Support vector machine and artificial neural network modelling of orogenic gold prospectivity mapping in the Swayze greenstone belt, Ontario, Canada

Highlights

• Translating the orogenic mineral system of the Swayze greenstone belt into mappable expressions.

• Support vector machines and artificial neural network were used to predict gold prospectivity.

• Stratified k-fold cross validation resulted in a mean accuracy of > 87% after k =10 folds.

• The best predictor layers were proximity to D₂ and D₃ high strain zones, lithological contacts, and D₂ folds.

• New deposit targets lie proximal to faulting, EM conductivities, and along chemically reactive rocks.

Abstract

Exploration for new mineral deposits has become increasingly difficult as new discoveries are being found under progressively deeper cover. To better understand and predict orogenic gold mineralization in the Archean Swayze greenstone belt, the essential ingredients of a mineral system are considered: 1) the source of gold and transport fluid ligands, 2) fluid pathways, 3) traps, and 4) the processes responsible for gold precipitation. The aim of this study is to use a mineral systems approach to help generate exploration targeting models using a spatial statistical method - weights of evidence (WofE) and data-driven machine learning tools, namely radial basis function neural networks (RBFNN) and support vector machine (SVM). The mineral prospectivity maps generated using the RBFNN and SVM machine learning methods were trained using the K-Fold cross-validation approach whereby 10 subsets of the data were used to train and test the model performance. The mean area under receiver operator curve after 10-fold cross-validations were 91% and 94% for the RBFNN models, and the SVM models obtained accuracies of 91% and 87%. Feature importance estimations obtained from both methods indicate that D₂ and D₃ high-strain zones, lithological contacts and D₂ folds (i.e., synclines and anticlines) were found to be important predictor layers for targeting potential prospective zones of gold mineralization. The machine learning algorithms used in this study are novel and pragmatic methods that use the full potential of geoscience datasets in mapping orogenic gold prospectivity.

Introduction

Mineral prospectivity mapping (MPM) can be described as utilising spatial computational techniques to determine the probability of finding a mineral prospect in an area using selected geological layers that show proximity to mineralization (Carranza, 2009). The field of MPM is not new, with early work on a two-staged least-square approach described by Agterberg and Cabilio (1969). Since then researchers have developed various tools such as Boolean overlay, fuzzy logic, weights of evidence, and artificial neural networks designed to predict the location of new mineral deposits (Bonham-Carter et al., 1988, Bonham-Carter, 1994, Carranza, 2004). Recent software developments have enabled geoscientists to generate prospectivity maps using geographical information system (GIS) packages. These packages are convenient for spatial data integration and to weight the relationship of known deposits with input predictor layers, making this a powerful tool for mineral exploration (Bonham-Carter, 1994, Carranza, 2009). The methods used for MPM can be classified as either data driven or knowledge-based approach (Bonham-Carter, 1994, Porwal and Kreuzer, 2010). Data-driven techniques are empirical and rely on statistical relationships between input layers and known deposits, while knowledge-driven techniques are more heuristic and rely on expert opinion (Parsa et al., 2017a, Parsa et al., 2017b, Parsa et al., 2017c, Parsa et al., 2018a, Parsa et al., 2018b, Yousefi et al., 2019, Roshanravan et al., 2020). Traditional MPM is done at deposit to district scales (several square kms), where deposit controls are well-defined and understood. MPM is an important tool to assist in the discovery of mineral deposits, as mineral deposit discovery has dramatically decreased even with increasing exploration expenditure. One reason for this is due to exploration being focused on more brownfield areas where mines and deposits already exist (Schodde, 2004, Schodde, 2017, Davies et al., 2020, Groves et al., 2020).

