OOSTERVELD, M.M., CAMPBELL, D. and HAZELL, K.R.

Geology related to statistical evaluation parameters

for a diamondiferous beach deposit. APCOM 87. Proceedings of the Twentieth International Symposium on
the Application of Computers and Mathematics in the Mineral Industries. Volume 3: Geostatistics. Johannesburg,
SAIMM, 1987. pp. 129-136.
Geology Related to Statistical Evaluation Parameters
for a Diamondiferous Beach Deposit
M.M. OOSTERVELD*, D. CAMPBELL* and K.R. HAZELL**
*Ore Evaluation Department, De Beers Consolidated Mines Ltd, Kimberley, South Africa
**Geological Department, CDM (Pty) Ltd, Oranjemund, Namibia
Diamonds are mined from ancient raised beach deposits in southern Namibia
where complex geological controls have resulted in patchy diamond distribu-
tion, largely related to irregularities (trapsites) in the bedrock.
The distribution of the number of diamonds per sample is a discrete, reverse-
J shape distribution which can be characterized by two parameters. These
parameters have been transformed into a Beta and Theta Contribution which
can be interpreted as reflecting the' quantity' and' quality' of the trapsites.
The relationship of statistical parameters, obtained during the sampling of
an ore deposit, to the geology of the deposit is an important element of ore
reserve estimation. The large-scale regional trends in the contributions have
been explained in terms of the geology.
Introduction
Two models are necessary for evaluating an
ore deposit. The geological model provides
an understanding of the distribution of the
ore minerals. This model is often complex
owi ng to the interact i on of the many and
Diamonds have been mined, since their
discovery in 1928, from ancient beach
deposits in Namibia in a narrow coastal
belt which stretches 100 km northwards from
va ri ed geological factors which can
i nfl uence the formation of the orebody or
its subsequent deformation.
The statistical model is obtained from
taken from the deposit. These
determi ni st i c mode 1 s genera lly represent
only a few aspects of the deposit; however,
they usua lly refl ect the maj or geo 1 ogi ca 1
controls. The evaluation of a deposit is
considerably enhanced if a relationship can
be found between the statistical and
geo 1 ogi ca 1 pa rameters.
In this paper the relationship between
the geology and the statistical parameters
is di scussed for a compl ex di amondiferous
beach depos it.
the mouth of
deposits were
one-metre-wide
the Orange Ri ver. The
initially sampled by
trenches excavated 500 m
apart across the stri ke of the deposit.
Each trench was divided into sections 5 m
long and a total of more than 40 000
sections was sampled.
Two important requirements for the
evaluation of a diamond deposit are:
(i) the density distribution of the
diamonds;
(ii) the size (mass) distribution of the
diamonds.
Two characteristics of the density
distribution and one parameter of the size
distribution were used for with
the geology.
STATISTICAL EVALUATION PARAMETERS FOR A DIAMONDIFEROUS BEACH DEPOSIT 129

General geology
The COM raised beach complex comprises a
series of shingle and basal
conglomerates overlain by and interbedded
with a variety of aeolian, marine, and to a
1 esser extent, terrestri a 1 sands. The
deposits, which vary in width from 3 000 m
in the south to 200 m in the north, rest on
narrow marine abrasion platforms cut into
Precambrian schists and phyllites.
The schist bedrock has been extensively
gullied by wave action and, where well
developed, the bedrock gullies assume a
characteristic pattern, apparently
controlled by the slope and structure of
the bedrock. In places the erosive action
of the waves has abraded the coastl ine and

-25
'" -20
<!)
b
Q)
El
.S
Q)
;-
<!)
-15
c;;
<!)
'"
<!)
;-
0
.0
ro
15
-10

<!)
:r:
-5
-0
-3
resulted in the formation of scarps and
cliffs in the bedrock.
The COM raised beaches reflect a
succession of marine transgressions each
followed by a major regression. Above sea
level six distinct transgressions may be
recogni sed as evi denced by the appropri ate
raised beach or fossil strand line.
However, it is likely that certain basal
conglomerates represent relict deposits
which predate the six beaches currently
recognised, and it is possible that there
has been considerable reworking of these
deposits. A schematic cross section of the
deposits is shown in Figure 1. The beaches
are named A to F, with the F Beach being
the oldest and highest beach and the A
-3
050 045 040 035 030 025 020 015 010 005 000 5 10 15 20
SHElF V
25 30 35 39
,
,

,
,
,
-I
" ,
5m--r:.. : SHElF I
1 .. 25m.:
! ..
SHElF ill SHElF IV
TOTAL DISTANCE = 445m
FIGURE 1. Schematic cross-section through beach deposits.
130
GEOSTA TISTICS: GEOLOGY- RELATED PROBLEMS
Beach being the youngest and lowest beach.
