilr-transformation was applied on the raw data of seven analyzed elements for estimation of the covariance matrix (Filzmoser et al., 2009). Then, as the ilr-transformation sacrificed one variable (Egozcue et al., 2003), the resultant covariance matrix was back-transformed to the clrspace for deriving interpretable robust factors (Filzmoser et al., 2009). Several criteria should be set prior to factor analysis, of which the number of factors could be determined by a significant amount of total variability to be explained by factors (Treiblmaier and Filzmoser, 2010). In this study, the number of factors was set to be three, because these three factors not only explained 85.6 % of the total variability, but also yielded in precise discrimination of element associations. Besides, the method of factor analysis, the method of factor rotation and the method of

10

ACCEPTED MANUSCRIPT obtaining factor scores should be subjectively determined, for which the principal factor analysis (PFA) (Reimann et al., 2002), the varimax rotation approach (Kaiser, 1958) and the Bartlett method (e.g., Filzmoser et al., 2009) were used, respectively. Significant loadings were returned by the absolute threshold value of 0.5 (e.g., Sun et al., 2009), because this value is a medium value, which

T

allows the slightly high values to be incorporated in the interpretation of factors (Parsa et al., 2016b).

RI P

The results of RFA are shown in Table 1. The first factor explains 54 % of total variability and represents an As-Sb elemental association with positive loadings, while it represents a Ni-Cr-Co element association with negative loadings (Table 1). In the second factor, the robust positive

SC

loadings represent Cu enrichment (Table 1). The robust positive loadings in the third factor reflect a Pb-Zn association (Table 1). The compositional biplots of the first and the second factor clearly depict Ni-Cr-Co and As-Sb element associations (Fig. 3). Although each of the seven analyzed elements can

NU

be used to prospect Ni mineralization in the study area, based on the results of loading matrix (Table 1) and compositional biplots (Fig. 3), Ni, Cr and Co were selected as the significant geochemical

MA

signatures (i.e., Ni mineralization-related elements) for prospecting Ni mineralization in the study area (e.g., Parsa et al., 2016b, c).

ED

4.2. Statistical analyses of elements applied

As nickel mineralization in the Ahangaran district is associated with relative enrichment of Ni, Cr and Co (Table 1 and Fig. 3), the descriptive statistics of these three elements were further evaluated

PT

(Table 2). According to the Table 2, the maximum contents of Ni, Cr and Co in the stream sediment data of the study area are 965 ppm, 1259 ppm and 68 ppm respectively, which generally could not be

AC CE

regarded as a noticeable geochemical enrichment (cf. Levinson, 1979). Although, there are two known hydrothermal Ni occurrences in the study, the low chemical contents of Ni, Cr and Co in the stream sediment data of the study area are due to extensive erosion processes associated with the arid climate condition, extreme dilution effect of aeolian sands and strong winds of the study area (Sadid, 2011). Besides, in such terrain conditions, stream sediment geochemical samples are severely affected by the dilution processes (Mokhtari et al., 2014) and thus element contents in stream sediment samples are explicitly lower than those of rock samples taken from mineralization sources (Spadoni et al., 2005). As Reimann et al. (2012) mentioned, the more weathered samples (sediments and soils) have somewhat lower median values for most elements and the strength of their geochemical signal in an anomaly map can be disproportionate due to the varying responses of these elements to weathering processes. The statistical characteristics of the elements, such as the skewness and kurtosis, suggest that the raw data (the data obtained from analysis of samples without any transformations) of Ni, Cr and Co do not follow normal distributions (Table 2). Besides, the histograms of the raw data (Fig.4), obviously demonstrate that the elements follow positively skewed distributions. Thus, in the data set, it could be inferred that there are multiple populations, which may be related to the influence of a variety of geological processes (Reimann et al., 2002).

11

ACCEPTED MANUSCRIPT To assess the spatial distribution of the investigated elements (Ni, Cr and Co), interpolation of raw geochemical variables was carried out. Geochemical mapping should be performed by the selection of a proper unit cell size (Carranza, 2009; Zuo, 2012). Selection of the cell size in geochemical mapping can be based on the density of samples (Hengl 2006; Zuo 2012). In this regard, a unit cell size of 200

T

m × 200 m was used to grid all the geochemical maps in this study. Due to the straightforward

RI P

implementation (Zuo, 2011), the ordinary inverse distance weighting (IDW) method, was used for gridding the raw data. Figure 5 illustrates the spatial distribution of the raw geochemical data of Ni, Cr and Co. As shown in this figure, the spatial distribution patterns of Cr, Ni and Co are quite similar

SC

and relatively enriched in Cretaceous ophiolites and the locations of known Ni occurrences.

NU

4.3. Multifractal modeling 4.3.1 C-A fractal modeling

The C-A log-log plots, consisting of the values of gridded geochemical maps of Ni, Cr and Co

MA

(Ct), versus the number of cells with the values of gridded geochemical map greater than or equal to Ct were derived (Figs. 6a, c, e). Based upon the least square method with the minimum regression errors (MSE) and the maximum regression coefficient (R2) values, four, three and four different line

ED

segments were fitted to the C-A log-log plots of Ni, Cr and Co, respectively (Figs. 6a, c, e). Thus, based on the C-A log-log plots, four, three and four different geochemical populations were

PT

delineated for Ni, Cr and Co, respectively (Figs. 6b, d, f). Table 3 represents the delineated geochemical populations based on the C-A log-log plots, along with their regression errors, regression

AC CE

coefficients and regression equations.

