Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa

OREGEO-01590; No of Pages 20 Ore Geology Reviews xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Ore Geology Reviews journal homepage: www.elsevier.com/locate/oregeorev

Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa Bijal Chudasama a,⁎, Alok Porwal a,b, Oliver P. Kreuzer c,d, Kris Butera c a

Centre of Studies in Resources Engineering, Indian Institute of Technology Bombay, Powai, 400076, India Centre for Exploration Targeting, University of Western Australia, Crawley, 6009 WA, Australia Corporate Geoscience Group, PO Box 5128, Rockingham Beach, WA 6969, Australia d Economic Geology Research Centre, James Cook University, Townsville, QLD 4811, Australia b c

a r t i c l e

i n f o

Article history: Received 22 April 2015 Received in revised form 27 July 2015 Accepted 18 August 2015 Available online xxxx Keywords: West Africa Craton Ghana Paleoproterozoic Kumasi Basin Orogenic gold Mineral systems Prospectivity modelling Fuzzy inference systems

a b s t r a c t This paper describes the geology and tectonics of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, as applied to predictive mapping of prospectivity for orogenic gold mineral systems within the basin. The main objective of the study was to identify the most prospective ground for orogenic gold deposits within the Paleoproterozoic Kumasi Basin. A knowledge-driven, two-stage fuzzy inference system (FIS) was used for prospectivity modelling. The spatial proxies that served as input to the FIS were derived based on a conceptual model of gold mineral systems in the Kumasi Basin. As a first step, key components of the mineral system were predictively modelled using a Mamdani-type FIS. The second step involved combining the individual FIS outputs using a conjunction (product) operator to produce a continuous-scale prospectivity map. Using a cumulative area fuzzy favourability (CAFF) curve approach, this map was reclassified into a ternary prospectivity map divided into high-prospectivity, moderate-prospectivity and low-prospectivity areas, respectively. The spatial distribution of the known gold deposits within the study area relative to that of the prospective and non-prospective areas served as a means for evaluating the capture efficiency of our model. Approximately 99% of the known gold deposits and occurrences fall within high- and moderate-prospectivity areas that occupy 31% of the total study area. The high- and moderateprospectivity areas illustrated by the prospectivity map are elongate features that are spatially coincident with areas of structural complexity along and reactivation during D4 of NE–SW-striking D2 thrust faults and subsidiary structures, implying a strong structural control on gold mineralization in the Kumasi Basin. In conclusion, our FIS approach to mapping gold prospectivity, which was based entirely on the conceptual reasoning of expert geologists and ignored the spatial distribution of known gold deposits for prospectivity estimation, effectively captured the main mineralized trends. As such, this study also demonstrates the effectiveness of FIS in capturing the linguistic reasoning of expert knowledge by exploration geologists. In spite of using a large number of variables, the curse of dimensionality was precluded because no training data are required for parameter estimation. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The Birimian metavolcanic and metasedimentary successions of Ghana form part of the Paleoproterozoic West African Craton that also underlies much of Burkina Faso, Niger, Cote d'Ivoire, Mali, Sierra Leone, Liberia, Guinea, and Senegal. Some of the world's largest gold deposits are localised along a network of shear zones and faults that cut the Paleoproterozoic Birimian Supergroup of West Africa (Allibone et al., 2002b). Much of this mineral wealth is concentrated in Ghana, the second largest gold producer in Africa (Brown, 2015). The main gold belts in southwestern Ghana are the Ashanti Belt (total

⁎ Corresponding author.

endowment ca. 170 Moz Au), the Sefwi–Bibiani Belt (N 30 Moz Au) and the Asankrangwa Belt (N10 Moz Au) (Fig. 1; Table 1). Compared to the established gold belts, such as Ashanti and Sefwi– Bibiani, exploration activities within the Asankrangwa Belt (Fig. 2a) and enclosing Kumasi Basin have been limited and mainly focused around artisanal and past colonial operations. Possible explanations for the scarcity of detailed exploration are: (1) extensive recent alluvial and colluvial cover, and (2) relatively poor knowledge of the geology and structure of the Kumasi Basin. Moreover, (3) it was not until the mid-1990s that the gold potential of the Asankrangwa Belt (a shear zone along the central axis of the Kumasi Basin, Fig. 2a) was fully recognised. The discovery in 2006 by Asanko Gold Inc. of the outcropping Esaase gold deposit (Table 1) illustrates the potential of the

http://dx.doi.org/10.1016/j.oregeorev.2015.08.012 0169-1368/© 2015 Elsevier B.V. All rights reserved.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

2

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 1. Study area: Kumasi Basin — location and extent (Jessell et al., 2012); the rectangle marks the area shown in Figs. 2 and 3.

Asankrangwa Belt and wider Kumasi Basin to yield further multimillion ounce gold deposits. Based on this premise, Asanko Gold Inc. commissioned Corporate Geoscience Group to undertake a geological framework, prospectivity and targeting study of this emerging and potentially well-endowed belt. The work undertaken by Corporate Geoscience Group in close collaboration with Asanko Gold Inc.'s technical team resulted in (1) new geological syntheses and interpretations of the Asankrangwa Belt and wider Kumasi Basin, (2) a better understanding of the controls on gold mineralization, and (3) an updated conceptual model. This input was crucial for the subsequent GIS-based gold prospectivity modelling employing

both data-driven weights of evidence (Miller et al., in press) and knowledge-driven fuzzy inference system (FIS) approaches. A two-stage, knowledge-driven, Mamdani-type fuzzy inference system (FIS) was implemented to model the orogenic gold prospectivity in the Kumasi Basin. The key components of the mineral system were modelled in the first stage FIS, and their outputs were combined using a conjunction operator in the second stage in order to derive the prospectivity map. The FIS model aims to capture the linguistic reasoning of exploration geologists in the form of fuzzy if–then rules and objective mathematical functions, rather than in the form of explicit class weights and map weights as in traditional fuzzy models.

Table 1 Grade and resource figures for selected, significant gold deposits and camps in Ghana. Deposit/camp

Obuasi Tarkwa Ahafo Prestea Iduapriem Damang Edikan Akyem Esaase Bibiani Wassa Nkran Chirano Bogoso Konongo Nzema a b c d

Region

Ashanti Belt Ashanti Belt Sefwi–Bibiani Belt Ashanti Belt Ashanti Belt Ashanti Belt Kumas Basin Ashanti Belt Asankrangwa Belt Sefwi–Bibiani Belt Ashanti Belt Asankrangwa Belt Sefwi–Bibiani Belt Ashanti Belt Ashanti Belt Ashanti Belt

Average gradea

Current resouceb

(g/t Au)

(oz Au)

5.32 1.15 1.96 4.79 1.39 1.92 1.10 1.72 1.45 3.40 2.21 2.34 2.46 2.74 3.80 1.30

27,360,000 9,568,000 10,120,000 1,084,000 6,610,000 5,260,000 7,411,000 7,180,000 5,910,000 1,700,000 3,519,000 3,470,000 2,271,000 2,000,000 942,000 1,900,000

Historic productionc

Total endowmentd

References

29,500,000 9,700,000 5,000,000 9,000,000 2,850,000 4,100,000 600,000 ? 0 4,000,000 1,277,422 590,000 1,264,626 1,108,234 1,600,000 400,000

56,860,000 19,268,000 15,120,000 10,084,000 9,460,000 9,360,000 8,011,000 7,180,000 5,910,000 5,700,000 4,796,422 4,060,000 3,535,626 3,108,234 2,542,000 2,300,000

AngloGold Ashanti (2014), SRK Consulting (2008) Gold Fields (2014, 2015) Newmont (2014a) www.gsr.com AngloGold Ashanti (2014) Gold Fields (2014, 2015) www.perseusmining.com.au Newmont (2014b) www.asanko.com Resolute Mining (2014a,b) www.gsr.com www.asanko.com, SRK Consulting (2011) www.kinross.com www.gsr.com, SRK Consulting (2014) Signature Metals (2015) www.endeavourmining.com

For current resources only. Most recent, publically available global resource figures including Measured, Indicated and Inferred Resources and Reserves. If applicable and/or reported. Many of the historic production figures are incomplete. Sum of current resource and historic production.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

3

Fig. 2. (a) Geology of Kumasi Basin, and Asankrangwa Belt, (b) structures of the Kumasi Basin, and the Asankrangwa Belt.

2. Geological setting of the Kumasi Basin The Kumasi Basin forms part of the Archean to Paleoproterozoic (Birimian) Man-Leo Shield (Lompo, 2010; Jessell et al., 2012) that in Ghana is exclusively composed of metasedimentary and metavolcanic rocks of the Paleoproterozoic Birimian Supergroup. Locally, these are unconformably overlain by metasedimentary rocks of the Tarkwaian Group that represent erosion products of the Birimian Supergroup (Leube et al., 1990). Within the Birimian Supergroup in Ghana, seven generally NE–SWstriking, linear greenstone belts of 2250 to 2170 Ma mafic volcanic arc (Davis et al., 1994; Hirdes et al., 1992, 1996; Loh and Hirdes, 1996, 1999; Hirdes and Davis, 1998; Feybesse et al., 2006) and associated sedimentary rocks are separated by major faults from intervening 2150- to 2100-Ma sedimentary rocks (Feybesse et al., 2006). The latter were deposited within foreland molasse basins, such as the Kumasi Basin, which over time evolved into a back arc basin (Leube et al., 1990; Davis et al., 1994; Hirdes and Davis, 2002; Perrouty et al., 2012). The Birimian Supergroup in the Kumasi basin comprises metasedimentary rocks (wacke, sand- and siltstone and phyllites), fault-bound slices of metavolcanic rocks (basaltic, dacitic and rhyolitic flows and minor interbedded volcaniclastics) and metaluminous granitoids of the Eburnean plutonic suite that intruded the basin sequences from ca. 2116 to 2088 Ma (Adadey et al., 2009). The Birimian metavolcanic rocks are mainly tholeiites and calc-alkaline basalts that recorded lower to upper greenschist facies metamorphism (Eisenlohr, 1989; Leube et al., 1990; Eisenlohr and Hirdes, 1992; Davis et al., 1994; Hirdes et al., 1996; Oberthür et al., 1998; Béziat et al., 2000; Hirdes and Davis, 2002;

