You are on page 1of 18

Chapter 4

WHAT IS A V A R I O G R A M ?

S Y N O P S I S — In this chapter we will try to show that the idea of a uariogram


introduces itself naturally when one starts to think about estimation and
variability problems. We will first define what the variogram function is,
then show that it incorporates several geological features which are impor­
tant in mineral deposit valuation, such as the continuity, size and shape of a
zone of influence. Then a little arithmetic will show us that the variogram
function intervenes in all estimation variances and block variances compu­
tations. This will be illustrated by two exercises in order to show what kind
of real results one can expect from the variogram, and finally before going
through the theoretical development we will show how one can compute a
variogram in a simple case, and introduce a most frequently used variable in
ore reserve computation, the accumulation.

4.1 SPATIAL C O R R E L A T I O N

All t h e geological features w h i c h w e have m e n t i o n e d in t h e previous


chapter could in fact be summarized under o n e n a m e . T h e y represent the
similarity or rather the dissimilarity w h i c h exists b e t w e e n the grade at o n e
p o i n t and t h e grade s o m e distance away. T h e error of estimation is thus a
function of t h e similarity w h i c h can be e x p e c t e d b e t w e e n samples and a
block. This e x p e c t e d variation can be called Spatial Similarity or rather
Spatial Correlation. We n o w see that w h a t w e have t o d o is t o find a w a y t o
characterize that similarity of grades within a deposit or, rather, within a
h o m o g e n e o u s area of the d e p o s i t where w e can suspect that t h e geological
relationships are t h e same and where c o n s e q u e n t l y the precision of an
estimation procedure for a given geometrical pattern depends only o n the
pattern and n o t o n t h e particular l o c a t i o n in the h o m o g e n e o u s area.

4.2 DEFINITION OF THE V A R I O G R A M

The m o s t natural w a y t o compare t w o values, say t w o grades Z(x) and


Z(x + h) at t w o points, x and x 4- h, h feet away from x, is t o consider their
difference. Since w e are n o t really interested in the sign of the difference
but rather in the absolute value, w e should consider the value \Z(x) —
Z(x + ft)|. This value expressing the dissimilarity b e t w e e n t w o particular
points is of very little interest. We rather w a n t the average difference be­
t w e e n t w o points h feet apart. Thus w e should consider the value {Average
\Z(x) — Z(x 4- h) |} for all possible p o i n t s x and x + h. Since absolute values
74

are difficult in calculus, w e will rather consider the squared differences and
thus select as our dissimilarity f u n c t i o n :

2y(h) = A V E [Z(x) - Z(x + h)]2


(4.1)
w h i c h should be properly expressed as:

2y(h) [Z(x)-Z(x + h)] dx


2
(4.2)
v K
where V is the v o l u m e of t h e ^ e p o s i t V.
This is the variogram, 2y(h). It is a function of a vector, in other words, a
distance and the orientation of that distance, and it expresses h o w grades
differ in average according t o the distance in that direction.
This definition involves an integral, in fact, a triple integral calculated over
the w h o l e deposit. In practice w e will have t o estimate that integral through
the available data^ S u p p o s e w e have a total of N(h) pairs of samples sepa­
rated b y a vector h, t h e n w e will estimate the variogram b y :
1 N
<*> -+
27(A) = I [Z( )-Z( +h)]
Xi Xi
2
(4.3)
N(h) i=i

As a simple case, take samples regularly distributed along a line as in


Fig. 5 2 ; w e have n samples at an interval of d feet, thus w e will have (n ~ 1)
pairs t o c o m p u t e y(d), (n — 2) t o c o m p u t e 7 ( 2 d ) and so on.

F i g . 5 2 . S c h e m a t i c c o m p u t a t i o n o f a v a r i o g r a m , u s i n g pairs o f s a m p l e s a g i v e n d i s t a n c e
apart.

Y(h)

Fig. 5 3 . Typical variogram from a lead—zinc d e p o s i t s h o w i n g a g o o d c o n t i n u i t y . The


m i n e r a l i z a t i o n is d i s s e m i n a t e d .
75

4.3 THE V A R I O G R A M AS A GEOLOGICAL F E A T U R E S DESCRIPTOR

All the geological characteristics w h i c h w e have m e n t i o n e d so far have t o


appear quantitatively in the variogram. This should help t o m a k e precise the
"feeling" of a geologist, specially at the beginning of an operation.

