Nội dung

Turkish Journal of Earth Sciences Turkish J Earth Sci (2013) 22: 574-587 © TÜBİTAK doi:10.3906/yer-1206-1 http://journals.tubitak.gov.tr/earth/ Research Article Three-dimensional subsurface modeling of mineralization: a case study from the Handeresi (Çanakkale, NW Turkey) Pb-Zn-Cu deposit 1, 1 2 3 Sinan AKISKA *, İbrahim Sönmez SAYILI , Gökhan DEMİRELA Department of Geological Engineering, Faculty of Engineering, Ankara University, Tandoğan, Ankara, Turkey 2 Fe-Ni Mining Company, Balgat, Ankara, Turkey 3 Department of Geological Engineering, Faculty of Engineering, Aksaray University, Aksaray, Turkey Received: 05.06.2012 Accepted: 19.12.2012 Published Online: 13.06.2013 Printed: 12.07.2013 Abstract: The main goal of 3D modeling studies in the mining sector is to address the complex geological, mineralogical, and structural factors in subsurface environments and detect the ore zone(s). In order to solve this complexity, use of quality data (e.g., a wide range of boreholes at regular intervals) is necessary. However, this situation is not always possible because of certain restrictions such as intensive vegetation, high slope areas, and some economic constraints. At the same time, with the development of computer technology, the unused and/or insufficiently considered data need to be gathered and reviewed. This assessment may lead to the detection of potential new zone(s) and/or could prevent unnecessary costs. In this study, the target area that was chosen had inadequate and unusable data, and we used the data as effectively as possible. The Handeresi area is located in the Biga Peninsula of northwestern Turkey. In this area, the Pb-Zn-Cu occurrences take place in carbonate levels of metamorphic rocks or at the fractures and cracks of other metamorphic rocks. The area is being explored actively now. In this study, using the borehole data, we attempted to model the subsurface of this area in 3D using commercial RockWorks2006® software. As a result, there were 3 ore zones that were seen intensively in this area. One of them indicates the area in which the adits are now operating. The others could be new potential zones. Key words: NW Anatolia, Biga Peninsula, inverse distance weighting, kriging, lead, zinc, copper, interpolation method 1. Introduction Three-dimensional (3D) interpretation of subsurface characteristics has been used in the mining sector for a long time. Before the use of state-of-the-art computer software, portrayal of the 3D features was done using two-dimensional (2D) specialized maps, cross-sections, and fence diagrams. Currently, it is possible to construct 3D subsurface models easily using 3D Geoscientific Information Systems (3D GSIS), which have efficient datamanagement capabilities (Rahman 2007). In the last 2 decades, the number of 3D subsurface modeling studies has increased due to the use of computer software (e.g., Renard & Courrioux 1994; de Kemp 2000; Xue et al. 2004; Feltrin et al. 2009; Ming et al. 2010; Akıska et al. 2010b). High-definition 3D models are constructed using the interpolation algorithms of those software programs; in addition, the determination of underground mines and their conditions of formation can be obtained. Because of connections between the study areas and the data that have some differences in all 3 dimensions, 3D GSIS is important (Rahman 2007). The application areas of 3D GSIS are: determining ore and oil deposits (e.g., Houlding * Correspondence: akiska@eng.ankara.edu.tr 574 1992; Sims 1992; Feltrin et al. 2009; Wang et al. 2011), hydrogeological studies (e.g., Turner 1992; Houlding 1994), various civil engineering projects (e.g., Özmutlu & Hack 1998; Veldkamp et al. 2001; Elkadi & Huisman 2002; Rengers et al. 2002; Özmutlu & Hack 2003; Zhu et al. 2003; Hack et al. 2006, Bistacchi et al. 2008), modeling structural factors (Renard & Courrioux 1994; de Kemp 2000; Galera et al. 2003; Zanchi et al. 2009), and establishing settlement areas (e.g., Rahman 2007). The main aim of 3D modeling of ore deposits is to determine the complex geological, structural, and mineralogical conditions in these areas and to detect the location of these deposits in the subsurface environment. With the help of recent 3D subsurface