Next Article in Journal
Recent Advances in Antibacterial Metallic Species Supported on Montmorillonite Clay Mineral: A Review
Next Article in Special Issue
Mapping of the Structural Lineaments and Sedimentary Basement Relief Using Gravity Data to Guide Mineral Exploration in the Denizli Basin
Previous Article in Journal
Study on the Controlling Factors and Laws of Deep Carbonate Geothermal Reservoirs in the Jizhong Depression
Previous Article in Special Issue
Particle Swarm Optimization (PSO) of High-Quality Magnetic Data of the Obudu Basement Complex, Nigeria
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Minerals
Volume 13
Issue 10
10.3390/min13101267
minerals-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Mineral Prospectivity Mapping for Orogenic Gold Mineralization in the Rainy River Area, Wabigoon Subprovince

by
Pouran Behnia
*,
Jeff Harris
,
Ross Sherlock
,
Mostafa Naghizadeh
and
Rajesh Vayavur
Mineral Exploration Research Centre, Harquail School of Earth Sciences, Laurentian University, 935 Ramsey Lake Road, Sudbury, ON P3E 2C6, Canada
*
Author to whom correspondence should be addressed.
Minerals 2023, 13(10), 1267; https://doi.org/10.3390/min13101267
Submission received: 12 August 2023 / Revised: 22 September 2023 / Accepted: 25 September 2023 / Published: 28 September 2023
(This article belongs to the Special Issue Application of Geophysical Data Interpretation in Geological and Mineral Potential Mapping)
Browse Figures
Versions Notes

Abstract

:
Random Forest classification was applied to create mineral prospectivity maps (MPM) for orogenic gold in the Rainy River area of Ontario, Canada. Geological and geophysical data were used to create 36 predictive maps as RF algorithm input. Eighty-three (83) orogenic gold prospects/occurrences were used to train the classifier, and 33 occurrences were used to validate the model. The non-Au (negative) points were randomly selected with or without spatial restriction. The prospectivity mapping results show high performance for the training and test data in area-frequency curves. The F1 accuracy is high and moderate when assessed with the training and test data, respectively. The mean decrease accuracy was applied to calculate the variable importance. Density, proximity to lithological contacts, mafic to intermediate volcanics, analytic signal, and proximity to the Cameron-Pipestone deformation zone exhibit the highest variable importance in both models. The main difference between the models is in the uncertainty maps, in which the high-potential areas show lower uncertainty in the maps created with spatial restriction when selecting the negative points.

Graphical Abstract

1. Introduction

Mineral prospectivity mapping (MPM) has seen many advancements over the past 30 years, beginning with the original work by Bonham-Carter [1,2]. MPM involves the use of geoscience data in the form of evidence or predictor maps representing vectors for various types of mineralization. These data are input into machine learning algorithms to produce maps that identify areas with higher probabilities for various types of mineral deposits. These maps should not only accurately predict the known deposits in a given study area but also reveal areas that are prospective for discovering new deposits. The final stage of the MPM process is evaluating how predictive the map is with respect to the training set of known deposits. This can be accomplished using cross-validation, the efficiency of classification/prediction graphs [3], and receiver operator curves (ROC) [4,5].
Machine learning algorithms used to model the geoscience data comprise two basic types: data-driven and knowledge-driven [1,2]. Data-driven techniques require a set of training data (e.g., mineral deposits or prospects) for modeling. Meanwhile, knowledge-driven techniques require the geoscientist to weigh each predictor map with a subjective weight; the predictor maps are then summed to produce an MPM map. Many types of data-driven algorithms exist, including weights of evidence [1,2,6,7,8,9,10,11], logistic regression [6,10,12,13], evidential belief modeling [14,15], support vector machine (SVM) [16,17], neural networks [18,19,20,21,22,23,24,25], and random forest, among others. Knowledge-driven techniques comprise algorithms such as Boolean, index overlay, and fuzzy logic [1,2,26,27,28,29].
Recently, numerous studies have described the random forest algorithm [30,31] as a highly powerful approach for producing MPM maps [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46]. As such, this paper adopts the random forest data-driven machine learning algorithm to produce an MPM map for a portion of the Rainy River greenstone belt in Ontario, Canada, based on orogenic gold. The random nature of the random forest algorithm regarding bootstrapping and random permutation of predictor maps through each decision tree prevents overfitting of the model.
This research is being conducted under the Metal Earth Project at the Laurentian University in Sudbury, Ontario, Canada. The fundamental objective is to elucidate why some greenstone belts are fertile while others are essentially barren. This is accomplished by collecting new 3D geophysical, seismic, lithogeochemical, and geochronological data, as well as collecting and compiling existing geoscience data to produce mineral prospectivity maps for various mineral commodities.
Gold deposits in the Superior geologic province exhibit a wide range of mineralization as well as structural and geometric characteristics [47,48]. In the Abitibi geologic subprovince, a significant portion of the gold deposits are syn-volcanic and synmagmatic, formed during the volcanic construction of the Abitibi belt. However, over 60% of the gold endowment originates from quartz-carbonate veins formed during late-stage tectonic inversion of the extensional basins [49]. In the Wabigoon subprovince, many gold deposits and prospects are interpreted as orogenic gold, including the areas of mineralization within the Cameron Lake area hosted by the Cameron-Pipestone deformation zone [50,51]. This type of gold mineralization is the focus of the present study.

