Bull. Min. Res. Exp.

(2019) 160: 177-195

BULLETIN OF THE
MINERAL RESEARCH AND EXPLORATION
Foreign Edition 2019 160 ISSN : 0026-4563

Bulletin of the Mineral

E-ISSN : 2651-3048

CONTENTS
Syn-sedimentary tectonic markings in the Oligocene Datça-Kale-Ac Göl basin, Western Anatolia
.............................................................Gülşen ELMAS, Gürol SEYİTOĞLU, Nizamettin KAZANCI and Veysel IŞIK/ Research Article 1

Research and Exploration

Sedimentological properties and depositional environments of the Holocene sequence in Yenikap, Istanbul
.................................................................................Meltem SEZERER BULUT, M. Namk YALÇIN and Oya ALGAN/Research Article 21

Paleoenvironmental features and ostracod investigation of paleogene-Neogene sequences in Babaeski- Lüleburgaz- Muratli-Çorlu Region
(Southeastern Thrace, Turkey)
.......................................................................................................................................................................Ümit ŞAFAK/Research Article 45

Petrological and geochemical features of Biga Peninsula Granitoids, NW Anatolia, Turkey

................................................................................................. Ümit AYDIN, Pnar ŞEN, Öner ÖZMEN and Erdal ŞEN/Research Article 81

The gas hydrate potential of the Eastern Mediterranean Basin

...................................................................................................................Şükrü MEREY and Sotirios Nik. LONGINOS/Research Article 117

Rare earth elements and yttrium geochemistry of the geothermal elds in the Eastern Black Sea Region (Ordu, Rize, Artvin), NE Turkey
..................................................................Esra HATİPOĞLU TEMİZEL, Fatma GÜLTEKİN and Arzu FIRAT ERSOY/Research Article 135

Appearance features of clayey mixtures having y ashes

..................................................................................Fatma DAĞCI, Nazl İpek KUL GÜL and Niyazi Uğur KOÇKAL/Research Article 155

Statistical approach by factor and cluster analysis on origin of elements from The Hamit Plutonic (Turkey) rock samples
.................................................Füsun YALÇIN, Daniel G. NYAMSARI, Nurdane İLBEYLİ and Rifat BATTALOĞLU/Research Article 163

Mineral resource estimation using a combination of drilling and IP-Rs data using statistical and cokriging methods
............................................................................................................. Kamran MOSTAFAEİ and Hamidreza RAMAZİ/Research Article 177

Landslide susceptibility mapping using information value and frequency ratio for the Arzew sector (North-Western of Algeria)
............................................................................................ Roukh ZINE EL ABIDINE1 and Nadji ABDELMANSOUR/Research Article 197

Geological structures mapping using aeromagnetic prospecting and remote sensing data in the karstic massif of Beni Mellal Atlas, Morocco

http://bulletin.mta.gov.tr
......................................................Ikram BOUTIRAME, Ahmed BOUKDIR, Ahmed AKHSSAS and Ahmed MANAR/Research Article 213

Simulation of a salt dome using 2D linear and nonlinear inverse modeling of residual gravity eld data
................................................................................................................Soheyl POURREZA and Farnush HAJIZADEH/Research Article 231
Nummulites sireli Deveciler (junior homonym of N. sireli Alan) renamed as Nummulites ercumenti nom. nov.
......................................................................................................................................................................... Ali DEVECİLER/Discussion 245

ACNOWLEDGEMENT ................................................................................................................................................................................ 247

Bulletin of the Mineral Research and Exploration Notes to the Authors ..................................................................................................................... 249

Mineral resource estimation using a combination of drilling and IP-Rs data using statistical and cokriging
methods

Kamran MOSTAFAEIa* and Hamidreza RAMAZIa

Faculty of Mining and Metallurgical Engineering, Amirkabir University of Technology (Tehran Polytechnic), Iran.
a

Research Article

Keywords: ABSTRACT
Correlation, Cu grade, The aim of this research is the mineral resource estimation using a combination of drilling and IP-Rs
IP-Rs, Statistical methods,
data. Therefore the approach of this paper is to study the correlation of induced polarization (IP) and
Cokriging.
Electrical resistivity (Rs) data with drilling data in order to grade estimation and mineral resource
estimation. Reducing the boreholes number and optimization of the boreholes location is another aim
of this research. The Abassabad copper mine located in Miami-Sabzevar mineralization belt northeast
Iran was chosen as a case study. Within the borehole locations, geophysical profiles were designed
and surveyed. After IP-Rs data inversion, 2D sections were prepared. The 3D block models of IP-Rs
were constructed by geostatistical methods. The correlation between IP-Rs and drilling data were
examined by statistical and geostatistical methods using regression, multivariate regression analysis,
and cokriging. Based on the mentioned methods copper grade was estimated and the 3D block models
of Cu grade were constructed. Obtained models were checked and compared with real Cu model
compiled according to drilling data which was done after geophysical measurements. Results showed
that the regression between IP data and Cu grade was more appropriate with least error. Rs data are not
suitable for Cu estimation, due to changing intervals which led to increasing estimation error. Based
Received Date: 10.07.2018 on the suggestions of this paper, we could reduce the number of boreholes to 30% of the initial number
Accepted Date: 13.11.2018 and optimize the boreholes locations.

1. Introduction studies results, such as boreholes and exploratory

drifts, which are located in a grid. (Wang Q. et al.,
Ore reserve modelling and estimation play crucial 2011).
part in mineral exploration and exploration planning
and require an extensive database that make it a costly There are numerous researches regarding ore
and time-consuming process. Mineral exploration modeling and grade estimation including ore
is a complex process which is carried out through reserve estimation by fuzzy modeling based on the
integration of different methods. Building a 3D model spatial variability (Tutmez et al., 2007); ore reserve
of a deposit is the main goal of mineral exploration. estimation using fractal methods in China (Wang Q.
The main issue is to acquire more information in et al., 2010) using Wavelet Neural Network (WNN)
less time and by lower costs. Grade estimation is the and Artificial Neural Network (ANN) methods for
most important phase which can have a major effect mineral deposit evaluation (Li et al., 2010); grade
on mining feasibility and its future management estimation and reserve evaluation for Iron ore deposit
(Tahmasebi and Hezarkhani, 2012). Ore modeling and in Iran (Shademan et al., 2013); grade estimation and
reserve estimation was done according to exploration modeling using support vector machine methods (Li
Citation info: Mostafaei, K., Ramazi, H. 2019. Mineral resource estimation using a combination of drilling and IP-Rs data using statistical
and cokriging methods. Bulletin of the Mineral Research and Exploration, 160, 177-195. http://dx.doi.org/10.19111/
bulletinofmre.502794.
* Corresponding author: Kamran MOSTAFAEI, [email protected]

