Geologic modeling of metal zonation and trends- A geology-based approach to account for local variations in mineralization

In many mineral deposits metal zonation is present and can be related to grade trends. If the geology of the host rock possesses certain geochemical and/or geophysical properties (such as porosity) then the resulting mineral deposit could form zoned ore bodies (disseminated ore) with locally varying mineralization. Disseminated metal of varying degrees poses a significant challenge to grade estimation since it is possible that no clear boundary exists in which to differentiate distinct populations. Therefore, properly defining grade/geology domains may prove elusive. Current resource estimation workflows can sometimes address this problem reasonably well by using pairwise relative variograms and localizing the kriging estimate. However, these geostatistical techniques may not fully account for all of the geology components or account for local variations in mineralization by assuming a single direction of continuity. By using geologic knowledge and current software technology to model the metal zonation and associated trends, local variations in mineralization can be properly accounted for in grade estimation using nonlinear block search paths. A subset of the Galore Creek gold data was used as a test case. Geology overview Located in the Late Triassic Galore Creek alkalic copper-gold (Cu-Au) porphyry district in northwestern British Columbia, the Central zone deposit represents the end member of the silicaundersaturated class of alkalic-porphyry systems (Micko, 2010). The principal economic resource is the Central zone deposit, which consists of three high-grade centers: the Au-rich north and south gold lenses and the Cu-dominated central replacement zone (Micko et al., 2014). The primary host rocks in the Central zone consist of a series of late Triassic volcanic and volcaniclastic units within an unnamed stratigraphic succession. A post mineral syenite porphyry intrudes the volcanic and volcanoclastic units immediately to the west of the Central zone with sills cutting the volcanic/volcaniclastic units. Alteration consists of several potassic, calc-potassic and calcic assemblages. Several moderately dipping (to the northwest) north-striking faults cut the volcanic units and provide a fluid pathway for mineral deposition and the observed alteration assemblages. Mineralization and domain analysis Gold mineralization within the Central zone is primarily associated with potassic and calcpotassic alteration. In particular, the calc-potassic alteration contains the majority of higher gold grades with potassic and calcic-alteration assemblages hosting decreasing concentrations of mineralization. Furthermore, the highest gold grades are generally associated with a northeast striking bornite core within the calc-potassic alteration zone. According to Micko (2010), these data suggest that the interaction of sulfate-rich (SO4²−(aq)) fluids with varying amounts of Fe²+- bearing minerals in host rocks increased H2S/ SO4²−, leading to the formation of reduced S and precipitation of sulfide minerals. While various alteration and mineralogical assembles are associated with different distributions of gold, no assemblage could be separated as a unique gold population when viewed in a geological context. Because the geospatial location of gold observed was not coincident with or completely controlled by any of these factors as continuous mineralization spans across adjacent alteration and mineralogical zones, no distinct gold domains could be derived from alteration or mineralogy. In other words, no single measurable or distinct geologic factor could account for the observed gold mineralization. Additionally, gold mineralization can gradually or more abruptly end within any particular alteration or mineral assemblage. Therefore, more extensive work was required to better define the gold domains. All ore deposits require a path for fluids to migrate away from an origin source. Three moderately dipping north-striking faults served as fluid conduits for pregnant solutions to migrate vertically within the Central zone. Alteration, mineralogy and gold mineralization appear to follow these faults, producing grade trends, but only account for part of the geospatial distribution of gold. Furthermore, volcaniclastic rocks contained the necessary permeability for fluids to migrate laterally away from feeder structures (Fig. 1). Combined, the structures, redox gradients and the observed rock permeability within volcaniclastic horizons help explain the broad alteration patterns and geospatial distribution of gold (disseminated ore). Figure 1 A north-looking sectional view of generalized fluid migration channels within the Central zone. White arrows indicate proposed directions of fluid flow from depths along feeder structures (blue lines) and laterally along permeable volcaniclastic horizons, which concentrated the gold. V1 and V2 volcanic/volcanoclastic units are marked. Pink horizontal shapes represent post-mineral sills. Distance units are in meters. Figure 2 A north-looking sectional view of inferred fluid flow. The solid arrows indicate directions of fluid flow through proposed feeder structures and permeable volcaniclastic rocks. The dashed arrows indicate perpendicular directions of permeable seepage. Distance units are in meters. A distinct grade pattern appears where fluids migrated through volcaniclastic horizons and along identified faults from the drill logs. Here, the gold grades increase to form curvilineargrade trends along fluid pathways. A closer look at the permeable volcaniclastic horizons show distinct patterns of zonation perpendicular to the proposed directions of primary fluid flow across these units. Gold mineralization exists adjacent to directions of primary fluid