OREGEO-01356; No of Pages 13
Ore Geology Reviews xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Ore Geology Reviews
journal homepage: www.elsevier.com/locate/oregeorev
Chatham Rise nodular phosphate — Modelling the prospectivity of a lag
deposit (off-shore New Zealand): A critical tool for use in resource
development and deep sea mining
Simon H.H. Nielsen a,⁎, Campbell McKenzie a, Alexandra Miller a, Gregor Partington a, Constance Payne a,
Elisa Puccioni a, Michelle Stokes a, Charlene Wildman a, Robin K.H. Falconer b, Ray Wood b
a
b
Kenex Limited, PO Box 41136, Eastbourne, Wellington 5047, New Zealand
Chatham Rock Phosphate Ltd., PO Box 231, Takaka 7142, New Zealand
a r t i c l e
i n f o
Article history:
Received 1 June 2014
Received in revised form 15 October 2014
Accepted 19 October 2014
Available online xxxx
Keywords:
Mineral prospectivity modelling
Mineral system approach
Mineral exploration
Resource modelling
Nodular phosphate
Chatham Rise
a b s t r a c t
After almost five decades of episodic exploration, feasibility studies are now being completed to mine the deepwater nodular phosphate deposit on the central Chatham Rise. Weights of evidence (WofE) and fuzzy logic
prospectivity models have been used in these studies to help in mapping of the exploration and resource potential, to constrain resource estimation, to aid with geotechnical engineering and mine planning studies and to provide background geological data for the environmental consent process. Prospectivity modelling was carried out
in two stages using weights of evidence and fuzzy logic techniques. A WofE prospectivity model covering the area
of best data coverage was initially developed to define the geological and environmental variables that control
the distribution of phosphate on the Chatham Rise and map areas where mineralised nodules are most likely
to be present. The post-probability results from this model, in conjunction with unique conditions and confidence
maps, were used to guide environmental modelling for setting aside protected zones, and also to assist with mine
planning and future exploration planning. A regional scale fuzzy logic model was developed guided by the results
of the spatial analysis of the WofE model, elucidating where future exploration should be targeted to give the best
chance of success in expanding the known resource. The development work to date on the Chatham Rise for nodular phosphate mineralisation is an innovative example of how spatial data modelling techniques can be used not
only at the exploration stage, but also to constrain resource estimation and aid with environmental studies,
thereby greatly reducing development costs, improving the economics of mine planning and reducing the environmental impact of the project.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
The New Zealand agricultural sector relies heavily on imported rock
phosphate-based fertilizer for farming efficiency, resulting in an opportunity for the production of locally based sources of phosphate. Sources
of rock phosphate in the SW Pacific region have been depleted over the
last 60 years, leading to rising imports of phosphate from further afield
and adding to the urgency of exploring the potential of local sources
(Cullen, 1979; Von Rad, 1984).
The only known phosphate mineralisation with economic potential
in New Zealand is a limestone gravel lag-based nodular phosphate deposit that is located in open waters at a depth of up to 400 m, more
than 400 km offshore off the east coast of the Canterbury region of the
South Island (Fig. 1). The phosphate deposit is located on the crest of
the Chatham Rise, a structurally simple bathymetric high rising over
the Hikurangi Plateau to the north and the Bounty Trough to the
⁎ Corresponding author. Tel.: +64 4 562 6253.
E-mail address: [email protected] (S.H.H. Nielsen).
south. It has been described as a replacement deposit comprising a
mix of Late Oligocene and Late Miocene chalk pebbles and hardground
rubble phosphatised in the Late Miocene and subsequently concentrated by erosion and selective removal of non-indurated chalk as the crest
of the Rise was karstified (Cullen, 1987; Kudrass and von Rad, 1984).
The Chatham Rise stretches due east from the Banks Peninsula as a
single structural entity for over 1000 km. The crest of the central Rise
plateaus at 400 m water depth, stepping down to more than 1000 m
east of the Chatham Islands (Wood et al., 1989). The slope to the
north to the Hikurangi Plateau and Trough has a relatively constant gradient, while the southern slope to the Bounty Trough appears as a series
of steps into what may be the failed rift arm of the Bounty Trough
(Fig. 1).
The Chatham Rise phosphate deposit has been subject to episodic
exploration since its discovery in the 1950s (Reed and Hornibrook,
1952). In the late 1960s Global Marine Inc. spent several weeks dredging on the Chatham Rise (Pasho, 1976), using the results from an initial
wide-ranging survey to target a second survey on the area with highest
nodule concentrations. The bulk of the exploration took place in the late
http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
0169-1368/© 2014 Elsevier B.V. All rights reserved.
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
2
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
Fig. 1. Chatham Rise overview map, with the Rise and surrounding basins indicated by bathymetric contours and local bathymetric highs indicated by their names. Mainland New Zealand
can be seen in the upper left corner. STF = Subtropical Front (simplified from Hayward et al., 2004). Grey arrows represent potential deepwater flow paths across the crest of the Rise.
1970s and early 1980s by collaborating German and New Zealand
scientists onboard the research vessels Valdivia (Kudrass and Cullen,
1982) and Sonne (von Rad and Kudrass, 1984). The surveys focussed
on the areas previously determined to be the most prospective.
Renewed interest by Chatham Rock Phosphate Ltd. in the 2010s led
to a series of surveys in 2011/2012, revisiting the areas surveyed by the
Valdivia and the Sonne. Subsequently, prospectivity and resource studies
based on new and historical data were undertaken. A 20-year seabed
mining permit was granted in the late 2013, and feasibility studies are
in the final stages of completion. The aim of these studies is to integrate
geological prospectivity with environmental studies and resource
assessment, to effectively develop the resource with minimal environmental impact, as well as plan future exploration where it is most likely
to increase the current resource and show a positive economic return.
Computer-based geographical information systems (GIS) provide a
variety of tools and statistical techniques that allow the mapping of
mineral prospectivity to be carried out and combined with environmental and economic risk analysis. Such techniques and tools have been
used by the petroleum industry for a number of years, and the mineral
exploration industry has taken this further with the help of spatial data
modelling using GIS (Bonham-Carter et al., 1989; Carranza, 2014; Deng,
2009; Harris et al., 2001; Lusty et al., 2012; Mejía-Herrera et al., 2014;
Partington and Sale, 2004).
A mineral system model adapted for phosphate deposition on the
Chatham Rise phosphate has been developed from the mineral system
concept normally applied to a range of hydrothermal and magmatic
styles of mineralisation (Hronsky and Groves, 2008; Lord et al., 2001;
Wyborn et al., 1994). This adapted model has been used to constrain
the modelling of the prospectivity of the central Chatham Rise through
the development of a range of predictive maps as proxies for the physical and chemical processes that have been combined to form the phosphate mineralisation. While the factors leading to the formation of
phosphatised lag gravels are obviously different from those leading to
many metallogenic ore deposits, the components of the mineral system
model are similar: There need to be a source of phosphate, a means to
transport it to the site of deposition, a trap to allow the phosphate rich
fluids to replace chalk pebbles and hardground rubble, and means to
concentrate and preserve the deposit.
The mineral system approach used with spatial data modelling differs from conventional ore deposit models by focussing on the similarities rather than on differences between deposit-types. Since the mineral
system approach is process based, it is not restricted to particular settings or mineral type, and has here been adapted to a mixed sedimentary and chemical process system. When applied to mineral deposits, the
mineral system approach requires identification at various scales of the
critical deposit-forming processes that can be mapped and that are
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
3
Fig. 2. Late Cretaceous contours (Wood et al., 1989) versus modern-day crest orientation, defined by outer bounds of the 500 m depth contour.
characteristic of the mineral system in question. The predictive maps
developed for the central Chatham Rise area represent aspects of possible source, transport, trap, and phosphate deposition. Developing mineral system concept for a nodular phosphate deposit poses challenges,
as unlike most metallogenic deposits, the deposit in question here is
surficial and laterally extensive, and may be formed by processes that
are not easily recognised or mapped.
