Alluvial Mining
Alluvial Mining
CONCLUSIONS
87
ACKNOWLEDGEMENTS
90
REFERENCES
91
Fig. No. Title
LIST OF TABLES
The deve lopment and di stri but i on of i::oncentrat ions of heavy mi nera 1sin
alluvial systems is the result of the interplay between a complex array of
processes. These range from the 1arge sca le contro 1s of plate tecton i cs,
down to grain-sca l e processes. An understanding of these pr ocesses is not
only fundamental to the exploration for, and evaluation of, alluvial placer
deposits, but a l soto the exploitat i on of heavy minerals in streams for
prospecting purposes.
Table I lists some of the more important detrital heavy minerals, and
shows properties, common detrital shapes, and sources.
MINERAL COMPOSITION
RELATIVE CRYSTAL
DENSITY HARDNESS SYSTEM
U"'""AL
SHAPE
NAIIVC CLH'CNTS
Gold Au 15.0-19.3 2.5-3 cubic round or flat graln', rods or flakes hypo.... meso- and epithermal veins
Platinum Pt 14.0-19,0 4-4.5 cubic angular grains and flakes mafites/ultramarrtes
Other P.G.M. Pd,Ir,Os 12.0-21.1 4.5-7 cubic angular grains mafiles/ultrarnafites
Diamond C 3.5 10 cubic mostly octahedra cratonic kimberllleJ\amproite
OXIDES AND HYDROXIDES
Magnetite Fe304 5.16 5.5-6.5 cubic octahedra · well-rounded equant grains mafic igneous rocks
Chromite FeCr204 4.3-4.6 5-6 cubic octahedra - well-rounded equant grains mafites/ultrarnafiles
Other Spinels XY204 3.6-4.6 7.5-4.6 cubic octahedra· well-rounded equant grains Ig. rocks and aluminous metasedts.
Cassiterite Sn02 6.8-7.1 6-7 Tetrag. prismatic xtals, angular - rounded grains S-type granites
Rutile 1102 4.18-4.25 S-S.7 Tetrag. prismatio xtals, angular - rounded grains plutonic Ig .• mel and sedl rocks
Ilmenite FeTi03 4.5-5 5-6 Hex. sub-rounded· rounded equant graln8 mafites/ultramafites
Leucoxene alteration after Urn. 3 .5-4.5 variable Amorphous rounded - angUlar Irregular coaled grains sedimentary rocks
Corundum AI203 3.95-4.15 9 Hex. ang.- rounded Irregular fragments syenite, leldsp.pegm81.melshale and 1st
Tanl- Columbite (Fe.Mn)(Nb,Ta)206 5.2-7.95 &6.5 ~rth. fracture fragments to sub-rounded grains granite pegmatite
Baddeleyite il02 5.5·6 6.5 Mono. rounded Mbeans" granitoids
Uraninite U02 8-10 5·6 cubic angular graIns granite pegmatite
Thorianite Th02 9.3 6.5 cubic angular grains granite pegmatite ,
Pyrochlore (NaCa)2(Nb.Ta)206(0.OH.F)
SILICATES, TUNGSTATESANO PHOSPHATES
4.2·6.4 5-5.5 cubic angular grains alkaline Deamatltes
,
W
Zircon ZrSi04 4.65-4.7 7.5 Tetrag. prisms acid - intermed . Igneous rocks
Garnet X3Y2{Si04) 3 3 .6·4.3 6.5-7.5 cubic rhombic dodecahedra or fractured to rounded malnly mel rocks. also some ultramafic rocks
Wolframite (Fe.Mn)W04 7-7.5 5-5.5 Mono. submetaJlic cleavage fragments S-type (and some I-type) granites
Kyanite AI2Si02 3.6-3.7 4·7 Tri prismatic to rounded stumpy grains reg. high-grade aluminous metasedts.
Tourmaline complex silicate 3-3 .2 7.5 Trig. rounded to angular fractures granites, grelsen and granite pegmatites
Monazite (Ce,la.Y,Th)P04 4.6·5.4 5-5.5 Mono. mainly ellipsoidal. rounded grnlns accessory of some granites and syenites
Xenolime YP04 4.59 4·5 Tetrag. equant to rectangular flakes accessory of some granites and syenites
Ap~ite Ca5-'f94}~{E, C1.0H) 3.17-3.23 5 Hax. mainlY' ellipsoidal. rounded grains accesso!y in most Igneous rocks
Boyle (1979) has estimated that gold produced from alluvial placers,
including palaeoplacers, accounts for two-thirds of the metal ever mined.
The most famous palaeoplacers (or "fossil" placers) are those of the
Witwatersrand in South Afri ca, but numerous other examp les exi st. The
Tarkwa goldfield in Ghana is considered by many workers to be of alluvial
origin, and is hosted in a clast i c sequence of Precambrian age ( Boyle,
1979). The auriferous-uraniferous conglomerates of Jacobina, (Brazil) are
also of Precambrian age, and share many similarities to the Witwatersrand
deposits (Andrade Romas and Fraenkel, 1974). Carboniferous conglomerates
of the Gays River area in Nova Scot i a conta i n go ld in the form of nuggets
and flakes. The source of this gold appears to be auriferous quartz veins
in rocks outcropping in the vicinity (Boyle, 1979).
Recent gold placers include those of the Pilgrim's Rest area in the
Transvaal, which are currently undergoing evaluation for potential dredge-
mining by Gold Fields of South Africa (Bentley, pers. camm.) .
In general, tin placer deposits are more productive and more economic to
mine than lode deposits. Cassiterite, from which most tin is smelted, is
conta ined in placer concentrates, wh i ch a 1so yi e 1d va 1uab le amounts of
ni ob i um, tantalum, and rare-earth minerals. The main source of tin is the
Southeast Asian tin belt, extending more than 2400 km from northern Burma,
through western Tha i 1and and Ma 1ay,s i a to Bill i ton Is land (I ndones i a) , The
high concentration of cassiterite-bearing granites coupled with deep and
rap i d chemi ca 1 weatheri ng has re leased large quant ites of primary
cassiterite from lode sources to produce eluvial, alluvial and marine
placers (Hails, 1976).
Heavy mineral stream surveys are also used in exploration for immobile
elements such as Au (e.g., Tooms, 1987), Sn (e.g., Sirinawan et a1., 1987),
W (e. g., Petersen and Stenda 1, 1987), U (e. g., Fri ck, 1987) and Ta-Nb
(e.g . , Watts et al., 1963). In arid environments mechanical dispersion of
elements dominates (section 2.1.3) and, heavy mineral concentrates of the
coarser fractions may produce anomalies. A classic example is the
dispersion of Pb and ln around the Gamsberg orebody in Bushmanland , South
Africa. Pb is dispersed in streams largely as grains of ang1esite (PbS04)
derived from · gossanous outcrop. ln occurs both as secondary minerals
derived by oxidation of the primary sphalerite, and as the spinel gahnite
(lnA 1204) , a product of amph i bo 1i te-fac ies metamorph i sm of the orebody
(McLaurin, 1978 ; J.M . Moore, pers. comm.).
In a regi ona 1 exp 1orat ion programme for ti n (cass iteri te) and tungsten
(wolframite, schee1ite) carried out in a Hercynian granite, migmatite and
schist terrain in northwestern Spain, Zantop and Nespereira (1978) reported
on the fa i1 ure of geochemi ca 1 stream sed iment samp 1i ng techn i ques. At a
density of 2 - 4 samples per square km, orientation samples in the vicinity
of known mineralizations failed to produce anomalies. However an efficient
method of detecting anomalies was found to be that of panning of stream
sediments, to obtain a heavy mineral concentrate . Visual determination of
the heavy mi nera 1s was performed by a techn i ci an at a rate of 10 to 15
minutes per sample. This method was found to have the following advantages
over geochemical methods:
1. 3 CLASSIFICATION OF PLACERS
In this review, the term placer refers to a surficial heavy mineral deposit
formed by mechan i ca 1 concentrat i on of mi nera 1 part i c les from weathered
debris.
Numerous class ifi cat ions of placer depos its have been proposed, usua lly
using such criteria as their major ore minerals qualified in some instances
by environmental descriptors. Examples might be as follows; alluvial gold
placers, heavy mi nera 1 beach placers etc. Such class ifi cat ions are of
little predictive use. Vlasov (1968) and Kartashov (1971) developed a
concept of classification in which the first order criterion is distance
from source. Hence, placers may be grouped as near source, all uv i a 1 or
marine. More simply they may be grouped as "autochthonous" (i.e. proximal
to source) or "allochthonous" (i .e. distal). This system is an
improvement, particularly in combination with the system of Emery and
Noakes (1968) who divided placer minerals into "heavy" heavy minerals
(gold, tin, platinum), "light" heavy minerals (ilmenite, rutile, zircon
and monazite) and diamonds (the extreme durability and value of which sets
them apart). The i r concept imp 1i ed that the heav iest mi nera 1s wou 1d
produce more proximal (or "autochthonous") placers. However, the fact that
the "light" heavy minerals are most commonly mined in beach environments is
more probably a function of the greater tonnages required for these
relatively low value commodities, than the fact that they are absent in
more proximal settings.