2. Study Area

Figure 1 is an overview of the Wabigoon tectonic province and Rainy River transect. Figure 2 presents the regional geology of the Rainy River transect and the study area in the north [52]. This area was selected based on the availability of geoscience data used in the modeling. Synvolcanic gold is present in the southern portion of the area. However, due to insufficient synvolcanic gold bodies in the area to apply a data-driven method, we applied the MPM—for the orogenic gold mineralization in the northern region.
The area comprises volcanic lithologies organized as anastomosing belts surrounding large granitoid batholiths [52,53]. These are divided into four main sequences [54]: (1) a lower mafic tholeiitic unit; (2) intermediate to felsic volcanic rocks comprising volcaniclastic rocks, intrusions, and calc-alkalic flows; (3) an upper unit comprising mafic tholeiitic flows; and (4) an upper sequence of sedimentary rocks comprising turbiditic and alluvial/fluvial sedimentary rocks. All lithologies were affected by regional greenschist-grade metamorphism and amphibolite-grade metamorphism related to the emplacement of the granitoid batholiths. There are also several major regional structures that -could act as potential conduits for ore-bearing fluids [55].
The volcanic units are bound in the north by the Sasbaskonk batholith and in the south by the Quetico Fault (Figure 2). The volcanic units around the transect are intruded by the Flemming-Kingsford and Jackfish Lake plutons [55]. The southern part of the Rainy River area is bounded by the Quetico subprovince, which comprises metasedimentary rocks of turbiditic origin and rare conglomerates. The east–west striking dextral and sub-vertical Quetico shear zone forms the boundary of the Quetico subprovince and the Rainy River greenstone belt [54,56,57].
The largest gold deposit in the area is the Rainy River deposit, operated by NewGold Inc. This deposit is hosted by a subaqueous felsic volcanic complex with gold strongly correlated with a sericite alteration assemblage [51,58,59]. Work by Pelletier et al. [58,59] suggests that mineralization is early, pre-deformation, and likely synvolcanic in origin. The mafic volcanic rocks host most of the orogenic gold mineralization [51,58,59], particularly in the northern portion of the map area (Figure 2), which tends to have undeveloped mineral occurrences. Launay et al. [55] provide additional details on the geology and mineralization of the study area.

3. Materials and Methods

3.1. Random Forest (RF)

A random forest is an ensemble of individual decision trees [30,31] that can be used for classification and regression. A decision tree is composed of several leaves (nodes) and branches (edges). In each node, a decision is made, and a split is performed. The splitting process is continued until the node is pure. To train the classifier, a bagging procedure called bootstrapping is employed [60]. In this method, a subset of the training data is randomly selected by each tree to train the classifier; the remaining data, also called out-of-bag (OOB) data, are used for the validation [60]. Each tree uses a random subset of all variables (predictor maps in this study) that helps reduce the correlation between the trees. Combining the trees provides the majority vote. The final outputs are a classification map (in the case of MPM a two-class map) and a probability map for each class. The probability (prospectivity) is calculated by the number of trees that predicted gold divided by the total number of trees at each data point (pixel); this yields a number between 0 and 1. For example, if there are 10 trees at location A and no trees predicted the presence of gold, the probability would be 0; if 10 trees predicted gold, the probability would be 1. Additional details on the RF algorithm and how it can be applied to MPM have been previously described [30,31,35,37,46].
We used EnMap-BOX 1.4 [60] to apply the RF classification. The number of features was set to the square root of all features (i.e., 6), and the impurity function was set to the Gini coefficient. We tested 300–1000 trees to assess the learning curves for each forest. In most cases, the OOB was stabilized after 500 trees with an average accuracy of 72%–78%.

3.2. Predictor Maps

Based on the characteristics of orogenic gold mineralization in the area, the following lithological, structural, and alteration criteria were considered as vectors to mineralization. We also included the geophysical data to evaluate their potential predictive power for the mineralization of interest.

3.2.1. Lithology

The main host rocks for mineralization in the area are mafic to intermediate and felsic to intermediate volcanics (Figure 2). These units were extracted from the geology map [55] and used as the lithologic predictor maps. As some of the gold occurrences were associated with diorite-monzogranite granitoids, these rocks were also included as another predictor map. Figure 3a–c shows the binary maps of the three lithologies used to create the MPM. In addition, contacts between these rock units were selected, and proximity zones up to 500 m were created around them with an interval of 100 m (Figure 3d).

3.2.2. Structure

A structural compilation was obtained from the Ontario Geological Survey (OGS) digital database. Several regional-scale structures exist in the Rainy River area [55]. Features such as deformation zones (fault and shear zones) and fold hinges can act as potential conduits and traps for the migration of mineralized fluids and the deposition of orogenic gold, respectively. Hence, these features are considered important vectors to mineralization. To create the structural predictive maps, proximity zones up to 4 km with an interval of 200 m were created around the folds based on the field geologist’s expert knowledge (Figure 3e). The same proximity zones of up to 3 km were created around the deformation zones. Since the Cameron-Pipestone deformation zone appears to be more important than the other zones (many gold prospects occur close to this deformation zone), we applied it as an individual predictor map and extended the proximity zones up to 4 km (Figure 3f,g).

3.2.3. Alteration

The geochemical data for gold and its pathfinder elements were sparse and, thus, excluded from the integration model. However, as more analyses were available on the major elements, these data were extracted from the Metal Earth database and applied to calculate the chlorite-carbonate-pyrite alteration index (CCPI = 100 (MgO + FeO)/(MgO + FeO + Na2O + K2O)) using the ioGAS software, version 8.1. A centered log-ratio (clr) transformation [61] was applied to the data to minimize the closure effects.
As the samples were not sufficiently dense to warrant interpolation, we employed a zone of influence approach. Anomalous samples were selected using Q-Q plots, and proximity zones up to 1 km with a 250 m interval were created around the samples (Figure 3h).

3.2.4. Airborne Magnetic Data

The airborne magnetic data used for this study was extracted from the Ontario Geological Survey (OGS) Master Grid compilation [62]. The downloaded magnetic grid—originates from a compilation of many surveys that were leveled to a common datum of 300 m with a cell size of 200 m. In our MPM study, we used the total field reduced to the pole and the analytical signal images (Figure 4a,b).

3.2.5. Susceptibility and Density Data

The analytical processing of geophysical data was carried out on 3D density and susceptibility models. These models were inverted from the GSC gravity Bouguer anomaly and the total magnetic anomaly [63] on a 1 × 1 km horizontal grid, respectively. To prevent spatial aliasing artifacts, the total magnetic anomaly and digital elevation maps of the study area were smoothed by a Gaussian window before sampling on a 1 × 1 km horizontal grid. The vertical grids of the 3D density and susceptibility models were exponentially increased toward the deeper parts of the model. Both gravity and magnetic inversions were carried out by imposing model smoothness constraints. The inverted density and susceptibility 3D models were re-sampled/interpolated into a regular 500 × 500 × 500 m grid to generate co-localized density and susceptibility volumes suitable for further data analysis. Figure 4c,e,g depicts the density slices at elevations of −250, −6250, and −12,250 m, respectively. Figure 4d,f,h depicts the susceptibility slices at elevations of −250, −6250, and −12,250 m, respectively.
All predictor maps (totaling 36) were resampled to 50 m to ensure the retention of details from the lithology maps when converting to the raster format. Table 1 provides a summary of the predictor maps used in the RF modeling procedure.