The Chatham Rise nodular phosphate deposit is a unique mining
opportunity, with no similar deposits known locally. Globally, similar
deposits are rare but examples can be found onshore in the Gingin
Chalk of Western Australia (Simpson, 1920) and offshore Baja California
(D'Anglejan, 1967). While historical data are available open-file from
the New Zealand Petroleum and Minerals (NZPAM) database, as
well as from the National Institute for Water and Atmosphere (NIWA)
and the Institute of Geological & Nuclear Sciences Ltd (GNS Science),
surveying on the Rise has been done episodically and without connection or consistency between the exploration programmes. Therefore,
the first task and the most time consuming one was to build a digital database of all available geological, geochemical and geophysical data.
Also, as the New Zealand outer continental shelf regions are generally
underexplored, data are sparse. The resulting GIS covers more than
30,000 km2 of the Chatham Rise crest region, with the spread of shallow,
high-resolution data limiting quantitative estimation of prospectivity to
the central region (Fig. 1).
2. Geology of the Chatham Rise
The oldest rocks of the Chatham Rise are indurated greywacke and
argillite of the Jurassic–Cretaceous Torlesse Supergroup affinity. These
rocks were later rifted into east–west oriented half-grabens generally
hinged to the north during the second episode of the Early Cretaceous
Rangitata orogeny (Wood et al., 1989). Minor movement has occurred
on the same faults during the Cenozoic. Movement along a second set
of north–south trending sinistral transcurrent faults appears to give
the Rise the appearance of a gentle bend (Fig. 1). After the Rangitata
rifting, the first sediments deposited were basin-fill initiated around
100 Ma, where 1000–2500 m or more of sediments were deposited in
newly formed half-grabens, with the thickest accumulations in the
western region of the Rise (Wood et al., 1989), leading to a relatively
complex east–west oriented crest structure which is generally oriented
similarly to the modern crest (Fig. 2).
The geology of the study area is summarised in Fig. 3. The early
Cenozoic succession is block faulted and down-thrown 10–65 m towards the north (Falconer et al., 1984). Rocks sampled from this time interval generally come from the Matheson Bank region, and include
cross-bedded foraminiferal limestone from a shallow (b100 m) water
depth. Radiolarian ooze and nannofossil foraminiferal chalk has also
been recovered from the study area. The chalk appears similar to the
Amuri limestone of Northern Canterbury, while the late Cenozoic succession is very thin or absent over large parts of the central Rise
(Wood et al., 1989). Several hundred metres of Miocene sediment
may have been deposited and subsequently eroded as uplift began in
the Late Oligocene with the development of the Alpine Fault plate
boundary (Field and Browne, 1989), resulting in less than 50 m of Miocene–Pliocene preserved in the study area. The Oligocene chalk may
have been locally exposed as well (Kudrass and von Rad, 1984). This is
overlain by a thin veneer of Pleistocene glauconitic-foraminiferal sand
containing phosphatised nodules of the eroded Cenozoic limestone
(Wood et al., 1989).
The paleoceanographic setting of the Chatham Rise is mainly described from scientific deep sea drilling reports and based on sediment
cores on the deep southern flank south of Mernoo Bank (DSDP Site 594,
Kennett et al., 1983) and northern flank north of Matheson Bank (ODP
Site 1125, Carter et al., 1999). The Chatham Rise is situated in a transitional region between oceanic and terrigenous influences (Kennett
et al., 1983). It is described as a significant barrier to ocean circulation,
with a crest that corresponds with the Subtropical Front, a convergence
zone between Subtropic Water flowing around northern New Zealand
as the East Cape Current, and Subantarctic Water flowing around southern New Zealand as the Southland Current (Heath, 1976). A slow
(0.08 m/s on average) northwards net drift rises from 700 m depth to
flow over and across the deeper parts of the saddle (Heath, 1976). A
zone of high productivity lies over the Rise following the Subtropical
Front (Fig. 1) adding a significant biogenic component to the many
sources of sediment to the crest of the Rise, namely (Carter et al.,
1999): biopelagic snow from vigorous water mixing along the Subtropical Convergence; suspended terrigenous sediment supplied by rivers
feeding into the East Cape and Southland Currents; distant airfall tephra
from central North Island volcanic eruptions; and occasional ice-rafted
debris from icebergs.
The position of the Subtropical Front likely remained fixed during
glacial/interglacial changes. However, during glacial periods, more
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
4
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
Fig. 3. Geology of the study area.
Simplified from Falconer et al. (1984).
vigorous intermediate water and more sluggish deep water circulation
would lead to relatively oxygen-starved and organic matter rich conditions on the sea bottom of the Rise (Carter et al., 1999). The main
depocentres for sediment removed off the top of the Rise appears to
be the topographically low regions along the north and south flanks,
where thick successions of rhythmic upper Neogene hemipelagic and
biopelagic sediments have been trapped and cover the last 10 Ma
(Carter et al., 1999; Kennett et al., 1983).
The phosphate mineralisation on the Chatham Rise has been described as thin, unconsolidated, surficial or near-surface gravel enclosed
in a matrix of completely unconsolidated glauconitic sandy mud
(Cullen, 1987). The nodules consist of indurated Oligocene as well as
Lower and Middle Miocene chalky limestone, which are suggested to
be phosphatised in the Late Miocene (Zobel, 1984). Phosphate
mineralisation has been mapped as intermittently straddling 400 km
of the longitudinal crest of the Chatham Rise from 177° E to 177° W.
However, the only region described as having commercial potential is
found between 179° E and 180°, where exploration efforts have been
concentrating (Fig. 1). Smaller areas of mineralisation may occur around
Matheson Bank (Cullen, 1987) and possibly elsewhere, but exploration
to date has not verified this. The thickness of the nodule layer is 20–
30 cm on average, but can be just a few centimetres to the thickest
part of the deposit at 70 cm (Cullen, 1987). The areas surveyed in detail
by the Sonne and Valdivia, covering about 380 km2, are believed to
contain the potential for up to 30 million tonnes of rock phosphate.
Extrapolation to cover the entire area where phosphate mineralisation
could occur on the central Rise suggests the possibility for about
100 million tonnes of rock phosphate (Cullen, 1987).
The phosphate nodules comprise indurated limestone replaced by
phosphate minerals, deposited as uncemented gravel clasts, with a generally subangular to subrounded grain shape. The size of the nodules
ranges from 2 mm to greater than 150 mm (coarse sand to cobbles),
but with the peak in the size range between 10–40 mm (Cullen,
1987), which have been dated by the age range of enclosed foraminifera
to Lower and Middle Miocene (Zobel, 1984). This is close in age to the
phosphatised, nodular interval at the top of the Upper Eocene–Lower
Oligocene Amuri Limestone of northern Canterbury (Wood et al., 1989).
The phosphate nodules display two phases of boring and burrowing
benthic fauna, also a sign of a complex history. The first phase consists of
filled burrows that riddle the phosphatised limestone of most nodules,
while a later phase of boring is reflected in the open tubular cavities
that penetrate and define the outline of larger nodules (Cullen, 1987).
3. Mineral system model for phosphate deposition on the
Chatham Rise
3.1. Source of phosphate
In the Chatham Rise region, a component of the phosphate along
with carbonate may be sourced from sluggish dysoxic deepwater flow
in a highly productive, stratified ocean before the establishment of the
modern frontal system and ocean circulation system (Fig. 4; Hayward
et al, 2004). Such warm conditions with sluggish deep-water flow
may have persisted throughout the Eocene and during the Early
Miocene (Hayward et al., 2004; Nelson and Cooke, 2001) and allow
phosphate to become super saturated in the warm water.
