- Email: [email protected]
Application of geochemical zonality coefficients in mineral prospectivity mapping
Computers & Geosciences 37 (2011) 1935–1945
Contents lists available at ScienceDirect
Computers & Geosciences journal homepage: www.elsevier.com/locate/cageo
Application of geochemical zonality coefficients in mineral prospectivity mapping Mansour Ziaii a,n, Emmanuel John M. Carranza b, Mahdi Ziaei a a b
Faculty of Mining, Petroleum, and Geophysics, Shahrood University of Technology, Iran Department of Earth Systems Analysis, Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, The Netherlands
a r t i c l e i n f o
abstract
Article history: Received 2 September 2010 Received in revised form 20 April 2011 Accepted 25 May 2011 Available online 15 June 2011
Indices of geochemical zonality (Vz) of multielements around mineral deposits and their spatial associations with particular geological, geochemical, and structural factors are critical aspects that must be considered in mineral exploration. Values of Vz indices allow distinction between sub-ore and supra-ore anomalies, which are associated with outcropping and blind deposits, respectively. In this paper, we used a map of a Vz index (Zn*Pb/Cu*Ag) in weights-of-evidence analysis of regional-scale prospectivity for porphyry–Cu deposits in the area covered by 1:100,000 scale map sheet of Jebal-Barez (Kerman province, southern Iran). For comparison, we used a Cu map instead of the Vz map in the weights-of-evidence (WofE) analysis. The Vz-in-WofE prospectivity model outperforms the Cu-in-WofE prospectivity model. In fact, prior to writing this paper, blind porphyry–Cu mineralization was intersected at depth by borehole exploration in a high prospectivity zone delineated by the Vz-in-WofE model. The results demonstrate the usefulness of the Vz-in-WofE for regional-scale targeting of blind mineral deposits. & 2011 Elsevier Ltd. All rights reserved.
Keywords: Geochemical zonality Sub-ore anomaly Supra-ore anomaly Weights of evidence GIS Jebal-Barez (Iran)
1. Introduction In the former Soviet Union (FSU), metallometric methods have been developed for mapping geochemical anomalies associated with mineral deposits (Ivanov and Meituv, 1972; Shcheglov, 1979). Many of those methods led to successful exploration results in the FSU, China, and other countries (Khorin et al., 1992). During the period 1981–1985, 238 mineral deposits were discovered by using suitable metallometric methods (Guanghua et al., 1996). In particular, the method of productivity mapping efficiently differentiates between mineralized and nonmineralized zones (Beus and Grigorian, 1977; Solovov, 1987, 1990; Grigorian, 1992; Moon, 1999; Carranza, 2004a). This method, although widely used in the FSU and China, did not attract significant attention in other countries. Most metallometric methods have also not been widely practiced by using geographic information systems (GIS). Most metallometric methods have been developed for stream sediment geochemical surveys, which are generally not useful for exploration in the following situations: (1) presence of significant heavy metal contamination of streams due to mining activities; (2) low relief areas where hydromorphic dispersion predominates over mechanical dispersion of metals from certain mineral deposits. Yet, the geochemical zonality (Vz) method, which is among the metallometric methods developed in the FSU for analysis of lithogeochemical data, does not have those limitations.
n
Corresponding author. Fax: þ98 273 333 5509. E-mail address: [email protected] (M. Ziaii).
0098-3004/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.cageo.2011.05.009
However, it has also not attracted significant attention in many countries outside the FSU because it has not been demonstrated yet in conjunction with GIS-based mapping of mineral prospectivity. Several methods currently exist for GIS-based mapping mineral prospectivity, whereby various relevant evidential data layers are integrated (Bonham-Carter, 1994; Pan and Harris, 2000; Carranza, 2008). The most widely used of those methods is weights-ofevidence (WofE) analysis (Bonham-Carter et al., 1989; Agterberg et al., 1990). Raines and Mihalasky (2002) and Raines et al. (2007) have shown that WofE analysis provides necessary information for outlining of geologically permissive terranes for assessment of undiscovered mineral deposits. Carranza (2004b) has also shown that WofE analysis is useful in mapping mineral prospectivity of areas with few known prospects. In addition, WofE analysis has also been used to define threshold values separating anomaly and background in geochemical datasets (Bonham-Carter et al., 1989; Carranza, 2010). In this paper, we also apply WofE for mapping mineral prospectivity. However, the main objective of this paper is to show that, compared to using conventional geochemical anomalies of pathfinder elements, using Vz anomalies improves mapping of mineral prospectivity.
2. Concept of geochemical zonality method Kitaev (1991) proposed a multidimensional geochemical field analysis based on the notion that geological space is composed of
1936
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
geochemical fields representing zonality of associations of chemical elements. This analysis takes into account dispersion of chemical elements to separate multielement anomalies according to values of Vz. Grigorian (1992) has shown that patterns of Vz around exposed deposits are distinct from patterns of Vz associated with blind deposits. Recognition of zonality of geochemical halos associated with blind deposits can be achieved via four cases of complementary analyses (Ziaii et al., 2009b): (1) analysis of element associations representing supra-ore and sub-ore halos of mineral deposits; (2) analysis of a single component, implying false anomaly; (3) analysis of mean values of indicator elements outside significant geochemical anomalies to eliminate background noise in data analysis; and (4) mapping of multiplicative geochemical anomalies (i.e., Vz indices). Fig. 1 shows vertical variations in three Vz indices (Zn*Pb/Cu*Ag, Pb*Zn/Cu*Mo, Zn*Pb*Bi/Cu*Mo*Ag) associated with porphyry–Cu deposits in areas of the same landscape–geochemical conditions in different countries. Values of each Vz index decrease downward uniformly despite considerable differences in local geological settings of individual porphyry–Cu deposits, suggesting the existence of uniform vertical Vz in primary halos of porphyry–Cu deposits (Grigorian, 1992; Ziaii et al., 2009a). Therefore, vertical variations in values of Vz indices allow distinction of levels of mineralization and their primary (supra-ore, upper-ore, ore, lower-ore, and sub-ore) halos (Solovov, 1987, 1990; Grigorian, 1992) (Fig. 1). Moreover, it can be deduced from Fig. 1 that similar values of a Vz index imply similar depths of mineralization and primary halos within an ore field. Thus, primary halos of mineral deposits at different depths are characterized by specific values of a Vz index. The practical exploration significance of a Vz index is for recognition of erosional surfaces representing vertical levels of geochemical anomalies. That is, because in a Vz index, element data products used as numerators represent supra-ore to ore element associations whereas those used as denominators represent ore to sub-ore element associations. With respect to the present level of erosion, high values of a Vz index imply the presence of subcropping to blind deposits whereas low values of the index imply outcropping or already eroded deposits. Among the Vz indices shown in Fig. 1, the Vz1 (or Zn*Pb/Cu*Ag) is the best indicator of blind porphyry–Cu deposits in general (i.e., those not enriched in a secondary metal such as Mo or Au) (Ziaii, 1996; Ziaii et al., 2009a). Although the Vz method was developed originally for analysis of lithogeochemical data (Kitaev, 1991), Grigorian (1992) has demonstrated that it can be applied to stream sediment geochemical data to analyze erosional surfaces of multiplicative anomalies representing different exhumation levels of mineral deposits. Successful recognition of anomalous erosional surfaces is related, therefore, to the landscape conditions in mineralized regions (Ziaii et al., 2009b).
3. The test area 3.1. Geological and metallogenic setting of Jebal-Barez (southern Iran) All known important porphyry–Cu deposits in Iran occur in calcalkaline–shoshonitic rocks within the Central Iranian volcano– plutonic magmatic arc (Fig. 2) (Taghizadeh and Mallakpour, 1976; Shahabpour, 1982; Hezarkhani and Williams-Jones, 1998; Hassanzadeh, 1993; Aftabi and Atapour, 2000; Tangestani and Moore, 2002; Atapour and Aftabi, 2007). Magmatic activity in the Kerman Cenozoic magmatic arc (Shafiei, 2010) began with Early Eocene volcanism and continued to the Middle–Late Eocene. Some of the highly faulted Eocene volcanic rocks contain copper deposits (Huber, 1969; Taghizadeh and Mallakpour, 1976). The Eocene volcanic rocks show transition from tholeiitic to calcalkaline type and are characteristic of an island arc-type tectonic setting. Island-arc volcanism was followed by intrusive activity during the Oligocene (Jebal-Bareztype granitoids) (Fig. 3). Magmatic activity continued into the Middle Oligocene as represented by the Hezar volcanic complex, which consists mainly of shoshonitic rocks with continental-arc magma characteristics. The density of porphyry–Cu mineralization within the Kerman Cenozoic magmatic arc increases from southeast to northwest, and the majority of deposits (e.g., Sarcheshmeh, Meiduk, Darreh Zar, Sar Kuh, Now Chun) are located in elevated (uplifted), thickened arc crust (45–50 km). Most of the porphyry–Cu deposits are associated with well developed potassic (K-feldspar and biotite), sodic (albite), sericitic, silicic, propylitic, and locally argillic alteration. Mineralization occurs as quartz stockworks, veins, and disseminated sulfides in both the Miocene porphyritic stocks and the Eocene mafic and intermediate volcanic rocks. Common hypogene ore minerals are chalcopyrite and pyrite with subordinate molybdenite; bornite and magnetite are scarce. Supergene oxidation and secondary Cu-enrichment are present in all the deposits, but are well developed only in the giant and largest deposits such as Sarcheshmeh, Meiduk, and Darreh Zar (Huber, 1969; Taghizadeh and Mallakpour, 1976; Shafiei, 2010). The study area is located in the Shar-E-Babak–Bam ore field in the southern part of the Central Iranian volcano–plutonic magmatic arc (Fig. 2). The Meiduk and Sarcheshmeh porphyry–Cu deposits, which are active mines, are located close to the study area. The geological map of the study area (Fig. 3) is part of the 1:100,000 scale geological map over the Jebal-Barez district and surrounding areas (Valeh, 1972). The NW–SE trending Kuh-e Jebal-Barez mountain ranges, which cut through the study area, form part of the Sahand–Basman Tertiary volcanic belt and metallogenic province within the Central Iranian volcano– plutonic magmatic arc (Figs. 1 and 2). This volcanic belt was
Fig. 1. Vertical geochemical zonality (Vz) models for porphyry–Cu deposits based on typical standard porphyry–Cu deposits in Kazakhstan, Bulgaria, Armenia, and Iran (from Ziaii, 1996).
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Fig. 2. Location of the study area in the southern part of the Sahand–Basman volcanic belt (Iran).
Fig. 3. Simplified geological map of the study area (adapted from Valeh (1972)). Black dots represent porphyry–Cu mineral deposits/occurrences.