177
 Bull. Min. Res. Exp. (2019) 160: 177-195

et al., 2013), combination of geostatistical and fractal parameters by geoelectrical data; II) estimation of
methods in order to assessment of mineral resource grade and chemical parameters using well logging
in the Tongshan porphyry copper deposit (Wang G. et geophysical data and III) evaluation of lithology and
al., 2013) and management of resources and reserve dynamic parameters using geophysical methods that
in Brazil (Seccatore et al., 2014). It is noted that grade used in geotechnical investigations.
estimation and ore reserve estimation of these studies
were based on drilling results. The attempt of this research was to combine
geophysical induced polarization (IP) and resistivity
Geophysical methods especially IP-Rs methods (Rs) and drilling data to enhance copper ore modeling
are widely used in earth sciences including mineral and grade estimation. In contrast of drilling, IP-Rs
exploration, engineering geology, environmental surveying is fast, continuous and cheap so it can
studies and etc. We can be mentioned some case cover more areas. Therefore combination of IP-Rs
as follow; Gold-silver deposit exploration by IP in and drilling, can save time and cost in exploration.
Russia (Gurin et al., 2015), Uranium mineralization Therefore, in the first stage of this study, IP-Rs surveys
detection in India (Biswas and Sharma, 2016), were carried out in the available borehole locations
Bitumen exploration in Iran using IP-Rs (Mashhadi et and then inversion and modeling of IP-Rs data was
al., 2017), detection deeply buried cave in Spain using conducted. After that the relationship between grade
combination of IP-Rs and gravity (Martinez-Moreno and IP-Rs has been checked and established in the
et al., 2013), sinkhole investigation in urban area borehole locations using statistical methods including
using ERT and GPR (Sevil et al., 2017), investigation regression and multivariate regression analyses (MRV)
of coal washing waste pile in Iran by geoelectrical and cokriging. Finally, the 3D block model of the
methods (Jodeiri et al., 2014) , investigation of landfill deposit was constructed that used for grade prediction
leaching plume using 2D and 3D ERT (Maurya et al., and ore reserve calculation in the all of the study area.
2017). Studying the spatial relationships of IP and Cu on one
hand and applying the spatial regression to predicting
In mineral exploration, geophysical surveys are the Cu grade based on the other regionalized variable
predominantly carried out for anomaly separation and is a new method which has been assessed in one case
delineation of geological structures. However, there study.
are some studies about grade and reserve estimation
by geophysical methods including copper grade 2. Methods
estimation in blast holes using prompt gamma neutron
activation analysis (PGNAA) in Chuquicamata copper 2.1. Induced Polarization and Resistivity (IP-Rs)
mine in Chile (Charbucinski et al., 2003), investigation
of organic pollutions effect on IP-Rs measurements In mineral exploration, geophysical methods are
in laboratory based on its results detection of used for measurement of physical features of bodies or
pollution zone in Aveiro, Portugal (Martinho and rocks, also especially, for identification of differences
Almeida, 2006), ore reserve estimation by VES and between studied targets. Geophysical exploration can
chemical analyses (Ehinola et al., 2009), correlation be used to detection the mineralization properties by
between geoelectrical data and aquifer parameters measuring physical properties directly (Gadallah and
in evaluation of ground water potential (Batte et al., Fisher, 2009). Recently, application of geophysical
2010), estimation of Ni grade using crosshole seismic exploration is increasing due to optimization in
velocity tomography in Canada (Perozzi et al., 2012), cost and time (Mostafaie and Ramazi, 2015). The
coal quality estimation using borehole geophysical base of geophysical methods is the identifying the
data (Webber et al., 2013), reserve estimation of contrasts of physical properties in materials. (Telford
limestone and sand using geoelectrical data (Ushie et al., 1990). Geophysical method(s) selection for a
et al., 2014) and predicting the pyrite oxidation and mineral deposit exploration depended on physical
transport process in coal waste pile using resistivity properties of mineral target and its accompanied rocks
methods in Iran (Jodeiri et al., 2016), At large, geological setting, and also its topography. Integrated
these studies include three subjects: I) estimation of geophysical methods are commonly used in mineral
hydraulic parameters of aquifer and hydrogeological exploration to obtain qualified results and more

178
 Bull. Min. Res. Exp. (2019) 160: 177-195

certain results (Mandal et al., 2013; Biswas et al., IP-Rs studies. For more information on CRSP array
2014; Mandal et al., 2015). Electrical resistivity (Rs) please refer to Ramazi and Mostafaie, 2013; Ramazi
and induced polarization (IP) are the most commonly and Jalali, 2014; Mostafaie and Ramazi, 2015; Amini
used and the oldest subset of geophysical exploration and Ramazi, 2016a, b.
(White et al., 2003; Dahlin and Loke, 2015). The
measured parameter in geoelectrical surveys is 2.2. Regression
apparent resistivity that shows by Rs (Loke and
Dahlin, 2002). The base of IP method is the study of Regression is a subset of statistical methods that
secondary electric fields that generated in ground by used to estimate the relationship between variables.
electric currents. IP observation are performed in time Regression has several techniques for modeling and
or frequency domains. In the time domain, the decay analyzing variables that focus on identifying the
voltage is measured as a function of time while in the relationship between a dependent variable and one
frequency domain, apparent resistivity p, is measured or more independent variables (Howarth, 2001).
at two or more frequencies, generally below 10 Hz. Regression techniques are widely used for prediction,
The measured parameter in IP survey is apparent and also used to understand which among the
chargeability (Dahlin et al., 2002). independent variables are related to the dependent
variable, and to explore the forms of these relationships
There are many arrays in IP-Rs surveying, (Armstrong, 2012).
but dipole - dipole, pole - dipole, Wenner and
Schlumberger arrays are the most commonly used. According to data and research goal, there are
Dipole - dipole array that is more conventional in several types of regression such as linear, logistic,
profiling is sensitive to horizontal resistivity variations polynomial, ridge and etc., (Faul et al., 2009). Linear
but very low sensitivity to vertical variations. It also regression is the simplest type of the regressions
has low depth of investigation compared to other where there are two correlated random variables; X
conventional arrays like Wenner and Schlumberger. and Y. The outcomes generate a cloud of data in the
The other disadvantage of dipole - dipole array is that plane X-Y and we want to determine the best affine
its pseudo-sections could be very different from the function Y=aX+b that fits the observations (Howarth,
real geological structures (Loke, 2015).Wenner and 2001). In most cases, including this study, linear
Schlumberger arrays cannot be used for profiling. regression dose not satisfactory results so we should
Although the conventional arrays (e.g. Dipole-Dipole, use the nonlinear regression that include several
pole- Dipole, Wenner, and Schlumberger) have been methods. The polynomial models can be used in those
used in many mineral exploration studies successfully, situations where the relationship between variables
is curvilinear. Sometimes a nonlinear relationship
in some topographical and geological conditions
in a small range of explanatory variable can also be
especially in thin and high dip-angle mineralized
modeled by polynomials (Helsel and Hirsch, 2002).
vines, these arrays may not lead to satisfying results
This model depends on the number of variables. For
(Mostafaie and Ramazi , 2015).
one variable the polynomial model is given by eq.1.
Regarding the mentioned factors and problems,
Y=B0+B1x1+B2x2 +…+Bn xn (eq.1)
Ramazi and Mostafaei (2013) tried to design a new
array, which lacks some of the mentioned problems,
Where
provides a good image of the subsurface, and can
be used practically in the field, especially for one Y; dependent variable, x; independent variable, B0
channel measurement systems. The designed array is to Bn; constant number
named CRSP that is the abbreviation of (Combined
Resistivity Sounding and Profiling). CRSP is a 2.3. Multivariate Regression Analyses (MRA)
combination of sounding and profiling which
can lead to good results in mentioned geological Functions obtained from regression analyses are
conditions (Ramazi and Mostafaie, 2013). This array usually used to describe the relationship between
(CRSP array) has successfully been applied in many response and predictor variables (Chiou et al., 2016).
exploration and/or site investigation projects using Multivariate regression analysis is a method with one