In recent phosphate forming environments there is low sedimentary
input and high organic productivity associated with seasonal upwelling
of nutrient-rich deepwater (Waples, 1982). Today, upwelling is associated with the Subtropical Front, which is situated over the Chatham
Rise and the deflection of the southern Southland Current and the
northern East Cape Current towards the east (Fig. 1). The presence
of such front leads to increased turbulence and mixing of surface waters
leading to increased organic production and higher influx of phosphatebearing organic matter to the Rise. The past presence of a front is indicated from latitudinal differences in benthic foraminifera assemblages
on either side of the Rise from at least early Late Miocene onwards
(Hayward et al, 2004), although a proto-front may have been present
during the first Antarctic glaciation in the Late Eocene and Early to
mid Oligocene (Nelson and Cooke, 2001). Since decreased oxygen levels
and increased input of organic matter may be linked to times of vigorous frontal activity on the Rise, it may be that the main phosphatising
events are limited to global cooling events and Antarctic glaciations
(Nelson and Cooke, 2001).
The source of phosphate may ultimately be upwelling systems that
may not easily be mapped as regional-scale features that focus
nutrient-carrying deep-water flow into the phosphatised areas. What
can be mapped as a proxy instead are regions where the bathymetry
allow larger volumes of organic-rich deep water to pass such as regional
saddles where slow deepwater flow has been suggested (Heath, 1976).
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
5
Fig. 4. Neogene water column structure and depositional model. A) Limestone deposition in quiet, oxidised environment; B) suboxic conditions at Rise crest leading to dissolution and
phosphatisation; C) enhanced circulation leading to oxic conditions, terrigenic deposition and glauconitisation; D) erosion and sediment redistribution under modern conditions.
Modified from Hayward et al., 2004.
Such areas are suggested where the width of the crest is reduced
(Fig. 1).
3.2. Transport of phosphate from the source to the trap site
If the source of phosphate can be viewed as the deep currents bringing organic matter from upwelling zones at the sea surface to the
Chatham Rise, the bottom contour-following currents are believed to
be the mechanism for redistributing the organic matter to the trap
sites where the phosphate replaces the lag gravel on the Rise. The direction and relative intensity of these currents depend on the grade of
slope and the presence of features that may constrict flow such as
fault scarps and incisions. While the Rise in general may have been relatively stable since the Late Cretaceous to early Palaeogene rifting
ceased (Cook et al., 1989), Oligocene faulting indicated in seismic profiles (Falconer et al, 1984) would have created seabed relief for currents
to follow and exposed high ground for hardground formation and
phosphatisation (Kudrass and von Rad, 1984). Regional faults are
mapped from relatively low-resolution industrial seismic surveys
(Wood et al., 1989) showing a general east–west trend of the major
lineaments (Fig. 2). A higher resolution fault and lineament set was
derived from the 1984 Sonne shallow seismic data (Falconer et al.,
1984; Falconer, pers. comm.) and recently acquired high-resolution
multibeam bathymetry (Wood., 2012; Wood et al., 2012) (Fig. 2).
In times of sluggish ocean circulation and/or sea level highstands,
bottom currents may be weak enough to allow transport and draping
of fine sediment on exposed highs. During times of more intense
ocean circulation and/or sea-level lowstands, these flows may be strong
enough to waft unconsolidated fine sediment away especially from
slopes and high ground.
3.3. Formation of trap
The Chatham Rise has been covered by subsequently lithified and
eroded Late Eocene, Oligocene and Early Miocene chalks and foraminiferal limestones (Fig. 3), which are the host-rocks for phosphatisation
since nodules contain remnants of chalk and foraminifera from the
age and environment represented by the chalk deposits (Cullen, 1987;
Kudrass and von Rad, 1984). In marine carbonate shelf environments,
extensive deposits of nodular phosphate are associated with pauses in
sedimentation and consequential hardground formation (Pedley and
Bennett, 1985). In addition, oxygen-starved conditions are necessary
for phosphate replacement and nodule formation (e.g. Arning et al.,
2009). The time-equivalent and laterally extensive Amuri Limestone
of Canterbury contains horizons of – and is overlain by – phosphatised
nodules (Morris, 1987), suggesting episodes of region-wide dysoxic
bottom conditions allowing extensive phosphatisation. The Oxygen
Minimum Zone is today situated at about 1000 m depth (Fig. 4;
Hayward et al., 2004), indicating that extensive phosphatisation has
not been possible on the Chatham Rise since the modern water column
structure was established.
The presence of phosphorous rich bottom water and oxygen minimum conditions may pass over an area without a trace in the geological
record except for the subsequent presence of phosphate mineralisation.
The intervals of phosphate production may be brief, and possibly alternating with periods of sediment bypass and winnowing away the matrix sediment, further concentrating the phosphatised hardground
rubble (Pedley and Bennett, 1985).
During the Eocene and Early Miocene times of sluggish, dysoxic
deep-water flow, conditions were optimal for deposition of chalky
oozes that would dilute deposited phosphate. Cooling of the surface
water and intensification of upwelling at Subtropical Convergence
would halt carbonate production leading to sediment starvation and
hardground formation at the sea bed. When bottom current intensified,
removal and dissolution of soft chalk would concentrate resistant and
phosphatised hardground formations. Episodes of intense upwelling at
the surface would increase organic matter contribution, leading to
pulses of organic matter to the sea bed and widespread replacement
of the exposed chalk and hardground by phosphate, and subsequent redistribution and removal by bottom currents. Major periods when this
is likely to have happened repeatedly are during times of rapid changes
in oceanic climate, such as the Eocene/Oligocene glaciation, the Late
Oligocene warming and the late Early Miocene end of the Miocene
Warm period (Zachos et al., 2001).
Since phosphate mineralisation straddles the crest and required
dysoxic waters to form, the time of formation may be limited to the
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
6
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
time before the modern well stratified water column was established,
when sluggish, dysoxic waters were able to flow through saddles on
the crest of the Rise (Hayward et al., 2004; Heath, 1976) (Fig. 1). The
lower age limit is determined by the earliest deposition of carbonates
on the Rise, while oxygenation of shallow and intermediate water
masses would not be established until the modern ocean currents
came into play in the Late Miocene when Antarctic Intermediate Waters
first flooded the Chatham Rise. This sets an upper age limit on the phosphate entrapment on the Crest of the Rise to be no younger than the
Late Miocene. Predictive map themes of trap formation should therefore
be based on the extent of chalk facies of Eocene to early Late Miocene
age.
3.4. Deposition and exposure of phosphate nodules
The physiochemical processes leading to phosphatisation by replacement of hardgrounds depend regionally on the presence of rapid
physiochemical changes in conditions to allow phosphate to be deposited, exposed carbonate to be replaced, and locally on high surface
area to volume ratios for effective adsorption of phosphate from the
mineralising bottom waters. Irregularities in the karst surface would
lead to more exposed areas becoming more phosphatised than others.
Since the phosphatised carbonate is resistant to weathering, the effect
would be self-reinforcing leading to protruding highs that would
break down mechanically through increased slope instability or, unique
to this deposit, by the impact of iceberg keels during the Pleistocene glaciations. There are no indications of lateral transport otherwise with
currents not being strong enough to move the phosphate nodules
(Kudrass and von Rad, 1984). The basins between the protruding
karst surfaces are traps for fine siliciclastic cover sediment, leading to
the presence of karst and deposits of phosphatised nodules to stand
out as areas of mappable local rough relief.
Another factor in the surface distribution of the deposit is the water
depth and slope orientation. Deeper waters are likely sites for accumulation of siliciclastic sediments as deep currents would slow down as the
accommodating water column volume increases, and slides and slumps
tend to move sediments downhill. The flanks of the Rise would be host
to sediment drapes as well. Orientation of slopes will preferentially facilitate or hinder sedimentation depending on their facing in relation
to dominating bottom current directions, leading some slopes to be potential traps for cover sediments while other would be swept clean
frequently.