Feybesse et al., 2006; Ganne et al., 2012; Jessell et al., 2012). The Birimian metasedimentary rocks, derived from adjacent Palaeoproterozoic greenstone belts (Davis et al., 1994; Jessell et al., 2012), include laterally gradational facies of volcaniclastic rocks, wackes, argillites, cherts, manganiferous and carbon-rich chemical sedimentary rocks, and carbonates (Junner, 1935, 1940; Leube et al., 1990; Oberthür et al., 1998; Hirdes and Davis, 2002). After emplacement and deposition, the Birimian volcanic and sedimentary rocks were intruded by belt and basin granitoids, respectively, that were emplaced during discrete magmatic pulses between approximately 2180 and 2088 Ma (Allibone et al., 2002b). The older “belt granitoids” (ca. 2180 Ma: Allibone et al., 2002b) are of tonalitic and metaluminous composition (Hirdes and Leube, 1989; Leube et al., 1990; Hirdes and Davis, 2002), whilst the younger “basin granitoids” (ca. 2116–2088 Ma: Allibone et al., 2002a) are peraluminous and relatively enriched in Rb and K (Hirdes and Leube, 1989; Leube et al., 1990). The entire Birimian sequence was uplifted and eroded during the Eburnean orogeny at around 2100–2090 Ma (Bonhomme, 1962; Taylor et al., 1988; Leube et al., 1990; Hirdes et al., 1996; Oberthür et al., 1998; Hirdes and Davis, 2002). The Tarkwaian Group (2132–2097 Ma: Davis et al., 1994; Oberthür et al., 1998) consists of detrital sediments, including clasts, derived from the erosion of, and unconformably overlying, rocks of the Birimian Supergroup (Leube et al., 1990; Hirdes and Davis, 2002). Given the respective age relationships, the Tarkwaian sedimentary cycle was at least partly coeval with the Eburnean Orogeny that involved emplacement of granitoid rocks known as the Eburnean Plutonic Suite into both Birimian Supergroup and Tarkwaian group. However there have been no confirmed occurrences of Tarkwaian rocks within the Kumasi Basin.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

4

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 3. Kumasi Basin study area and its geological setting. From Feybesse et al. (2006).

The Kumasi Basin lies between the Ashanti Belt in the southeast and the Sefwi–Bibiani Belt in the northwest. To the north, the Kumasi Basin is concealed by the Neoproterozoic Volta Basin, whilst to the south it is concealed by Phanerozoic sediments (Loh et al., 2010) and the Atlantic Ocean. The Asankrangwa gold belt (Fig. 2a) occurs in a complex, 10–20 km wide, NE–SW striking shear zone system parallel to the central axis of the Kumasi Basin. It represents a fundamental crustal structure that has been repeatedly reactivated through the entire geological history of the region, from initial rifting through orogenesis. The Asankrangwa gold belt has been interpreted as a SE-verging fold and thrust belt that developed in response to reactivation and inversion of pre-existing basinforming structures during (1) D2, when the intervening stratigraphic sequences were tightly- to isoclinally-folded and transposed into the dominant NE–SW structural grain, and (2) D4/D5. The Asankrangwa belt comprises Birimian meta-sediments, minor granitic intrusions and mafic igneous rocks. The lithological units are cut by primary and secondary structures that dip steeply towards northwest. It is this structural and geological architecture that have been important in the localisation of gold mineralisation, and provides the capacity for enhanced fluid flow forming major alteration systems in association with gold bearing systems. The controls on gold mineralisation are similar to those in the Sefwi and Ashanti Greenstone Belts. Overall, the Asankrangwa Gold Belt is interpreted as an inverted half-graben in which growth faulting controlled the accumulation of Birimian metasedimentary rocks above the inferred Birimian metavolcanic rocks in the footwall (Siddorn and Lee, 2005). 3. Tectonic setting and evolution of the Kumasi Basin The Kumasi Basin (Fig. 3) is bound to the west (i.e., along its contact with the Sefwi–Bibiani Belt) by the Bibiani and Ketesso faults and to the

east (i.e., along its contact with the Ashanti Belt) by the Ashanti fault, two semiparallel, more than 200 km-long, lithospheric-scale shear zone systems (Duodu et al., 2009; Perrouty et al., 2012). These faults (1) separate domains of lower greenschist to non-metamorphosed facies in the Kumasi Basin, from domains of up to amphibolite facies in the Sefwi and Ashanti belts (Feybesse et al., 2006, p 162), and (2) control — directly or indirectly (in the form of 2nd order faults) — the location of large to very large Birimian gold systems such as Obuasi (+60 Moz Au) (Blenkinsop et al., 1994; Allibone et al., 2002b), Bibiani (+7 Moz Au) and Chirano (+5 Moz Au). The Kumasi Basin developed between 2150 Ma and 2100 Ma, initially as a foreland basin where Birimian Supergroup metasedimentary rocks were deposited in a shallow marine environment (Feybesse et al., 2006) We interpret that, over time, the Kumasi Basin evolved into a back-arc basin that was subducted beneath active volcanic arcs to the west (Sefwi–Bibiani Belt) and to the east (Ashanti Belt). Ongoing thinning of the crust triggered asthenospheric upwelling with the arrival of a mantle plume causing underplating of the thinned crust by mafic magmas (Siddorn and Lee, 2005). The resultant intermediate to felsic peraluminous melts produced in the lower crust were emplaced in the upper crust as the Eburnean Plutonic Suite, intruding Birimian Supergroup and Tarkwaian Group lithologies resulting into thickening of the crust. In the following paragraph, we review the Paleoproterozoic evolution of the Western African Craton, of which the Kumasi Basin is an important element, based on tectonic models proposed by Allibone et al. (2002a,b), Feybesse et al. (2006), Perrouty et al. (2012) and Jessell et al. (2012). The evolution of the Birimian in Ghana began at the margin of and with the breakup of the Archean São Luis Craton around 2350– 2300 Ma. The period 2350–2150 Ma recorded extensive magmatic activity and terrane accretion along the margin of the São Luis Craton,

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

5

Fig. 4. Palaeogeographic reconstruction and geodynamic model for the Ghanian province. From Feybesse et al. (2006).

eventually leading to cratonisation and formation of the first segments of West African continental crust such as the Sefwi–Bibiani and Ashanti Belts (Hirdes et al., 1992, 1996; Davis et al., 1994; Vidal et al., 1996; Fetter et al., 1997; Kosuch et al., 1997; Castellana and Long, 1998; Feybesse et al., 2000, 2006). Subsequent extensional deformation between 2150 and 2100 Ma led to the opening of, and deposition of flysh-type sediments in, the Kumasi Basin, which initially developed as a foreland molasse basin (Feybesse et al., 2006). Eventually the Kumasi basin evolved into a back

arc basin, as mentioned above, with active subduction under the Ashanti Belt and Sefwi–Bibiani Belt along eastern and western margins respectively This marked the beginning of the Eburnean Orogeny which commenced at around 2130 Ma and continued up to 1980 Ma (Davis et al., 1994; Oberthür et al., 1998; Allibone et al., 2002a,b; Feybesse et al., 2006; Jessell et al., 2012) and is characterised by several phases of deformation. Jessell et al. (2012) and Feybesse et al. (2006) describe the Eburnean Orogeny to be comprising of two major deformation phases:

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

6

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 5. The space–time diagram showing the geodynamic synthesis of the Birimian Supergroup, Ghana, West Africa. Data Sources: Metallogeny and Deformation History: Allibone et al. (2002a, b); Perrouty et al. (2012); Magmatism: Attoh et al. (2007); Bossiere et al. (1996); Diallo et al. (2001); Dioh et al. (2006); Doumbia et al. (1998); Gasquet et al. (2003); Hirdes et al. (1992); John et al. (1999); Klemd et al., (2002); Leube et al. (1990); Ndiaya et al. (1997); Oberthür et al. (1998); Pawlig et al. (2006); Taylor et al. (1992); Yao and Robb (2000);Metamorphism: Harcouet et al. (2005); Pawlig et al. (2006); Pigos et al. (2003);Gold Events: Le Mignot et al. (2014); Oberthür et al. (1998); Pigos et al. (2003); White et al. (2014).

D1 and D2. The first phase of deformation, D1, occurred from 2130 Ma to 2100 Ma and caused crustal thickening due to a NW–SE oriented compressional stress regime (Boher et al., 1992; Milési et al., 1992; Allibone et al., 2002a; Feybesse et al., 2006; Vidal et al., 2009; Jessell et al., 2012). The second phase of deformation, D2, was a period of strike-slip movement that formed sinistral to reverse sinistral shearing and caused reactivation along the pre-existing D1 thrust faults (Feybesse et al., 2006; Jessell et al., 2012). Fig. 4 graphically represents the sequence of events in the tectonic evolution of the Western African Craton. On the other hand, the tectonic model proposed by Allibone et al. (2002a,b; summarized below) is based on five phases of deformation during the Eburnean Orogeny that shaped the structural framework of the Kumasi Basin. However, the model proposed by Allibone et al. (2002a,b) is very much Ashanti Belt-centric in that the underlying geological, tectonic and structural observations and interpretations stem

from this belt alone and, thus, may only be indirectly applicable to the Kumasi Basin. Based on our observations from the Asankrangwa Belt the following elements of the Allibone et al. (2002a,b) model are most likely to apply to the Kumasi Basin. • D1 (2130 Ma; Stage 5a in Fig. 4): Basin inversion due to NW–SE compression resulted in development of upright, tight, NE-trending isoclinal folds (F1) and a weak bedding-parallel cleavage (S1) in the Birimian metasedimentary rocks. • D2 (2105 Ma; Stage 5a in Fig. 4): WNW–ESE to NW–SE directed compression eventually resulted in fold lock up and reactivation of pre-existing basin growth faults. Dextral strike-slip and dip-slip movement along NE–SW-striking shear zones caused further crustal thickening due to thrusting and stacking of thrust slices and due to emplacement of syn-tectonic granitoids. Consequently, a NE–SW trending shear zone developed along the central axis of the Kumasi

Table 2a Targeting criteria and spatial proxies for regional-scale modelling of orogenic gold potential of the Kumasi. Critical process: source Constituent Spatial proxies processes

Rationale

Energy Fluids Ligands Metals

Sources of energy, fluids, ligands and metals had to Not applicable be present to generate the known gold deposits. For the purpose of this study, the distribution of sources was assumed to have been homogeneously distributed as evidenced by gold deposits of similar age and character occurring in many parts of the Kumasi Basin and Ghana as a whole.

Magmatism (intrusions), metamorphism (grade) Metamorphic, magmatic, (meteoric?) Supracrustal rocks, pre-existing S-rich deposits Supracrustal rocks

Primary Data description data & sources

Predictor Method for maps predictor map generation

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

Critical process: source Constituent Spatial processes proxies Pathways

Rationale

D2 thrust faults

The location of gold deposition in the Kumasi Basin was controlled by major NE–SW-striking D2 thrust faults; these structures represent repeatedly reactivated basin growth faults Topographic Indicate presence of zones of ridges silicification due to hydrothermal alteration; these ridges indicate passage of hydrothermal fluids.