modeling studies, some information can be gained not only about detecting ore locations but also about the formation conditions of the deposits (e.g., Feltrin et al. 2009). Optimizing the subsurface data collected from various sources (boreholes, geophysical methods, well logs, etc.) could minimize the costs of many operations. Because of the complex spatial relationship existing in the subsurface environment, regularly spaced boreholes and good-quality data are necessary to resolve this complexity (Hack et AKISKA et al. / Turkish J Earth Sci al. 2006). However, this situation is not always possible due to economic constraints and the difficulties of field conditions. In Turkey, some institutions such as the General Directorate of Mineral Research and Exploration of Turkey (MTA), Turkish Petroleum Corporation (TPAO), General Directorate of State Hydraulic Works (DSİ), and others have thousands of meters of borehole log data. These data were interpreted using old technology, and much of the information is no longer used today. In fact, some of these data were taken casually and could not be associated with the subsurface characteristics (especially due to the technological deficiencies at that time) and/or could not be interpreted due to inadequacies in the number of boreholes in the study areas. With the development of new technology, these unused data need to be gathered and reviewed. New ore zones can then be detected, and unnecessary costs can be prevented by means of the reviewed data. The goals of this study were to determine the potential Pb-Zn ore zones in the subsurface environment of the Handeresi Cu-Pb-Zn deposit by means of the surface and borehole geologic data, and to provide focus on mining N 26 Stra nd W S zo ne BLACK SEA K m azd as ağ sif Kırşehir massif Afyon zone Bo rno va fly sch AEGEAN SEA zon e Tavşa nlı zo ne Lycian nappes Pam phli an BLACK SEA TU URKEY U R Y ANKARA e tur Su AEGEAN SEA one Sakarya zone İSTANBUL MİR İ İZMİR ISPARTA MEDITERRANEAN SEA 42 lz nbu İsta ture Intrapontide su Thracian basin Menderes massif Study area rea 2. Geological setting The study area is located in the Biga Peninsula of northwestern Turkey. It is situated between the Edremit (Balıkesir) and Yenice (Çanakkale) districts, and lies to the south of Kazdağ Massif in the western section of the Sakarya Zone (Figure 1). This zone is represented by PreJurassic basement rocks that are deformed, and it includes metamorphosed and unmetamorphosed Jurassic-Tertiary units. The area consists of Devonian (Okay et al. 1996) granodiorite rocks called Çamlık granodiorite, a PermoTriassic (Okay et al. 1990) metamorphic sequence called the Karakaya Complex, and Oligo-Miocene (Krushensky 1976) granitoid and volcanic rocks. The common rocks in the metamorphic sequence are sericite-graphite schists, phyllites, and quartzites with metasandstone and marble lenses (Figure 2). The Pre-Jurassic clues of the basement units are strongly overprinted by Alpidic deformations (Okay et al. 2006). 34 ja E operations in specific areas in spite of obstacles such as insufficient borehole units, structural factors, and intensive vegetation. In addition, it is hoped that this study will make a contribution to more detailed modeling studies. 0 200 km Study area 36 Figure 1. Simplified tectonic map showing the location of the main Tethyan sutures and neighboring tectonic units in western Turkey (after Okay et al. 1990; Harris et al. 1994). 575 AKISKA et al. / Turkish J Earth Sci 35/518100 4402200 N 8 BOREHOLES ROAD W RIVER E ACTIVE-DEACTIVE MINING FAULT-PROBABLE FAULT Q S Tha SYNCLINE-ANTICLINE A’ A Q Tha CROSS-SECTION LINE Tha ALLUVIUM VOLCANITES Pzmd METADIABASE Pzş A SCHIST HDK Pzş R E Sİ 27 H A N D E 13 Pzmk 26 Pzmk Pzmk 23 Pzmd Pzş + 20 B’ 5 1 24 + 18 YG 25 B 22 21 Pzş Pzmk METASANDSTONE 16 HDYU 2 3 BB 12 14 4 19 15 Pzmk 10 17 + 9 A’ 8 Pzş 6 + Pzmk Pzmd + Pzmd Pzmk Pzmd 7 + 11 d Pzm Pzmd Pzş Pzmk Pzş 0 100 200 m 4400800 35/519100 Figure 2. Geologic map of the Handeresi area including the cross-section lines between the boreholes; coordinates are given in UTM coordinate system (modified from Yücelay 1976). 