MPM over the Swayze greenstone belt has been carried out by Harris (2002) with a primary focus on effective integration and representation of geochemical datasets during prospectivity analysis. The present study seeks to build on Harris (2002) by considering the mineral systems approach and new geophysical datasets, and assess prospectivity of gold mineralization within the Swayze greenstone belt. This study tests two data-driven machine learning methods and assesses the gold mineralization prospectivity within the Swayze greenstone belt. The Swayze greenstone belt is regarded as the southwestern extension of the well-mineralized Abitibi greenstone belt. However, previous studies suggested that the Swayze greenstone belt is not as well mineralized as the world-class gold deposits found in the Abitibi greenstone belt (Heather, 2001, Hastie, 2017, Breemen et al., 2006). The aims of this study are: 1) to outline the critical processes responsible for orogenic gold deposition in the Swayze greenstone belt, 2) to determine the targeting and geological features representing key geological processes, 3) to determine the mappable targeting criteria by reviewing the available geoscience data, 4) to use data-driven tools to map mineral prospectivity in the Swayze greenstone belt. The two tools used to predict mineral prospectivity are support vector machine (SVM), and the radial basis function neural networks (RBFNN). The weights of evidence Bayesian statistical method was used to aid in objective estimation of spatial correlations between gold deposits and predictor layers.

The Archean Swayze greenstone belt is found within the Superior province (Fig. 1), southwest of the Abitibi greenstone belt, and is connected to the Abitibi by a narrow septum of metavolcanic and metasedimentary rocks associated with the mineral prolific Porcupine-Destor Fault zone in the north and the Cardillac-Larder lake deformation zone in the southern part of Abitibi (Ayer, 1995, Breemen, 2006). The Porcupine-Destor fault zone hosts most of the historic gold mineralization in the Timmins Area – about 11,819 tons of gold can be traced for 300 km, and gold is associated with strong schistosity and chlorite-sericite alterations (Milne, 1972, Ayer, 1995, Powell et al., 1995, Heather, 2001, Bierlein et al., 2006). Both the Cardillac-Larder lake and the Porcupine-Destor fault zones involve complex tectonic histories, suggesting reactivation of the faults from reverse fault to strike slip (Robert, 1989, Powell et al., 1995).

The Swayze greenstone belt hosts a range of deposits including Cu-Pb-Zn, Ni-Cu-Fe, Ag and Au, and contains alteration-styles, deformations, and mineralizations that are very similar to the mineral-rich Abitibi greenstone belt. However, the Swayze greenstone belt lacks significant base metals and precious metals such as Au and Ag (Ayer, 1995, Heather, 2001). According to Hastie (2017), the gold deposit types found in the Swayze greenstone belt have been referred to as intrusion-related, syenite-associated, greenstone-hosted and gold-rich VMS deposits. A detailed account of the structural features, metamorphic history, alteration, and mineralogy of orogenic gold deposits and an intrusion-related gold deposit in the Swayze greenstone belt have been summarised in Maepa and Smith (2020). The Swayze greenstone belt has undergone multiple deformational events and volcanism spanning from 2739 ± 1 to 2695 Ma, and most of the host rocks have undergone regional greenschist metamorphism (Breemen et al, 2006). There are many known orogenic gold deposits, such as the Jerome and Namex deposits, which are hosted in highly deformed and sheared rock units, along lithological contacts, within metasomatic quartz veins and near feldspar porphyry units (Fumerton and Houle, 1995, Hastie, 2017). The world-class intrusion related Côté Au-(Cu) deposit is the only well studied and documented intrusion-related gold deposit in the Swayze greenstone belt. It is hosted within tonalite, diorite and quartz intrusion. There are currently only seven intrusion-related gold deposit prospects that are hosted in the Swayze greenstone belt. The Swayze greenstone belt comprised of a range of extrusive and intrusive rock types including felsic to ultramafic rocks overlain by metasedimentary rocks typically metamorphosed from greenschist to subgreenschist facies with local amphibolite facies found adjacent to synvolcanic and syntectonic intrusions (Heather, 2001, Breemen et al., 2006). The igneous rocks consist of plutonic and volcanic rocks with mafic to felsic volcanic assemblages in the Abitibi and Swayze greenstone belts with compositions ranging from rhyolitic to komatiitic, while sedimentary rocks occur as clastic and chemical metasedimentary rocks (Heather, 1998, Heather, 2001, Ayer et al., 2002, Breemen et al., 2006). The plutonic rocks range from synvolcanic, syntectonic and post-tectonic in age and are found within the greenstone belt as well as in the surrounding granitoid complexes, while the sedim (Ayer et al., 2002) entary rocks are mainly found at the top of the succession (Heather, 1998, Heather, 2001, Ayer et al., 2002, Breemen et al., 2006). According to Ayer et al. (2002), both geological mapping and geochronological studies done in the southern Abitibi greenstone belt support a coherent, upward-facing, autochthonous stratigraphy.