Primary data

Data description & sources

Predictor maps

Geophysical data (electromagnetics, magnetics and radiometrics), geological maps

Proximity to D2 thrust faults D2 thrust faults were interpreted based on all available geophysical and geological data, including Asanko Gold Inc. proprietory data plus data available from the Ghana Geological Survey

SRTM DEM

Ridges were extracted from public domain SRTM data and a detailed DEM owned by Asanko Gold Inc. using a ridge detection algorithm developed by Fathom Geophysics Pty Ltd capable of mapping ridge density, orientation and slope

Proximity to topographic ridges with elevations of 30–50 m greater than average elevation and NE–SW orientation (parallel to D2 thrust faults)

Method for predictor map generation (a) Extraction of D2 thrust faults from geological maps and geophysical grids and export into a new map layer; (b) calculation of Euclidean distance from each D2 fault (a) Identification of areas with elevations of 30–50 m greater than the average elevation; (b) extraction of NE–SW-trending ridge lines; (c) calculation of Euclidean distance from each ridge

Table 2c Targeting criteria and spatial proxies for regional-scale modelling of orogenic gold potential of the Kumasi Basin. Critical process: source Constituent Spatial proxies processes

Rationale

Primary data

Data description & sources

Predictor maps

Method for predictor map generation

Physical traps

NE–SW-striking D2 thrust faults were reactivated during D4/D5 (main gold event) as sinistral strike-/dip-slip faults Bends represent zones of dilational or contractional deformation that were critical for focussing fluid flow and localising gold deposition Strong competency contrasts between competent (e.g., granite) and incompetent (e.g., shale) rocks disturbs regional stresses leading to fracturing of competent rocks and giving rise to local zones of dilation and fluid focusing at or close to lithological contacts Redox reactions between hydrothermal fluids carrying soluble gold complexes and Fe-rich rocks may lead to their destabilization and precipitation of gold

Geophysical data (electromagnetics, magnetics and radiometrics), geological maps

Bends were interpreted both (a) manually from geological maps and structural interpretations, and (b) automatically from geophysical data using an algorithm developed by Fathom Geophysics Pty. Ltd.

Proximity to bends along D2 thrust faults

(a) Extraction of those D2 thrust fault segments that are characterised by a change of orientation diverging from NE–SW; (b) calculation of Euclidean distances from these flexures along D2 Thrusts.

Geophysical data (electromagnetics, magnetics and radiometrics), geological maps

Competent and incompetent rock types and their contacts were identified using all available geophysical and geological data, including Asanko Gold Inc. proprietory data plus data available from the Ghana Geological Survey

Density estimation of competency contrast values

(a) Extraction, as line features, of all geological contacts between different lithological units; (b) subtraction of competency values of rock units along both sides of each contact; (c) density estimation of competency contrast values

Geophysical data (electromagnetics and magnetics), geological maps

The distribution of Fe-rich Paleoproterozoic rocks was extracted from geological maps and interpreted from geophysical data, including Asanko Gold Inc. proprietory data plus data available from the Ghana Geological Survey

Proximity to Fe-rich Paleoproterozoic rocks

(a) Extraction, as linear features, of positive magnetic anomalies; (b) Calculation of Euclidean distance from magnetic anomalies (a) Extraction, as line features, of all geological contacts between different lithological units; (b) subtraction of chemical reactivity values of rock units along both sides of the contact; (c) density estimation of reactivity contrast values

Bends along D2 thrust faults

Lithological contacts (weighted by competency contrast density)

Chemical traps

Fe-rich Paleoproterozoic rocks Lithological contacts (weighted by reactivity contrast density)

Density estimation of reactivity contrast values

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

Table 2b Targeting criteria and spatial proxies for regional-scale modelling of orogenic gold potential of the Kumasi.

7

8

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

basin (Siddorn and Lee, 2005). • D3 (2090 Ma; Stage 6 in Fig. 4): A relatively minor, or far-field, deformation event causing NE–SW shortening with localised formation of crenulation cleavages. • D4 and D5 (1980 Ma; Stage 6 in Fig. 4): The D4 and D5 deformation events were characterised by transcurrent faulting and thereby transformed the stress regime from NW–SE compression to sinistral strike slip faulting. Brittle reactivation of and sinistral dip-slip movement occurred along the D2 shear zones due to N–S shortening during the D4 deformation phase. Sinistral strike-slip faults developed during the D5 phase in a NNE–SSW compression stress field. These faults were formed due to reactivation of the pre-existing D2 thrust faults. A geodynamic synthesis of the Birimian Supergroup including stratigraphy, magmatism, metamorphism, mineralization events, deformation history and tectonic events is shown in the form of a space–time diagram shown in Fig. 5. 4. Orogenic gold mineral system model, Kumasi Basin Orogenic gold mineralization in southern Ghana have been subject to extensive study (e.g., Junner, 1932; Hirdes and Leube, 1989; Leube et al., 1990; Appiah, 1991; Milési et al., 1991; Eisenlohr and Hirdes, 1992; Mumin et al., 1994; Oberthür et al., 1994; Allibone et al., 2002a, b; Feybesse et al., 2006; Perrouty et al., 2012). Most gold deposits in this region are hosted by shear zones cutting Birimian metavolcanic and metasedimentary rocks and/or granitoids of the Eburnean Plutonic Suite (Allibone et al., 2002a,b; Perrouty et al., 2012). Shear zone hosted gold mineralization is either in the form of (1) high grade quartz veins, or vein stockworks, in fractured rock associated with brittle deformation (Junner, 1932; Appiah, 1991; Allibone et al., 2002a,b), or (2) low grade disseminated sulphide mineralization comprising arsenopyrite and/or pyrite, in within shear zones which have undergone ductile deformation (Junner, 1932; Hirdes and Leube, 1989; Leube et al., 1990; Milési et al., 1991; Mumin et al., 1994; Oberthür et al., 1994; Allibone et al., 2002a,b; Perrouty et al., 2012). The structural and geodynamic evolution of the Kumasi Basin during the Eburnean orogeny had a significant control over gold mineralization (Eisenlohr and Hirdes, 1992; Feybesse, 1999; Allibone et al., 2002a,b, 2004; Feybesse et al., 2006). For the present study, interpretations on the ore controls and processes that led to gold mineralization in the Kumasi basin have been derived largely from Allibone et al. (2002a,b, 2004) and Feybesse et al. (2006) and knowledge derived from field work in Asankrangwa Belt along with Asanko team. The deformation events in the following discussion refer to those described by Allibone et al. (2002a,b, 2004). Two distinct gold events have been recognised in the Kumasi Basin: • An early, apparently minor, gold mineralisation event at ca. 2110 Ma and coincident with ductile deformation during D2. During this

Fig. 7. Pathways — piece-wise linear membership function plot for proximity to D2 thrust faults.

time, gold deposition was controlled by and localised along NE–SWstriking, shear zones undergoing dextral dip-slip and that are marked by the development of graphitic mylonite; • A later event that coincided with brittle–ductile deformation during D4/D5 produced auriferous quartz veins and stockworks in Birimian metasedimentary and metavolcanic (?) rocks and Eburnean granitoids.

Alternating seismic and interseismic episodes of deformation created an environment favourable for gold mineralization. The initial D1 and D2 phases of deformation resulted in formation of crustal scale structures such as basin forming faults that facilitated channelization and migration of ore bearing fluids. The third deformation, D3, was a period of gradual build-up of tectonic stresses in the basin, characterised by ductile deformation, which thereby induced circulation of fluids through the D2 thrusts under increasing stresses. Between D2 and D4 there was possibly significant uplift of the new orogen and, thus, a change in physical conditions, which may have resulted in increased seismic activity during the D4–D5 event due to reactivation of and sinistral strike-slip and dip-slip movement along D2 thrusts. Thus Kumasi Basin was transformed into a low-stress domain due to strain partitioning between faults and the rock masses, leading to the development of abundant structural “traps” such as en echelon faults, shearzones, fractures and microfracture sets, stockworks, fault gouge and breccia, and cataclasites during progressive deformations (Allibone et al., 2002a,b; Feybesse et al., 2006). Active deformation along the existing structures and associated secondary structures increased permeability in parts of the Kumasi Basin and created spaces which led to sporadic breaching of overpressured fluid reservoirs due to seismic failure, sudden fluid escape from these reservoirs along structures that were actively being deformed and subsequent physical and chemical reactions eventually triggering gold precipitation. Thus although the

Fig. 6. FIS-based prospectivity modelling flow chart.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 8. Pathways — piece-wise linear membership function plot for proximity to topographic ridges.

structural architecture of the Kumasi Basin evolved continuously throughout the Eburnean orogeny, the most significant gold mineralization event took place post peak metamorphism towards the close of Eburnean orogeny. For gold prospectivity modelling of the Kumasi Basin, a conceptual genetic model of orogenic gold mineralization in the Kumasi Basin was developed to understand and analyse various processes involved in the formation and localization of gold deposits. The indications of these processes having operated at a given location, termed targeting criteria, are inferred by their responses in the available exploration data, which are mapped to generate spatial proxies of the mineralization processes. An orogenic gold mineral system comprises three essential components, viz. source, pathways and traps (Hagemann and Cassidy, 2000; Joly et al., 2012). Traps include “fluid throttling” and “metal scrubbing” mechanisms (cf. McCuaig et al., 2010). For the prospectivity modelling of the Kumasi Basin, we considered only pathways and traps as essential components, as source is generally critical only at the craton scale (McCuaig et al., 2010). The broad spatial distribution of the known gold deposits in Ghana, and the broad similarity of their morphology, structural settings and timing direct towards the presence of a broad relatively homogenous gold source rather than local point sources such as individual plutons or batholiths as in the case of, for example, porphyry or intrusion-related gold systems. As such, the broad spatial distribution of the known gold deposits suggests that, at the regional scale, which is the scale of our prospectivity modelling, sources of energy, fluids, ligands and metals are available and likely to have a homogenous spatial distribution. Moreover source processes are commonly not well understood, not easily observable/ mappable and are cryptic. The expressions of these processes in the available datasets are difficult to identify. Hence the main controls on

9

Fig. 10. Physical traps — piece-wise linear membership function plot for proximity to D2 flexures.