The four geochemical populations of Ni are representative of high anomaly, anomaly, background and low background geochemical contents (Table 3). Concentration values (C) of greater than 570 ppm are delineated as highly anomalous geochemical contents for Ni, while the values of 210 < C < 570 ppm are recognized as anomalous concentrations for Ni (Table 3 and Figs. 6a, b). The three delineated geochemical populations for Cr represent high anomaly, anomaly and background geochemical populations (Table 3 and Figs. 6b, d). The highly anomalous geochemical contents of Cr are delineated based on the (C) values of greater than 789 ppm. Likewise, the anomalous geochemical contents of Cr are delineated based on the values of 249 < C < 789 ppm (Table 3 and Figs. 6b, d). The four delineated geochemical classes of Co are considered to represent highly anomalous, anomalous, background and low background geochemical contents (Figs. 6e, f). The highly anomalous and anomalous geochemical contents of Co were separated based on the (C) values of > 45 ppm and 21 < C < 45 ppm, respectively (Table 3 and Figs. 6e, f). 4.3.2 Multifractal spectrum modeling

Multifractal spectrum parameters of the selected elements (Ni, Cr and Co) were measured based

12

ACCEPTED MANUSCRIPT on six iterations with different cell sizes ranging from 200 m to 1200 m with steps of 200 m. Partition functions of different q values ranging from -5 to 5 with steps of 0.5 were then generated. Log-log plots of partition function χ (q) values versus different cell sizes (ε) show a series of straight lines for the selected elements, supporting their multifractal nature (Fig.7). Mass exponents τ(q) of different q

T

values were then calculated by measuring the slopes of different straight lines of Fig.7, with the least square fitting method and the R2 values of greater than 0.99 and MSE values of lower than 0.001.

RI P

The multifractal spectrum curves f(α) of the selected elements, calculated by the moment method (Halsey et al., 1986) are continuous and asymmetric with convex shapes (Fig.8). The geochemical

SC

distributions of the selected elements are of different degrees of multifractality, because the multifractal spectrum curves are different for each element (Fig.8).

The multifractal spectrum curves of the selected elements are all of right deviations (R < 0) and

NU

short left branches suggesting some degrees of local depletion. Thus, it could be deduced that the study area, which is mostly covered by ultrabasic ophiolitic rocks, is favorable for Ni mineralization

MA

and just few parts of the study area are locally depleted. Table 4 summarized the multifractal parameters of the studied elements. According to the table 4, the selected elements have similar geochemical behaviors as their multifractal parameters are quite similar (Table 4).

ED

The widths of the multifractal spectrum (Δα) of the selected elements are quite similar (Table 4). The highest Δα of Ni suggests the highest degree of irregularity and therefore multifractality (Table 4). Moreover, the values of τ'' (1) for the selected elements are all lower than 0 (Table 4), suggesting

PT

that the spatial distribution of the selected elements satisfy multifractal statistics (Cheng, 1997). The highest values of ΔαL and ΔαR, suggest the higher degrees of local singularity and thus highest degrees

AC CE

of element enrichment and depletion, respectively (cf. Arias et al., 2012). 4.3.3 Local singularity mapping

Singularity mapping (SM) can be implemented on either raw or gridded data. For areas with irregularly-spaced sampling, gridding the data not only allows for better visualization of the spatial characteristics of the data set, but also improves the accuracy of SM (Yuan et al., 2015). Therefore, SM was applied to gridded datasets for Ni, Cr and Co (Fig.5). The selection of edge sizes and intervals in SM is based on the expanse of the study area, the scale of the unit cell size and the scale of local structures of interest (Cheng, 2007). In this regard, the minimum edge size is limited by the scale of unit cell size of gridded data, while the maximum cell size could be determined by the geological characteristics of the study area. Larger intervals are appropriate for the determination of regional geochemical characteristics as well as regional background of the study area, however smaller intervals are used for the determination of local properties of the data (Cheng, 2007). In order to identify the local characteristics of the geochemical data set, a set of five edge sizes with the minimum edge size of rmin = 200 m, the maximum edge size of rmax = 1000 m, and the interval of ri+1 – ri = 200 m was used.

13

ACCEPTED MANUSCRIPT A Matlab-based program was coded and implemented for the calculation of the singularity indices (α values). Fig. 9 depicts the stretched spatial distribution maps of singularity indices (α values) of Ni, Cr and Co. The α values of Ni, Cr and Co vary in the ranges of 1.24 to 2.91, 1.23 to 2.86 and 1.46 to 2.41, respectively (Fig.9). Aiming to generate a binary map of anomalous geochemical patterns, the

T

stretched distribution maps of singularity indices were classified in two categories of (a) α < 2 and (b)

RI P

α ≥ 2, which are representatives of (a) local enrichment and (b) local depletion or no association with mineralization, respectively (Fig.10).

SC

4.4. Ground-truth analysis of the delineated geochemical anomalies

Inspection of the delineated geochemical anomalies of the Ni mineralization-related elements, generated by the C-A fractal model (Fig. 6) and local singularity analysis (Fig. 10), show that the

NU

anomalies are correlated with the two known Ni of the study area. To further assess the efficiency of C-A and local singularity analysis in defining background and anomalies, field surveys were carried

MA

out in the delimited anomaly patterns of the two modeling methods and far from the locations of two known Ni occurrences. Ni mineralization was observed in some outcrops of listwanitic Cretaceous ophiolitic rocks, which has delineated as anomalous zones (Fig. 11). The results of field surveying

ED

demonstrated that the delimited anomalous areas could be regarded as plausible exploration targets for detail exploration surveys.

It should be noted that neither all parts of the Cretaceous ophiolitic rocks are mineralized nor all

PT

parts of these rock units are classified as mineralization-related geochemical anomalies. But as Cretaceous ophiolites are host rocks of the hydrothermal Ni mineralization in the study area (Sadid,

AC CE

2011), and because the average contents of Ni, Cr and Co are generally higher in these rock units (cf. Levinson, 1979), the delineated anomaly patterns generally mark the distribution of mafic and ultramafic rocks (Reimann et al., 2012). Therefore, as Carranza (2011) proposed, the binary map pattern manifesting the outcrops of ultrabasic ophiolitic rocks was used for comparing the efficiency of modeling methods in the delineation of anomaly patterns. The geochemical populations delineated by the C-A fractal model (Table 3 and Fig.6) were used to construct binary anomaly maps. The binary anomaly maps for Ni were constructed by considering the anomalous and highly anomalous geochemical populations as the anomaly class and the background and low background geochemical populations as the anomaly class in binary geochemical anomaly map of Ni. Likewise, the binary anomaly maps of Cr were built by considering the anomalous and highly anomalous geochemical populations as the anomaly class and the background geochemical population as the background class. Similarly, the binary anomaly map of Co were constructed by considering the anomalous and highly anomalous geochemical populations as the anomaly class and the background and low background geochemical populations as the background class. Besides, on the basis of singularity mapping, Figs. 10a, b and c were considered as the binary anomaly map of Ni, Cr and Co, respectively.