Continental Eluvial Au, Pt, Sn, WO, Ta, Percolating waters. Soil sampling, shal· Open cast, hydraulic
Nb, gem stones (all chemical and biologi- low pitting, churn sluicing, hand mining
varieties) cal reactions, heat, drilling
wind and rain
Colluvial Au, Pt, Sn, WO, Ta, Surface creep. wind, Stream and soil sam- Hydraulic sluicing
Nb, gem stones (all rainwash, elutriation. pIing. shallow pitting bulldozer and
varieties) Frost and trenching loader-hand mining
Fluvial Au, Pt, Sn, rarely Ta, Flowing streams of Stream sampling, Bucket dredging in
Nb. diamonds and water geophysics, pitting, active beds, bucket
corundums churn auger and pit- dredging, hydraulic
digging drills, sluicing and dozer-
Banka Drills loader operations in
old stream beds
Desert Au, Pt, Sn, WO, Ta, Wind with minor Shallow pitting, Various earth~moving
Nb, gem stones (all stream flow_ Heat and chum and pit digging combinations
varieties) frost drills, geophysics
Glacial Au (rarely) Moving streams of Stream sampling and Hydraulic sluicing
ice and melt waters pitting
Transitional Strandline Ti, Zr, Fe, ReO, Au, Waves, currents, Hand augering and Suction cutter dredg-
Pt, Sn wind, tides sludging, sample ing, bulldozer and
splitting allowable loaders, bucket~wheel
dredging
Coastal Ti, Zr, Fe, ReO Wind and rain splash Power augers Suction cutter dredg·
Aeolean (hollow), sample split. ing, buldozers and
ting allowable buried loaders
Deltaic Ti, Zr, Fe, ReO Waves , currents, Hand augering and Specially designed
wind, tides and chan· sludging, sample shallow depth dred·
nel flow . splitting allowable ges having great
mobility
Marine Drowned Au, Pt, Sn, diamonds, Eustatic, isostatic Geophysics (seismic Bucket line dredging,
placers minor Ti , Zr, Fe, ReO and tectonic refraction and reflec- jetting, clamshell,
industrial sand and movements-net rise tion) bottom sam- rarely suction-cutter
gravel in sea level pIing, remote sensing, dredging
hammer, jet, vibro
and bank. drills,
positioning
1acks important deta i 1s of env ironment and genet i c process requ ired by a
scheme which can truly be used as an exploration tool.
The concepts of where and how are centra 1 to the theme of th is rev i ew. The
next sect i on deals with the genesis of alluvial placers.
PART2 GENESIS OF ALLUVIAL PLACERS
Proximal Distal
l'j
/
Chemical Maturity
..
Textural Maturity
..
Grain Size
Sorting
..
FIG. 1. The evolution of placer deposits.
2. 1 . 1 Weathering Processes
The weathering processes responsible for the breakdown of primary rocks can
be subdivided into: i) chemical weathering; and ii) mechanical
-10-
(b) Gibbs free energy values for weathering reactions 1-10 in Table I.3a (after Curtis
19761.
Thus:
o
6 Gf = -1036.2 + 1012.3
Frost wedging involves the alternate freezing and thawing of water which
has penetrated into cracks in rocks. Water has the un i que property of
expanding by about 9 per cent as it freezes, thus exerting an outward
force, and increasing the dimensions of the cracks. This process is most
common in periglacial environments in sub-arctic and mountainous regions.
Un load ing produces exfo 1i at i on domes caused by the different i a 1 reduct ion
in pressure as overlying rock is stripped away. It is particularly well-
developed on granites.
2. 1 • 4 Examples
The stann iferous placers of the Tujuh arch i pe 1ago, Indones i a, have been
described by Aleva (1986). 95% of the mineable cassiterite in this
province lies on the source bedrock indicating that minimal transport has
occurred. The generati on of these placers has been the resu 1t of the
interaction of four factors:
The liberation of the brittle cassiterite from its encasing minerals in the
host rock, has been efficiently effected by deep tropical weathering, which
alters most sil icates to kao l inite . An example of this reaction is as
follows:
Alkalies and earth alkalies are removed by leaching, leaving quartz and
cassiterite as residual primary minerals. Granitic rocks are thus altered
to a sapro 1 ite of p 1ast i c to tough clayey cons i stency composed of quartz
(and cassiterite) grains in kaolin, with many relict igneous textures
ret a i ned. Th i s sapro 1ite may reach tens of metres in thi ckness and is
covered by a residual sandy layer. The sedimentary rocks develop a thinner
sapro 1ite of on ly a few metres because their const ituent mi nera 1s are
already largely in equilibrium with near-surface conditions.
During the Quaternary, ice sheets periodically encroached from the north
to produce a periglacial environment in the region . Wet phases dominated
the i ntergl aci a 1 stages, and the 1ast co 1d phase termi nated about 10 000
years BP with amelioration of the climate. The ground preparation for
placer development in a cryergic system depends largely upon frost wedging,
and on disintegration during rapid and marked variations in temperature of
material composed of several mineral species with different coefficients of
thermal expansion. This process builds up a layer of unconsolidated debris
known locally as "head", which moves downslope by processes of mass
wasting, most notably solifluction.
LEGEND
1,,"'1 PLACER DEPOSITS 135m CONTOUR .: ••••••••• 230m (ONTUJR
* c
FIG. 3. Map of the North St. Austell area showing disposition of stockwork,
sheeted vein and replacement deposits, and of placer deposits (after
Carnm and Hosking, 1985).
2.2 TECTONIC SETIINGS OF ALLUVIAL PLACERS
For example, the primary source rocks of diamonds are kimberlites and
certain lamproites, and diamondiferous examples of these igneous intrusions
have a strong geographical association with cratonic areas (Dawson, 1980).
In some regions such as Ghana and Brazi 1, economic concentrations of
alluvial diamonds have been derived from metamorphosed Precambrian
sediments, and, although the primary sources of these diamonds have never
been estab 1i shed, thei r occurrence adds to the associ at i on of diamond
mineralization with shield areas. The distribution of significant diamond
placers is shown in Figure 4. There is a clear correspondence with
cratonic areas.
Q.,
II
6~~~
'" "-
~" }
"·E " '20 'GO
• • <>.::
,
I
-I
U-II
I
~
,,·s o (}'i"J
DIU4GlfD -nA[£ l
IGO
'" " ~D · w 11 lO·[
" '"
FIG. 4. a) Primary source rocks of diamond. Diagonal lines represent
Precambrian shield areas and solid black shows areas of outcrop
of kimberlite and lamproite.
,~."Y:
., ~
{}(I~
CJ » o
-W. )
PUCQmbfian PalaeOloic
I
.....
N
I
'i-,Q'"
~
.,<, N Toromakau
.,0v'''
t Ross Flat
t~ w
I
Clutno -Kawe rou Ri ... er
- Fault
Otago
(USIOhl;
t
$otO~ ct"<r>qei
~-=""T
14 00Cl.., \&111ft
_A>_ ,..,__
z
g
Blue 5(lur
F~l'!Jr~
Renewed uplift to form the present day Alpine Mountains induced reworking
of the Blue Spur fanglomerates into the Otago basins which had formed on
the flanks of the rising mountain chain. Sedimentary reworking in these
basins occurred in response to base-level changes due primarily, to
episodic regional uplift, as well as eustatic sea-level fluctuations.
Adams et al. (1978 ; see also section 2.3.2) suggested that the complex
response to a single uplift would produce the multifold· reworking of
channel alluvium that is required for placer formation. Although such
rejuvenat ion may destroy ex i st i ng placers, it creates a su i tab 1e
environment for the formation of new ones. The broad distribution of gold
through the stratigraphic column (Figure 7), therefore records the
sequence of base- 1eve 1 changes and sed iment-supp ly changes that occurred
during the period of episodic uplift.
-25-
Hence, tectonic uplift was the primary controlling factor in the genesis of
the Otago alluvial gold deposits. This conclusion may be extended to other
major alluvial gold deposits, including those of the Witwatersrand, in
which the gold is associated with angular unconformities· (Els, 1991).
Although the Witwatersrand Basin is now part of the Kaapvaal Craton, it is
likely to have developed as a foreland basin in a collision setting
(Stanistreet and McCarthy, 1991).
A study of the late Quaternary alluvial diamond placer on the Bir im River
in Ghana by Hall et al. (1985) assessed the contributions of both
morphogenic and grain-scale concentrating processes in the deposit's
formation. Quaternary climatic changes are cons i dered to have been
significant in governing the development of many West African diamond
placers. The Birim deposit is the downstream continuation of more proximal
alluvial and colluvial placers of the Akwatia diamond field.
Geology
Geomorpho logy
The Birim River floodplain lies in a wide shallow valley, 30-40 m below
broad interfluves which rise to between 155 to 180 m above s'ea level. The
interfluves are capped by a thin laterite which is regarded as the remnant
of a Tertiary planation surface. The Birim floodplain displays alternating
broad and restricted reaches from 200 to 1500 min wi dth. The bedrock
-26-
• KUMASI
0;..,~_ _ _ _ _ _ ::.;3,O ml
o 5'0 km
UPPE
D.
, SEKOND I AN &
ACCRAIAN
o L OWER 8 1RRIM IA N
EZJ OAHOMEYAN
o VO LTA . AN
BA SIC
.~ TOGO SE AlES I NT RUSIV E S
FIG. 8. Geological map of southern Ghana, showing the location of the Birim
catchment (broken line). All rocks except intrusives and Tertiary
deposits are of Proterozoic age (after Hall et al., 1985).
The river f lows in a box-l ike channel on average 30 m wide and 7 m deep .