4 . 3 . 1 The continuity

T h e c o n t i n u i t y is reflected b y the rate of growth of j(h) for small values


of h. In a sedimentary deposit, changes usually occur very slowly. We obtain
variograms like t h e o n e o f Fig. 5 3 (a lead—zinc d e p o s i t ) , w i t h a gentle, regular
growth from zero. In other cases the mineralization m a y occur as nuggets,
or blobs, o f t e n c o n c e n t r a t e d in veinlets, t h e n rapid changes m a y happen over
a very short distance. This will be called later a N u g g e t Effect. We see such a
variogram (from a m o l y deposit) in Fig. 5 4 . T h e c o n t i n u i t y m a y also be
c o m p l e t e l y non-existent. Then w e have a variogram like t h o s e of Fig. 5 5 :
whatever t h e distance b e t w e e n t w o samples, t h e y are simply i n d e p e n d e n t of
each other. This is c o m m o n in gold deposits and this is w h y that t y p e is
called t h e Pure Nugget Effect t y p e .

ii
V(h)

T -
Co i i i i -i •
1
Fig. 5 4 . Variogram f r o m a m o l y b d e n u m d e p o s i t s h o w i n g c o n t i n u i t y curtailed b y a high
l o w - s c a l e v a r i a b i l i t y ( C o ) . T h e m i n e r a l i z a t i o n o c c u r s in t h i n (a f e w m m ) v e i n l e t s .

Y(h) f I

T i l I I I I I 1

0' 200' 400' 600' 800'


Fig. 5 5 . Variograms f r o m a n o t h e r m o l y b d e n u m d e p o s i t s h o w i n g n o c o n t i n u i t y . Mineral­
i z a t i o n o c c u r s in t h i c k ( 1 ) v e i n s .
76

4 . 3 . 2 The zone of influence

T h e z o n e of influence is the z o n e b e y o n d w h i c h the influence of a sample


disappears. It is normal t o characterize that z o n e of influence in a given
direction b y t h e distance at w h i c h t h e variogram eventually reaches a plateau.
Typical z o n e s of influence can be seen in Figs. 5 6 and 5 3 . In Fig. 5 6 w e
have a 2 0 0 ' z o n e o f influence in an iron deposit. In Fig. 5 3 the range of in­
fluence (we will say in short, range) was o n l y 3 5 ' . In Fig. 5 5 o n e can say
that it is less than o n e sampling interval, i.e. virtually zero.
Also n o t e that in s o m e instances the further apart the samples, the m o r e
different t h e y are. This happens frequently in h y d r o thermal deposits: the
variogram s h o w n in Fig. 57 is from a Cu—Ni deposit. When a range exists,
the variogram is said t o be of a transition t y p e .

50' 100' 150' 200' 250' 300'

Fig. 5 6 . A very " g o o d " variogram f r o m an iron ore deposit s h o w i n g a 2 0 0 (vertical) z o n e


of influence.

200' 400' 600' 800' DISTANCE (log)

F i g . 5 7 . V a r i o g r a m o f n i c k e l g r a d e in a s u l f i d e v e i n .

4 . 3 . 3 The anisotropics

The anisotropics are easily depicted b y c o m p u t i n g the variogram in dif­


ferent directions. On Fig. 5 8 for instance one sees four variograms from four
directions in a porphyry m o l y b d e n u m . T h e y can be considered as identical,
77

Fig. 5 8 . F o u r variograms c o r r e s p o n d i n g t o four different d i r e c t i o n s in a p o r p h y r y m o l y b ­


d e n u m d e p o s i t . (1 = E—W, 2 = N — S , 3 — v e r t i c a l , 4 = c u b e d i a g o n a l ) s h o w i n g n o c h a n g e
of behaviour with direction.

ft HORIZONTAL RANGE d J
I 1 1 1 1 1 1— —P* 1

DISTANCE

F i g . 5 9 . H o r i z o n t a l a n d v e r t i c a l v a r i o g r a m s in a n i r o n m i n e , s h o w i n g d i f f e r e n c e o f b e ­
haviour with direction.