4. Spatial analysis methodology
Two prospectivity models were produced. In order to quantitatively
define the most important predictive parameters for phosphate
mineralisation, a WofE model was established over the area of highest
data concentration (Fig. 1) and which covers the most sampled seabed
sedimentary units. The results of this model were applied to more extensive fuzzy logic model covering all of the seismically mapped area
of the Chatham Rise (Fig. 3).
4.1. Weights of evidence spatial data analysis
The WofE spatial modelling technique relies on the spatial analysis
of each predictive map representing different aspects of the mineral system model using a set of training points of known areas of phosphate
mineralisation. It was originally developed for medical diagnosis
(Spiegelhalter and Knill-Jones, 1984), but has been adopted for mineral
prospectivity modelling (Bonham-Carter, 1994; Bonham-Carter et al.,
1989). The predictive maps are commonly binary, divided simply into
areas of favourable and unfavourable conditions (Deng, 2009). The
spatial correlation is calculated based on the area covered by the data
variable tested and the number of training points falling within that
area. A contrast value (C) is then calculated based on the ratio between
the areas of the favourable and unfavourable conditions. A simple student test (StudC) is the ratio of C over its standard deviation (Cs). In
this study C values N 2 and StudC values N 1.5 are considered significant
positive correlations (Bonham-Carter, 1994; Peters and Partington,
2008).
The modelling was carried out using the Spatial Data Modelling tools
for ESRI's ArcMAP 10 GIS software (Sawatzky et al., 2010). The study
area grid was set up at a 250 × 250 m resolution, the finest resolution
the model should be viewed at based on data availability and coverage.
The grid covers the parts of the seismic facies map of Falconer et al.
(1984) that intersected the majority of the available exploration data,
spanning a total area of 6539 km2. Although the phosphate may originally have been a single extensive deposits, subsequent erosion and
deposition of cover sediments have resulted in irregular surface distribution of phosphate nodules (Kudrass and von Rad, 1984). Therefore,
based on observations of deposit irregularity and the geostatistical analyses underlying resource estimates (Kudrass, 1984; Sterk, 2014), a unit
cell size of 2 km2 was chosen for the analysis and modelling. This resulted in a prior probability of 0.01407 or a 1.4% chance of randomly finding
mineralisation within any single 2 km2 area before additional evidence
is applied to the study area.
No phosphate mines exist in the area, so a set of training points was
developed from sample points with high concentrations of phosphate
nodules. All points with phosphate coverage calculated to be above
50 kg/m2 were initially chosen. From these 424 points, 46 training
points were chosen by filtering out points that were a unit cell radius
or less from other points, while ensuring a good geographical spread
of training data.
4.2. Fuzzy logic data analysis
The fuzzy logic spatial data modelling technique is an easily understood and effective method used for prospectivity modelling in less
data rich study areas (Bonham-Carter, 1994; Carranza, 2008). Fuzzy
logic does not rely on training data, but uses expert knowledge rather
than spatial statistics to map prospectivity. This knowledge is often
expressed in useful but imprecise terms such as near, sometimes, and
maybe. Fuzzy logic provides a means to convert such semantic descriptions into a numerical map. Instead of simple Boolean logic True and
False situations, fuzzy logic allows degrees of truth expressed as a membership function between 0 (no membership) and 1 (full membership).
Membership values below 0.01 will exclude the predictive map in
question; at 0.1 the map will only be included if all other maps give
high weights to the cell in question. Values of 0.1–0.5 will deem a region
somewhat unfavourable depending on the total value. A membership of
0.5 indicates a perfectly ambiguous situation and the map will have no
impact on the final probability value. Values of 0.5–0.9 will weigh a
map favourable in the final prospectivity map.
The Overlay Spatial Analyst Tools that are part of ESRI's ArcGIS software (Sawatzky et al., 2010) were used to create the fuzzy logic model.
The fuzzy logic model study area covers the entire central Chatham Rise
region from the east side of Reserve Bank to the northwest of Matheson
Bank near the slope and shelf region of the Chatham Islands (Fig. 1). The
total area is 30,993 km2, covering the full extent of the Falconer et al.
(1984) seismic facies as well as the majority of the area samples by
the 1960s surveys by Global Marine (Pasho, 1976) (Fig. 3).
5. Weights of evidence spatial analysis of predictive maps
representing proxies of the mineral system model
5.1. Analysis of predictive maps representing phosphate source
Deep water currents carrying phosphate-rich waters from surface upwelling are here considered the source of phosphate for mineralisation.
The paths of general deepwater flow can be mapped based on the general
shape of the Chatham Rise. The Viewshed tool of ArcMap was used to
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
define areas lying within and facing the broad north–south trending saddles that cross the central and eastern Rise (Fig. 1). For the WofE model,
95% of the training points occur in the broad area that is either within
or facing the central saddle, giving a C value of 2.7 and a StudC value of
3.7 (Table 2). This result was applied to the fuzzy logic model, in which
areas facing the saddles were given a weight of 0.7, and areas outside
were weighted 0.3 (Table 3). The regional bathymetric dataset was not
of good quality, and it is recommended to update this if more detailed
studies are needed.
5.2. Analysis of predictive maps representing phosphate transport
Structural features are a very important control on the transport and
deposition of mineralisation in a majority of mineral systems, for
example by channelling and focussing mineralising fluids along faults.
However, as the mineral system model here involves a surficial deposit
formed through surface exposure and deposition, transport mineral system processes must be mainly controlled by topography and interaction
with bottom water currents. These features have some overlap with
trap mineral system processes. Faults create topographic relief and can
act as a major influence on geomorphology, which on the Chatham
Rise can promote exposure to phosphate rich waters (trap feature) as
well as create gradients for bottom currents to follow (transport
feature). Fault data were compiled from regional and local seismic
mapping and found to have highest spatial correlation with a 1400 m
distance buffer, capturing 93% of the training data with a C value of 2.7
and StudC of 5.6 (Table 2). This result confirms that although the
mineralisation is formed by surface processes, underlying geological
structures still have a significant control on the distribution of the
phosphate. This was extended to the fuzzy logic model using the
regional fault data only and assuming that the fault zone itself as well
as terrain distant to faults would have no significant impact on
prospectivity; 0–400 m distance and distances over 1400 m were
given a weight of 0.5 (neutral), while 400–1400 m was given a weight
of 0.7 (Table 3). As the fault set is rudimentary with detailed work
only in the central WofE model area (Fig. 2), no other aspects of faulting
such as age, orientation or intersects, were considered.
Water depth and orientation of slopes are important controls on the
direction and strength of dominant bottom current flow directions,
which facilitate hardground formation, phosphatisation and nodule exposure. These factors overlap somewhat with the trap predictive maps,
but are regarded at a larger scale. At deeper levels contour-following
currents dominate leading to the deposition of muddy siliciclastic drifts.
The relatively shallow crest is influenced by episodic strong currents
keeping it generally free of sediment cover, facilitating hardground formation in the past and keeping the nodule lag clear of sediment cover
today. A categorical test of reclassified depth intervals found that 81%
of the training points are found in shallow water, but not the shallowest
interval as this is outcrops of siliciclastic basement.
Slope orientation influences prospectivity in conjunction with water
depth, as slopes control the intensity and direction of bottom currents.
Also, slope directions subjected to the most persistent currents will
more likely form hardgrounds, become phosphatised and be clear of
cover sediments today, while slopes in current shadows will be sediment traps and thereby be detriment to phosphatisation and nodule exposure at the surface. The slope orientation was tested on a high
resolution bathymetric grid of limited extent. E–SE facing slopes captured 32% of the training data with good spatial correlation. Since
these variables are interdependent, they were combined with fuzzy
OR, essentially extending the range of favourable water depth when
the slopes are oriented favourably as well. The combined bathymetry
and aspect layer captured 89% of the training data with a C value of
1.3 and a StudC of 3.1 (Table 2). This was extended to the fuzzy logic
model by setting the weight of areas without favourable bathymetry
or slope aspect to 0.3, while areas with either or both variables being
favourable weighted by 0.7 (Table 3).