1937
1938
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
formed during closure of the Neo-Tethys and subduction of oceanic materials (Stocklin, 1968). The Sahand–Basman metallogenic province is subdivided into two metallogenic zones, namely Jebal-Barez and Karkas. The Jebal-Barez metallogenic zone consists of two distinct ore fields, namely Shar-E-Babak–Bam and Sarcheshmeh. From a geological point of view, porphyry–Cu mineralization in the study area occurs mainly in diorite/granodiorite and less commonly in quartz-monzonite. Shafiei (2010) presented and discussed a metallogenic model for the generation of the Miocene ore-hosting porphyries in a postsubduction and/or collisional setting in the Kerman Cenozoic magmatic arc (SE Iran) (Fig. 2). Regional mapping of the Kerman Cenozoic magmatic arc reveals distinctive patterns of argillic and phyllic rocks that can be associated with regional structural features and tectonic processes, and that can be used in regional mineral assessments. Semicircular patterns, 1–5 km in diameter, of mapped phyllic- and argillic-altered rocks are typically associated with Eocene to Miocene intrusive igneous rocks, some of which host known porphyry–Cu deposits, such as at Meiduk and Sarcheshmeh. Linear phyllic-altered rock patterns associated with extensive faults and fractures indicate potential epithermal or polymetallic vein deposits (Chahar-Gonbad deposit). Based on argillic and phyllic alteration patterns, about 50 potential porphyry–Cu deposits were mapped in an eroded, exhumed, and dormant part of the magmatic arc in NW Iran, whereas only 11 potential porphyry–Cu deposits were mapped in the volcanically active part of the magmatic arc in SE Iran (Figs. 1 and 2). In the volcanically active part of the magmatic arc in SE Iran, intermediate Quaternary volcanic rocks cover most of the Eocene rocks, which likely contain porphyry–Cu, epithermal, and polymetallic vein deposits (Romanco and Sadat, 2000; Mars and Rowan, 2006). Thus, based on regional alteration patterns and structures in the study area, analysis of metallogenic zones can reflect levels of exhumation (or horizon of erosional surfaces) of altered rocks and hydrothermal deposit types (or multimineralogical and geochemical types). The region where the study area is located has a semiarid climate, mountainous topography, and poor vegetation cover. Most of the known porphyry–Cu deposits in this region are characterized by well-developed zonal patterns of mineralization and hydrothermal alterations. These zonal patterns exhibit significant differences in terms of major oxide and trace element content reflecting variations in mineralogical and geochemical compositions of the mineralized and hydrothermally altered zones. 3.2. Spatial datasets From various spatial databases of the Geological Survey of Iran (GSI), we used the following datasets for regional-scale datadriven predictive mapping of prospectivity for porphyry–Cu deposits in the Jebal-Barez.
Locations of 15 porphyry–Cu deposits/occurrences (Fig. 3). Fault/fracture lineaments digitized from the 1:100,000 scale
geological/structural map of Jabal-Barez (Valeh, 1972) and from a shaded-relief image of total magnetic field intensity. Lithologic units from the 1:100,000 scale geological/structural map of Jebal-Barez (Valeh, 1972). Map of hydrothermally altered rocks interpreted from Landsat ETMþ bands. A subset of stream sediment geochemical data (300 samples analyzed for Cu, Zn, Ag, Pb, Au, W, As, Hg, Ba, Bi, Mo) pertaining to the study area. This geochemical data subset represents a total drainage basin area of ca. 760 km2 (i.e., sampling density of one sample per 2–3 km2).
4. Predictive mapping of prospectivity for porphyry–Cu deposits 4.1. Spatial evidence of prospectivity for porphyry–Cu deposits There are currently two major porphyry–Cu mines located close to the study area (Fig. 3). The analyses presented in this paper aim to obtain a regional-scale answer to the question ‘‘Which parts of the Jebal-Barez are prospective for undiscovered porphyry–Cu deposits?’’ based on empirical spatial associations of known porphyry–Cu deposits with evidential maps of porphyry– Cu prospectivity, including Vz1. Two sets of evidential maps were used. In the first set, we used an anomaly map of Cu (i.e., a pathfinder element for porphyry–Cu deposits), whereas in the second set, we used an anomaly map of Vz1. Both sets of evidential maps commonly include the lithological map, the alteration map derived from Landsat data, and a map of distance to fault lineament intersections. The objective of using these two sets of evidential maps is to compare the performance of geochemical anomalies of a pathfinder element and Vz anomalies in mapping of mineral prospectivity. We applied WofE analysis to quantify the spatial associations of known porphyry–Cu deposits in the study area with individual layers of spatial evidence. In this analysis, we used a unit cell size of 500 m for spatial representation of porphyry–Cu deposits based on the method proposed by Carranza (2008, 2009) for objective selection of unit cell size in data-driven predictive mapping of mineral prospectivity. In WofE analysis with a large number of deposits (say 420), a maximum positive contrast (C) for presence/ absence of evidence is considered a cutoff level for converting evidential data into binary predictor maps. However, in WofE analysis with small number of deposits, say 20 (e.g., Carranza and Hale, 2000, 2002; Carranza, 2004b), the studentized C (i.e., the ratio of C to its standard deviation) is used to judge the strength or statistical significance of spatial association and to a select cutoff level for converting evidential data into binary predictor maps. A studentized C greater than 1.96, for example, suggests a statistically significant positive spatial association at 98% confidence level (Bonham-Carter et al., 1989). We use this criterion for judging the strength of spatial association and for converting individual evidential data layers into binary predictor maps. 4.1.1. Lithologic evidence Six lithologic map units contain known porphyry–Cu deposits (Table 1 and Fig. 3). Unit Ert comprises rhyolitic pyroclastics. Unit Evt comprises mostly partly silicified intermediate volcanics and tuffs. Unit D comprises dacites occurring mainly near the margins granite–granodiorite–quartz-diorite (Gd) in contact with porphyrites (P). Unit Eta3 comprises agglomerate, ash and tuff. Of these lithologic units, unit P has the strongest and the only statistically significant positive spatial association with the known porphyry–Cu deposits. These results show that on a regional scale, lithology is a strong spatial evidence of porphyry–Cu prospectivity. In fact, based on the highest value of studentized C per evidence layer (Table 1), lithology is the second strongest spatial evidence of porphyry–Cu prospectivity in the study area. 4.1.2. Structural evidence Faults generally play important roles in porphyry–Cu mineralization (Ovchinnikov, 1971; Grigorian, 1992). In the study area, faults and fractures are probable structural controls on porphyry–Cu mineralization (Mars and Rowan, 2006). The fault/fracture lineament map, which contains 438 line segments of various types of faults, was initially subjected to cumulative increasing distance approach of WofE analysis. However, the result (not shown) was
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
1939
Table 1 Results of WofE calculations for individual layers of spatial evidence of porphyry–Cu prospectivity. Evidence layer
Evidence class
Area (km2)
Wþ
sW þ
W–
sW–
C
sC
studC
Lithology
Ert Evt Gd Eta3 D P
162.9 47.4 351.4 24.0 16.3 10.2
2 1 8 1 1 2
0.480 0.079 0.150 0.773 1.200 2.522
0.712 1.011 0.358 1.022 1.033 0.791
0.100 0.005 0.148 0.038 0.049 0.132
0.280 0.270 0.381 0.270 0.270 0.280
0.580 0.085 0.297 0.810 1.249 2.654
0.765 1.046 0.523 1.057 1.068 0.839
0.758 0.081 0.569 0.767 1.170 3.165
Distance to fault/fracture intersections (m)
o 100
87.6
6
1.293
0.423
0.395
0.336
1.688
0.540
3.126
100–200 200–300 300–400 400–500 4500
168.0 240.2 304.0 360.1 762.4
6 8 8 10 15
0.612 0.541 0.297 0.353 0.000
0.416 0.360 0.358 0.321 0.000
0.266 0.390 0.258 0.467
0.336 0.381 0.381 0.450
0.879 0.931 0.555 0.819
0.535 0.524 0.523 0.553
1.644 1.778 1.061 1.482
Alteration
Present Absent
83.1 679.3
7 8
1.525 0.521
0.395 0.356
0.521 1.524
0.356 0.396
2.045 2.045
0.532 0.532
3.843 3.843
Cu
1.00–0.80 0.80–0.60 0.60–0.40 0.40–0.20 0.20–0.00
61.9 301.6 546.0 750.5 762.4
3 4 9 15 15
0.931 0.401 0.180 0.016 0.000
0.592 0.503 0.336 0.261 0.261
0.163 0.165 0.282
0.291 0.305 0.414
1.094 0.566 0.463
0.660 0.589 0.533
1.658 0.961 0.867
Vz
1.00–0.80 0.80–0.60 0.60–0.40 0.40–0.20 0.20–0.00
111.5 265.2 444.2 615.8 762.4
5 8 10 13 15
0.850 0.438 0.138 0.072 0.000
0.458 0.359 0.320 0.280 0.261
0.275 0.370 0.276 0.471
0.319 0.381 0.451 0.712
1.125 0.809 0.413 0.543
0.558 0.523 0.553 0.765
2.017 1.545 0.748 0.710
No. of deposits
W þ and W– are positive and negative weights of evidence, respectively. sW þ and sW– are standard deviations of W þ and W–, respectively. C is spatial contrast (¼ W þ –W–). sC is standard deviation of C. studC is studentized C. Rows in bold are spatial evidence attributes per evidence layer with highest values of studC greater than 1.96, which corresponds to 98% confidence level (Bonham-Carter et al., 1989). The evidence classes corresponding to the rows in bold were the bases for creating binary predictor maps from the respective evidence layers.
not useful because the contrast value was statistically nonsignificant, indicating that not all types faults are spatially associated with the known porphyry–Cu deposits. Instead, a subset of intersecting faults/fractures was used. The results of analysis show that zones within 100 m of fault/fracture intersections have the strongest and the only statistically significant positive spatial association with the known porphyry–Cu deposit (Table 1). These results are consistent with the findings of Carranza and Hale (2002) that zones around fault/fracture discontinuities are favorable loci for porphyry–Cu mineralization. These results show that on a regional scale, distance to fault/fracture intersections is a strong spatial evidence of porphyry–Cu prospectivity and, in fact, it is the third strongest spatial evidence based on the highest value of studentized C per evidence layer (Table 1).