According to Ayer et al. (2002), there are six recognized stratigraphic supracrustal groups consisting of ultramafic and felsic assemblages that are found within the Swayze with ages ranging from 2740 Ma to 2695 Ma. These are the Chester, Marion, Biscotasing, Trailbreaker, Swayze Group and Ridout groups that have been correlated with coeval assemblages across the southern Abitibi such as the Pacaud, Deloro, Kidd-Munro, Tisdale, Blake River and the Timiskaming assemblages, respectively. The oldest rocks are found in the Chester Group which is dated at 2739 ± 1 Ma, and consists of mafic volcanic rocks overlain by sedimentary rocks that have been disrupted by younger diorite and tonalite intrusions of the Kenogamissi granitoid complex (Breemen, 2006). At the top of the Swayze greenstone belt stratigraphic succession lies the Ridout Group that consists of younger sedimentary and volcanic rocks with an age spanning from 2742 Ma to 2688 Ma. According to Breemen et al. (2006), these rocks are spatially associated with D₂ and D₃ high-strain zones. Details on the compositions of each stratigraphic group can be found in Heather (2001), Ayer et al., (2002), and Breemen et al., (2006).

As outlined by Heather, 2001, Ayer, 1995, the Swayze greenstone belt has undergone a complex and protracted structural history involving the development of multiple foliations (S₁, S₂, S₃., etc.), folds (F₁, F₂, F₃., etc.), faulting, as well as the occurrences of major deformational events (D₁, D₂, D₃., etc.) with deformation events occurring between 2740 Ma and 2660 Ma. According to these authors, the oldest deformational event D₁ consists of preserved penetrative foliations, intrafolial folds, and contact shear zones. The second deformational event resulted in the D₂ high-strain zones (Fig. 2) that have a general east–west orientation and are spatially correlated and synchronous with gold mineralization in the northern and southern Swayze greenstone belt as well as in the Abitibi gold camp (Heather, 2001, Breemen et al., 2006, Maepa and Smith, 2020). The Ridout high-strain zone (HSZ) has been interpreted as a syn-to late D₂ structure having local association with Ridout Group sedimentary rocks that have been mapped to be associated with the Timiskaming sedimentary assemblages in the Abitibi greenstone belt (Heather, 2001, Ayer et al., 2002). According to Heather (2001), two types of D₂ high-strain zones are observed in the Swayze greenstone belt, one localised along the limbs of F₂ folds as well as major lithological contacts and another found along the isoclinal geometry of F₂ synclines. The D₃ deformation events in the Swayze greenstone belt have prominent N, NE-, ENE- and ESE directions (Heather, 2001). The D₃ deformational events are associated with NE-striking fractures that are locally associated with felsic and mafic dikes. The syntectonic (D₂) alterations in the Swayze greenstone belt include chlorite, sericite, Fe-carbonate, sulphidation, carbonatization and tourmalization, while the late tectonic alteration has resulted in chloritization, epidotization, hematization, and silicification (Harris et al., 2000, Heather, 2001).