7
Sloping terrain facilitates bottom current flow which keeps the
hardground surface sediment free and transports the elements needed
for phosphatisation. Flat terrain is detrimental to nodule prospectivity
as this indicates sediment-filled basins. Too steep slopes may also be
detrimental as these may be drifts or sediment wedges at scarps, or
on the regional scale it indicates terrain on the actual flanks of the
Chatham Rise in deeper water and covered with sediment drapes.
This was reflected in a categorical analysis of a reclassified slope map,
which gave a poor correlation with the shallowest slope angles and
negative correlation or no training data in areas with relatively steep
slopes. Slopes based on the regional bathymetry datasets had good correlation with 84% of the training data at angles below 0.1° (C = 1.7,
StudC = 4.5). This was extended to the fuzzy logic model but using a
maximum angle of 0.2° with a weight of 0.8, while steeper slopes
were considered unprospective with a weight of 0.1.
5.3. Analysis of predictive maps representing phosphate traps
The actual precipitation of phosphate depends on the convergence
of several factors including low oxygen levels, increased phosphorous
concentration and/or flux to the area (for example from mixing upwelling cold and P2O5 saturated deep water with relatively warm surface
water at a front), paths of local bottom water currents, and lack of
other sedimentation. The phosphate nodules were formed as a replacement and growth around carbonate hardground gravel clasts on or near
the seabed. The predictive maps created outline the distribution of
carbonate on the seabed as well as the distribution of local ridges. A categorical test of the seismic facies from Falconer et al. (1984) resulted in a
strong positive correlation between training data and the extent of Oligocene chalk as well as with Miocene chalk on the south side of the
crest. A negative correlation was found with Late Eocene chalk and
chert, while a negative correlation with Miocene chalk north of the
crest may be the result of the facies map being oversimplified. Combing
the Oligocene and southern Miocene facies with logic OR resulted in a
predictive map covering 96% of the training data with a C of 3.5 and a
StudC value of 4.9. For the fuzzy logic model, facies older than Late
Eocene were given a very low weight of 0.01. The Late Eocene marl
was weighted neutrally at 0.5. Early Oligocene chalks were weighted
0.8, while the softer Miocene chalks were weighted slightly lower at
0.6. Pleistocene cover sediments were weighted 0.1.
Local highs were mapped at a finer resolution from the bathymetric
data than was done for the transport process predictive maps. The analysis used the Flow Focus tool of ArcMap, which assigns a value to a cell
based on how many of the neighbouring cells have higher values than
the centre cell. In a bathymetric grid, the cells with the lowest Flow
Focus values must be local high points. These were then buffered with
a distance buffer and found to have best correlation at 2000 m: 96% of
the training points were captured this way, with a C value of 4.5 and a
StudC of 6.2 (Table 2). The quality and resolution of the regional bathymetry were not good enough to produce a meaningful predictive
map based on local bathymetry for the fuzzy logic model.
Also mapped were the presence of ridges and the position of the
crest axis interpreted at the time of the hardground formation and
phosphatisation. This was done on grids based on unconformity depth
contours from regional seismic data from the Chatham Rise (Wood
et al., 1989). These grids, which can be considered to be maps of
palaeo-bathymetry, were inverted and the ridges were outlined using
the Flow Accumulation tool of ArcMap. Due to the poor coverage of seismic data, the result was not useful at the scale and coverage of the
prospectivity models. However, the results did map the structural trends
of the Rise, which were useful for later resource estimation modelling.
5.4. Analysis of predictive maps representing phosphate deposition
Since the phosphate mineralisation is uniform and distinctly different from the host sediment, phosphate content has generally been
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
8
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
derived from the relative proportion of the coarse sediment fraction
(N1 mm), extrapolated from analysis of chemical phosphate on size
fractions of bulk samples. Consequently, phosphate grade is calculated
as the mass of nodules per square metre, based on the weight of the
coarse fraction modified by the percentage of phosphate nodules in
the fraction and the size of the area sampled. The phosphate nodule
grade maps were not used in the WofE model because the four main
sampling programmes used at least four different sampling methodologies as well as different sample processing techniques and a thorough
analysis of the implications were not available at the time of modelling.
Instead, observed percentage of nodules in samples, generally observed
shipboard, was used as an indicator of economic nodule abundance. The
point data interpolated to an area grid using inverse distance squared
weighting with a 5 km search radius, using circular search similar to
the procedure used for resource estimation of the Chatham Rise
phosphate (Kudrass, 1984; Sterk, 2014). Cut-offs were selected
from the sample value histogram. Combining nodule abundances by
weight N15% captured 93% of the training data with a C of 3.7 and a
StudC value of 6.1 (Table 2).
For the fuzzy logic model, the Global Marine dredge sample set was
added to the sample data, which were also converted to kg/m2 grade
estimates, constrained by a new resource estimate (Sterk, 2014).
Based on the structural orientation inferred from the Flow Accumulation analysis described above, the study area was divided into five sectors with distinct structural orientations. These were used to guide
inverse distance squared weighting interpolation of the grade data
using oriented 12 km by 6 km search ellipses. A cut-off grade was assumed to be 20 kg/m2 and to maintain the robustness of the interpolation only areas interpolated from at least 5 sample values were
included. The grid was also clipped by unprospective geology (outcrops
of basement and Palaeogene clastic sediment, as well as Pleistocene
cover sediment). Robust estimates of N20 kg nodules/m2 were given a
weight of 0.8, while samples outside of this were given a neutral weight
of 0.5 (Table 3).
6. Weights of evidence and fuzzy logic modelling
The spatial analysis of each predictive map is as important for exploration targeting as the spatial data modelling itself, as the analysis identifies data that best predicts mineralisation as well as interrelationships
between datasets that can improve exploration targeting and improve
future data acquisition (Partington, 2010). The analysis also highlights
where more research may be needed and where additional data can
be collected to improve the model. For example, what is the spatial relationship between the inferred ridges on the Chatham Rise crest and
the extent of mineralisation? Are there deposits associated with past
ridges when the Chatham Rise had a different configuration from
today? The most important variables that can be mapped for predicting
nodular phosphate deposits on the Chatham Rise based on the WofE
spatial analysis are listed in Table 1.
A WofE prospectivity model was created using the predictive maps
described above, representing all stages of the phosphate nodule mineral system model (Table 2). For example, spatial association with saddle
areas provides information on source; associations with faults, geology
and features of crest architecture provide information on transport
pathways and traps; and distribution of phosphate nodules provides information on the efficiency of deposit formation. The final predictive
maps were chosen as having the best regional coverage, a significant association with the mineral system model considered, being useful for
establishing parameters and weights to use in the fuzzy logic model,
and with minimal duplication of the predictive map patterns. The
predictive themes were added to the final model after their spatial correlation values were calculated (Table 2).
Based on the WofE map themes, similar maps were developed
for the laterally more extensive fuzzy logic model. Weights were
constrained by the C values derived from the weights of evidence
Table 1
Summary of the key exploration criteria derived from spatial correlation analysis for phosphate nodule deposition on the Chatham Rise.
Exploration variable
Chatham Rise nodular phosphate deposition
Bathymetry
Shallow water depth, eastern and south-eastern slope aspects,
local highs, association with regional saddle features.
Seismic facies
Chalk-containing surface facies (Oligocene and Miocene units).
Pleistocene cover is negatively correlated, as is Eocene and
older, generally siliciclastic, facies.
Lithology
Proportion of phosphate nodules in the coarse sediment
fraction
Faults and lineaments Exposure from fault offsets, seabed tilting, linear zones
of weakness. Iceberg scouring likely too young to influence
overall phosphate nodule distribution.
model as well as on the impact of the variable in question missing
from an area (for example not being close to a fault does not preclude
phosphatisation, while being in areas of basement outcrop do). The
fuzzy logic predictive themes and their weights are summarised in
Table 3.