The ri ver bed compri ses coarse grave 1s, and it's banks, sandy and clayey
silts. The channe 1 pattern is i rregu 1ar ly meandering, but the abundant
bedload, and the lack of the typical characteristics of a meandering river
-27-
Br..... r"DUI.'l'
lo .. tow f",.c.
9=::...:.':~~_"_. _ ____ ~1o~ flooOr>l:::':'"=-__ T~"a c. •. _ __"c-=c._"_ _____"'=.=-:c.:H~....::..:'~M~'~"~.~.___
'"
......\ Brown
...... r. llIY soil __ _ _ _ _~--;-,
: · -~--.....,aIRI IoI
n,
A
A
'"
o m
o
Sod
one1
,
Gtey bto.t molll.d
!lancIY clayey s~t OIly-brown lighlly
moW,d day .00's
•
StronOly moWed
ell.,.e.,. sand
Gre., blue mollled
cll yey .-0
Dark ore.,. sandy
claye.,. Silt wtlh
traIn
~f~~;E~~r~g'''h
••nd
Coane s..-cl and line sancl ""'th line
to mediuM ..,....., WIth Grey QUartz Sind II
..... 10 woooy trasn atw:I !tash
occasional pel)bles.
Abu>danl rlrMl tr.sn eo...,. Sind, line
,.n d PI<tC" ot wood ,,"Ivlll and limOef.
"
CII.t -sUOOQrtfld coa rs. Coar•• IVI"'e' w it h
grave' w"h Il"ly elay pebbles and cobbles
m.,,; • . PilN;. . 01 woo d ana Si nd m.trlx
o
" A 122 183m. 0 81
8
122 183ml
o
" '"
C
FlG. 10. Vertical sections in the Birim floodplain deposits. A, B and C
refer to bedrock levels shown in Fig. 9 (after Hall et aI., 1985).
-28-
(meander loop cut-offs, point bar deposits etc.) indicates that the general
model of channel migration and floodplain development by lateral accretion
is not applicable .
Climatically, the region falls into the humid tropical morphogenic region
of Suther 1and (1985 ; sect ion 2.3.1), but radiocarbon dat ing of organi c
material, and stratigraphic corre lation, has produced the following
inferred sequence of environmenta 1 and f 1uv i a 1 cond iti ons in the Birim
Valley for the late Quaternary.
Pre· 21000 years BP.. Formation of the lower terraces (Figure 9) during
several humid to sub-humid climatic oscillations.
21000 - 13500 years BP.. A contraction of the dry climatic belts and a
return to humid pluvial conditions. Extensive erosion of pre-existing
sed iments and scouri ng of bedrock. Depos i t ion of bas a 1 grave 1 with high
proportions of cobbles and pebbles (Unit I in Figure 10) .
7000 - 45000 years BP..Reduced peak discharges of the river, and limited
reworking of uppermost gravels.
Diamond Distribution
Within the gravels, grades increase downwards and, as with other alluvial
diamond placers, relatively high grades and larger stones are often found
close to bedrock. Higher grades may also occur well above bedrock and this
is ascribed to continued aggradation and multiple periods of reworking.
There is a strong correlation of higher grades with coarser gravels.
However, where the gravels are less than 1.Sm thick they are usually
barren.
·Spat i a lly, the grades decrease downstream from over 1. Scar. 1m 3 to 1ess
than O.S car. 1m 3 over the 32 km from locality 1 on Figure 8. Superimposed
upon this trend are major grade deviations reflecting local variations in
grave 1 1itho logy . Hi gher diamond grades are a 1so associ ated with points
where the floodplain widens to form a braided fan as described by Collinson
(1986) .
The disparity in hydrau 1i c equ i va lence between the diamonds and grave 1s
suggests that the primary process of diamond concentration is interstice
entrapment (section 2.4.2). Penetration of diamonds into the gravels may
have been aided by shear-sorting (section 2.4.1.1) induced by high bed-
shear stresses.
The richest placers were probably formed between 10S00 and 7000 years BP.
as a result of increased precipitation following the prolonged "Ogolian"
dry phase and the deposition of the coarse basal gravels . This combination
of circumstances had the following effects:
Hence the domi nant factor contro 11 i ng the geomorph i c evo 1ut i on of th i s
cratonic tectonic region was climatic.
Placer mi nera 1 concentrat ions are the product of part icu lar interact ions
between three distinct sets of variab les: the nature of the bedrock ;
basin dynamics ; and external processes acting at surface which are mainly
controlled by climate. Since geomorphology is the product of the same set
of variables, it follows that a geomorphological approach is essential to
an understanding of placer deposits at all scales . The global distr i bution
of placer deposits is largely a product of variation, both at present and
in the recent geological past, in geomorphological processes, acting at the
Earth's surface, given that suitable primary minera l sources exist.
2 •3 •1 Influence Of Climate
2) cold non-glacial;
-31-
3) humid temperate ;
5) humid tropical.
!;Sl J.
~ J'
b:)l
[]Jl
bd'
!Ell
flG. 11. Morphogenetic regions of the world: la, glacial regions effected
by ice-sheets' 1b, glacial regions in mountainous areas; 2, cold
non-glacial regions; 3, humid temperate regious ; 4, semi-arid and
arid regions; 5, humid tropical regions (after Sutherland, 1985).
1) Glacial Regions
These are the areas presently glacierized as well as those areas that were
extensively glaciated during the Quaternary . A subdivision is indicated
between areas of ice-sheet glaciation and mountain glaciation, the latter
being restricted by topographic control.
Thirdly, glaciers can disperse heavy minerals with little size sorting over
considerable distances to give rise to large volumes of very low grade
sediment. This protore may be reworked by post-glacial rivers into
economically valuable placers. The extensive gold placers along the North
Saskatchewan River in Canada are an example. Other possible examples are
some of the alluvia l diamond deposits of South Africa. Harger (1909) was
the first to suggest that at least some of the Vaal Rivet diamonds are
derived from pre-Karoo Kimberlites via Permo-Carboniferous Dwyka
"conglomerates". This hypothesis is based on the fact the diamond-bearing
gravels generally occur in areas where Ventersdorp andesites are associated
with the overlying Dwyka diamictite, and the premise that the diamonds have
been reworked from low-grade concentrations in the diamictite, into higher
grade gravels, by processes of heavy mineral concentration. This
suggestion is supported by the fact that the Dwyka diamictite is locally
diamondiferous (e.g., at Bosluispan in Bushmansland) . Although the Vaal
-33-
These areas lie beyond the Quaternary ice sheets, but are characterised by
the presence of permafrost. The greatest extent of these regions lies in
Siberia, Alaska and Canada.
A thin active soil layer with limited vegetation cover and seasonal thawing
combine to produce a short period of intense fluvial activity in late
spring and early summer. This temporal concentration together with the
contrast in state between the soil in the thawed valley bottoms and frozen
interfluves, results in fluvial incision.
Examples of such placers include gold in the Yukon River area of Alaska and
NW Canada (e. g., Cobb, 1973 ; Boyle, 1979). Sutherland (1985) also
recorded examples from the Siberian shield.
This region includes much of mainland Europe and the U.S.A, with a more
restricted distribution in the southern hemisphere . River flow is
-34-
Qnaternary climatic change has meant that even the most arid modern
environments have probably experienced some fluv i al activity in the last
three million years, and no distinction is therefore made between semi-arid
and arid regions.
The wide dispersion of diamonds in central and southern Africa may be due
to transport during arid phases of the Caenozoic, but concentration has
been during alluvial or "wet" phases. It is suggested that in arid
phases , weathering and fluvial processes are more effective in l i berat i ng
and transport ing heavy minera 1s, wh i 1e during "wet" or all uv i a 1 phases, the
heavy minerals are more effectively concentrated.
Sutherland (1985) stated that much of the humid tropical zone has
period i ca lly experi enced semi -ari d cond it ions throughout the Quaternary.
During such periods, fluvial activity may have been particularly efficient
in transport i ng 1iberated heavy minera 1s. The return of humi d cond it ions
and consequent increase in vegetat i on cover wou 1d reduce sed iment supp ly
and encourage rework i ng of materi ali ntroduced into the fl uv i a 1 sys tem
during the preceding arid phase. Such a combination of circumstances might
produce placer deposits of considerable extent.
-36-
During the latter part of the Cretaceous and the early Tertiary, humid
tropi ca 1 processes appear to have operated in mi d- to - high 1at itudes,
produc i ng deep 1y weathered clayey rego 1i ths (Dury, 1971 ; Ki ng , 1951 ;
Partridge and Maud , 1987). Significant changes occurred in the mid-
Tertiary as the climate shifted towards a regime characterized by a
sequence of glacial and interglacial periods exhibiting a periodicity of
around 0.1 Ma (Shackelton et al., 1984). These glacial cycles involve
major climatic changes at all latitudes (Figure 12).
The role of basin dynamics and tectonics in the formation of pl acers has
been studied by Adams et a1. (1978 ) i n an experimental model of an alluvia l
drainage basin. The concept of geomorph i c thresho 1ds is ment i oned in
section 3.2.1 where it is noted that channel pattern (the basis on wh i ch
rivers are classified) is controlled by slope thresholds, for a given
discharge, within a fluvial system. This concept has great significance to
the morphology of a drainage basin at times of tectonic uplift and drainage
rejuvenation. Figure 13 shows how a meandering river first becomes
incised, then braided, and finally stabilizes in response to a single
rejuvenation event (in this case a lowering of base level) .