4.1.3. Presence of hydrothermal alteration By applying principal component (PC) analysis to Landsat ETMþ data (e.g., Moon et al., 2006), it is possible to extract spatial and spectral signatures that may reflect the presence of hydrothermally altered rocks associated with porphyry–Cu deposits in the Central Iranian volcanic belt (Ranjbar et al., 2004). Results of PC analysis of six Landsat ETM þ spectral bands show that PC3 enhances vegetation because it has high loadings on band 4 (Table 2). PC5 enhances the presence of hydroxyl minerals because it has high loadings of different signs on bands 5 and 7. PC4 enhances the presence of iron-oxides because it has high loadings of different signs on bands 1 and 3. Because of the arid conditions in the study area, it less likely that mineralized areas are strongly characterized by the presence of secondary iron-oxides. In contrast, outcropping porphyry–Cu deposits are likely associated with hydrothermally altered rocks. Thus, the image of PC5, possibly reflecting the presence of hydrothermal
Table 2 Principal components (PC) of six bands of Landsat ETMþ data. Landsat ETMþ bands
PC1
PC2
PC3
PC4
PC5
PC6
Band Band Band Band Band Band
0.22 0.30 0.43 0.24 0.56 0.47
0.18 0.12 0.16 0.05 0.26 0.24
0.03 0.04 0.06 0.72 0.05 0.50
0.50 0.15 0.58 0.36 0.35 0.19
0.07 0.10 0.29 0.09 0.54 0.62
0.48 0.39 0.43 0.19 0.45 0.15
1 2 3 4 5 7
Values in the table are loadings (eigenvectors) on each variable per PC. PC1 with loadings of the same signs represents overall brightness in all bands. Signs of variable loadings in PC2 reflect differences of data in bands covering the visible spectral wavelengths (bands 1–3) and bands covering the infrared spectral wavelengths (bands 1–3). PC3 represents the presence of vegetation because of highest loading on band 4. PC4 represents the presence of iron-oxides because it has high loadings of different signs on bands 1 and 3. PC5 represents the presence of hydroxyl minerals because it has high loadings of different signs on bands 5 and 7. PC6 is similar to PC2 but the former is related to noise in the data. Among the six PCs, PC5 is used to map hydrothermal alteration for mapping of prospectivity for porphyry–Cu deposits in the study area.
argillic alteration in rocks, is considered as a spatial evidence of porphyry–Cu prospectivity. However, based on the spectral characteristics of clays in general (i.e., reflection in band 5, absorption in band 7), the image of PC5 is negated (i.e., multiplied by –1) to portray hydrothermally altered rocks as bright pixels (Fig. 4). The negated PC5 image of the six Landsat ETMþ spectral bands was generalized into two classes—altered rock present and altered rocks absent—after applying a cumulative decreasing value approach of WofE analysis to determine the range of negated PC5 values with the strongest spatial association with the known porphyry–Cu deposits. The classified image was subjected to a new WofE analysis to determine weights for each of the ‘‘present’’ and ‘‘absent’’ classes (Table 1). The positive and
1940
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Fig. 4. Negated PC5 image showing the presence of hydrothermally altered rocks as bright pixels. White dots are locations of known porphyry–Cu deposits/occurrences.
negative spatial associations of the known porphyry–Cu deposits with the present and absent classes, respectively, are statistically significant. This shows that, on a regional-scale, the presence of hydrothermal alteration is a strong spatial evidence of porphyry–Cu prospectivity and, in fact, it is the strongest spatial evidence based on the highest value of studentized C per evidence layer (Table 1).
4.1.4. Geochemical evidence Individual unielement data were interpolated by kriging. The interpolated map of Cu (Fig. 5) was used as spatial evidence of porphyry–Cu prospectivity. For direct comparison of results of WofE analysis, values in the interpolated Cu map and in the interpolated Vz1 map were individually rescaled linearly to the range [0,1]. The rescaled interpolated Cu map was subjected to a cumulative decreasing value approach of WofE analysis, after classifying the map into 20th percentile classes based on sample values per sample but not unit cell interpolated values (Carranza, 2010). Zones of the study area with the highest 20th percentile Cu values have the strongest and the only statistically significant positive spatial association with the known porphyry–Cu deposits (Table 1). Multiplicative geochemical data of Cu*Ag and Pb*Zn were first created. Application of fuzzy c-means clustering (Bezdek et al., 1984) shows two cluster centroids in the multiplicative geochemical data (Fig. 6). Centroid C1 represents background values of Cu*Ag and Pb*Zn. Centroid C2 represents anomalous Pb*Zn values and background Ag*Cu values. The two clusters are separated by a clear threshold value of Pb*Zn. Therefore, high Vz1 values would suggest supra-ore to ore multielement anomalies associated with blind to subcropping porphyry–Cu mineralization (Fig. 1). With this assumption, the study area can be subdivided into at least five zones where possible porphyry–Cu deposits are exhumed to
different levels (Fig. 7). Of those five zones, zones A1 and A2 would be favorable for exploration of subcropping to blind porphyry–Cu deposits whereas the other zones would be favorable for exploration of outcropping porphyry–Cu deposits. In addition to zones A1 and A2, the interpolated and rescaled Vz1 values show a zone in the northwestern part of the study area where subcropping to blind porphyry–Cu deposits possibly occur (Fig. 8). The rescaled interpolated Vz1 map was subjected to a cumulative decreasing value approach of WofE analysis, after classifying the map into 20th percentile classes based on sample values per sample but not unit cell interpolated values (Carranza, 2010). Zones of the study area with the highest 20th percentile Vz1 values have the strongest and the only statistically positive spatial association with the known porphyry–Cu deposits (Table 1). Based on the highest value of studentized C per evidence layer (Table 1), The Vz1 map, compared to the Cu map, is a better spatial evidence of porphyry–Cu prospectivity. However, both geochemical evidence layers (i.e., the Vz1 map and Cu map), compared to the three geological evidence layers (lithology, distance to fault/ fracture intersections, and alteration), have weaker positive spatial associations with the known porphyry–Cu deposits in the study area. This result is explicable, however, because the geochemical evidence layers represent stream sediments, which are materials transported away from their sources, whereas the geological evidence layers represent in situ materials like the mineral deposits under examination. 4.2. Integration of spatial evidence layers Based on the results of WofE analyses of spatial associations between the known porphyry–Cu deposits and the individual layers of spatial evidence (Table 1), binary predictor maps were
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
1941
Fig. 5. Geochemical contour map for Cu. White dots are locations of porphyry–Cu deposits/occurrences.
Fig. 6. Scattergrams of multiplicative Ag*Cu and Pb*Zn data showing two centroids, C1 and C2, in the data.
prepared and the weights (Wþ, W–) were assigned to the corresponding predictor criteria class (Table 3) in each map. The weighted binary predictor maps were integrated with each other to update prior probability of porphyry–Cu deposit occurrence (for details, see Bonham-Carter et al., 1989; Agterberg et al., 1990). Using 500 m as unit cell size for a study area covering 762.4 km2 and containing 15 known porphyry–Cu deposits, the estimated prior probability of porphyry–Cu deposit occurrence is 0.005. Two integration experiments were performed: (1) using
lithological, structural, alteration, and Cu predictor maps; and (2) using lithological, structural, alteration, and Vz1 predictor maps. Results of the first experiment are referred to as the Cuin-WofE model; the second experiment, Vz1-in-WofE model. Cumulative frequency curves of posterior probabilities in both models exhibit similar shapes and inflection points at ca. 96 cumulative percent and at ca. 88 cumulative percent (Fig. 9). However, the inflection points on the cumulative frequency curve for the Vz1-in-WofE model correspond to higher posterior
1942
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Fig. 7. Classification of the study area into five zones based on Vz1 (or Zn*Pb/Cu*Ag) values derived from stream sediment geochemical data.
Fig. 8. Contour map of Vz1 (or Zn*Pb/Cu*Ag) values. White dots represent locations of porphyry–Cu deposits/occurrences.
probabilities, compared to those of the inflection points on the cumulative frequency curve for the Cu-in-WofE model. This shows that the Vz1 map, compared to the Cu map, has stronger spatial association with the known porphyry–Cu deposits and is a better predictor of porphyry–Cu prospectivity. Based on the
inflection points, we classified the respective posterior probability maps into three prospectivity or favorability classes (Fig. 9). Posterior probabilities of porphyry–Cu deposit occurrence in moderate and high favorability zones are greater than the estimated prior probability. Some portions of low favorability
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
1943
Table 3 Weights of binary predictor maps used in WofE analysis of porphyry–Cu prospectivity. Predictor
Criteria
Wþ
sW þ
W–
sW–
C
sC
studC
Lithology Fault/fracture intersections Hydrothermal alteration Cu values Vz1 values
Presence/absence of unit P Within/beyond 100 m Presence/absence Z80/ o 80 percentile Z80/ o 80 percentile
2.522 1.293 1.525 0.931 0.850
0.791 0.423 0.395 0.592 0.458
0.132 0.395 0.521 0.163 0.275
0.280 0.336 0.356 0.291 0.319
2.654 1.688 2.045 1.094 1.125
0.839 0.540 0.532 0.660 0.558
3.165 3.126 3.843 1.658 2.017
W þ and W– are positive and negative weights of evidence, respectively. sW þ and sW– are standard deviations of W þ and W–, respectively. C is spatial contrast (¼ W þ –W–). sC is standard deviation of C. studC is studentized C.
Fig. 9. Porphyry–Cu prospectivity models: (a) Cu-in-WofE and (b) Vz1-in-WofE. Inflection points on each cumulative frequency curves of posterior probability models were used for classifying prospectivity into three favorability classes. See the text for explanations about borehole and subareas marked as I–III.
zones have posterior probabilities of porphyry–Cu deposit occurrence greater than the estimated prior probability. The shapes and areas of moderate and high favorability zones for both models are strongly similar, but the distributions of moderate and high favorability zones differ. The Cu-in-WofE model shows that high favorability zones are mainly in the northern part of the study area, whereas the Vz1-in-WofE model shows the opposite (Fig. 9). To describe the quality of the classified prospectivity maps (Fig. 9), we derived a prospectivity index by taking the ratio of the cumulative percentage of known porphyry–Cu deposits delineated to the cumulative percentage of area covered by high to low prospectivity classes (Table 4). A similar but different quality measure of prospectivity maps was used by Brown et al. (2000) and Porwal et al. (2006). The prospectivity index is akin to the success rate as explained in Table 1. A prospectivity index of 1 would imply a random chance of mineral deposit occurrence, whereas higher prospectivity indices imply higher deviations
from that random chance. Actually, for delineated high prospectivity zones, the Vz1-in-WofE model has a success rate of 60%, whereas the Cu-in-WofE model has a success rate of 40% (Table 4). Thus, based on the prospectivity index it is clear that the high prospectivity class of the Vz1-in-WofE model outperforms the high prospectivity class of the Cu-in-WofE model (Table 4). The results show that the Vz1-in-WofE model is better than the Cu-in-WofE model for constricting and focusing of mineral exploration targets. Based on either model, there seems to be at least three subareas (I–III) that warrant further exploration (Fig. 9). Based on the Cu-in-WofE model it is difficult to say whether any of those three subareas would be favorable for exploration of blind porphyry–Cu deposits. In contrast, based on the Vz1-in-WofE model, subarea III would be favorable mainly for exploration of outcropping porphyry–Cu deposits whereas subareas I and II would be favorable for exploration of blind to subcropping
1944
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Table 4 Performance of models of porphyry–Cu prospectivity in the study area. Model
Prospectivity classes
Cumulative % deposits contained (A)
Cumulative % area covered (B)
Prospectivity index (A–B)
Cu-in-WofE
High High–moderate High–moderate–low
40 80 100
4 12 100
10 6.7 1
Vz1-in-WofE
High High–moderate High–moderate–low
60 80 100
3 12 100
20 6.7 1
Column (A) refers to the success rate or goodness-of-fitting evidential maps with a number (here n ¼15) of known deposits. Column (B) refers to the size of the proverbial ‘‘haystack’’ (i.e., exploration target area), such that the smaller it is the better the chance of finding the proverbial ‘‘needle’’ (i.e., mineral deposit type of interest) in it. The prospectivity index relates to the likelihood that mineral deposits exist in areas with certain prospectivity classes.
porphyry–Cu deposits. In fact, prior to writing of this paper, blind porphyry–Cu mineralization was intersected by borehole exploration in subarea I (Fig. 9) (Borna, 2007; NICICO, 2008).
with those of other methods of geochemical anomaly mapping, is warranted.