t h e c o n t i n u i t y of t h e ore is thus t h e same in all directions. This is t h e sim­


plest case but, unfortunately, t h e least frequent! A m o r e c o m m o n situation
is seen o n Fig. 5 9 . The vertical and horizontal range of influence in an iron
deposit differ w i d e l y . This can be interpreted by saying that a given vertical
distance is equivalent t o the same horizontal distance multiplied b y an
anisotropy factor k (greater than o n e in this case). For instance if the ani-
s o t r o p y factor is 3 , t h e n ten vertical feet are equivalent t o thirty horizontal
feet as far as grade variations are concerned.
Other t y p e s of anisotropics are e n c o u n t e r e d . For instance, w h e n a layering
of t h e mineralization exists, then the global variation in any direction can be
t h o u g h t of as the sum of t w o i n d e p e n d e n t sources of variation, o n e isotropic
c o m p o n e n t plus a zonal c o m p o n e n t , w h i c h d e p e n d s o n l y o n t h e n u m b e r of
layers crossed. The difference b e t w e e n the grades of t w o points depends
78

Y(h)

( 2 ) vertical (3) oblique

(1) horizontal

i i n i
50' 100' 200' DISTANCE (log)

F i g . 6 0 . D i f f e r e n c e o f c o n t i n u i t y in t h r e e d i r e c t i o n s for t h e t h i c k n e s s o f a c o a l s e a m , as
s h o w n b y t h e t h r e e v a r i o g r a m s in t h e s e d i r e c t i o n s .

Y(h)

(1) vertical
variogram

( 2 ) horizontal
variogram

"I 1
10m 30m 50 m DISTANCE

F i g . 6 1 . D i f f e r e n c e o f c o n t i n u i t y in t h e h o r i z o n t a l a n d v e r t i c a l d i r e c t i o n a t t h e E x o t i c a
M i n e , as s h o w n b y t h e v a r i o g r a m s .

thus n o t o n l y o n their relative distance, b u t also o n their difference of


elevation in the layered system. N o t e that the layering m a y n o t be visible.
Such variograms are m o r e c o m p l e x t o interpret. T w o different t y p e s can be
seen in Fig. 6 0 and 6 1 .

4.3.4 Conclusion

We see that the variogram takes i n t o a c c o u n t all the geological features


w h i c h w e m e n t i o n e d earlier. Thus w e can e x p e c t that an estimation pro­
cedure based on it can be geologically reliable. We will n o w s h o w h o w the
estimation variance is a function of the variogram.
79

4.4 THE V A R I O G R A M AS THE F U N D A M E N T A L F U N C T I O N IN E R R O R


COMPUTATIONS

A l t h o u g h w e have n o t y e t given t h e theoretical probabilistic m o d e l s


describing t h e grade o f a d e p o s i t as a r a n d o m f u n c t i o n , w e can write a f e w
lines of statistics in order t o appreciate w h a t is e x a c t l y required in error
estimation.

4 . 4 . 1 The variance of the error of estimation

L e t us try t o appreciate t h e variance of t h e error of estimation e,- w h i c h


w e m e n t i o n e d before. The Z(V ) are true u n k n o w n grades of the blocks
t

V i9 t h e Z*(Vi) are t h e linear c o m b i n a t i o n s Z*(VJ) = 2 " a Z ( X ) of t h e= 1 y ;

k n o w n sample grades at l o c a t i o n s Xj(j = 1 . . . n). T h e n :


9 9

VAR(e ) f = VAR[Z(y )-Z*(V )] f i

V A R (e,-) = VAR[Z{V )]-2COV[Z(V ) Z*(V )]


i i 9 i +VAR[Z*(V )] Z

VAR(e,) = VAR[Z(V )] t -2 I a COY [(Z(V )


} t 9 Z(Xj)]
j
+ 11 aja k COY[Z(Xj) 9 Z(X )] k (4.4)
j k
If w e consider each o f these V A R , and COV separately, w e will see that
t h e y should be c o m p u t a b l e from t h e variogram.

4 . 4 . 2 The variance of the grade of blocks: VAR [Z(V )]


t

T h e variability of the grade of blocks should d e p e n d on t h e average dif­


ference in grade w h i c h exists b e t w e e n any t w o p o i n t s inside the block. If
this average difference is l o w , then the variance of the b l o c k should be l o w .
If it is high, t h e variance will be high. Since t h e difference b e t w e e n the
grades at t w o p o i n t s is expressed b y the variogram, w e can say that w e
should be able t o c o m p u t e the variance of the blocks from the variogram.
This will be strictly c h e c k e d and proved later.

