Global exploration for high-grade iron ore increased rapidly in the early to mid-2000s in response to increased demand for seaborne-traded iron ore, which was, in turn, largely driven by China’s rapid economic growth (Barkas, 2015; Hurst, 2016). In the Hamersley province, Pilbara craton, Western Australia, major and rapid growth in iron ore production drove increased levels of both exploration expenditure (Fig. 1A) and declared resources (Fig. 1B). Exploration activity in the Hamersley province focused on expanding resources in the near-mine environment and using ore genesis models to guide evaluation of covered or underexplored parts of the province (Dalstra and Flis, 2008). Several major iron orebodies were delineated as an outcome of this exploration activity, many of which were historically known deposits expanded through systematic drilling activity (Kepert, 2015). A small number of deposits can, however, be considered as new discoveries within the Hamersley province through this period of intense exploration, including Nyidinghu (Fowers et al., 2013) and buried channel iron deposits at Caliwingina Creek (Dalstra et al., 2009) and Solomon (Clarke et al., 2009; Kepert et al., 2010).

In parallel with the growth in global iron ore exploration activity was an increase in research into the geology and genesis of high-grade iron ore (e.g., Hagemann et al., 2008, 2016; Angerer et al., 2014). Many studies have focused on the geology and ore genesis of high-grade, commonly microplaty textured hematite deposits, represented by Mt. Tom Price and Mt. Whaleback in the Hamersley province (e.g., Taylor et al., 2001; Thorne et al., 2004; Webb et al., 2004; Morris, 2012). Additional studies have been completed on analogous deposits in West Africa (Cope et al., 2005), Carajas, Brazil (Figueiredo e Silva et al., 2008; Lobato et al., 2008), India (Mukhopadhyay et al., 2002), and South Africa (Beukes et al., 2003). These studies have generally proposed hypogene processes for the formation of high-grade hematite mineralization, with deposits variably overprinted by later, supergene processes (Beukes et al., 2003; Thorne et al., 2008).

Fewer studies have been undertaken on the volumetrically more significant martite-goethite deposits in the Hamersley province (Fig. 2), represented, for example, by Hope Downs (Lascelles, 2006) and Roy Hill (Clout and Fitzgerald, 2011). Martite-goethite ores are commonly interpreted to have formed during late Mesozoic weathering of banded iron formations (Morris, 1980; Thorne et al., 2008), due to their spatial association with Cretaceous and Tertiary drainage patterns and paleosurfaces (Harmsworth et al., 1990). This paper seeks to add to research into major martite-goethite orebodies by documenting the discovery, regional geologic setting, and nature and controls on mineralization of the 1.8-billion-tonne (Gt) South Flank iron ore deposit located in the Hamersley province (Fig. 3A).

The Hamersley province is located on the southern margin of the Pilbara craton in the northwest of Western Australia (Fig. 3A). The predominantly sedimentary rocks of the Hamersley Group form part of the Mt. Bruce Supergroup (Trendall and Blockley, 1970). They were deposited conformably on the basalt-dominated Fortescue Group (Thorne and Trendall, 2001) and are overlain by the Turee Creek Group (Thorne et al., 1995; Martin et al., 2000). The Mt. Bruce Supergroup rests unconformably on 3.72 to 2.85 Ga granite-greenstone terrains of the Pilbara craton (Van Kranendonk et al., 2002) and is unconformably overlain by the Wyloo Group, a late Paleoproterozoic sedimentary and volcanic succession (Tyler and Thorne, 1990; Thorne and Seymour, 1991) deposited in a number of subbasins spanning the period ca. 2050 to 1680 Ma (Krapez et al., 2015, 2017). The Hamersley Group comprises intercalated banded iron formations (BIF), shales, and carbonates overlain by bimodal volcanic rocks and associated subvolcanic intrusions, dated to between 2629 and 2449 Ma (summarized by Trendall et al., 2004). It has been divided into eight formations with major BIF horizons occurring at the base (Marra Mamba Iron Formation), toward the center (Brockman Iron Formation), and at the top (Boolgeeda Iron Formation; Trendall and Blockley, 1970; Fig. 3B).

Rocks of the Mt. Bruce Supergroup record the opening and closing of a marginal basin (Blake and Barley, 1992). Initial development of a WNW-ESE–trending rift along the southern margin of the Pilbara craton between about 2770 and 2690 Ma, accompanied by basaltic volcanism and associated sedimentation, was followed by sedimentation within a siliciclastic-starved, ensialic, deep-water basin. Convergence along the southern margin resulted in the formation of a subduction-related orogeny accompanied by voluminous bimodal volcanism and foreland sedimentation (Turee Creek Basin; Krapez et al., 2017). Subtle evidence of an orogenic event dated to between ~2450 and 2400 Ma is based on (1) bimodal age distribution of metamorphic monazite and xenotime growth in metasedimentary rocks throughout the Pilbara (ca. 2430–ca. 2400 Ma and ca. 2195–ca. 2145 Ma; Rasmussen et al., 2005), and (2) timing of orogenic Au mineralization at Paulsens mine, which is hosted in deformed rocks of the Fortescue Group (2403 ± 5 Ma; Fielding et al., 2017). The presence of internal unconformities within the Turee Creek Group is also indicative of deformation accompanying sedimentation.

Regional metamorphism at ca 2.2 Ga (Rasmussen et al., 2005; Shibuya et al., 2010) coincides with the Ophthalmian orogeny, which was related to the collision of the Glenburgh terrane and the Pilbara craton (Johnston et al., 2013). The Ophthalmian orogeny is associated with the development of regional-scale, WNW-trending, upright to N-verging asymmetric folds. Dolerite sills dated to 2208 ± 10 Ma (Muller et al., 2005) are folded by the Ophthalmian event, placing a maximum age on Ophthalmian folding. The sills are truncated by the unconformity at the base of the Lower Wyloo Group.

The Lower Wyloo Group is bounded at the top and bottom by regional unconformities and represents a major intracontinental rift event, associated with the development of the Cheela Springs large igneous province (LIP), dated to between ca. 2050 and ca. 2030 Ma (Horseshoe Basin; Krapez et al., 2015). In the western part of the Hamersley province, Ophthalmian folds were reoriented about NW-trending axes during the Panhandle event (Taylor et al., 2001). Both Ophthalmian and Panhandle folds are cut by NW-trending 2009 ± 16 Ma dolerite dikes (Muller et al., 2005), which in turn are truncated by the unconformity at the base of the Upper Wyloo Group: this places a minimum age on the Panhandle event. The unconformity at the top of the Lower Wyloo records the collision between the Yilgarn craton and the amalgamated Pilbara craton and Glenburgh terrane between ca. 2025 and ca. 1950 Ma (Glenburgh orogeny; Johnston et al., 2013).

Hydrothermal growth of xenotime in high-grade iron deposits records the thermal effects of orogenic and extensional events occurring to the south and east of the Hamersley province during the remainder of the Paleoproterozoic (Rasmussen et al., 2007) and a number of mafic dike suites were emplaced during the Meso- to Neoproterozoic (Wingate, 2017). There is no preservation of Palaeozoic rocks in the Hamersley province.

Domal uplift of the Hamersley province commenced in the late Mesozoic (Czarnota et al., 2014) and has driven three main periods of sedimentation and detrital Fe ore accumulation together with at least three periods of lateritic hardcap development (Kneeshaw and Morris, 2014). These lateritized paleosurfaces are well preserved, forming smooth undulating tops to the ranges, and are referred to as Hamersley surfaces (Campana et al., 1964; Twidale, 1997).