576 AKISKA et al. / Turkish J Earth Sci Some vein- and skarn-type lead, zinc, and copper deposits are located in the Permo-Triassic metamorphic sequence (Yücelay 1976; Çağatay 1980; Çetinkaya et al. 1983; Tufan 1993; Akıska 2010). Mineralized zones occur in carbonate levels of metamorphic rocks or at the fractures and cracks of other metamorphic rocks. The main ore mineral paragenesis is galena, sphalerite, chalcopyrite, pyrite, arsenopyrite, and hematite assemblage, while gangue minerals are grossularitic and andraditic garnets, manganiferous hedenbergitic pyroxenes, epidote, quartz, and calcite (Akıska 2010; Akıska et al. 2010a; Demirela et al. 2010). When the Handeresi Pb-Zn-Cu deposit was explored in the early 1970s by the MTA, 27 boreholes were drilled for mineral exploration. The Handeresi deposit is one of the most important Pb-Zn-Cu occurrences in Turkey, with total mineral resources of 3.5 Mt at an average grade of 7% Pb, 4% Zn, and 3000 g/t Cu (Yücelay 1976). In this area, mining activities have been maintained for about 40 years. The deposit is currently mined by Oreks Co. Ltd., which has produced ores from 4 adits in this area. 3. Methods As pointed out previously, subsurface modeling studies are quite important in mining sectors, and detailed modeling studies can be achieved as a result of developments in computer technology. Particularly, assigning adit directions accurately in an underground mining area can reduce the various operating costs associated with mining. GEOLOGICAL MAP Several software programs that are used for surface and subsurface modeling include several spatial interpolation algorithms such as nearest neighbors, inverse distance weighting (IDW), kriging, and triangulated irregular network-related interpolations (Li & Heap 2008). Each software program has its own advantages and disadvantages, but all of them have almost every one of the interpolation algorithms used in modeling studies (Rahman 2007). In this study, modeling was accomplished with commercial RockWorks2006® software, which enables the use and capability of the interpretation in conjunction with all of the data in much less time. This program includes 2 main windows: Borehole Manager and Geologic Utilities. Borehole Manager contains the borehole procedures such as entry, management, and analysis of borehole data. Geologic Utilities has mapping, gridding, and contouring properties (Rahman 2007). In this study, Borehole Manager is used for subsurface modeling and Geologic Utilities is used for surface modeling (Figure 3). In this study, the database is created using borehole, topographic, and geologic data, and a digital elevation model (DEM) is generated via topographic data. DEM is used especially in GIS applications constructing 3D surface modeling. A 3D topographic map is generated by determining the unknown points from certain points through various interpolation methods. That is why choosing the right interpolation method is very important for creating DEMs. Many researchers have revealed the relationship between DEM accuracy and the interpolation SOURCE TOPOGRAPHIC MAP BOREHOLE DATA PROCESSES LITHOLOGY KRIGING LITHO BLEND GEOREFERENCED GEOLOGICAL MAP DEM SURFACE MODELING SOLID MODELING GEOCHEMICAL ANALYSES (Pb%, Zn%, Cu%) IDW SOLID MODELING WITH CUT-OFF GRADES BOREHOLE POINT MAPS IDW 2D CROSSSECTIONS 3D BOREHOLE DATA SUBSURFACE MODELING Figure 3. Organigram of the modeling processes (modified from Kaufmann & Martin 2008) (the words in the parallelograms indicate the interpolation methods). 577 AKISKA et al. / Turkish J Earth Sci technique (Zimmerman et al. 1999; Binh & Thuy 2008; and references therein). Fencík and Vajsáblová (2006) investigated the accuracy of the DEM using the kriging interpolation technique with different variogram models in the Morda-Harmonia area. As a result of this study, the authors concluded that the most appropriate variogram was a linear model. Chaplot et al. (2006) created DEMs using various interpolation techniques (kriging, IDW, multiquadratic radial basis function, and spline) in regions of France and Laos. The authors concluded that all interpolation techniques showed similar performance in the regions with dense sample points, while IDW and kriging were better than the others in regions with low density sample points. However, the study carried