::: , ....
,,: -.
Braided
Channel
-- ,
iii) as the tributaries stabilize to the new base - level, sediment loads
decrease, and a new phase of channel erosion occurs forming a low
terrace (Figure 13D).
Adams et al. (op. cit.) derived an empirical model in which the processes
involved in crossing geomorphic thresholds are superimposed on those
involved in complex response as follows:
i) when the lower main channel is degrading (Figure 13B and D) heavy
minerals are in transport. When aggrading (Figure 13C) the heavy
minerals are stored in the alluvial fill;
ii) superimposed upon this sequence of heavy mineral transport and storage
are shorter per i ods of transport and storage controlled locally by
internal thresholds of gradient and discharge. During periods of
aggradation, the channel is braided and heavy minerals are stored in
the drainage basin. During periods of incision, the channel is
meandering or straight, and heavy minerals are reworked and
concentrated.
of mild aggradation and degradation that result from the exceeding of the
internal geomorphic thresholds which may follow rejuvenation.
The Miocene age of the Auchas gravels corresponds to the uplift event ci ted
by Partridge and Maud (1978) as being responsible for their Post-African I
erosion cycle. Note that rejuvenation need not be in response to tectonic
uplift, but may be due to a drop in base-level (a eustatic lowering of sea-
level for examp l e) as is suggested by Dingle and Hendey (1984) for the late
Oligocene - early Miocene rejuvenation event in the western part of
southern Africa.
The nature of the bedrock influences the nature of sediment liberated into
a drainage system, and hence the nature of the materi a 1 with wh i ch heavy
minerals must "compete " for transport and deposition. Many fluvial placers
are associated with gravels for reasons outlined in section 2.4.2, and it
follows that a bedrock which 1iberates coarse clasts is required by many
fluvial placer-forming processes.
For example, consider the alluvial diamond deposits of the northern Cape
and western Transvaal in South Africa (Figure 14) . The alluvial diamond
depos i ts are vi rtua lly confi ned to a bedrock of Ventersdorp Supergroup
lithologies, comprising mainly andesitic lavas. These rocks are resistant
-41-
v V
V
SRSWEIZER RENEKE V V
B BLOEMHOF V
CHRISTIANA V V
C V
WaWARRENTON V
WiWINDSORTON
BWBARKLY WEST
DeDELPORTSHOOP
D DOU GLAS
S SCHUTSEKAMA
R RITCHIE
K KIMBERLEY
•
0
ALLtNIAL
KAROO ROCKS
DIGGINGS
~ CAMPBELL GROUP
K
•
}fODDER
FIG. 14. Geological map showing the alluvial diamond deposits of the northern
Cape and western Transvaal, South Africa. Note the association of
deposits with Ventersdorp lava bedrock.
ii) where Ventersdorp bedrock forms a topographic high (an exhumed pre-
Karoo high) the drainage is always highly incised thus forming
gorges. On the downstream side of such gorges, large alluvial fans
have formed, which are the sites of some of the richest
deposits.
Examples of the first case are found at Delportshoop, and of the second at
Schutsekama (Figure 14).
2.4 GRAIN-SCALE CONCENTRATING PROCESSES
2 . 4. 1 Hydraulic Sorting
The settling velocities of grains immersed in water will depend upon their
density if shape and surface texture are similar. Consequently, the size
of gra i ns that sett 1e together on the bed wi 11 be an inverse funct ion of
their density. Figure 15 is a schematic illustration of the quartz
settling equivalents of four heavy minerals. It can be seen that a
spherical cassiterite grain has a diameter which is only 0.52 of its quartz
equivalent (see Callaghan, 1979 for formulae). Gold, with a greater
density contrast, has a settling equivalent diameter only 0.32 times that
of quartz.
-44-
Unit
: sen ling
, veJoclly
FIG. 15. Spheres of selected minerals that have the hydraulic settling
equivalence of quartz. Values of density are given for each mineral
(after Reid and Frostick, 1985).
QU,Htz diameter, mm
0.1 0.5 1.0 2.0
Shorn.lov & /
Shumovskiy 's Data /
30 ... gold. Au o·
/
g cassiterite, 5n
/ ....'0<:;1 \:) ,,-
II magnetite. Fe / • 1· ./
• ~menite . Ti / .,)q Q"'./ -""
't': c.,(' \"~ -"' ...
./ /""" ""..
\.\eY /"
/'
/ "-""
/ , (\\1
/
,/ /,/
/
~)/,: p ·5.2
A /,/.
/" T
II Empirical
\I .;; Stokes' & Rubey's
formulae for spheres
,Ti p . 4.7
o~~--__----~----~~
a 5 10 15
Quartz (rounded) settling velocity. cnu "
W: ( /? /?) D~
~
-45-
W= 2/3g . /? - /? Os
~
It has long been recognised that the individual laminae of river deposits
often consist of like-size and like-shape particles, regardless of density.
Th i sind i cates alack of sett 1 i ng equ i va lence for the 1i ght and heavy
minera 1sin such a 1ayer. In fact, the heavy mi nera 1s present are much
larger than predicted by cons ideration of their quartz sett 1ing
equivalence.
100
-="_2'~""~""'---
:,r --
I,( c Settling
"",,
equivalence
.," "~I
~ . 50
.." ," '
/'/ heavy minerals
P" 5.2
o
I( ---
FIG. 17. Typical particle size curves for: a) light and heavy minerals on
beach bars of Lake Turkana that stand proud of the still-water level
and on which the swash rakes the surface as a shallow flow to
produce entrainment equivalence; and b) for those bars whose
crest is just below the still-water level and where breaking waves
throw bottom sediment into suspension to produce settling
equivalence. c) Schematic relationship between the average median
- sized quartz and magnetite - ilmenite particles on the two types
of swash bar showing the difference in relative size of each mineral
generated by the processes involving entrainment and settling
equivalence (after Reid and Frostick, 1985).
Komar and Wang (1984) have demonstrated that like-size particles congregate
because larger particles are removed. They stand proud on the bed and by
do i ng so, they are subject to greater lift and drag.
/
/
/
(f)
(f) 4 .-
/
/ ,
w ... ",'"
Z.,- ~_--'
~ ot~;/~/~/~~~~~~~~§~~i
a::
2
o 10 20 30 40 50 60 70
U',,(cm-sec-1l
Since experience shows that heavy minerals tend to dominate the finer
fractions of a sediment (Blatt et al . , 1972), the operation of dispersive
stress may lead to a concentration of heavy minerals several grain
diameters below the bed surface.
-50-
Whilst not dismissing the role of shear sorting in the genesis of placers,
Komar and Wang (1984) found a wide range in dispersive stress, mineral by
mineral, in each of their samples from a beach in Oregon. This leads to
the conclusion that, if dispersive stress is of consequence, it is an
ancillary process. Its role is to feed larger light-mineral grains to the
top of the mobile layer where protrusion tends to promote entrainment as
outlined previously .
As noted in section 2.3.3, alluvial gold and diamonds are often associated
with coarse-grained deposits (e . g. Witwatersrand gold and Vaal River
diamonds). The size contrast between heavy minerals and coarse framework
is usua lly too great to invoke sett 1i ng or entra inment equ i va lence to
explain the genesis of these deposits. This suggests that such deposits
were laid down during two distinct sedimentation events; the larger clasts
during extreme flood events; the heavy mineral-bearing matrix, later
(possibly during the waning stage of the flood).
Minter and Toens (1970) demonstrated experimentally that heavy minerals can
move into the interstices of a simulated river gravel comprising an open
framework of densely packed beds of pebbles. Furthermore, they found that
the concentrations for heavy minerals in transported bed-loads is increased
during passage across a layer of gravel. Smith and Minter (1980) provided
convincing evidence of this process having occurred in the coarse-grained
Witwatersrand "bankets" (pebb le cong lomerates) . They showed that the
conglomerates have concentrations of heavy minerals ten times greater than
that of neighbouring sandstones, thus indicating their trap-site qualities.
Figure 19shows the concentrations of gold and uranium in sand and gravel
facies of a Witwatersrand palaeoplacer. The sandstones have, in general,
much lower concentrations of the heavy grains.
A major control on the size of the material which passes into a stream-bed
framework and forms a matrix, is the size distribution of the surface
pores, i.e . the interstices of the surface, or armour layer. The largest
pore dictates the maximum grain size of the matrix, and in the case of a
close ly packed framework th i s wi 11 be 0 ..4 times the med i an framework
particle size . (Fraser, 1935) .
-51-
100
10 ~
,....
I
...
0\• J'
0
01)
, 1 f- ~
::J 0 •
-< 0
•
• •• •
0
0.1 ~ • 0 • 0
o conglomerate
• sandstone
0.01 • I I
1 10 100 1000
-1
U, g.t
Frostick et al., (1984) showed that the sub-armour framework also effects
the quantity and size distribution of the matrix. In a fining-upward
gravel bed in which the ratio of median particle diameter (D50) of the
armour layer to that of the sub-armour layer is about 0.5, the amount of
matri x (and heavy minera 1s) wi 11 be over three times greater than that
occurring ina coarsen i ng upwards bed wi th an armour 1ayer : sub armour
layer size ratio of about 2 (Figure 20). This is explained by the fact
that where the sub-surface pores are smaller than those of the armour layer
(i.e. a coarsening-upward profile), there is a tendency for coarser matrix
particles to lodge not far into the framework and thus clog the pores.