179
 Bull. Min. Res. Exp. (2019) 160: 177-195

dependent and many independent variables (Braglia et techniques considering the spatial correlation between
al., 2012; Zhang and Goh, 2016). In this method, it is data. Spatial correlation data, including distance and
assumed that a dependent variable (Y) is expressed as direction, can be expressed in mathematical form,
the function of independent variable (Xi): considering the spatial structure. Spatial structure
is studied by means of a variogram in geostatistics
Y=f (Xi) (Mostafaie et al., 2014). Geostatistical methods was
developed for estimation of the regionalized variables
The regression type depends on the type of such as grade in ore body at a known location.
obtained function so that if the function was linear the Regionalized variables are variables of a phenomenon
regression is called linear and if the function was non- defined in space (and/or time) that possesses a certain
linear the regression is called non- linear (Granian et structure (Kumar et al., 2007). Kriging is an optimal
al., 2015).
interpolation based on regression against observed z
The general form of the model is as follows: values of surrounding data points, weighted according
to spatial covariance values. Cokriging is kriging
(eq.2) using information from one or more correlated
secondary variables, or multivariate kriging in general
That in this function y is dependent variable, x1 to (Bohling, 2005). The cokriging procedure is a natural
xn are independent variables and a0 to an are regression extension of kriging when a multivariate variogram or
coefficient. In linear regression analysis, the regression covariance model and multivariate data are available
coefficients are calculated by least square method. In (Wackernagel, 2003). Cokriging and kriging are subset
linear regression analysis, correlation coefficient (R2) of the geostatistical methods used for estimation and
can be obtained from the following equation: interpolation. These methods are generalized form
of univariate and multivariate linear regression for
estimation at a point, an area or within a volume.
Cokriging methods are used to take advantage of the
covariance between two or more regionalized variables
that are related, and are appropriate when the main
In this equation Ŷi is calculated value of ith sample
attribute of interest is sparse, but related secondary
of dependent variable, Ȳ is mean of the variable and
information is abundant (Deutsch and Journel, 1998).
Yi is the value of ith sample of the dependent variable
(Braglia et al., 2012). If R2 is close to 1, it means the The information available on a natural phenomenon
result is desirable if R2 is closed to zero this means
is rarely limited to the values assumed by a single
that dependent and independent variables have not
variable over set of sample points. In the most real
correlation (Mogaji, 2016).
studies (especially earth science) involve more
than on variable, so we have to use the multivariate
Multivariate regression has been shown to suitable
generalization of kriging which is named cokriging
model for various earth sciences studies. Over the
(Chiles and Delfiner, 2012). Using auxiliary variables
past few years, multivariate regression analysis was
we can improve the precision of a main variable.
used widely in different earth sciences for predicting
auxiliary variables usually are cheap in measuring
various objectives (e.g., Noori et al., 2010 ; Khanlari et
so we can reduce the number of observations for
al, 2012; Mokhtari, 2014; Habibi et al., 2014; Granian
et al., 2015). Therefore, multivariate linear regression the main variable that it is expensive and needed for
has been used to obtain the relationship between IP, Rs optimizing the interpolations (Knotters, et al., 1995).
and copper grade in the Abassabad copper mine. A variable of interest is cokriged at a specific location
from data about itself and about auxiliary variable in
2.4. Cokriging the neighborhood. The data set may not cover all data
variables at all sample locations. Ordinary cokriging
Geostatistics aims at providing quantitative requires at least on data value about the variable
descriptions of natural variables distribute in space of interest, while simple cokriging, relying on its
or in time and space (Chiles and Delfiner, 2012). knowledge of the mean, can be performed with data
Geostatistics is the application of statistical estimation solely about the auxiliary variables (Wackernagal,

180
 Bull. Min. Res. Exp. (2019) 160: 177-195

2003). As mentioned before, cokriging method multi-collected neighborhood, full neighborhood

consisted of one primary and one secondary variable. (Chiles and Delfiner, 2012, Madani and Emery, 2018).
Moreover, spatial structure should be studied in any In collected cokriging only retained secondary data
geostatistical method. So in the cokriging analyses, are the ones available at the target location (Madani
the cross semi-variogram (or cross-variogram) should and Emery, 2018). The collected cokriging and full
be determined in prior (Hooshmand, et al., 2011). cokriging have been used in this study.
There are some method for coregionalization model
investigation such as: linear model of coregionalization For more information about Cokriging in ore
(LMC); Markov-type model; intrinsic linear model modeling and grade estimation refer to Knotters et al.,
(Madani and Emery, 2018). The linear model of 1995; Boezio et al., 2011 and Xu et al., 2015.
coregionalization (LMC) is the one of the approach to
simultaneously model direct and cross variogram in a 3. Case Study-Abassabad Copper Mine
multivariate setting. An LMC is suitable in cokriging
and cosimulation (Leuangthong et al., 2008, Goulard The Abassabad cooper mine is an active mine that
and Voltz, 1992). is located 120 km East of Shahrud, Semnan Province,
Northeastern Iran. This mine is located 10 km north
The measurements available for different variables of Abassabad village, named Abassabad copper mine.
in given domain may be located either at the same The location map are presented in figure 1.
sample points or at different points for each variables.
Complete heterotopy: the variables have been From the geological point of view, Abassabad is
measured on different sets of sample points and have a part of a wide mineralization belt named Miami-
no sample locations in common. Partial heterotopy: Sabzevar copper belt. There are many lithology units
some variables share some sample locations. Isotopy: in all of the Miami-Sabzevar copper belt including
data is available at all sampling points (Wackernagal, 2 major geological units: igneous rocks (porphyritic
2003). The sampling is partially heterotopic in andesite and tracchyandesite) and sedimentary rocks
this study. By considering the data located in (limestone with marl). Mineralization has occurred
a neighborhood of the target location and data in the contact of andesite and limestone. The geology
distribution situation, the search strategies to select map of Abassabad was presented in figure 2. Based
neighboring data will be selected. There are some on the geology map there are some lithology units
search strategies including: collected neighborhood, include porphyritic tracchyandesite, conglomerate,

Figure 1- Location of Abassabad copper mine (case study) in Iran.

181
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 2- Geological map of Abbasabad copper mine (study area) and IP-Rs profiles location.