Acknowledgments 5. Discussion and conclusions Mineral deposits and associated primary halos are characterized by variations in chemical compositions along both vertical and horizontal directions. Most methods of analyses of anomalies in stream sediment data are concerned with variations in surficial chemical compositions along horizontal directions. However, the concept of the geochemical zonality method allows distinction between sub- and supra-ore anomalies. Therefore, application of Vz indices in mineral prospectivity allows further interpretation about whether delineated favorable areas are attractive for exploration of outcropping or blind mineral deposits. This added-value information from Vz indices is essential in planning for exploration activities. However, integration of maps of Vz indices with other maps used as spatial evidence is important in filtering-out false indications of mineral prospectivity portrayed in every layer of evidence. This is shown by the fact that inflection points on the cumulative frequency curve for the Vz1-in-WofE model relate to higher posterior probabilities, compared to those of the inflection points on the cumulative frequency curve for the Cu-in-WofE model (Fig. 9). This is also demonstrated by comparing the Vz1 map (Fig. 8) with the obtained mineral prospectivity map (Fig. 9b). The former map shows high Vz1 values in the northwestern part of the study area, whereas the latter map portrays low prospectivity in that part of the study area. Thus, care must also be taken in the interpretation of Vz maps because, although Vz indices can reflect geochemical variations in vertical directions, a large proportion of variations in geochemical compositions of earth materials are, in fact, mostly unrelated to mineral deposit. This caveat is particularly important if the method of geochemical zonality is applied to stream sediment geochemical data, which represent earth materials upstream from a sampling site. Although methods for GIS-based mapping mineral prospectivity are now mostly well established, it is important to review which methods of geochemical data analysis result in anomaly maps that, in turn, lead to better models of mineral prospectivity. We have shown in this study case that, instead of using anomalies of pathfinder elements, using geochemical zonality anomalies as one of several evidential maps results in improved mapping of mineral prospectivity. In addition, whereas weights-of-evidence analysis was used in this study, other methods of data representation and integration for mineral prospectivity mapping can be used. Finally, further testing of the geochemical zonality method in mineral prospectivity mapping, and comparing its performance
The research was funded by the School of Mining, Petroleum, and Geophysics at the Shahrood University of Technology. Helpful comments by two reviewers (Dr. Gary Raines and an anonymous individual) led to substantial improvements of the manuscript. We particularly thank Dr. Gary Raines for providing a wide range of insightful ideas and comments throughout the program that stimulated the direction of research and helped draw it to a close. References Aftabi, A., Atapour, H., 2000. Regional aspects of shoshonitic volcanism in Iran. Episodes 23, 119–125. Agterberg, F.P., Bonham-Carter, G.F., Wright, D.F., 1990. Statistical pattern integration for mineral exploration. In: Gaa´l, G., Merriam, D.F. (Eds.), Computer Applications in Resource Estimation. Pergamon Press, Oxford, pp. 1–21. Atapour, H., Aftabi, A., 2007. The geochemistry of gossans associated with Sarcheshmeh porphyry copper deposit, Rafsanjan, Kerman, Iran: implications for exploration and the environment. Journal of Geochemical Exploration 93, 47–65. Beus, A.A., Grigorian, S.V., 1977. Geochemical Exploration Methods for Mineral Deposits. Applied Publishing Ltd., Wilmette, IL 287 pp. Bezdek, J.C., Ehrlich, R., Full, W., 1984. FCM: The fuzzy c-means clustering algorithm. Computers & Geosciences 10, 191–203. Bonham-Carter, G.F., 1994. Geographic Information Systems for Geoscientists: Modelling with GIS. Pergamon, Ontario 398 pp. Bonham-Carter, G.F., Agterberg, F.P., Wright, D.F., 1989. Weights of evidence modelling: a new approach to mapping mineral potential. In: Agterberg, F.P., Bonham-Carter, G.F. (Eds.), Statistical Applications in the Earth Sciences. Geological Survey of Canada, pp. 171–183 Paper 89-9. Borna, B., 2007. Discovery of copper and molybdenum porphyry in the Jebal-Barez, SW Iran. Report of the Geological Survey of Iran, 150 pp. (in Farsi). Brown, W.M., Gedeon, T.D., Groves, D.I., Barnes, R.G., 2000. Artificial neural networks: a new method for mineral prospectivity mapping. Australian Journal of Earth Sciences 47, 757–770. Carranza, E.J.M., 2004a. Usefulness of stream order to detect stream sediment geochemical anomalies. Geochemistry: Exploration, Environment, Analysis 4, 341–352. Carranza, E.J.M., 2004b. Weights-of-evidence modelling of mineral potential: a case study using small number of prospects, Abra, Philippines. Natural Resources Research 13, 173–187. Carranza, E.J.M., 2008. Geochemical Anomaly and Mineral Prospectivity Mapping in GIS. Handbook of Exploration and Environmental Geochemistry, vol. 11. Elsevier, Amsterdam 351 pp. Carranza, E.J.M., 2009. Objective selection of suitable unit cell size in data-driven modeling of mineral prospectivity. Computers & Geosciences 35, 2032–2046. Carranza, E.J.M., 2010. Mapping of anomalies in continuous and discrete fields of stream sediment geochemical landscapes. Geochemistry: Exploration, Environment, Analysis 10, 171–187. Carranza, E.J.M., Hale, M., 2000. Geologically constrained probabilistic mapping of gold potential, Baguio district, Philippines. Natural Resources Research 9, 237–253. Carranza, E.J.M., Hale, M., 2002. Where are porphyry copper deposits spatially localized? A case study in Benguet province, Philippines. Natural Resources Research 11, 45–59.
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Grigorian, S.V., 1992. Mining Geochemistry. Nedra Publishing House, Moscow 310 pp. (in Russian). Guanghua, Z., Zongqi, O., Hui, L., 1996. Exploration Models for Major Types of Gold Deposits in China. Geological Publishing House, Beijing 280 pp. (in Chinese with abstract in English). Hassanzadeh, J., 1993. Metallogenic and Tectono-magmatic Events in SE Sector of the Cenozoic Active Continental Margin of Central Iran (Shahr-E-Babak, Kerman Province). Ph.D. Dissertation, University of California, Los Angeles, 204 pp. Hezarkhani, A., Williams-Jones, A.E., 1998. Controls of alteration and mineralization in the Sungun porphyry copper deposits, Iran: evidence from fluid inclusions and stable isotopes. Economic Geology 93, 651–670. Huber, H., 1969. Geological Map of Iran Sheet No. 6, Southeast Iran, scale 1:1,000,000. National Iranian Oil Company, Tehran. Ivanov, V.V., Meituv, G.M., 1972. Geological and Geochemical Investigations of Ore Provinces. Nedra, Moscow 152 pp. (in Russian). Khorin, G.I., Borodin, B.P., Matveev, A.A., 1992. Recommended Methods for Lithogeochemical Prospecting of Ore Deposits Through Dispersion Flow. Institute of the Mineralogy, Geochemistry and Crystallography of Rare Elements, Moscow 164 pp. Kitaev, N.A., 1991. Multidimensional analysis of geochemical fields. Mathematical Geology 23, 15–32. Mars, J.C., Rowan, L.C., 2006. Regional mapping of phyllic- and argillic-altered rocks in the Zagros magmatic arc, Iran, using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data and logical operator algorithms. Geosphere 2, 161–186. Moon, C.J., 1999. Towards a quantitative model of downstream dilution of point source geochemical anomalies. Journal of Geochemical Exploration 65, 111–132. Moon, C.J., Whateley, M.K.G., Evans, A.M. (Eds.), 2006. Introduction to Mineral Exploration. Blackwell Publishing, Oxford. NICICO, 2008. Drill-holes Exploration in Kerver Area. National Iranian Copper Industries Company (NICICO) Reports, 30 pp. Ovchinnikov, L.N., 1971. Prognostic evaluation of world reserves of metals in land deposits. Dokl, RAS, No. 3, 683–686. Pan, G.C., Harris, D.P., 2000. Information Synthesis for Mineral Exploration. Oxford University Press, Inc., New York 461 pp. Porwal, A., Carranza, E.J.M., Hale, M., 2006. A hybrid fuzzy weights-of-evidence model for mineral potential mapping. Natural Resources Research 15, 1–15. Raines, G.L., Connors, K.A., Chorlton, L.B., 2007. Porphyry copper deposit tract definition—a global analysis comparing geologic map scales. Natural Resources Research 16, 191–198.
1945
Raines, G.L., Mihalasky, M.J., 2002. A reconnaissance method for delineation of tracts for regional-scale mineral-resource assessment based on geologic-map data. Natural Resources Research 11, 241–248. Ranjbar, H., Shahriari, H., Honarmand, M., 2004. Integration of ASTER and airborne geophysical data for exploration of copper mineralization: a case study of Sar Cheshmeh area. In: XXth International Society for Photogrammetry and Remote Sensing Congress, 12–23 July 2004, Istanbul, Turkey, Geo-imagery Bridging Continents, vol. 4, pp. 701–706. Romanco, E.F., Sadat, M.A., 2000. Metallogenic zoning of the Sistan and Baluchestan, Iran. Journal of Geology Science 1, 20–23. Shafiei, B., 2010. Lead isotope signatures of the igneous rocks and porphyry copper deposits from the Kerman Cenozoic magmatic arc (SE Iran), and their magmatic-metallogenetic implications. Ore Geology Reviews 38, 27–36. Shahabpour, J., 1982. Aspects of Alteration and Mineralization at the Sar-Cheshmeh Copper–molybdenum Deposit, Kerman, Iran. Ph.D. Dissertation, Leeds University, 342 pp. Shcheglov, A.D., 1979. Fundamentals of Metallogenic Analysis. Mir Publishers, Moscow 335 pp. (Trans. from Russian by V. Shiffer). Solovov, A.P., 1987. Geochemical Prospecting for Mineral Deposits. Mir, Moscow, 288 pp. [V.V. Kuznetsov, Trans.; Engl. ed.]. Solovov, A.P. (Ed.), 1990. Handbook on Geochemical Prospecting for Useful Minerals. Nedra Publishing House, Moscow 336pp. [in Russian]. Stocklin, J., 1968. Structural history and tectonics of Iran: a review. American Association of Petroleum Geologists Bulletin 52, 1229–1258. Taghizadeh, N., Mallakpour, M.A., 1976. Mineral Distribution Map of Iran, scale 1:2,500,000. Geological Survey of Iran, Tehran. Tangestani, M.H., Moore, F., 2002. Porphyry copper alteration mapping at the Meiduk area, Iran. International Journal of Remote Sensing 23, 4815–4825. Valeh, N., 1972. Geological Map of Iran Sheet 7647—Jebal-Barez, scale 1:100,000. Geological Survey of Iran, Tehran. Ziaii, M., 1996. Lithogeochemical Exploration Methods for Porphyry Copper Deposit in Sungun, NW Iran. Unpublished M.Sc. Thesis, Moscow State University (MSU), Moscow, 98 pp. (in Russian). Ziaii, M., Abedi, A., Ziaei, M., 2009a. Geochemical and mineralogical pattern recognition and modeling with a Bayesian approach to hydrothermal gold deposits. Journal of Applied Geochemistry 24, 1142–1146. Ziaii, M., Pouyan, A.A., Ziaei, M., 2009b. Neuro-fuzzy modelling in mining geochemistry: identification of geochemical anomalies. Journal of Geochemical Exploration 100, 25–36.