The oldest package (so-called CzD1 or “Red Ochre Detritals”) is typically preserved in karst sinkholes and small basins formed above the contact between Marra Mamba Iron Formation and Wittenoom Formation and comprises alluvium dominated by clasts of massive hematite and kaolinite-altered shale set in a matrix of fine-grained hematite. A rare drill intercept of lignite from within the CzD1 sequence taken from the Ethel Gorge area contains a number of microfossils, most of them belonging to species that were extant over long periods of time, but revealed no evidence of angiosperm pollen, suggesting that the sample is possibly Mesozoic in age (ca. 174–100 Ma; Hannaford, 2016). The CzD2 package includes the economically important channel iron deposits (CIDs), which have been dated to mid-Miocene in age (Danisik et al., 2013). These consist of fluviatile deposits comprising pelletoids with hematite cores and goethite rims, abundant goethitized wood/charcoal fragments, and minor amounts of clay set in a porous goethitic matrix (Morris and Ramanaidou, 2007). Outside these discrete fluvial systems, CzD2 deposition is characterized by organic-rich clays and lignite overlain by massive siderite, typically altered to kaolinite clays overlain by goethite in the weathering profile. Lacustrine-fluvial micritic limestone and calcrete, silicified in the weathering zone, of the Oakover Formation forms the uppermost unit of CzD2. A further erosional phase, starting in the mid-Pliocene and extending to the present day, resulted in the accumulation of alluvial and colluvial material of the CzD3 sequence (Kneeshaw and Morris, 2014). The prominent steep talus cones developed beneath high BIF cliffs attest to the current aggressive erosional regime.

There are four main types of iron mineralization in the Hamersley province (e.g., Thorne et al., 2008; Angerer et al., 2014): (1) hypogene martite-microplaty hematite (M-mplH) developed, for example, at Mt. Whaleback and Mt. Tom Price; (2) mimetic-textured martite-goethite mineralization, which is the topic of this paper (refer to Ramanaidou and Morris, 2010, for the distinction between this and lateritic weathering); (3) channel iron deposits (CIDs), exemplified by Yandicoogina and Robe River (Ramanaidou et al., 2003); and (4) detrital deposits dominated by fine-grained hematitic sediments (CzD1) and colluvial/alluvial accumulations derived from erosion of mineralized BIF (CzD3: Clout and Simonson, 2005; Morris and Kneeshaw, 2011).

Iron deposits of the central Hamersley province (Fig. 4) include both supergene martite-goethite mineralization (Lascelles, 2006; Bodycoat, 2007) and detrital iron deposits. There is no documented evidence of hypogene-style iron mineralization in this region, characterized by microplaty hematite, or the suite of alteration minerals, such as apatite, chlorite, and dolomite/ankerite, which are typically associated with this style of iron mineralization (Thorne et al., 2004, 2014).

Regional exploration of the Hamersley province followed lifting of the export embargo on iron ore in Western Australia by the state government in 1960, and followed initial discoveries of iron ore by Hilditch, Hancock, and others (Lee, 2015). The subsequent issuing of iron ore tenements by the Western Australian government in 1961 led to a major increase in exploration by both industry and government (Harmsworth et al., 1990) and the establishment of initial mining operations at Mount Tom Price in 1966 and Mount Whaleback in 1969. Exploration during this period targeted bedrock-hosted, high-grade microplaty-hematite mineralization hosted by the Brockman Iron Formation. Drilling of mineralization in the Marra Mamba Iron Formation commenced in the 1970s around Mining Area C, after recognition of the low P content of these ores (Kepert, 2015). Numerous martite-goethite deposits were identified and drilled at this time, but, due to a number of factors, exploration and mine development were deferred for several years (Kepert, 2015).

Earliest exploration around South Flank, which is located on the southern side of the Weeli Wolli anticline (Fig. 4), was conducted by Mt. Goldsworthy Mining in the period 1964 to 1978 and comprised reconnaissance mapping, which identified outcropping hardcap iron mineralization, and limited shallow drilling to test both bedrock and detrital iron ore potential (Table 1). Synclinal control on the distribution of bedrock-hosted, high-grade iron mineralization at South Flank was recognized from this work, and a potential “reserve” of 872 million tonnes (Mt) estimated, based on 32 drill holes across part of the deposit (Burton and Hackett, 1978). BHP Iron Ore (Goldsworthy) Pty Ltd undertook an extensive remapping program of the Weeli Wolli anticline area in 1994 (Lipple et al., 1994). Reverse-circulation drill testing between 1993 and 1995, to test for iron mineralization in scree, red ochre detrital sediments, and bedrock, intersected thin (less than 30 m) or unmineralized detrital deposits and discontinuous soft limonitic iron mineralization in the Mt. Newman Member (Tehnas and Kepert, 1996). Results of these programs indicated moderate potential for iron mineralization at South Flank (Kepert, 2001).

No further geologic work took place at South Flank until 2007, by which time the nearby operations at Mining Area C had commenced production from martite-goethite orebodies hosted by Marra Mamba Iron Formation, and the geology and structural controls on these orebodies had been studied (Bodycoat, 2007). South Flank was reevaluated as part of a regional program to assess iron potential in the well-mineralized central part of the Hamersley province (Fig. 4). An exploration model for martite-goethite mineralization was developed as the basis for assessing under- or unexplored areas. The model included elements of traditional deposit models, developed from analogous deposits at the camp or district scale (cf. McCuaig and Hronsky, 2014). In this case, orebody-scale geologic controls on martite-goethite mineralization at Mining Area C and other major deposits in the district (e.g., Hope Downs, Jinidi), such as host rocks, structural controls, and geochemical and mineralogical signatures, were used as inputs to the model (Fig. 5). Components of the minerals systems approach to exploration targeting (Wyborn et al., 1994; McCuaig et al., 2010), applicable to the camp or district scale, were incorporated in the conceptual model for iron mineralization. These include the following:

• Structure and potential fluid pathways: At the regional scale, martite-goethite mineralization is commonly associated with major E-W–trending folds (Fig. 4), which can have major faults subparallel to fold axes. At the deposit scale, mineralization is associated with tight, inclined to overturned D2 folds, and associated reverse faults developed at the contact between Mt. Newman and West Angela members (Thorne et al., 2008). High-angle faults crosscut stratigraphy and potentially enhanced permeability through host iron formations. A structural model was developed for South Flank, which hypothesized that N-verging synclines and associated thrust faults, which had not been identified in earlier geologic assessments, are critical to the development of large highgrade Fe orebodies.

• Suitable trap/host rocks: At Mining Area C and other large martite-goethite deposits, iron mineralization is preferentially developed in the Mt. Newman Member of the Marra Mamba Iron Formation (Bodycoat, 2007; Thorne et al., 2008). Similarly, iron mineralization is normally best developed in the D2 and D4 units of the Dales Gorge Member in Brockman Iron Formation hosted martite-goethite deposits (Barrett et al., 2015).

• Preservation: An important element of the model for martite-goethite mineralization is preservation of prospective host iron formations and iron ore beneath younger detrital cover sequences, which are extensively developed across South Flank. Importantly, the model developed highlighted unknown or conjectural factors in the formation of very large martite-goethite orebodies, which may have constrained the exploration search space (cf. Hronsky, 2009). These include fluid sources, potential causes of large-scale desilicification, which is commonly cited as the principal iron ore-forming process (e.g., Angerer et al., 2014), and timing of mineralization with respect to deformation.