flow but gradually decreases with increasing distance away from these fluid pathways. This relatively gradual decrease in grade can be explained by the slower percolation of fluids (less fluid flow) away from the primary sources of fluid flow. The volcaniclastic rocks are permeable and allowed fluid to percolate through perpendicular to the main horizons, which acted as fluid conduits (Fig. 2). Lower fluid flow carried less gold in solution from the main paths of fluid migration, which resulted in the observed grade trends and associated metal zonation patterns. Further evidence supporting the presence of metal zonation can be found by constructing a contact plot between mineralized and unmineralized material. The purpose for generating contact plots is not to observe the behavior of the gold grade at the mineralized boundary, but rather, to observe the behavior of the gold grade within the mineralized zone (Fig. 3). The gold grades gradually increase as distance from the edge of mineralization increases. In other words, the gold grade increases toward the center of mineralization (recognized as feeder structures). The presence of a trend within the mineralized zone suggests that an assumption of stationarity within the gold domain may not be appropriate for gold mineralization in the Central zone. Metal zonation and associated grade trends pose difficulties to resource geologists employing conventional geostatistics due to the requirement of second-order stationarity within a given domain. Traditional geostatistical methods account for trends or metal zonation in simulation by removing the large-scale deterministic trend component from the random variable to satisfy the assumption of stationarity (Qu, 2018). In grade estimation, localizing the grade estimates and modeling pairwise relative variograms are two approaches to dealing with trends or metal zonation. However, these geostatistical approaches may not fully consider the geological controls on mineralization, which is critical for understanding the geospatial distribution of the metals of interest. However, by incorporating geological knowledge into the modeling workflow, metal zonation and their associated trends can be modeled and included for use in the resource estimate. Figure 3 A contact plot between mineralized (>0.1 ppm) and unmineralized ( Figure 4 A north-looking sectional view of modeled metal zonation for gold in the Central zone using Leapfrog software. Blue solid represents the boundary of mineralization. Inner yellow solid represents higher-grade core where gold grades increase systematically. Background colors represent the volcanic/volcaniclastic and post-mineral porphyry units. Distance units are in meters and gold assay data is in ppm. Modeling workflow The first step in the modeling process to account for trends is to construct closed three-dimensional solids representing the observed metal zonation (Fig. 4). The purpose for modeling the metal zonation is to account for the local variations in mineralization in 3D space that are due to the presence of faults (for example, in fluid pathways or trends) and rock permeability (metal zonation) in the case at the Central zone. Ultimately, the modeled solids will be used as input to derive the dynamic block searches for grade estimation. Hence, the modeler is using geologic knowledge to design a dynamic search path for the blocks to use in grade estimation. The boundary of mineralization should be modeled first to establish the zone of interest and to provide a limit for which blocks are to be estimated. This zone, within the Central zone, forms a broad halo around the feeder structures and represents the extent to which gold has been transported via solution through permeable host rocks. However, this boundary is not a hard boundary for composites used in grade estimation and represents gold deposition under similar geologic conditions and to provide the grade estimation within a geologically defined curvilinear block search path (Fig. 5). This boundary must be continuous and must account for the observed geological controls on mineralization. Hence, random shapes should not be considered, but rather, geologic knowledge must be used when constructing the boundary (such as for structures, permeability and fracture zones). Following the construction of the mineralized boundary, an inner core should also be constructed for the purpose of defining a dynamic block search path for inner blocks or blocks that are close to the core of mineralization. The inner core reflects local geologic conditions surrounding the feeder structures within the Central zone and possesses a lower coefficient of variation compared to the boundary zone (1.75 for the boundary zone compared to 1.27 for the inner core). Once again, this inner core should only be constructed where continuity exists and geological controls are understood and will provide a geologically defined curvilinear search path to use in grade estimation for inner blocks. Additionally, it must be noted that modeling the observed metal zonation with geological inputs will account for any large-scale trends in the data. Within the Central zone a post-mineral porphyry body disrupts the mineralization and creates zones of unmineralized rock cutting through the primary host rocks. These postmineral sills must be accounted for in the model, but not at the early stages of the metal zonation modeling workflow. Constructing the metal zonation solids should temporarily ignore the post-mineral units to generate continuous modeled solids. By doing so, the modeler is providing an uninterrupted, continuous search path for use later in the grade estimation. For the paths to be used correctly, continuous solids are required (Fig. 5). Figure 5 A north-looking sectional view of desired estimation search paths (drill holes shown). White arrows