14

ACCEPTED MANUSCRIPT By comparison of the binary anomaly classes of C-A fractal modeling and the binary ground-truth map of the geological controller of Ni mineralization, the cells with true positive (TP), true negative (TN), false positive (FP) and false negative (FN) classifications were counted for assessment of the quantities of type I (T I) error, type II (T II) error, and the OA. These quantities represent the ability of

T

the methods applied (i.e., the C-A fractal modeling method and the singularity multifractal model) in

RI P

recognizing background and anomalous populations (Table 5).

The lowest T I error, which is associated with the delineated Co anomalies (Table 5), demonstrates that the background populations of Co, which delineated by the C-A fractal modeling

SC

method, are more reliable than the delineated Ni and Cr background populations. Likewise, the lowest T II error, which is associated with the delineated anomalous populations of Cr by using the C-A fractal modeling method (Table 5), shows that the anomalous populations of Cr are more reliable than

NU

the delineated Ni and Co anomalous populations. The highest value of OA is associated with the delineated geochemical populations of Ni. Thus, the result of classification of Ni anomalies by using

MA

the C-A fractal modeling is more reliable than that of geochemical anomalies of Cr and Co (Table 5). The T I error, T II error and OA of binary anomaly map of Ni, Cr and Co, which derived by the singularity multifractal model (Fig. 10), were then calculated by comparing these anomalies with the

ED

ground-truth map of Ni-mineralization related geological evidence (Table 6). The lowest T I error is associated with the delineated Co anomalies (Table 6) suggesting that the delineated background populations of Co by the singularity multifractal modeling method, are more reliable than those of the

PT

delineated background populations of Ni and Cr. Likewise, the lowest T II error is associated with the geochemical anomalies of Cr delineated by the singularity multifractal modeling method (Table 6),

AC CE

suggesting that they are more reliable compared with the delineated anomalous geochemical populations of Ni and Co. The highest value of OA for the singularity multifractal method is associated with the delineated geochemical populations of Ni. It demonstrates that the results of classification of Ni anomalies by using the singularity multifractal method are more reliable than the results of classification of geochemical anomalies of Cr and Co (Table 6). However, the anomaly maps of Co and Cr, obtained from the C-A fractal and the singularity multifractal modeling methods, have the lowest T1 and T2 errors, respectively, the values of OA for the anomaly maps of Ni in both C-A fractal and the singularity multifractal models are higher than that of the Co and Cr anomaly maps. Thus, as Carranza (2008) mentioned, these suggest that the anomaly map of Ni is optimal for prospecting the mineralization in the study area because the higher the OA is, the more reliable classification would be.

5.

Discussion and conclusion In this study, a district-scale stream sediment geochemical survey has been conducted as a

preliminary exploration stage for recognizing the significant geochemical signatures of hydrothermal Ni mineralization in the Ahangaran district, an arid terrain covered mainly by the aeolian sands (Guillou et al., 1980). Stream sediment samples are a composite of upstream materials (Spadoni et al.,

15

ACCEPTED MANUSCRIPT 2005; Spadoni, 2006; Mokhtari and Nezhad, 2015), and so, geochemical anomalies of which exhibit complex spatial patterns (Macklin et al., 1994; Spadoni et al., 2005; Spadoni, 2006; Yousefi et al., 2013; Mokhtari et al., 2014). Moreover, due to the arid climate and hydromorphic conditions, the chemical contents of stream sediment materials have extremely influenced by the strong dilution

T

effect of aeolian sands of the study area (Sadid, 2011). Thus, weak and complex geochemical

RI P

anomalies are accompanied in the stream sediment data as a result of characteristics of the study area (Zuo et al., 2015). Consequently, traditional methods such as setting threshold values according to the global average concentrations of elements in specific rock units are inadequate in recognizing

SC

mineralization-related anomalies in the study area (cf. Zuo et al., 2013; Shuguang et al., 2015). For the precise recognition of mineralization-related geochemical anomalies, application of multivariate and multifractal methods has been exploited.

NU

Ni, Cr and, Co were identified as the elements, which were related to hydrothermal Ni mineralization by the application of robust factor analysis of compositional data (Table 1) and

MA

compositional biplots (Fig.3). The distribution map of these elements revealed that they are enriched within the Cretaceous ophiolitic rocks (Fig.5). Application of different fractal and multifractal models has, then, manifested different aspects of spatial and statistical distribution of the mineralization-

ED

related elements.

The concentration–area (C-A) fractal method was applied for recognition of regional background (Fig. 6 and Table 3) and reflected that the delineated geochemical populations of the mineralization-

PT

related elements are almost identical in the shape and extent of geochemical anomalies (Fig.6). The continuous multifractal nature of the studied elements was revealed by the multifractal spectrum