According to Heather (2001), the grade of metamorphism varies from amphibolite to greenschist facies transition that coincides with the timing of McOwen and Fawn D₁ high-strain zones, and the emplacement of Kenogamissi granitoid complex. Furthermore, there is retrograde amphibolite to greenschist facies metamorphism along the D₂ and D₃ high-strain zones. In addition, Heather, 2001, Breemen et al., 2006, suggest that the amphibolite facies metamorphism is pre-F₂ folding and D₂ events.

Thurston (2002) suggested an autochthonous/parautochthonous development of Archean greenstone belts in the Superior Province with several lithotectonic assemblages representing a variety of geodynamic settings including island arc and back arc settings (Condie, 1986). Work by Calvert and Ludden (1999) suggest that the Abitibi greenstone belt was formed in a subduction-related geodynamic setting.

The Swayze greenstone belt hosts various gold deposit types as outlined by Hastie (2017), ranging from orogenic gold deposits (mesothermal, syenite associated, banded iron formation, and greenstone-hosted gold deposits), to intrusion-related gold, similar to gold mineralization hosted in the Abitibi greenstone belt. The intrusion related Côté Au(-Cu) deposit is hosted in tonalite, diorite and quartz-diorite intrusions. The Ridout high-strain zone occurs approximately 3 km north of the Côté Au(-Cu) deposit (Katz, 2016). U-Pb zircon geochronological ages constrain the ages of the Yeo Formation to 2739 ± 1 Ma and 2734 ± 2 Ma (Breemen et al., 2006, Katz, 2016). According to Katz (2016), gold is disseminated in tonalite and dioritic rocks, and occurs in sheeted veins and stockworks, and the age of mineralization is between 2737 ± 11 Ma and 2741 ± 11 Ma, which is consistent with the timing of D₂ events. An in-depth description and detailed analysis on the Côté Au(-Cu) deposit can be found in Katz (2016).

This study will focus only on orogenic gold deposits in the Swayze greenstone belt with characteristics as outlined by Groves et al. (1998). These gold deposits in the Swayze greenstone belt are hosted in a variety of rocks including feldspar porphyry intrusions and diorites, polymictic conglomerates, iron-rich tholeittic and breccias, mafic to ultramafic metavolcanics rocks as well as quartz carbonate veins (Fumerton and Houle, 1995, Hastie, 2017, Maepa and Smith, 2020). The host rocks are typically regionally metamorphosed to greenschist metamorphic facies, brecciated, and intensely altered, with mineralization found near lithological contacts, folds, fractures, joints, lower-order faults, and high-strain zones (Heather, 2001). Typical alteration signatures include carbonate, silicification, hematite, chlorite, sericite, and biotite alteration (Hastie, 2017).

This study seeks to outline the orogenic mineral systems of the Swayze greenstone belt and to use the orogenic gold deposit prospects as target datasets for training spatial modelling tools and creating prospectivity maps.

The rate of mineral deposit discovery has been drastically decreasing despite massive levels of exploration expenditure (Schodde, 2014, Schodde, 2017). The low mineral-discovery rate has led researchers to explore deeper mineral targets in greenfield environments. However, to improve mineral exploration in greenfield environments, the mineral systems approach and processes that operated at regional to district scale to distribute mineral deposits has been developed (Wyborn et al., 1994, Knox-Robinson and Wyborn, 1997). Fyfe and Kerrich (1976) were the first to introduce the “systems approach”, however it has become more popular since Wyborn et al., (1994) re-introduced following the establishment and success of the petroleum systems approach used in the oil and gas industry for petroleum exploration (Magoon and Dow, 1994, Wyman et al., 2016). According to Wyborn et al. (1994), the critical processes that make up a mineral system include: 1) the source of the metals or ores (environments: igneous, sedimentary, or metamorphic), 2) source of the fluid responsible for leaching and transporting metals (meteoric, metamorphic or magmatic fluids), 3) fluid pathways and traps (high-strain zones and crustal scale faults), 4) thermal and pressure gradient and 5) processes that facilitate deposition. Although mineral systems cannot be observed directly, their expressions and proxies can be translated using available geoscience data in GIS (McCuaig et al., 2010, McCuaig and Hronsky, 2014, Kreuzer et al., 2019). Recently, the mineral systems approach has been applied in conjunction with MPM as a conceptual approach to help geoscientists to define and express key parameters of a mineral system for the purpose of improving mineral exploration targeting (McCuaig et al., 2010, Hagemann et al., 2016, Wyman et al., 2016, Tessema, 2017, Ford et al., 2019, Groves et al., 2020). Mineral deposits are rare events requiring that their associated mineral systems exist within a fertile lithosphere in the appropriate geodynamic setting (McCuaig and Hronsky, 2014, Groves et al., 2020).