out by Peralvo (2004) in 2 watersheds of the Eastern Andean Cordillera of Ecuador showed a different result. According to this study, the IDW interpolation method produced the most incorrect DEM. In the evaluation of these studies referred to by Binh and Thuy (2008), the authors noted that the studies showed contradictory results due to the differences in technological application levels, research methods, and the types of topography in different countries. In their study, Binh and Thuy (2008) created DEMs via 3 interpolation techniques in 4 different areas in Vietnam using digital photogrammetry and total station/GPS research methods. As a result, regularized spline interpolation is the most suitable algorithm in mountainous regions, while IDW or an ordinary kriging interpolation algorithm with the exponential variogram model is recommended in hilly and flat regions (Binh & Thuy 2008). When all the data are evaluated together, even though some points are important while choosing the interpolation method that creates the DEM, there are no specific rules for choosing the interpolation algorithms. Nevertheless, considering the work done by Binh and Thuy (2008), because the area in this study includes flat and hilly areas, the kriging interpolation method is preferred for surface modeling. Kriging (Krige 1951; introduced by Matheron 1960) is the generic name of generalized least-squares regression algorithms (Li & Heap 2008). This method is a wellknown geostatistical interpolation method that weights the surrounding measured values to derive a prediction for an unmeasured location (Cressie 1990). This algorithm is an estimation process that determines the unknown values using the known values and variograms. Kriging is considered the most reliable method for geological and mining applications (Rahman 2007). The most important advantage of kriging, compared with other estimation methods, is that the weights are determined via certain mathematical operations instead of randomly. The data are analyzed systematically and objectively; as a result of this analysis, weights that will be used in variogram functions 578 are calculated (Tercan & Saraç 1998). Another advantage of this method is that it gives the error estimation via the kriging variance. The kriging variance does not depend on the exact values of the data; it is a function between the numbers of data and the distances of data (Tercan 1996). Very close estimation of data generated by the kriging interpolation algorithms to the real values depends on the number of samples, the frequency of data, and the degree of accuracy of the variogram model and parameters (Brooker 1986; Chaouai & Fytas 1991). The method creates variogram models of the data set that represent the relation of the variance of the data pairs with distance. This variogram indicates the extent of the spatial autocorrelations and the variogram models that could be isotropic or anisotropic, depending on the directional variability of the data. The unknown values are predicted based on the variogram model (Cressie 1990). Most frequently used in several variogram models are spherical, exponential, linear, and Gaussian models (Burrough & McDonnell 1998). While doing subsurface modeling, RockWorks2006 performs solid modeling. Solid modeling is a grid process in 3 dimensions, which creates a cube from regularly spaced nodes derived from irregularly spaced data. During 3D modeling, the subsurface is divided into cells that have specific dimensions called voxels, and the geologic units that correspond to these cells form the cubes. Each voxel created is identified by the corner points, called nodes. Each node has an x, y, and z location coordinate, and a g value, which in this study is a geochemical analysis value. In this study, 2 different solid models are made. The first model is applied to “ORE ZONE”, shown in the boreholes. The second model is applied to Pb%, Zn%, and Cu% values obtained from these ore zones. In the first model, RockWorks2006 uses a solid modeling algorithm that is designed specifically to interpolate lithologies in the boreholes. Using this algorithm, which is called “litho blend”, the subsurface is separated into block diagrams and all lithologies