represent desired block search paths as defined by the input metal zonation solids (blue and yellow). Notice the search path preserves, the higher-grade core (from drill holes) and the lower-grade outer zones by accounting for the dynamic changes in grade. Distance units are in meters. Figure 6 Sectional-view looking north of selected blocks and their associated estimation search paths. Modeled metal zonation solids in blue and yellow. Selected blocks (red) are shown with the calculated block estimation search paths (white-dash lines). Block search paths are determined by the modeled solids and are dynamic. Figure 7 A sectional view (looking north) of gold-grade (ppm) block estimation with drill holes (outlined in white). Post-mineral dikes outlined in pink. Thin yellow line marks the boundary of the inner solid and the thin blue line marks the boundary of the outer solid used to guide the dynamic block search path. High and low grades are preserved at the boundaries of the solids as expected in a deposit with metal zonation. Distance units are in meters. The last step in the modeling process involves merging the solids. This is a critical step in the modeling workflow since the mineralized zone will consist of a single-estimation domain representing the mineralized material. Combining the solids into a single solid provides the necessary estimation search path for all blocks within the domain. Hence, blocks located in the interior portions of the mineralized zone will have estimation search paths dictated by the path along the inner solid, while blocks located in the outer portions of the mineralized zone will have estimation search paths dictated by the path along the outer solid (Fig. 6). Blocks that reside in between the inner and outer solids will have estimation search paths dictated by the shape of the inner and outer solids. This ensures that the block search paths will not abruptly change on a block-by-block basis and will, in fact, largely preserve the metal zonation correctly. Grade estimation After the solids have been merged, the resulting solid can be used to generate experimental variograms and also used in grade estimation using HxGN MinePlan software. For further details describing the files and procedures required to set up and run the dynamic grade estimation, refer to Cardwell and Cartwright (2016). Table 1 lists several of the kriging parameters used for gold-grade estimation. However, the kriging setup was not necessarily optimized as kriging optimization is not the focal point of this article. Instead, the purpose of developing this kriging setup was to test if the software could handle the required geological inputs (metal zonation). Since the modeled solids captured the nature of gold zonation, hard boundaries were not used. A maximum of three composites per drill hole were used with no anisotropy to provide some level of smoothing for estimation. Two kriging runs were performed; the first run for the highergrade core used no outlier restrictors or capping, and the second run for the lower-grade edges of the mineralized boundary used a high-grade restrictor of 0.21 ppm to prevent grade smearing. Results from the kriging runs proved successful as the estimation executed as designed. Figure 7 shows an example from a section of the kriged model where the block grades clearly followed the modeled zonation (solids). Areas of grade transition were preserved as expected (visual validation from the drill holes) in a deposit with metal zonation, as were the higher-grade core and lower-grade perimeter. By using geologic knowledge to help construct the observed metal zonation it is possible to perform grade estimation by using the curvilinear nature of metal zonation (in the presence of trends). However, it must be noted that no swath plots, global-bias plots, change of support, gradetonnage curves, and so on, were considered for this study since statistical validation was beyond the scope of this article. Conclusions Metal zonation has proven difficult to account for and poses challenges to grade estimation. Current geostatistical methods address this challenge by essentially removing the trends or metal zonation during simulation, or by considering local stationarity when performing grade estimation. However, necessary geological input can be missing when applying these geostatistical methods, which could produce estimates lacking confidence. By understanding the geology and modeling the trends and metal zonation appropriately, domaining and geostatistical methods can be modified to increase confidence in final grade estimates. HxGN MinePlan 3D software allows the user to incorporate solid modeling that accounts for trends and metal zonation into the estimation workflow for increased confidence in grade estimation. Table 1 Basic estimation parameters used for grade estimation. Five meter blocks (cubed) were used for estimating. References Boisvert, J.B., Manchuk, J.G. & Deutsch, 2009, C.V. Kriging in the Presence of Locally Varying Anisotropy Using Non-Euclidean Distances, Math Geosci v. 41, pp. 585-601. Cardwell, J. N., Cartwright, Alyson, 2016, Dynamic Unfolding–Complex Geology Case Study of Tenke-Fungurume Deposits, Mining Engineering, v. 68, no. 1, pp. 20-27. Micko, J., 2010, The Geology and Genesis of the Central Zone Alkalic Copper-Gold Porphyry Deposit, Galore Creek District, Northwestern British Columbia, Canada, Doctor of Philosophy, University of British Columbia. Micko, J., Tosdal, R. M., Bissig, T., Chamberlain, C. M., and Simpson, K. A., 2014, Hydrothermal Alteration and Mineralization of the Galore Creek Alkalic Cu-Au Porphyry Deposit, Northwestern British Columbia, Canada, Economic Geology, v. 109, no. 4, pp. 891-914. Qu, J., 2018, A practical Framework to Characterize Non-Stationary Regionalized Variables, Doctor of Philosophy, University of Alberta.