There is no consensus regarding the source of mineralizing fluids and transport ligands associated with orogenic gold deposits due to uncertainty and lack of agreement among various authors. Previous studies postulated several sources of fluid, example; mantle-derived fluid, magmatic activity, fluid formed during metamorphic devolatilization, dewatering of sedimentary basin and hydrosphere (Gaboury, 2019, Robb, 2005, Hagemann et al., 2016). However, according to Pirajno (2008), the role of igneous or magmatic activity remains uncertain even though oxygen and deuterium isotope data plot in the magmatic and metamorphic fluid source regions (Ridley and Diamond, 2000). Furthermore, studies by Pitcairn et al., (2006) in the Otago province of New Zealand show that the magmatic-hydrothermal source model cannot fully explain the role of magmatic activity as the source of gold, and magmatic activity cannot provide a universal model for formation of orogenic gold deposits (Pitcairn et al., 2006, Phillips and Powell, 2010, Wyman et al., 2016, Groves et al., 2020). The supracrustal metamorphic source model which favours metamorphic devolatilization of hydrated supracrustal rocks at greenschist to amphibolite metamorphic facies has become a generally accepted model for orogenic gold deposits because it corresponds to late metamorphic and deformational timing of gold deposition globally and no specific host rock is required (Goldfarb and Groves, 2015, Goldfarb et al., 2001, Groves et al., 2016, Wyman et al., 2016). Evidence from metamorphic devolatilization models of crustal sediments (Phillips and Powell, 2010), fluid inclusions (Ridley and Diamond, 2000) and thermodynamic modelling (Tomkins, 2013, Zhong et al., 2015) shows that sedimentary rocks release large amounts of fluid during transition from greenschist to amphibolite facies metamorphism supporting the devolatilization of crustal sedimentary rocks as sources for orogenic fluids (Gaboury, 2019).

The Swayze greenstone belt is characterized by gold associated with quartz carbonate veining (Ayer, 1995, Heather, 2001) with dominant alterations including Fe-carbonate, sericite, and chlorite alterations (Ayer, 1995, Hastie et al., 2015), which serve as good evidence for fluid generated during prograde metamorphism (Pirajno, 2008). All the Archean aged rocks in the Swayze greenstone belt have undergone sub-greenschist to amphibolite facies metamorphism (Heather, 2001) and the metamorphic devolatilization model suggests that the transition between greenschist to amphibolite facies can generate metamorphic fluids (Phillips and Powell, 2010). The exact source of mineralization fluids in the Swayze greenstone belt is unknown. There have not been any fluid inclusion studies known by the authors in the Swayze greenstone belt for orogenic gold deposits that may suggest a different source for hydrothermal fluids. Only the intrusion-related Côté Au(-Cu) deposit has been extensively studied and sources of fluids are said to be of magmatic origin (Katz, 2016). Field observations in the northern Swayze greenstone belt documented by Ayer (1995) identified the presence of hydrothermal silicification and carbonatization that is strongly associated with sericite and chlorite alterations and epidotization.