A range of E-W–striking hills (the Djadjiling Range), which extend about 25 km across the northern parts of the deposit, characterizes the terrain around South Flank. Largely unmineralized lower units of the Marra Mamba Iron Formation (Nammuldi and MacLeod members) are well exposed in steeply incised gullies within the Djadjiling Range; however, detrital deposits cover much of the prospective Mount Newman Member—the central and southern parts of the deposit are concealed beneath a flat sediment-filled valley. A high-sensitivity aeromagnetic and gravity gradiometer survey was completed over the extents of the project area in 2007 to provide geophysical datasets to support further exploration (Fig. 6A, B). Interpretation of the survey (Fig. 6C) enabled identification of a number of important elements of the martite-goethite mineralization model, notably (1) the main stratigraphic members of the Marra Mamba Iron Formation, including distribution of the Mount Newman Member, (2) regional-scale E-W–trending folds, which have a strong spatial relationship to iron mineralization in analogous deposits at Mining Area C (Fig. 5), and (3) a series of NE- and NW-trending faults that crosscut stratigraphy. The interpreted survey was the basis for a detailed mapping program completed in 2007 to 2008, which concentrated on determining the prospect-scale distribution and geometry of Marra Mamba Iron Formation stratigraphy at South Flank, documenting main structures, and ground-truthing the geophysical interpretation, particularly zones of magnetite destruction associated with gravity anomalies that may represent iron mineralization.

A proof of concept drilling program, on 1,200-m-spaced lines with holes spaced at 50 m along each line, commenced in late 2008 across central and southern parts of the deposit. The program aimed to test the structural model for South Flank, in particular stratigraphic thickening caused by folding and reverse faulting, and to assess bedrock-hosted iron mineralization in areas beneath detrital cover. Results from the drilling indicated the presence of thickened Mt. Newman Member, resulting from folding and associated reverse faults, the potentially wide extent and high Fe grade of martite-goethite mineralization at South Flank, and gave encouragement for further 600-m-spaced infill drilling and drilling to define the full extents of iron mineralization.

Since 2010, 550 km of reverse-circulation drilling and diamond drilling (to define depth extents of iron mineralization) has been completed at South Flank in several programs to progressively infill and define the orebody. In parallel, the geometallurgical characteristics and geotechnical and hydrogeological parameters of the deposit have been assessed. The aims and results of each of these programs are summarized in Table 1. Each of these programs was enabled and preceded by detailed environmental and heritage surveys and approvals.

South Flank is located 90 km west-northwest of the town of Newman and forms a major iron deposit distinct from Mining Area C (comprising current mining operations of the North Flank and Packsaddle orebodies), which is located 8 to 10 km to the north (Fig. 4). Iron mineralization at South Flank is located on the southern flank of the Weeli Wolli anticline, an EW-trending, doubly plunging anticline in the core of which are exposed rocks of the Jeerinah Formation, Marra Mamba Iron Formation, and Wittenoom Formation (Fig. 7). Rocks belonging to the Mt. Sylvia Formation, Mt. McRae Shale, and Brockman Iron Formation crop out to the north in the Packsaddle Ranges, and to the south at Mt. Robinson and in the Governor Range. The intervening strike valleys are largely filled with Cenozoic detrital material. Proterozoic bedrock has been metamorphosed to prehnite-pumpelleyite facies (prehnite-pumpelleyite-epidote-actinolite Zone III of Smith et al., 1982). Smith et al. (1982) interpret peak metamorphic P-T conditions of 0.5 to 1 kbar and 100° to 300°C. The major stratigraphic units encountered at South Flank are described below in order of increasing age.

CzD3: Pliocene to Recent sediments of CzD3 comprise a mix of colluvial and alluvial sediments reaching thicknesses of about 100 m. The sediments range from proximal cobble to pebble scree fans to distal silty valley fills and playa deposits (Kepert, 2001). The colluvial/alluvial fans comprise polymictic conglomerate with subangular to subrounded, poorly sorted clasts of iron ore, BIF, and shale in a general silty to soil-rich matrix. The siltstones are generally homogeneous, red to reddish-brown in color, and consist mainly of finegrained, iron-rich silt with up to 20% of maghemite-rich clasts. The unit becomes more gravelly toward the base: these portions are clast supported and contain loose to cemented pisolites, in addition to the maghemite-rich clasts, set in a silty matrix. Horizons within the gravelly siltstone may be unusually pisolite rich and cemented with goethite, producing what is informally termed “welded pisolites.” Goethite-cemented ore pebble colluvial deposits or “canga” are the product of Pliocene to Recent lateritic weathering.

CzD2: Fluvial and lacustrine sediments of the CzD2 package typically reach 50 m in thickness and are dominated by sands, clays, and fluvial to lacustrine limestone and associated calcrete, the latter typically developed toward the top of the sequence. Light-colored sands can contain rounded pebbleto cobble-sized clasts of quartz and are variably intermixed with clay. Clay-rich zones vary in color from red through yellowish-brown to white, and Fe hydroxides dominate over Fe oxides. Paleolateritization of CzD2 material is reflected by the development of a breccia unit, which consists of angular to subrounded clasts of ferricrete together with pisolites, set in a vuggy matrix dominated by goethite with minor white clay, underlain by mottled saprolitic clays. The major goethite-hematite channel iron deposits that occur throughout the Hamersley province form part of the CzD2 detrital package (Morris and Ramanaidou, 2007), but are poorly developed in the South Flank area.

CzD1: The CzD1 package comprises dominantly fluviatile sediments and can exceed 100 m in thickness at South Flank. The sediments range from proximal conglomerates to distal clays and appear to be a mix of detrital material derived mainly from the Marra Mamba Iron Formation and terra rossa soils developed over the Paraburdoo Member (Morris, 1994). Rare drill intersections of lignite- and pyrite-bearing clays and sideritic material occur beneath the equivalent hematitic sediments at Hope Downs (Kepert, 2001). Hematite-rich conglomerate dominates the CzD1 package and comprises angular to subrounded clasts of massive hematite and kaolinite-altered shale in a matrix of fine hematitic silt with minor aluminous clay. The conglomerates mostly overlie the West Angela Member and mineralized Mount Newman Member with a moderately dipping unconformity on an irregular, gullied surface (Fig. 8A). Conglomerate layers are interbedded with more silty horizons on a submeter scale and the proportion of siltstone increases away from the bedrock source. Limited transport of the hematite siltstone material is inferred from the abundance of fine, sand-sized martite grains, which retain an octahedral form inherited from the magnetite precursor phase. Bedding angles commonly exceed 50° (i.e., in excess of the angle of repose), suggesting postdepositional tilting and/or slumping (Fig. 8B). These hematitic conglomerates and siltstones typically occupy narrow, elongate, internally draining (i.e., doubly plunging) depressions developed in the Wittenoom Formation adjacent to the contact with the Marra Mamba Iron Formation. Such basins are particularly well developed immediately to the north of iron ore deposits at North Flank, Mining Area C. It is possible that the steeply dipping bedding reflects ongoing karstic dissolution of carbonates in the underlying Wittenoom Formation and slumping/collapse of the CzD1 sediments into karst cavities. These Cenozoic sediments were deposited unconformably on a lateritized paleosurface of probable Mezozoic to Palaeogene age, referred to as the Hamersley surface (Campana at al., 1964; Twidale, 1997), which was developed on the underlying Paleoproterozoic to Archean bedrock.

Wittenoom Formation: The Wittenoom Formation is divided into three units. From top to bottom, these are the Bee Gorge, Paraburdoo, and West Angela members (Blockley et al., 1993; Simonson et al., 1993). The Bee Gorge Member comprises thinly laminated argillite with subordinate intercalated beds of dolomite, chert, volcaniclastic material, and BIF. Measured sections range from 111 to 227 m in thickness, and in the Weeli Wolli anticline area the unit is typically about 140 m thick. It contains three informal stratigraphic marker horizons, listed in descending stratigraphic order: the Spherule Marker Bed (2565 ± 9 Ma; Trendall et al., 2004), the Main Tuff interval, and the Crystal-rich Tuff (2561 ± 8 Ma; Trendall et al., 1998). Although the Main Tuff interval reaches 16.4 m in thickness, individual carbonate layers are typically less than 4 m thick, and the individual BIF horizons are even thinner.