limestone, sandstone and shale. Quaternary alluvium Based on the MRA results, Cu grade was estimated
occur in some locations. Sedimentary rocks consist and the 3D model of Cu grade was obtained. Based on
of limestone, and in some parts clay minerals were the obtained model, 7 boreholes were proposed. The
converted to marl. Mineralogy studies shows that accuracy of grade estimation and prepared models
limestone is a geochemical barrier to settle the copper was checked out by new exploratory boreholes. Then
content. Limestone caused deposition of copper prepared models -using regression and MRA methods-
minerals so mineralization has occurred in the contact were reviewed and remodeled by adding new drilling
of andesite and limestone. According various studies data. For more accuracy and comparison, cokriging
such as; to mineralogy, petrology, alteration and was used. Relationship of IP values and Cu grade was
economic geology, this deposit is Manto-type copper examined and Cu grade was estimated and then, 3D
deposit (Salehi et al., 2016). model of Cu was prepared using cokriging. Finally
estimated Cu grade and obtained Cu models were
compared. Prepared models based on the IP-Rs data
4. Results
and Cu grade were compared with real block model
of Cu grade.
In this research, the integration of IP-Rs data and
grade data for construction of 3D block model and
4.1. Geophysical Results
resource estimation was investigated. For this purpose,
the Abassabad copper mine was selected which In the first stage, location of profiles were detected
is an active copper mine in the Semnan Province, then data surveying was done. Data inversion was
Northeastern Iran. In the first step, geophysical done by Res2Dinv and the 2D imaging of profiles is
surveying was performed in the exploratory prepared. The 3D model of data was prepared based
boreholes. Then, inversion of obtained data carried on geostatistical methods. Also for the determination
out using Res2Dinv software package and 2D sections of the IP-Rs data thresholds, statistical methods and
were prepared. Based on the geostatistical methods, fractal methods were used. The result of 3D modeling
3D block model of IP-Rs were prepared in the study was checked out by geostatistical methods and drilling
area. In the second step, the relationship of IP-Rs results.
data and the copper grade was checked out. Based
4.1.1. Designing and Surveying Data
on correlation between IP and Cu grade obtained by
regression, Cu grade was estimated in all of the study Based on the exploratory borehole plan, the IP-Rs
area and the 3D model of Cu grade was prepared. profiles were designed. Geophysical profile position
Then the relationship of IP-Rs data and Cu grade were was chosen in a way that it covers most of exploration
calculated by multivariate regression analyses (MRA). boreholes (Figure 3). According to the mineralization

182
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 3- Map of study area in Abbasabad mine and the location of available BH and geophysical profiles also proposed BH after
geophysical modeling.

type, mine situation and surveying conditions, CRSP 4.1.2. Inversion Results
(combined resistivity sounding and profiling) array
was selected for this investigation. CRSP array is a As mentioned earlier in this paper, IP-Rs data
combined array which can lead to useful results in inversion was done in the first stage resulted in 2D
the various topographical and geological conditions IP-Rs imaging. The resistivity and IP data sets were
(Ramazi and Mostafaie, 2013). Therefore, IP-Rs data inverted using the RES2DINV software. To prepare
IP-Rs sections, the resistivity and IP data sets were
have been surveyed by using CRSP array. Generally,
inversed by the Newton and Gauss–Newton methods,
4 profiles were designed and surveyed. Three profiles
from the RES2DINV software package ( Loke and
P1, P2 and P3 are parallel. P1 and P2 were surveyed
Dahlin, 2002). In the inversion process number of
along borehole profiles with 10 meters electrode
iterations were 5. The best iteration was selected
spacing. P3 and P4 were surveyed for the evaluation
based on the RMS error and geological situations.
of IP-Rs results, therefore along these profiles there
The RMS error level is between 3.2% to 5.5% for
are no exploration boreholes. P3 was located 50
IP data, and RMS error is between 8% to 12.5% for
meters south of P1 and P2, and is also parallel to them. resistivity data. The inversion results determined
The electrode spacing of P3 is 20 meters. P4 is located as IP and resistivity sections shown by the surfer
500 meters from the western part of the others by 10 software. Inversion results and compiled sections
m electrode spacing. Finally, 2000 points were read are presented in figure 4 for profiles 1 to 4. Anomaly
in the length of all profiles. In each point, resistivity values were defined based on the fractal methods.
and induced polarization in time domain have been Due to the data, the “concentration–area” method
recorded. Also the drilling data of 20 boreholes was (Ferdows and Ramazi, 2015) has been used in this
available. research for separation of the anomalous value from
the background. The obtained threshold correlated
Data acquisition has been carried out according to with drilling results and the obtained threshold was
designed plan field in 09/01/2017 to 09/22/2017. In modified. The threshold of the inversed IP data is
order to check the data quality, several measurements equal to 25 mV/V and the threshold of inversed Rs
were randomly repeated in the field. The surveyed data is 250 Ωm. According to the obtained results
data was revised, the data accuracy was checked out, of the geophysical study and those correlated with
and then data processing was done. the drilling data, it may be say that resistivity and IP

183
 Bull. Min. Res. Exp. (2019) 160: 177-195

C D
Figure 4- Inverted resistivity (Rs) and induced polarization (IP) sections with topography; A) Profile1, B) Profile2, C)Profile3, D) Profile4.

section could detect mineralization zones so well. In a 3D model, we can investigate the area between the
the profile locations, the mineralization was traced profiles and also all of the study area.
with acceptable accuracy by a high value of resistivity
4.1.3. 3D Modeling
and IP. The efficiency of IP-Rs in the Abassabad
copper mine is high and this investigation reduced and After inversion and preparation of 2D sections
optimized the drilling operation efficiently. For further presented in figure 4, the 3D models of IP-Rs data were
investigation and optimization of the exploration obtained based on the geostatistical methods. In this
boreholes, we need a 3D model of the study area. By study, data of P1, P2 and P3 were used for modeling,

184
 Bull. Min. Res. Exp. (2019) 160: 177-195

because these profiles were parallel and the distance related software such as (SGeMS) (Bohling, 2007)
between them was less. As mentioned, P4 is located experimental variogram for data were calculated
500 meters from the western part of the other profiles, and presented scientifically. Variogram for various
so we cannot use it in modelling. In geostatistics, parameters such as different azimuth and dip were
the variables with spatial structure are investigated. calculated. The appropriate theoretical models based
In the other word, spatial structure is essential for on the least square differences were fitted to the
using geostatistical methods. The variogram is a variogram (Figures 5 to7).
fundamental tool in geostatistics for investigating
Variogram models and parameters for maximum
spatial structure because it provides critical
range; median range and minimum range that are
parameters for various Kriging estimators. Accuracy
perpendicular each other and essential for modeling
of the proposed parameters from the variogram are of are presented in table 1 briefly. The angels of
crucial importance and can have significant positive or anisotropy were obtained for X, Y and Z respectively,
negative influence on the estimated blocks (Mostafaie -30, 0 and 120.
et al., 2014). In order to study the spatial structure,
the data were reviewed and variography was carried In the next step, 3D modeling was done based on
out. According to the above factors and applying the the obtained parameters of the variography. Datamine

Figure 5- Variogram model for minimum range of data (azth: 330, dip: 30).

Figure 6- Variogram model for median range of data (azth: 150, dip: 60).

185
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 7- Variogram model for maximum range of data (azth: 60, dip: 0).

Table 1- Obtained parameters of presented variogram.