4 . 4 . 3 The covariance of the grade of a Mock and the grade of a sample:


COV[Z(V ),
t Z(Xj)]

This value expresses the correlation w h i c h exists b e t w e e n the grade of a


block and t h e grade of a given sample. It should again obviously be equal
t o the average correlation b e t w e e n any p o i n t in the b l o c k and a p o i n t in
the sample. This correlation being expressed b y t h e variogram, w e thus
can state that o n c e w e k n o w the variogram w e will be able t o k n o w t h e
covariances.
80"

4 . 4 . 4 The covariance of the grades of two samples: COV[Z(Xj), Z(X )]


k

After w h a t w e have just said w e see that this is just a particular case of
C O V [ Z ( V j ) , Z(Xj)]. This time w e k n o w that t h e correlation or rather, t h e
lack of correlation b e t w e e n t w o samples is directly expressed b y the vario­
gram function. Thus C O V [ Z ( X ) , Z(X )]
y k should be s o m e constant minus
y(h ) if h is the distance b e t w e e n samples Xj and X .
jk jk k

4.5 CONCLUSION

We have n o w c o m p l e t e d our intuitive e x a m i n a t i o n of the problem of ore


reserves estimation. We have reached the stage where w e k n o w theoretically,
h o w t o c o m p u t e the magnitude of the error involved in one procedure or the
other. Thus w e should be able t o select the best m e t h o d , the one w h i c h will
give t h e m i n i m u m variance. What is left n o w is t o firmly establish t h e theory
which is called the T h e o r y of Regionalized Variables and, secondly, t o
e x a m i n e w h i c h practical c o m p u t a t i o n a l steps and interpretations are in­
volved in practical cases. The actual o p t i m i z a t i o n procedure — kriging —
will be seen later in Chapter 9. Before w e proceed, w e will s h o w in t w o exer­
cises h o w t h e variogram helps t o c o m p u t e all w h a t w e n e e d and h o w t o c o m ­
pute it in a simple case. T h o s e w h o are n o t concerned w i t h the theoretical
basis or practical aspects of variograms and variance c o m p u t a t i o n s can
directly proceed t o Chapter 9.

4.6 EXERCISES

4 . 6 . 1 Variances and the variogram

This exercise is the follow-up of Exercise 3 . 2 . 5 . In that previous exercise,


w e had s h o w n w h i c h physical realities our c o n c e p t s were covering, namely
w h a t b l o c k variances and estimation variances are. A t that t i m e , in order t o
c o m p u t e these values, w e have had t o use block grades w h i c h are normally
only available after t h e material has been m i n e d out, i.e. t o o late. In fact w e
never k n o w the real grade of blocks, so that in daily mining practice it is
n o t possible t o c o m p u t e the variances w h i c h w e have just obtained, in the
w a y w e did.
Geostatistics has s h o w n that this is, however, possible. From o n e drilling
campaign o n e can estimate the variogram of the deposit and then fit a m o d e l
t o it.
This defines a few parameters like the z o n e of influence in all directions
(which w e call the range, a ) , the sampling error (which w e call the nugget
effect, C ) and the variance of the samples (which w e call the sill, C 4- C ) .
0 0

Using these parameters and charts w e can obtain the variances w h i c h w e had
before. H o w these charts have been c o m p u t e d is irrelevant so far. All w h a t
w e n e e d t o k n o w is that t h e y can be d e d u c e d from samples only.
81

The variogram parameters in this case are as f o l l o w s : the horizontal range


is 4 0 0 ' in any direction. The vertical range is 1 0 0 ' , t h e nugget effect is
0 . 2 0 ( % ) C u and t h e variance is 0 . 8 0 ( % ) C u . In other w o r d s , w e have, as a
2 2

p o i n t variogram, using standard geostatistical n o t a t i o n s :

ah
400' a v = 100' C = 0.60

C 0 = 0.20 C 4- C 0 = 0.80
These values have b e e n estimated from a 2 0 0 ' x 2 0 0 ' x 5 0 ' drilling grid.
The corresponding variograms are s h o w n on Fig. 6 2 . We will later see h o w
the p o i n t variogram is d e d u c e d from a 5 0 ' sample variogram.