The Paraburdoo Member comprises evenly bedded dolomite with minor amounts of chert. Individual layers of dolomite range from a few centimeters to less than 1 m in thickness. The exact thickness of the member is unknown, but Kepert (2001) estimates it to be 350 to 500 m thick to the west of the Weeli Wolli anticline. It is typically very poorly exposed and forms the floors of sediment-filled valleys between the Marra Mamba and Brockman iron formations.

The dominant lithologies of fresh West Angela Member comprise interbedded dolomite and shale-rich dolomite (Fig. 9). Chert is a minor component of the upper part of the member but becomes more common toward the base. A 10-to 20-m-thick interval of interbedded chert, BIF, and shale toward the base of the member gives rise to three characteristic gamma-ray peaks (AS1 to AS3: Simonson et al., 1993). The West Angela Member has been best described where it is intersected in drilling, proximal to martite-goethite ore deposits. In these areas, the carbonate content is leached and the member presents as manganiferous, Fe-rich, poorly bedded shale interbedded with chert. The West Angela Member is informally subdivided into an upper, more shale rich unit (WA2) and a lower, more chert rich unit (WA1; Fig. 9, Table 2). The lower WA1 subunit is commonly mineralized and extends from the thick, podded, magnetite-bearing BIF horizon at the top of the Marra Mamba Iron Formation to the base of AS3 shale horizon. The upper WA2 subunit is characterized by a noisier and higher-amplitude gamma-ray log, a greater shale component, and elevated Mn levels (Table 2).

Marra Mamba Iron Formation: The Marra Mamba Iron Formation conformably overlies the Roy Hill Member of the Jeerinah Formation and is subdivided into three members on the basis of changes in relative proportions of BIF, chert, and shale or shale-rich BIF (Simonson et al., 1993; Kepert, 2001). From top to bottom, these are the Mt. Newman, MacLeod, and Nammuldi members (Fig. 9). The average compositions of the three members are displayed graphically in Figure 10 and quantified in Table 2.

The Mount Newman Member is primarily composed of carbonate- and silicate-rich magnetite BIF with minor shale interbeds. Nine macrobands of BIF are separated by eight macrobands of more shale rich material (NS1–NS8), and these form important stratigraphic markers. Compared with BIF mesobands in the MacLeod and Nammuldi members, the Mt. Newman Member is more iron rich, better banded, and has cream-colored (rather than yellow) chert bands. The more chert rich bands typically exhibit a pinch-and-swell structure termed “podding,” which is characteristic of this formation (Fig. 8C). The geochemistry of the shale bands and rare textural preservation of glass shards suggest derivation from volcaniclastic components. The Mount Newman Member is about 60 m thick and is divided into three subunits to separate the relatively Fe oxide poor and shale rich center, denoted N2 (taken from the base of NS3 to the top of NS6), from the more BIF rich portions above and below. The N1 subunit is at the base and N3 is at the top of the Member. The N3 subunit is generally the best mineralized of the three (Fig. 10, Table 2).

The MacLeod Member is shale rich and very susceptible to erosion as a result. It contains several horizons of chert pods, and the base of the unit is placed informally at the base of the Potato Bed to facilitate mapping and drill-hole logging (Kepert, 2001). This is a recognizable marker unit in outcrop and corresponds to a readily identifiable boundary in gamma-ray logs. The MacLeod Member is typically about 75 m thick in the central part of the Hamersley province. The Potato Bed is a 12-m-thick unit comprising abundant chert pods set in a shale-rich, riebeckite-bearing BIF. A second distinctively podded layer, the Football Chert, comprises a 4- to 6.5-m-thick unit of large (<70 cm), elliptical, white chert pods. A further distinctive unit is the Stilpnomelane Bed. This is about 11 m thick and is composed of dark green to black, stilpnomelanerich tuff with limonitic nodules (<2-cm diameter) after diagenetic pyrite.

The Nammuldi Member is typically about 90 m thick and consists of poorly bedded yellow and brown chert with thin BIF horizons. Undulating bedding and chert pods and rods are commonly developed. The base of the member is marked by a distinctive 1- to 2-m-thick band of dark chert containing lath-shaped crystal molds, possibly after gypsum, carbonate, or barite (Kepert, 2001). The lower 1 to 15 m of the member is carbonate rich and can be locally extensively silicified. A persistent zone of rodded chert, up to 10 m thick and interbedded with shale or tuff, marks the transition to more chert rich material in the main body of the member.

Jeerinah Formation: The Jeerinah Formation crops out in the core of the Weeli Wolli anticline. It forms the uppermost unit of the Fortescue Group and is intruded by dolerite sills. The uppermost part of the Jeerinah Formation comprises pyritic black shales of the Roy Hill Member. Massive to well-bedded felsic tuffs and tuffaceous shales, interbedded with pyritic chert, occur at the top of the member and have been dated to 2629 ± 5 Ma (Trendall et al., 2004). A distinctive 5-m-thick, black- and white-striped chert, the Marker Chert, occurs toward the top of the Roy Hill Member but below the tuffaceous horizon (Fig. 9).

South Flank is located on the southern limb of one of the largest outcropping domes of Marra Mamba Iron Formation in the central Hamersley province—the Weeli Wolli anticline. Lipple et al. (1994) identified fold complexity within the anticline, where subordinate asymmetric folds occur in relatively narrow zones along the southern and northern limbs, with an intervening broad, gently plunging synclinal fold of lower strain along the regional fold hinge. Areas of highest topography generally coincide with anticlinal zones, which in places have exposed the upper parts of the Jeerinah Formation, whereas synclinal zones occupy lower elevations and comprise the Mt. Newman and West Angela members, commonly covered by detrital sediments. The asymmetric geometry of subordinate folds has strongly influenced the form of the landscape, with the gently S dipping limbs reflected in the landscape, whereas the steeply N dipping to subvertical limbs more commonly form steep slopes on the south side of gullies.

The regional structural setting of the central Hamersley province is dominated by regional-scale (10- to 20-km wavelength), open, doubly plunging folds, with inclined to upright E-W–striking axial planes (Fig. 7), which were developed during the Ophthalmian orogeny (D2). The interplay of these folds with the weathering-resistant nature of the BIFs has resulted in the present prominent outcrop pattern within the region (O’Sullivan, 1992). Major anticlines are typically represented by low ranges formed by the Marra Mamba Iron Formation, whereas synclines are preserved as higher-relief ridge-lines formed by the Brockman Iron Formation. Low-relief, sediment-filled valleys have formed on the limbs of the major folds over the generally steeply dipping carbonates of the Wittenoom Formation. Interpretation of three seismic reflection lines acquired through the central Hamersley province shows these regional-scale folds have likely developed as fault-propagation folds, with basal detachments in the Fortescue Group or along basement highs (granite domes), propagating northward toward the Ophthalmian foreland (Hollingsworth et al., 2002). In the western Hamersley province, the regional-scale dome-and-basin fold pattern has been strongly influenced by the basement granite-greenstone geometry, either during basin inversion or through gravity-driven vertical tectonics (Rocklean Movement; Kepert, 2001) in addition to overprinting by the Panhandle event (Taylor et al., 2001).