AC CE

curves (Fig.8) and multifractal parameters (Table 4). The multifractal spectrum method demonstrated that distribution pattern of the elements are similar. However, due to higher values of ΔαL, ΔαR and Δα, it was resulted that the spatial-frequency distribution of Ni was of higher complexity and irregularity in comparison with those of Cr and Co (Table 4). Comparison of different methods has shown that with the C-A fractal method, different patterns of element enrichments and depletions were revealed as compared with the local singularity mapping (SM) multifractal method (Fig.9). However, the delineated patterns of the studied elements by the SM method are identical (Fig.9). Comparison of the results of the C-A and SM methods with the geological evidence of mineralization (Table 5 and Table 6) show that the type I (T I) and type II (T II) errors, which are related to the uncertainty of background and anomaly populations, respectively, are lower in the SM. Besides, the overall accuracies (OA) of the SM method in delineating geochemical populations are better than those of the C-A method (Table 5 and Table 6). Therefore, following the results of Arias et al. (2012) and Zuo and Wang (2015), it was demonstrated that the application of SM not only allows for unraveling different patterns of element distributions (Figs.9, 10), but also the delineated anomalies of the method are better correlated with the geological evidence of mineralization (here, Cretaceous ophiolitic rocks) in comparison with the C-A fractal method. Besides, due to the reliability

16

ACCEPTED MANUSCRIPT of the delineated geochemical anomalies of the SM (Fig.10) for the mineralization-related elements, these anomalies could be used in mineral prospectivity modeling as geochemical evidence layers.

T

Acknowledgements

RI P

The authors express their gratitude to the geological survey of eastern Iran (GSI), particularly Mr. Hassan Azmi for providing the data to this research. The authors are also appreciative of Professor Huseyin Yilmaz and two anonymous reviewers for their constructive comments and suggestions, which significantly improved the

SC

manuscript.

NU

References

Afzal, P., Ahari, H. D., Omran, N. R., Aliyari, F., 2013b. Delineation of gold mineralized zones using concentration–volume fractal model in Qolqoleh gold deposit, NW Iran. Ore Geology Reviews 55, 125-133.

MA

Afzal, P., Asl, R.A., Adib, A., Yasrebi, A.B., 2015. Application of Fractal Modelling for Cu Mineralization Reconnaissance by ASTER Multispectral and Stream Sediment Data in Khoshname Area, NW Iran. Journal of the Indian Society of Remote Sensing 43, 121-132.

Afzal, P., Fadakar Alghandis, Y., Khakzad, A., Moarefvand, P., Rashidnejad Omran, N., 2011. Delineation of mineralization

ED

zones in porphyry Cu deposits by fractal concentration-volume modelling. Journal of Geochemical Exploration 18, 220-232.

Afzal, P., Harati, H., Fadakar Alghalandis, Y., Yasrebi, A.B., 2013a. Application of spectrum–area fractal model to identify

PT

of geochemical anomalies based on soil data in Kahang porphyry-type Cu deposit, Iran. Chemie der ErdeGeochemistry 73, 533-543.‫‏‬

Afzal, P., Khakzad, A., Moarefvand, P., Rashidnejad Omran, N., Esfandiari, B., Fadakar Alghalandis, Y., 2010.

AC CE

Geochemical anomaly separation by multifractal modeling in Kahang (Gor Gor) porphyry system, Central Iran. Journal of Geochemical Exploration 104, 34–46. Agterberg, F. P., 2007. Mixtures of multiplicative cascade models in geochemistry. Nonlinear Processes in Geophysics 14, 201-209.

Agterberg, F.P., 1994. Fractals, multifractals and change of support. In: Dimitrakopoulos, R. (Ed.), Geostatistics for the Next Century. Kluwer, Dordrecht 223–234. Agterberg, F.P., 1996. Multifractal modeling of the sizes and grades of giant and supergiant deposits. International Geology Review 37, 1-8.‫‏‬ Agterberg, F.P., 2012. Multifractals and geostatistics. Journal of geochemical exploration 122, 13 –122. Agterberg, F.P., Cheng, Q., Brown, A., Good, D., 1996. Multifractal modeling of fractures in the Lac du Bonnet Batholith, Manitoba. Computers & Geosciences 22, 497–507. Aitchison, J., 1986. The statistical analysis of compositional data. Chapman and Hall, London. 416p. Aitchison, J., Greenacre, M., 2002. Biplots of compositional data. Applied Statistics 51, 375-392. Alavi, M., 1994. Tectonics of Zagros Orogenic belt of Iran, new data and interpretation. Tectonophysics 229, 211–238. Ali, E. H. M., Khakzad, A., Lotfi, M., Jafari, M. R., 2015. Prospecting of Ni mineralization using fractal models based on lithogeochemical data in Patang area, Eastern Iran. Arabian Journal of Geosciences 8, 9667-9677. Arias, M., Gumiel, P., Martín-Izard, A., 2012. Multifractal analysis of geochemical anomalies: a tool for assessing prospectivity at the SE border of the Ossa Morena Zone, Variscan Massif (Spain). Journal of Geochemical Exploration 122, 101-112. Bølviken, B., Stokke, P. R., Feder, J., Jössang, T., 1992. The fractal nature of geochemical landscapes. Journal of Geochemical Exploration 43, 91-109.

17

ACCEPTED MANUSCRIPT Buccianti, A., 2015. The FOREGS repository: Modelling variability in stream water on a continental scale revising classical diagrams from CoDA (compositional data analysis) perspective. Journal of Geochemical Exploration 154, 94 - 104. Buccianti, A., Grunsky, E., 2014. Compositional data analysis in Geochemistry: Are we sure to see what really occurs during natural processes? Journal of Geochemical Exploration 141, 1-5. Buccianti, A., Lima, A., Albanese, S., Cannatelli, C., Esposito, R., De vivo, B., 2015. Exploring topsoil geochemistry from

T

the CoDA (Compositional Data Analysis) perspective: The multi-element data archive of the Campania Region (Southern Italy). Journal of Geochemical Exploration 159, 302-316.

RI P

Carranza, E.J.M., 2008. Geochemical anomaly and mineral prospectivity mapping in GIS. Handbook of Exploration and Environmental Geochemistry. 11. Elsevier, Amsterdam.

Carranza, E.J.M., 2009. Objective selection of suitable unit cell size in data-driven modeling of mineral

SC

prospectivity. Computers and Geosciences 35, 2032 – 2046.

Carranza, E.J.M., 2010. Catchment basin modeling of stream sediment anomalies revisited: incorporation of EDA and fractal analysis. Geochemistry: Exploration, Environment, Analysis 10, 365 – 381.