Contents lists available at ScienceDirect
Computers & Geosciences journal homepage: www.elsevier.com/locate/cageo
Application of geochemical zonality coefficients in mineral prospectivity mapping Mansour Ziaii a,n, Emmanuel John M. Carranza b, Mahdi Ziaei a a b
Faculty of Mining, Petroleum, and Geophysics, Shahrood University of Technology, Iran Department of Earth Systems Analysis, Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, The Netherlands
a r t i c l e i n f o
abstract
Article history: Received 2 September 2010 Received in revised form 20 April 2011 Accepted 25 May 2011 Available online 15 June 2011
Indices of geochemical zonality (Vz) of multielements around mineral deposits and their spatial associations with particular geological, geochemical, and structural factors are critical aspects that must be considered in mineral exploration. Values of Vz indices allow distinction between sub-ore and supra-ore anomalies, which are associated with outcropping and blind deposits, respectively. In this paper, we used a map of a Vz index (Zn*Pb/Cu*Ag) in weights-of-evidence analysis of regional-scale prospectivity for porphyry–Cu deposits in the area covered by 1:100,000 scale map sheet of Jebal-Barez (Kerman province, southern Iran). For comparison, we used a Cu map instead of the Vz map in the weights-of-evidence (WofE) analysis. The Vz-in-WofE prospectivity model outperforms the Cu-in-WofE prospectivity model. In fact, prior to writing this paper, blind porphyry–Cu mineralization was intersected at depth by borehole exploration in a high prospectivity zone delineated by the Vz-in-WofE model. The results demonstrate the usefulness of the Vz-in-WofE for regional-scale targeting of blind mineral deposits. & 2011 Elsevier Ltd. All rights reserved.
Keywords: Geochemical zonality Sub-ore anomaly Supra-ore anomaly Weights of evidence GIS Jebal-Barez (Iran)
1. Introduction In the former Soviet Union (FSU), metallometric methods have been developed for mapping geochemical anomalies associated with mineral deposits (Ivanov and Meituv, 1972; Shcheglov, 1979). Many of those methods led to successful exploration results in the FSU, China, and other countries (Khorin et al., 1992). During the period 1981–1985, 238 mineral deposits were discovered by using suitable metallometric methods (Guanghua et al., 1996). In particular, the method of productivity mapping efficiently differentiates between mineralized and nonmineralized zones (Beus and Grigorian, 1977; Solovov, 1987, 1990; Grigorian, 1992; Moon, 1999; Carranza, 2004a). This method, although widely used in the FSU and China, did not attract significant attention in other countries. Most metallometric methods have also not been widely practiced by using geographic information systems (GIS). Most metallometric methods have been developed for stream sediment geochemical surveys, which are generally not useful for exploration in the following situations: (1) presence of significant heavy metal contamination of streams due to mining activities; (2) low relief areas where hydromorphic dispersion predominates over mechanical dispersion of metals from certain mineral deposits. Yet, the geochemical zonality (Vz) method, which is among the metallometric methods developed in the FSU for analysis of lithogeochemical data, does not have those limitations.
n
Corresponding author. Fax: þ98 273 333 5509. E-mail address: [email protected] (M. Ziaii).
0098-3004/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.cageo.2011.05.009
However, it has also not attracted significant attention in many countries outside the FSU because it has not been demonstrated yet in conjunction with GIS-based mapping of mineral prospectivity. Several methods currently exist for GIS-based mapping mineral prospectivity, whereby various relevant evidential data layers are integrated (Bonham-Carter, 1994; Pan and Harris, 2000; Carranza, 2008). The most widely used of those methods is weights-ofevidence (WofE) analysis (Bonham-Carter et al., 1989; Agterberg et al., 1990). Raines and Mihalasky (2002) and Raines et al. (2007) have shown that WofE analysis provides necessary information for outlining of geologically permissive terranes for assessment of undiscovered mineral deposits. Carranza (2004b) has also shown that WofE analysis is useful in mapping mineral prospectivity of areas with few known prospects. In addition, WofE analysis has also been used to define threshold values separating anomaly and background in geochemical datasets (Bonham-Carter et al., 1989; Carranza, 2010). In this paper, we also apply WofE for mapping mineral prospectivity. However, the main objective of this paper is to show that, compared to using conventional geochemical anomalies of pathfinder elements, using Vz anomalies improves mapping of mineral prospectivity.
2. Concept of geochemical zonality method Kitaev (1991) proposed a multidimensional geochemical field analysis based on the notion that geological space is composed of
1936
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
geochemical fields representing zonality of associations of chemical elements. This analysis takes into account dispersion of chemical elements to separate multielement anomalies according to values of Vz. Grigorian (1992) has shown that patterns of Vz around exposed deposits are distinct from patterns of Vz associated with blind deposits. Recognition of zonality of geochemical halos associated with blind deposits can be achieved via four cases of complementary analyses (Ziaii et al., 2009b): (1) analysis of element associations representing supra-ore and sub-ore halos of mineral deposits; (2) analysis of a single component, implying false anomaly; (3) analysis of mean values of indicator elements outside significant geochemical anomalies to eliminate background noise in data analysis; and (4) mapping of multiplicative geochemical anomalies (i.e., Vz indices). Fig. 1 shows vertical variations in three Vz indices (Zn*Pb/Cu*Ag, Pb*Zn/Cu*Mo, Zn*Pb*Bi/Cu*Mo*Ag) associated with porphyry–Cu deposits in areas of the same landscape–geochemical conditions in different countries. Values of each Vz index decrease downward uniformly despite considerable differences in local geological settings of individual porphyry–Cu deposits, suggesting the existence of uniform vertical Vz in primary halos of porphyry–Cu deposits (Grigorian, 1992; Ziaii et al., 2009a). Therefore, vertical variations in values of Vz indices allow distinction of levels of mineralization and their primary (supra-ore, upper-ore, ore, lower-ore, and sub-ore) halos (Solovov, 1987, 1990; Grigorian, 1992) (Fig. 1). Moreover, it can be deduced from Fig. 1 that similar values of a Vz index imply similar depths of mineralization and primary halos within an ore field. Thus, primary halos of mineral deposits at different depths are characterized by specific values of a Vz index. The practical exploration significance of a Vz index is for recognition of erosional surfaces representing vertical levels of geochemical anomalies. That is, because in a Vz index, element data products used as numerators represent supra-ore to ore element associations whereas those used as denominators represent ore to sub-ore element associations. With respect to the present level of erosion, high values of a Vz index imply the presence of subcropping to blind deposits whereas low values of the index imply outcropping or already eroded deposits. Among the Vz indices shown in Fig. 1, the Vz1 (or Zn*Pb/Cu*Ag) is the best indicator of blind porphyry–Cu deposits in general (i.e., those not enriched in a secondary metal such as Mo or Au) (Ziaii, 1996; Ziaii et al., 2009a). Although the Vz method was developed originally for analysis of lithogeochemical data (Kitaev, 1991), Grigorian (1992) has demonstrated that it can be applied to stream sediment geochemical data to analyze erosional surfaces of multiplicative anomalies representing different exhumation levels of mineral deposits. Successful recognition of anomalous erosional surfaces is related, therefore, to the landscape conditions in mineralized regions (Ziaii et al., 2009b).
3. The test area 3.1. Geological and metallogenic setting of Jebal-Barez (southern Iran) All known important porphyry–Cu deposits in Iran occur in calcalkaline–shoshonitic rocks within the Central Iranian volcano– plutonic magmatic arc (Fig. 2) (Taghizadeh and Mallakpour, 1976; Shahabpour, 1982; Hezarkhani and Williams-Jones, 1998; Hassanzadeh, 1993; Aftabi and Atapour, 2000; Tangestani and Moore, 2002; Atapour and Aftabi, 2007). Magmatic activity in the Kerman Cenozoic magmatic arc (Shafiei, 2010) began with Early Eocene volcanism and continued to the Middle–Late Eocene. Some of the highly faulted Eocene volcanic rocks contain copper deposits (Huber, 1969; Taghizadeh and Mallakpour, 1976). The Eocene volcanic rocks show transition from tholeiitic to calcalkaline type and are characteristic of an island arc-type tectonic setting. Island-arc volcanism was followed by intrusive activity during the Oligocene (Jebal-Bareztype granitoids) (Fig. 3). Magmatic activity continued into the Middle Oligocene as represented by the Hezar volcanic complex, which consists mainly of shoshonitic rocks with continental-arc magma characteristics. The density of porphyry–Cu mineralization within the Kerman Cenozoic magmatic arc increases from southeast to northwest, and the majority of deposits (e.g., Sarcheshmeh, Meiduk, Darreh Zar, Sar Kuh, Now Chun) are located in elevated (uplifted), thickened arc crust (45–50 km). Most of the porphyry–Cu deposits are associated with well developed potassic (K-feldspar and biotite), sodic (albite), sericitic, silicic, propylitic, and locally argillic alteration. Mineralization occurs as quartz stockworks, veins, and disseminated sulfides in both the Miocene porphyritic stocks and the Eocene mafic and intermediate volcanic rocks. Common hypogene ore minerals are chalcopyrite and pyrite with subordinate molybdenite; bornite and magnetite are scarce. Supergene oxidation and secondary Cu-enrichment are present in all the deposits, but are well developed only in the giant and largest deposits such as Sarcheshmeh, Meiduk, and Darreh Zar (Huber, 1969; Taghizadeh and Mallakpour, 1976; Shafiei, 2010). The study area is located in the Shar-E-Babak–Bam ore field in the southern part of the Central Iranian volcano–plutonic magmatic arc (Fig. 2). The Meiduk and Sarcheshmeh porphyry–Cu deposits, which are active mines, are located close to the study area. The geological map of the study area (Fig. 3) is part of the 1:100,000 scale geological map over the Jebal-Barez district and surrounding areas (Valeh, 1972). The NW–SE trending Kuh-e Jebal-Barez mountain ranges, which cut through the study area, form part of the Sahand–Basman Tertiary volcanic belt and metallogenic province within the Central Iranian volcano– plutonic magmatic arc (Figs. 1 and 2). This volcanic belt was
Fig. 1. Vertical geochemical zonality (Vz) models for porphyry–Cu deposits based on typical standard porphyry–Cu deposits in Kazakhstan, Bulgaria, Armenia, and Iran (from Ziaii, 1996).
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Fig. 2. Location of the study area in the southern part of the Sahand–Basman volcanic belt (Iran).
Fig. 3. Simplified geological map of the study area (adapted from Valeh (1972)). Black dots represent porphyry–Cu mineral deposits/occurrences.