Azimuth Dip Model Range(m) Sill(mv/v)2 Nugget effect(mv/v)2
330 30 Spherical 18 50 10
150 60 Spherical 50 50 0
60 0 Spherical 102 50 0

Studio3 software package was used to prepare 3D and drilled (Figure 3). Drilling data confirmed the
models. The input data of Datamine Studio3 software geophysical results. In the drilling planning of this
were two-dimensional modeling results obtained mine, boreholes were drilled at a distance of 30
from inversion. As we know the output of inversion meters. By geophysical modelling the mineralization
including X, Y, Z, IP-Rs for each point that were zone and it’s characterizes including; depth, thickness,
prepared to input dataset as continuous data according continuity were determined. Geophysical modelling
to the setting of Datamine database. The input data showed that mineralization zone has an acceptable
were composited and the composite length was continuity. So the boreholes are not needed at a distance
selected 5 meters. Then, the 3D block models of IP- of 30 meters, and using the boreholes at intervals
Rs were prepared by Datamine Studio software. The more than 60 meters we can obtained required data to
two models include induced polarization (IP) and mineralization identify. Therefore using geophysical
Resistivity (Rs) are presented as a 3D block model in results we were able to reduce the borehole number
figures 8 and 9, respectively. It should be noted that significantly
the block size were identical and equal to 5 meters
in the whole of the model. In these models, IP-Rs According to drilling results, the database includes
parameter distribution has been shown with high boreholes and geophysical models was constructed.
accuracy and the mineralization zone is marked in In order to better correlation the drilling and IP-Rs
different directions. Maximum value of the IP is 45 data combined and composited. After data composite,
mv/v and the maximum value of resistivity is 850 Ωm. dataset was selected to correlation studying (Table 2).
The threshold of IP is 25 mv/v and the Rs threshold
is 250 Ωm. IP-Rs models show that high values of 4.2. Regression Results
IP-Rs are the anomalous values. In the other word,
anomaly value is the location with high value of the As mentioned before, one of the main goal of
IP-Rs. Based on the 2D section and the 3D model this research is to estimate Cu grade based on the
of geophysical results, 7 boreholes were proposed IP-Rs data in order to use regression method. Firstly,

186
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 8- The results of modeling as a 3D block model for IP data in study area.

Figure 9- The results of modeling as a 3D block model for Rs data in study area.

Table 2- The statistical description of used data in this study.

Parameter Symbol No. of samples Max. Ave. Min. Std. dev.
Induce polarization IP 106 46.45 25.75 11.370521 7.91
Resistivity Rs 106 735.1 362.7 81.585678 179.5
Cu grade Cu 106 5.4 0.74 0.002 1.08

primary statistical analysis of data was performed. In The R-sq of equational.1 is about 75%, and the S
regression analysis, the relationship among variables is about 4.9% in this analyze that it is acceptable. It is
is examined. For this purpose, correlation between IP notable that S is the standard deviation of the distance
and Cu grade were checked out. So, IP was assumed between the data values and fitted values.
as an independent variable and Cu assumed as
dependent. Therefore there is a function; Y=f(X) that Based on the IP, copper reserve was estimated in
X= IP and Y= Cu. all of the study area according to equation1 and the 3D
model of estimated copper was constructed (Figure 11).
After checking the correlation between IP and Cu, In this copper mine, Cut of grade was considered as
the best equation was selected and the obtained result 0.1%, so the model consists of two group: values less
is presented in figure 10. As mentioned previously, than 0.1% and values more than 0.1% in the presented
there are several type of regression. So the polynomial models. The dimensions of the block model are selected
regression type is selected as the best type of regression 5*5*5 meters. Based on the block model, estimated Cu
and the obtained equation (eq.3) is as fallow; reserve is about 2.11 million ton (Table 3).
Y= 0.0043x2 0.13x | 1.002
The correlation between Rs and Cu grade was
where; X=IP, Y=Cu eq.3 checked out. The correlation coefficient between Rs

187
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 10- Diagram of Cu grade versus IP.

Figure 11- 3D model of the estimated Cu based on the regression results.

Table 3- Result summaries of estimated Cu using regression.

Min (%) Max (%) Cutoff grade (%) Mean (%) Block size(m) Estimated ore(million ton)
0.05 4.2 0.1 0.78 5x5x5 2.11

and Cu grade is less than 50% that is not a satisfactory examination of one dependent and more independent
value. Therefor its results were not used in copper variables relation. In this paper, IP-Rs values were
estimation. considered as independent variables and Cu grade were
considered as dependent variable in order to predict
4.3. Multivariate Regression Analyses (MRA) Result
the Cu grade by IP-Rs parameters using Minitab
software. As mentioned in the previous parts-2.2
In the previous part, we checked out the correlation
between IP and Cu grade, and now we want to regression- there are several type of regression, also in
investigate the correlation between IP-Rs and Cu the multivariate regression analyses, there are several
grade simultaneously. In the other word, we want to type of regression. After the studying these types
check the IP-Rs influence on Cu estimation together. polynomial type is selected. Then the best equation
So we used the Multivariate Regression Analyses with high R-Sq. was selected based on trial and error
(MRA), because as mentioned, MRA was used for method. The obtained equation is presented as eq.4.

188
 Bull. Min. Res. Exp. (2019) 160: 177-195

Cu (%) = 5587 + 6.80 IP - 0.120 Rs - 6002 (Rs) 0.1 - and Cu is more than correlation between Rs and Cu,
0.143 (IP) 2 - 51.0 Ln (IP) + 620 Ln (Rs) + 0.00130 also the Rs data variation range is more than IP which
(IP) 3 + 288 (Rs) 0.3 (eq.4) causes an increase in estimation error level. Therefore
Correlation Coefficient = 67.3% Rs data were not used in cokriging estimation. In this
paper, primary variable was Cu grade and secondary
Based on the eq.2 the Cu grade in the study variable was IP data. Due to abundance of IP data,
area was estimated and the Cu 3D block model was we attempted to estimate Cu grade using IP data. As
prepared (Figure 12). The Cu reserve was estimated mentioned in cokriging cross-variogram is necessary,
about 2.46 million ton (Table 4). so variography was done based on the linear model
of coregionalization (LMC). At first, variography of
We mentioned cokriging method as consisted of IP and Cu grade was performed. Subsequently, cross-
one primary and one secondary variable. The primary variogram of IP and Cu grade was calculated and
statistical analyses shoed that correlation between IP required parameters were obtained (Figures 13 to 15).

Figure 12- 3D model of estimated Cu based on the MRA results.

Table 4- Results summaries of estimated Cu using MLR.

Min (%) Max (%) Cutoff grade (%) Mean (%) Block size(m) Estimated ore(million ton)
0 8 0.1 1.43 5x5x5 2.46

Figure 13- Variogram model for IP data.

189
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 14- Variogram model for Cu data.

Figure 15- Cross-variogram model for IP and Cu.

The standardized variograms and their parameters (Figure 16) and the results presented in table 6. In
presented in (Table 5). the estimation maximum and minimum of data equal
to 3 and 15 respectively and also the search radius
Then according to the variogram parameters, considered 60 meters in the cokriging estimation
Cu grade was estimated and the 3D block model process. Ordinary type of cokriging (OK) was used
of Cu grade was constructed by SGeMS software for estimation. The secondary data (IP data) covers

Table 5- The summaries of used parameters in cokriging based on the standardized variograms.
Experimental semivariogram Fitted model sill Range(m) Nugget effect
IP Spherical 1 150 0.18
Cu Spherical 1 65 0.35
Cross Spherical 1 185 2.6

190
 Bull. Min. Res. Exp. (2019) 160: 177-195

Figure 16- The 3D block model of Cu estimation using cokriging.

Table 6- Results summaries of estimated Cu using cokriging.

Min (%) Max (%) Cutoff grade (%) Mean (%) Block size (m) Estimated ore (million ton)
0 4.44 0.1 0.92 5*5*5 1.85

the whole domain because we want to estimate the Cu Rs were prepared. IP-Rs results were compared with
grade based on the 3D block model of IP data (Figure drilling results. Then in the locations that there is no
8 in 4.1.3). Given that the secondary data is more than boreholes, 2 profiles were designed and surveyed.
primary data, so we used the full cokriging and also Also these data were inverted and sections were
collected cokriging with Markov Model 2 (MM2). prepared (see 4.1.2). In the next stage, 3D models of
The results of full cokriging that it was used all data IP-Rs data were prepared by geostatistical methods
are better and acceptable, so we presented the full and the correlation between Cu grade and IP-Rs was
cokriging results in this paper. For more details see examined. To checking obtained results, 7 exploratory
in section 4.2. Cokriging and presented explanations boreholes were proposed, and drilled. Drilling results
in (Wackernagal, 2003; Chiles and Delfiner, 2012; showed that the geophysical method could detect
Madani and Emery, 2018). anomalous region for Cu mineralization, significantly.
Moreover, the correlation between Cu grade and IP-
5. Discussion Rs is appropriate and applicable (see 4.1.2).