3.2.6. Training (Au) Data

Mineral occurrence data were obtained from the Ontario Mineral Inventory database (OMI). Our target variable was orogenic gold bodies, comprising 116 orogenic gold prospects and occurrences available in the study area. We selected 83 gold prospects and occurrences as the training data and retained 33 gold occurrences as the test data to validate the MPM. Although random forest applies internal cross-validation, it is useful to validate the model with test data that were not included in the model.
In addition to the location of gold deposits, random forest requires additional negative or non-gold locations. However, selecting these points is relatively challenging. Rahimi et al. [64] used a clustering algorithm and selected random points from unfavorable clusters. Behnia [13] created random points from the low favorability zones of an MPM created using the weights of evidence method. Meanwhile, Carranza and Laborte [34] and Sun et al. [65] selected the negative points following a point pattern analysis. In this study, we selected the negative points using two strategies. In set-1, 83 random points were created some distance away from the known orogenic gold bodies, following a point pattern analysis. Figure 5 presents the results of the point-pattern analysis for orogenic gold prospects/occurrences in the area. The distance between each deposit and its nearest neighbor and the corresponding probability of finding a gold deposit within that distance were calculated and plotted. There was a 100% probability that another deposit occurred within 7 km of a deposit (Figure 5). However, using this distance to create random points away from the existing deposits causes the negative points to localize in a small area. Thus, instead, we selected a 2 km distance, corresponding to an 84% probability of finding a neighboring deposit next to any given deposit, to create 83 random points. For set-2, 83 random points were created with no restriction. This was repeated five times for each set for a total of ten training data sets, in which the gold prospects/occurrences were the same while the non-gold points differed. Hence, each time the model was trained, it received 83 gold and 83 non-gold points. Figure 6 shows how the two datasets were created. Locations of the known gold prospects/occurrences (train and test) are also shown.

4. Results

4.1. Probability Maps

A 10-fold repetition of the model, each using a different suite of randomly selected non-gold points, resulted in ten probability maps. To better compare the two models, the mean probability value for each set was generated, creating two MPMs (Figure 7a,b). Although the two maps look similar, they also have certain differences. For example, some areas, such as the northern regions and area (A) in Figure 7a, exhibit a higher probability in set-1, while set-2 has a higher probability in the east–south area (B) in Figure 7b. Moreover, uncertainty due to the use of different sets of training data was mapped based on the standard deviation of each of the two sets comprising five MPMs (Figure 8a,b). The uncertainty range is similar for the two sets; however, it is slightly lower in set-1 (0–0.274 vs. 0–0.287). As shown in Figure 8, the uncertainty is low in areas of gold mineralization and in areas with low probability/prospectivity in the eastern and western regions. Set-1 has lower uncertainty in high-probability areas where several gold prospects/occurrences exist, particularly in the north–central (black circle, Figure 8) and north–eastern regions. Areas of high uncertainty primarily correspond with areas lacking gold mineralization.
To assess the correlation between the MPMs, Pearson correlation coefficients were calculated. The correlations between the maps in set-1 and set-2 range from 0.881 to 0.955 and 0.867 to 0.923, respectively, showing a slightly higher correlation for set-1. The correlations between all ten maps range from 0.835 to 0.955.
The variable importance can be measured using different methods, including the mean decrease impurity and the mean decrease accuracy. Impurity-based feature importance measures can be biased in favor of variables with high cardinality (i.e., variables with many unique values). Hence, we used mean decrease accuracy or permutation feature importance, which do not have such a bias [66]. In this method, accuracy is measured on the OOB samples for each tree. The OOB values for a single variable are then randomly permuted, and the accuracy is recalculated. The average of the differences in the original and permuted accuracies for a variable represents the raw importance of that variable [60,66]. The normalized variable importance is the raw importance value divided by the standard deviation of the variable.
The variable importance in ten experiments (set-1 and 2 random points) was relatively different. To better compare the results, we applied the average results from each set (Figure 9). The density (depth: 5250–12,250 m) exhibits the highest predictive power in both models. Proximity to lithological contacts and mafic to intermediate volcanics show high predictive power, particularly in set-2, whereas felsic to intermediate volcanics and diorite-monzogranite granitoids exhibit low importance. Proximity to the Cameron-Pipestone zone and to other deformation zones has medium to high importance. Moreover, the analytical signal has high (set-1) to medium (set-2) predictive power, whereas the total magnetic field shows lower importance. Susceptibility has medium to high predictive power, while proximity to folds and alteration zones have the lowest prediction power. This was expected as the alteration map lacked sufficient data throughout the study area; as such, the values were missing for most of the area. The main difference between the two sets is that in set-1, the analytical signal shows higher importance than the lithological contacts and mafic to intermediate volcanics, while in set-2, the opposite occurs. Furthermore, susceptibility shows higher variability in set-1, ranging from high to low, while in set-2, excluding an elevation of −250 m, medium to low predictive power is observed.

4.2. Validation

Area–frequency curves were used to assess the MPM performance. In these graphs, the cumulative frequency percent of the training/test data is plotted against the cumulative area percent of the probability map from high to low values. Figure 10 shows the efficiency of the classification and prediction curves for the RF mineral prospectivity maps. More specifically, Figure 10a presents plots for mean set-1 and mean set-2, whereas 10b shows a plot for the mean of all ten prospectivity maps. Figure 10b shows that the area under the curve (AUC) is high for classification defined by the original training points used for classification (0.958) and prediction (0.904) defined by an independent test set not used for classification indicating that the models do not suffer from overfitting. Approximately 80% of the gold occurrences are predicted within 6.4% of the most prospective areas on the probability map; this falls to 15% when examining the prediction rate.
The classification accuracy was assessed using cross-validation. The average F1 accuracy is high, ranging from 0.952 to 0.982 when assessed by the training data and 0.814 to 0.893 when assessed by the test data. Although the accuracy changes slightly in each experiment, significant differences do not exist between the two training sets.