1937
1938
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
formed during closure of the Neo-Tethys and subduction of oceanic materials (Stocklin, 1968). The Sahand–Basman metallogenic province is subdivided into two metallogenic zones, namely Jebal-Barez and Karkas. The Jebal-Barez metallogenic zone consists of two distinct ore fields, namely Shar-E-Babak–Bam and Sarcheshmeh. From a geological point of view, porphyry–Cu mineralization in the study area occurs mainly in diorite/granodiorite and less commonly in quartz-monzonite. Shafiei (2010) presented and discussed a metallogenic model for the generation of the Miocene ore-hosting porphyries in a postsubduction and/or collisional setting in the Kerman Cenozoic magmatic arc (SE Iran) (Fig. 2). Regional mapping of the Kerman Cenozoic magmatic arc reveals distinctive patterns of argillic and phyllic rocks that can be associated with regional structural features and tectonic processes, and that can be used in regional mineral assessments. Semicircular patterns, 1–5 km in diameter, of mapped phyllic- and argillic-altered rocks are typically associated with Eocene to Miocene intrusive igneous rocks, some of which host known porphyry–Cu deposits, such as at Meiduk and Sarcheshmeh. Linear phyllic-altered rock patterns associated with extensive faults and fractures indicate potential epithermal or polymetallic vein deposits (Chahar-Gonbad deposit). Based on argillic and phyllic alteration patterns, about 50 potential porphyry–Cu deposits were mapped in an eroded, exhumed, and dormant part of the magmatic arc in NW Iran, whereas only 11 potential porphyry–Cu deposits were mapped in the volcanically active part of the magmatic arc in SE Iran (Figs. 1 and 2). In the volcanically active part of the magmatic arc in SE Iran, intermediate Quaternary volcanic rocks cover most of the Eocene rocks, which likely contain porphyry–Cu, epithermal, and polymetallic vein deposits (Romanco and Sadat, 2000; Mars and Rowan, 2006). Thus, based on regional alteration patterns and structures in the study area, analysis of metallogenic zones can reflect levels of exhumation (or horizon of erosional surfaces) of altered rocks and hydrothermal deposit types (or multimineralogical and geochemical types). The region where the study area is located has a semiarid climate, mountainous topography, and poor vegetation cover. Most of the known porphyry–Cu deposits in this region are characterized by well-developed zonal patterns of mineralization and hydrothermal alterations. These zonal patterns exhibit significant differences in terms of major oxide and trace element content reflecting variations in mineralogical and geochemical compositions of the mineralized and hydrothermally altered zones. 3.2. Spatial datasets From various spatial databases of the Geological Survey of Iran (GSI), we used the following datasets for regional-scale datadriven predictive mapping of prospectivity for porphyry–Cu deposits in the Jebal-Barez.
Locations of 15 porphyry–Cu deposits/occurrences (Fig. 3). Fault/fracture lineaments digitized from the 1:100,000 scale
geological/structural map of Jabal-Barez (Valeh, 1972) and from a shaded-relief image of total magnetic field intensity. Lithologic units from the 1:100,000 scale geological/structural map of Jebal-Barez (Valeh, 1972). Map of hydrothermally altered rocks interpreted from Landsat ETMþ bands. A subset of stream sediment geochemical data (300 samples analyzed for Cu, Zn, Ag, Pb, Au, W, As, Hg, Ba, Bi, Mo) pertaining to the study area. This geochemical data subset represents a total drainage basin area of ca. 760 km2 (i.e., sampling density of one sample per 2–3 km2).
4. Predictive mapping of prospectivity for porphyry–Cu deposits 4.1. Spatial evidence of prospectivity for porphyry–Cu deposits There are currently two major porphyry–Cu mines located close to the study area (Fig. 3). The analyses presented in this paper aim to obtain a regional-scale answer to the question ‘‘Which parts of the Jebal-Barez are prospective for undiscovered porphyry–Cu deposits?’’ based on empirical spatial associations of known porphyry–Cu deposits with evidential maps of porphyry– Cu prospectivity, including Vz1. Two sets of evidential maps were used. In the first set, we used an anomaly map of Cu (i.e., a pathfinder element for porphyry–Cu deposits), whereas in the second set, we used an anomaly map of Vz1. Both sets of evidential maps commonly include the lithological map, the alteration map derived from Landsat data, and a map of distance to fault lineament intersections. The objective of using these two sets of evidential maps is to compare the performance of geochemical anomalies of a pathfinder element and Vz anomalies in mapping of mineral prospectivity. We applied WofE analysis to quantify the spatial associations of known porphyry–Cu deposits in the study area with individual layers of spatial evidence. In this analysis, we used a unit cell size of 500 m for spatial representation of porphyry–Cu deposits based on the method proposed by Carranza (2008, 2009) for objective selection of unit cell size in data-driven predictive mapping of mineral prospectivity. In WofE analysis with a large number of deposits (say 420), a maximum positive contrast (C) for presence/ absence of evidence is considered a cutoff level for converting evidential data into binary predictor maps. However, in WofE analysis with small number of deposits, say 20 (e.g., Carranza and Hale, 2000, 2002; Carranza, 2004b), the studentized C (i.e., the ratio of C to its standard deviation) is used to judge the strength or statistical significance of spatial association and to a select cutoff level for converting evidential data into binary predictor maps. A studentized C greater than 1.96, for example, suggests a statistically significant positive spatial association at 98% confidence level (Bonham-Carter et al., 1989). We use this criterion for judging the strength of spatial association and for converting individual evidential data layers into binary predictor maps. 4.1.1. Lithologic evidence Six lithologic map units contain known porphyry–Cu deposits (Table 1 and Fig. 3). Unit Ert comprises rhyolitic pyroclastics. Unit Evt comprises mostly partly silicified intermediate volcanics and tuffs. Unit D comprises dacites occurring mainly near the margins granite–granodiorite–quartz-diorite (Gd) in contact with porphyrites (P). Unit Eta3 comprises agglomerate, ash and tuff. Of these lithologic units, unit P has the strongest and the only statistically significant positive spatial association with the known porphyry–Cu deposits. These results show that on a regional scale, lithology is a strong spatial evidence of porphyry–Cu prospectivity. In fact, based on the highest value of studentized C per evidence layer (Table 1), lithology is the second strongest spatial evidence of porphyry–Cu prospectivity in the study area. 4.1.2. Structural evidence Faults generally play important roles in porphyry–Cu mineralization (Ovchinnikov, 1971; Grigorian, 1992). In the study area, faults and fractures are probable structural controls on porphyry–Cu mineralization (Mars and Rowan, 2006). The fault/fracture lineament map, which contains 438 line segments of various types of faults, was initially subjected to cumulative increasing distance approach of WofE analysis. However, the result (not shown) was
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
1939
Table 1 Results of WofE calculations for individual layers of spatial evidence of porphyry–Cu prospectivity. Evidence layer
Evidence class
Area (km2)
Wþ
sW þ
W–
sW–
C
sC
studC
Lithology
Ert Evt Gd Eta3 D P
162.9 47.4 351.4 24.0 16.3 10.2
2 1 8 1 1 2
0.480 0.079 0.150 0.773 1.200 2.522
0.712 1.011 0.358 1.022 1.033 0.791
0.100 0.005 0.148 0.038 0.049 0.132
0.280 0.270 0.381 0.270 0.270 0.280
0.580 0.085 0.297 0.810 1.249 2.654
0.765 1.046 0.523 1.057 1.068 0.839
0.758 0.081 0.569 0.767 1.170 3.165
Distance to fault/fracture intersections (m)
o 100
87.6
6
1.293
0.423
0.395
0.336
1.688
0.540
3.126
100–200 200–300 300–400 400–500 4500
168.0 240.2 304.0 360.1 762.4
6 8 8 10 15
0.612 0.541 0.297 0.353 0.000
0.416 0.360 0.358 0.321 0.000
0.266 0.390 0.258 0.467
0.336 0.381 0.381 0.450
0.879 0.931 0.555 0.819
0.535 0.524 0.523 0.553
1.644 1.778 1.061 1.482
Alteration
Present Absent
83.1 679.3
7 8
1.525 0.521
0.395 0.356
0.521 1.524
0.356 0.396
2.045 2.045
0.532 0.532
3.843 3.843
Cu
1.00–0.80 0.80–0.60 0.60–0.40 0.40–0.20 0.20–0.00
61.9 301.6 546.0 750.5 762.4
3 4 9 15 15
0.931 0.401 0.180 0.016 0.000
0.592 0.503 0.336 0.261 0.261
0.163 0.165 0.282
0.291 0.305 0.414
1.094 0.566 0.463
0.660 0.589 0.533
1.658 0.961 0.867
Vz
1.00–0.80 0.80–0.60 0.60–0.40 0.40–0.20 0.20–0.00
111.5 265.2 444.2 615.8 762.4
5 8 10 13 15
0.850 0.438 0.138 0.072 0.000
0.458 0.359 0.320 0.280 0.261
0.275 0.370 0.276 0.471
0.319 0.381 0.451 0.712
1.125 0.809 0.413 0.543
0.558 0.523 0.553 0.765
2.017 1.545 0.748 0.710
No. of deposits
W þ and W– are positive and negative weights of evidence, respectively. sW þ and sW– are standard deviations of W þ and W–, respectively. C is spatial contrast (¼ W þ –W–). sC is standard deviation of C. studC is studentized C. Rows in bold are spatial evidence attributes per evidence layer with highest values of studC greater than 1.96, which corresponds to 98% confidence level (Bonham-Carter et al., 1989). The evidence classes corresponding to the rows in bold were the bases for creating binary predictor maps from the respective evidence layers.
not useful because the contrast value was statistically nonsignificant, indicating that not all types faults are spatially associated with the known porphyry–Cu deposits. Instead, a subset of intersecting faults/fractures was used. The results of analysis show that zones within 100 m of fault/fracture intersections have the strongest and the only statistically significant positive spatial association with the known porphyry–Cu deposit (Table 1). These results are consistent with the findings of Carranza and Hale (2002) that zones around fault/fracture discontinuities are favorable loci for porphyry–Cu mineralization. These results show that on a regional scale, distance to fault/fracture intersections is a strong spatial evidence of porphyry–Cu prospectivity and, in fact, it is the third strongest spatial evidence based on the highest value of studentized C per evidence layer (Table 1).