6.1. Weights of evidence model results
The final WofE model was developed using the Arc-SDM toolbox developed for ArcGIS 10 (Sawatzky et al., 2010). Inputs to the model are
the selected predictive maps, their spatial statistics and the training
data. The modelling produces up to five grids that measure geological
potential and uncertainty. To be able to query the combinations of inputs that produced the prospectivity result, a grid response map was
produced which contains the intersection of all of the input themes in
a single integer theme, called a unique conditions grid. Each row of
the attribute table contains a unique row of input map values, indicating
which predictive maps contributed to the final prospectivity value of a
given grid cell. Any combination of map variables can thereby be queried to assist the user in establishing the final targets. Especially important for exploration, the areas missing different types of data can also be
identified. The unique conditions grid can be combined with other
maps, for example of exploration expenditure, to simplify cost analysis
of future exploration needs (e.g. Kreuzer et al., 2014).
A confidence map was also produced, showing the confidence that
the reported post-probability is not zero for a given grid cell. This
is achieved by dividing the post-probability with the total standard
deviation, an approximate Student T test (Sawatzky et al., 2010). The
variances of the weights and variance due to missing data are summed
to give the total variance of the post-probability in these maps.
Combining the unique conditions grid with the confidence maps produced during the weights of evidence modelling allows areas where
the confidence in the result is low and where data are missing to be
mapped. Thereby the model can be used not only to target new prospective ground, but also to plan new exploration programmes where new
or additional data are required.
The final stage of the modelling involved reclassification of the postprobability maps to define areas of high-priority targets for phosphate
mineralisation for sampling and resource estimation studies. Unique
to this model is the presence–absence nature and extensive coverage
Table 2
Predictive maps used in the WofE model, with weights and spatial correlation values.
Map
Variable
W+
W−
C
StudC
Saddle region
Shallow bathymetry and slope aspect
Gentle slopes
Proximity to faults
Oligocene and Miocene chalk near surface
Proximity to local highs
Area with significant nodule abundance
Source
Transport
Transport
Transport
Trap
Trap
Deposit
0.5
0.3
0.6
0.9
0.9
1.6
1.2
−2.2
−0.9
−1.1
−1.8
−2.6
−2.9
−2.4
2.7
1.3
1.7
2.7
3.5
4.5
3.7
3.7
3.1
4.5
5.6
4.9
6.2
6.1
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
Table 3
Predictive maps used in the fuzzy logic model, with classes and fuzzy weights.
Map
Variable
Fuzzy class
Fuzzy weight
Saddle region
Source
Bathymetry and aspect
Transport
Slope
Transport
Proximity to faults
Transport
Seismic facies
Trap
2 (inside)
1 (outside)
2 (favourable)
1 (unfavourable)
2 (b0.2°)
1 (N0.2°)
3 (400–1400 m)
2 (0–400 m)
1 (N1400 m)
6 (no data)
5 (Pleistocene cover)
4 (Miocene soft chalk)
3 (Oligocene chalk)
2 (Eocene marl)
1 (Basement)
2 (N20 kg/m2)
1 (b20 kg/m2)
0.7
0.3
0.7
0.3
0.8
0.1
0.7
0.5
0.5
0.5
0.1
0.6
0.8
0.5
0.01
0.8
0.5
Nodule grade
Deposit
of the nodular lag deposit in focus, which differs significantly from
the generally narrow, highly concentrated targets in metallogenic
prospectivity modelling. To achieve maximum target coverage, the
prior probability of 0.01407 was set as a lower cut-off, as the broadest
area the prospectivity model can validly target. Within the prospective
area the model has identified 67 target areas covering 891 km2, or
13.6% of the 6538 km2 study area (Fig. 5), with 93.5% of the training
points located within the prospective part of the study area. The prospective areas of the model occur on local topographic highs near the
central saddle, along the strike of major faults and where the terrain is
generally flat. 727 km2 (81.6%) of the target area is also covered by a
resource with N 20 kg/m2 nodule coverage, indicating that the
prospectivity model adds 164 km2 of new potentially prospective
ground (Fig. 5).
The model does have issues of conditional dependency, as several of
the predictive maps used have been derived from similar data sources
and have similar map patterns. Probability values may therefore be
overestimated and should not be used as actual statistical probability
of finding phosphate mineralisation. The conditional dependence issue
can be minimised by excluding or combining dependent predictive
9
maps. However, this could remove important information from a
dataset that is already sparse. The probability values can instead be
used to objectively rank the targets by ordering the posterior probability
values of the target areas.
Statistically testing the predictive capacity of the prospectivity
model required creating a set of measured data points with known
high nodular phosphate concentrations. From the phosphate coverage
dataset, all points with phosphate coverage value above 50 kg/m2 within the study area were chosen to test the predictive efficiency of the
model (424 points). The curve for the 46 points of the training data
gave a Success Rate value of 96.7% and the efficiency of prediction
based on the 424 high nodule concentration points is 93.5%. These measures confirm that the model has high predictive efficiency.
6.2. Fuzzy logic model results
The fuzzy logic prospective modelling was completed to produce
targets based largely on predictive map derivatives of bathymetry and
geology, using the fuzzy logic modelling tools of the Arc-SDM toolbox
for ArcMap 10. A cut-off of 0.51 was used for targeting, based on a comparison with mineralised target areas from the WofE model. The highest
prospectivity was indicated for targets that included the N 20 km2 resource estimate based on all available sample data as well as proximity
to faults, marked with red in Fig. 6.
In addition to the resource estimate data and fault map, the model
used low slope angles, entrainment in the saddle regions, and shallow
bathymetry with SE-facing slopes as important parameters for nodule
deposition on the Chatham Rise. The most prospective regions are likely
to be found in the vicinity of major faults and within the very prospective Early Oligocene chalk facies. Importantly, the model also maps
areas that are less likely to be mineralised: i.e., areas containing outcrops
of basement and areas with thick cover of Late Neogene, especially
Pleistocene, sandy sediments. Prospectivity is also likely to decrease as
one searches north or south away from the central crest where slope angles are steeper, young sediment drapes are thicker, and water depth
likely to have been too deep to favour extensive phosphatisation in
the past.
The targets cover more than 9600 km2 or 30% of the model area
(Fig. 6). 57 targets were defined with areas greater than 1 km2. The 10
Fig. 5. Weights of evidence prospectivity, ranked from above prior probability (blue, 0.01407) to maximum (red, 0.85), as well as targets outlined in grey. The black grid represents where
within the target areas a 1 km2 resolution resource model indicates N20 kg/m2 nodule coverage. (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: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
10
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
Fig. 6. Fuzzy logic modelling targeted prospective areas, compared with the areas of elevated nodule abundance discovered by the 1967–1968 Global Marine surveys (dotted outlines,
based on Cullen, 1987). The red area has the highest prospectivity, based mainly on the indication of nodule presence N20 kg/m2 and proximity to regional faults. (For interpretation
of the references to colour in this figure legend, the reader is referred to the web version of this article.)
targets with N100 km2 area make up 90% of the total target area. The
highest prospectivity targets (red in Fig. 6) are based on the resource
and fault predictive maps especially, and cover 2400 km2 or 13% of the
total model area.
7. Predictive mapping and resource modelling
Resource modelling requires evaluation based on statistics and
sample distribution. Because the Chatham Rise deposit is essentially a
2-dimensional resource resulting from multiple inputs, there is no simple way to constrain the resource distribution. The prospectivity model,
on the other hand, takes into account the varying mappable geological
parameters that control deposition and can be used to more accurately
constrain the resource estimation interpolations and prevent grade
smearing into less or more importantly non-prospective areas.
Using the weights of the different facies, a map can be produced that
compares grade estimated using resource modelling to the facies model
(Fig. 7A). This can then be used to exclude areas of phosphate grades
projected into unfavourable geology, as a hard boundary for resource
distribution in addition to other limiters such as an economic grade
cut-off (Fig. 7B). However, the most effective approach to resource
modelling may be to base it on the post-prospectivity map, thereby including all the variables that are favourable for deposit occurrence. This
would allow grade to occur over basement facies, for example, if other
variables were in favour and the presence of a nodular lag deposit on
a basement outcrop was due to the removal of younger, finer-grained
deposits by bottom currents.