Geologic modeling of metal zonation and trends- A geology-based approach to account for local variations in mineralization

Authored by:
Justin Cardwell

To view the past issues of the UCA magazine you must be a member of SME/UCA.

EnergyX signs binding $2 billion lithium supply deal with Trafigura

Mining Engineering’s Weekly Rundown: Episode 73

Get your SME Foundation VIP Access Conference Pass to MINEXCHANGE 2027

EnergyX signs binding $2 billion lithium supply deal with Trafigura

Mining Engineering’s Weekly Rundown: Episode 73

Get your SME Foundation VIP Access Conference Pass to MINEXCHANGE 2027

To view the past issues of the UCA magazine you must be a member of SME/UCA.

Consultants provide expertise, skills and experience to the industry

Trona mine to consider addition of a micronuclear power plant

Related Articles

Get your SME Foundation VIP Access Conference Pass to MINEXCHANGE 2027

“Check the box” and help support the SME Foundation

The Art and Science of the Flowsheet

UCA Division officer nominations

EnergyX signs binding $2 billion lithium supply deal with Trafigura

Mining Engineering’s Weekly Rundown: Episode 73

Get your SME Foundation VIP Access Conference Pass to MINEXCHANGE 2027

EnergyX signs binding $2 billion lithium supply deal with Trafigura

Mining Engineering’s Weekly Rundown: Episode 73

Get your SME Foundation VIP Access Conference Pass to MINEXCHANGE 2027

Use Single Sign On