5. Discussion

The probability maps created using different non-deposit sites (set-1 and 2) are similar in appearance, with correlation coefficients ranging from 0.835 to 0.955. Based on the efficiency of classification curves (Figure 10), the probability maps created using the two sets of training data (5 maps for each set) have similar performances. Moreover, based on cross-validation, considerable differences do not exist in the F1 accuracy of the ten classification maps. The F1 accuracy for all the maps is very high when assessed by the training data used in the model (95.2%–98.2%) and moderate when assessed by the test data (81.4%–89.33%).
The variable importance for the two sets of training data is similar, with certain differences. Density (depth: 5250–12,250 m) has the highest predictive power in both models, followed by the analytic signal, susceptibility, proximity to lithological contacts, and Cameron-Pipestone deformation zone in set-1 and proximity to lithological contacts and mafic to intermediate volcanics in set-2. The high correlation with density is likely related to the presence of mafic-ultramafic intrusions in the Kakagi Lake area that localize the deformation corridors and, thus, the mineralization.
One main difference between the two sets is in the uncertainty maps. That is, the uncertainty in set-1 compared to set-2 is lower in areas of high probability where several orogenic mineralizations occur. This is likely due to restricting the creation of random points to 2 km from the known gold mineralization in set-1.
Figure 11a,b exhibits the most prospective areas (top 5%) based on the MPM produced from the average of set-1 and set-2 data points, respectively, overlaid on the lithology map. The prospective areas (A to H) are common in both models, except that areas G, H, and small patches in the south–east do not show up on the map created using set-1 training data. These areas have fewer or no gold prospects/occurrences in the vicinity, suggesting that selecting non-gold points a certain distance away from the mineralized bodies results in higher probabilities in the areas surrounding these bodies compared to other areas. Table 2 presents a geologic description of each prospective zone (A–H). Mafic to intermediate volcanics are present in all areas except C, in which mafic and ultramafic intrusions and felsic to intermediate volcanics predominate. All areas are associated with deformation zones. Although density does not exhibit a clear pattern, it is approximately medium to high and corresponds with higher values with increased depth. Moreover, the uncertainty is very low to low in all areas, and all but area H and small patches in the south–east have associated gold prospects/occurrences.

6. Conclusions

Random forest classification was applied to orogenic gold mineralization using 36 predictor maps and two sets of training data. The AUC of the area-frequency curves for classification and prediction is high, indicating that the model is not overfitted. The two models, despite visual similarity and very close accuracies, have differences regarding uncertainty, feature importance, and top prospective areas, indicating that the way in which the negative points are selected is important and affects the results. Further studies are required to determine how to select these points. Selecting mineralized bodies other than the mineralization of interest with different geological, geochemical, and geophysical characteristics may be a better option for selecting the negative points.
Based on the variable importance values, the presence of mafic to intermediate volcanics and lithological contacts are important vectors to mineralization. Moreover, proximity to deformation zones, particularly the Cameron-Pipestone, shows high predictive power.
Most of the orogenic gold bodies are within the top 5% of high prospectivity areas (A–F in Figure 11). However, some high-probability areas have fewer (G) or no gold mineralization (H) and can, thus, serve as targets for further exploration.

Author Contributions

Conceptualization, P.B. and J.H.; methodology, P.B.; formal analysis, P.B., M.N. and R.V.; writing—original draft preparation, J.H., P.B., R.S. and M.N.; writing—review and editing, P.B., R.S. and J.H.; visualization, P.B.; supervision, R.S. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Canada First Research Excellence Fund (CFREF-2015-00005) and is the Metal Earth publication MERC-ME-2023-28.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Acknowledgments