8. Data confidence and exploration planning
As important as the assessment of prospectivity for phosphate may
be, the ability of the WofE technique to quantify data uncertainty is
also significant for this project. This enables the distribution of data involved in the estimates to be mapped and quantify where the data coverage is too sparse to ensure confidence in the result. This is not possible
for the expert opinion-driven fuzzy logic modelling.
The fuzzy logic model output can be used to aid future exploration
planning, by highlighting areas most likely to be prospective. For the
Chatham Rise model, the WofE targets generally fall within the area of
high confidence in the data. This is common for most prospectivity
models and may be an artefact of historical surveying generally focussing on areas of known mineralisation (Fig. 8). Targets that are outside
the area of statistically convincing results may be prioritised for exploration if they overlap fuzzy logic targets. Likewise, areas without WofEbased targets that are in regions with low confidence can be prioritised
for follow-up exploration especially if the fuzzy logic model shows prospective ground there. Likewise, areas shown with confidence to have
no targets can be avoided.
Combining the confidence map with targets and resource estimates,
which are generally calculated simply based on grades from sample
data, can likewise guide exploration towards adding data where the return will be the greatest: Areas with WofE targets and high data confidence can be assumed to be valid, lending validity to resource
estimates from these (compare Figs. 7b and 8). Areas with fuzzy logic
targets but not high WofE data confidence are prime subjects for further
exploration, including data acquisition, and resource estimates from
these will need to be treated with caution. Areas with high data confidence and no targets are effectively sterilised, barring changes to the
mineralisation model or geology.
9. Predictive maps and environmental modelling
The final prospectivity maps and associated predictive maps can be
combined with cadastral and environmental maps to define where
mining is possible. This way, the Chatham Rise predictive model results
can be applied to environmental modelling since part of the modelling
outlines the extent of hardground formation; hard substrates provide
living space for a specialised subset of benthic organisms, which will
be impacted when the hardground rubble is removed from the seabed
as part of the mining process. Model output can be used to map out
and evaluate the impact of the planned mining on hardground dwellers
(area mined compared to the total area of living space). Conversely,
mapping the distribution of hardground dwellers and animals that
feed on them may be added to the phosphate model as potential predictive layers.
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
11
Fig. 7. A. Facies prospectivity from Table 3 compared with a grade map. B. Grade map, cut by unprospective lithologies and with b20 kg/m2 cut out.
10. Conclusions
Spatial data modelling techniques were applied to the offshore
Chatham Rise phosphate nodule deposit. Challenges to the modelling include inconsistent data formats, scarcity of data and uncertainty of locations of especially older data. Two models were
produced, both combining individual predictor themes of geology,
geochemistry and geophysical data into singular predictive maps. A
weights of evidence (WofE) model was first produced that covers
the area of highest data density. The result from the WofE modelling
was then used to guide a fuzzy logic model that covers a wider if less
well explored region.
The models confirm that aspects of the topography and especially
change in relief (nearness to local highs, faults) are as important
as understanding the geology. For future exploration, collecting
good quality, high-resolution bathymetric data should be a primary
concern, as is mapping near-surface structures using sub-bottom
or shallow-seismic profiling. Having good coverage of samples
tested for the presence of phosphatised hardground nodules is also
important.
The benefits of this type of analysis include a need for qualitycontrolled input data, effective data compilation and through the production of unique conditions and confidence grids, a guide for future exploration programmes. Since the deposit is a lag deposit, the output of
the models is relevant for showing where the deposit can be found as
much as where it is absent. Combining the WofE and fuzzy logic models
post-probability maps with a map of statistical confidence in the results
can be used to limit exploration to areas where exploration will give the
most return, limiting expenditure as well as environmental impact. The
models can therefore be integral to exploration and mine planning as
well as to resource and environmental modelling over the central Chatham Rise.
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
12
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
Fig. 8. Prospective ground (post-probability N prior probability) versus data confidence.
Acknowledgements
This work is based on historic exploration data compiled from the
New Zealand Petroleum and Minerals Department of the Ministry of
Economic Development, as well as on exploration financed by Chatham
Rock Phosphate Ltd. Additional thanks to Hermann Kudrass for inspiration and geological discussions and to Feriel Falconer for further comments and suggestions to the manuscript. Also thanks to J.I. Escavy
and one anonymous reviewer for comments and suggestions to improve the manuscript.
References
Arning, E.T., et al., 2009. Genesis of phosphorite crusts off Peru. Mar. Geol. 262, 68–81.
Bonham-Carter, G.F., 1994. Geographic information systems for geoscientists: modelling
with GIS. Pergamon, Oxford (398 p.).
Bonham-Carter, G.F., Agterberg, F.P., Wright, D.F., 1989. Weights of evidence modelling: a
new approach to mapping mineral potential. In: Agterberg, F.P., Bonham-Carter, G.F.
(Eds.), Statistical Applications in the Earth Sciences 89-9. Geological Survey of
Canada, pp. 171–183.
Carranza, E.J.M., 2008. Geochemical anomaly and mineral prospectivity mapping in GIS.
Handbook of Exploration and Environmental Geochemistry 11. Elsevier, (351 p.).
Carranza, E.J.M., 2014. Evidential belief predictive modeling of mineral prospectivity using
few prospects and evidence with missing values. Nat. Resour. Res. http://dx.doi.org/
10.1007/s11953-014-9250-z.
Carter, R.M., McCave, I.N., Richter, C., Carter, L., et al., 1999. Proceeding of the Ocean Drilling Program, Initial Reports, 181. Ocean Drilling Program, College Station, TX http://
dx.doi.org/10.2973/odp.proc.ir.181.2000.
Cook, R.A., Wood, R.A., Campbell, H.J., 1989. The Chatham Rise: an exploration frontier.
1989 New Zealand Petroleum Conference Proceedings, pp. 35–41.
Cullen, D.J., 1979. Mining minerals from the sea floor: Chatham Rise phosphorite. New
Zealand Petroleum and Minerals, Mineral Report Series, MR2744, (9 pp.).
Cullen, D.J., 1987. The submarine phosphate resource on central Chatham Rise. Division of
Marine and Freshwater Science Reports 2, pp. 1–22.
D’Anglejan, B.F., 1967. Origin of marine phosphorites off Baja California, Mexico. Mar.
Geol. 5, 15–44.
Deng, M., 2009. A conditional dependence adjusted weights of evidence model. Nat.
Resour. Res. 18, 249–258.
Falconer, R.K.H., von Rad, U., Wood, R.A., 1984. Regional structure and high-resolution
seismic stratigraphy of the central Chatham Rise (New Zealand). In: Von Rad, U.,
Kudrass, H.-R. (Eds.), 1984: Geology of the Chatham Rise Phosphorite Deposits East
of New Zealand: Results of a Prospection Cruise with R/V Sonne (1981). Geologische
Jahrbuch, D 65, pp. 29–56.
Field, Browne, 1989. Cretaceous and Cenozoic sedimentary basins and geological evolution of the Canterbury region, South Island, New Zealand. New Zealand Geological
Survey Basin Studies 2, (94 pp.).
Harris, J.R., Wilkinson, L., Heather, K., Fumerton, S., Bernier, M.A., Ayer, J., Dahn, R., 2001.
Application of GIS processing techniques for producing mineral prospectivity maps
— a case study: mesothermal Au in the Swayze Greenstone Belt, Ontario, Canada.
Nat. Resour. Res. 10, 91–124.
Hayward, B.W., Grenfell, H.R., Carter, R., Hayward, J.J., 2004. Benthic foraminiferal proxy
evidence for the Neogene palaeoceanographic history of the Southwest Pacific, east
of New Zealand. Mar. Geol. 205, 147–184.