4.1.3. Presence of hydrothermal alteration By applying principal component (PC) analysis to Landsat ETMþ data (e.g., Moon et al., 2006), it is possible to extract spatial and spectral signatures that may reflect the presence of hydrothermally altered rocks associated with porphyry–Cu deposits in the Central Iranian volcanic belt (Ranjbar et al., 2004). Results of PC analysis of six Landsat ETM þ spectral bands show that PC3 enhances vegetation because it has high loadings on band 4 (Table 2). PC5 enhances the presence of hydroxyl minerals because it has high loadings of different signs on bands 5 and 7. PC4 enhances the presence of iron-oxides because it has high loadings of different signs on bands 1 and 3. Because of the arid conditions in the study area, it less likely that mineralized areas are strongly characterized by the presence of secondary iron-oxides. In contrast, outcropping porphyry–Cu deposits are likely associated with hydrothermally altered rocks. Thus, the image of PC5, possibly reflecting the presence of hydrothermal
Table 2 Principal components (PC) of six bands of Landsat ETMþ data. Landsat ETMþ bands
PC1
PC2
PC3
PC4
PC5
PC6
Band Band Band Band Band Band
0.22 0.30 0.43 0.24 0.56 0.47
0.18 0.12 0.16 0.05 0.26 0.24
0.03 0.04 0.06 0.72 0.05 0.50
0.50 0.15 0.58 0.36 0.35 0.19
0.07 0.10 0.29 0.09 0.54 0.62
0.48 0.39 0.43 0.19 0.45 0.15
1 2 3 4 5 7
Values in the table are loadings (eigenvectors) on each variable per PC. PC1 with loadings of the same signs represents overall brightness in all bands. Signs of variable loadings in PC2 reflect differences of data in bands covering the visible spectral wavelengths (bands 1–3) and bands covering the infrared spectral wavelengths (bands 1–3). PC3 represents the presence of vegetation because of highest loading on band 4. PC4 represents the presence of iron-oxides because it has high loadings of different signs on bands 1 and 3. PC5 represents the presence of hydroxyl minerals because it has high loadings of different signs on bands 5 and 7. PC6 is similar to PC2 but the former is related to noise in the data. Among the six PCs, PC5 is used to map hydrothermal alteration for mapping of prospectivity for porphyry–Cu deposits in the study area.
argillic alteration in rocks, is considered as a spatial evidence of porphyry–Cu prospectivity. However, based on the spectral characteristics of clays in general (i.e., reflection in band 5, absorption in band 7), the image of PC5 is negated (i.e., multiplied by –1) to portray hydrothermally altered rocks as bright pixels (Fig. 4). The negated PC5 image of the six Landsat ETMþ spectral bands was generalized into two classes—altered rock present and altered rocks absent—after applying a cumulative decreasing value approach of WofE analysis to determine the range of negated PC5 values with the strongest spatial association with the known porphyry–Cu deposits. The classified image was subjected to a new WofE analysis to determine weights for each of the ‘‘present’’ and ‘‘absent’’ classes (Table 1). The positive and
1940
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Fig. 4. Negated PC5 image showing the presence of hydrothermally altered rocks as bright pixels. White dots are locations of known porphyry–Cu deposits/occurrences.
negative spatial associations of the known porphyry–Cu deposits with the present and absent classes, respectively, are statistically significant. This shows that, on a regional-scale, the presence of hydrothermal alteration is a strong spatial evidence of porphyry–Cu prospectivity and, in fact, it is the strongest spatial evidence based on the highest value of studentized C per evidence layer (Table 1).
4.1.4. Geochemical evidence Individual unielement data were interpolated by kriging. The interpolated map of Cu (Fig. 5) was used as spatial evidence of porphyry–Cu prospectivity. For direct comparison of results of WofE analysis, values in the interpolated Cu map and in the interpolated Vz1 map were individually rescaled linearly to the range [0,1]. The rescaled interpolated Cu map was subjected to a cumulative decreasing value approach of WofE analysis, after classifying the map into 20th percentile classes based on sample values per sample but not unit cell interpolated values (Carranza, 2010). Zones of the study area with the highest 20th percentile Cu values have the strongest and the only statistically significant positive spatial association with the known porphyry–Cu deposits (Table 1). Multiplicative geochemical data of Cu*Ag and Pb*Zn were first created. Application of fuzzy c-means clustering (Bezdek et al., 1984) shows two cluster centroids in the multiplicative geochemical data (Fig. 6). Centroid C1 represents background values of Cu*Ag and Pb*Zn. Centroid C2 represents anomalous Pb*Zn values and background Ag*Cu values. The two clusters are separated by a clear threshold value of Pb*Zn. Therefore, high Vz1 values would suggest supra-ore to ore multielement anomalies associated with blind to subcropping porphyry–Cu mineralization (Fig. 1). With this assumption, the study area can be subdivided into at least five zones where possible porphyry–Cu deposits are exhumed to
different levels (Fig. 7). Of those five zones, zones A1 and A2 would be favorable for exploration of subcropping to blind porphyry–Cu deposits whereas the other zones would be favorable for exploration of outcropping porphyry–Cu deposits. In addition to zones A1 and A2, the interpolated and rescaled Vz1 values show a zone in the northwestern part of the study area where subcropping to blind porphyry–Cu deposits possibly occur (Fig. 8). The rescaled interpolated Vz1 map was subjected to a cumulative decreasing value approach of WofE analysis, after classifying the map into 20th percentile classes based on sample values per sample but not unit cell interpolated values (Carranza, 2010). Zones of the study area with the highest 20th percentile Vz1 values have the strongest and the only statistically positive spatial association with the known porphyry–Cu deposits (Table 1). Based on the highest value of studentized C per evidence layer (Table 1), The Vz1 map, compared to the Cu map, is a better spatial evidence of porphyry–Cu prospectivity. However, both geochemical evidence layers (i.e., the Vz1 map and Cu map), compared to the three geological evidence layers (lithology, distance to fault/ fracture intersections, and alteration), have weaker positive spatial associations with the known porphyry–Cu deposits in the study area. This result is explicable, however, because the geochemical evidence layers represent stream sediments, which are materials transported away from their sources, whereas the geological evidence layers represent in situ materials like the mineral deposits under examination. 4.2. Integration of spatial evidence layers Based on the results of WofE analyses of spatial associations between the known porphyry–Cu deposits and the individual layers of spatial evidence (Table 1), binary predictor maps were
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
1941
Fig. 5. Geochemical contour map for Cu. White dots are locations of porphyry–Cu deposits/occurrences.
Fig. 6. Scattergrams of multiplicative Ag*Cu and Pb*Zn data showing two centroids, C1 and C2, in the data.
prepared and the weights (Wþ, W–) were assigned to the corresponding predictor criteria class (Table 3) in each map. The weighted binary predictor maps were integrated with each other to update prior probability of porphyry–Cu deposit occurrence (for details, see Bonham-Carter et al., 1989; Agterberg et al., 1990). Using 500 m as unit cell size for a study area covering 762.4 km2 and containing 15 known porphyry–Cu deposits, the estimated prior probability of porphyry–Cu deposit occurrence is 0.005. Two integration experiments were performed: (1) using
lithological, structural, alteration, and Cu predictor maps; and (2) using lithological, structural, alteration, and Vz1 predictor maps. Results of the first experiment are referred to as the Cuin-WofE model; the second experiment, Vz1-in-WofE model. Cumulative frequency curves of posterior probabilities in both models exhibit similar shapes and inflection points at ca. 96 cumulative percent and at ca. 88 cumulative percent (Fig. 9). However, the inflection points on the cumulative frequency curve for the Vz1-in-WofE model correspond to higher posterior
1942
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Fig. 7. Classification of the study area into five zones based on Vz1 (or Zn*Pb/Cu*Ag) values derived from stream sediment geochemical data.
Fig. 8. Contour map of Vz1 (or Zn*Pb/Cu*Ag) values. White dots represent locations of porphyry–Cu deposits/occurrences.
probabilities, compared to those of the inflection points on the cumulative frequency curve for the Cu-in-WofE model. This shows that the Vz1 map, compared to the Cu map, has stronger spatial association with the known porphyry–Cu deposits and is a better predictor of porphyry–Cu prospectivity. Based on the
inflection points, we classified the respective posterior probability maps into three prospectivity or favorability classes (Fig. 9). Posterior probabilities of porphyry–Cu deposit occurrence in moderate and high favorability zones are greater than the estimated prior probability. Some portions of low favorability
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
1943
Table 3 Weights of binary predictor maps used in WofE analysis of porphyry–Cu prospectivity. Predictor
Criteria
Wþ
sW þ
W–
sW–
C
sC
studC
Lithology Fault/fracture intersections Hydrothermal alteration Cu values Vz1 values
Presence/absence of unit P Within/beyond 100 m Presence/absence Z80/ o 80 percentile Z80/ o 80 percentile
2.522 1.293 1.525 0.931 0.850
0.791 0.423 0.395 0.592 0.458
0.132 0.395 0.521 0.163 0.275
0.280 0.336 0.356 0.291 0.319
2.654 1.688 2.045 1.094 1.125
0.839 0.540 0.532 0.660 0.558
3.165 3.126 3.843 1.658 2.017
W þ and W– are positive and negative weights of evidence, respectively. sW þ and sW– are standard deviations of W þ and W–, respectively. C is spatial contrast (¼ W þ –W–). sC is standard deviation of C. studC is studentized C.
Fig. 9. Porphyry–Cu prospectivity models: (a) Cu-in-WofE and (b) Vz1-in-WofE. Inflection points on each cumulative frequency curves of posterior probability models were used for classifying prospectivity into three favorability classes. See the text for explanations about borehole and subareas marked as I–III.
zones have posterior probabilities of porphyry–Cu deposit occurrence greater than the estimated prior probability. The shapes and areas of moderate and high favorability zones for both models are strongly similar, but the distributions of moderate and high favorability zones differ. The Cu-in-WofE model shows that high favorability zones are mainly in the northern part of the study area, whereas the Vz1-in-WofE model shows the opposite (Fig. 9). To describe the quality of the classified prospectivity maps (Fig. 9), we derived a prospectivity index by taking the ratio of the cumulative percentage of known porphyry–Cu deposits delineated to the cumulative percentage of area covered by high to low prospectivity classes (Table 4). A similar but different quality measure of prospectivity maps was used by Brown et al. (2000) and Porwal et al. (2006). The prospectivity index is akin to the success rate as explained in Table 1. A prospectivity index of 1 would imply a random chance of mineral deposit occurrence, whereas higher prospectivity indices imply higher deviations
from that random chance. Actually, for delineated high prospectivity zones, the Vz1-in-WofE model has a success rate of 60%, whereas the Cu-in-WofE model has a success rate of 40% (Table 4). Thus, based on the prospectivity index it is clear that the high prospectivity class of the Vz1-in-WofE model outperforms the high prospectivity class of the Cu-in-WofE model (Table 4). The results show that the Vz1-in-WofE model is better than the Cu-in-WofE model for constricting and focusing of mineral exploration targets. Based on either model, there seems to be at least three subareas (I–III) that warrant further exploration (Fig. 9). Based on the Cu-in-WofE model it is difficult to say whether any of those three subareas would be favorable for exploration of blind porphyry–Cu deposits. In contrast, based on the Vz1-in-WofE model, subarea III would be favorable mainly for exploration of outcropping porphyry–Cu deposits whereas subareas I and II would be favorable for exploration of blind to subcropping
1944
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Table 4 Performance of models of porphyry–Cu prospectivity in the study area. Model
Prospectivity classes
Cumulative % deposits contained (A)
Cumulative % area covered (B)
Prospectivity index (A–B)
Cu-in-WofE
High High–moderate High–moderate–low
40 80 100
4 12 100
10 6.7 1
Vz1-in-WofE
High High–moderate High–moderate–low
60 80 100
3 12 100
20 6.7 1
Column (A) refers to the success rate or goodness-of-fitting evidential maps with a number (here n ¼15) of known deposits. Column (B) refers to the size of the proverbial ‘‘haystack’’ (i.e., exploration target area), such that the smaller it is the better the chance of finding the proverbial ‘‘needle’’ (i.e., mineral deposit type of interest) in it. The prospectivity index relates to the likelihood that mineral deposits exist in areas with certain prospectivity classes.
porphyry–Cu deposits. In fact, prior to writing of this paper, blind porphyry–Cu mineralization was intersected by borehole exploration in subarea I (Fig. 9) (Borna, 2007; NICICO, 2008).
with those of other methods of geochemical anomaly mapping, is warranted.