We thank Gaëtan Launay for providing valuable advice on conceptual mineral deposit models. We also thank four anonymous reviewers for their constructive comments.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Bonham-Carter, G.F. Geographic Information Systems for Geoscientists: Modeling with GIS. Pergamon; Elsevier Science Inc.: New York, NY, USA, 1994. [Google Scholar]
  2. Bonham-Carter, G.F.; Agterberg, F.P.; Wright, D.F. Integration of geological datasets for gold exploration in Nova Scotia. PE&RS 1988, 54, 1585–1592. [Google Scholar]
  3. Chung, C.-J.F.; Fabbri, A.G. Validation of Spatial Prediction Models for Landslide Hazard Mapping. Nat. Hazards 2003, 30, 451–472. [Google Scholar] [CrossRef]
  4. Chen, Y.; Wu, W. A prospecting cost-benefit strategy for mineral potential mapping based on ROC curve analysis. Ore Geol. Rev. 2016, 74, 26–38. [Google Scholar] [CrossRef]
  5. Nykänen, V.; Lahti, I.; Niiranen, T.; Korhonen, K. Receiver operating characteristics (ROC) as validation tool for prospectivity models—A magmatic Ni–Cu case study from the Central Lapland Greenstone Belt, Northern Finland. Ore Geol. Rev. 2015, 71, 853–860. [Google Scholar] [CrossRef]
  6. Raines, G.L.; Mihalasky, M.J. A Reconnaissance Method for Delineation of Tracts for Regional-Scale Mineral-Resource Assessment Based on Geologic-Map Data. Nat. Resour. Res. 2002, 11, 241–248. [Google Scholar] [CrossRef]
  7. Carranza, E.J.M. Weights of Evidence Modeling of Mineral Potential: A Case Study Using Small Number of Prospects, Abra, Philippines. Nat. Resour. Res. 2004, 13, 173–187. [Google Scholar] [CrossRef]
  8. Harris, D.; Zurcher, L.; Stanley, M.; Marlow, J.; Pan, G. A Comparative Analysis of Favorability Mappings by Weights of Evidence, Probabilistic Neural Networks, Discriminant Analysis, and Logistic Regression. Nat. Resour. Res. 2003, 12, 241–255. [Google Scholar] [CrossRef]
  9. Harris, J.R.; Sanborn-Barrie, M.; Panagapko, D.A.; Skulski, T.; Parker, J.R. Gold prospectivity maps of the Red Lake greenstone belt: Application of GIS technology. Can. J. Earth Sci. 2006, 43, 865–893. [Google Scholar] [CrossRef]
  10. Harris, J.R.; Wilkinson, L.; Heather, K.; Fumerton, S.; Bernier, M.A.; Ayer, J.; Dahn, R. Application of GIS Processing Techniques for Producing Mineral Prospectivity Maps—A Case Study: Mesothermal Au in the Swayze Greenstone Belt, Ontario, Canada. Nat. Resour. Res. 2001, 10, 91–124. [Google Scholar] [CrossRef]
  11. Zuo, R.; Zhang, Z.; Zhang, D.; Carranza, E.J.M.; Wang, H. Evaluation of uncertainty in mineral prospectivity mapping due to missing evidence: A case study with skarn-type Fe deposits in Southwestern Fujian Province, China. Ore Geol. Rev. 2015, 71, 502–515. [Google Scholar] [CrossRef]
  12. Chung, C.F.; Agterberg, F.P. Regression models for estimating mineral resources from geological map data. Math. Geosci. 1980, 12, 473–488. [Google Scholar] [CrossRef]
  13. Carranza, E.J.M.; Hale, M. Logistic regression for geologically-constrained mapping of gold mineralization potential, Baguio district, Philippines. Explor. Min. Geol. 2001, 10, 165–175. [Google Scholar] [CrossRef]
  14. Carranza, E.J.M.; Woldai, T.; Chikambwe, E.M. Application of Data-Driven Evidential Belief Functions to Prospectivity Mapping for Aquamarine-Bearing Pegmatites, Lundazi District, Zambia. Nat. Resour. Res. 2005, 14, 47–63. [Google Scholar] [CrossRef]
  15. Carranza, E.J.M. Data-Driven Evidential Belief Modeling of Mineral Potential Using Few Prospects and Evidence with Missing Values. Nat. Resour. Res. 2015, 24, 291–304. [Google Scholar] [CrossRef]
  16. Zuo, R.; Carranza, E.J.M. Support vector machine: A tool for mapping mineral prospectivity. Comput. Geosci. 2011, 37, 1967–1975. [Google Scholar] [CrossRef]
  17. Abedi, M.; Norouzi, G.-H.; Bahroudi, A. Support vector machine for multi-classification of mineral prospectivity areas. Comput. Geosci. 2012, 46, 272–283. [Google Scholar] [CrossRef]
  18. Singer, D.A.; Kouda, R. Application of a feedforward neural network in the search for Kuroko deposits in the Hokuroku district, Japan. Math. Geosci. 1996, 28, 1017–1023. [Google Scholar] [CrossRef]
  19. Harris, D.; Pan, G. Mineral Favorability Mapping: A Comparison of Artificial Neural Networks, Logistic Regression, and Discriminant Analysis. Nat. Resour. Res. 1999, 8, 93–109. [Google Scholar] [CrossRef]
  20. Brown, W.M.; Gedeon, T.D.; Groves, D.I.; Barnes, R.G. Artificial neural networks: A new method for mineral prospectivity mapping. Aust. J. Earth Sci. 2000, 47, 757–770. [Google Scholar] [CrossRef]
  21. Porwal, A.; Carranza, E.J.M.; Hale, M. Artificial neural networks for mineral potential mapping. Nat. Resour. Res. 2003, 12, 155–171. [Google Scholar] [CrossRef]
  22. Porwal, A.; Carranza, E.J.M.; Hale, M. A Hybrid Neuro-Fuzzy Model for Mineral Potential Mapping. Math. Geosci. 2004, 36, 803–826. [Google Scholar] [CrossRef]
  23. Behnia, P. Application of Radial Basis Functional Link Networks to Exploration for Proterozoic Mineral Deposits in Central Iran. Nat. Resour. Res. 2007, 16, 147–155. [Google Scholar] [CrossRef]
  24. Nykänen, V. Radial Basis Functional Link Nets Used as a Prospectivity Mapping Tool for Orogenic Gold Deposits Within the Central Lapland Greenstone Belt, Northern Fennoscandian Shield. Nat. Resour. Res. 2008, 17, 29–48. [Google Scholar] [CrossRef]
  25. Parsa, M.; Carranza, E.J.M.; Ahmadi, B. Deep GMDH Neural Networks for Predictive Mapping of Mineral Prospectivity in Terrains Hosting Few but Large Mineral Deposits. Nat. Resour. Res. 2022, 31, 37–50. [Google Scholar] [CrossRef]
  26. An, P.; Moon, W.; Rencz, A.N. Application of fuzzy theory for integration of geological, geophysical, and remotely sensed data. CJEG 1991, 27, 1–11. [Google Scholar]
  27. Lusty, P.A.J.; Scheib, C.; Gunn, A.G.; Walker, A.S.D. Reconnaissance-scale prospectivity analysis for gold mineralization for gold mineralization in the southern Uplands-Dwon-Longford Terrane, Northern Island. Nat. Resour. Res. 2012, 21, 359–382. [Google Scholar] [CrossRef]
  28. Knox-Robinson, C.M. Vectorial fuzzy logic: A novel technique for enhanced mineral prospectivity mapping, with reference to the orogenic gold mineralization potential of the Kalgoorlie Terrane, Western Australia. Aust. J. Earth Sci. 2000, 47, 929–941. [Google Scholar] [CrossRef]