It is envi saged that the di amonds found
in the beach deposits were derived by
reworking of the diamondiferous gravels of
the Orange River delta during marine
transgressions.
Areas of diamond concentration
The distribution of diamonds is different
for each of the six recognised marine
transgressions. These different distri-
butions are primarily related to the length
of still stand of the sea whi ch i nfl uenced
the degree of abras i on of the ma ri ne
p 1 a tform and the amount of reworki ng tha t
took place. The distribution of diamonds
is a 1 so i nfl uenced by the amount of
diamondiferous gravel available during the
transgression. This entrainment of gravel
was probably inversely related to the rate
of the rise in the level of the sea during
the transgression.
Diamonds are associated with gravel
horizons. The elements of a typical beach
depos it a re a beach crest, a beach slope
and a shelf, and diamonds may occur in each
of the three elements of the beach.
However, the shelf gravels carry by far the
most significant quantities of diamonds
because they are deposited in a high energy
environment, where the erosive power of the
waves creates irregularities in the bedrock
forming trapsites, and where the turbulence
of the water results in a concentrating
action.
Thus, the distribution of the diamonds is
directly related to the presence of a
diamond-carrying gravel, to the 'quantity'
and 'qua 1 ity' of the traps ites and to the
degree of reworking.
Bedrock gullies (trapsites)
There are two distinct gully sets, one
trendi ng northeast-southwest and normal to
the coast and another trending roughly
north-south. The former set, or
'consequent' gullies, reflect the control
exerted by the seaward slope of the marine
abrasion platform while the latter set, or
'subsequent' gullies, are controlled by the
well developed foliation present in the
schi st.
Four stages in gully development are
recognised, youthful, mature, post-mature
and wave-cut platform.
The major differences in bedrock morpho-
logy, both within a shelf and between
she 1 ves , refl ect the 1 ongev ity of the
sti 11 stands of the sea, the competence of
the bedrock, the differing resistance to
wave attack of various lithologies and the
erosive power of the sea during each
still stand.
The type of gully formed also depends
upon the amount of boulder material
available. If few boulders are available
structural features will be exploited
(subsequent gullies) while an abundance of
boulders will increase the erosive action
of the waves and consequent gullies will be
formed irrespective of the competence of
the bedrock.
Diamond concentration within gullies (trap sites)
The concentration of diamonds is largely
related to the degree of reworking during
deposition which is in turn determined by
the degree of turbulence and the length of
stillstand. Thus high concentrations can
occur on both smooth wave-cut platforms as
a result of a lengthy stillstand as well as
on gullied bedrock where vigorous
turbulence has taken place.
It has been found that adjacent similar
trapsites can contain vastly differing
numbers of diamonds. This can be ascribed
to the complex hydraulic conditions which
prevail in the surf zone during deposition.
STATISTICAL EVALUATION PARAMETERS FORA DIAMONDIFEROUS BEACH DEPOS IT 131
Statistical evaluation parameters
The grade can be calculated from the
density and size frequency distribution
Grade G = o x S [l]
where 0 is an estimate of the average
number of stones per unit and S is an
estimate of the mean value of the stone
size.
Diamond density distribution
At the sample size at which the COM deposit
was sampled the di screte frequency
di stri buti on of the number of di amonds per
samp 1 e is that of the reverse J-shape type
with a relatively large number of barren
samples and a very long upper tail
(Figure 2).
A model which represents this diamond
density distribution well is the compound
Poisson distribution developed by Sichel.
The derivation of this model and the
estimation of the parameters is shown in
Sichel
1
.
<J)
<l)
"i5.
El
c<l
<J)
""' 0
....
<l)
.0
a
::l
z
80
60
40
20
o
o
Ihhll
20 40
Number of stones
FIGURE 2. Stone density histogram
132
60
The density of the compound Poisson
distribution is given by
where
o ae/21l="EJ
S/1l="EJ
S = ae/2
[2J
[3J
[4J
and where e and a are distribution
parameters while S is a function of e and
a.
The procedure followed at COM was to
divide the beaches into geologically
homogeneous ore blocks (superb locks) and to
estimate both the a and e parameters by
means of a joint maximum likelihood
estimati on method wi th both a and e
unknown.
It was found that to obtain a reasonable
estimate of the e parameter, a large number
of samples was required and in most
superblocks the number of samples was
insufficient. A smoothed value for e was
obtained by calculating a moving average of
e/v'1-8 for groups of 50 samples. The
a parameter was then calculated from a
maximum likelihood equation with e fixed a
priori. The S was calculated from [4J.
As shown above, the stone density can be
split into two useful components. Because
these two components are multiplied [3J, a
simple addition of the components can be
obtained by converting to logs.
log 0 = log S + log ( 1 / ~ ) [5J
Log S and log (1/11 - e) are called the
Beta and Theta Contributions respectively
and are discussed below with respect to the
geology.