the location of gold deposit formation recognised are active fluid pathways (commonly large-scale, deep-seated faults) and “traps” (commonly damage zones associated with fault jogs, bends and splays) that promote metal deposition (Cox et al., 2001; Sibson, 2001). The mineral system components and their respective spatial proxies are given in Tables 2a–2c. Key structural pathways identified for transportation of ore bearing fluids in the Kumasi basin are the D2 thrust faults. This deformation was coeval with the first minor phase of gold mineralization in the Kumasi Basin. The D2 thrust faults developed along the crustal-scale basin forming faults during basin inversion in a NW–SE directed compressional regime, although at least some of these would have been new faults that formed at the time of D2. These are long-lived, deep-seated, fundamental fault systems that were reactivated during basin inversion under compressive stress regimes in convergent tectonic settings, and they are large both with respect to their down-dip and strike extent that enables them to tap a large volume of crust. Hence these faults are interpreted to have been instrumental in facilitating the migration of fluids from source to trap regions. Thus, D2 faults are identified as the critical spatial proxy for gold transportation pathways that channelize gold bearing fluids from source to trap regions. A second spatial proxy for pathways are the topographic ridges, which stand 30 to 50 m above the average elevation of the surrounding countryside. Their resistance to weathering can often be linked to silicification and presence of hydrothermal alteration assemblages such as quartz–sericite–iron carbonate–pyrite assemblages formed due to silica flooding (Shadrach Ainoo, Asanko Gold Inc., pers. comm., 2014). Silicification is a prominent form of hydrothermal alteration brought about by the passage of hydrothermal fluids. A majority of large deposits (e.g., Esaase, Nkran, Dynamite Hill) in the Asankrangwa Belt are located on ridges. Although there are alternative explanations for topographic ridges that should be considered. As an example, many

Fig. 9. Physical traps — piece-wise linear membership function plot for density of competence contrast along lithological contacts.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

10

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 11. Chemical traps — piece-wise linear membership function plot for proximity to Fe-rich Paleoproterozoic.

topographic ridges in the Ashanti Belt are associated with the weathering of mafic rocks. In the Kumasi Basin also, a major topographic ridge between Ayanfuri and Pampe is associated with mafic rocks. Laterite (and bauxite) surface may also cause topographic ridges. However, even in cases where topographic ridges do not reflect silicification, they can represent compositional contrasts in the bedrock (e.g., between less competent and more competent lithologies) which can make them potentially prospective. Thus probable pathways can be located along these ridges, especially along NW–SE-trending topographic ridges that align along the trend of the D2 structures. Consequently the two key spatial proxies identified for pathways include the D2 thrust faults and topographic ridges aligned along the trend of the D2 structures. Gold forms complex compounds with sulphur and is transported in soluble form as Au(HS)2−, HAu(HS)02 and Au(HS)0 complexes (Pirajno, 2009). A geological condition, be it tectonic settings, structural framework, lithological facies or geochemical environment, that leads to the breaking of these complexes and precipitation of gold is a favourable trap region. Accordingly there can be physical traps and chemical traps. In the context of the Kumasi Basin, favourable physical traps are expected to be generated during its tectonic evolution. During the Eburnean orogeny the NE–SW trending D2 thrust fault were reactivated as sinistral strike-slip faults as a result of the D4–D5 deformation events. There was change in compressional regime from NW–SE to N–S during the D4–D5 event and flexures developed along the original D2 thrusts, forming pressure shadow regions. Thus as fluids migrate through the D2 thrusts, a sudden drop in pressure along these flexures may cause the precipitation of gold. Accordingly the D2 flexures serve as an important spatial proxy for physical traps. Also a drop in pressure is witnessed along the ductile–brittle transition.

As fluids flow from a ductile deformation regime, they are subjected to high pressures, but as they migrate to a brittle deformation region (through dilational jogs and fractures) the pressure drops and leads to the escape of volatiles and gold deposition (Joly et al., 2012). In the Asankrangwa Belt, this process occurs in areas of juxtaposition of granites and metasedimentary rock units. Deformation causes the incompetent metasedimentary rocks to fail by shearing but given that these rocks are very tight due to previous shortening and deformation no, or only very little, new space was created. The granite bodies, on the other hand, fail by fracturing creating low stress sites and space for gold deposition. The fracturing event triggers increased permeability and suction and thus focuses fluid flow into the newly created fractures/stockwork. So lithological contacts weighted by competence contrast density can be utilized as another spatial proxy for physical traps (Groves et al., 2000; Brown, 2002; Lindsay et al., 2015). Chemical precipitation of Au involves changes in the geochemical composition of fluids. Thus changes in redox, Eh–pH conditions, or presence of highly reactive rocks and Fe-rich rocks could cause gold deposition. Two spatial proxies identified for chemical traps include Paleoproterozoic iron-rich rock units such as dolerites and BIFs, and lithological contacts weighted by the density of contrast in their respective chemical reactivity. The chemical reactivity is estimated using the relative chemical reactivity scale provided in Brown (2002). In the Kumasi Basin however chemical scrubbing may not be a significant metal deposition mechanism. Carbonaceous shales may have played a role in metal deposition, but this role is not yet clear whether the carbon is primary (and thus would have been present at the time of the gold-bearing fluids entering this rock package) or whether it is an alteration/

Fig. 12. Chemical traps — piece-wise linear membership function plot for density of reactivity contrast along lithological contacts.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

11

Fig. 13. Piece-wise linear membership function plot for gold potential.

metasomatic feature and expression of the gold mineralising system. In the latter scenario the reactivity would not necessarily have been in place at the time of gold deposition. 5. Fuzzy inference systems (FIS)-based orogenic gold prospectivity modelling The theory of fuzzy sets (Zadeh, 1973) forms the basis of FIS. A fuzzy set is one whose membership ranges from 0 to 1. A linguistic value is the label of a fuzzy set. For example, ‘proximity’ to a fault is the label of a fuzzy set of all pixels that are ‘close’ to a fault. For a given dataset, all pixels are members of this fuzzy set, but with varying values of membership in the range [0, 1]. A fuzzy membership value defines the degree of truth of the statement that a given pixel value is ‘close to a fault’. A fuzzy membership function converts a given pixel value to a fuzzy membership value bound by 0 and 1. Formally, let X be a set whose elements are represented generically as x. A fuzzy set à in X is a set of ordered pairs: ~ ¼  x; μ ~ ðxÞjx∈X
; A A where μà is the membership function (degree of compatibility or degree of truth) of x in Ã. The fuzzy set à becomes a classical set if its membership value is restricted to either 0 or 1. An FIS comprises a set of if–then rules written in a natural language which represent the expert's inductive reasoning for predicting the state of a system based on the combination of conditions represented

Table 3 Input variables and linguistic values. Input variable (spatial proxy) Premise variables • Pathway proxies 1. Proximity to D2 thrust faults. 2. Proximity to topographic ridges aligned along D2 thrust faults. • Physical trap proxies 3. Proximity to flexures along D2 thrust faults. 4. Density estimation of competence contrast values. • Chemical trap proxies 5. Proximity to Fe-rich Paleoproterozoic rocks. 6. Density estimation of reactivity contrast values. Consequent variables • Gold potential

Linguistic values Proximal, intermediate, distal Proximal, intermediate, distal

Proximal, average, distal High, average, low

Proximal, average, distal High, average, low

Very high, high, average, low, very low

in terms of linguistic variables (Porwal et al., 2014). An FIS can be used to capture an exploration geologist's deductive logic for predicting the mineral potential from a combination of linguistic predictor variables. Each of the linguistic values in an if–then rule is the label of a fuzzy set whose membership for the relevant data is estimated using a predefined membership function. The data are examples of spatial proxies (see McCuaig et al., 2010, for a detailed discussion). Several such fuzzy if–then rules that predict the mineral potential in linguistic terms (e.g., high/moderate/low) for different combinations of the linguistic values of the input spatial proxies are compiled to form an FIS. The outputs of individual rules are fuzzy values that are combined using an appropriate fuzzy aggregation technique to derive the output of the FIS. However, this output is a fuzzy area under the curve which is defuzzified to extract a single output. This output is a measure of the mineral potential in terms of a single number between 0 and 1. The following steps are involved in the implementation of an FIS (drawn from Porwal et al., 2014). • Generate a conceptual model of the targeted mineral systems and identify the targeting criteria. Process exploration datasets using GIS tools to create appropriate spatial proxies (or predictor maps) for the targeting criteria. The spatial proxies are the input variables to the FIS. The output variables are mineral-system-component potential, namely, source potential, pathway potential and trap potential. Optionally, energy and ligand source-potential can be added, depending on the targeted mineral system and the availability of relevant spatial proxies. • Select the linguistic values for each input spatial proxy. For example, Fe content is a proxy for chemical scrubber for gold precipitation in orogenic gold systems. The possible linguistic values for Fe can be {very high, high, average, low, very low}. The qualifier ‘very’ is termed a hedge (Zadeh, 1972). Similarly select the linguistic values for each of the output mineral-system-component potential (‘high source potential’, ‘low trap potential’, etc.). • Define a membership function for translating the numerical or categorical values of each input spatial proxy into fuzzy values. • Build up an FIS for each mineral system component, namely, energy source, metal source, ligand source, active pathway and physical and chemical traps. The FIS should capture the reasoning of an exploration geologist, and therefore should capture all realistically possible combinations of linguistic values of the proxies. The choice of the coordinating operator (conjunction or disjunction) between two input spatial proxies is non-trivial and should be made only after a careful consideration of the geological relations including conditional dependencies between the two with respect to the targeted mineral system. • Calculate the firing strength of each rule. The firing strength of a rule is the combined fuzzy membership value of the if-part of the rule. The fuzzy membership values of the input spatial proxies in

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

12

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Table 4 FIS rules. Premise (IF) part • Pathways 01. IF 02. IF 03. IF 04. IF 05. IF 06. IF 07. IF 08. IF 09. IF

Consequent (THEN) part

D2 thrust is proximal D2 thrust is intermediate D2 thrust is distal D2 thrust is distal D2 thrust is intermediate D2 thrust is intermediate D2 thrust is distal D2 thrust is proximal D2 thrust is proximal

• Physical traps 10. IF 11. IF 12. IF 13. IF 14. IF 15. IF 16. IF 17. IF 18. IF

Competence contrast density is high Competence contrast density is low Competence contrast density is average Competence contrast density is high Competence contrast density is low Competence contrast density is high Competence contrast density is average Competence contrast density is low Competence contrast density is average

• Chemical traps 19. IF Fe-rich rocks are distal 20. IF Fe-rich rocks are proximal 21. IF Fe-rich rocks are average 22. IF Fe-rich rocks are proximal 23. IF Fe-rich rocks are distal 24. IF Fe-rich rocks are average 25. IF Fe-rich rocks are proximal 26. IF Fe-rich rocks are average 27. IF Fe-rich rocks are distal

AND AND AND AND AND AND AND AND AND

Topographic ridge is proximal Topographic ridge is intermediate Topographic ridge is distal Topographic ridge is proximal Topographic ridge is proximal Topographic ridge is distal Topographic ridge is intermediate Topographic ridge is distal Topographic ridge is intermediate

THEN THEN THEN THEN THEN THEN THEN THEN THEN

Pathway prospectivity is very high Pathway prospectivity is average Pathway prospectivity is very low Pathway prospectivity is average Pathway prospectivity is high Pathway prospectivity is average Pathway prospectivity is low Pathway prospectivity is high Pathway prospectivity is very high