Heath, R.H., 1976. Oceanic circulation in the head of the Hikurangi Trench, East Coast,
New Zealand. N. Z. J. Mar. Freshw. Res. 10, 651–674.
Hronsky, J.M.A., Groves, D.I., 2008. Science of targeting: definition, strategies, targeting
and performance measurement. Aust. J. Earth Sci. 55, 3–12.
Kennett, J.P., von der Borch,, C.C., et al., 1983. Initial Reports of the Deep Sea Drilling
Program, 90. U.S. Govt. Printing Office, Washington, pp. 653–744.
Kreuzer, O.P., et al., 2014. Comparing prospectivity modelling results and past exploration
data: a case study of porphyry Cu–Au mineralisation systems in the Macquarie Arc,
Lachlan Fold Belt, New South Wales. Ore Geol. Rev. http://dx.doi.org/10.1016/j.
oregeorev.2014.09.001.
Kudrass, H.-R., 1984. The distribution and reserves of phosphorite on the central Chatham
Rise (SONNE-17 cruise 1981). In: Von Rad, U., Kudrass, H.-R. (Eds.), 1984: Geology of
the Chatham Rise Phosphorite Deposits East of New Zealand: Results of a Prospection
Cruise with R/V Sonne (1981). Geologische Jahrbuch, D 65, pp. 179–194.
Kudrass, H.-R., Cullen, D.J., 1982. Submarine phosphorite nodules from the Central Chatham Rise off New Zealand — composition, distribution and reserves (VALDIVIACruise 1978). Geol. Jahrb. D 51, 3–41.
Kudrass, H.-R., von Rad, U., 1984. Underwater television and photography observations,
side-scan sonar, and acoustic reflectivity measurements of phosphorite-rich areas
on the Chatham Rise (New Zealand). In: Von Rad, U., Kudrass, H.-R. (Eds.), 1984:
Geology of the Chatham Rise Phosphorite Deposits East of New Zealand: Results of
a Prospection Cruise with R/V Sonne (1981). Geologische Jahrbuch, D 65, pp. 69–90.
Lord, D., Etheridge, M.A., Willson, M., Hall, G., Uttley, P.J., 2001. Measuring exploration
success: an alternative to the discovery-cost-per-ounce method of quantifying exploration success. Soc. Econ. Geol. Newsl. 45 (1), 10–16.
Lusty, P.A.J., Scheib, C., Gunn, A.G., Walker, A.S.D., 2012. Reconnaissance prospectivity
analysis for gold mineralisation in the Southern Uplands-Down Longford terrane,
Northern Ireland. Nat. Resour. Res. 21, 359–382.
Mejía-Herrera, P., Royer, J.J., Caumon, G., Cheillietz, A., 2014. Curvature attribute from
surface-restoration as predictor variable in Kupferscheifer copper potentials. Nat.
Resour. Res. http://dx.doi.org/10.1007/s11053-014-9247-7.
Morris, J.C., 1987. The Stratigraphy of the Amuri Limestone Group, East Marlborough,
New ZealandPh.D. Thesis University of Canterbury, (439 pp.).
Nelson, C.S., Cooke, P.J., 2001. History of the oceanic front development in the New
Zealand sector of the Southern Ocean during the Cenozoic — a synthesis. N. Z. J.
Geol. Geophys. 44, 535–553.
Partington, G., 2010. Developing models using GIS to assess geological and economic risk:
an example from VMS copper gold mineral exploration in Oman. Ore Geol. Rev. 38,
197–207.
Partington, G.A., Sale, M.J., 2004. Prospectivity mapping using GIS with publicly available
earth science data — a new targeting tool being successfully used for exploration in
New Zealand. Australasian Institute of Mining and Metallurgy Conference Volume,
New Zealand Branch Annual Conference, Auckland, pp. 160–167.
Pasho, D.W., 1976. Distribution and morphology of Chatham Rise phosphorites. NZ
Oceanogr. Inst. Mem. 77, 29.
Pedley, H.M., Bennett, S.M., 1985. Phosphorites, hardgrounds and syndepositional solution subsidence: a palaeoenvironmental model from the Miocene of the Maltese
Islands. Sediment. Geol. 45, 1–34.
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013
S.H.H. Nielsen et al. / Ore Geology Reviews xxx (2014) xxx–xxx
Peters, K., Partington, G., 2008. Prospectivity modelling of mineralisation systems in
Papua New Guinea using weights of evidence analysis. PACRIM Congress Conference
Proceedings, pp. 211–216.
Reed, J.J., Hornibrook, H. de B., 1952. Sediments from the Chatham Rise. N. Z. J. Sci.
Technol. B34 (3), 173–189.
Sawatzky, D.L., Raines, G.L., Bonham-Carter, G.F., 2010. Spatial data modeller. http://www.
ige.unicamp.br/sdm/ArcSDM10/source/ReadMe.pdf.
Simpson, E.S., 1920. On Gerksuitite at Gingin, Western Australia. Mineral. Mag. J. Mineral.
Soc. 89, 23–39.
Spiegelhalter, D.J., Knill-Jones, R.P., 1984. Statistical and knowledge-based approaches to
clinical decision-support systems, with an application to gastroenterology. J. R. Stat.
Soc. (A) 147, 35–77.
Sterk, R., 2014. Chatham Rise project. NI43-101 Technical Report and Mineral Resource
Estimate on the Chatham Rise Project in New Zealand. RSC Mining & Mineral Exploration (158 pp.).
Von Rad, U., 1994. Outline of SONNE Cruise SO-17 on the Chatham Rise phosphorite deposits East of New Zealand. In: Von Rad, U., Kudrass, H.-R. (Eds.), 1984: Geology of
the Chatham Rise Phosphorite Deposits East of New Zealand: Results of a Prospection
Cruise with R/V Sonne (1981). Geologische Jahrbuch, D 65, pp. 5–23.
Von Rad, U., Kudrass, H.-R., 1984. Geology of the Chatham Rise phosphorite deposits east
of New Zealand: results of a prospection cruise with R/V Sonne (1981). Geol. Jahrb. D
65, 252.
13
Waples, D.W., 1982. Phosphate-rich sedimentary rocks: significance for organic facies and
petroleum exploration. J. Geochem. Explor. 16, 135–160.
Wood, R.A., Andres, P.B., Herzer, R.H., et al., 1989. Cretaceous and Cenozoic Geology of the
Chatham Rise Region, South Island. 3. New Zealand Geological Survey Basin Studies,
New Zealand (76 pp.).
Wood, R., et al., 2012. Environmental Survey of the Chatham Rock Phosphate's Permit
Area CSL 50270, March 2012. GNS Science Consultancy Report 2012/144, (98 pp.).
Wood, R., 2012. Survey of Chatham Rock Phosphate's Permit Area CSL 50270, December
2011. GNS Science Consultancy Report 2012/06, (116 pp.).
Wyborn, L.A.I., Heinrich, C.A., Jaques, A.L., 1994. Australian Proterozoic mineral systems:
essential ingredients and mappable criteria. Australas. Inst. Min. Metall. Publ. Ser. 5,
109–115.
Zachos, J., et al., 2001. Trends, rhythms, and aberrations in global climate 65 Ma to present. Science 292, 686–693.
Zobel, B., 1984. Foraminiferal age of phosphorite nodules from the Chatham Rise (SO-17
cruise). In: Von Rad, U., Kudrass, H.-R. (Eds.), 1984: geology of the Chatham Rise
phosphorite deposits east of New Zealand: results of a prospection cruise with R/V
Sonne (1981). Geologische Jahrbuch, D 65, pp. 99–106.
Please cite this article as: Nielsen, S.H.H., et al., Chatham Rise nodular phosphate — Modelling the prospectivity of a lag deposit (off-shore New
Zealand): A critical tool for use i..., Ore Geol. Rev. (2014), http://dx.doi.org/10.1016/j.oregeorev.2014.10.013