The Miami-Sabzevar copper belt is a wide Based on the drilling, the database includes 20
mineralization zone that it is currently under boreholes (available boreholes and drilled boreholes
reconnaissance and exploration. This vast areas of this after geophysical modeling) was constructed. The
belt entails high cost and time-consuming exploration correlation between Cu grade and IP-Rs were revised
activity. Thus the main goal of this research was to and investigated. For this purpose regression, MRA
devise an exploration plan to overcome this hurdle. and cokriging were used and according to the obtained
Geophysical methods are among the cheapest and correlation, Cu grade was estimated in the study area.
fastest so we decided to use geophysical methods The 3D block model of the estimated Cu grade was
combined with drilling results in ore modeling. For constructed (see; 4.2, 4.3, 4.4).
this purpose, the active mine of Madan Bozorg in the
As mentioned before the case study is an active
Miami-Sabzevar belt was selected as a case study.
copper mine, so the extraction block model based on
At first, the application of the IP-Rs methods the measured copper grade is available. The mineral
was investigated through devising survey profiles. In reserve estimation results according to the measured
the borehole locations 2 profiles were designed and copper grade presented briefly (Table.7). The obtained
surveyed. In this location, there were 20 boreholes. results showed that the estimated copper is about 1.95
Obtained data was inverted and 2D sections of IP- million ton with 0.71% mean.

191
 Bull. Min. Res. Exp. (2019) 160: 177-195

Table 7- Results summaries of estimated Cu based on the measured copper grade in Abbasabad copper mine.
Min (%) Max (%) Cutoff grade (%) Mean (%) Block size (m) Estimated ore (million ton)
0.006 6.8 0.1 0.71 5*5*5 1.95

The obtained models based on the combination of of Rs data that increases error rate. However, Rs data
IP-Rs and drilling data were compared were compared can be used for geological prediction.
with actual Cu models, according to this we can say the
obtained results are acceptable.to better comparison, Results of this paper shows that the location and
we calculated the estimation error of obtained models number of additional boreholes can be optimized
over the actual model (Table 8). by the combination of geophysical data and drilling
results. We could reduce the number of boreholes and
Table 8- The error level of estimation methods over actual model. the cost of the exploration operation significantly.
Based on these results, we devised exploration plan for
No. Method Estimation error
1 Regression 8.2% the other areas of the Miami-Sabzevar mineralization
2 MLA 26.1% belt.
3 Cokriging 5.1%

References
Results showed that the model constructed by the
Amini, A., Ramazi, H. 2016a. Anomaly enhancement in
cokriging method is close to reality with minimum 2D electrical resistivity imaging method using
error. Beside cokriging method, the results of the a residual resistivity technique, the journal
regression method (correlation between IP and Cu) is of the Southern African Institute of Mining
better. Due to the smoothing effect of the cokriging, and Metallurgy, vol. 116, 1–8. http://dx.doi.
the estimated copper is less than the actual result. org/10.17159/2411-9717/2016/v116n2a7
The results of MRA methods are of most error due
Amini, A., Ramazi, H. 2016b. Application of Electrical
to consideration of the simultaneous effect of IP-Rs Resistivity Imaging for Engineering Site
values. The Rs data variation range is more than of Investigation, A Case Study on Prospective
IP data, so estimation of Cu grade entails more error Hospital Site,Varamin, Iran, Acta Geophysica,
rather than regression and cokriging methods. 64(4), 2200–2213. http://doi.org/10.1515/
acgeo-2016-0100.
Conclusion Armstrong, J. S. 2012. Illusions in regression analysis
International Journal of Forecasting, 28, 689–694
IP-Rs methods have been used successfully in the
copper detection in the Abassabad copper mine with Batte, A. G., Barifaijo. E., Kiberu. J. M., Kawule, W.,
highest efficiency. Through geostatistical methods, Muwanga, A., Owor, M., Kisekulo, J. 2010.
we were able to construct 3D models for IP-Rs data Correlation of geoelectric data with aquifer
that surveyed in 2D which were confirmed by drilling parameters to delineate the groundwater potential
of hard rock terrain in central Uganda. Pure and
results.
Applied Geophysics, 167(12), 1549–1559. http://
The number of boreholes could be decreased doi.org/10.1007/s00024-010-0109-x
from 20 to 7 in two places with acceptable results Biswas, A., Sharma, S. P. 2016. Integrated geophysical
by geophysical modeling that led to significant cost studies to elicit the structure associated with
saving. Uranium mineralization around South Purulia
Shear Zone, India: A Review. Ore Geology
Correlation between Cu grade and IP-Rs were Reviews, 72, 1307-1326.
examined and calculated using regression, MRA
Biswas, A., Mandal, A., Sharma, S. P., Mohanty, W. K.
and cokriging methods and based on them, Cu grade
2014. Delineation of subsurface structure using
was estimated and the 3D block model of Cu was
self-potential, gravity and resistivity surveys from
constructed. South Purulia Shear Zone, India: Implication
to uranium mineralization. Interpretation, 2(2),
Results proved that Cu estimation using IP data is T103-T110.
of better quality than Rs data, because of high variation