AND AND AND AND AND AND AND AND AND

D2 flexures are proximal D2 flexures are distal D2 flexures are average D2 flexures are distal D2 flexures are proximal D2 flexures are average D2 flexures are proximal D2 flexures are average D2 flexures are distal

THEN THEN THEN THEN THEN THEN THEN THEN THEN

Physical trap prospectivity is very high Physical trap prospectivity is very low Physical trap prospectivity is average Physical trap prospectivity is average Physical trap prospectivity is high Physical trap prospectivity is high Physical trap prospectivity is high Physical trap prospectivity is average Physical trap prospectivity is low

AND AND AND AND AND AND AND AND AND

Reactivity contrast density is low Reactivity contrast density is high Reactivity contrast density is average Reactivity contrast density is low Reactivity contrast density is average Reactivity contrast density is low Reactivity contrast density is average Reactivity contrast density is high Reactivity contrast density is high

THEN THEN THEN THEN THEN THEN THEN THEN THEN

Chemical trap prospectivity is very low Chemical trap prospectivity is very high Chemical trap prospectivity is average Chemical trap prospectivity is average Chemical trap prospectivity is average Chemical trap prospectivity is low Chemical trap prospectivity is high Chemical trap prospectivity is high Chemical trap prospectivity is average

the if-part can be calculated using fuzzy conjunction or fuzzy disjunction. Fuzzy conjunction (or t-norm) operators such as AND and Fuzzy Algebraic Product (= ∏ ni = 1μ i , where μ is the fuzzy membership value and n is the number of input variables)

implement generalize intersection and return the minimum and product, respectively, of the input fuzzy membership values. On the other hand, fuzzy disjunction (t-conorm or s-norm) operators generalize union and return output maximum or Algebraic Sum

Fig. 14. Pathways FIS rules: the fuzzy inference system used for prospectivity modelling of orogenic gold in the Kumasi Basin. An example for a pixel with values [550, 550] on the two predictor maps for pathways: (i) distance to D2 thrust faults, (ii) distance to topographic ridges, respectively, is shown. The values are marked by red lines. The yellow colour indicates the activated fuzzy membership values for each function (the fuzzy functions are blank if the activated fuzzy membership value for a given input value is 0). In the output consequent membership function, the blue part indicates the minimum implication, that is, the remaining area of an output consequent function after it has been truncated by the firing strength (= minimum ∗ (And Operator) of the all fuzzy membership values in the IF part) of the rule. The last function in the out prospectivity column shows the aggregate of all output fuzzy functions. The red line on this function indicates the defuzzified crisp value which indicates the possibility of occurrence of pathways in the given pixel. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

13

Fig. 15. Physical traps FIS rules (for the explanation of colours refer to the caption of Fig. 14). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(= 1 − ∏ni = 1(1 − μi), where μ is the fuzzy membership value and n is the number of input variables) of the input fuzzy membership values. • The firing strength of each fuzzy-if then rule is used to scale down the fuzzy output of the rule. There are various methods for scaling down the output such as Min–Max or Max-Product. In the former method, the output fuzzy set defined by the then-membership function is truncated to level of the firing strength. In the latter method, the then-membership function is multiplied by the firing strength, and hence is scaled down in proportion to the firing strength (see the explanation in Porwal et al., 2014, Fig. 1). In both cases the output is a fuzzy set represented by an area under curve and not a fuzzy membership value represented by a single point.

• To determine the final output of an FIS, the areas-under-curve produced by individual rules are aggregated. Again there are several methods available; the most commonly used method is Max. In this method, the output is the union of all areas (see the explanation in Porwal et al., 2014, Fig. 1). Finally the aggregate area is converted into a point value. This procedure is called defuzzification. The point value can be estimated using a number of defuzzification procedures, the most common being the centre of gravity of the aggregated area. The output point value of an FIS gives the potential of a specific mineral system component. • The fundamental basis of the mineral systems approach is that all essential processes must occur to form a deposit, and therefore,

Fig. 16. Chemical traps FIS rules (for the explanation of colours refer to the caption of Fig. 14). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

14

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 17. Pathway prospectivity map.

the output prospectivity can be estimated using a conjunction operator such as Algebraic Product or AND as follows: Prospectivity ¼ MINðSource potential; Pathway potential; Trap potentialÞ; or Prospectivity ¼ Source potential  Pathway potential  Trap potential:

The reader is referred to Porwal et al. (2014) for further, more detailed theoretical background to and details of the algorithms and procedures used in FIS-based prospectivity modelling. The modular approach to prospectivity modelling described by Porwal et al. (2014) involves evaluating every aspect of each of the orogenic-gold-mineralsystem components and analysing its influence on localization of gold mineralization. A two-stage FIS was used (Fig. 6). In the first stage individual FIS were designed for each of key component, namely, pathways, physical traps and chemical traps. In the second stage the outputs of each of the above FIS were combined using a conjunction operator. The predictor maps used to define the input variables for the individual FIS have been tabulated in Tables 2a–2c. We used piece-wise linear functions to assign fuzzy membership values to the input and the output variables and compute their influence on mineralization (Figs. 7–13). The functions were selected after careful consideration of their ability to capture the perception of the exploration geologist

Fig. 18. Physical trap prospectivity map.

about the decay of influence of the geological features with distance or other values. For example, based on field observations and discussion with the Asanko Gold Inc.'s geologists, we interpret a buffer of 100 m around a D2 thrust as a zone where the influence of the thrust is expected to be definitely present. From 100 m to 500 m, we interpret that the influence decays linearly, and beyond 500 m, we do not expect the thrust to have any influence. This rationale is captured through the piece-wise linear fuzzy membership function for the linguistic value “proximal” (Fig. 7, curve a). The linguistic values used for defining the membership functions are given in Table 3. Based on generalized geological settings of orogenic gold mineralization and the controls of the geology, tectonic settings, deformation events and crustal scale structures on mineralization of gold in the Kumasi basin, several ‘if–then’ rules were generated for analysing the prospectivity of each ground unit area (cell size-100 m2) for gold mineralization (Table 4 and Figs. 14, 15, 16). The rules were incorporated in the fuzzy inference system to predict the chances of the ore forming processes having occurred in the ground unit area and thereby assign prospectivity values on a fuzzy scale from zero to one, where zero is non-prospective and one is prospective. The rules used for designing the FIS are given in Table 4. The ‘if’ part of the rules evaluates the values of the input variables and the potential for each component is generated in the ‘then’ part. The MIN (AND) operator was used to combine the ‘if–then’ rules and the final prospectivity value for each component was obtained by calculating the centroid of the combination of the ‘then’ part of the rules. That generated three intermediate predictor maps showing the prospectivity for each of the mineral system component, viz. pathways, physical traps and chemical traps (Figs. 17, 18, 19).

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 20. Trap prospectivity map.

Fig. 19. Chemical trap prospectivity map.

The next step involved integration of all three fuzzy inference systems to obtain the final mineral prospectivity map. As for traps, the presence of either a physical or a chemical trap is sufficient to facilitate mineralization. Hence the output maps of physical and chemical traps were combined using an ‘OR’ operator to generate a single map representing prospectivity values with respect to traps (Fig. 20). Lastly the prospectivity maps of pathways and traps were combined using the ‘PRODUCT’ operator to produce the final fuzzy prospectivity map for orogenic gold in the study area (Fig. 21).

6. Results and validation We used a cumulutive area fuzzy favourability (CAFF) plot (Fig. 22) to separate the prospective and non-prospective areas. The slopes of different segments of a CAFF plot represent the fractal dimensions of different spatial populations present in a given area (Porwal et al., 2003). If the entire study area comprises of a single population then the CAFF plot is expected to be a straight line with a constant slope. However, if there are two populations in a given area, namely, mineralized and barren, then the CAFF plot is expected to be characterised by two segments with different slopes. The inflexion points at which the slopes change indicate changes in the fractal dimensions, and are used as thresholds for differentiating areas with different prospectivity values and to reclassify the continuous-scale prospectivity model into a binary or ternary model. In the present study, the CAFF plot (Fig. 22) shows inflexion points at prospectivity values of approximately 0.2 and 0.5, indicating presence of three populations. These are classified as high prospectivity, moderate prospectivity and low prospectivity populations.

15

In the ternary prospectivity map, only 7% of the total area is identified as highly prospective, 24% is moderately prospective whilst most of the area (i.e., 69%) falls in the low prospectivity domain (Fig. 21). The performance of the model has been evaluated using the 1159 known gold occurrences and deposits in the Kumasi Basin including 40 in-situ and 1119 alluvial occurrences. The model efficiently captures 77.5% (31) of the in-situ deposits and occurrences in the high prospectivity and moderate prospectivity areas (Table 5). If we consider all deposits and occurrences, then the model captures 99% (1148 deposits) in the high prospectivity and moderate prospectivity areas (Fig. 21, Table 5). There are a total of 32 deposits with estimated gold endowment. Out of those, 78% (25) deposits are captured in high prospectivity and moderate prospectivity areas. 7. Discussion The method for gold prospectivity modelling presented in this paper and applied to the Kumasi Basin consists of the following steps: (1) formulation of a genetic model conceptualizing the processes of orogenic gold mineralization in the study area (2) identification of key controls on the location of gold mineralization, and (3) discerning spatial proxies of these processes and controls to develop an FIS. The resulting prospectivity model captures the possibility of the occurrence of gold mineralising processes having occurred, rather than merely focussing on finding analogues of the existing deposits, as discussed in more detail below. Prospectivity modelling is commonly performed using either datadriven or knowledge-driven approaches. Data-driven modelling (e.g., weights-of-evidence) is governed by spatial and statistical analyses of characteristic geological features of the known deposits and

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

16

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 21. Orogenic gold prospectivity map with an overlay of all known deposits (both insitu and alluvial).

recognising these features in a predetermined area based on the assumption that the deposits that have already been discovered in an area are adequate samples of the outputs of the targeted mineral systems. Since these models rely on the spatial statistical association of known deposits with geological features to estimate prospectivity, the emphasis is on identifying geological features associated with the known mineral deposits. A knowledge-driven approach like an FIS, focuses on capturing the processes of mineralization rather than the features associated with mineral deposits. Prospectivity modelling is implemented without any

Fig. 22. Cumulative area fuzzy favourability plot (CAFF).