Computing block variance with the F-chart


Without discussing the reason for this formula, let us just m e n t i o n it in
order t o apply it.
The variance o\ of a b l o c k h' x h' x /' w h e n t h e variogram has parameters
C , C and a is:
0

ol F \ ^ , - ^ \ I C
h

a a a ''

T h e vertical distances m u s t be c o m p a r e d t o the vertical range and the


horizontal distances t o the corresponding horizontal range.
For blocks 1 0 0 ' x 1 0 0 ' x 50':

50 100 100
ol = C 1-F
ll00'400'400

a\ = 0 . 6 0 [ 1 - F ( 0 . 5 , 0.25)/C]

a% = 0.60(1-0.325)

o% = 0.405

Co *.03

200 400 600 800 1000 1200 1400 1600 1800 F E E T

Fig. 6 2 . Variogram of 5 0 ' core in three directions for t h e s i m u l a t e d d e p o s i t of Exercise


3.2.5.
82

In fact in our case, blocks are n o t really blocks, but rather the average of
a certain n u m b e r of D . D . H . and w h e n that n u m b e r is small, the correction
factor t o be added is n o t negligible: it is the nugget effect divided b y the
n u m b e r of samples in a block, i.e., for 1 0 0 ' blocks it is 0 . 0 5 / 4 = 0 . 0 1 2 5 s o
that o% = 0 . 4 1 8 .
For blocks 1 5 0 ' x 1 5 0 ' x 5 0 ' :

2 „™L 50 150 150\ / ~

ol = 0.60[1-F(0.5,0.375)/C]

ol = 0.60(1-0.38)

o 2
B = 0.372

For blocks 200'x 200'x 50':

ol = 0.60 [ 1 - ^ ( 0 . 5 , 0.5)]

o2
B = 0.60(1-0.45)

ol = 0.330

For blocks 250'x 250'x 50':

ol = 0.60 [ 1 - F ( 0 . 5 , 0.625)]

ol = 0.60(1-0.51)

ol = 0.294

Comparison with experimental values


These values compare well with the values obtained from the previous
histograms assuming a complete perfect knowledge of the deposit.

°100 — 0.-157 versus 0.420

0150 = 0.413 versus 0.372


n2
- 0.362 versus 0.330
°200

°250 0.318 versus 0.294

One of the reasons for the small discrepancy is that there exists m o r e than
o n e w a y t o divide the d e p o s i t i n t o blocks and the theoretical variance is the
average value of all the possible divisions, while the experimental variance
corresponds t o o n l y o n e division. We c o u l d c o m e back t o the original D . D . H .
intersection values and group t h e m differently t o obtain other estimates of
°\oo • • •

Estimates from a limited number of blocks


C o m i n g back t o Figs. 44—47 w e also have estimates of t h e previous vari­
ances, c o m p u t e d o n half of level 1 of our deposit. Keeping this v o l u m e
83

constant, the n u m b e r of blocks decreases as their size increases. T h e results


obtained w i t h 3 0 t o 4 0 0 blocks values are given b e l o w and also compare well
with the forecasting of t h e m o d e l .

a'100
2
= 0.46

a?c
r
150 = ft 0.42

a?
/
2n0n
0 = 0.35

a?co
;
250 = 0.32

Remark on the variance of blocks


T h e variances w h i c h are c o m p u t e d from a limited n u m b e r of blocks are
n o t equal t o the theoretical o n e s d e d u c e d from the variogram. T h e reason is
that t h e values o b t a i n e d from t h e variogram are e x p e c t a t i o n s for all t h e
possible blocks in the deposit, s o our estimate is in turn a random variable,
the average value of w h i c h is the forecasted value.