192
 Bull. Min. Res. Exp. (2019) 160: 177-195

Boezio, M. N. M., Costa, J. F. C. L., Koppe, J. C. 2011. function parameter for geoelectrical data ( case
Ordinary Cokriging of Additive Log-Ratios study : Hamyj copper deposit , Iran ). Journal of
for Estimating Grades in Iron Ore Deposits, Geophysics and Engineering, 12. 909–921.
Proceedings of the 4th International Workshopon
Gadallah, M. R., Fisher, R. 2009. Exploration geophysics.
Compositional Data Analysis 1–10.
Berlin. Springer. http://doi.org/10.1007/978-3-
Bohling, G. 2005. KRIGING, C&PE 940, 1–20. ttp:// 540-85160-8
people.ku.edu/~gbohling/cpe940. Goulard, M., Voltz, M. 1992 Linear Coregionalization
Bohling, G. 2007. S-GeMS Tutorial Notes in Model: Tools for Estimation and Choice of
Hydrogeophysics: Theory, Methods, and Cross-Variogram Matrix, Mathmatical Geology,
Modeling, Boise State University, Boise, Idaho, 24(3),269-286.
1–26.
Granian, H., Hassan, S., Asadi, H. H., John, E., Carranza,
Braglia, M., Carmignani, G., Frosolini, M., Zammori, F. M. 2015. Multivariate regression analysis of
2012. Data classification and MTBF prediction lithogeochemical data to model subsurface
with a multivariate analysis approach. Reliability mineralization: a case study from the Sari Gunay
Engineering and System Safety, 97(1), 27–35. epithermal gold deposit, NW Iran. Journal of
http://doi.org/10.1016/j.ress.2011.09.010 Geochemical Exploration, 148, 249–258. http://
doi.org/10.1016/j.gexplo.2014.10.009
Charbucinski, J., Malos, J., Rojc, A., Smith, C. 2003.
Prompt gamma neutron activation analysis Gurin, G., Tarasov, A., Ilyin, Y., Titov, K. 2015. Application
method and instrumentation for copper grade of the Debye decomposition approach to analysis
estimation in large diameter blast holes. Applied of induced-polarization profiling data (Julietta
Radiation and Isotopes, 59, 197–203. http://doi. gold-silver deposit, Magadan Region) Russian
org/10.1016/S0969-8043(03)00163-5 Geology and Geophysics 56, 1757–1771. http://
doi.org/10.1016/j.rgg.2015.11.008
Chiles, J.P., Delfiner, P. 2012 Geostatistics: modeling
Spatial Uncertainty, Hoboken, N.J., Wiley. Habibi, M. J., Mokhtari, A. R., Baghbanan, A., Namdari,
S. 2014. Journal of Petroleum Science and
Chiou, J., Yang, Y., Chen, Y. 2016. Multivariate functional
Engineering Prediction of permeability in dual
linear regression and prediction. Journal of
fracture media by multivariate regression analysis.
Multivariate Analysis, 146, 301–312. http://doi.
Journal of Petroleum Science and Engineering,
org/10.1016/j.jmva.2015.10.003
120, 194–201. http://doi.org/10.1016/j.
Dahlin, T., Loke, M. H. 2015. Negative apparent chargeability petrol.2014.06.016
in time-domain induced polarisation data. Journal
Helsel, B. D. R., Hirsch, R. M. 2002. Hydrologic Analysis
of Applied Geophysics, 123, 322–332. http://doi.
and Interpretation; Chapter A3 Statistical
org/10.1016/j.jappgeo.2015.08.012.
Methods in Water Resources. Techniques of
Dahlin, T., Leroux, V., Nissen, J. 2002. Measuring techniques Water-Resources Investigations of the United
in induced polarisation imaging. Journal of States Geological Survey.
Applied Geophysics, 50(3), 279–298. http://doi.
Hooshmand, A., Delghandi, M., Izadi, A., Aali, K. A. 2011.
org/10.1016/S0926-9851(02)00148-9.
Application of kriging and cokriging in spatial
Deutsch, C.V., Journel, A.G. 1998. GSLIB: Geostatistical estimation of groundwater quality parameters,
Software Library and User’s Guide, second African Journal of Agricultural Research, 6(14),
edition. Oxford, UK: Oxford University Press. 3402–3408.
Ehinola, O. A., Oladunjoye, M. A., Gbadamosi, T. O. 2009. Howarth, R. J. 2001. A History of Regression and Related
Chemical composition , geophysical mapping and Model-Fitting in the Earth Sciences (1636?
reserve estimation of clay deposit from parts of -2000), Natural Resources Research 10(4),241-
Southwestern Nigeria, Journal of Geology and 286.
Mining Research , 1(3), 57–66.
Jodeiri, B., Ramazi, H., Faramarz, S., Ardejani, D.,
Faul, F., Erdfelder, E., Buchner, A., Lang, A. G. 2009. Moradzadeh, A. 2014. Integrated Time-Lapse
Statistical power analyses using G*Power 3.1: Geoelectrical–Geochemical Investigation at a
Tests for correlation and regression analyses. Reactive Coal Washing Waste Pile in Northeastern
Behavior Research Methods, 41 (4), 1149–1160. Iran, NE Iran, Mine Water and the Environment,
http://doi.org/10.3758/BRM.41.4.1149. 33: 256-265.
Ferdows, M. S., Ramazi, H. 2015. Application of the Jodeiri, B., Faramarz, S., Ardejani, D., Ramazi, H.,
fractal method to determine the membership Moradzadeh, A. 2016. Predicting pyrite oxidation

193
 Bull. Min. Res. Exp. (2019) 160: 177-195

and multi-component reactive transport processes Shear Zone, India. Journal Geological Society of
from an abandoned coal waste pile by comparing India, 82(6), 601-606.
2D numerical modeling and 3D geoelectrical
Mandal, A., Mohanty, W.K., Sharma S.P., Biswas, A., Sen,
inversion, International Journal of Coal Geology,
J., Bhatt, A. K. 2015. Geophysical signatures
164: 13-24
of uranium mineralization and its subsurface
Khanlari, G. R., Heidari, M., Momeni, A. A., Abdilor, Y. validation at Beldih, Purulia District, West Bengal,
2012. Prediction of shear strength parameters India: a case study.Geophysical Prospecting,
of soils using arti fi cial neural networks and 63(2), 713-726. https://doi.org/10.1111/1365-
multivariate regression methods. Engineering 2478.12205.
Geology, 131–132, 11–18. http://doi.
Martinho, E., Almeida, F. 2006. 3D behaviour of
org/10.1016/j.enggeo.2011.12.006
contamination in landfill sites using 2D
Knotters, M., Brus, D. J., Voshaar, J. H. O. 1995. A resistivity/IP imaging: Case studies in Portugal.
comparison of kriging , co-kriging and kriging Environmental Geology, 49(7), 1071–1078.
combined with regression for spatial interpolation http://doi.org/10.1007/s00254-005-0151-7
of horizon depth with censored observations,
Martínez-Moreno, F.J., Pedrera, A., Ruano, P., Galindo-
Geoderma, 67, 227–246.
Zaldívar, J., Martos-Rosillo, S., González-
Kumar, D., Ahmed, S., Krishnamurthy, N. S., Dewandel, B. Castillo, L., Sánchez-Úbeda, J.P., Marín-Lechado,
2007. Reducing ambiguities in vertical electrical C. 2013. Combined microgravity, electrical
sounding interpretations: A geostatistical resistivity tomography and induced polarization
application, Journal of Applied Geophysics, to detect deeply buried caves: Algaidilla cave
62(1), 16–32. http://doi.org/10.1016/j. (Southern Spain). Engineering Geology, 162, 67-
jappgeo.2006.07.001 78, doi: 10.1016/j.enggeo.2013.05.008.
Leuangthong, O., Daniel Khan, K., Deutsch, C.V. 2008. Mashhadi, S.R., Mostafaei, K., Ramazi, H. 2017. Improving
Solved problem in geostatisitcs, Chapter 9: bitumen detection in resistivity surveys by
Multiple Variable. using induced polarisation data, Exploration
Li, X., Xie, Y., Guo, Q., Li, L. 2010. Adaptive ore grade Geophysics. https://doi.org/10.1071/EG17032.
estimation method for the mineral deposit Mogaji, K. A. 2016. Geoelectrical parameter-based
evaluation. Mathematical and Computer multivariate regression borehole yield model
Modelling, 52(11–12), 1947–1956. http://doi. for predicting aquifer yield in managing
org/10.1016/j.mcm.2010.04.018. groundwater resource sustainability. Integrative
Li, X., Li, L., Zhang, B., Guo, Q. 2013. Neurocomputing Medicine Research, 10(4), 584–600. http://doi.
Hybrid self-adaptive learning based particle org/10.1016/j.jtusci.2015.12.006
swarm optimization and support vector regression
Mokhtari, A. R. 2014. Hydrothermal alteration mapping
model for grade estimation. Neurocomputing,
through multivariate logistic regression
118, 179–190. http://doi.org/10.1016/j.
analysis of lithogeochemical data. Journal of
neucom.2013.03.002.
Geochemical Exploration, 145, 207–212. http://
Loke, M.H., 2015, Tutorial: 2-D and 3-D electrical imaging doi.org/10.1016/j.gexplo.2014.06.008
surveys: (Revision date: 17th October 2015),
Mostafaie, K., Ramazi, H. 2015. Application of electrical
(www.geotomosoft.com)
resistivity method in sodium sulfate deposits
Loke, M. H., Dahlin, T. 2002. A comparison of the exploration, case study: Garmab, Iran. Journal of
Gauss – Newton and quasi-Newton methods in Biodiversity and Environmental Sciences, 6(2),
resistivity imaging inversion, Journal of Applied 2220–6663. Retrieved from http://www.innspub.
Geophysics, 49, 149–162. net
Madani, N., Emery, X. 2018. A comparison of search Mostafaie, K., Ramazi, H. R., Jalai, M. 2014. Application
strategies to design the cokriging neighborhood for of Integrated Geophysical and Geostatistical
predicting coregionalizaed variables, Stochastic Methods in Amiriyeh Site Classification.
Environmental Research and Risk Assessment, Geodynamics Research International Bulletin
https://doi.org/10.1007/s00477-018-1578-1 (GRIB), (2) 2,1-15.
Mandal, A., Biswas, A., Mittal, S., Mohanty, W. K., Sharma, Maurya, P.K., Ronde, V.K., Fiandaca, G., Balbarini, N.,
S. P. Sengupta, D., Sen, J., Bhatt, A. K. 2013. Auken, E., Bjerg, L.P., Christiansen, A.V. 2017
Geophysical anomalies associated with uranium Detailed landfill leachate plume mapping using
mineralization from Beldih mine, South Purulia 2D and 3D electrical resistivity tomography- with