are modeled (RockWare 2006). In this study, all lithologic units are modeled; however, only “ORE ZONE” is used for the purpose of the study. In the second model, in order to model the percentage distribution of ore zones in the boreholes, Pb%, Zn%, and Cu% values are modeled with 3D solid modeling. In this modeling study, to generate the block diagrams in the subsurface, the IDW interpolation method is preferred, which makes a distinction with respect to the similarity of degrees of the measured points. In other words, in the estimation of the unknown points, it gives more weight to the closest known points instead of the remote ones. IDW is very versatile and an easily understandable programmable method. In addition, it gives very accurate AKISKA et al. / Turkish J Earth Sci results in wide-range data interpretation (Lam 1983). The most important feature of this method is that it is able to quickly interpolate the scattered data in the regular grids or the irregularly spaced data (Li & Heap 2008). 4. Geostatistical analysis of the surface data The number of x, y coordinates and z elevation data points in the area is 5292. These data are digitized from the topographic map from Yücelay (1976). The topographic map has a 10-m contour interval of the study area. The information about the survey method was not given by Yücelay (1976). As mentioned above, the modeling studies are carried out using RockWorks2006 commercial software. However, for the geostatistical analysis, more comprehensive software is needed. Therefore, the geostatistical analysis is done with the Geostatistical Analyst Tool in ArcGis9® (Johnston et al. 2001). The changes depending on the distance of the difference between regionalized variable values are revealed with the variogram function in geostatistics (Tercan 1996). When the variogram is calculated in different ways, it sometimes exhibits different behaviors (Armstrong 1998). Anisotropy is used for calculating the directional effects in the semivariogram model, which is made for surface calculations. It is characteristic of a random process that indicates higher autocorrelation in one direction than another (Johnston et al. 2001). In this study, the surface data indicate anisotropy. The range values are different while the sill values are the same in the variograms calculated in different directions in this study. This also shows that the surface has geometric anisotropy (Armstrong 1998). It can be seen that the major axis of the anisotropic ellipse is trended NE-SW (Figure 4a). Experimental variograms have been calculated in 4 directions, which are N-S, E-W, NE-SW, and NW-SE. The lag size is 100 m and the angle tolerance is 45°. The experimental variogram has been fitted by an “exponential variogram” model (Figure 4b) that represents the direction of maximum continuity 55° from the north. In the experimental variogram, the sill a) b) value is 5392.6 m, the range is 1280.39 m, and the nugget value is 10 m. Neighborhood estimation, which defines a circle (or ellipse) including the predicted values on unmeasured points, is used to restrict the data (Johnston et al. 2001). While interpolating each grid node, the search ellipse defines the neighborhood of points to consider. Outside the search ellipse, the data points are not taken into account (Fencík & Vajsáblová 2006). In most cases, the search ellipse range and direction coincides with the anisotropy range and direction. At the same time, to prevent the tendency of particular directions, this circle (or ellipse) is divided into sectors. In this study, for determining the search ellipse, the anisotropy range and direction are used automatically and the ellipse is divided into 4 sectors. The maximum number of samples chosen is 6 for neighborhood estimation. Cross-validation is used to control all numbers of data points (5292 points) used in interpolation. The graphic and table obtained after the cross-validation analysis are shown in Figure 5 and Table 1, respectively. For perfect prediction, the estimation errors should be symmetrically distributed, and linear regression of exact values on estimated values should be close to a 45° line (Saraç & Tercan 1996). The needed criteria for the best created DEM were given by Johnston et al. (2001): • Standardized mean nearest to 0. • Smallest root mean square (RMS) prediction error. • Average standard error nearest to the RMS prediction error. • Standardized RMS prediction error nearest to 1. Both the predicted values are nearly the same as measured values and the prediction error values indicate the satisfactory result of the interpolation (Figure 5; Table 1). 