Computing the estimation variances with the charts


Using the same parameters as previously b u t n o w considering the esti­
m a t i o n variance chart, for a square b l o c k estimated b y its central D . D . H . w e
find, since t h e formula is:

a\ = C + Ca 2
(-,-)
0
\a a j
where C is the nugget effect of a D . D . H . intersection of 5 0 ' :
0

For blocks 1 5 0 ' x 1 5 0 ' x 50':


h = 15P = I _ _50 _
a 400 ° d 7 5
' a ~ 100 " °- 5

after Fig. 1 5 9 :
a 2
= 0.050

so that ol = 0 . 0 5 + ( 0 . 6 0 ) ( 0 . 0 5 0 )

a\ = 0 . 0 8 0 ^a e = 0 . 2 8 3 % Cu

For blocks 2 5 0 ' x 2 5 0 ' x 5 0 ' :

h 250 / 50

« Io~o = °-
= 625;
« = Wo = °- 5

after Fig. 1 5 9 :

o 2
= 0.12

ol = 0 . 0 5 + ( 0 . 6 0 ) ( 0 . 1 2 ) = 0 . 1 2 2 - > a = 0 . 3 4 9 % Cu e

These theoretical values c o m p a r e again well w i t h t h e experimental esti­


mates obtained from a single partition of t h e w h < ^ Hor*r»si+. c i n ^ o w o f m m H
84

after Figs. 5 0 and 5 1 :

for the 1 5 0 ' p a t t e r n a\ = 0.073 or oe = 0 . 2 6 9 % Cu;

for t h e 2 5 0 ' pattern ol = 0.104 or a e = 0 . 3 2 4 % Cu.

S o again, the variogram is sufficient t o c o m p u t e the precision associated


t o a given drilling pattern.

4 . 6 . 2 Back-of-cigarette-pack geostatistics

In this exercise, w e w o u l d like t o s h o w t h e t y p e of information w h i c h is


c o m p u t a b l e o n the back of a cigarette pack in a few m i n u t e s as long as o n e
has c o m p u t e d the variogram of a vein t y p e d e p o s i t and f o u n d that it follows
a particular m o d e l , very c o m m o n in practice, w h i c h is called the D e Wijsian
model.
The data. Let us suppose the f o l l o w i n g deposit, illustrated in Fig. 6 3 . It is
a subvertical sulphide vein w h i c h has been sampled b y three rows of drill­
holes 2 5 ft. apart and where the estimated dimensions of levels are as indi­
cated on t h e sketch. The variogram is readily c o m p u t e d from these three
lines of samples and f o u n d t o p l o t as a straight line on a log scale (for dis­
tance). It can be given the e q u a t i o n y(h) = 1.2 Inh + b. ( N o t e , o n e writes
1.2 = 3 a )
The questions w h i c h w e w a n t t o answer are as f o l l o w s :
— k n o w i n g that it has been established that the breakeven p o i n t is 2 . 1 5 % Cu,
w h a t are the chances of going broke if the d e p o s i t is m i n e d ?
— k n o w i n g that it is planed t o m i n e this b y b l o c k caving and that everyday
a block of 2 0 ' x 2 5 ' x 2 5 ' will be m i n e d , w h i c h variation should w e e x p e c t
for the daily fluctuations, m o n t h l y and quarterly?
— k n o w i n g that a one-day b l o c k p r o d u c t i o n will be estimated from o n e
central intersection (see Fig. 6 4 ) w h a t is the precision w i t h w h i c h this block
will be k n o w n ?

Fig. 6 3 . Schematic representation o f the deposit of Exercise 4 . 6 . 2 .


85

Fig. 6 4 . S c h e m a t i c v i e w o f m i n i n g b l o c k s for t h e d e p o s i t o f Exercise 4 . 6 . 2 .

The tools to answer. We first have t o c h e c k for the isotropy of our vario­
gram. This is easily d o n e as f o l l o w s : c o m p u t e the variance of the samples
o (%/D) = X(X —X) /(n
2
t — 1) where X is grade of a sample of length / and
2
t

n is the number of samples. This is an experimental value. It can be s h o w n


that if the deposit is isotropic this value should be equal t o 1.2 In L/Z where
L is the so-called linear equivalent of the deposit, given b y L — a 4- b 4- c/2 =
5 0 0 ' 4- 3 5 0 ' 4- 2.072 = 8 6 0 ' . If b o t h theoretical and experimental values c o m ­
pare well (there is n o test y e t t o qualify the "well") then the d e p o s i t is
isotropic.
N e x t , there is a formula — t o be d e m o n s t r a t e d later — w h i c h states that
for a b l o c k V o f height H length L thickness /,-, w i t h n samples h feet
t i9 i9 (

apart in a central line t h e estimation variance is:

In our case numerical estimation i m m e d i a t e l y gives, k n o w i n g that samples


are h = 2 5 ' apart:

o2
ni = 0 . 2 5 3 1 (%) 2

< - 0 . 1 5 7 1 (%) 2

a 2
n a = 0 . 1 8 9 8 (%) 2

The overall estimated grade of the deposit is:

4- m .|3

Yi + V 4- V 2 3

m = 2 . 4 3 4 % Cu
The estimation variance of this m e a n is:
86

m
' lv, + V 2 + V 3 ° m
> [Vi + v 2 + v :

w h i c h gives o = 0 . 0 7 0 8 ( % ) C u .
m
2

Therefore t h e standard deviation of the m e a n grade of the deposit is:


am = 0.266%
K n o w i n g this, w e can normalize t h e breakeven value 2 . 1 5 t o determine the
probability o f going broke:

= iX^m 1=s (2.15-2.434)


a 0.266
From the normal probability table, this yields a probability of 8 5 . 6 % of
success and 14.4% chances of failure.
The estimation variance of a 2 5 ' x 2 5 ' x 2 0 ' block c o m e s from Fig. 1 5 3 ,
and for a/h = 2 5 / 2 0 = 1.25 w e read a / 3 a = 0 . 8 ; and 3 a = 1 . 2 , hence 2

ol = 0.96 ^ 1 (%) and a = l % C u .


2
e

This shows a high uncertainty on the grade of individual blocks estimated


this way.
The variance of daily production within a month ( 2 0 days) will be given
by the following formula t o be discussed in chapter 7:

L 5 0 0 + 2 5 4- 2 0 / 2
oHv/V) = 3aln y = 1.2In 2 5 + 2 - 6 + 2 Q / 2 2 . 6 3 (%)*

t h e n : o(v/V) = 1.62%
This s h o w s that o n e should e x p e c t big daily fluctuations. N o w let us see if
w e c o u l d e x p e c t this t o average o u t over a longer period of time.
The variance of quarterly production over the mine life:

so that over three-month periods the grade fluctuation will still s h o w a


standard deviation of abouty^O.46 = 0.7% Cu.

4.7 C O M P U T I N G A N I S O T R O P I C V A R I O G R A M (after t h e F o n t a i n e b l e a u s u m m e r
s c h o o l , F o n t a i n e b l e a u is w h e r e t h e C e n t e r o f M a t h e m a t i c a l M o r p h o l o g y is)

Having seen an e x a m p l e of the kind of c o m p u t a t i o n s w e can m a k e from a


variogram, w e should get a feel a b o u t h o w t o c o m p u t e o n e . This will be our
last step in our "empirical" approach t o the solution of the problem.
87

035 035 0 33 033 034 031 035 0 37 041 041

035 035 035 035 0 33 041

037 035 037 035 0 37 037 039 039 041

037 040 042 034 036 041 034

037 041 033

035 042 033

039 031

O30

F i g . 6 5 . Drill h o l e v a l u e s o n a r e g u l a r grid t o s e r v e as d a t a o f E x e r c i s e 4 . 7 .

We will consider a simple stratiform deposit w i t h regularly gridded sam­


ples and occasionally missing ones. T h e different grades are given on Fig. 6 5 .
One should always remember that c o m p u t i n g a variogram along a line or
in one direction means that one considers the grade variations t o b e h o m o ­
geneous in that direction. It m a y occur that in a d e p o s i t these variations are
n o t t h e same in all directions. For instance, variations along strike are less
rapid than across strike. Thus o n e should c o m p u t e a variogram for a n u m b e r
(4 or 8) directions in order t o c h e c k the possible anisotropics. In this case
w e will consider t h e E—W ( 1 ) , N - S ( 2 ) , S E - N W ( 3 ) , S W - N E (4) directions,
for intervals going from 1 t o 3 . When p l o t t i n g the variograms one should
ensure that in directions 3 and 4 the increment in distance is a>/2~ if a is
the grid size.
T h e c o m p u t a t i o n formula is simply for direction a:

y«(h) = ~ } I [Z(x i + h )-z( )]


X i
2

where N(h) is the n u m b e r of samples e n c o u n t e r e d for l a g h .