A series of subordinate folds (~100–1,000-m wavelength) are superimposed on the regional-scale folds but appear to be slightly older, representing the first response to far-field stress during the Ophthalmian orogeny (also D2). These manifest as open to tight, asymmetric, N-verging (axial planes dip ~30°–50°S) folds which predominantly trend east-west; however, the trend of these folds varies significantly across South Flank from northwest-southeast through to east-northeast–west-southwest. Historically, these changes in fold orientation were ascribed to overprinting by two later deformation events (D4 and D5 of Lipple et al., 1994); however, structural interpretations conducted over the last decade show these folds are likely to have been developed synchronously with the Ophthalmian orogeny (NW-SE– and ENE-WSW–trending folds) and possibly during a later period of extension (ENE-WSW–trending folds).

Thrust faults, with up to ~200 m of apparent movement (Kepert, 2001), have developed locally in response to overtightening of these asymmetric folds, typically nucleating in the West Angela Member with displacement rapidly decreasing down stratigraphy into the lower Mt. Newman and MacLeod members. Faulting has largely been accommodated by bedding-parallel shearing in shale units and along stratigraphic boundaries, commonly making discrete fault planes difficult to identify. A shallowing of fault dips from the south to north limb of the Weeli Wolli anticline suggests thrust faults were formed prior to or synchronously with regional-scale folding. Mesozoic-Cenozoic dissolution of carbonates in the footwall of the faults has resulted in modification of fault planes, which now commonly display folded geometries mirroring the form of bedrock in the footwall. In extreme cases, particularly on the North Flank, this has resulted in complete reversal of fault dips that now, in places, dip to the north.

Northeast- and NW-trending faults, possibly related to dike emplacement, crosscut the stratigraphy. Displacements along these structures are usually difficult to identify; however, changes in fold geometry and complexity across them suggest these were early-formed structures, possibly vertically accreted from the basement. These structures may have later formed important fluid conduits or barriers to fluid movement during ore formation, as Fe mineralization can terminate against these faults.

Based on orientation and crosscutting relationships, Tyler (1991) defined eight suites of dolerite dikes in the southeastern Hamersley province. These dolerite dikes are typically highly weathered, topographically recessive, and manifested as steep-sided gullies, cutting through both the Marra Mamba and Brockman iron formations. Near surface, they comprise pink- to orange-colored, kaolinite-rich clays. At least three generations of mafic dikes occur in the Mining Area C region, with two of these occurring in the Weeli Wolli anticline (Fig. 7). A NW-SE–trending dike suite has been tentatively correlated with the 1075 Ma Round Hummock Dolerite Suite; however, dikes in this orientation are rarely observed in the upper Weeli Wolli Formation, so they may represent the feeders to mafic flows and sills therein. In the west, the second dike orientation (NE-SW) is represented by a single intrusive body of unknown age; however, the southern extent of this body appears to terminate in Warakurna-age (1075 Ma) dolerite sills of the Edmund Basin. Both dike orientations appear to postdate folding.

Continuity of asymmetric folds and associated thrust faults from east-west through to northwest-southeast trends suggests existing heterogeneity in the underlying rocks has strongly influenced the geometry of F₂ structures through strain partitioning, producing the pronounced sigmoidal fold pattern developed on the Weeli Wolli anticline. At the regional scale, ENE-trending lineaments appear to have influenced the geometry of the longer-wavelength folds in the Mining Area C region. The origin of the ENE-trending lineaments may be related to accretion of the Sylvania inlier (part of the Kurrana terrane) to the Pilbara craton (J. Hronsky, pers. commun., 2017) or development of the 2940 Ma Mosquito Creek Basin (Hickman, 2004). A seemingly solitary, open and upright syncline developed through the center of the Weeli Wolli anticline appears to have developed in response to extension focused along one such ENE-trending lineament.

The oldest structures observed at South Flank formed during diagenesis (D1), synchronous with vertical loading and basin extension. Such structures include weak bedding-parallel fabric and differential bedding-parallel (Fig. 8C) extensional boudins (“pods and rods”) and stacked pods, rare intrafolial isoclinal folds, and extensional shear zones. In addition to tectonic pods, early diagenetic pods occur at specific horizons within the sequence (Kepert, 2001), which historically have proven to be useful marker horizons during both mapping and drilling. Rasmussen et al. (2013) propose these chert pods are likely to have formed due to differential compaction of BIF precursor sediments due to variable silicification at or just below the seafloor.

Karsting developed in the Wittenoom Formation has resulted in postdepositional (Mesozoic-Cenozoic) modification of the CzD1 (Kneeshaw and Morris, 2014) and, possibly, CzD2 sediments. Removal of significant volumes of carbonates has resulted in deformation, in the form of folds and joints, as the sediments subsided to accommodate the volume loss in the underlying bedrock. These karst-generated folds often manifest as steeply dipping (up to 70°) alluvial beds and basal unconformities with dip angles typically close to that of the underlying bedrock. Locally, this has resulted in the development of apparently internally draining basins. Dewatering of the sediments may have also contributed to the observed CzD1 geometries (Kepert, 2001).

South Flank comprises a continuous and tabular body of iron mineralization that covers a strike length of greater than 25 km (Fig. 7). Mineralization is open to the west but is affected by a fold closure where mineralized Mt. Newman Member plunges beneath dolomite of the Wittenoom Formation, and is open to the east, where it continues onto adjoining leases.

Two distinct populations of Fe grade occur at South Flank (Fig. 11): unmineralized BIF (Fe = 20%–45%) and iron ore (Fe greater than 50%, with a range of Fe content 50%–68%). All material with Fe content greater than 50% is considered mineralized. Five types of iron mineralization can be classified, based in part on host unit: (1) bedrock iron mineralization, hosted mostly within the Mount Newman Member of the Marra Mamba Iron Formation, which forms 90% of economic mineralization at South Flank; (2) detrital iron mineralization (CzD1), also known as Red Ochre Detritals; (3) detrital iron mineralization (CzD2), also referred to as pisoidal mineralization; (4) detrital iron mineralization (CzD3); and (5) hardcap mineralization, which can form on all of the above types of mineralization or directly on BIF. Each mineralization type is discussed below; however, the focus of this section is to describe bedrock iron mineralization.

South Flank largely comprises strata-bound and tabular martite-goethite and martite-ochreous goethite mineralization hosted predominantly within the Mount Newman Member (units N1, N2, N3) of the Marra Mamba Iron Formation. Greater than 70% of the mineral resource (at a 54% Fe cutoff) is within the N2 and N3 units, with lesser quantities of ore in N1 (15%). The overlying lower West Angela Member (WA1) and the underlying MacLeod Member host minor amounts of ore (approximately 12% in total).

Iron mineralization shows a strong spatial correlation with several E-W–trending, N-verging synclines, and is associated with low-angle E-W–trending thrust faults, which have caused repetition of the stratigraphy resulting in thicker zones of mineralization (Fig. 12A). Mineralization is typically best developed where the Mount Newman Member is present in the hanging wall and footwall adjacent to thrust planes. This structural configuration commonly results in ore thicknesses of up to 150 m. Mineralization commonly extends to about 200 m below surface but can reach depths of about 300 m in the cores of steep synclines. In addition to syncline- and thrust-associated mineralization, iron enrichment at South Flank can occur on the southern limb of the regional-scale Weeli Wolli anticline, where it is typically 30 to 40 m thick and can extend to depths of up to 200 m below surface (Fig. 12B).

Bedrock iron mineralization comprises both martite-goethite and martite-ochreous goethite types (Fig. 13). The relative abundance of each ore type is related to host stratigraphy. Well-mineralized intervals of the N2 and N3 units of the Mount Newman Member tend to contain martite-goethite with subordinate martite-ochreous goethite ore, whereas mineralization in N1 and WA1 is martite-ochreous goethite dominant. Both martite-goethite and martite-ochreous goethite mineralization comprise gray to dark brown martite-rich layers interbedded with yellow to brown goethite and ochreous goethite-rich layers (Fig. 13B). The top of the N3 unit contains the highest-grade iron mineralization and corresponds with a 15-m-thick zone of martite-dominant mineralization (Fig. 13C). Below this interval, the proportion of martite-ochreous goethite increases.