NU

Carranza, E.J.M., 2011. Analysis and mapping of geochemical anomalies using log-ratio transformed stream sediment data with concord values. Journal of geochemical Exploration 110, 167-185. Carranza, E.J.M., 2011. Analysis and mapping of geochemical anomalies using log-ratio transformed stream sediment data

MA

with concord values. Journal of geochemical Exploration 110, 167-185. Cheng, Q., 1997. Multifractal modeling and lacunarity analysis. Mathematical Geology 29, 919–932. Cheng, Q., 1999. Multifractality and spatial statistics. Computers & Geosciences 25, 949-961. Cheng, Q., 2006. GIS based fractal/multifractal anomaly analysis for modeling and prediction of mineralization and mineral

ED

deposits. In: Harris, J.R. (Ed.), GIS for the Earth Sciences. Geological Association of Canada, St. John's, Newfoundland 285–296.

Cheng, Q., 2007. Mapping singularities with stream sediment geochemical data for prediction of undiscovered mineral

PT

deposits in Gejiu, Yunnan Province, China.Ore. Geology Reviews 321, 314-324.‫‏‬ Cheng, Q., Agterberg, F.P., 1996. Multifractal modeling and spatial point processes. Mathematical Geology 27, 831–845. Cheng, Q., Agterberg, F.P., 2009. Singularity analysis of ore-mineral and toxic trace elements in stream sediments.

AC CE

Computers and Geosciences 35, 234– 244. Cheng, Q., Agterberg, F.P., Ballantyne, S.B., 1994. The separation of geochemical anomalies from background by fractal methods. Journal of Geochemical Exploration 54, 109 – 130. Cheng, Q., Xia, Q., Li, W., Zhang, S., Chen, Z., Zuo, R., Wang, W, 2010. Density/area power-law models for separating multi-scale anomalies of ore and toxic elements in stream sediments in Gejiu mineral district, Yunnan Province, China. Biogeosciences 7, 3019-3025.‫‏‬ Cheng, Q., Xu, Y., Grunsky, E., 1999. Integrated spatial and spectral analysis for geochemical anomaly separation. In: Lippard, S.J., Naess, A., Sinding-Larsen, R. (Eds.), Proceeding of the Fifth Annual Conference of the International Association for Mathematical Geology, Trondheim, Norway 6–11th August. Cheng, Q., Xu, Y., Grunsky, E., 2000. Integrated spatial and spectrum method for geochemical anomaly separation. Natural Resources Research 9, 43-52.‫‏‬ de Souza, J., Rostirolla, S.P., 2011. A fast MATLAB program to estimate the multifractal spectrum of multidimensional data: Application to fractures. Computers & Geosciences 37, 241-249. Egozcue, J.J., Pawlowsky-Glahn, V., Mateu-Figueras, G., Barceló-Vidal, C., 2003. Isometric logratio transformations for compositional data analysis. Mathematical Geology 35, 279–300. El-Makky, A.M., 2011. Statistical analyses of La, Ce, Nd, Y, Nb, Ti, P, and Zr in bedrocks and their significance in geochemical exploration at the Um Garayat Gold Mine Area, Eastern Desert, Egypt. Natural Resources Research 20, 157-176. Evertsz, C.J.G., Mandelbrot, B.B., 1992. Multifractal measures (Appendix B). In: Peitgen, H.O., Jurgens, H., Saupe, D. (Eds.), Chaos and Fractals. Springer, Verlag, New York. Feder, J., 1988. Fractals. Plenum Press, New York.

18

ACCEPTED MANUSCRIPT Filzmoser, P., Hron, K., Reimann, C., Garrett, R., 2009. Robust factor analysis for compositional data. Computers and Geosciences 35, 1854-1861. Ghorbani, M., 2013. The economic geology of Iran: mineral deposits and natural resources. Springer Science and Business Media.‫‏‬ Grassberger, P., Procaccia, I., 1983. Measuring the strangeness of strange attractors. Physica D 9, 189–208.

T

Guillou, Y., Maurizot, P., Vaslet, D., de la Villeon, H., 1980. Geological map of Iran, 1:100, 000 sheet. Geological Survey of Iran, Tehran.

RI P

Halsey, T.C., Jensen, M.H., Kadanoff, L.P., Procaccia, I., Shraiman, B.I., 1986. Fractal measures and their singularities: the characterization of strange sets. Physical Review A 33, 1141.

Halls, C., Zhao, R., 1995. Listvenite and related rocks: perspectives on terminology and mineralogy with reference to an

SC

occurrence at Cregganbaun, Co. Mayo, Republic of Ireland. Mineralium Deposita 30, 303-313. Hawkes, H.E., Webb, J.S., 1962. Geochemistry in Mineral Exploration. Harper and Row, New York. He, F. P., Wang, Z., Fang, C., Wang, L., Geng, X., 2014. Identification and assessment of Sn-polymetallic prospects in the

NU

Gejiu western district, Yunnan (China). Journal of Geochemical Exploration 145, 106-113. Helvoort, P. J., Filzmoser, P., Van Gaans, P. F., 2005. Sequential factor analysis as a new approach to multivariate analysis of heterogeneous geochemical datasets: an application to a bulk chemical characterization of fluvial deposits (Rhine–

MA

Meuse delta, The Netherlands). Applied Geochemistry 20, 2233-2251.‫‏‬ Hengl, T., 2006. Finding the right pixel size. Computers and Geosciences 32, 1283-1298. Kaiser, H.F., 1958. The varimax criteria for analytical rotation in factor analysis. Psychometrika 23, 187–200. Kumru, M. N., Bakac, M., 2003. R-mode factor analysis applied to the distribution of elements in soils from the Aydın

ED

basin, Turkey. Journal of Geochemical Exploration 77, 81-91. Levinson, A. A., 1979. Introduction to Exploration Geochemsitry. Second Edition, Applied Pub. Luz, F., Mateus, A., Matos, J.X., Gonçalves, M.A., 2014. Cu- and Zn-soil anomalies in the NE Border of the South

Research 23, 195-215.