194
 Bull. Min. Res. Exp. (2019) 160: 177-195

correlation to ionic strength measured in screens. Telford, W. M., Geldart, L. P., Sheriff, R. E. 1990.
Journal of Applied Geophysica, 138:1-8. Applied Geophysics. Cambridge University
Press, Cambridge. http://doi.org/10.1180/
Noori, R., Khakpour, A., Omidvar, B., Farokhnia, A. 2010.
minmag.1982.046.341.32
Expert Systems with Applications Comparison
of ANN and principal component analysis- Tütmez, B., Tercan, A. E., Kaymak, U. 2007. Fuzzy
multivariate linear regression models for Modeling for Reserve Estimation Based on Spatial
predicting the river flow based on developed Variability. Mathematical Geology, 39(1),87-111.
discrepancy ratio statistic. Expert Systems With http://doi.org/10.1007/s11004-006-9066-4
Applications, 37(8), 5856–5862. http://doi.
Ushie, F., Harry, T., Affiah, U. 2014. Reserve Estimation
org/10.1016/j.eswa.2010.02.020
from Geoelectrical Sounding of the Ewekoro
Perozzi, L., Gloaguen, E., Rondenay, S., McDowell, G. Limestone at Papalanto , Ogun State , Nigeria ,
2012. Using stochastic crosshole seismic velocity Journal of Energy Technologies and Policy 4(5),
tomography and Bayesian simulation to estimate 28–33.
Ni grades: Case study from Voisey’s Bay, Canada.
Wackernagal, H. 2003. Mutivariate Geostatistics; an
Journal of Applied Geophysics, 78, 85–93. http://
introduction with applications, Springer sciences.
doi.org/10.1016/j.jappgeo.2011.06.036
Wang, G., Pang, Z., Boisvert, J. B., Hao, Y., Cao, Y., Qu,
Ramazi, H., Mostafaie, K. 2013. Application of integrated
J. 2013. Quantitative assessment of mineral
geoelectrical methods in Marand (Iran)
resources by combining geostatistics and
manganese deposit exploration. Arabian Journal
fractal methods in the Tongshan porphyry Cu
of Geosciences, 6(8), 2961–2970. http://doi.
deposit ( China ). Journal of Geochemical
org/10.1007/s12517-012-0537-2
Exploration, 134, 85–98. http://doi.org/10.1016/j.
Ramazi, H., Jalali, M. 2014. Contribution of geophysical gexplo.2013.08.004
inversion theory and geostatistical simulation
Wang, Q., Deng, J., Liu, H., Yang, L., Wan, L., Zhang, R.
to determine geoelectrical anomalies. Stud.
2010. Fractal models for ore reserve estimation.
Geophys. Geod. 59: 97–112.
Ore Geology Reviews, 37(1), 2–14. http://doi.
Salehi, L., Rasa, I., Alirezaei, S., Kazemi Mehrnia, A. 2016. org/10.1016/j.oregeorev.2009.11.002
The Madan Bozorg, volcanic-hosted copper
Wang, Q., Deng, J., Liu, H., Wang, Y., Sun, X., Wan, L.
deposit, East Shahroud; an example of Manto type
2011. Fractal models for estimating local reserves
copper deposits in Iran, Journal of Geoscience,
with different mineralization qualities and spatial
25(98): 93-104.
variations. Journal of Geochemical Exploration,
Seccatore, J., Marin, T., Tomi, G. De, Veiga, M. 2014. 108(3), 196–208. http://doi.org/10.1016/j.
A practical approach for the management of gexplo.2011.02.008
resources and reserves in Small-Scale Mining.
Webber, T., Costa, J. F. C. L., Salvadoretti, P. 2013. Using
Journal of Cleaner Production, 84, 803–808.
borehole geophysical data as soft information
http://doi.org/10.1016/j.jclepro.2013.09.031
in indicator kriging for coal quality estimation.
Sevil, J., Gutierrez, F., Zarroca, M., Desira, G., Carbonela, International Journal of Coal Geology, 112, 67–
D., Guerrero, J., Linares, R., Roque, C., Fabregat, 75. http://doi.org/10.1016/j.coal.2012.11.005
I. 2017. Sinkhole investigation in an urban area
White, R. M. S., Collins, S., Loke, M. H. 2003. Resistivity
by trenching in combination with GPR, ERT and
and IP arrays, optimised for data collection and
high-precision leveling. Manteled evaporate karst
inversion. Exploration Geophysics, 34(4), 229.
of Zaragoza city, NE Spain. Engineering Geology
http://doi.org/10.1071/EG03229
213: 9-20
Xu, H., Sun, J., Russell, B., Innanen, K. 2015. Porosity
Shademan, Kh M., Madani,H., Hassani, H., Moarefvand,
prediction using cokriging with multiple
P. 2013. Determining the Best Search
secondary datasets, CREWES Research Report,
Neighbourhood in Reserve Estimation , using
27, 1–13.
Geostatistical Method: A Case Study Anomaly
No 12A Iron Deposit in Central Iran, Journal Zhang, W., Goh, A. T. C. 2016. Geoscience Frontiers
Geological Society of India , 81(12), 581–585. Multivariate adaptive regression splines and
neural network models for prediction of pile
Tahmasebi, P., Hezarkhani, A. 2012. A hybrid neural
drivability. Geoscience Frontiers, 7(1), 45–52.
networks-fuzzy logic-genetic algorithm for grade
http://doi.org/10.1016/j.gsf.2014.10.003.
estimation. Computers and Geosciences, 42, 18–
27. http://doi.org/10.1016/j.cageo.2012.02.004

195