Banded iron formation primary textures, such as beds and fine laminations, are preserved within ore zones (Fig. 13B) and exert a strong control on the location of mineralization. The transition from BIF to iron mineralization tends to be gradual along strike and downdip within the Mount Newman Member but rapid across strike. This is potentially due to the original compositional bedding of the BIF and, in particular, the presence of shale-rich bands, which likely acted as aquitards during mineralization. The most prominent shale band in the Mount Newman Member occurs toward the base of the N2 unit, and this horizon is also where mineralization commonly transitions to unmineralized rocks throughout the South Flank deposit. Exceptions to this relationship can occur where the N1 unit is proximal either to the paleosurface or to the main E-W–trending thrusts. In these situations, N1 is normally well mineralized.

Within the southernmost central part of South Flank, a small zone of high-grade (Fe > 60%) magnetite-kenomagnetite–rich martite with minor goethite ore occurs. Mineralization is hosted within N1 and N2 and is spatially related to a regional NE-SW–trending structure (evident from gravity and magnetic datasets) where it intersects a smaller east-west fault. At a microscopic scale, this material comprises martite-rich aggregates, with relict magnetite cores, overprinted by goethite, which typically forms rims around the martite. This style of mineralization has not been observed in other parts of South Flank, nor has it been documented in comparable deposits in the central Hamersley district, and it is interpreted to represent the transition from unoxidized, magnetite-rich iron formation to martite-goethite iron mineralization.

Based on detailed geologic logging of extensive and deep drilling, which has tested the full depth extents of the South Flank orebody, no wall-rock alteration associated with iron mineralization has been identified, other than oxidation and weathering-related leaching within adjacent iron formations. There is no evidence at South Flank of hypogene martite-microplaty hematite mineralization and associated alteration assemblages of magnetite-carbonate-chlorite ± talc ± stilpnomelane ± apatite within host iron formations, which has been described at Mount Tom Price (Taylor et al., 2001) and at Mount Whaleback (Webb et al., 2004). With increasing depth, mineralization at South Flank transitions from martite-ochreous goethite and martite-goethite types to martite-goethite dominant, to weakly mineralized iron formation, comprising quartz-magnetite-martite with minor goethite, and, finally, to unmineralized iron formation. The transition typically occurs over a scale of tens to hundreds of meters.

The CzD1 detrital sediments are the oldest preserved unit of the Cenozoic detrital sediments (Morris, 1994) and commonly form channels incised into the West Angela Member, which follow lithological strike. The largest of these bodies at South Flank is 4 km long, 400 to 600 m wide, and up to 40 m thick; however, due to erosion, there is only partial preservation. Lithologically, CzD1 detrital sediments comprise interbedded conglomerates, siltstones, and minor shales. All units contain hematite and lesser goethite in a red hematitic silt- and clay-rich matrix with 2- to 8-mm angular clasts of BIF, chert, and hematite. Locally at South Flank, these units reach ore grade (Fe > 59%) due to high hematite content. Apart from a typically massive, approximately 1 m thick basal conglomeratic unit, which may represent a paleoregolith horizon, CzD1 detrital sediments are moderately stratified. Based on this, the CzD1 is interpreted to have been deposited in a fluvial environment, with channels incised mostly into deeply weathered lower Wittenoom Formation.

Mineralization in CzD2 consists of angular to subrounded clasts of ferricrete together with pisolites set in a vuggy matrix dominated by goethite with minor white clay. It can locally reach up to 30 m in thickness and is commonly associated with thicker intervals of clay (>10 m) and commonly overlain by limestone and associated calcrete.

The CzD3 detrital sediments are the youngest preserved unit of the Cenozoic detrital sediments (Morris, 1994) and are widespread through South Flank and across the Hamersley province. At South Flank, colluvium composed of BIF ore fragments and pisolitic and welded pisolite sections of CzD3 detrital sediments can commonly contain Fe grades greater than 50%. Hence, these units can be locally economic, particularly where they overlie bedrock ore; however, they normally contain high levels of the deleterious elements SiO₂ and Al₂O₃. Lithologically, iron-enriched CzD3 comprises hematitic-goethitic pisolites of various sizes, which vary from uncon-solidated silt and soil-supported through to hard-welded pisolith, comprising pisolites cemented by a goethite matrix. The unit appears to have formed through erosion and redistribution of lateritic pisolites that formed due to weathering.

Recent weathering and extended exposure of iron ore and BIF above the water table have produced extensive ferricrete across the Pilbara (Morris, 1994). This zone of recent weathering has been designated the hardcap zone (e.g., Morris, 1980; Ramanaidou and Morris, 2010) and at South Flank comprises (1) an upper indurated zone of ferricrete commonly forming a 1- to 4-m-thick carapace, and (2) a lower hydrated zone rich in goethite, normally 20 to 30 m thick.

Ferricrete typically follows topography and normally occurs at South Flank as a surficial indurated carapace (Fig. 8D). Depending on the degree of hydration, the upper zone can range from pure hematite (with minor maghemite) through to a complex mix of hematite, maghemite, and goethite. Goethite in this zone is usually colloform or botryoidal in texture and forms crosscutting zones or laminations within voids. Primary banded textures are commonly preserved.

The lower hydrated zone ranges in thickness from 10 to 30 m and is best developed in rock types more susceptible to weathering (e.g., West Angela Member) and proximal to faults, where it can reach 70 m in thickness. The lower hydrated zone largely comprises porous goethite and vitreous goethite, with destruction of primary textures, particularly in proximity to the overlying ferricrete. The base of the hydrated zone is commonly marked by a notable decrease in S content and a reduction in vitreous goethite abundance (Banerjee et al., 2017).

A very large geochemical dataset for South Flank is available from resource drilling completed across the deposit. Three-meter composite samples have been collected from reverse circulation drill chips and analyzed by X-ray fluorescence at Bureau Veritas Laboratories in Perth. Loss on ignition (LOI) was determined by thermogravimetric analysis by the same laboratory. Selected major oxide element and trace element analyses of unmineralized and mineralized rock units at South Flank are presented in Table 2.

Bedrock geochemistry: Iron mineralization at South Flank is characterized by an increase in Fe and a reduction in SiO₂ compared to unmineralized lithologies (compare Figs. 10, 14; refer to Table 2). Primary lithology appears to exert an important control on the Fe grade and levels of deleterious elements such as P and Al₂O₃. Iron formations with the highest original Fe content and lowest levels of Al₂O₃ host the highest grade of iron ore. For example, the N3 unit of the Mount Newman Member contains the highest average primary (original) Fe content (35%) and the lowest Al₂O₃ content (0.5%), whereas N3-hosted mineralization contains the highest Fe grades (commonly greater than 64% Fe) with low SiO₂ and Al₂O₃ contents (Fig. 14). Primary lithology is also a control on iron ore mineralogy, with higher proportions of martite-dominant mineralization hosted by iron formations with the highest primary Fe contents. Shale-rich units within the West Angela Member and MacLeod Member generally host lower-grade, goethite-dominant Fe mineralization characterized by A₂O₃ contents of about 3% to 4%.