regard of the spatial distribution of the known mineral deposits and thus is not likely to be biased by the attributes of already discovered deposits. In FIS modelling, the emphasis is on predicting the occurrence of processes and components that constitute a particular mineral system and eventually produce a mineral deposit. So, if the processes and the components driving a mineral system have been correctly represented and incorporated in the FIS, then it should be able to predict the known deposits very well. The FIS approach models mineralisation processes in terms of their possible expressions in various exploration data sets. These expressions are identified as spatial proxies of mineralization processes. In the FIS approach, the spatial proxies are combined using either ‘AND’ or ‘OR’ operators depending on whether all the processes manifested by these spatial proxies are necessary or whether a certain combination can suffice for mineralization. The above operators also handle the conditional dependencies amongst the spatial proxies. For instance, in the current study we considered topographic ridges, a pathway proxy, to be an expression of silicification induced by gold-related hydrothermal alteration. However, such ridges can also represent felsic intrusive bodies that can be equally resistant to weathering and are known to form ridges. To differentiate between topographic ridges representing felsic intrusions and those representing silicification, we used an ‘AND’ operator to conjunct topographic ridges with D2 thrust faults in the FIS. The capture efficiency of an FIS model can be validated using the spatial distribution of known deposits because the same is not used for estimating the model parameters. This feature gives FIS an edge over datadriven approaches to mineral prospectivity modelling. In data-driven modelling, the model parameters are estimated based on the spatial distribution of the known deposits, assuming that the known deposits used to train the model are a representative sample of the targeted mineral system. This may not always be true. As a result, it is often difficult to verify if a data-driven model has good generalization capability, or if it is overfitting to the known deposits, even though hold-back validation techniques are used. On the other hand an FIS does not use the spatial statistical correlations of the known deposits, and thus less likely to overfit the prospectivity model to the known mineral deposits. An FIS-derived prospectivity model can also be evaluated for its effectiveness in identifying prospective areas and controls on mineralization. Orogenic gold deposits in the Kumasi Basin are products of the tectonic evolution of the basin and its regional structural grain. According to the genetic model, gold mineralization is expected to display a strong NE–SW control exerted by D2 thrust faults, a relationship that is also illustrated by the prospectivity map. Hence, the FIS can identify trends of key controls on gold mineralization and classifies these as either prospective or non-prospective. Further this paper presents a new approach of using an FIS to map the different components of an orogenic gold mineral system and combining their outputs to generate the final prospectivity map. This approach allows the model to capture and identify prospective areas based on the input layers derived from the processes that control mineralization. Since the prospectivity map picks up the trends defined in the conceptual model, the existing mineral deposits are used for the purpose of validation of this approach and of the prospectivity map. The aspects evaluated for validation are the capture efficiency of the model and trends of mineralization shown by the deposits. The model has very high capture efficiency as it recognises 99% of the known gold occurrences in the high and moderate prospectivity zones that collectively make up 31% of the study area (Table 5). There is a strong spatial association between the known gold deposits and the metallogenic trends as identified by the FIS model. These trends were identified using Fry analysis (Fry, 1979), or spatial autocorrelation of point data. This technique employs a method of recording the distance and bearing from each point of a particular dataset to every other point. For n points there are n2–n spatial relationships (i.e., translations) that can be illustrated in graphs (i.e., Fry or translations plots) and rose diagrams (Fry, 1979; Kreuzer et al., 2007). A Fry plot is a means of analysing spatial distribution and association of

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

17

Table 5 Capture efficiency of the FIS model. Total no. of gold occurrences

In situ and alluvial In situ In situ deposits with endowment

1159 40 32

High prospectivity areas 15,995.2 km2 (7%)

Moderate prospectivity areas 50,983.6 km2 (24%)

Low prospectivity areas 147,175.5 km2 (69%)

No. of deposits (%)

No. of deposits (%)

No. of deposits (%)

823 (71%) 19 (47.5%) 17 (53.13%)

325 (28%) 12 (30%) 8 (25%)

12 (1%) 9 (22.5%) 7 (21.87%)

mineral deposits with one another and of identifying major trends in their distribution. The Fry plot shows two dominant orientations, namely, NNE–SSW and NE–SW (Fig. 23). The NE–SW trend can be related to the D2 thrust faults, which seem to exhibit major influence on mineralization in the Asankrangwa Gold Belt as inferred from geological mapping, drill core logging and 3D modelling of drillhole assay data. However, NNE–SSW trend is rather cryptic and does not have welldefined surface expression. There does exist an alignment of highergrade and larger tonnage gold deposits along this orientation. Moreover, many post-gold mineralization (that is, post-Proterozoic) dyke swarms that generally strike NW–SE but individual dykes within these swarms have a well-defined NNE–SSW orientation, commonly only over short (up to 10 km long) sections, in particular where they cross cut the Asankrangwa Gold Belt. Here they switch orientation from NNW–SSE to NNE–SSW, although there is one such dyke further north that strikes NNE–SSW and that is at least 80 km long. Although post-mineralisation in age, these dykes could follow deep-seated, mantle-tapping structures. Further, the boundary between the Kumasi Basin and Sefwi Belt

that, over a strike length of ca. 45 km, has a well-defined NNE–SSW orientation — the section that hosts Bibiani and Chirano gold mines. It is worth noting that certain segments of the NW–SE-striking D2 shear zones of the Asankrangwa Gold Belt often form bends that strike NNE–SSW. Many of the topographic ridges have a well-defined NNE– SSW orientation both over short and very long (N100 km) distances. It is likely that the NNE–SSW trend is the orientation of yet-unmapped mantle-tapping structures of Paleoproterozoic age, or older, that could have played a critical role in gold mineralization in the study area. The orientation analysis shows that, at a regional scale (segment lengths 0–300 km), the dominant trends are along NNE–SSW, NE–SW and N–S, whilst there are minor spikes along E–W (Fig. 24A). At smaller scales of camp- to district (segment lengths 0–50 km), the trend becomes more prominently NE–SW, although significant spikes persist along NNE–SSW (Fig. 24B). However, new trends along NNW–SSE and NW–SE appear at this scale and become prominent (Fig. 24B). At the deposit-scale (segment lengths 0–10 km), predominant spikes occur along NNW–SSE and E–W directions, whilst there are subsidiary spikes

Fig. 23. Fry translations of in-situ gold deposits.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

18

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Fig. 24. Orientation diagrams of the Fry translations shown in Fig. 23: (A) regional scale (0–300 km), (B) camp to district scale (0–50 km), (C) deposit to camp scale (0–10 km).

along NE–SW and NW–SE (Fig. 24C). The NNE–SSW trend still persists but becomes insignificant (Fig. 24C). The prominent NNE–SSW trend at the regional scale can possibly relate to ancient crustal-scale structures that could have formed mantletapping plumbing systems for source processes. The NE–SW trend is most likely related to the D2 thrust faults, which seem to exhibit major influence on mineralization. The N–S trending spike is again cryptic — there is no well-defined expression of this trend on the surface. However, like NNE–SSW, this could represent a crustal feature related to gold mineralization that runs through the entire belt. The trends at the camp-to-district scales are noisy, but the major spike along NE–SW persists, with additional spikes appearing along NW–SE and NNW–SSE that can be associated to the flexures caused due to sinistral strike-slip faults generated by reactivation of the D2 thrusts during the D4 deformation event. At the deposit-scale, the NNW–SSE trend becomes predominant, with significant spikes along E–W. This highlights the possible influences of cross faults, splays, jogs and bends along D2 thrust faults associated to the D4 deformation event becomes more discernible as evidenced by the trends along NW–SE and approximately at NNW– SSE directions. The analyses therefore indicate that, at the scale of the concession area, the NW–SE- and NNW–SSE-trended structures are likely to be more prospective. The above discussion shows that the Fry analysis and rose diagrams derived for the existing deposits point towards the structural trends identified in the genetic model, and hence the conceptual genetic model so formulated holds grounds in terms of recognising key characteristics of mineralization and incorporating it in the prospectivity map. 8. Summary and conclusions • This paper presents new insights into the geology of, and controls on orogenic gold deposits in, the Kumasi Basin. • A two-stage Mamdani-type fuzzy inference system is used to model the gold prospectivity of the Kumasi Basin. The output model captures about 99% of all known gold deposits and occurrences, including in-situ and alluvial, within high-prospectivity and moderate-prospectivity areas that occupy 31% of the total study area. Further, 77.5% of the in-situ orogenic gold deposits are captured in high prospectivity and moderate prospectivity areas.

• The high- and moderate-prospectivity areas in the prospectivity map are elongate features that are spatially coincident with areas of structural complexity along and reactivation during D4 of NE– SW-striking D2 thrust faults and subsidiary structures, implying a strong structural control on gold mineralization in the Kumasi Basin. New exploration targets have been identified and are being investigated by Asanko Gold Inc.