5. Three-dimensional subsurface modeling of mineralizations The study area covers 1.4 km2 (1 × 1.4 km) and elevation ranges from 270 m to 520 m. The surface has been divided γ×10-4 1.3 1.04 0.78 0.52 0.26 0 1.62 3.24 4.86 6.48 8.1 9.72 11.34 12.96 Distance, h×10-2 Figure 4. (a) Anisotropic ellipse showing NE-SW trend and the direction of the variogram, (b) experimental variogram in the direction of the major axis of the anisotropic ellipse. 579 AKISKA et al. / Turkish J Earth Sci Surface data (kriging) 550.00 R² = 0.9999 Predicted values 500.00 450.00 400.00 350.00 300.00 250.00 250.00 300.00 350.00 400.00 450.00 Measured values 500.00 550.00 Figure 5. Cross-validation scatter plot of the surface data. into 10 × 10 × 10 m blocks (490,000 total voxels). In the northwestern and southeastern parts of the area, hilly topography with gentle slopes is seen while the Handeresi River flows from the northeast to southwest. In order to create surface modeling, the topographic map (1/1000) of the study area (Yücelay 1976) is digitized, and x, y, and z values are entered into the Geologic Utilities section of RockWorks2006. Using these values, the software creates a grid-based file. While creating the grid file, the kriging method is used as an interpolation algorithm. One of the important features of RockWorks2006 software is that it is able to choose the most appropriate variogram that analyzes all the data automatically in the kriging interpolation method calculations. In this study, the “Exponential with nugget” variogram determined in accordance with the analysis of the software is preferred. Choosing the “Exponential with nugget” variogram automatically shows that the results in this analysis are also compatible with the results of geostatistical analysis. The 3D topographic surface modeling is intersected with a georeferenced geological map. Finally, the borehole point maps and the adits are done by drawing 3D borehole multilogs (Figure 6). The subsurface has been divided into 5 × 5 × 5 m blocks. “ORE ZONE” applied to the litho blend algorithm and Pb%, Zn%, and Cu% values applied to IDW algorithm have 4,010,151 and 2,154,921 total voxel values, respectively. Table 1. The summary statistics of the prediction errors using kriging interpolation with “Exponential with nugget” variogram. Prediction errors Samples Mean RMS Average standard error Mean standardized RMS standardized 580 5292 –0.007349 0.7163 1.6 –0.0002796 0.06106 In order to create subsurface modeling, 27 borehole data are taken from Yücelay (1976). The shallowest drilling is 60 m (S-01) and the deepest drilling is 245.65 m. (S-13). Total drilling depth is 4239.15 m while the average drilling depth is 157 m. The ore zones are observed in 6 out of 27 boreholes (S-04, S-06, S-14, S-15, S-19, and S-21). In order to determine Pb%, Zn%, and Cu% values in the ore zones, geochemical analysis was carried out according to the methods of Yücelay (1976). All of these values ​​are entered into the RockWorks2006 software separately without any modification. The “ORE ZONE”, which is detected from drilling cores, is modeled via solid modeling. Here, while creating the block diagrams, the software makes the solid models of the lithologies using the litho blend algorithm (RockWare 2006). This algorithm is used to interpolate and extrapolate numeric values that represent “ORE ZONE” in the lithology class. Grid nodes between the boreholes are assigned a value that corresponds to the “ORE ZONE” section in the lithology class and relative proximity of each grid node to surrounding boreholes (Sweetkind et al. 2010). The model is intersected with topographic surface modeling. However, because of the insufficient number of drilled boreholes, the accuracy of the modeling of areas that are outside of the drilled area (Figure 7) is