Iron shows a strong negative correlation with SiO₂ within the N3 unit of the Mount Newman Member (Fig. 15), reflecting the transition from iron formation with Fe content in the range of 25% to 35% to iron ore (Fe content greater than 50%). Phosphorous, Al₂O₃, and LOI contents are low in both unmineralized host rock and iron mineralization and increase moderately with increasing Fe grade, but are negatively correlated with Fe in samples with Fe content greater than 60%. At these high grades of Fe, major element content of ore samples reflects ore mineralogy. Goethite-dominant ore samples are higher in LOI, Al₂O₃, and P, compared to martite-dominant samples with equivalent Fe content (Fig. 15). The association of elevated P with goethite in iron mineralization at South Flank is also observed in other goethite-rich deposits in the Hamersley province (Dukino et al., 2000; Wells and Ramanaidou, 2011).

The spatial distribution of bedrock iron mineralization shows a strong, linear east-west pattern (Fig. 16), which is largely controlled by distribution of the Mount Newman Member. Zones of highest Fe grade and thickest mineralization correspond to cores of F2 synclines or settings where the Mount Newman Member is repeated through thrusting.

Hardcap geochemistry: The geochemical relationships within hardcap mineralization show a broadly similar pattern to those described for bedrock iron mineralization; however, there is increased spatial variability in both Fe and deleterious element contents. A comparison of the major element oxide geochemistry of hardcap mineralization to bedrock mineralization for each unit of the Mount Newman Member highlights significant variability between the two styles of mineralization (Fig. 17), with the following trends apparent: (1) both Fe and P are lower within hardcap mineralization; (2) SiO₂, Al₂O₃, LOI, and S are significantly elevated in hardcap compared to bedrock mineralization; and (3) hardcap mineralization is characterized by highly variable Fe, SiO₂, Al₂O₃, LOI, and S contents. Banerjee et al. (2017) interpret these trends to be the result of recent weathering and movement of the water table.

Cenozoic detrital sediments geochemistry: Typical lithologies from each of the detrital units at South Flank, namely iron pisolite-rich gravelly siltstone (CzD3), ferruginous pisolitic material with goethite cement (CzD2), and hematitic siltstone (CzD1), all have highly variable major element oxide contents (Table 2), reflecting the range of lithotypes comprising each of these units, which, in turn, reflects multiple source rocks and modes of deposition of the detrital units. Mineralized samples from each of these units show significantly lower levels of Fe and substantially higher SiO₂ and Al₂O₃ contents compared to bedrock mineralization.

South Flank shares several characteristics in common with BIF-hosted martite-goethite and martite-ochreous goethite deposits elsewhere in the Hamersley province (cf. Thorne et al., 2008; Angerer et al., 2014). These are summarized below and, in combination with the overall tonnage, Fe content, and wide geographic extent of martite-goethite and martite-ochreous goethite mineralization in the Hamersley province, define an important type of iron mineralization.

Martite-goethite mineralization is the dominant ore type in the Hamersley province (Fig. 2), and individual deposits form some of the largest accumulations of iron anywhere in the world. There is in excess of 15 billion tonnes (Gt) of martite-goethite iron ore in the central Hamersley province alone, defined as the area shown in Figure 4, hosted in up to 10 >1-Gt deposits. Iron mineralization can commonly occur over strike lengths of up to 50 km, as demonstrated by semi-continuous ore, hosted by the Brockman Iron Formation, extending from Packsaddle, in the west, through Eastern Packsaddle, to Jinidi in the east (Fig. 4). The size of individual deposits and overall extent of martite-goethite mineralization across the Hamersley province suggest uniform ore-forming processes across a 400- × 250-km-wide transect (Fig. 3). The extent, size and, therefore, contained Fe of martite-goethite mineralization within this transect are likely to be larger than other comparable iron ore districts globally (cf. Hagemann et al., 2016).

Brockman and Marra Mamba iron formations are the major host rocks to martite-goethite and martite-ochreous goethite mineralization. Boolgeeda Iron Formation, the youngest iron formation in the Hamersley Group, hosts minor iron mineralization, which is generally high in P and Al₂O₃ (Howard and Darvall, 2011). Subunits within both the Brockman Iron Formation (Dales Gorge and Joffre members) and Marra Mamba Iron Formation (Mount Newman Member) host the largest and highest-grade Fe orebodies. Preferential siting of ore within these subunits is interpreted to be the result of (1) higher initial Fe content of primary iron formation, and (2) high levels of permeability caused by multiple shale bands within the footwall and hanging walls of these units (Fig. 9), which enabled focused movement of fluids during mineralization.

While other units within the Marra Mamba Iron Formation can host Fe ore—for example, the lowermost Nammuldi Member in the Chichester Ranges (e.g., Hannon et al., 2005)—mineralization tends to be relatively thin and of low grade.

Regional structural controls on the basin-wide distribution of martite-goethite mineralization include the following: (1) regional-scale fold patterns that, in turn, control occurrence of Brockman and Marra Mamba iron formations within ~300 m of the surface, (2) local structural complexity, in particular the size and number of synclines present in an area and prevalence of crosscutting faults and dikes, (3) paleotopography influencing surface fluid infiltration and regional/local hydrological gradients, and (4) preservation versus erosion.

The importance of F2 fold structures for hosting high-grade martite-goethite iron mineralization was identified early in the exploration of the Hamersley province (Harmsworth et al., 1990), and they have formed an important structural target for subsequent exploration activity. A number of workers (e.g., Kneeshaw and Kepert, 2002; Hodkiewicz et al., 2005; Thorne et al., 2008, 2014) have highlighted the importance of structures at the deposit scale in controlling thickening of preferentially mineralized iron formations and in acting as conduits for iron-mineralizing fluids. Analysis of structural controls on iron mineralization at South Flank supports findings from these studies and further highlights the importance of the following: (1) synclines, especially with steep limbs, which are generally better mineralized than anticlines—this structural setting potentially enhances local movement of iron-mineralizing fluids; (2) crosscutting faults and dikes, which can act as a focus for mineralization with higher grade and thicker ore on one side of the controlling structure, indicating a horizontal vector to fluid flow, and can also act as aquitards forming boundaries to mineralization; and (3) thrust faults, which are strongly linked to iron mineralization at South Flank and act to thicken iron formations and to increase permeability.

Martite-goethite deposits are typically characterized by simple ore mineralogy and associated geochemical trends, which are strongly related to the primary composition of host iron formations. Detailed investigation of the Brockman Iron Formation-hosted Marillana deposit, located 50 km northeast of South Flank, has identified early siderite-pyrite alteration, related to the regional-scale Poonda fault, which is overprinted by later goethite-hematite Fe mineralization (Crowe et al., 2015). Similarly, Hannon et al. (2005) recognize two styles of mineralization hosted by the Nammuldi Member of the Marra Mamba Iron Formation in the Chichester Ranges: hypogene microplaty hematite overprinted by supergene martite-goethite and martite-ochreous goethite iron mineralization. Detailed logging of drill core in both weathered and unweathered parts of South Flank has not identified either of these alteration/mineralization styles.

Key processes in ore formation at South Flank are (1) conversion of magnetite-rich bands in iron formation to martite, resulting in high-Fe and low-Al₂O₃ ore with corresponding low LOI, and (2) alteration of silicate- and carbonate-rich sections of iron formation to goethite, with preservation of altered shale bands. These processes coincide with an overall reduction in the stratigraphic thickness of mineralized versus barren iron formation (cf. Morris, 1985) and have the effect of reducing SiO₂ content while increasing LOI and Al₂O₃, although it is noted that mimetic replacement of gangue minerals by goethite results in some volume preservation.

Vertical zoning of mineralogy and texture has been described in other martite-goethite deposits in the Hamersley province (Clout, 2002). This zonation is broadly developed at South Flank, where hardcap characterized by textural preservation and assemblages of hematite-goethite-maghemite overlies a hydrated zone comprising porous vitreous goethite where primary textures have been replaced. Iron mineralization is dominated by martite-goethite-ochreous goethite, with relative abundance of these minerals controlled by host iron formation as well as depth and water-table movements over time.