Consequent ly, there may be a cons i derab 1e amount of unfi 11 ed pore space
below such a matrix plug. The heavy mineral concentrating process is then
short-lived and such gravels are unlikely to be sites of preferential heavy
mi nera 1 accumu 1at i on. Converse ly, where grave 1s fine upwards, the pores
-52-
E 100
o
50
100
fining downwiud fining upward
O+-~==~I~==~~
o 10 20 30 40
(lJ
N
....
VI
(lJ
""0
50 channel-bed
c: armour-Ia yer
::::l pores
~
FIG.20. Size distributions of both the armour layer pores, and of the
ingressed matrices where the sub-armour layer framework particles
are differentiated according to size, of a gravel-bed stream. The
inset shows the total quantity of matrix that accumulates where the
the bed sediment either fines or coarsens. D 50 refers to median
clast size (after Frostick et al., 1984).
tend to be packed uniformly and over a longer period of time, and the
opportun ity for heavy mineral accumulation is therefore increased.
This process may explain the distribution of diamonds in the Vaal and Riet
River gravels in the northern Cape Province of South Africa. Gravel
profiles from the Vaal (Matheys , 1990) and Riet (Dawson, 1990) Rivers
closely correspond to the proximal-to-distal facies sequences of alluvial
fans (Collinson, 1986 ). Both sequences occur where the rivers emerge from
the confines of a gorge. Both comprise a proximal facies of massive
-53-
bou 1der-grave 1s, a mi d-f an fac ies of interbedded grave 1s and sands, and a
distal facies of coarse, trough-cross bedded pebbly sands. These alluvial
fans represent the large-scale morphological features built up by bedload
streams and they are characterised by a braided channel pattern. Matheys
(op.cit.) noted a decrease in per carat value of diamonds from the Vaal
River in a downstream direction, from the proximal facies at Gong Gong to
the mid-fan facies at Delpoortshoop (Figure 14). This decrease is ascribed
to a decrease in diamond size . A positive correlation exists between clast
and diamond sizes suggesting that during the waning water stage of flood
events, the largest diamonds were deposited with the coarsest and heaviest
matrix in the proximal areas, and progressively finer stones were deposited
further and further downstream. A fining-upwards profile has been noted in
some of the basal gravel beds which would enhance accumulations as
described above.
This model is not in conflict with that proposed by McGowan and Groat
(1973) for the Precambrian Van Horn alluvial fan in west Texas. These
authors found the greatest concentration of heavy minerals in the trough-
cross bedded distal facies of the fan. However, these were relatively fine-
grained ilmenite and magnetite grains which would have been carried in
suspension by flood events. Matheys (op.cit.) also found abundant fine-
grained heavy minerals (chiefly ilmenite) in the more distal facies of the
Vaal River gravels.
Geomorphological Example
Scale
(after Smith & Minter 1980)
& bars
ml:ander bends
DIRECTION OF FLOW •
? ~~~~~
Centimetres
%MAGNEnTE I
<5 5-15 16-30 31-45 46-60 61-75
Rolling-down of
boundary layer
Wake vortex
The areas of higher and lower pressures and velocities generated around the
obstac 1es contro 1 the behaviour of higher and lower dens ity gra i ns such
that sorting can take place. The areas of high concentration which bound
the wake region are clearly related to the large fluid forces recorded
immediately around the obstacle's flanks. The extremes of both velocity
and turbulence will ensure supply of both heavy and light sediment grains
to this zone . Downstream, rapid deceleration of fluid sorts the heavy and
light minerals. Smaller, denser grains will be more resistant to tractive
forces than 1arger, lower dens ity gra i ns. They wi 11 thus be concentrated
by the hydraulic processes of settling and selective entrainment of lights.
The area of turbulence and variable flow velocity associated with the point
of free-stream reattachment (i.e. the point at which bed pressure rises to
free-stream levels, immediatetly downstream of the wake of the obstacle)
also concentrates heavy minerals (Figure 22) .
Zones where heavy minerals are deficient reflect areas of low fluid
velocity. For instance, the magnitude of lift and drag may be insufficient
to draw grains into the low pressure wake zone. Only lighter grains, which
are characteri sed by lower entra i nment thresho lds, are carri ed into such
areas, so that their concentration increases above background levels.
Flow separation may also occur at larger scales (Figure 21), for example
at a channe 1 confl uence . Fi gure 24 shows the re lat i ve concentrat i on of
magnetite at an experimental channel confluence with a ratio of tributary
channel discharge: main channel discharge of 1.05. The bed morpho logy
-57-
a, = 1.05
BED ELEVATION
(mm)
avalanche face edge
o~
•
1.0
MAGNETITE
CONCENTRATION
FIG. 24. Bed elevation and magnetite isopleth maps at a channel confluence.
Discharge ratio (Or = OtlOm, where Ot = tributary channel
discharge and Om = main channel discharge) is 1.05. Magnetite
concentration is expressed as a ratio with the background
concentration of the free stream (after Best and Brayshaw, 1985).
compri ses three domi nant elements : a centra 1 scour ; ava lanche faces at
the mouth of each channel ; and a bar below the downstream junction corner .
It is around the flanks of the bar that significant quantities of magnetite
accumulate (Best and Brayshaw, 19B5). This pattern of accumulation may
again be explained by the distortions of the local flow field caused by
flow separation, and the associated processes .
At the downstream corner of the channe 1 junct i on, a large zone of slowly
recirculating separated flow occurs (Best and Reid, 19B4). This zone is
associated with fluid pressures and velocit i es that are much lower than the
surrounding flow (Figure 25). Additionally, a strong shear layer is formed
that bounds the separation zone, and al ong which powerful vortices are
generated. The vort i ces are respons i ble for the turbulent character of the
flow, engendering close similarities to the character istics of flow around
an isolated obstacle , described above .
-58-
Q, = 1.05
I I I
Flow separation produces two loci for the concentration of heavy minerals.
Firstly, the shear layer that bounds the separation zone (Figure 25)
induces the necessary shear stresses to provide a lag of heavy minerals by
selective entrainment of coarse, light mineral grains. Secondly, the
separation zone provides an area of low pressure and shear stress which
determines paths of transported grains. It also provides an area of
accumulation.
Heavy mi nera 1 gra ins enteri ng the ma in channe 1 from the tributary, are
deflected by the main stream (see vectors in Figure 25). The majority of
sediment grains are supplied to a narrow transport zone at the downstream
junction corner. Higher density grains are entrained along the shear
layer, but heavy minerals are deposited in an area of rapidly decreasing
shear stress within the separation zone. Best and Brayshaw (op.cit) found
that concentrations of heavy minerals increases with increasing tributary
discharge - main channel discharge ratio.
ii) the path of the fluid towards the separation zone carries entrained
heavy minerals into an area of lower shear stresses; and
on the downstream side of ripples, dunes and bars (e.g . Smith and
Minter, 1980)
CONTROLS
BASIN GRAIN-SCALE
SCALE TECTONIC SEITING CLIMATE DYNAMICS PROCESSES
CONTINENTAL
LARGE (SYSTEM)
INTERMEDIATE (BAR)
SMALL (BED)
TABLE IV. Some observed sites of alluvial heavy mineral concentration drawn
from the literature, and the key influences which control this
distribution at the different scales.
-62-
Q..
.~~>~1.'
Q "
o· wr ,.. 110· 160·
Not all significant placer deposits are included within the regions
identified in Figure 26, largely because of climatic change since the time
of their formation. In the Vaal River basin of South Africa, over 10 Mct.
of alluvial diamonds have been recovered. It is now known that the major
period of diamond transport from the Kaapvaal craton to the west coast of
South Africa was during the Miocene, as evidenced by the vertebrate
assemblage collected at Arrisdrift on the lower Orange River (van Wyk and
Piennar, 1986 ; Corbett, 1989). At present, the kimberlite diamond sources
in the Vaal River basin are typically sealed by calcrete duricrusts and
there is little release of diamonds into the present drainage. It is
therefore clear that target areas may be selected on the basis of studies
of palaeogeomorphology, which might indicate periods of heavy mineral
liberation, transport, reworking and concentration (e.g., Partridge and
Maud, 1987 ; Corbett, 1989 ; Thomas et al., 1985 ; de Wit, in prep.).
Local target selection will depend largely upon the type of fluvial system
under consideration. A number of prospecting models are presented based
upon the criteria discussed in previous sections. The models are specific
to classes of fluvial systems, since each class is chara:cterized by a
particular assortment of morphological and process parameters. It is
therefore pertinent to briefly outline various types of rivers prior to
presenting the models. Note that rivers evolve in response to tectonic and
climatic changes and therefore may change their character with time.
3 •2 • 1 Classification Of Rivers
~ .,
;:..r' __..,---~
./~
---
~-~-------....,...
- ~,--_
-"'-"'>~-
-
straight
meondering
anastomosing
The scheme is nat entirely satisfactory since the classes are nat mutually
exclusive, being based on slightly differing criteria. For example
braiding refers to multiple channels separated by alluvial islands or bars,
whereas meanderi ng refers to sinuous channels (Fi gure 27). It is thus
possible to have a braided channel bed within a meandering channel system.
Such comp 1ex i ties are, however, rare. Bra i ded streams are .genera 11y wi de
and sha 11 ow with near ly stra i ght channe 1 systems, and meanderi ng streams
tend to be relatively deep and narrow and are thus seldom braided. Cross-
sections of braided, meandering and straight channels are shown in Figure
28.