PT

Portuguese Zone (Iberian Variscides, Portugal) identified by multifractal and geostatistical analyses. Natural Resources

Macklin, M.G., Ridgway, J., Passmore, D.G., Rumsby, B.T, 1994. The use of overbank sediment for geochemical mapping

AC CE

and contamination assessment: results from selected Welsh flood plains. Applied Geochemistry 9, 698–700. Mandelbrot, B.B, 1983. The Fractal Geometry of Nature (updated and augmented edition). Freeman, New York 495. Mokhtari, A. R., Rodsari, P. R., Fatehi, M., Shahrestani, S., Pournik, P., 2014. Geochemical prospecting for Cu mineralization in an arid terrain-central Iran. Journal of African Earth Sciences 100, 278-288. Mokhtari, A., Nezhad, S.G., 2015. A modified equation for the downstream dilution of stream sediment anomalies. Journal of Geochemical Exploration 159, 185-193. Panahi, A., Cheng, Q., 2004. Multifractality as a measure of spatial distribution of geochemical patterns. Mathematical Geology 36, 827–846. Parsa, M., Maghsoudi, A., Yousefi, M., Sadeghi, M. 2016a. Prospectivity modeling of porphyry-Cu deposits by identification and integration of efficient mono-elemental geochemical signatures. Journal of African Earth Sciences 114, 228-241. Parsa, M., Maghsoudi, A., Yousefi, M., Sadeghi, M., 2016b. Recognition of significant multi-element geochemical signatures of porphyry Cu deposits in Noghdouz area, NW Iran. Journal of Geochemical Exploration 165, 111-124. Parsa, M., Maghsoudi, A., Ghezelbash, R., 2016c. Decomposition of anomaly patterns of multi-element geochemical signatures in Ahar area, NW Iran: A comparison of U-spatial statistics and fractal models. Arabian journal of Geosciences 9, 1-16. Pison, G., Rousseeuw, P.J., Filzmoser, P., Croux, C., 2003. Robust factor analysis. Journal of Multivariate Analysis 84, 145172. Reimann, C., de Caritat, P., Team, G.P., Team, N.P., 2012. New soil composition data for Europe and Australia: Demonstrating comparability, identifying continental-scale processes and learning lessons for global geochemical mapping. Science of the Total Environment 416, 239–252.

19

ACCEPTED MANUSCRIPT Reimann, C., Filzmoser, P., 2000. Normal and lognormal data distribution in geochemistry: death of a myth. Consequences for the statistical treatment of geochemical and environmental data. Environmental geology 39, 1001-1014. Reimann, C., Filzmoser, P., Garrett, R. G., 2002. Factor analysis applied to regional geochemical data: problems and possibilities. Applied Geochemistry 17, 185-206. Sadeghi, M., Billay, A., Carranza, E.J.M., 2015. Analysis and mapping of soil geochemical anomalies: implications for

T

bedrock mapping and gold exploration in Giyani area, South Africa. Journal of Geochemical Exploration 154,184-193. Sadeghi, M., Morris, G. A., Carranza, E. J. M., Ladenberger, A., Andersson, M., 2013. Rare earth element distribution and

RI P

mineralization in Sweden: An application of principal component analysis to FOREGS soil geochemistry. Journal of Geochemical Exploration 133, 160-175.

Sadid, S., 2011. Generation of Ni and Au prospection in Birjand area, eastern Iran. Geological survey of Iran (in Persian).

SC

Shuguang, Z., Kefa, Z., Yao, C., Jinlin, W., Jianli, D., 2015. Exploratory data analysis and singularity mapping in geochemical anomaly identification in Karamay, Xinjiang, China. Journal of Geochemical Exploration 154, 171-179.‫‏‬ Sinclair, A. J., 1974. Selection of threshold values in geochemical data using probability graphs. Journal of Geochemical

NU

Exploration 3, 129–149.

Sinclair, A.J., 1976.Applications of probability graphs in mineral exploration. 4. Associations of Exploration Geochemists. Spadoni, M, 2006. Geochemical mapping using a geomorphologic approach based on catchments. Journal of Geochemical

MA

Exploration 90, 183 – 196.

Spadoni, M., Voltaggio, M., Cavarretta, G, 2005. Recognition of areas of anomalous concentration of potentially hazardous elements by means of a subcatchment-based discriminant analysis of stream sediments. Journal of Geochemical Exploration 87, 83-91.‫‏‬

ED

Stanley, C.R., Sinclair, A.J., 1989.Comparison of probability plots and gap statistics in the se le ct ion of threshold for exploration geochemistry data. Journal of Geochemical Exploration 32,‫‏‬355 –357. Stocklin, J., 1975.On the origin of ophiolite complexes in the southern Tethys region. Tectonophysics 25, 303–322.

PT

Sun, X., Deng, J., Gong, Q., Wang, Q., Yang, L., Zhao, Z., 2009. Kohonen neural network and factor analysis based approach to geochemical data pattern recognition. Journal of Geochemical exploration 103, 6-16. Sun, X., Gong, Q., Wang, Q., Yang, L., Wang, C., Wang, Z., 2010. Application of local singularity model to delineate