Isotopic studies done by Kerrich (1986) on the mineralization in the Abitibi greenstone belt suggest that gold vein mineralization postdates late alkaline magmatic activities, and the possible source of hydrothermal fluids is late tectonic devolatilization during metamorphism with minor magmatic contributions (Sibson and Poulsen, 1988). Since the Swayze greenstone belt is interpreted to be an extension of the Abitibi, these sources of hydrothermal fluids might hold true in the Swayze greenstone belt as well. Assay result observations by Hastie (2017) of altered feldspar porphyry veins in the Kenty deposit suggest that gold was originally bound in pyrite and was later remobilized into favourable structural traps, i.e., the mobilization of gold might be due to metamorphic and deformational events. Recent work by Hastie et al. (2020) argues that gold remobilization in the Kenty deposit (Fig. 2) might be due to gold being redeposited from a polymetallic sulfide melt and considers unlikely that metamorphic fluids could be the source since metamorphic fluids can only hold less than 100 ppb Au in solution at the Kenty deposit scale (Wagner et al., 2016).

Orogenic mineral systems are usually found in greenschist to amphibolite facies metamorphosed supracrustal rocks that are structurally controlled with evidence of deep circulating crustal fluids and strong spatial associations with intrusive rocks (Pirajno, 2008). There is uncertainty and lack of agreement on the source of orogenic gold deposits (Tomkins, 2013, Gaboury, 2019) with studies suggesting that gold is associated with specific rock types; syenites (Robert, 2001), tonalite-trondhjemite-granodiorite (Katz et al., 2017); and iron-formations (Lambeck et al., 2011) among others (Tomkins, 2013, Gaboury, 2019). Other suggested sources of gold include primary diagenetic pyrite within sedimentary rocks such as black shale (Pitcairn et al., 2006, Large et al., 2012, Thomas et al., 2011). Globally, orogenic gold deposits are found in tectonically complex accretionary settings related to subduction of oceanic slab with an underlying sedimentary wedge (Goldfarb and Groves, 2015, Groves et al., 2016; Wyamn et al., 2016; Groves et al., 2020). At this geodynamic setting, there is devolatilization of the subducting slab and the oceanic sediment wedge resulting in a release of gold and Ag, As, Bi, Sb, Te and W bearing fluids from pyrite and pyrrhotite in the sediments (Groves et al., 2020). Although Xue et al. (2013), suggest that granitic and other igneous rock types can be the sources of orogenic gold deposits in Archean environments, several authors regard the sedimentary model as a more robust and generally acceptable source of gold (e.g., Gaboury, 2019, Tomkins, 2013). Gold can be transported by hydrothermal fluids as metal–ligand complexes including chloride (Cl^-) and hydrogen sulfide (HS^-). The hydrothermal fluids are aqueous, dilute, carbonic acid and low salinity fluids with >6 wt% of NaCl equivalent that contain CO₂ ± CH₄ ± N₂ ± H₂ and are generally enriched in S (~1000 ppm) but low in Cl (~60 ppm) suggesting metamorphic fluids were formed in the crust (Hagemann and Cassidy, 2000, Kerrich et al., 2000). Orogenic gold deposits that are found within the Swayze greenstone belts are generally associated with sheared and brecciated felsic intrusive rocks (quartz-feldspar porphyries), quartz veins and intermediate volcanic rocks (diorite). For example, the Jerome mine contains gold hosted in a sheared contact between quartz-feldspar porphyry and epiclastic Timiskaming-type sedimentary rocks (Ayer, 1995). Gold is also associated with chalcopyrite, molybdenite, tetrahedrite, sphalerite and some principal alterations including biotitic, hematization, chloritic, carbonatization and silification. At the Rundle mine (a developed prospect with reserves), gold is hosted in cataclastic schist and vein stockworks, and the deposit is associated with the feldspar porphyry intrusive rocks (Hastie et al., 2015).