Trul y straight channels are uncommon in aggrading rivers. They occur most
often in degradational (downcutting) settings and therefore seldom provide
opportunity for the preservation of their deposits . Anastomosing rivers
have rece i ved 1itt le attent ion in the 1i terature a lthough Smith and Smith
(1980) recognised recent examples from Alberta, and Smith and Eriksson
(1979) described probable examples from a Permo-Carboniferous proglacial
Typo Morphology Sinuosity Load type Bedload per- Width/ Erosive behaviour Deposilional
cent (oC total depth behaviour
load) ratio
M{'f\n(lerinli{ !;ingle channels > 1.3 suspension <11 <40 channel incision, point-bar
or meander formation
mixed load widening
nrnide(l two or more < 1.3 bedload >11 >40 channel wiul:'ning channf'l
ch:mncis with aggril<!alion,
bars and small mid -channel
island. bar formation
Slrni~ht sin~lcchnnnel with < 1.5 suspension, <11 <40 minor channf'1 sirlc ·chilOnd
I
pools nml rirnes, mixed or widening <Inc! bar formation
mennrlering thalweg uedload incision '"
U1
I
Anastomosing two or more chnnllrls >2.0 suspension < 3 <10 slow meander slow hnnk
with Illq.~e, slnhle islilnds load widening accr(>tion
- - -- - - --- --- _ .-
0"'
!
"*,,,,",t-:;C .No d.
A B c
o 0' b b'
O~:;g;
,. ":':.
2.5
r .
' . ~ ..
. •
...)-\j-f'. ,.."'""'7':• . ~
' ''''''0,'<:::1'
. ':rf't;7-0-;
.. .. =
..·.•..•"r\)-r<i=:)
:~.; ..:
-.,
.50 : " -;
v '::.:~ .~•• "",lo.._~~
1.5 . - ... . . .
, ,,' •
....... ...... ., ... / .,
"
FIG, 28, Typical channel patterns and cross sections in braided (A),
straight (B) and meandering (C) reaches of the Brahmaputra River
(after Coleman, 1969),
Leopold and Wolman (1957) have shown that an empirical relationship exists
between channel slope, bankfull discharge and channel pattern (Figure 29).
For a given bankfull discharge, braiding is favoured by high slopes, and
meandering by low slopes. For a given channel slope, braid ing is favoured
by larger bankfull discharges and meandering by lesser bankfull discharges.
-67-
, I
.
~
~
0;
<
<
,g 0 .001
u
0 .0005
0 .0001
0,00005 Ll_----'--L_-L...L_---'--L_-L-L"-...J
10' 10' 10' 10'
Banll. full <!!s.:horqe ( cfs)
Braided river systems are favoured over meandering systems for heavy
mineral concentration because of their gravel-dominated alluvium and
associated fluvial process. In fact Collinson (1986) refers to braided
systems as "bedload streams", indicating that the coarser grain-sizes
dominate the associated deposits. This assertion is borne out by
consideration of the fact that the majority of economic alluvial placers
and paleoplacers are associated with the deposits of braided streams.
Examples include the diamondiferous gravels of the northern Cape (Dawson,
1990) and the Witwatersrand gold (Minter, 1978) in South Africa, and the
stanniferous placers of Ranong Province, Thailand (Aleva et al., 1973).
Basins into which fans build are quite variable in character. They may be
-68-
McGowan and Groat (1973) have produced an alluvia 1 fan prospect i ng mode 1
for fine-grained gold based on the distribution of other heavy minerals in
the Van Horn sandstone of west Texas. They noted that fine-grained heavy
minerals (mainly ilmenite and magnetite) occur throughout the fan, but are
concentrated in the distal-fan facies where sedimentation units are thin.
There is a close spatial association between heavy mineral accumulations
and trough crossbeds.
Minter (1978) developed this model in his synthesis of the heavy mineral
distribution within the Witwatersrand pa laeoplacers. Gold and uranium
occur in braided river environments in the mid-fan to distal facies of wet
alluvial fan-complexes. Gold concentrations are highest in a strike-
para lle 1 band across the fans, between 4 and 10 Km from the entry front
(Figure 30) . The gold here i s mostly concentrated in pebble-supported
conglomerates with a maximum clast size of between 2 to 4 cm. The highest
uranium values are displaced 2 Km down the palaeoslope (Figure 30). This
spatial distribution is controlled largely by the size of heavy and light
minerals available . Generally, the clast size of alluvial fans decrease
exponentially downslope (Collinson, 1986). It is therefore reasonable that
given the source size distributions, and the differential rates of heavy
mineral comminution, a strike-parallel band of accumulation would be formed
where the rat i 0 of 1oca 1 mean heavy mi nera 1 size to loca 1 mean 1i ght
mineral size was appropriate for heavy mineral deposition either by
i nterst i ce entrapment (more prox ima lly) or by entra i nment sort i ng (more
distally). This point would be shifted down palaeoslope for uraninite
compared to gold because its lesser density would demand a smaller light
mineral entrainment equivalence. The location of this band within the
palaeodepositional system might be predicted if the initial settling
velocity distributions of the gold were known.
-69-
1 1
1 I
1 1
1 I
r r
,.\ ~
~I ~I
'-1 ,-\
r r
V'I,.. "'I-
;:'\ ~\
""\ ""\
"I- "I-
';.1
~\ -'\
~,
"'.<-' ",'..po'",
",'~"\< Sample Sample
01-,>-.......... locafions ~>-'.... locafions
2 km .... _D Low 'Olold ....... Cl Low uranium
L....L.J E2J HIOh Qold
F1G.30. Heavy mineral enrichment at the system scale. The Steyn and Basal
placers of the Witwatersrand goldfield have high gold concentration
occnrring sourceward of high uranium concentrations. This is a
function of the sizes of light and heavy minerals available, and the
average roughness size and friction velocities down the fans.
(Modified after Minter, 1978).
Frostick and Reid (1985) draw attention to the fact that at the heads of
fans, coarse-grained deposits act as natural sieves for the entrapment of
sa ltating heavy minerals. The efficacy of interstice entrapment wi 11
decline away from the apical zones of alluvial fans because of the decrease
in grain size . Because of the rapid flow associated with these areas , only
coarse heavy minerals (larger diamonds, or gold nuggets for example) mi ght
be accumulated here (e.g. Matheys, 1990).
S0lJTH
A. PLAN VIEW _ _ _ _...:NORTH o
-- ----- ...
Re!o h~e Il1lel'lSlIy of fluvlol processes
Re he l brolded pollern
Surge dePOSits
f\
Lonqlludmal bars
CJ
(
•
Oecreose in grain size B. CRO~S ~ECTION
Decreos.e in thickness uf sec imentalic:n lJ'lils
FIG. 31. Alluvial fan prospecting model (modified after McGowan and Groat,
1973).
-71-
The all uv i a 1 fan prospect i ng mode 1 of Figure 31 shows that heavy minera 1
gra ins i zes decrease in the down-fan direct ion, the coarse-gra i ned heavy
minera 1s be i ng concentrated by i nterst i ce entrapment, the fine-gra i ned,
mainly by processes of hydraulic equivalence. Minter (1978) showed that
th i s trend may be extended to bra i ded systems genera 11y. The Steyn
palaeoplacer provides the longest, most extensively accessible, palaeoslope
exposure available in the Witwatersrand Bas i n and affords an ideal
opportunity to examine facies and associated changes in the heavy mineral
population, related to the decrease in hydraulic energy down the
palaeoslope.
~ :1
:1
PYRITE
- "•
HOOULE3
:1
:1
ZIIIICON
.
GRAlfitS
'e
~
Z
\
\~-" I
/~.-
I· '7"
c
~---~
: :1, tI. :1 J • I •
I
I
I
I
I
I
/
I
....
'..
i ,
, , ' ....
• S
allr
~
I / I ~ ............. -"4
..! =1, •.11 .. • • I I I .~ - - .....
::1 ~
0 ' /0
.... , . .
. I
I
; 0,I
:1 ... . ::1
. • (J)\
~
~~
.,
'0
/
.of
;
of
/
A'·.. , - , r- 0'
'-
;
\- 2.,6 - -
" 1
;
;
\ 2:4=~""''''~Y
,. ,,0 -
" 0
.
• HIGH GOLD CONTENT •
• HIGH URANIUM CONTENT. . "<---r-'''rl
" .
.. !!
~
..'" -
~
• ;;
;;
en " I
. U
,. ..• I
;; .. ;// I ••
"
Too
Spa, ..
ire
e. o II(
e/
I.
~
0
B •2 • • • .. ~
• ..• •.. ZOm
'---'
o Sampl. locatlonl
3"
" PhI Ical. contourl
Sam,le I ..
• 2,0
TABLE VI. The six principal facies assembhges in gravel- and sand- dominated
braided river deposits (after MiaJJ, 1978). Lithofacies codes as
in Table vn.
..
I chonnel
or 5, ~.
::::.=. ~
Iii!- I chonnel
'" .',
' ., debris flow "
:',9;-~~':': '. deposit !~ ~----:. system
.';';-~;~" ';:. :
..
.:' ~:. ~: ~;} ~ stream flow
"::.j~~~;~;:r
! cOl1)pound
,
,:~':f:;(~
bar Mt!it"''''' I
... V( I .,.