AC CE

geochemical anomalies in Xiong'ershan gold and molybdenum ore district, Western Henan province, China. Journal of Geochemical Exploration 107, 21-29.‫‏‬ Swets, J.A., 1988. Measuring the accuracy of diagnostic systems. Science 240, 1285–1293. Thompson, M., Howarth, R.J., 1976. Duplicate analysis in geochemical practice. Part 1: theoretical approach and estimation of analytical reproducibility. Analyst 101, 690–698. Treiblmaier, H., Filzmoser, P., 2010. Exploratory factor analysis revisited: How robust methods support the detection of hidden multivariate data structures in IS research. Information and Management 47, 197-207. Tripathi, V.S., 1979. Factor analysis in geochemical exploration. Journal of Geochemical Exploration, 11, 263-275. ‫‏‬Turkey, J .W. , 1977. Exploratory data analysis. Reading 7 Addison-Wesley. Wang, W., Zhao, J., Cheng, Q., 2011. Analysis and integration of geo-information to identify granitic intrusions as exploration targets in southeastern Yunnan district, China. Computers and Geosciences 37, 1946– 1957. Wang, W., Zhao, J., Cheng, Q., Liu, J. 2012. Tectonic–geochemical exploration modeling for characterizing geo-anomalies in southeastern Yunnan district, China. Journal of Geochemical Exploration 122, 71-80. Xiao, F., Chen, J., Zhang, Z., Wang, C., Wu, G., Agterberg, F. P., 2012. Singularity mapping and spatially weighted principal component analysis to identify geochemical anomalies associated with Ag and Pb-Zn polymetallic mineralization in Northwest Zhejiang, China. Journal of Geochemical Exploration 122, 90-100.‫‏‬ Xie, S., Bao, Z., 2004. Fractal and multifractal properties of geochemical fields. Mathematical Geology 36, 847–864. Yousefi, M., Carranza, E. J. M., 2015a. Fuzzification of continuous-value spatial evidence for mineral prospectivity mapping. Computers & Geosciences 74, 97–109. Yousefi, M., Carranza, E. J. M., 2015b. Prediction-area (P-A) plot and C-A fractal analysis to classify and evaluate evidential maps for mineral prospectivity modeling. Computers & Geosciences 79, 69-81.

20

ACCEPTED MANUSCRIPT Yousefi, M., Carranza, E. J. M., 2016. Data-Driven Index Overlay and Boolean Logic Mineral Prospectivity Modeling in Greenfields Exploration. Natural Resources Research 25, 3-18. Yousefi, M., Carranza, E.J.M., Kamkar-Rouhani, A., 2013. Weighted drainage catchment basin mapping of stream sediment geochemical anomalies for mineral potential mapping. Journal of Geochemical Exploration 128, 88-96. Yousefi, M., Kamkar-Rouhani, A., Carranza, E.J.M., 2012. Geochemical mineralization probability index (GMPI): a new

mineral potential mapping. Journal of Geochemical Exploration 115, 24-35.

T

approach to generate enhanced stream sediment geochemical evidential map for increasing probability of success in

RI P

Yousefi, M., Kamkar-Rouhani, A., Carranza, E.J.M., 2014. Application of staged factor analysis and logistic function to create a fuzzy stream sediment geochemical evidence layer for mineral prospectivity mapping. Geochemistry: Exploration, Environmental, Analysis 14, 45-58.

SC

Yuan, F., Li, X., Zhou, T., Deng, Y., Zhang, D., Xu, C., Zhang, R., Jia, C., Jowitt, S.M., 2015. Multifractal modelling-based mapping and identification of geochemical anomalies associated with Cu and Au mineralisation in the NW Junggar area of northern Xinjiang Province, China. Journal of Geochemical Exploration 154, 252-264.

NU

Zuo, R, 2011. Identifying geochemical anomalies associated with Cu and Pb–Zn skarn mineralization using principal component analysis and spectrum–area fractal modeling in the Gangdese Belt, Tibet (China). Journal of Geochemical Exploration 111, 13-22.

MA

Zuo, R., 2012. Exploring the effects of cell size in geochemical mapping. Journal of Geochemical Exploration 112, 357-367.‫‏‬ Zuo, R., Cheng, Q., Xia, Q., Agterberg, F.P., 2009. Application of singularity mapping technique to identify local anomalies using stream sediment geochemical data, a case study from Gangdese, Tibet, western China. Journal of Geochemical Exploration 101, 225–235.

ED

Zuo, R., Wang, J., 2015. Fractal/multifractal modeling of geochemical data: A review. Journal of Geochemical Exploration 164, 33-41.

Zuo, R., Wang, J., Chen, G., Yang, M., 2015. Reprint of “Identification of weak anomalies: A multifractal perspective”.

PT

Journal of Geochemical Exploration 154, 200-212. Zuo, R., Xia, Q., Wang, H., 2013. Compositional data analysis in the study of integrated geochemical anomalies associated

AC CE

with mineralization. Journal of Geochemical Exploration 28, 202-211.

21

ACCEPTED MANUSCRIPT Table captions Table 1. Rotated component matrix of robust factor analysis. Significant loadings (bolded values) are selected based on the absolute threshold values of 0.5. Table 2. Descriptive statistics of the raw geochemical data of Ni, Cr and Co. All the concentrations are in ppm. Table 3. Geochemical populations derived by the C-A fractal method along with their regression coefficient

RI P

Table 4. Multifractal parameters of Ni, Cr and Co in the study area.

T

(R2), regression error and regression equation.

Table 5. True positives (TP), true negatives (TN), false positives (FP), false negative (FN), Type I error (T I), Type II error (T II) and the overall accuracy (OA) of concentration–area (C-A) fractal model in delineation

SC

of geochemical populations of Ni, Cr and Co.

Table 6. True positives (TP), true negatives (TN), false positives (FP), false negative (FN), Type I error (T I),

AC CE

PT

ED

MA

geochemical populations of Ni, Cr and Co.

NU

Type II error (T II) and the overall accuracy (OA) of local singularity multifractal model in delineation of

22

ACCEPTED MANUSCRIPT Table 1. Rotated component matrix of robust factor analysis. Significant loadings (bolded values) are selected based on the absolute threshold values of 0.5.

F2

F3

0.871

0.123

0.234

Co

-0.821

-0.342

0.337

Cr

-0.653

0.413

Cu

-0.219

0.867

-0.154

Ni

-0.818

-0.438

-0.312

Pb

0.298

0.217

0.597

Sb

0.84

0.376

0.232

Zn

0.211

-0.223

0.879

Var.

54

16.1

15.5

Cum. Var.