There is currently no published robust dating that determines timing of deposition of martite-goethite and martite-ochreous goethite mineralization in the Hamersley province. Identification of martite-goethite ore clasts in the Yarraloola Conglomerate and CzD1 has been used to suggest a Mesozoic or older age for mineralization (Kneeshaw and Morris, 2014). The major NE-SW–trending dike at South Flank, which crosscuts Ophthalmian-age structures, appears to predate mineralization, as it influences grade and mineralogy across the intrusive contact but shows no evidence of hydrothermal alteration. This dike has not been dated but most likely belongs to the 1075 Ma suite (Tyler, 1991), which potentially places an upper limit on the age of martite-goethite mineralization at South Flank.

Martite-goethite mineralization at South Flank is strongly associated with zones of structural complexity, notably upright to N-verging asymmetric synclines and associated low-angle reverse faults. This structural relationship has been documented in analogous deposits at Mining Area C (Bodycoat, 2007) and Hope Downs (Paquay and Ness, 1998); however, martite-goethite mineralization can also be developed in largely undeformed rock packages containing iron formations (e.g., Chichester Ranges; Thorne et al., 2008). This relationship between iron enrichment and deformation suggests either a syn- or postdeformation timing for mineralization. Absence of hypogene alteration zones around and in unweathered rocks beneath martite-goethite deposits in the central Hamersley province and lack of evidence of active structural controls on fluid flows during mineralization suggest a “passive” structural control (Angerer et al., 2014), in which earlier-formed structures were important in establishing thickened zones of iron formation and permeability-enhancing faults and synclines. In summary, in the absence of available age dating, absolute timing of martite-goethite mineralization in the Hamersley province is currently unconstrained.

Martite-goethite and martite-ochreous goethite deposits are widely developed across the Hamersley province, and are also associated with iron formations in the Pilbara and Yilgarn cratons (e.g., Yarrie-Nimingarra: Podmore, 1990; Koolyanobbing: Angerer et al., 2012; Weld Range: Duuring and Hagemann, 2013). Similar martite-goethite and martite-ochreous goethite mineralization has also been described in India (Mukhopadhyay et al., 2008), in the Quadrilátero Ferrífero, Brazil (Rosiere et al., 2008), and in other major global iron ore districts (summarized by Hagemann et al., 2016). In each of these regions, martite-goethite and martite-ochreous goethite mineralization is commonly developed as a weathering-related, supergene overprint of earlier, hematite-dominant hypogene mineralization, with variably developed wall-rock alteration, typically comprising assemblages of chlorite, carbonate, and magnetite (e.g., Hagemann et al., 2016). Ramanaidou and Morris (2010) distinguish weathering-related lateritic iron ores, including blue-dust ores, which overlie most iron deposits located in tropical climatic environments, from supergene mimetic martite-goethite ores. There are fewer examples of very large iron deposits composed entirely of martite-goethite and martite-ochreous goethite mineralization described in the literature. Examples include the lateritic Capanema mine, Minas Gerais, Brazil (Ramanaidou, 2009), and Hope Downs (Lascelles, 2006) and Mining Area C (Bodycoat, 2007) in the Hamersley province.

Detailed logging of extensive drilling laterally and at depth at South Flank has not identified hypogene iron mineralization or its associated wall-rock alteration in iron formations. Similarly, hypogene iron mineralization has not been described in other major iron deposits within the central Hamersley province, for example, Hope Down (Lascelles, 2006) and Mining Area C (Bodycoat, 2007), despite extensive drilling and mining of these orebodies. This suggests either that (1) hypogene iron mineralization in the central Hamersley province has been entirely replaced by subsequent supergene martitegoethite mineralization at South Flank and other iron deposits, (2) hypogene iron mineralization did not develop in the central Hamersley province and the occurrence of multiple giant martite-goethite deposits is the result of ore-formation processes specific to this region, or (3) hypogene iron mineralization has not yet been recognized or discovered in the central Hamersley province. Each of these hypotheses is tenable, and further research is required to constrain genesis of major martite-goethite iron deposits.

The discovery and assessment of the full potential of South Flank are the result of developing an exploration model for martite-goethite iron mineralization at the camp scale, based on geologic observations on similar deposits in the district, and a systems approach, which identified key processes and their associated geologic characteristics in ore formation. Acquisition and interpretation of high-precision airborne magnetic and gravity gradiometer data were critical in providing data in concealed areas to support observations from detailed mapping; however, deposit-scale targeting was achieved through integration of a number of geologic datasets. Results from wide-spaced drilling, which targeted the intersection of preferentially mineralized stratigraphic units and favorable structural settings for iron mineralization, demonstrated the potential for South Flank to host significant tonnages of highgrade ore. Importantly, initial drilling intersected structurally thickened Mount Newman Member, resulting from a combination of folding and associated reverse faulting. Without this, South Flank would be unlikely to contain significant tonnages of iron ore.

Subsequent exploration of South Flank has involved a 550-km drilling program to evaluate the full extent of the deposit for future mining. Due to the structural complexity and chemical variability of the ore in both Fe and deleterious elements (Al₂O₃, SiO₂), 50- × 50-m is the minimum level of drill-hole spacing required to provide the level of geologic data to support the evaluation. In areas of greater complexity or to support resource estimation and variography, 25- by 25-m drilling has been completed. Expenditure in exploration and resource drilling of South Flank has been significant; however, the extensive programs completed have enabled the full potential of the deposit to be realized.

South Flank and other giant (>1 Gt of ore) martitegoethite deposits in the central Hamersley province are characterized by the following: (1) common structural association with synclines and associated reverse faults, (2) preferential host-rock settings within the Mount Newman Member of the Marra Mamba Iron Formation and the Dales Gorge Member of the Brockman Iron Formation, (3) simple ore mineralogy and geochemistry, and (4) absence of hypogenestyle iron mineralization and associated wall-rock alteration. Giant martite-goethite deposits, of which South Flank is an example, potentially represent a distinct deposit style. While this type of mineralization is the most important in terms of overall resource endowment and current and future production from the Pilbara, as known resources of high-grade microplaty hematite (e.g., Mt. Tom Price and Mt. Whaleback) are approaching depletion, it is also the least studied, and key questions remain regarding its geology and genesis.

South Flank has many characteristics common to iron deposits proposed to have formed by supergene processes (e.g., Morris, 1985; Paquay and Ness, 1998; Taylor et al., 2001), namely structural setting, ore mineralogy and chemistry, and preservation of primary iron formation textures within ore zones. However, further detailed studies are ongoing at South Flank and other similar deposits in the Hamersley province to better understand ore formation, especially given increasing evidence of potentially multiple alteration/mineralization events in martite-goethite deposits (Hannon et al., 2005; Crowe et al., 2015). The focus of these studies is on dating mineralization events, understanding mechanisms of ore deposition, and determining controls on the spatial distribution of ore at the district and regional scales to develop a holistic system-based model.

BHP is thanked for permission to publish this paper. We acknowledge and recognize the geoscientists involved in the discovery and evaluation of South Flank, from the very earliest stages of reconnaissance through the past decade of intensive exploration and geologic evaluation. We particularly recognize Cam Leslie, Natalia Tolstova, Asmita Mahanta, Gaurav Dua, Garreth Carter, Chris Williams, George Joyce, and Clayton Wright for their contribution to the geology and safe completion of major drilling programs, and James Shaw for his support for the 2008 and 2009 phases of work. The manuscript was greatly improved as a result of constructive reviews by Steffen Hagemann and Jens Gutzmer, and guidance from Antonio Arribas, who are all gratefully acknowledged. We respectfully acknowledge the Banjima people—the traditional owners of South Flank.