Diamond size distribution
It has been found at COM that diamonds in
geo log i ca 11 y homogeneou s areas gene ra 11 y
display a lognormal size distribution. The
GEOSTATISTICS: GEOLOGY-RELATED PROBLEMS
mean
the
size parameter S was
correlation between
obtained
geology
statistical
distribution.
pa rameters from the
for
and
size
The mean stone size was calculated using
the method as described by Siche1
2
. The
maximum 1 i ke 1 i hood estimation
(t'-estimator) for the mean of the
two-parameter lognormal population was
applied described by
3
By as Oosterveld .
employing this estimator the variance of
the mean value distribution was minimised.
Because it was found that the observed
sample variance of the logarithm of the
size was generally smaller than unity, and
the number of observations in the sample
usually exceeded 20, it was possible to use
the simpler t'-estimator instead of the
theoretically more correct t-estimator.
Statistical parameters related to geology
The Beta Contribution characterises the low
end of the J-shaped distribution such that
the lower the value the greater the number
of barren samples.
Therefore, if positive samples are
equated to trapsites, it can be stated that
an increase in the Beta Contri buti on
(decrease in barren samples) indicates an
increase in the number (quantity) of
trapsites.
An increase in the Beta Contribution
could also be related to an increase in the
supply of diamonds which is a function of
the rate of the marine transgression and
the availability of diamonds at the time of
deposition.
The Theta Contribution characterises the
upper end or tail of the J-shape distri-
bution such that the higher the value the
longer the tail. In geological terms, a
higher Theta Contribution impl ies more
diamonds in trapsites and it can be said
that the 'quality' of the trapsites has
improved.
The Beta and Theta Contributions and the
stone density and diamond size are
schematically represented in Figures 3 to 6
in which bar charts represent the value of
the variables for each block as a function
of di stance from the mouth of the Orange
River.
From the geological description, it is
apparent that these deposits and the
distribution of diamonds are the result of
extremely complex interactions between a
number of geological and hydraulic
pa rameters. Therefore, no attempt has been
made to explain small changes in the
contributions but rather to explain the
larger scale regional trends in terms of
the geology.
Observations in the trends of the various
contributions with possible geological
explanations are given below.
A
B
c
E
F
100
Kilometres from Orange River mouth
FIGURE 3. Beta Contribution to the stone density
STATISTICAL EVALUATION PARAMETERS FOR A DIAMONDIFEROUS BEACH DEPOSIT 133
A bedrock platforms giving a more even
B
! l rmcrrImmfD cl
'-
.3 <-----J..L.I u....L.-J.lb nf.L...W...J.fdJJ...1.J..Ll..LmU-l-J.JUIT-L...ll.L.JIf..J.......J....f]!'--!.J.......l! d1f---'-I...J..
r
.. _D I
1
__ .... ... . ...... El
ITb rDUJJlUJTJTInrrn crnrurJill .
o
Kilometres from Orange River mouth
FIGURE 4. Theta Contribution to the stone density
Beta Contribution ('quantity' of trapsites)
Whi 1 e there are differences in the scale
and trend of the Theta Contribution between
the beaches, there is a certain degree of
similarity in the Beta Contribution. This
degree of similarity in trap 'quantity' is
to be expected when dealing with beaches in
the same orebody with similar bedrock
characteristics.
Furthermore, though the deposit is
divided up into six recognised beaches, the
abrasion platforms upon which these beaches
lie might have been developed during
earl ier marine transgressions. Thus two of
the recognised beaches might lie on one old
abrasion platform and therefore have
similar trapsite development.
The Beta Contribution is highest on the F
and E beaches. Both the F and E beaches
are deposited on relatively smooth mature
134
distribution of the diamonds. These
deposits are also more massive and
therefore there are fewer barren samples.
This is probably related to supply, with
slow marine transgressions resulting in a
hi gh degree of gravel entra i nment from the
Orange River delta. In addition, it is
likely that there has also been reworking
of older deposits.
The Beta Contribution is very low in the
south of the E beach. This could be
explained by the rapid accumulation of
material from the delta allowing little
time for diamonds to be concentrated at the
base of the deposits in bedrock trapsites.
On the D beach, the Beta Contribution is
low in the south and high in the north.
The D beach is characterised by a thin,
poorly developed gravel band indicating a
rapid marine transgression allowing only a
small supply of gravel (and diamonds) from
the submari ne delta. Therefore though
trapsites might exist it is likely that few
diamonds were available at time of
depos it i on.
Further to the north, the coast steepens
and the hi gher Beta Contri buti on of the D
beach might be the result of reworking of
older deposits or of sampling conglomerates
from an earlier transgression.
On the A, Band C beaches the Beta
Contribution is variable with no apparent
correlation with major geological features.