54

70.1

85.6

NU

SC

RI P

T

F1

As

-0.393

AC CE

PT

ED

MA

Element

23

ACCEPTED MANUSCRIPT

Cr

Co

Minimum

13

34

5

Maximum

965

1259

68

Mean

170.27

247.93

21.35

Std. dev.

170.671

211.965

11.716

Skewness

2.048

2.472

1.380

Kurtosis

4.268

6.709

2.143

RI P

Ni

SC

Element

T

Table 2. Descriptive statistics of the raw geochemical data of Ni, Cr and Co. All the concentrations are in ppm.

AC CE

PT

ED

MA

NU

Std. dev. Standard deviation

24

ACCEPTED MANUSCRIPT Table 3. Geochemical populations derived by the C-A fractal method along with their regression coefficient (R2), regression error and regression equation.

Regression equation y= -12.483x + 96.242 y = -1.945x + 29.257 y = -0.674x + 22.633 y = -0.220x + 20.894 y = -13.143x + 104.91 y = -1.541x + 27.356 y = -0.104x + 20.574 y = -15.749x + 77.301 y = -3.383x + 29.765 y = -0.945x + 22.235 y = -0.491x + 21.115

NU

SC

RI P

T

Regression error (MSE) 0.003 0.002 0.004 0.009 0.001 0.004 0.009 0.003 0.003 < 0.001 0.008

MA

Co

Regression coefficient (R2) 0.97 0.99 0.97 0.79 0.98 0.98 0.73 0.98 0.99 0.99 0.86

ED

Cr

High anomaly Anomaly Background Low background High anomaly Anomaly Background High anomaly Anomaly Background Low background

Range (in ppm) 570 - 949 210 - 570 51 – 210 13- 51 789 – 1239 249 – 789 36 – 249 45 – 67 21 – 45 12 – 21 5 - 12

PT

Ni

Population

AC CE

Element

25

ACCEPTED MANUSCRIPT

αmin

1.986

1.986

1.992

αmax

2.04

2.021

2.006

α0

1.997

1.999

1.998

ΔαL

0.053

0.012

0.005

ΔαR

0.011

0.02

0.008

Δα

0.042

0.033

0.013

R

-0.582

-0.24

-0.207

f (α)max

1.328

1.327

1.327

f (α)min

1.161

1.242

1.296

Δf (α)

0.166

0.084

0.031

τ'' (1)

-0.041

-0.033

-0.004

RI P

Co

SC

Cr

NU

Ni

AC CE

PT

ED

MA

Element

T

Table 4. Multifractal parameters of Ni, Cr and Co in the study area.

26

ACCEPTED MANUSCRIPT Table 5. True positives (TP), true negatives (TN), false positives (FP), false negative (FN), Type I error (T I), Type II error (T II) and the overall accuracy (OA) of concentration – area (C-A) fractal model in delineation of geochemical populations of Ni, Cr and Co. FP 1458 1446 1246

FN 2419 2179 2574

TN 8271 8511 8116

TI 0.461 0.457 0.394

T II 0.226 0.203 0.240

OA 0.738 0.720 0.724

PT

ED

MA

NU

SC

RI P

T

TP 1701 1713 1913

AC CE

Element Ni Cr Co

27

ACCEPTED MANUSCRIPT Table 6. True positives (TP), true negatives (TN), false positives (FP), false negative (FN), Type I error (T I), Type II error (T II) and the overall accuracy (OA) of local singularity multifractal model in delineation of geochemical populations of Ni, Cr and Co. FP 1056 865 175

FN 2174 1892 3670

TN 8625 8798 7020

TI 0.333 0.273 0.055

T II 0.201 0.176 0.343

OA 0.8 0.768 0.722

PT

ED

MA

NU

SC

RI P

T

TP 2106 2294 2984

AC CE

Element Ni Cr Co

28

ACCEPTED MANUSCRIPT Figure captions Figure 1. Major structural zones in Iran (after Alavi, 1994), outcropping ophiolitic rocks and location of the study area.

T

Figure 2. Generalized geological map of the study area (after Guillou et al., 1980) and stream sediment sample

Figure 3. Biplots of robust factor 1 versus robust factor 2. Figure 4. Histogram of the raw geochemical data for Ni, Cr and Co.

RI P

locations.

Figure 5. Stretched maps showing the dispersion patterns of (a) Ni, (b) Cr and (c) Co.

SC

Figure 6. C-A log-log plots and the delineated geochemical populations of (a, b) Ni, (c, d) Cr and (e, f) Co. Figure 7. Log-log plots showing the partition functions {χ(q)} versus different cell sizes (ε) for (a) Ni, (b) Cr

NU

and (c) Co.

Figure 8. Multifractal spectrum curves {f (α)} of Ni, Cr and Co.

Figure 9. Stretched maps showing the dispersion patterns of singularity indices {(α)} of (a) Ni, (b) Cr and (c)

MA

Co.

Figure 10. Binary maps showing the anomalous geochemical populations based on the singularity indices of α < 2 for (a) Ni, (b) Cr and (c) Co.

ED

Figure 11. (a, b, c) Nickel mineralization within listwanitic-altered rocks and (d) a general view of mineralized

AC CE

PT

outcrops in listwanitic-altered rocks of the study area.

29

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC CE

PT

ED

Fig. 1

30

AC CE

Fig. 2

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

31

Fig. 3

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

32

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

Fig. 4

33

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

Fig. 5

34

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

Fig. 6 35

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

Fig. 7

36

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC CE

PT

Fig. 8

37

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

Fig. 9

38

AC CE

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

Fig. 10

39

PT

AC CE

Fig. 11

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

40

ACCEPTED MANUSCRIPT Highlights

PT

ED

MA

NU

SC

RI P

T

Stream sediment data were applied for prospecting Ni mineralization. Mineralization-related elements were identified by robust factor analysis. Regional anomalies were recognized by the concentration area fractal model. Local enrichments were manifested by the singularity analysis.

AC CE

   

41