arguable. Using the Pb%, Zn%, and Cu% geochemical analysis results, the model files are constructed separately using the IDW interpolation method. The parameters of the IDW interpolation method and 3D grade results are shown in Table 2 and Figure 8, respectively. The ore zones determined in the model files, due to the existence of ore zones in almost all boreholes, do not provide any focus area. One of the aims of this study is to lead to more detailed studies and to focus the ore zone(s) into more restricted areas. Determining of the area(s) in which Pb, Zn, and Cu mineralizations above the cut-off grades is thought to be ensured, as much as possible, close to the purpose described above. For this purpose, using Pb%, Zn%, and Cu% values with the above cut-off grades in all ore zones creates a database in RockWorks2006. In this software, this kind of subsurface data (such as those representing geochemistry, geotechnical measurements, etc.) is possible to model in 3D (I-data tool; RockWare 2006). Using this tool, RockWorks2006 interpolates the downhole interval-base data into a solid model. Solid modeling is implemented separately for the chemical analysis belonging to each element (Pb.mod file for Pb% modeling, Zn.mod file for Zn% modeling, and Cu.mod file for Cu% modeling; Figure 9a). As mentioned above, the ore zones that exist in almost all the boreholes reflect this modeling study, and large areas are detected for each element in the subsurface environment. Choosing a target area is difficult when the results are considered together. That is why using the intersections of all element zones AKISKA et al. / Turkish J Earth Sci a) S-11 S-07 S-22 S-05 S-066 S-20 S-09 S-23 S-08 S-17 S-26 S-14 S-13 S-27 S-10 S-12 S-16 S-24 S-15 S-02 S-25 S-18 S-033 S-01 S-19 S-04 S-21 Lithology FAULT ZONE MARBLE 35.518200 METADIABASE 35.518600 METASANDSTONE ORE ZONE SCHIST SERPENTINITE 4402000 35.519100 400 S W 400 300 300 200 200 100 100 BB ADIT YG ADIT HDYU ADIT HDK ADIT N E b) S-11 S-07 S-08 0 S-05 S-06 S-22 S-20 S-23 S-09 S-17 S-14 S-10 S-15 S-12 S-03 S-13 S-16 S-02 S-04 S-19 S-26 S-01 S-27 S-24 S-25 S-18 S-21 Figure 6. (a) 3D topographic surface modeling with borehole points, adits, and the geological map of the Handeresi area (to avoid confusion, –500 m offset is applied to the geological map (Figure 2) and +500 m offset is applied to the 3D topographic surface modeling along the z-axis). (b) 3D boreholes and adits of the Handeresi area. 581 AKISKA et al. / Turkish J Earth Sci Lithology FAULT ZONE MARBLE METADIABASE METASANDSTONE ORE ZONE SCHIST SERPENTINITE ? ? Figure 7. Topographic surface modeling, boreholes, and “ORE ZONE”, which is modeled with solid modeling of the Handeresi area (to avoid confusion, +500 m offset is applied to boreholes and upper surface modeling image along to z- axis); side of view: from NE. (Pb%, Zn%, and Cu%) above the cut-off grades is more suitable for choosing a target area. If there are not enough boreholes without regular intervals, and if we do not detect the ore zones more precisely using these data separately, intersecting the areas above the cut-off grades gives the most promising fields. The probability of the presence of ores in these fields is the greatest. The purpose here is primarily to evolve model files in which the values above the cut-off grade get “1” and the values below the cut-off grade get “0”, and then to determine the “1” value in the file resulting from multiplying these model files with each other. In this latest model file, the areas having a “1” value indicate intersection of above the cut-off grade of Pb%, Zn%, and Cu%. In order to determine intersecting area(s) with the help of some arithmetic operations, new models need to be established. These operations are described below. Rockworks Utilities includes several modeling tools, such as generating or making changes to a solid model. These are displayed under the Solid menu. The Solid/ Boolean Operations/Boolean Conversion tool converts the real number solid model file to a Boolean (true/false)type solid model file. In this process, the tool assigns a “1” if G-values of nodes fall within a user-defined range or assigns a “0” if they do not (RockWare 2006). In this study, the percentage