-J
l£;£;~r\?'i bors
I~ I
"lD':'-J0'-06C\ I m
fi'::J.:i.I:':.[:'; ~,
,,?,:l>l-.'1
~(ft~~ t 0
~;~:{~:i-i~i.i r
FIG. 33. Vertical profile models for braided stream deposits. Facies codes
to left of each column are given in Table VII. Arrows show
small-scale cyclic sequences. Conglomerate clasts are not shown to
scale. (Alter Miall, 1978).
-75-
Faefes
Code Lithofacies SedImentary 5tructures Interpretation
Jensen and 8ateman (1979) noted that "p l acer deposits do not form in the
downstream meanders of sluggish, old-age streams, because the stream
velocity is insufficient to transport heavy minerals." Kartashov (1970)
also noted that the more proximal reaches of rivers were more favourable to
placer deve 1opment, and that finer, and 1i ghter heavy minera 1s wou 1d be
transported to, and concentrated at, less proximal sites than coarser and
heavier heavy minerals. Thus gold placers might be expected to form close
to source, while rutile, ilmenite, zircon and monazite will continue to be
transported over considerable distances, leading to their major
concentration in coastal regions (e.g., Richard's Bay, South Africa).
The coarsest fractions are deposited in the deepest parts of the channel
and the finer sediments in the shallower, upper part of the point bar. As
-77-
the channe 1 mi grates 1atera lly (by eros i on of the outer bank), the point
bar likewise accretes laterally, forming a tabular deposit that fines
upwards. In most cases, the bed-forms are dunes in the deepest part of the
channel, and ripples in the shallower upper portion. Hence point-bar
deposits show an upward change from large-scale to small-scale cross-
stratification as well as a grain-size decrease.
/1
,,/' /
==
CD
to ~"
CD
"
~ ~ ,. • .,. ~I ' . i th ••• d ........ at beddi ..... \JowtIc:oI .urlll_ depo";' t,_ _ rbonll
CD CO Spot •• eak:io.o", cot"boMIhI _ .. ;0.... llaod .. ~bt., cMpOtil..t in boc.II--.,
CD
"
~ ... lnrw9I1.brol. bul"ro.. in low. port.
S-,aul 01.......
OI'-.,...hiIIIIM II _ _
I• •.
OI''''~'
CD
~.
to
CD to
"
.....
CO
...CD" CD
"'"
inlo ,;11.10111. ''''''''',brent bwll'O'W&" point-bar ....... till"".
'" ;:r
;:r
rt ~"
'" ~"'"
};' 0 CD
'<l
,..". ...
...rt til
n ~
",... "'" =
rt
"..... ==
CD
" ...=
CD
=
rt
l' ...." ...CDCO ~ '"
rrafed ot bot. Gnd scoller" lI'Ir!lU9f*&l
""'tI
porIed D bed-loao _ ,1".. bM form"
·dun..•. StrOl'lQ. ¥O,i a~ t ............
rt
= ,... Trou<;l'" o;:ton-.trollfl ,d. IO-90at! ;"'0 lu_
CIl
...0rt
",..
rt
0 .... '<l
CIl
(lQ
=
CD
... " "
11'Iio::1I. COMorr.d cros.-ur",. II .... 110. .
orld micfdr ••
SlItsl.". dents ,.".". r~ tonc:""rol. wh_
~ ..... _ d ........
(- CD
"111 101\01101'11
To some extent, the selection of sample sites is obvious from the models
presented in the previous sections. However, it is stressed that, as in
geochemical sampling, some orientation work is required in order to
discover the optimum heavy mineral sampling sites. In arid regions, for
example, a surficial lag may develop by wind deflation over unconsolidated
fluvial sediments, making this the most attractive sampling horizon.
Eva 1uat i on of "trap" sites has been performed by vari ous workers in
different environments, working with different heavy minerals. Clearly,
such stud i es are of 1imited use in new areas . However, certa i n samp 1e
sites do appear repeatedly in the 1 iterature as representing "good" trap
sites. A review of some of these sites follows.
The poorest samples were collected at sites with relatively few clasts,
a lthough some of these contained concentrat ions of finer probab ly non-
kimberl itic heavy minerals. Figure 36 summarises the trap sites in terms
of efficiency of concentrating the kimberlitic indicators.
rlow DlrecllOn
~*'
#~"b
POOR TO
MOOERATE
distinct obstrUction Sorting IS moderate to poor.
Bou'ders are rare or absent. Clasts mam[y range
rrom cobbJe$ to pebbles and may not Include well.
rounded l'/pes MatrIx contams sand and Silt.
FIG. 36. Broad field classification of heavy mineral trap sites. Diagrams
are of river bed cross-sections parallel to main water flow
direction. (After Muggeridge, 1989).
For example, McQuivey and Keefer (1969) showed how the intensity of
turbu l ence from the trough to the crest of ripples may be used to explain
the segregat i on of magnet ite and other fi ne heavy mi nera 1s beh i nd the
crests of ripples. Such studies are particularly useful in indicating
sites of concentration in suspension load-dominated meandering streams.
-81-
The experiments of Linkholm (1968) explained the abundance of gem and the
near absence of industrial quality diamonds on the west coast of southern
Africa. A ball mill was charged with gravel and steel balls, together with
s i x industrial and six gem quality diamonds. It was then run for a total
period of 950 h. Each hour, the contents was washed over a 60 mesh screen
and the oversize returned to the mi 11. After only 7h, the industrial
diamonds had disintegrated. After 950 h, the total weight loss of the six
gems was hardly measurable at (0.01%), whilst the gravel loss was 40%.
Surface textures are microrelief features of the grain surface and are
independent of size, shape or roundness. Less abrasion and transport are
required to modify these details in comparison with the change of grain-
size, shape and roundness. For examp Ie Wentworth (1922) determined that
about 560 m of fluvial transport would remove glacial striations from
limestone pebbles, and Bond (1954) noted that the frosting of sands of the
Kalahari desert is lost in less than 64 Km of transport down the Zambezi
River . The app 1i cat i on of surface texture stud i es of heavy mi nera 1s to
determine the distance from source is discussed in two examples : i)
kimberlitic indicators i and ii) detrital gold.
-82-
Kimberlitic Indicators
Mosig (1980) sub-dived picroilmenite and pyrope garnet into three classes
respectively, based upon their degree of wear. This classification can be
summarised as follows:
Picroilmenite
ii) Fresh worn picroilmenite. Grains 2-4 km from source are sub-rounded
with matt surfaces, and some are freshly fractured.
Garnet
iii) Worn garnet. At distances greater than 3 km, garnets become smooth
and rounded, and some may exhibit frosting.
Applying these criteria to the detrital gold from the Witwatersrand, the
transport distances for these grains appears to be in the order of 10 to 30
km. It should be noted, however, that much of the , gold of the
Witwatersrand may have been transported in association with sand-rich
sediment charges .
3 . 4 . 2 Population Ratios
Mos i g (1981) reports that the elongate, fibrous chrome diopsides from South
Australian kimberlites are broken down to sizes too small to recover beyond
3 km of transport in streams. Numeric ratios of chrome diopside to, say,
garnet may therefore give an indication of proximity to source.
3.4. 3 Conclusions
It is concluded that, whilst the use of transport trends will remain a tool
in mineral exploration, its application is limited to the broad scale, or
to projects in which some degree of control exists, derived from
orientation studies.
CONCLUSIONS
Detrita 1 heavy mi nera 1s are a 1imited group with d i st inct phys i ca 1 and
chemical properties. Their presence in alluvial systems of all kinds may
be of economic value both as mineral deposits in their own right (i.e.
placers) and as pathfinders to deposits drained by rivers.
Economi ca lly, the most important all uvi alp 1acers compri se go 1d, diamond
and cass iterite depos its. The genes i s of these is a consequence of the
complex interplay between tectonic setting and geomorphic evolution. It
can generally be stated that major alluvial gold placers develop on the
flanks of rising mountain chains during orogenesis, where the primary lode
gold was emplaced in the overriding plate during subduction of oceanic
crust, prior to collision. Where conditions are unfavourable for the
emplacement of I-type granitoids (i.e. where sUbduction of oceanic crust is
limited or absent) then primary hydrothermal gold is unlikely to be
emplaced, and · such orogenies are unlikely to spawn major deposits of
alluvial gold. A good example of such a collision belt is provided by the
European Alps. Here, the development of oceanic crust was limited and
the plate margin motion was dominantly lateral. Subduction of oceanic
crust was not therefore developed on any substantial scale, except over an
area between Yugoslavia and Iran (Evans, 1975). This has resulted in a
1ack of I -type magmati sm and the consequent absence of hydrotherma 1 go 1d
deposits (as well as related deposits such as porphyry copper). Molasse
basins marginal to the European Alps are not therefore likely to host
significant gold placers. Of course smaller gold placers may develop
locally under suitable conditions, adjacent to any pre-existing gold
deposit. An example might be the Caeonozoic alluvial gold deposits at
Bourkes Luck on the Blyde River in the eastern Transvaal, which have formed
in response to a rejuvenation event (or events) from primary Archaean
sources in the Pilgrims Rest goldfield 12 km upstream.
Studies of the dynamics of river systems and their facies models, aided by
empirical studies of heavy mineral distribution within rivers, have enabled
the development of prospecting models for the two main types of river
system ; bra i ded and meanderi ng. These mode 1s are intended to gu i de
exploration programmes to the most promising sites of heavy mineral
accumulation, and to indicate likely trends in their distribution. Grain
size is found to be an important parameter in governing this distribution.