Two new cryptic regional metallogenic trends, namely, NNE–SSW and N–S, have been identified based on Fry analysis, these trends do not have well-defined surface expression and need further investigations using high resolution gravity, magnetics and seismic data sets. Conflict of interest None. Acknowledgements Useful discussions with Shadrach Ainoo (Exploration Manager Ghana at Asanko Gold Inc.) are gratefully acknowledged. Insightful comments and suggestions by Stephane Perrouty and an anonymous reviewer helped to improve the manuscript significantly. References Adadey, K., Clarke, B., Théveniaut, H., Urien, P., Delor, C., Roig, J.Y., Feybesse, J.L., 2009, Geological map explanation — map sheet 0503 B (1:100 000), CGS/BRGM/Geoman, Geological Survey Department of Ghana (GSD). No MSSP/2005/GSD/5a. Allibone, A., McCuaig, T.C., Harris, D., Etheridge, M., Munroe, S., Byrne, D., 2002a. Structural controls on gold mineralization at the Ashanti Gold Deposit, Obuasi, Ghana. Soc. Econ. Geol. Spec. Publ. 9, 65–93. Allibone, A., Teasdale, J., Cameron, G., Etheridge, M., Uttley, P., Soboh, A., Appiah-Kubi, J., Adanu, A., Arthur, R., Mamphey, J., Odoom, B., Zuta, J., Tsikata, A., Pataye, F., Famiyeh, S., Lamb, E., 2002b. Timing and structural controls on gold mineralization at the Bogoso Gold Mine, Ghana, West Africa. Econ. Geol. 97, 949–969. Allibone, A., Hayden, P., Cameron, G., Duku, F., 2004. Paleoproterozoic gold deposits hosted by albite- and carbonate-altered tonalite in the Chirano District, Ghana, West Africa. Bull. Soc. Econ. Geol. 99 (3), 479–497. AngloGold Ashanti, 2014. Integrated Report. Available online at: http://www. anglogoldashanti.com/en/Media/Reports/Annual%20Reports/AGA-IR14.pdf. Accessed on 12 Sept 2015. Appiah, H., 1991. Geology and mine exploration trends of Prestea goldfields, Ghana. J. Afr. Earth Sci. (Middle East) 13, 235–241.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx Attoh, K., Corfu, F., Nude, P.M., 2007. U–Pb zircon age of deformed carbonatite and alkaline rocks in the Pan-African Dahomeyide suture zone, West Africa. Precambrian Res. 155 (3), 251–260. Béziat, D., Bourges, F., Debat, P., Lompo, M., Martin, F., Tollon, F., 2000. A Paleoproterozoic ultramafic–mafic assemblage and associated volcanic rocks of the Boromo greenstone belt: fractionates originating from island-arc volcanic activity in the West African craton. Precambrian Res. 101 (1), 25–47. Blenkinsop, T.G., Schmidt Mumm, A., Kumi, R., Sangmor, S., 1994. Structural geology of the Ashanti Gold Mine. Geol. Jahrb. D 100, 131–153. Boher, M., Abouchami, W., Michard, A., Albarède, F., Arndt, N.T., 1992. Crustal growth in West Africa at 2.1 Ga. J. Geophys. Res. B Solid Earth Planets 97, 345–369. Bonhomme, M., 1962. Contribution à l'étude géochronologique de la plate-forme de l'Ouest Africain, Annales de la Faculté des Sciences. Géologie, Minéralogie 5. Université de Clermont-Ferrand (62 pp.). Bossière, G., Bonkoungou, I., Peucat, J.J., Pupin, J.P., 1996. Origin of Paleoproterozoic conglomerates and sandstones of the Tarkwaian Group in Burkina Faso, West Africa. Precambrian Res. 80, 153–172. Brown, W., 2002. Artificial Neural Networks: A New Method for Mineral-prospectivity Mapping (PhD Thesis) University of Western Australia. Brown, D., 2015. Gold mining in Ghana. Gold Investing News (www.goldinvestingnews. com (accessed on 15 June 2015)). Castellana, C.H., Long, L., 1998. Origin of the Coronel João Sá pluton: implications for the tectonic evolution of the NE Brazil. Abstr. — Geol. Soc. Am. Abstr. 30 (7), 236. Cox, S.F., Knacksted, M.A., Braun, J., 2001. Principles of structural controls on permeability and fluid flow in hydrothermal systems. SGA Rev. 14, 1–24. Davis, D.W., Hirdes, W., Schaltegger, E., Nunoo, E.A., 1994. U/Pb age constraints on deposition and provenance of Birimian and gold-bearing Tarkwaian sediments in Ghana, West Africa. Precambrian Res. 67, 89–107. Diallo, D.P., 2001. Le paléovolcanisme de la bordure occidentale de la boutonniere de Kédougou, Paléoprotérozoïque du Sénégal oriental: Incidences géotectoniques. J. Afr. Earth Sci. 32, 919–940. Dioh, E., Béziat, D., Debat, P., Grégoire, M., Ngom, P.M., 2006. Diversity of the Palaeoproterozoic granitoids of the Kédougou inlier (eastern Sénégal): petrographical and geochemical constraints. J. Afr. Earth Sci. 44 (3), 351–371. Doumbia, S., Pouclet, A., Kouamelan, A., Peucat, J.P., Vidal, M., Delor, C., 1998. Petrogenesis of juvenile-type Birimian (Palaeoproterozoic) granitoids in Central Côted'Ivoire, West-Africa: geochemistry and geochronology. Precambrian Res. 87, 33–63. Duodu, A., J., Loh, G.K., Boamah, K.O., Baba, M., Hirdes, W., Toloczyki, M., Davis, D.W., 2009. Geological map of Ghana 1:1000000. Geological Survey Department of Ghana (GSD). Eisenlohr, B.H., 1989. The structural geology of Birimian and Tarkwaian rocks in Ghana. Bundesanstalt fuer Geowissenschaften und Rohstoffe, Hannover, Report Number 106448 (66 pp.). Eisenlohr, B.N., Hirdes, W., 1992. The structural development of the early Proterozoic Birimian and Tarkwaian rocks of southwest Ghana, West Africa. J. Afr. Earth Sci. 14, 313–325. Fetter, A.H., Van Schmus, W.R., Sariva Dos Santos, T., Arthaud, M., Nogueira Neto, J., 1997. Geologic history and framework of Ceara state: NW Borborema province, NE Brazil. Geol. Soc. Am. Abstr. 29 (6), 409. Feybesse, J.L., 1999. Tectonic controls and geometry of the hydrothermally altered and mineralized ore bodies of the Rank area E, Kenyase, Bosumkese, Yamfo trend, Teekyere west, and Subenso prospects. BRGM report 99/076, p. 18. Feybesse, J.L., Billa, M., Milesi, J.P., Lerouge, C., Le Goff, E., 2000. Relationships between metamorphism–deformation–plutonism in the Archean–paleoproterozoic contact zone of West Africa. 31st International Geological Congress, Rio. CD-ROM, Abstract. Feybesse, J.L., Billa, M., Guerrot, C., Duguey, E., Lescuyer, J.L., Milési, J.P., Bouchot, V., 2006. The Paleoproterozoic Ghanaian province: geodynamic model and ore controls, including regional stress modeling. Precambrian Res. 149, 149–196. Fry, N., 1979. Random point distributions and strain measurement in rocks. Tectonophysics 60 (1), 89–105. Ganne, J., De Andrade, V., Weinberg, R., Vidal, O., Dubacq, B., Kagambega, N., Naba, S., Baratoux, L., Jessell, M.W., Allibon, J., 2012. Modern-style plate subduction preserved in the Palaeoproterozoic West African Craton. Nat. Geosci. http://dx.doi.org/10.1038/ ngeo1321. Gasquet, D., Barbey, P., Adou, M., Paquette, J.L., 2003. Structure, Sr–Nd isotope geochemistry and zircon U–Pb geochronology of the granitoids of the Dabakala area (Côte d'Ivoire): evidence for a 2.3 Ga crustal growth event in the Palaeoproterozoic of West Africa? Precambrian Res. 127 (4), 329–354. Goldfarb, R.J., Groves, D.I., Gardoll, S., 2001. Orogenic gold and geologic time: a global synthesis. Ore Geol. Rev. 18, 1–75. Gold Fields Limited, 2014. Mineral resource and mineral reserve supplement to the integrated annual report for the year ended 31 December 2014. Available online at: https://www.goldfields.com/reports/annual_report_2014/pdf/full-mineral. pdf Last Accessed on 12 Sept 2015, Gold Fields Limited, 2015. Gold Fields West Africa. (Author: Baku, A.) Available online at: https://www.goldfields.com/pdf/ presentations/2015/west-africa-overview.pdf [last accessed: 23 June 2015]. Groves, D.I., Goldfarb, R.J., Knox-Robinson, C.M., Ojala, J., Gardoll, S., Yun, G.Y., Holyland, P., 2000. Late-kinematic timing of orogenic gold deposits and significance for computer based exploration techniques with emphasis on the Yilgarn Block, Western Australia. Ore Geol. Rev. 17 (1–2), 1–38. Hagemann, S.G., Cassidy, K.F., 2000. Archean orogenic lode gold deposits. Rev. Econ. Geol. 13, 9–68. Harcouet, V., Guillou-Frottier, L., Bonneville, A., Feybesse, J.-L., 2005. Pre-mineralization thermal evolution of the Paleoproterozoic gold-rich Ashanti belt, Ghana. In: McDonald, et al. (Eds.), Mineral Deposits and Earth Evolution. Geological Society, London, Spec. Pub vol. 248, pp. 103–118.