Theta Contribution ('quality' of trapsites)
The Theta Contribution is highest on the F
and B beaches and the trends in the
contribution are similar for the two
beaches. Both these beaches are backed by
cl iffs cut into the bedrock duri ng marine
transgressions. The base of the cliffs
forms an excellent environment for the
concentration of diamonds resulting in high
GEOSTATISTICS: GEOLOGY-RELATED PROBLEMS
Cl
bJ)
5
Theta values.
Very high Theta Contributions occur in
the north on beaches Band A. These hi gh
values are not only caused by the B beach
cliff but also by the high degree of
reworking of older deposits in this area.
In the north there is an increase in the
coastal slope and a consequent reduction in
beach wi dth whi ch has the effect of
'telescoping' the deposits resulting in
more reworking of the gravels.
There are also indications of crustal
down warping which has affected the older
normally more elevated, beaches causing
them to be reworked. Hi gh Theta values may
also be obtained in embayments which form
good concentrating environments and which
occur in the north of the deposit.
The Theta Contribution is low on most
beaches between approximately 40 km and
A
B
c
D
70 km north of the Orange River mouth.
This is the result of less favourable gully
development caused by incompetent bedrock
where fewer good trapsites were formed. It
has been noted during mining operations
that the less competent bedrock occurs from
42 km to 75 km north of the Orange Ri ver
mouth.
Stone density
The stone density (Figure 5) is a
combination of both the Beta and Theta
Contributions. Therefore a hi gh stone
density could either be a function of a
high Beta Contribution or a high Theta
Contribution or both.
Thus high stone densities on the F and B
beaches are the result of good diamond
concentrations against the cl iffs as well
as a long stand of sea level and reworking
of older deposits as evidenced by the
smooth bedrock.
The low stone densities between 40 km and
60 km in the D, E and F beaches are
probably caused by poor trapsite
deve 1 opment on incompetent bedrock. Though
offset from the older beaches the low stone
densities on the B beach also correspond to
softer bedrock areas.
The low stone densities on the southern
half of the D beach are the result of a
rapid marine transgression and a poor
supply of diamonds.
Stone size
The stone size is a geological parameter
E which decreases rapidly north of the Orange
River on all beaches (Figure 6).
o 20 40 60
Kilometres from Orange River mouth
FIGURE 5. Stone density
F
80 100
Samples of alluvial gravels in the Orange
River appear to indicate that the initial
influx of diamonds to the coast had a
larger stone size. This might offer an
explanation for the average stone size on
the older F and E beaches being much larger
than the more recent A and B beaches.
STATISTICAL EVALUATION PARAMETERS FORADIAMONDIFEROUS BEACH DEPOS IT 135
A
B
c
It has been found that certain general
trends in the statistical parameters
obtained from the sampling of
diamondiferous beach deposits in Namibia
can be related to the geology of the
deposits.
Acknowledgements
The authors wish to thank Prof. H.S. Sichel
and Or W.J. Kleingeld for assistance in
D preparing this paper. Thanks are also due
to the members of the CDM Geol ogi ca 1
Department for many hours of discussion.
E
F
100
Kilometres from Orange River mouth
FIGURE 6. Stone size
Grade
The grade of the deposit is a combination
of the stone density and the stone size.
The high grades of the F beach a re the
result of a high stone density and stone
size wh i 1 e on the younger beaches the high
stone densities are transformed into lower
grades by a small stone size.
Conclusions
The relationship of statistical parameters,
obtained during the sampling of an ore
depos it, to the geology of the depos i t is
an important feature of ore reserve
estimation. However, the geological
controls influencing the distribution of
ore minerals are invariably complex and can
only be modelled in broad terms by the
statistical parameters.
136
The permission of the Executive Director
and the General Manager of CDM (Pty) Ltd of
Namibia to publish this paper is gratefully
acknowledged.
References
1. SICHEL H.S. Statistical Valuation of
Diamondiferous Deposits. In: Proceedings of
the International Symposium on the
Application of COmputers in the Mineral
Industry. South African Institute of Mining
and Metallurgy, Johannesburg, 1972.
pp.17-25.
2. SICHEL H.S. The estimation of means
and associated confidence limits for small
samples from lognormal populations.
In: Sympos i um on Ma themati ca I Stati s tics and
Computer Applications in Ore Evaluation.
South Afri can In st itute of Mi n i ng and
Meta 11 u rgy 1966, pp. 106-122.
3. OOSTERVELD M.M. Ore reserve estimation
and depletion planning for a beach diamond
deposit. In: Proceedings of the Tenth
International Symposium on the Application
of COmputers in the Mineral Industry.
South African Institute of Mining and
Metallurgy, Johannesburg, 1972. pp. 17-25.
GEOSTATISTICS; GEOLOGY-RELATED PROBLEMS