Gold is hosted in a variety of rocks that have been sheared and deformed with some mineralization associated with major deformational events (D₂). Felsic intrusions are believed to be the possible sources of the metal-rich fluids and possibly related to the introduction, remobilization, and concentration of metals in the Swayze greenstone belt (Marmont, 1983). According to Marmont (1983), the possible processes that account for the association of gold with felsic intrusions are: 1) Magmatic association - gold has spatial relationship with felsic intrusions and dikes, so the intrusions may be the source of gold; 2) Metamorphic association - felsic intrusions may have acted as “heat” sources during contact metamorphism, providing heat that caused metal bearing fluid convection; and 3) Assimilation - during emplacement, intrusions may have assimilated mineralized xenoliths from country rocks. However, Bierlein et al. (2006), suggest that no specific rock type can be considered as the source of gold because available gold can be leached and transported through faults and deposited when the conditions are favorable. Also, Kerrich (1991) noted that some felsic intrusions in the Abitibi greenstone belt pre-date gold deposition and therefore cannot be regarded as sources of gold but rather they serve as physical traps for vein formation (Goldfarb and Groves, 2015).

Orogenic mineral systems require a fluid plumbing system that propagates gold bearing fluids to their depositional sites (McCuaig and Hronsky, 2014). Metallogenic provinces hosting world class orogenic gold deposits such as the Abitibi subprovince and the Yilgarn craton have been widely studied and researchers agree that crustal to lithospheric scale faults, shear zones or high-strain zones are required to focus fluids towards their depositional sites (Groves et al., 2005, Goldfarb et al., 2005, Robert et al., 2005, Vearncombe and Zelic, 2015, Groves et al., 2020). Research on fault systems also indicate that the irregularity of major shear zones (such existence of kinks and bends) allows for a transfer of large volumes of fluids through the earth’s crust (Vearncombe and Zelic, 2015). According to Connolly, 2010, Cox, 2016, fluid decompression pressure due to metamorphic process may be the most important event in generating orogenic gold deposits because it helps trigger seismicity resulting in fault reactivation, which allow fluid flow to shallower levels of the crust (Gaboury, 2019). Like in other Archean orogenic gold deposits, there is evidence that the main pathways for fluids in the Swayze greenstone belt are D₂ high-strain zones, fractures, and faults. Mineralization in the Abitibi greenstone belt is spatially associated with the D₂ Porcupine-Destor faults and the Cardillac-Larder fault zones. Similarly, the Slate Rock and the Ridout high-strain zones are spatially associated with gold mineralization in the Swayze greenstone belt (Ayer et al., 2002, Heather, 2001, Hastie, 2017, Maepa and Smith, 2020). It is therefore suggested that the large scale D₂ crustal faults may be the regional-scale geological controls that focus fluids to lower-order structures in the Swayze greenstone belt (Maepa and Smith, 2020). Similar to the Abitibi greenstone belt, the second and third order structures as well as breccias and fractures are the district to prospect scale controls of gold mineralization (Heather, 2001, Love and Roberts, 1991).

Since there is a strong association between structures and gold deposit distribution, the field-based mappable structures as well as faults derived from aeromagnetic data can be used to outline the pathways of deep to shallow circulating fluids. However, there are structures considered to be post-mineralization, and thus, spatial analysis techniques such as Fry analysis and directional distribution were used to determine prominent fault orientations that are spatially and statistically associated with gold mineralization at regional and prospect scales (Parsa and Maghsoudi, 2018, Maepa and Smith, 2020). An evaluation of controls for gold mineralization using Fry analysis (shown on Fig. 3), fractal processes and distance distributions between gold and faults in the Swayze greenstone reveals that gold displays bifractal behaviour at scales >8 km and >8 km. Conclusions by Maepa and Smith (2020) suggest that 97% of gold being found at distances less than 4 km from second-order faults and fractures, while 96% of gold occur at distances less than 7 km from first-order D₂ high-strain zones. The D₂ high-strain zones and faults are spatially and statistically correlated with mineral deposit distributions with the ESE-WNW, NE – SW, and ENE-WSW strike, which appear to be the preferred orientations of mineral deposits at local to regional scales (Fig. 3).