19

Hirdes, W., Davis, D.W., 1998. First U–Pb zircon age of extrusive volcanism in the Birimian Supergroup of Ghana/West Africa. J. Afr. Earth Sci. 27, 291–294. Hirdes, W., Davis, D.W., 2002. U–Pb Geochronology of Paleoproterozoic rocks in the Southern part of the Kedougou–Kéniéba Inlier, Senegal, West Africa: evidence for diachronous accretionary development of the eburnean province. Precambrian Res. 118, 83–99. Hirdes, W., Davis, D.W., Eisenlohr, B.N., 1992. Reassessment of Proterozoic granitoid ages in Ghana on the basis of U–Pb zircon and monazite dating. Precambrian Res. 56, 89–92. Hirdes, W., Leube, A., 1989. On gold mineralisation of the Proterozoic Birimian Supergroup in Ghana/West Africa. Ghanaian-German Mineral Prospecting Project, Technical Cooperation Project 80 2046-6. Hirdes, W., Davis, D.W., Ludtke, G., Konan, G., 1996. Two generations of Birimian (paleoproterozoic) volcanic belts in northeastern Cote d'Ivoire (West Africa), as demonstrated by precise U–Pb mineral dating: consequences for the “Birimian controversy”. Precambrian Res. 80, 173–191. Jessell, M.W., Amponsah, P.O., Baratoux, L., Asiedu, D.K., Loh, G.K., Ganne, J., 2012. Crustalscale transcurrent shearing in the Paleoproterozoic Sefwi–Sunyani–Comoé region, West Africa. Precambrian Res. 212, 155–168. John, T., Klemd, R., Hirdes, W., Loh, G., 1999. The metamorphic evolution of the paleoproterozoic (Birimian) volcanic Ashanti belt (Ghana, West Africa). Precambrian Res. 98 (1), 11–30. Joly, A., Porwal, A., McCuaig, T.C., 2012. Exploration targeting for orogenic gold deposits in the Granites–Tanami Orogen: mineral system analysis, targeting model and prospectivity analysis. Ore Geol. Rev. 48, 349–383. Junner, N.R., 1932. The geology of the Obuassi Goldfield. Gold Coast Geol. Surv. Mem. 2, 41. Junner, N.R., 1935. Gold in the Gold Coast. Gold Coast Geol. Surv. Mem. 4 (76 pp.). Junner, N.R., 1940. Geology of the Gold Coast and western Togoland. Gold Coast Geol. Surv. Bull. 11, 40. Klemd, R., Schröter, F.C., Will, T.M., Gao, J., 2002. PT-evolution of glaucophane–omphacite bearing HP–LT rocks in the western Tianshan orogen, NW China: new evidence for ‘Alpine-type’ tectonics. J. Metamorph. Geol. 20, 239–254. Kosuch, M., Van Schmus, W.R., Brito Neves, B.B., Bretas Bittar, S.M., 1997. 2.1 Ga Transamazonian crust as a source for more recent melt events in northeast Brazil: 1 Ga and 0. 75 Ga. Geol. Soc. Am. Abstr. 29 (6), 412. Kreuzer, O.P., Blenkinsop, T.G., Morrison, R.J., Peters, S.G., 2007. Ore controls in the Charters Towers goldfield, NE Australia: constraints from geological, geophysical and numerical analyses. Ore Geol. Rev. 32 (1), 37–80. Le Mignot, E., Siebenaller, L., Béziat, D., Salvi, S., André-Mayer, A., Reisberg, L., Velasquez, G., Zimmernann, C., Franceschi, G., 2014. The Paleoproterozoic copper–gold deposit of Gaoua, Burkina Faso: evidence for a polyphased mineralization. Acta Geol. Sin. (Engl. Ed.) 88 (Suppl. 2), 970–972. Leube, A., Hirdes, W., Mauer, R., Kesse, G.O., 1990. The early Proterozoic Birimian Supergroup of Ghana and some aspects of its associated gold mineralization. Precambrian Res. 46, 139–165. Lindsay, M., Aitken, A., Ford, A., Dentith, M., Hollis, J., and Tyler, I. (2015) Reducing subjectivity in multi-commodity mineral prospectivity analyses: modelling the west Kimberley, Australia: Ore Geol. Rev. (in press). Loh, G., Hirdes, W., 1996. Geological map of southwest Ghana, 1:100000, sheets Axim and Sekondi. Ghana Geological Survey Bulletin 49, 63. Loh, G., Hirdes, W., 1999. Explanatory notes for the geological map of Southwest Ghana 1: 100,000: Sekondi (0402A) and Axim (0403B) sheets. Geol. Jb. B 93, 149. Loh, G., Hirdes, W., Anani, C., Davis, D.W., Vetter, U., 2010. Explanatory notes for the geological map of southwest Ghana 1:100,000, Geol. Jb. B 93, 150. Lompo, Martin. “Paleoproterozoic structural evolution of the Man-Leo Shield (West Africa). Key structures for vertical to transcurrent tectonics.”. J. Afr. Earth Sci. 58.1, 19–36. Lompo, M., 2010. Paleoproterozoic structural evolution of the Man-Leo Shield (West Africa). Key structures for vertical to transcurrent tectonics. Journal of African Earth Sciences 58, 19–36. McCuaig, T.C., Beresford, S., Hronsky, J.M.A., 2010. Translating the mineral systems approach into an effective exploration targeting system. Ore Geol. Rev. 38 (3), 128–138. Milési, J.P., Ledru, P., Ankrah, P., Johan, V., Marcoux, E., Vinchon, Ch., 1991. The metallogenic relationship between Birimian and Tarkwaian gold deposits in Ghana. Miner. Deposita 26, 228–238. Milési, J.P., Ledru, P., Feybesse, J.L., Dommanget, A., Marcoux, E., 1992. Early Proterozoic ore deposits and tectonics of the Birimian orogenic belt, West Africa. Precambrian Res. 58, 305–344. Miller, A.V.M., Partington, G.A., Kreuzer, O., Butera, K., Buckingham, A., Ainoo, S., 2015. Regional prospectivity modelling in data-poor areas: the Kumasi Basin, Ghana. Proceedings of AusIMM 2015 Conference, New Zealand (in press). Mumin, A.H., Fleet, M.E., Chryssoulis, S.L., 1994. Gold mineralization in As-rich mesothermal gold ores of the Bogosu–Prestea mining distric of the Ashanti Gold Belt, Ghana: remobilization of invisible gold. Mineral. Deposita 29, 445–460. Ndiaye, P.M., Dia, A., Vialette, Y., Diallo, D.P., Ngom, P.M., Sylla, M., Wade, S., Dioh, E., 1997. Données pétrographiques, géochimiques et géochronologiques nouvelles sur les granitoı¨des du Paléoprotérozoıque du Supergroupe de Dialé–Daléma (Sénégal Oriental): implications pétrogé netiques et géodynamiques. J. Afr. Earth Sci. 25 (2), 193–208. Newmont Mining Corporation, 2014a. Reserves and resources as of December 31, 2014. Available online at: http://www.newmont.com/files/doc_downloads/reserves_and_ resources/Reserves-Resources-for-Posting_Final.pdf. last Accessed on 13 Sept 2015. Newmont Mining Corporation, 2014b. Delivering on our commitments, 2014 Annual Report and Form 10-K. http://www.newmont.com/files/doc_financials/annual/ 861753_as-printed-Annual-Report_2014_v001_e83uds.pdf; Last accessed on 12 Sept 2015.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012

20

B. Chudasama et al. / Ore Geology Reviews xxx (2015) xxx–xxx

Oberthür, T., Vetter, U., SchmidtMumm, A., Wesiser, T., Amanor, J.A., Gyapong, W.A., Kumi, R., Blemkinsop, T.G., 1994. The Ashanti gold mine at Obuassi, Ghana: mineralogical, geochemical, stable isotope and fluid inclusion studies on the metallogenesis of the deposit. Jahrb. Geol. D 100, 31–129. Oberthür, T., Vetter, U., Davis, D.W., Amanor, J.A., 1998. Age constraints on gold mineralization and paleoproterozoic crustal evolution in the Ashanti belt of southern Ghana. Precambrian Res. 89, 129–143. Pawlig, S., Gueye, M., Klischies, R., Schwarz, S., Wemmer, K., Siegesmund, S., 2006. Geochemical and Sr–Nd isotopic data on the Birimian of the Kédougou–Kénieba inlier (eastern Senegal): implications on the Paleoproterozoic evolution of the West African craton. S. Afr. J. Geol. 109, 411–427. Perrouty, S., Aillères, L., Jessell, M., Baratoux, L., Bourassa, Y., Crawford, B., 2012. Revised Eburnean geodynamic evolution of the gold-rich southern Ashanti Belt, Ghana, with new field and geophysical evidence of pre-Tarkwaian deformations. Precambrian Res. 204–205, 12–39. Pigois, J.P., Groves, D.I., Fletcher, I.R., McNaughton, N.J., Snee, L.W., 2003. Age constraints on Tarkwaian palaeoplacer and lode-gold formation in the Tarkwa–Damang district, SW Ghana. Miner. Deposita 38 (6), 695–714. Pirajno, F., 2009. Hydrothermal Processes and Mineral Systems. Springer Science, p. 1241. Porwal, A., Carranza, E.J.M., Hale, M., 2003. Extended weights-of-evidence modeling for predictive mapping of base-metal deposit potential in Aravalli Province, Western India. Explor. Min. Geol. 10 (4), 155–163. Porwal, A., Das, R.D., Chaudhary, B., Gonzalez-Alvarez, I., Kreuzer, O.P., 2014. Fuzzy inference systems for prospectivity modeling of mineral systems and a case–study for prospectivity mapping of surficial Uranium in Yeelirrie Area, Western Australia. Ore Geol. Rev. http://dx.doi.org/10.1016/j.oregeorev.2014.10.016 (Available online 28 October 2014, ISSN 0169-1368). Resolute Mining, 2014a. Annual Report http://www.resolute-ltd.com.au/wp-content/uploads/2014/10/2014-Annual-Report.pdf. Accessed on 12 Sept 2015. Resolute Mining, 2014b. ASX Announcement http://www.resolute-ltd.com.au/wp-content/uploads/2014/08/576-150814-Reserve-Resource-Statement-June-2014.pdf; Last accessed on 12 Sept 2015. Sibson, R., 2001. Seismogenic framework for hydrothermal transport and ore deposition. SEG Rev. 14, 25–50. Siddorn, J., Lee, C., 2005. Structural geology of the Ashanti II concessions, Southwest Ghana. SRK Consulting Report No. SRK 2CP005.000/001 (70 pp.). Signature Metals Limited, 2015. Quarterly report - December 2014. Available online at: http://www.signaturemetals.com.au/ASXReleases.html; last accessed: 13 Sept 2015.

SRK Consulting, 2008. Independent technical report, Obuasi gold project, Obuasi, Ghana. In: Couture, J.-F., Chartier, D., Weiershauser, L. (Eds.), Technical Report for Pelagio Mines Incorporated Available online at: http://www.pelangio.com/i/pdf/techreports/2008-04-30_NI-43-101.pdf. Accessed on 13/09/2015. SRK Consulting, 2011. Technical report - Obotan gold project mineral resource estimate. In: Gleeson, P., Trivindren, N., Binoir, J. (Eds.), National Instrument 43-101 Technical Report for PMI Gold Corporation Limited. SRK Consulting, 2014. Technical report on resources and reserves, Golden Star Resources Ltd, Bogoso Prestea mine, Ghana. National Instrument 43-101. In: Oldcorn, R., Bray, C., Arthur, J., Bourassa, Y. (Eds.), Technical Report for Golden Star Resources Limited Available online at http://www.gsr.com/files/doc_downloads/Bogoso/U5859-GSR43-101_2013_BOGOSO_Final_140312_Clean.pdf; last accessed on 13 Sept 2015. Taylor, P.N., Moorbath, S., Leube, A., Hirdes, W., 1988. Geochronology and crustal evolution of Early Proterozoic granite–greenstone terrains in Ghana West Africa. International Conference and Workshop on the Geology of Ghana with Special Emphasis on Gold, Programme and Abstracts, Accra, Ghana, pp. 43–45. Taylor, P.N., Moorbath, S., Leube, A., Hirdes, W., 1992. Early Proterozoic crustal evolution in the birimian of Ghana: constraints from geochronology and isotope geochemistry. Precambrian Res. 56, 97–111. Vidal, M., Delor, C., Pouclet, A., Sim'eon, Y., Alric, G., 1996. Evolution géodynamique de l'Afrique de l'Ouest entre 2.2 et 2 Ga: le style “archéen” des ceintures vertes et des ensembles sédimentaires birimiens du nord-est de la Cˆote d'Ivoire. Bull. Soc. Géol. Fr. 167, 307–319. Vidal, M., Gumiaux, C., Cagnard, F., Pouclet, A., Ouattara, G., Pichon, M., 2009. Evolution of a Paleoproterozoic weak type orogeny in the West African Craton (Ivory Coast). Tectonophysics 477, 145–159. White, A., Burgess, R., Charnley, N., Selby, D., Whitehouse, M., Robb, L., Waters, D., 2014. Constraints on the timing of late-Eburnean metamorphism, gold mineralisation and regional exhumation at Damang mine, Ghana. Precambrian Res. 243, 18–38. Yao, Y., Robb, L.J., 2000. Gold mineralization in Palaeoproterozoic granitoids at Obuasi, Ashanti region, Ghana: ore geology, geochemistry and fluid characteristics. S. Afr. J. Geol. 103, 255–278. Zadeh, L.A., 1972. A fuzzy-set-theoretic interpretation of linguistic hedges. J. Cybern. 2 (3), 4–34. Zadeh, L.A., 1973. Outline of a new approach to the analysis of complex systems and decision processes. IEEE Trans. Syst. Man Cybern. SMC-3, 28–44.

Please cite this article as: Chudasama, B., et al., Geology, geodynamics and orogenic gold prospectivity modelling of the Paleoproterozoic Kumasi Basin, Ghana, West Africa, Ore Geol. Rev. (2015), http://dx.doi.org/10.1016/j.oregeorev.2015.08.012