value of the elements is assigned to each solid model file as a G-value. The Pb% model file (Pb.mod) is chosen in the Boolean Conversion tool, and a value of “1” is assigned to 7% (Pb cut-off grade) and higher values. All values below 7% are accepted as “0” and a new model file consisting of Boolean values (Pb_boolean.mod) is created. All of these processes are applied separately to Zn% values with a 4% cut-off grade and Cu% values with a 0.3% cut-off grade, and Boolean model files are created (Zn_boolean.mod and Cu_boolean.mod, respectively). For the next step, the Solid/Math tool, which includes the arithmetic operation, is applied to solid models. The Pb% Zn% Table 2. The parameters of the IDW interpolation method. 582 Weighting exponent 4 Max. points per voxel 64 Max. points per borehole 32 Sector width 90° Sector height 90° Cu% Figure 8. 3D grade model of the Pb%, Zn%, and Cu% distributions in the subsurface environment; side of view is the same in Figure 9. See Figure 11 for colored interval legends. AKISKA et al. / Turkish J Earth Sci The Model&Model tool applies arithmetic operations to the values of 2 model files and creates a new model file. In this study, the multiplication operation is applied to Pb_boolean.mod with Zn_boolean.mod files. As a result of the multiplication operation, the areas that include “1” values in both files (Pb and Zn cut-off grades and the higher values) indicate “1” values, while the other areas indicate “0” values in the newly created Boolean model file (Pb_Zn_boolean.mod). The multiplication operation is then applied for the generated file (Pb_Zn_boolean. mod) and the Cu_boolean.mod file. The created file (Pb_ Zn_Cu_boolean.mod) at the end of this process includes “1” values that are assigned to Pb%, Zn%, and Cu% cut-off grades and higher values; the others include “0” values. The visualization of this modeling and schematic representations of these processes are shown in Figures 9b and 10, respectively. In order to observe spatial distribution modeling in 2D, 2 cross-section lines (A-A’ and B-B’) are drawn containing the boreholes that take place in the possible ore zone areas (see Figure 2 for cross-section lines). In the first crosssection (A-A’), distribution of the ore zones is observed between and around S-01 and S-02 boreholes at elevations of about 225–300 m and around the S-17 borehole at elevations of 350–375 m. In the second cross-section (BB’), distribution of the ore zones is observed around the S-23 borehole at elevations of about 215–240 m (Figures 11a and 11b). HDK and HDYU adits are situated near the S-01 and S-02 boreholes and the YG adit appears between the S-18 and S-01 boreholes in cross-section A-A’ (Figure 6). There are not any operating ore zone(s) and adits in the ore zones near the S-17 borehole in cross-section A-A’ or near the S-23 borehole in cross-section B-B’. In the modeling study, it is important to correlate with measured values and predicted values, which are revealed a) Pb.mod Zn.mod Cu.mod b) Figure 9. (a) The view of the “Pb.mod”, “Zn.mod”, and “Cu.mod” files after the solid modeling process is applied. (b) Visualization of the intersected zone after the Boolean-type solid modeling and mathematical operations are applied to “Pb.mod”, “Zn.mod”, and “Cu.mod” files. options (Model&Model, Model&Constant, and Resample) within the Solid/Math tool are applied to arithmetic operations on the values in the solid model files previously created; this generates a new solid file (RockWare 2006). 0 0 0 1 0 1 1 1 1 0 1 1 1 1 0 0 1 1 0 0 0 0 1 0 0 x Pb_boolean.mod 0 0 0 0 0 0 1 1 1 0 1 1 1 1 0 0 1 1 0 0 0 0 0 0 0 0 1 1 1 0 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 = Zn_boolean.mod 0 0 1 0 0 Pb_Zn_boolean.mod x 0 0 0 0 0 1 1 1 0 1 1 1 1 1 1 1 0 0 1 1 Cu_boolean.mod 0 0 0 0 0 0 1 1 1 0 1 1 1 1 0 0 1 1 0 0 0 0 1 0 0 Pb_Zn_boolean.mod 0 0 0 0 0 = 0 0 0 0 0 0 1 1 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 Pb_Zn_Cu_boolean.mod Figure 10. Schematic representations of the mathematical operations that are implemented to Boolean-type files. The “1” values indicate levels above the cut-off grades, while “0” values indicate levels below them (the values on these figures were arbitrarily selected). 583