Fry analysis is a spatial autocorrelation technique that can be used to investigate if the spatial distribution of point objects such as gold mines and prospects occur along linear trends. According to this method, for every n number of points, there would be n²-n translations (Hanna and Fry, 1979) . Applications of Fry analysis in exploration targeting have proven useful in determining the overall trend of metallogenic processes and the spatial relationships of mineral deposits (Vearncombe and Vearncombe, 1999, Kreuzer et al., 2007, Carranza, 2009, Parsa et al., 2018a, Parsa et al., 2018b, Parsa and Maghsoudi, 2018). In this study, Fry translations of mineral deposits distributions are displayed using rose diagram as shown in Fig. 3. The scale-integrated Fry analysis of gold mines and prospects that was carried out in the Swayze greenstone belt shows that the orientations of mineral deposits vary from ESE - WNW trend to ENE-WSW strike (Maepa and Smith, 2020).

Assuming a hydrothermal fluid source, the interaction of hot aqueous fluids with the country rock would result in multiple alteration signatures such as sericite, calcite, chlorite, hematization, silicification, and carbonatization (Gaboury, 2019, Groves et al., 2020). As fluids flow through second and third-order structures and infiltrate planes of weaknesses between lithological contacts, the chemical interaction of hot aqueous fluids with Fe-rich and carbonaceous sedimentary units disturb the metal–ligand complexes resulting in metal precipitation (Goldfarb and Groves, 2015). Regardless of the metal–ligand complex in the fluid system, reducing the S content in the fluid, changes in the pH and fO₂ associated with fluid and rock interactions will affect the solubility of gold resulting in metal precipitation (Gaboury, 2019). Love and Roberts (1991) suggest that the movement of hydrothermal fluids at the Rundle deposit in the Swayze greenstone belt was controlled locally by permeable zones and competency contrast between the felsic porphyry rocks and ductile mafic volcanic rocks. Decreasing temperature and pressure conditions as hydrothermal fluids move through faults and interact with brittle and ductile country intrusive rocks of the Swayze greenstone belt resulted in metal precipitation. The presence of chlorite, calcite and quartz represent reactions of fluids with mafic volcanic and feldspar porphyry rocks which resulted in various alteration signatures that spatially correlate with gold precipitation (Love and Roberts, 1991).

Finally, to have world-class mineral deposits, there needs to be suitable post-tectonic processes that ensure the preservation of mineralization (Groves et al., 2020). Preservation is a critical aspect of mineral systems because it ensures that deposits remain in their current depositional environment and are not eroded away (Groves et al., 2005). Orogenic gold deposits are usually well-preserved because they form later in the orogenic event and at crustal depths >2 km (Goldfarb et al., 2001, Groves et al., 2020). The Archean volcanic greenstone sequence rocks of the Swayze greenstone belt are overlain by impermeable younger metasedimentary and volcanic rocks of the Ridout Group which help for preservation of mineral deposits. The sedimentary rocks of the Ridout groups form long linear units found to be spatially associated with the regional D₂ and the D₃ high-strain zones (Breemen et al., 2006). Furthermore, the sedimentary rocks of the Ridout group are especially localised near the D₂ Ridout high-strain zone and unconformably overlay the older volcanic and sedimentary packages (Heather, 2001). The fault-bounded sedimentary rocks of the Ridout group were analysed with detrital zircons and yielded a maximum depositional age of 2688 ± 2 Ma (Breemen et al., 2006). The burial by impermeable sedimentary rocks of the Ridout group and possible down-faulting geometry of structures possibly account for the degree of preservation in the Swayze greenstone belt.