Whilst it has been beyond the intended scope of this review to discuss
evaluation of placers, it should be clear that the genetic process history
of placers provides an indication of how to subdivide a deposit into
separate sampl ing "strata", each with its own characteristic grade and
tonnage. Examples mentioned in this review include the Auchas alluvial
diamond deposit on the Orange River in Namibia, which can obviously be
subdivided into the valley-fill and channel-fill facies (section 2.3 . 2);
the Otago alluvial gold deposits of South Island, New Zealand, which can be
subdivided into cycles of fluvial degradation developed in response to
intermittent tectonism (section 2.2.1); and the stanniferous placers of the
Tujuh archipelago in Indonesia, which can be subdivided into Kuli~ Kaksa
and Miencan concentrat ions, correspond i ng to the stages of placer evo 1ut i on
(sect ion 2.1. 4) . An understandi ng of the geomorph i c hi story therefore
provides descriptive and genetic terrain models which can also be used by
mining geologists.
ACKNOWLEDGEMENTS
I wou ld 1ike to thank De Beers Conso 1i dated Mi nes, Ltd. for 'permi ss ion and
financial support in the undertaking of the M.Sc . Exploration Geology
course and in particular my colleague Mike de Wit for his initial
encouragement.
Numerous people have willingly shared their experience and expertise, and
thanks are expressed to the following Rhodes University academic staff for
maintaining an open-door policy throughout the year: Prof. N. Hiller,
Prof. J.S. Marsh, Prof. H.V. Eales, Prof. R.E. Jacob, Dr. A.R. Butcher, Dr.
B. Teigler, and Mr C. Mallinson. I am particularly indebted to Prof. John
Moore, whose enthus i asm, constructive criti ci sms and abi 1ity to convey
often complex concepts in a readily understandable way, has been a constant
source of inspiration. Special thanks are also extended to Dr. Teigler for
proof-reading the text of this dissertation.
Adams, J., Zimpfer, G. L. and McLane, C.F. (1978) Basin dynamics, channel
processes, and placer formation a model study.
Econ. Geol., V. 73 : 416 - 426.
Aleva, G.J.J., Bon, H.E., Nossin, J.J. and Sluiter, W.J. (1973) A
contribution to the geology of part of the Indonesian
Tinbelt : the sea area between Singkap and Bangka
Islands and around Karimata Island. Bull. Geol.
Soc. Malaysia, No.6, pp. 257 - 272.
Atkinson , W.J., Smith, C.B. and Boxer, G.L. (1984) The discovery and
geology of the Argyle diamond deposits, Kimberley,
Western Australia. Proc. Australas. Inst. Min. Metall.
Ann. Conf . Darwin, Aug. 1984, 141 - 149.
Best, J.L. and Brayshaw, A.C. (1985) Flow separation - a physical process
for the concentration of heavy minerals within alluvial
channels. J. geol. Soc. London, V. 142: 747 - 755.
Bond, G. (1954) Surface textures of sand grains from the Victoria Falls
area. J. Sed. Pet., V.24 : 191 - 195.
Callaghan, C (1979) Tin placer deposits How and where to find them.
Geol. Surv. S. Afr. Open file Report No. 121, 36 pp .
Cobb, E.H. (1973) Placer deposits of Alaska. Bull. U.S. geol. Surv., No.1374 .
Dana, E.S. (1932) A Textbook of Mineralogy. John Wiley and Sons, New
York, 851 pp.
Emery, K.O. Noakes, L.J. (1968) Economic Placer Deposits of the Continental
Shelf, ECAFE Technical Bulletin Vol. 1, CCOP, Bangkok.
Frostick, L.E., Lucas, P.M. and Reid, I. (1984) The infiltration of fine
matrices into coarse grained alluvial sediments and its
implications for stratigraphical interpretation. J. geol.
Soc. London, V. 141 : 955 - 965.
Gurney, J.J . and Moore, R.O . (1991) Kimberlitic garnet, chromite and
ilmenite compositions applications to exploration.
-95-
Hails, J.R . (1976) Placer Deposits. In: Wolf, K.H. (Ed.) Handbook of
Strata-bound and Stratiform Ore Deposits, V.3: 123 -
239.
Hall, A.M., Thomas, M.F. and Thorp, M. B. (1985) Late Quaternary alluvial
placer development in the humid tropics: the case of
the Birim Diamond Placer, Ghana . J. geol. Soc. London,
v. 42 : 777 -787.
Henley, R.W., Norris, R.J. and Paterson, C.J. (1976) Multistage ore
genesis in the New Zealand geosyncline: a history of
post-metamorphic lode emplacement. Mineral. Deposita,
11 : 180 - 196.
Henley, R.W. and Adams, J. (1979) On the evolution of giant gold placers.
Trans. Instn. Min. Metall., 88 : B41 - 50.
King (1951) South African Scenery . Oliver and Boyd, Edinburgh, 379pp.
Komar, P.O. and Wang, C. (1984) Processes of selective grain transport and the
formation of placers on beaches, J. Geol . Chicago, 92 :
637 - 655.
Leopold, L.B. and Wolman, M.G. (1957) River channel patterns, braided
meandering and straight. U.S. Geol. Surv., Prof. Pap.
282 - B.
Matheys, F.G. (1990) The alluvial diamond deposits of the Lower Vaal River
between Barkly West and the Vaal-Harts confluence in the
northern Cape Province, South Africa, M.Sc. thesis,
Rhodes University (unpubl.).
McGowan, J.H. and Groat, C.G. (1973) Van Horn Sandstone, West Texas: An
alluvial "fan prospecting model for mineral exploration .
Bureau of Econ. Geology, Univ . Texas at Austin, 57 pp.
McQuivery, R.S. and Keefer, T.N. (1969) The relation of magnetite over ripples.
Prof. Pap . U.S. geol. Surv. 650 D, 244 - 247.
Moore, J.M. (1991) Course notes for modules EXT, M.Sc . Explor. Geol.
course, Rhodes Un i v. (unpubl.) 76 pp.
Mosley, M.P. and Schumm, S.A. (1977) Stream Junctions - a probable location
for bedrock placers. Econ. Geol., V. 72, p. 691 - 694.
Reed . L. E. and Sincla i r, I.G.L . (1991) The search for kimberl i te in the
James Bay Lowlands of Ontario. CIM Bulletin, V. 84 ,
No. 947 : 132 - 139.
Reid,I. and Frostick, L.E . (1985) Role of settling, entrainment and dispersive
equivalence and of interstice trapping in placer formation.
J. geol. Soc. London, V. 142, pp. 739 - 746.
-99-
Ruhe, R.V. and Cady, J.G. (1967) The relation of Pleistocene geolo9Y and
soils between Bently and Adair in southwestern lowa,
U.S. Dept. Agric, Tech. Bull. 1349.
Scott Smith, B.H., Danchin, R.V., Harris, J.W. and Stracke, J.W. (1984)
Kimberlites near Orroroo, South Australia. In
Kornprobst, J. (ed.), Kimberlites I: Kimberlites and
Related Rocks, pp. 121 - 142. Elsevier, Amsterdam.
Smith, D.G. and Smith, N.D. (1980) Sedimentation in anastomosed river systems:
examples from alluvial valleys near Banff, Alberta. J. Sed.
Pet., V. 50, No. I, p. 157 - 164.
Smith, N.D . and Minter, W.E.L (1980) sedimentological controls of gold and
uranium in Witwaterand palaeoplacers. Econ. Geol. Vol . 75
1 - 14.
Sneed, E.D. and Folk, R.L. (1958) Pebbles in the Lower Colorado River, Texas.
A study in morphogenesis. J. Geol. Chicago, V. 66 : 114 -
150.
Sobolev, N.V . (1980) Siberian diamonds - minerals from the Earth's mantle.
Indiaqua 26(3) : 9 - 15.
Toh, E.S.C. (1978) Comparison of exploration for alluvial tin and gold.
In : Jones, M.J. (Ed.), Proceedings, 11th Commonwealth
Mining and Metallurgical Congresss, Hong Kong: Trans,
Inst. Min . Met . , p. 259 - 278.
Tooms, J.S. (1987) Exploration for gold in the humid tropics. J. Geochem.
Explor., 29 : 437 - 439.
Van Wyk, J.P. and Piennar, L.F. (1985) Daimondiferous gravels of the lower
Orange River, Namaqualand. In: Anhaeusser, C.R. and
Maske, S. (Eds.), Mineral Deposits of Southern Africa,
Vol. II : 2209 - 2321. Geological Soc. of S.Africa,
Johannesburg.
Visser, J.N.J. and Loock, J.C. (1988) Sedimentary Facies of the Dwyka
Formation associated with the Nooitgedacht glacial
pavement, Barkly West district. S. Afr. J. Geol.,
V. 91 : 38 - 48.
Vlasov, K.A. (1958) Geochemistry and mineralogy of rare elements and genetic
types of their deposits, 3 : 614 - 633.
Ward, J.D. (1987) The Cenozoic succession in the Kuiseb Valley central
Namib Desert. Mem. 9, Geol. Surv. Namibia, 124 pp.
-102-
Watts, J .T., Tooms, J.S. and Webb, J.S. (1963) Geochemical dispersion of
niobium from pyrochlore-bearing carbonatites in
Northern Rhodesia. Inst. Min. Metall. Trans.
V. 72 : 729 - 747.