  29. Harris, J.; Grunsky, E.; Behnia, P.; Corrigan, D. Data- and knowledge-driven mineral prospectivity maps for Canada’s North. Ore Geol. Rev. 2015, 71, 788–803. [Google Scholar] [CrossRef]
  30. Breiman, L. Bagging Predictors. Mach. Learn. 1996, 24, 123–140. [Google Scholar] [CrossRef]
  31. Breiman, L. Random Forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
  32. Lachaud, A.; Marcus, A.; Vučetić, S.; Mišković, I. Study of the Influence of Non-Deposit Locations in Data-Driven Mineral Prospectivity Mapping: A Case Study on the Iskut Project in Northwestern British Columbia, Canada. Minerals 2021, 11, 597. [Google Scholar] [CrossRef]
  33. Cracknell, M.J.; Reading, A.M. Geological mapping using remote sensing data: A comparison of five machine learning algorithms, their response to variations in the spatial distribution of training data and the use of explicit spatial information. Comput. Geosci. 2014, 63, 22–33. [Google Scholar] [CrossRef]
  34. Rodriguez-Galiano, V.F.; Chica-Olmo, M.; Chica-Rivas, M. Predictive modelling of gold potential with the integration of multisource information based on random forest: A case study on the Rodalquilar area, Southern Spain. Int. J. Geogr. Inf. Sci. 2014, 28, 1336–1354. [Google Scholar] [CrossRef]
  35. Carranza, E.J.M.; Laborte, A.G. Data-driven predictive mapping of gold prospectivity, Baguio district, Philippines: Application of Random Forests algorithm. Ore Geol. Rev. 2015, 71, 777–787. [Google Scholar] [CrossRef]
  36. Harris, J.R.; Naghizadeh, M.; Behnia, P.; Mathieu, L. Data-driven gold potential maps for the Chibougamau area, Abitibi greenstone belt, Canada. Ore Geol. Rev. 2022, 150, 105176. [Google Scholar] [CrossRef]
  37. Carranza, E.J.M.; Laborte, A.G. Data-Driven Predictive Modeling of Mineral Prospectivity Using Random Forests: A Case Study in Catanduanes Island (Philippines). Nat. Resour. Res. 2016, 25, 35–50. [Google Scholar] [CrossRef]
  38. Behnia, P.; Harris, J.; Liu, H.; Jørgensen, T.R.C.; Naghizadeh, M.; Roots, E.A. Random forest classification for volcanogenic massive sulfide mineralization in the Rouyn-Noranda Area, Quebec. Ore Geol. Rev. 2023, 161, 105612. [Google Scholar] [CrossRef]
  39. McKay, G.; Harris, J.R. Comparison of the Data-Driven Random Forests Model and a Knowledge-Driven Method for Mineral Prospectivity Mapping: A Case Study for Gold Deposits Around the Huritz Group and Nueltin Suite, Nunavut, Canada. Nat. Resour. Res. 2015, 25, 125–143. [Google Scholar] [CrossRef]
  40. Harris, J.; Grunsky, E.C. Predictive lithological mapping of Canada’s North using Random Forest classification applied to geophysical and geochemical data. Comput. Geosci. 2015, 80, 9–25. [Google Scholar] [CrossRef]
  41. Zhang, S.; Xiao, K.; Carranza, E.J.M.; Yang, F. Maximum Entropy and Random Forest Modeling of Mineral Potential: Analysis of Gold Prospectivity in the Hezuo–Meiwu District, West Qinling Orogen, China. Nat. Resour. Res. 2019, 28, 645–664. [Google Scholar] [CrossRef]
  42. Zhang, S.; Carranza, E.J.M.; Xiao, K.; Wei, H.; Yang, F.; Chen, Z.; Li, N.; Xiang, J. Mineral Prospectivity Mapping based on Isolation Forest and Random Forest: Implication for the Existence of Spatial Signature of Mineralization in Outliers. Nat. Resour. Res. 2021, 31, 1981–1999. [Google Scholar] [CrossRef]
  43. Wang, J.; Zuo, R.; Xiong, Y. Mapping Mineral Prospectivity via Semi-supervised Random Forest. Nat. Resour. Res. 2020, 29, 189–202. [Google Scholar] [CrossRef]
  44. Parsa, M.; Maghsoudi, A. Assessing the effects of mineral systems-derived exploration targeting criteria for random Forests-based predictive mapping of mineral prospectivity in Ahar-Arasbaran area, Iran. Ore Geol. Rev. 2021, 138, 104399. [Google Scholar] [CrossRef]
  45. Hong, S.; Zuo, R.; Huang, X.; Xiong, Y. Distinguishing IOCG and IOA deposits via random forest algorithm based on magnetite composition. J. Geochem. Explor. 2021, 230, 106859. [Google Scholar] [CrossRef]
  46. Ford, A. Practical Implementation of Random Forest-Based Mineral Potential Mapping for Porphyry Cu–Au Mineralization in the Eastern Lachlan Orogen, NSW, Australia. Nat. Resour. Res. 2020, 29, 267–283. [Google Scholar] [CrossRef]
  47. Colvine, A.C.; Fyon, J.A.; Heather, K.B.; Marmont, S.; Smith, P.M.; Troop, D.G. Archean Lode Gold Deposits in Ontario; Ontario Geological Survey: Sudbury, ON, Canada, 1988; Paper 139; 136p. [Google Scholar]
  48. Hodgson, C.J. Mesothermal lode-gold deposits. In Mineral Deposits Modeling; Kirkham, R.V., Sinclair, W.D., Thorpe, R.I., Duke, J.M., Eds.; Geological Association of Canada: St. John’s, NL, Canada, 1993; pp. 635–678. [Google Scholar]
  49. Dubé, B.; Mercier-Langevin, P. Gold Deposits of the Archean Abitibi Greenstone Belt, Canada. In Geology of the World’s Major Gold Deposits and Provinces; Sillitoe, R.H., Goldfarb, R.J., Robert, F., Simmons, S.F., Eds.; Society of Economic Geologists, Inc.: Littleton, CO, USA, 2020; pp. 669–708. [Google Scholar] [CrossRef]
  50. Melling, D.R. The occurrence of gold in the Cameron–Rowan Lake area: Preliminary report on the Monte Cristo and Victor Island gold prospects in the Monte Cristo Shear Zone. In Summary of Field Work and Other Activities; Ontario Geological Survey: Sudbury, ON, Canada, 1986; pp. 252–260. [Google Scholar]
  51. Pelletier, M. The Rainy River gold deposit, Wabigoon Subprovince, Western Ontario: Style, Geometry, Timing and Structural Controls on ore Distribution and Grades. Master’s Thesis, Université du Québec, Institut national de la recherche scientifique, Quebec City, QC, Canada, 2016; 404p. [Google Scholar]
  52. Launay, G.; McRae, M.L.; Sherlock, R.L. Stratigraphy, Metallogeny and Crustal Architecture of the Rainy River Greenstone Belt. Laurentian University. 2021. Available online: https://merc.laurentian.ca/sites/default/files/mepresentationrainyriver2021_seg_v2.pdf (accessed on 10 June 2023).
  53. Percival, J.A.; Skulski, T.; Sanborn-Barrie, M.; Stott, G.M.; Leclair, A.D.; Corkery, M.T.; Boily, M. Chapter 6 Geology and tectonic evolution of the Superior province, Canada. In Tectonic Styles in Canada: The Lithoprobe Perspective; Percival, J.A., Cook, F.A., Clowes, R.M., Eds.; Geological Association of Canada: St. John’s, NL, Canada, 2012; Special Paper 49; p. 321. [Google Scholar]