Acknowledgments 5. Discussion and conclusions Mineral deposits and associated primary halos are characterized by variations in chemical compositions along both vertical and horizontal directions. Most methods of analyses of anomalies in stream sediment data are concerned with variations in surficial chemical compositions along horizontal directions. However, the concept of the geochemical zonality method allows distinction between sub- and supra-ore anomalies. Therefore, application of Vz indices in mineral prospectivity allows further interpretation about whether delineated favorable areas are attractive for exploration of outcropping or blind mineral deposits. This added-value information from Vz indices is essential in planning for exploration activities. However, integration of maps of Vz indices with other maps used as spatial evidence is important in filtering-out false indications of mineral prospectivity portrayed in every layer of evidence. This is shown by the fact that inflection points on the cumulative frequency curve for the Vz1-in-WofE model relate to higher posterior probabilities, compared to those of the inflection points on the cumulative frequency curve for the Cu-in-WofE model (Fig. 9). This is also demonstrated by comparing the Vz1 map (Fig. 8) with the obtained mineral prospectivity map (Fig. 9b). The former map shows high Vz1 values in the northwestern part of the study area, whereas the latter map portrays low prospectivity in that part of the study area. Thus, care must also be taken in the interpretation of Vz maps because, although Vz indices can reflect geochemical variations in vertical directions, a large proportion of variations in geochemical compositions of earth materials are, in fact, mostly unrelated to mineral deposit. This caveat is particularly important if the method of geochemical zonality is applied to stream sediment geochemical data, which represent earth materials upstream from a sampling site. Although methods for GIS-based mapping mineral prospectivity are now mostly well established, it is important to review which methods of geochemical data analysis result in anomaly maps that, in turn, lead to better models of mineral prospectivity. We have shown in this study case that, instead of using anomalies of pathfinder elements, using geochemical zonality anomalies as one of several evidential maps results in improved mapping of mineral prospectivity. In addition, whereas weights-of-evidence analysis was used in this study, other methods of data representation and integration for mineral prospectivity mapping can be used. Finally, further testing of the geochemical zonality method in mineral prospectivity mapping, and comparing its performance
The research was funded by the School of Mining, Petroleum, and Geophysics at the Shahrood University of Technology. Helpful comments by two reviewers (Dr. Gary Raines and an anonymous individual) led to substantial improvements of the manuscript. We particularly thank Dr. Gary Raines for providing a wide range of insightful ideas and comments throughout the program that stimulated the direction of research and helped draw it to a close. References Aftabi, A., Atapour, H., 2000. Regional aspects of shoshonitic volcanism in Iran. Episodes 23, 119–125. Agterberg, F.P., Bonham-Carter, G.F., Wright, D.F., 1990. Statistical pattern integration for mineral exploration. In: Gaa´l, G., Merriam, D.F. (Eds.), Computer Applications in Resource Estimation. Pergamon Press, Oxford, pp. 1–21. Atapour, H., Aftabi, A., 2007. The geochemistry of gossans associated with Sarcheshmeh porphyry copper deposit, Rafsanjan, Kerman, Iran: implications for exploration and the environment. Journal of Geochemical Exploration 93, 47–65. Beus, A.A., Grigorian, S.V., 1977. Geochemical Exploration Methods for Mineral Deposits. Applied Publishing Ltd., Wilmette, IL 287 pp. Bezdek, J.C., Ehrlich, R., Full, W., 1984. FCM: The fuzzy c-means clustering algorithm. Computers & Geosciences 10, 191–203. Bonham-Carter, G.F., 1994. Geographic Information Systems for Geoscientists: Modelling with GIS. Pergamon, Ontario 398 pp. Bonham-Carter, G.F., Agterberg, F.P., Wright, D.F., 1989. Weights of evidence modelling: a new approach to mapping mineral potential. In: Agterberg, F.P., Bonham-Carter, G.F. (Eds.), Statistical Applications in the Earth Sciences. Geological Survey of Canada, pp. 171–183 Paper 89-9. Borna, B., 2007. Discovery of copper and molybdenum porphyry in the Jebal-Barez, SW Iran. Report of the Geological Survey of Iran, 150 pp. (in Farsi). Brown, W.M., Gedeon, T.D., Groves, D.I., Barnes, R.G., 2000. Artificial neural networks: a new method for mineral prospectivity mapping. Australian Journal of Earth Sciences 47, 757–770. Carranza, E.J.M., 2004a. Usefulness of stream order to detect stream sediment geochemical anomalies. Geochemistry: Exploration, Environment, Analysis 4, 341–352. Carranza, E.J.M., 2004b. Weights-of-evidence modelling of mineral potential: a case study using small number of prospects, Abra, Philippines. Natural Resources Research 13, 173–187. Carranza, E.J.M., 2008. Geochemical Anomaly and Mineral Prospectivity Mapping in GIS. Handbook of Exploration and Environmental Geochemistry, vol. 11. Elsevier, Amsterdam 351 pp. Carranza, E.J.M., 2009. Objective selection of suitable unit cell size in data-driven modeling of mineral prospectivity. Computers & Geosciences 35, 2032–2046. Carranza, E.J.M., 2010. Mapping of anomalies in continuous and discrete fields of stream sediment geochemical landscapes. Geochemistry: Exploration, Environment, Analysis 10, 171–187. Carranza, E.J.M., Hale, M., 2000. Geologically constrained probabilistic mapping of gold potential, Baguio district, Philippines. Natural Resources Research 9, 237–253. Carranza, E.J.M., Hale, M., 2002. Where are porphyry copper deposits spatially localized? A case study in Benguet province, Philippines. Natural Resources Research 11, 45–59.
M. Ziaii et al. / Computers & Geosciences 37 (2011) 1935–1945
Grigorian, S.V., 1992. Mining Geochemistry. Nedra Publishing House, Moscow 310 pp. (in Russian). Guanghua, Z., Zongqi, O., Hui, L., 1996. Exploration Models for Major Types of Gold Deposits in China. Geological Publishing House, Beijing 280 pp. (in Chinese with abstract in English). Hassanzadeh, J., 1993. Metallogenic and Tectono-magmatic Events in SE Sector of the Cenozoic Active Continental Margin of Central Iran (Shahr-E-Babak, Kerman Province). Ph.D. Dissertation, University of California, Los Angeles, 204 pp. Hezarkhani, A., Williams-Jones, A.E., 1998. Controls of alteration and mineralization in the Sungun porphyry copper deposits, Iran: evidence from fluid inclusions and stable isotopes. Economic Geology 93, 651–670. Huber, H., 1969. Geological Map of Iran Sheet No. 6, Southeast Iran, scale 1:1,000,000. National Iranian Oil Company, Tehran. Ivanov, V.V., Meituv, G.M., 1972. Geological and Geochemical Investigations of Ore Provinces. Nedra, Moscow 152 pp. (in Russian). Khorin, G.I., Borodin, B.P., Matveev, A.A., 1992. Recommended Methods for Lithogeochemical Prospecting of Ore Deposits Through Dispersion Flow. Institute of the Mineralogy, Geochemistry and Crystallography of Rare Elements, Moscow 164 pp. Kitaev, N.A., 1991. Multidimensional analysis of geochemical fields. Mathematical Geology 23, 15–32. Mars, J.C., Rowan, L.C., 2006. Regional mapping of phyllic- and argillic-altered rocks in the Zagros magmatic arc, Iran, using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data and logical operator algorithms. Geosphere 2, 161–186. Moon, C.J., 1999. Towards a quantitative model of downstream dilution of point source geochemical anomalies. Journal of Geochemical Exploration 65, 111–132. Moon, C.J., Whateley, M.K.G., Evans, A.M. (Eds.), 2006. Introduction to Mineral Exploration. Blackwell Publishing, Oxford. NICICO, 2008. Drill-holes Exploration in Kerver Area. National Iranian Copper Industries Company (NICICO) Reports, 30 pp. Ovchinnikov, L.N., 1971. Prognostic evaluation of world reserves of metals in land deposits. Dokl, RAS, No. 3, 683–686. Pan, G.C., Harris, D.P., 2000. Information Synthesis for Mineral Exploration. Oxford University Press, Inc., New York 461 pp. Porwal, A., Carranza, E.J.M., Hale, M., 2006. A hybrid fuzzy weights-of-evidence model for mineral potential mapping. Natural Resources Research 15, 1–15. Raines, G.L., Connors, K.A., Chorlton, L.B., 2007. Porphyry copper deposit tract definition—a global analysis comparing geologic map scales. Natural Resources Research 16, 191–198.
1945
Raines, G.L., Mihalasky, M.J., 2002. A reconnaissance method for delineation of tracts for regional-scale mineral-resource assessment based on geologic-map data. Natural Resources Research 11, 241–248. Ranjbar, H., Shahriari, H., Honarmand, M., 2004. Integration of ASTER and airborne geophysical data for exploration of copper mineralization: a case study of Sar Cheshmeh area. In: XXth International Society for Photogrammetry and Remote Sensing Congress, 12–23 July 2004, Istanbul, Turkey, Geo-imagery Bridging Continents, vol. 4, pp. 701–706. Romanco, E.F., Sadat, M.A., 2000. Metallogenic zoning of the Sistan and Baluchestan, Iran. Journal of Geology Science 1, 20–23. Shafiei, B., 2010. Lead isotope signatures of the igneous rocks and porphyry copper deposits from the Kerman Cenozoic magmatic arc (SE Iran), and their magmatic-metallogenetic implications. Ore Geology Reviews 38, 27–36. Shahabpour, J., 1982. Aspects of Alteration and Mineralization at the Sar-Cheshmeh Copper–molybdenum Deposit, Kerman, Iran. Ph.D. Dissertation, Leeds University, 342 pp. Shcheglov, A.D., 1979. Fundamentals of Metallogenic Analysis. Mir Publishers, Moscow 335 pp. (Trans. from Russian by V. Shiffer). Solovov, A.P., 1987. Geochemical Prospecting for Mineral Deposits. Mir, Moscow, 288 pp. [V.V. Kuznetsov, Trans.; Engl. ed.]. Solovov, A.P. (Ed.), 1990. Handbook on Geochemical Prospecting for Useful Minerals. Nedra Publishing House, Moscow 336pp. [in Russian]. Stocklin, J., 1968. Structural history and tectonics of Iran: a review. American Association of Petroleum Geologists Bulletin 52, 1229–1258. Taghizadeh, N., Mallakpour, M.A., 1976. Mineral Distribution Map of Iran, scale 1:2,500,000. Geological Survey of Iran, Tehran. Tangestani, M.H., Moore, F., 2002. Porphyry copper alteration mapping at the Meiduk area, Iran. International Journal of Remote Sensing 23, 4815–4825. Valeh, N., 1972. Geological Map of Iran Sheet 7647—Jebal-Barez, scale 1:100,000. Geological Survey of Iran, Tehran. Ziaii, M., 1996. Lithogeochemical Exploration Methods for Porphyry Copper Deposit in Sungun, NW Iran. Unpublished M.Sc. Thesis, Moscow State University (MSU), Moscow, 98 pp. (in Russian). Ziaii, M., Abedi, A., Ziaei, M., 2009a. Geochemical and mineralogical pattern recognition and modeling with a Bayesian approach to hydrothermal gold deposits. Journal of Applied Geochemistry 24, 1142–1146. Ziaii, M., Pouyan, A.A., Ziaei, M., 2009b. Neuro-fuzzy modelling in mining geochemistry: identification of geochemical anomalies. Journal of Geochemical Exploration 100, 25–36.