The results are given in Table 4.1.
Drawing the four curves on the same graph (Fig. 6 6 ) w e find t h e m t o be
identical. We can thus consider t h e d e p o s i t t o be isotropic and c o m p u t e an
average variogram w h i c h is given in Table 4.II.
T o m a k e t h e exercise c o m p l e t e , w e will find an equation t o represent this
88

*(h)t

a 2a 3a 4a h

Fig. 6 6 . Variograms o f t h e data o f Fig. 6 5 , c o m p u t e d in four directions.

T A B L E 4.1

Variogram values for different directions for data o f Fig. 6 6

Lag h = l h = 2 h = 3

N(l) 7(1) N(2) 7(2) AT(3) 7(3)

Direction 1 24 4.1 20 8.4 18 12.1


2 22 4.25 18 8.2 15 10.9
3 19 5 16 12 10 17.3
4 18 6.5 14 11.3 8 15.4

T A B L E 4.II

Average isotropic variogram for data of Fig. 6 6

h N(h) y(h)

a 46 4.1
ay/2 37 5.7
2a 38 8.3
2a\[2 30 11.6
3a 33 11.5
3a>/2 18 16.3
89

variogram: the equation of the m o d e l c h o s e n will be linear. A n acceptable


m o d e l w o u l d be y(h) = Ah I a.

4.8 A N A L T E R N A T E V A R I A B L E TO T H E G R A D E : THE A C C U M U L A T I O N

4 . 8 . 1 The particular case of stratiform deposits

A l m o s t everything w h i c h has b e e n written before, supposes that the


samples are of equal length, at least for variogram c o m p u t a t i o n . There is
o n e simple case, however, where t h e y have different lengths b u t never­
theless yield easy c o m p u t a t i o n s . It is the case of stratiform deposits.
Imagine a stratiform or vein-type d e p o s i t for w h i c h w e have a n u m b e r
of intersections, going from hanging wall t o footwall. We have a series
of grades, thicknesses and coordinates. In m a n y cases these are the kind of
data which will lend themselves t o a very quick geostatistical estimation.
The variable w h i c h will be studied h o w e v e r is n o t directly the grade. One
should n o t compare the grades of intersections of different lengths. Rather
o n e considers a n e w variable, familiar t o gold-miners: (feet) x (%) or inch-
dwt. This is proportional t o a metal quantity. A simple c o n s t a n t factor trans­
forms (feet) x (%) i n t o p o u n d s of metals given a block size and a tonnage
factor, since the quantity of metal Q, in a b l o c k of height h area A and iy t

grade Z w i t h a tonnage factor F is:


t

Qi = hiAiZJF

Qi = (hiZfiAJF
One will c o m p u t e the variogram of accumulations. N o t e that the thick­
ness is also a regionalized variable and that t o evaluate the average thickness
of a block o n e will have t o weight the information from surrounding inter­
sections. One will estimate the thicknesses. Thus o n e needs the variogram
of thickness. Finally, in order t o estimate the grade of a block, o n e will
divide the estimated accumulation of a block b y the corresponding estimated
thickness. Examples of this procedure are given in David ( 1 9 7 1 ) , Journel

Fig. 6 7 . Variograms o f t h i c k n e s s a n d a c c u m u l a t i o n in skarn d e p o s i t .


90

»(h)

F i g . 6 8 . V a r i o g r a m o f t h i c k n e s s , g r a d e a n d a c c u m u l a t i o n in a c o p p e r v e i n d e p o s i t .

(1974), David and Dagbert ( 1 9 7 6 ) . A c o m p l e t e e x a m p l e will be s h o w n in


Chapter 8. Sandefur and Grant ( 1 9 7 6 ) n o t i c e that t h e accumulation is t h e
variable t o work w i t h in roll-front deposits as it is proportional t o a quantity
of U O 3 s per square f o o t .

4.8.2 Examples

In Figs. 6 7 and 6 8 s o m e accumulation variograms are s h o w n , together


w i t h the corresponding thickness and grade variograms. One sees that as it
happens in m a n y instances, grade appears as a random variable and accumu­
lation is just a reflection of t h e c o n t i n u i t y of thickness of t h e ore.

You might also like