  54. Blackburn, C.E.; Johns, G.W.; Ayer, J.A.; Davis, D.W. Wabigoon Subprovince. In Geology of Ontario; Ontario Geological Survey: Sudbury, ON, Canada, 1991; Volume 4, pp. 303–382. [Google Scholar]
  55. Launay, G.; McRae, M.L.; Sherlock, R.L. Preliminary results from stratigraphic and structural study of the Rainy River green-stone belt and the Quetico Subprovince. In Summary of Field Work and Other Activities; Ontario Geological Survey: Sudbury, ON, Canada, 2019; pp. 44-1–44-10, Open File Report 6360. [Google Scholar]
  56. Davis, D.W.; Poulsen, K.H.; Kamo, S.L. New Insights into Archean Crustal Development from Geochronology in the Rainy Lake Area, Superior Province, Canada. J. Geol. 1989, 97, 379–398. [Google Scholar] [CrossRef]
  57. Fernández, C.; Czeck, D.M.; Díaz-Azpiroz, M. Testing the model of oblique transpression with oblique extrusion in two natural cases: Steps and consequences. J. Struct. Geol. 2013, 54, 85–102. [Google Scholar] [CrossRef]
  58. Pelletier, M.; Mercier-Langevin, P.; Crick, D.; Tolman, J.; Beakhouse, G.P.; Dubé, B. Targeted Geoscience Initiative 4. Lode gold deposits in ancient and deformed terranes: Preliminary observations on the nature and distribution of the deformed and metamorphosed hydrothermal alteration associated with the Archean Rainy River gold deposit, northwestern Ontario. In Summary of Field Work and Other Activities; Ontario Geological Survey: Sudbury, ON, Canada, 2014; pp. 41-1–41-10, Open File Report 6300. [Google Scholar]
  59. Pelletier, M.; Mercier-Langevin, P.; Dubé, B.; Crick, D.; Tolman, J.; McNicoll, V.J.; Jackson, S.E.; Beakhouse, G.P. The Rainy River—Atypical Archean Au deposit, western Wabigoon Subprovince, Ontario. In Targeted Geoscience Initiative 4: Contributions to the Understanding of Precambrian Lode Gold Deposits and Implications for Exploration; Geological Survey of Canada: Ottawa, ON, Canada, 2014; pp. 193–207, Open File 7852. [Google Scholar]
  60. Jakimow, B.; Rabe, A.; van der Linden, S.; Wirth, F.; Hostert, P. EnMAP-Box Manual, Version 1.4, Humboldt-Universität zu Berlin, 2012, Germany. Available online: https://bitbucket.org/hu-geomatics/enmap-box-idl/wiki/imageRF%20-%20Manual%20for%20Application (accessed on 10 November 2021).
  61. Aitchison, J. The Statistical Analysis of Compositional Data; Chapman and Hall: London, UK, 1986; 416p. [Google Scholar]
  62. Ontario Geological Survey. Single Master Gravity and Aeromagnetic Data for Ontario; Ontario Geological Survey: Sudbury, ON, Canada, 1999. [Google Scholar]
  63. Geological Survey of Canada. The Canadian Gravity and Magnetic Anomaly Database: GSC. Available online: http://gdr.agg.nrcan.gc.ca/gdrdap/dap/search-eng.php (accessed on 12 October 2022).
  64. Rahimi, H.; Abedi, M.; Yousefi, M.; Bahroudi, A.; Elyasi, G.-R. Supervised mineral exploration targeting and the challenges with the selection of deposit and non-deposit sites thereof. Appl. Geochem. 2021, 128, 104940. [Google Scholar] [CrossRef]
  65. Sun, T.; Chen, F.; Zhong, L.; Liu, W.; Wang, Y. GIS-based mineral prospectivity mapping using machine learning methods: A case study from Tongling ore district, eastern China. Ore Geol. Rev. 2019, 109, 26–49. [Google Scholar] [CrossRef]
  66. Pedregosa, F.; Varoquaux, G.; Gramfort, A.; Michel, V.; Thirion, B.; Grisel, O.; Blondel, M.; Prettenhofer, P.; Weiss, R.; Du-bourg, V.; et al. Scikit-learn: Machine Learning in Python. J. Mach. Learn. Res. 2011, 12, 2825–2830. Available online: https://scikit-learn.org/stable/modules/permutation_importance.html#permutation-importance (accessed on 14 March 2022).
Minerals 13 01267 g001
Figure 1. Simplified geology of the Wabigoon and Rainy River transects.
Figure 1. Simplified geology of the Wabigoon and Rainy River transects.
Minerals 13 01267 g001
Minerals 13 01267 g002
Figure 2. Geologic map of the Rainy River area [52]. Deformation zones: BLdz = Brooks Lake, CPdz = Cameron-Pipestone, DPdz = DogPaw, HPdz = Helena-Pipestone, Kdz = Kakagi, MCdz = Monte Cristo, PHdz = Phinney, Qdz = Quetico, SBdz = Sabaskong Bay, SRdz = Seine River. A blue rectangle shows the study area selected for MPM.
Figure 2. Geologic map of the Rainy River area [52]. Deformation zones: BLdz = Brooks Lake, CPdz = Cameron-Pipestone, DPdz = DogPaw, HPdz = Helena-Pipestone, Kdz = Kakagi, MCdz = Monte Cristo, PHdz = Phinney, Qdz = Quetico, SBdz = Sabaskong Bay, SRdz = Seine River. A blue rectangle shows the study area selected for MPM.
Minerals 13 01267 g002
Minerals 13 01267 g003
Figure 3. Lithological, structural, and alteration predictor maps: (a) Mafic to intermediate volcanics; (b) felsic to intermediate volcanics; (c) diorite-monzogranite granitoids; (d) proximity to lithological contacts; (e) proximity to folds; (f) proximity to the Cameron-Pipestone deformation zone; (g) proximity to deformation zones; (h) proximity to chlorite-carbonate-pyrite alteration. Green points represent geochemical sample locations.
Figure 3. Lithological, structural, and alteration predictor maps: (a) Mafic to intermediate volcanics; (b) felsic to intermediate volcanics; (c) diorite-monzogranite granitoids; (d) proximity to lithological contacts; (e) proximity to folds; (f) proximity to the Cameron-Pipestone deformation zone; (g) proximity to deformation zones; (h) proximity to chlorite-carbonate-pyrite alteration. Green points represent geochemical sample locations.
Minerals 13 01267 g003
Minerals 13 01267 g004
Figure 4. Airborne magnetic data and a subset of susceptibility and density maps created by the inversion of gravity and total magnetic anomalies. (a) Total magnetic intensity reduced to pole; (b) Analytical signal; (c,e,g) density slices at elevations of −250 m, −6250 m, and −12,250 m, respectively. (d,f,h) susceptibility at elevations of −250 m, −6250 m, and −12,250 m, respectively. Orogenic gold is shown as yellow points.
Figure 4. Airborne magnetic data and a subset of susceptibility and density maps created by the inversion of gravity and total magnetic anomalies. (a) Total magnetic intensity reduced to pole; (b) Analytical signal; (c,e,g) density slices at elevations of −250 m, −6250 m, and −12,250 m, respectively. (d,f,h) susceptibility at elevations of −250 m, −6250 m, and −12,250 m, respectively. Orogenic gold is shown as yellow points.
Minerals 13 01267 g004
Minerals 13 01267 g005
Figure 5. Point pattern analysis on orogenic gold bodies. A distance of 2 km corresponds to an 84% probability of finding a neighboring deposit to any given deposit.
Figure 5. Point pattern analysis on orogenic gold bodies. A distance of 2 km corresponds to an 84% probability of finding a neighboring deposit to any given deposit.
Minerals 13 01267 g005
Minerals 13 01267 g006
Figure 6. (a) Example of non-gold points randomly generated 2 km away from the known gold prospects/occurrences (set-1). (b) Non-gold points randomly generated with no restriction (set-2). Gold occurrences used as test data are also shown.
Figure 6. (a) Example of non-gold points randomly generated 2 km away from the known gold prospects/occurrences (set-1). (b) Non-gold points randomly generated with no restriction (set-2). Gold occurrences used as test data are also shown.
Minerals 13 01267 g006
Minerals 13 01267 g007
Figure 7. Mean probability maps of set-1 (a) and set-2 (b). Note the higher probabilities in the northern parts and area (A) in set-1 and area (B) in Set-2.
Figure 7. Mean probability maps of set-1 (a) and set-2 (b). Note the higher probabilities in the northern parts and area (A) in set-1 and area (B) in Set-2.
Minerals 13 01267 g007
Minerals 13 01267 g008
Figure 8. Uncertainty map based on the standard deviation for set-1 (a) and set-2 (b). Note the lower uncertainty in set-1 in areas shown in black circles.
Figure 8. Uncertainty map based on the standard deviation for set-1 (a) and set-2 (b). Note the lower uncertainty in set-1 in areas shown in black circles.
Minerals 13 01267 g008
Minerals 13 01267 g009
Figure 9. Mean variable importance for set-1 (a) and set-2 (b). Density has the highest importance in both models, followed by proximity to lithological contacts, susceptibility, mafic-intermediate volcanics, analytic signal, and Cameron-Pipestone deformation zone.
Figure 9. Mean variable importance for set-1 (a) and set-2 (b). Density has the highest importance in both models, followed by proximity to lithological contacts, susceptibility, mafic-intermediate volcanics, analytic signal, and Cameron-Pipestone deformation zone.
Minerals 13 01267 g009
Minerals 13 01267 g010
Figure 10. (a) Efficiency of classification and prediction curves for the mean prospectivity maps of set-1 and set-2. Set-2 performs slightly better for classification and prediction. (b) Efficiency of classification and prediction for the mean of ten prospectivity maps.
Figure 10. (a) Efficiency of classification and prediction curves for the mean prospectivity maps of set-1 and set-2. Set-2 performs slightly better for classification and prediction. (b) Efficiency of classification and prediction for the mean of ten prospectivity maps.
Minerals 13 01267 g010
Minerals 13 01267 g011
Figure 11. The most prospective areas show the top 5% of the MPM for (a) set-1 and (b) set-2. Most of the known orogenic gold prospects/occurrences lie in these areas.
Figure 11. The most prospective areas show the top 5% of the MPM for (a) set-1 and (b) set-2. Most of the known orogenic gold prospects/occurrences lie in these areas.
Minerals 13 01267 g011
Table 1. Summary of predictor maps.
Table 1. Summary of predictor maps.
DataVariableSource
Lithology (4 maps)Mafic-intermediate volcanics
Felsic-intermediate volcanics
Diorite-monzogranite granitoids
Proximity to lithological contacts		Launay, 2021, Metal Earth [52]
Structure (3 maps)Cameron-Pipestone DZ *
Deformation zones (other)
Folds		Launay, 2021, Metal Earth [52]
Alteration (1 map)Chlorite-carbonate-pyrite index (CCPI)Geochemical data, Metal Earth
Airborne Magnetic (2 maps)Total magnetic intensity (TMI) reduced to pole,
Analytic signal 		Ontario Geological Survey
3D geophysical data (26 maps)Density: −250 m to −12,250 m with a 1 km interval
Susceptibility: −250 m to −12,250 m with a 1 km interval 		Geological Survey of Canada
* Deformation zone.
Table 2. Geologic description of the prospective areas in Figure 11.
Table 2. Geologic description of the prospective areas in Figure 11.
AreaGeologyDeformation ZoneDensityUncertaintyGold
Mineralization
AMafic-intermediate volcanics, felsic-intermediate volcanics, mafic and ultramafic intrusions, diorite-monzogranite granitoidsCameron-Pipestone, DogPaw Medium near surface,
higher in deeper parts		Very lowYes
BMainly mafic-intermediate volcanics, and mafic intrusionsDogPaw Medium to high in medium depthsVery low to lowYes
CMainly mafic and ultramafic intrusions, as well as felsic-intermediate volcanicsKakagi Medium to highVery low to lowYes
DMainly mafic-intermediate volcanics, felsic-intermediate volcanics, mafic intrusions, and diorite-monzogranite granitoidsMonte Cristo HighVery low to lowYes
EMainly mafic intrusions, felsic-intermediate volcanics, mafic-intermediate volcanics, and felsic intrusionsHelena-Pipestone, Phinney Medium to high, higher in deeper partsVery low to lowYes
FFelsic-intermediate volcanics, diorite-monzogranite granitoids, and mafic-intermediate volcanicsCameron-Pipestone VariableVery low to lowYes
GMafic-intermediate volcanics, felsic-intermediate volcanics, and mafic intrusionsCameron-Pipestone Very high (in deeper parts) to mediumVery lowYes
HMafic-intermediate volcanicsPhinney MediumVery lowNo
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Behnia, P.; Harris, J.; Sherlock, R.; Naghizadeh, M.; Vayavur, R. Mineral Prospectivity Mapping for Orogenic Gold Mineralization in the Rainy River Area, Wabigoon Subprovince. Minerals 2023, 13, 1267. https://doi.org/10.3390/min13101267

AMA Style

Behnia P, Harris J, Sherlock R, Naghizadeh M, Vayavur R. Mineral Prospectivity Mapping for Orogenic Gold Mineralization in the Rainy River Area, Wabigoon Subprovince. Minerals. 2023; 13(10):1267. https://doi.org/10.3390/min13101267

Chicago/Turabian Style

Behnia, Pouran, Jeff Harris, Ross Sherlock, Mostafa Naghizadeh, and Rajesh Vayavur. 2023. "Mineral Prospectivity Mapping for Orogenic Gold Mineralization in the Rainy River Area, Wabigoon Subprovince" Minerals 13, no. 10: 1267. https://doi.org/10.3390/min13101267

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop